MOLECULAR GENETICS OF ASTHMA AND ALLERGY

1 Introduction
2 Early Linkage Studies
3 Linkage studies of the chromosome 11q atopy gene
4 Association studies of the chromosome 11q atopy gene
5 Linkage studies of the chromosome 5 atopy gene
6 Association studies of the chromosome 5 atopy gene
7 Chromosome 12 atopy genes
8 Chromosome 13 atopy genes
9 Chromosome 17 atopy genes
10 Other asthma/atopy loci
11 Genome scans for asthma and atopy
12 Genome wide association scans for asthma and atopy
13 HLA and atopy
14 Tumour Necrosis Factor and atopy
15 Filaggrin
16 The IL-4 receptor
17 CARD4 and CARD15
18 The IL-9 Receptor
19 PAF acetylhydrolase
20 PLA2G2D and asthma
21 Drazen et al and others on 5-lipoxygenase variants
22 LTC4 synthase variants and aspirin sensitive asthma
23 Glutathione-S-transferase and asthmal
24 CTLA4, CD28, and IgE level
25 CASP10, asthma and IgE level
26 DPP10, asthma and IgE level
27 Tenascin-C and asthma
28 GPR154 and asthma
29 PTGDR and asthma (Oguma et al 2004)
30 IL1RL1 and asthma (Shimizu et al 2005)
31 Chromosome 11p atopy genes
32 IL18 and atopy
33 CHI3L1, YKL-40 and asthma
34 Van Eerdewegh et al on ADAM33 [2002]
35 Superoxide dismutase and asthma
36 Adenosine deaminase and asthma
37 Frohlander & Stjernberg and others on haptoglobin
38 Srivavasta and C3 Complement type
39 Oxelius and Gm
40 Angiotensin Converting Enzyme
41 The delta-F508 variant of CFTR
42 FOXP3 and XLAAD
43 Animal models of asthma and bronchial responsiveness

This document was last modified on 23-September-2009.


1 Introduction

A list of candidate genes for a role in asthma and allergy pathogenesis can be generated for all the mediators and their receptors described earlier, the receptors characteristic of particular effector cells involved in allergic inflammation and asthma, and of course IgE and its receptors. A number of these candidates are currently under investigation. Likely ones are some of the interleukin receptors eg the IL-5 receptor (located on chromosome 3p26), IL-4 receptor (16p12.1-p11.2); the immunoglobulin genes on 14q; the high affinity IgE receptor (assigned as part of a search described below to 11q13); the HLA region; the T-cell receptor (on 14q11.2 - alpha and delta subunits, 7p15-p14 - gamma subunit, and the beta subunit on 7q35) [Genome Data Base 1993]. Other regions of recent interest are the chromosome 5q cluster of interleukin genes (including IL-4 and IL-5), close to the beta-2-adrenoceptor gene (ADRB2), and interferon-gamma on 12q.

Some readers may find the organisation of this section slightly confusing. Although some of the earliest studies in this area started in the HLA region looking for allergen specific human equivalents of mouse Immune Response genes, I have chosen to start with the positional-based search for a (generalised) atopy gene, culminating in the era of genome scans. I then return to the older candidate based approach.


2 Early Linkage Studies

2.1 Zieve et al [1936]

This paper, "On the linkage relations of the genes for allergic disease and the genes determining the blood groups, MN types and eye colour in man" [1936], must be the first of its kind. A sample of 66 nuclear families (383 individuals) were ascertained. Diagnosis of asthma, hayfever, urticaria, eczema, angioneurotic oedema and gastro-enteritis was made by the investigators based on personal interview. Preliminary segregation analyses were performed [Wiener et al 1936], which were supportive of an intermediate dominant hypothesis, with (increasing) homozygotes suffering early onset disease, and heterozygotes more likely to develop disease after childhood. AB0 and MN blood group was determined in 56 complete families (317 subjects plus 16 in incompletely typed families). After excluding nonpaternity etc, based on the blood grouping, linkage analysis was carried out under the preferred segregation model using the method of Wiener [1932]. No linkage to blood group or eye colour was found.

Reanalysis of the pedigrees using MLINK and SIBPAL actually finds weak hints of linkage to AB0. Mean identity by descent sharing at the AB0 locus among affected sib pairs is 0.61 (SE=0.26, P=0.02), while maximum likelihood linkage analysis under an additive SML model (q=0.5, f2=0.6, f1=0.3, f0=0.0), finds a maximum lod score of 0.44 at recombination fraction r=0.00. In view of the small families and less than fully informative marker used, these results are all the more surprising.

2.2 Eiberg et al [1985]

The next study I am aware of dates to 1985. This abstract describes a preliminary linkage analysis for total sIgE level. A maximum lod score for Kidd blood group of 1.0 in males and for both sexes a lod score of 2.07 for Esterase D (recombination fraction=0.00). This would place a gene regulating overall IgE level at 13q14. This finding was not replicated by Cookson et al [1989]. More recently, Daniels et al [1996] and several other linkage studies have found evidence of linkage of atopy to this region.


3 Linkage studies of the chromosome 11q atopy gene

3.1 Oxford group

The group including William Cookson and Julian Hopkin has published several papers [Cookson & Hopkin 1988; Hopkin 1988; Cookson et al 1988; Cookson et al 1989a; Cookson et al 1989b; Cookson et al 1992; Young et al 1992; Sandford et al 1993], describing a large family study of atopy including RFLP and later STRP based linkage analysis. Initially they examined the nuclear families of 20 (skin) atopic asthmatics from a respiratory outpatient clinic (age 12-15 years) and 20 control families ascertained via similarly aged children admitted to hospital for appendicectomy (12), pneumothorax (6) or pneumonia (2). A further 158 subjects from three large extended families with strong histories of atopic disease were also recruited via letters to general practitioners and newspaper publicity. This total of 394 subjects was extended to a total of over 500 subjects in a later analysis [Hopkin 1988]. No statistical allowance for ascertainment bias was made in this initial analysis.

3.2 Initial segregation and linkage analysis

Subjects completed a modified version of the ATS-DLD-87 questionnaire which included items on frequency and seasonality of symptoms. All underwent skin prick testing for 11 common aeroallergens, and had blood collected for total and specific sIgE levels. Affected individuals satisfied any one of the following criteria:
  1. one or more skin prick wheals with a mean diameter 1 mm greater than the negative control;
  2. a total sIgE more than two standard deviations above the population mean;
  3. one or more positive (not defined) RAST results.

The norms for sIgE were age-sex-standardised [Hopkin et al 1993]. The definition of skin atopy as being 1 mm or greater than the control would be lower than that used in a number of other studies, and would proportionately increase the number of subjects diagnosed as being atopic. Similarly RAST results are commonly reported in semiquantitative form; the cutoff for a positive reaction may vary from kit to kit. I note this because other workers (eg Sibbald 1988) have criticised the definition of atopy used in this study.

Within the nuclear families examined, 80% of the parents defined as being atopic fulfilled two out of three criteria, 12% had an elevated total sIgE, and 8% a positive skin test alone. While 83% of all atopic subjects met questionnaire criteria for asthma or hayfever (defined as rhinitis and conjunctivitis), only 30% specifically reported diagnoses of these syndromes. Of the nonatopic subjects, 13% met these criteria. Within the nuclear families, significantly more of the parents of cases were atopic than were those of controls (X2(2 df)=11.2, P=0.004; see Table). This gives an unadjusted odds ratio for atopic asthma versus (any) parental history of atopy of 11 (95% confidence interval 2-60), and a population prevalence estimated from the control parents of 25% (11-38%), not dissimilar from most estimates in the literature, as discussed earlier.

Among the extended families, the authors felt that the segregation of atopy was most consistent with an autosomal dominant gene of fairly high penetrance (Table). In only four nuclear families out of 31 producing atopic offspring were the parents nonatopic as defined by this study; in each case, one of the grandparents was atopic, and in three families, the linking parent had a history consistent with childhood rhinitis.

Table 1. Parental affection status of cases and controls from 40 nuclear families [from Cookson & Hopkin 1988].
Group of childrenNumber of parents atopic
NoneOneBoth
Atopic asthmatics (20)2117
Control children (20)1181

Table 2. Affection status of offspring of mating types within three extended families with high prevalences of atopy [from Cookson & Hopkin 1988]
Mating TypeUXUUXAAxA
Atopic offspring4/?31/478/12
Percentage--66%75%

The linkage study was performed using seven extended families, three of these being those described above, others recruited from, among other sources, the authors' respiratory outpatient clinic. A two liability class model was used: the first class contained subjects where the diagnosis of atopy or normality was clearcut; the second class included those subjects where skin prick testing and total IgE were equivocal. For example, as sIgE diminishes with age, nonatopic individuals (by the authors' criteria) aged over 60 years were assigned to this second liability class. Penetrance for the first liability class was modelled as being 99%, and 75-90% for the second class. Two particular kindreds, with a high prevalence of cigarette use, contributed the bulk of this second liability class.

The strategy used was a chromosome by chromosome search - see Table. The seventeenth probe examined, p lambda-MS.51 (D11S97), showed 8 recombinants from 53 meioses. A further 69 meioses were then recruited. The maximum lod score for all the families was 5.58 at a recombination fraction of 10.5%. Subsequently, more families were included [Cookson et al 1989]. This led to a maximum lod score of 6.39 at a recombination fraction of 10%. For the original data, four families contributed a maximum lod score of 5.24 at a recombination fraction of 6%, with the second liability class members contributing little to the lod score. The remaining families, where cigarette use was high, also suffered from poor family structure, with a small third generation and many uninformative meioses.

Hopkin [1990] summarised further work. The number of families included now numbered 60. Most of these were relatively small kindreds. The lod score remained highly significant at over 5. There was no evidence of genetic heterogeneity on testing. This test would not be very powerful if most families were small.

This work gives us a model for atopic disease in terms of a single gene, backed up by the linkage analysis. The gene frequency is between 15-25%. The penetrance for any atopic symptoms was 85%, for wheeze 60%, and for diagnosed asthma 20%. If one uses the higher figure for the prevalence of atopy in the general population, this equates to a gene frequency of 0.16; if the prevalence is 30%, the gene frequency 0.19. Using a gene frequency of 0.16 and a penetrance of 85% gives VA of 14% and for VD of 1%. Using the formulas presented by James [1971] and Risch [1990], this suggests the population relative risk (recurrence risk compared to the general population risk) for atopy in the parents or siblings of atopic children would be 2.1, and for MZ twins would be 3.4 (see also Section 11.5.2). If both parents have no symptoms of atopy, the risk to the children of developing wheeze is approximately 7%.

In the same pericentromeric region of Chromosome 11, one finds CD20 (11q12-q13.1), uteroglobin, complement component 1 inhibitor (hereditary angioedema, 11q12-13.1), Multiple Endocrine Neoplasia I (11q12-13), muscle glycogen phosphorylase (11q12- q13.2), the gene associated with McArdle syndrome, NFKB3 (the beta-3 subunit of the nuclear factor of Kappa light chain gene enhancer in B cells) and various pepsinogens (3-5, Gp I) making up pepsinogen A (11q13). In addition, this region is homologous to the mouse 7F (T50H) region, associated with parental imprinting effects [Hall 1990]. A number of human disease genes existing in other homologous areas have already been demonstrated to exhibit maternal (Wilm's tumour) or paternal (Huntingdon's disease) imprinting; this might extend to atopy.

3.3 Further Linkage Analysis

In the period since the last paragraph was written, evidence supporting this latter prediction was published. This group published a further two papers in the Lancet, which included GM Lathrop as an author. Young et al [1992] described a separate analysis of sixty-four nuclear families. These were ascertained through children (age under 15 years) reporting eczema, hayfever or asthma. The same diagnostic criteria as before were used. The model assumed a gene frequency of 0.2, and a 95% penetrance. Genetic homogeneity was assessed using the test proposed by Smith [Ott 1991].

Under an assumption of equal recombination distance in both sexes, the maximum lod score was 3.8, but there was strong evidence for a male-female difference. Relaxing the equality constraint led to a maximum lod score of 5.2, with rM=0.18 and rF=0.001. The proportion of unlinked families was estimated at zero, with a 95% upper limit of 40%. The marked sex difference was left unexplained.

Cookson et al [1992] described sib-pair analyses that examined sharing of maternal and paternal marker alleles. All subjects genotyped up to August 1991 were included (N=723). Affected sib-pairs were significantly more likely to share a maternal Chromosome 11 than a paternal Chromosome 11, whether using an affected phenotype defined as any skin test 4 mm or more in diameter (maternal allele shared 30/43 and paternal alleles shared 24/42), a high total sIgE (maternal 51/81; paternal 28/59), or the combination of three measures described earlier (maternal 125/203; paternal 83/179). There was no significant sharing of alleles in unaffected sibling pairs in the informative families. A "hot spot" of male recombination could not be found to explain the sex difference. In families where the father was atopic, and the mother nonatopic, the sharing of maternal alleles was 19/32 (0.59, exact 95% CI 0.41- 0.76), which trends in the same direction as the previous results. The authors note that alternative explanations include phenotypic maternal effects, so a pair of sibs carrying this atopy gene are more likely to express it after transplacental or breast milk exposure to maternal IgE or other factors. Since the presence of atopy in offspring of atopic fathers is increased, they also concur that the disease must be genetically heterogenous.

Finally, this gives a possible explanation of the failure to replicate the original linkage between 11q13 markers and atopy in other studies (see Section 6.1.4 et seq). If sex of the parent is ignored, obviously the evidence for linkage will be attenuated: for atopy via three measures - the phenotype giving the strongest evidence - affected sib pairs shared 208 alleles out of 382 (0.54, exact 95% CI 0.49-0.60), not significantly different from the null expectation of 192.

The second paper [Sandford et al 1993] gives multipoint linkage information. Because of the presence of probable maternal inheritance and genetic heterogeneity, sib-pair methods of analysis were again used. Two novel STRP's in the 11q13 region were described. One CA-repeat marker was derived from FCER1B DNA, the other was derived from a cosmid clone.

Examination of the maternal or paternal origin of marker alleles in affected sib-pairs again gave strong evidence for maternal transmission (D11S97 paternal alleles shared 45/87 and maternal alleles shared 72/100; FCER1B-ca paternal 43/84, maternal 56/67; cosmid clone marker cCl11-319ca paternal 28/60 and maternal 56/67 - in each case maternal allele sharing is markedly greater than 50%). The best ordering of the markers was D11S97, cCl11-4, cCl11-319ca, CD20, FCER1B-ca, cCl11-222. The atopy locus mapped using maternally derived alleles into the (95% support) region running from cCl11-4 (with the lowest estimated recombination fraction r=0.005) to FCER1BI-ca (r=0.03). There was no evidence for allelic association (linkage disequilibrium). Critically, the authors note that in 16 of 88 families informative for loci within this region, atopy was not linked to 11q13. They commented that linkage might be spurious in approximately this number of families again, as the phenotype was so common. Therefore approximately 60% of atopy in these families would be attributable to the 11q gene. The FCER1B gene (MS4A2)is therefore a strong candidate gene, as presumably is CD20 (later excluded), though I can find few references ascribing a function to this cell surface marker. A table summarizes the subsequent replication attempts described below.

Table 3. Gene probes used by the Oxford group [1988, 1989, 1992, 1993].
Genetic ProbeChromComment
pYNZ21
pYNH 2
p-lambda-g3 7
collNJ3 7
p-lambda-MS.51
D11S97 11q12-13Initially described probe linked to atopy; first to show evidence of maternal imprinting
FCER1B-ca11q13Probe with strong evidence of linkage to atopy; strong candidate
CD2011q13Marker with strong evidence of linkage to atopy; Possible candidate (B-cell surface marker of unknown function).
CCl11-31911q13Probe with strong evidence of linkage to atopy
CCl11-411q13Probe with strong evidence of linkage to atopy
CCl11-22211q13Probe with strong evidence of linkage to atopy
InsHVR11p
Esterase D13previous report of linkage [Eiberg et al. Cytogen Cell Gen 1985;14:622]
WC2513
WC8313
p7f1213
AW10114Excludes variation around IgE H chain locus
3' a globin16
p79-2-2316
pYNZ2217
RMU3 17
Jeffries 33.15multipleposs linkage with asthma [Brereton et al 1990] Jeffries 33.6 multiple
M13 gene 3 tandem repeatmultiple
zeta intron HVRmultiple

3.4 Inacio et al [1990]

This abstract [1990] was the first published attempt at replication of the Oxford group's findings. Seventeen Portuguese families were examined (14 two generation and 3 three-generation, total N=83). Half (40/83) of the subjects were atopic. It is implied but not stated in the abstract that all these individuals were pollen sensitive. There was no evidence of linkage to the lambda-MS51 probe site on 11q. They note that "[t]he observed patterns were distinct and different from those reported in the literature". Apparently [D. Meyers, pers comm], this refers to the fact that the Pasteur Institute group could not identify the particular length restriction fragments (alleles) described by Hopkin et al [1989]. However, all the atopic individuals did carry the HLA DQA2-DQB2 haplotype (see below).

3.5 Shirakawa et al [1991, 1994]

A successful replication of Cookson et al was reported by a Japanese group at HGM 11 [1991]. There were 274 atopic asthmatic probands, but only 136 families were finally selected - the remainder were excluded because of "discrepancies between measurements" of the three measures of atopy used [as per Cookson et al]. Of these, only 4 families containing more than 15 meioses each (out of the 8 families eligible) were used to examine linkage to D11S97 (N=69 meioses). One family gave a lod score of 2.98 at r=0.01, the other three 0.6 to 0.7, so that the four families gave a total lod score of 4.88. The genetic model chosen is not described, but is presumably the same autosomal dominant model of Cookson et al. Shirakawa et al [1994] revises this maximum lod score to 9.35 for D11S97 and FCER1B-ca under the assumption of unequal rates of maternal and paternal recombination, and report evidence for maternal transmission in 2 families.

3.6 Amelung et al [1992]

These authors [1992] described a continuing study in Groningen (Netherlands). The group includes Deborah Meyers. Twenty families (N=117 typed individuals) were ascertained through a proband diagnosed as asthmatic in 1962-70. Linkage to both atopy and BHR was tested for, using both LINKAGE under two different genetic models, and sibpair analyses. The lod score was below -2 for r lt;0.12 with the PCR-based marker INT2. A sib-pair analysis found that of the (possible) 54 affected pairs, 28 shared the same INT2 allele as the affected parent, and 26 were discordant. In addition, they examined D6S105, tightly linked to the HLA-B region. In this case, a lod score of -2 was found at r=0.07.

3.7 Hizawa et al [1992]

In contrast to Shirakawa et al [1991], this Japanese group [1992] found no evidence for linkage between D11S97 and atopy in four families (N=60). A number of definitions of atopy were used, including that of Cookson et al. Using this definition in the LIPED analyses, linkage was excluded below r < 0.04. Tightening the definition led to stronger evidence for exclusion.

3.8 Lympany et al [1992]

This is another attempted replication of the 11q linkage findings [1992a; 1992b] in nine families (2 and 3 generation, N=89). The phenotypes studied were skin atopy (a panel of 10 common allergens), positive RAST (to the same 10 allergens), as well as BHR to methacholine using the Yan protocol. This group used VNTRs with the lambda-MS51 probe (D11S97). The heterozygosity at this locus was 77% (9 alleles). No evidence for linkage (using Liped) was found for any of seven phenotypes defined. For atopy, the overall maximum lod score was -0.3 at 30% with Hinf1 and -0.04 at 30% with Taq1. Segregation analysis using REGD found atopy "not inherited in a simple mendelian fashion". They note that they were unable to reproduce the 10.8 kb allele described by Cookson et al.

Further analysis with INT2 and PGYM found no association with atopy, or linkage - for PGYM LOD<-2 for r<0.07; for INT2 for r<0.04. Altering the definition of atopy, or restricting it to BHR did significantly alter these results.

3.9 Rich et al [1992]

Blumenthal's group in Minnesota [1992] examined three large pedigrees (N=67 typed at D11S97). Atopy was defined as by Cookson et al, save that skin wheals had to be 5 mm greater than control. D11S97, HLA-A and HLA-B typing results, were used for linkage analysis (LINKAGE). In addition, sib-pair analysis (126 pairs) were performed using SIBPAL. Linkage with D11S97 was excluded for r<0.05, and with the HLA region out to r=0.23. Sib-pair analyses found no evidence for linkage to either region.

3.10 Coleman et al [1993]

Further linkage analysis of 11q13 was also performed in 95 families (N=407) with at least two first-degree relatives with active atopic eczema [1993]. Atopy defined as a positive skin prick test, or elevated specific or total sIgE was present in 80% of parents and 86% of offspring, and asthma or hayfever in 70% of the eczema probands. Ignoring the possibility of maternal effects gave a multipoint lod score (D11S97, PGYM, CD20, centromere) of -7.8 for 11q13. Sib-pair analysis (101 pairs) found a nonsignificant absence of sharing of maternal alleles (D11S97 paternal 21/47, maternal 23/61 - 38% with exact 95% CI=26-51%; PGYM 30/62, 25/61 - 41% with 95%CI=29-54%; CD20 9/16, 14/28). Linkage using subclasses of parental eczema found persisting negative lod scores for families where the mother was unaffected or both parents were affected. For 12 informative families where the father was unaffected, the lod score was +0.8. The authors conclude that a maternal effect might exist, but that heterogeneity is masking it.

3.11 Collée et al [1993]

This sib-pair analysis appeared as a letter to the Lancet [1993]. There were 26 affected sib-pairs from 26 nuclear families, using as the disease definition elevated sIgE (>100 IU/ml for over 10 year olds), a positive RAST (0.35 PRU/ml) and two or more symptoms on a modified MRC questionnaire. Families were typed at 3 loci - D11S97, PGYM, and FCER1B-ca. The proportion of alleles shared IBD for D11S97 was 18/23 (78%, exact 95%CI=56-92%); PGYM, 33/44 (75%, 60-87%); and FCER1B-ca, 17/28 (61%, 41-78%). Breaking this down by parental origin showed a tendency for sharing to be greater for maternal than paternal alleles, but because of the small sample size, this was not significant (Cochran-MantelHaenszel statistic X2(2 df)=2.84).

3.12 Neely et al [1994]

This abstract [1994] describes a linkage study of 355 sib-pairs (170 individuals) from 11 (Amish) pedigrees studied by the group headed by David Marsh. Three 11q13 markers were used: FCER1B-ca, INT2 and PGYM. No evidence for linkage with any of four phenotypes (total sIgE, specific IgE to D pter, D far, and a panel of 20 common aeroallergens) in a Haseman-Elston regression analysis using SIBPAL was found.

More recently, this study was extended to include one additional pedigree, further individuals in the existing pedigrees, and total sIgE was reassayed. Reanalysis finds some evidence for linkage to INT2 (FGF3).

3.13 Lawrence et al [1994]

Holgate's group [1994] have performed linkage and segregation analysis of 131 nuclear families recruited for having three or more children, but not specifically for atopic disease. "Evidence for major loci [was] suggestive, but there was no evidence for imprinting or linkage to 11q13". In a subsequent paper [Watson et al, 1995], these authors reported more fully these combined segregation-linkage analyses under one and two trait locus models.

A number of markers on chromosomes 1, 5, 11, 12, 16 and 19 were examined: D1S104, IL2RB, IL9, TGFB, IFNA, D11S480 (3.9 cM from FCER1B), D11S527, D11S534, IGF1, D16S298, D19S112, D19S177. No maximized LOD score greater than 0.57 (or less than -1.00) was obtained.

3.14 Brereton et al [1994]

This South Australian linkage study [1994] of 12 extended pedigrees in which atopy appeared to segregate in an autosomal dominant fashion failed to find linkage or association to 11q.

3.15 van Herwerden et al [1995]

In a Victorian sample of 123 sib-pairs [1995], by contrast, evidence for linkage of FCER1B-ca to methacholine BHR and diagnosed asthma was found. This study was nested within the European Community Respiratory Health Survey as part of which a random sample of 4500 individuals aged 25-40 were screened via the IUATLD questionnaire. A subset of 757 of these underwent methacholine challenge and skin prick testing to 11 common allergens. Subjects with a history of asthma (episode in previous 12 months, use of asthma medication, or nocturnal dyspnoea), atopy (any SPT > 3 mm), or BHR (PD20<2 mg cumulative dose of methacholine) were asked if they had a sibling, who was invited to undergo testing. A total of 137 pairs were recruited. Linkage of asthma, atopy and BHR to FCER1B-ca was tested an affected sib-pair identity-by-state analysis.

There was significantly increased sharing of alleles in 67 concordant pairs concordant for asthma (observed number of shared alleles=98, expected=83.1, P=0.002), 106 pairs concordant for atopy (O=145, E=131.4, P=0.02), 53 BHR concordant pairs (O=80, E=65.7, P=0.001). Among the 17 BHR pairs where there was not concordance for atopy, the sharing was still greater than expected (O=28, E=21.7, P=0.004); this was not the case in the 70 atopy concordant pairs not concordant for BHR (O=93, E=86.8, P=0.124). No evidence for parental imprinting was found when reported family history, and its interaction with ibs sharing was examined.

In Wong et al [1997], results from another three markers is presented: D11S987, D11S1314, D11S937. In this case, no increased sharing in any subgroups was detected, but these markers are probably too distant from FCER1B to obtain significant results given the sample size.

The authors' conclusion that they have demonstrated linkage to a BHR gene rather than an atopy gene may be a little premature. I am skeptical of the ibs methods here, as these can give inappropriately low P-values, as opposed to ibd methods (which are perfectly applicable).

3.16 Martinati et al [1996]

This Italian study [1996] describes sib-pair linkage analysis in 45 nuclear families (213 subjects) ascertained through atopic asthmatic children (median age 12 years, range 2-47), and containing at least one additional atopic member. Atopy was defined as per Cookson et al [1989]. Analysis of maternal versus paternal sharing at two markers (FCER1B-ca and CCl11-319ca) was performed on 128 affected sib pairs.

No striking evidence for linkage was detected, although the sharing of maternal alleles is in the expected direction. The maternal and paternal sharing at FCER1B-ca was 28/47 (60%, exact 95%CI=44-74%) and 20/50 (40%) respectively; for CCl11-319ca, 12/19 (63%) and 21/39 (54%). The difference between maternal and paternal sharing at FCER1B-ca is almost significant (OR=2.21, Fisher exact 95%CI=0.96-5.03).

3.17 Trabetti et al [1998]

An extension [1998] of the above study examined the second intron RsaI and E237G polymorphisms of FCER1B in a significantly increased total of 168 families (659 individuals) recruited through a child attending an allergy and respiratory medicine clinic. There were 57 sib-pairs where both exhibited BHR, 117 where both exhibited one or more positive skin prick tests, and 137 concordant for either a positive SPT or a sIgE level > 100 IU/ml.

Within the sample, the E237G allele frequency was 4%, and the int 2 RsaI B allele frequency was 56%. There was slightly increased IBD sharing at the i2 polymorphism within sib pairs (55% for BHR, P=0.05), with slightly stronger evidence from an APM test using SimIBD (empirical P=0.02 for BHR; empirical P=0.01 for a positive SPT). There was no evidence of allelic association using the TDT (number of parent-offspring trios not reported).

It seems typical that despite the large sample size, the strength of the evidence for linkage to the region remains quite low.

3.18 Noguchi et al [1997]

A similar study [1997] used 68 nuclear families (306 subjects) ascertained through atopic asthmatic children attending a Tsukuba (Japan) pediatric allergy clinic and their affected sibs (median age 10.5 years, range 1-29). The geometric mean sIgE in children was 670 U/ml.

The affected sib-pair analysis for atopy (log sIgE 1 SD above population mean, or a positive RAST to any of six allergens; 85 pairs), and asthma (recurrent wheeze and dyspnoea in the previous 12 months, spontaneous or bronchodilator reversibility; 46 pairs) supported linkage to 5q31-33 (see below) but not FCER1B-ca (mean ibd sharing 52%, P=0.43; and 51%, P=0.26). The results for a Haseman-Elston analysis of tIgE (using SIBPAL) were more encouraging for FCER1B-ca (t=-1.51, df=88, P=0.067).

3.19 Folster-Holst et al [1998]

This study involved 12 families (98 individuals) ascertained through a proband with atopic dermatitis. Commenges' WPC (APM) test was used to screen a set of 15 Chromosome 11 markers, and an additional parametric analysis was performed for FCER1B using MLINK and TMLINK (maximizing the lod score over 4 models). The best nonparametric result was a P=0.005 for FCER1B, while the highest lod score was 3.55 under a two-locus model: a recessive FCER1B-linked locus (r=0), and a dominant "background" locus. This lod score was contributed by only 2 families. By contrast, a lod of 0.8 was the best they coukld obtain under a single locus model.

3.20 Trabetti et al [1998]

Another Italian study examined the second intron RsaI and E237G polymorphisms of FCER1B in 168 families (659 individuals) recruited through a child attending an allergy and respiratory medicine clinic. There were 57 sib-pairs where both exhibited BHR, 117 where both exhibited one or more positive skin prick tests, and 137 concordant for either a positive SPT or a sIgE level > 100 IU/ml.

Within the sample, the E237G allele frequency was 4%, and the int 2 RsaI B allele frequency was 56%. There was slightly increased IBD sharing at the int 2 polymorphism within sib pairs (55% for BHR, P=0.05), with slightly stronger evidence from an APM test using SimIBD (empirical P=0.02 for BHR; empirical P=0.01 for a positive SPT).

3.21 Palmer et al [1998]

This West Australian study examined 121 nuclear families: 95 (442 individuals) ascertained via a child regardless of asthma status, and 26 (134 individuals) through a child with severe symptomatic asthma Multipoint linkage analysis using Haseman-Elston sib-pair methods provided evidence of linkage between chromosome 5q markers (D5S393, D5S399) and total sIgE levels (P=0.04), specific sIgE levels (P=0.04), and eosinophil counts (P=0.03), while chromosome 11q markers (FCER1B-ca and D11S480) were weakly linked only to specific IgE (summed D. pter and mixed grass RAST scores) level (P=0.03). No linkage to BHR (Yan protocol log His DRS) was exhibited by either region. Evidence of linkage to 5q was slightly stronger for total sIgE adjusted for specific IgE level (P=0.004), but this diminished the evidence for linkage to FCER1B (P=0.3).

3.22 Laprise et al [2000]

This paper describes both a case-control and a linkage study. In the latter, four extended pedigrees (106 individuals) in which asthma was common. All subjects underwent SPT and methacholine inhalation challenge, and 27 met the study criteria for the diagnosis of asthma. There is some evidence of linkage between FCER1B-E237G genotype and various atopic phenotypes (lods 1-2).


4 Association studies of the chromosome 11q atopy gene

4.1 Oxford based studies

Two important association studies confirmed the Oxford group's original linkage findings and, to some extent, their explanations as to why replication by other groups was so difficult. One study used a sample from the original Oxford families, the other, families from Busselton, West Australia.

Shirakawa et al [1994] first sequenced FCER1B in six atopic and six nonatopic subjects. Three (6th exon) mutations were found in one atopic individual leading to substitutions of isoleucine for leucine at position 181 (I181L), and valine for leucine at position 183 (V183L). A PCR based assay for these two mutations was developed (AS-PCR - allele-specific DNA amplification).

In a "random sample" of 163 patients unselected for allergic disease (undergoing venesection for other purposes), 25 were found to carry the Ile181Leu mutation, but none, the Val183Leu. A total sIgE greater than 100 IU/ml was present in 41 (25%), of whom 11 carried the Leu181 mutation (OR=3.1, 95%CI=1.2-7.5). A similar association was found for the presence of grass pollen specific IgE (OR=2.6, 1.1-6.4).

The Leu181 mutation was also found to segregate in 10 of 60 of the atopic nuclear families described earlier. In each family, the mutation was transmitted from the mother (and was present in the proband). Among 14 nonproband offspring, four were atopic, and two carried Leu181; none of the ten nonatopic offspring carried the mutation. Furthermore, the two "sporadics" arose in bilineal atopy - that is the father was atopic, and did not carry Leu181.

In the Busselton family study [Hill et al 1994], 232 nuclear families (1020 individuals, 556 children) were typed for the Leu181 mutation in the FcERI beta-subunit gene. This was found in 28 subjects. There were 8 children carrying the gene where the parent of origin was the mother - three with asthma, the remaining five with hayfever. Specific IgE and skin prick test wheals to house dust mite (as well as a sum of RAST scores) were significantly higher than in controls on Wilcoxon test - the most appropriate in view of the clustered nature of the data [Shirley & Hickling 1981].

Subsequently [Hill & Cookson 1996], they have described an exon 7 coding polymorphism in the Busselton population, Gly237Glu (E237G, rs569108). An ASO-PCR assay was developed, and demonstrated this mutation to be present in 53 of 1004 subjects. These individuals exhibited larger wheals to mixed grass allergens and house dust mite, more asthma, as well as increased BHR to methacholine. Serum IgE levels were only slightly higher (68.0 v 46.4 IU/ml). There was no parent of origin effect for this polymorphism (17 paternal, 13 maternal).

Shirakawa et al [1996a, 1996b] have reported a case-control study of FCER1B, using 500 atopic patients (allergic - early and late onset - and nonallergic asthma, hayfever or eczema) ascertained via Osaka hospital clinics were compared to 100 controls (disease status presumably unknown) attending a "health examination company". Initially, they found no Leu181 mutations either by ASO-PCR in the entire sample, or via sequencing in 10 atopic subjects. Therefore they developed RFLPs in intron 2 of FCER1B (RsaI), CD20 and GIF. Significant differences in genotype frequencies were detected for the FCER1B polymorphism between the controls (AA: 89; AB: 10; BB: 1) and several different subgroups of the patients, most strongly for childhood onset allergic asthma (AA: 50; AB: 42; BB: 8), and not at all for the intrinsic asthma group (AA: 86; AB: 11; BB: 3). There were no such effects for CD20 or GIF.

When they tested for the Gly237Glu mutation [Shirakawa et al 1996b], this was found in 6/100 controls, but 16/100 adult atopic asthmatics (P=0.025) and 20/100 child asthmatics (P=0.005). It was not significantly increased in the nonatopic asthmatics (8/100), but was markedly increased in the subgroup with a total sIgE greater than 1000 IU/ml.

Weaker evidence for association in the region was described by Holgate et al [1996]. As part of the Southampton study described above, allelic effects of two markers were detected on different phenotypes: allele 168 of D11S527 with BHR (P=0.0003, Bonferroni corrected for 13 alleles P=0.004), and allele 235 of D11S534 with total IgE (P=0.007, Bonferroni corrected for 14 alleles P=0.09).

4.2 Cox et al on eczema [1998]

This paper from the Oxford group [1998] describes association studies of FCER1B with eczema. There were two subsamples, the first of 60 nuclear families (277 subjects) recruited through a single child attending a hospital clinic with atopic dermatitis, the second a similar panel (first described in 1969) of families selected to have two or more first degree relatives with "active atopic dermatitis". The latter group was that used by Coleman et al [1993]. The exact size of this group was not stated but based on the TDT tables contained at least 21 families.

There was strong evidence for association with the maternally transmitted allele for two Rsa1 polymorphisms in the 2nd intron and the 7th exon. Interestingly, the E237G allele was not associated with eczema. This probably reflects low power, as it was present in only 6% of probands, but argues against it being a functional mutation.

Table 4. TDTs for FCER1B polymorphisms in atopic dermatitis families.
PaternalMaternal
Panel A
TrNTTrNT
In 21710316
Ex 7916174
Panel B
In 224171023
Ex 718212411

4.3 Ulbrecht et al [1997]

These authors [1997] tested for allelic association in a case-control design comparing 129 unrelated individuals with a total sIgE level of 200 IU/ml or more, to 266 controls with total sIgE below 200 IU/ml. The subjects were genotyped at the FCER1B-ca microsatellite. They detected no significant differences in allele frequencies.

4.4 Haider & Hijazi [1998]

This report [1998] describes allele frequencies for the Intron 2 RsaI and Exon 7 UTR RsaI polymorphisms in FCERB1 among 78 asthmatics and 122 nonasthmatic members of 52 Kuwaiti Arab families. The overall allele frequencies were quite different from those published for Caucasian or Aboriginal Australians. There was complete linkage disequilibrium between the 2nd intron and 7th exon polymorphisms, and no difference in allele/haplotype frequency between the asthmatics and controls. A subgroup analysis of hayfever, eczema and skin-prick test positivity (SPT+) did find a trend for SPT+ asthmatics to carry the int2 B allele more often than did SPT- asthmatics (56% v 33%, P=0.01). This effect was much weaker in nonasthmatics (Breslow-Day test comparing asthmatics' with nonasthmatics' odds ratio X2(1 df)=3.41, P=0.06).

Table 5. Frequency of the RsaI int2 FCERB1 polymorphism among Kuwaiti Arab asthmatics and controls, from Haider and Hizawi [1998].
GroupNRsaI int2 B frequency
AsthmaticsSPT+540.56
SPT-240.33
NonasthmaticsSPT+300.53
SPT-920.51

4.5 Rohrbach et al [1998]

This paper describes an SSCP and exon amplification based screen for variants in FCER1B among 224 atopic asthmatic children and 227 related and unrelated controls. The I181L mutation was not detected, but 3.7% of cases and 2.6% of controls carried E237G. The authors also detected nine previously unreported variants.

4.6 Dickson et al [1999]

These authors sequenced FCER1B (all exons, some introns and part of the 5' UTR) in 71 subjects from the Australian sib pair study [van Herwerden et al 1995] described earlier. No I181L, V183L and E237G alleles were detected, and two novel noncoding polymorphisms were not associated to asthma or BHR.

4.7 Soriano et al [2000]

This case control study compared 146 asthmatics presenting to hospital acutely to 50 controls. Half the cases had presented during the soybean epidemic (see above). Several candidate genes were tested for association. No participants were found to carry an I181L allele, and the E237G allele frequency was not increased in cases (7.6% v. 8.5%, P=0.88).

4.8 Hizawa et al [2000]

This is another Japanese case-control study, that includes 226 asthmatics (54 nonatopic) and 226 healthy controls. None of 24 cases undergoing sequencing harboured a I181L, V183L or E237G variant allele, but a -109C>T SNP in the promotor region had a frequency of 65% in cases and 69% in controls (P=0.10). There was Hardy-Weinberg disequilibrium in the cases (a heterozygote excess, P=0.003), but not the controls. Paradoxically, the T/T genotype was associated with a higher total sIgE level within the asthmatics (P=0.0015).

The authors concluded that this polymorphism does not contribute directly to risk of asthma, but was a modulator of total sIgE level in individuals with atopy.

Table 6. Frequency of the C/T -109 genotype among Japanese asthmatics and controls, from Hizawa et al [2000].
Asthmatics (N=226)Nonasthmatics N=226
T/TC/TC/CT/TC/TC/C
Genotype frequency85123181089919
Expected under HWE95.0103.128.0109.895.520.8
total sIgE (IU/ml)427234275545456

4.9 Cardaba et al [2000]

This study was of 22 "olive-pollen allergic" families (N=88). The TDT was applied to several favourite genes: HLA DRB1, DQB1, TCR-Va 8.1, LTA and FCER1B. The FCER1B Exon 7 UTR RsaI *1 allele was associated with elevated total sIgE level (P=0.01) as well as olive pollen specific IgE.

4.10 Laprise et al [2000]

This describes a French-Canadian case-control study of asthma and E237G, as well as a family based study (see above). There were 100 cases (total sIgE>280 ug/l plus 3 or more positive SPTs to a panel of 6 allergens, 25 asthmatics) and 100 controls (nonatopic). No L181 alleles were found in any subjects. The G237 allele was much more common among the cases (Fisher exact P=0.00001) than the controls.

Table 7. Frequency of the FCER1B-E237G genotypes among French-Canadian atopic subjects and controls, from Laprise et al [2000].
Group E/EE/GG/GG237 FrequencyHWE P-val
Controls98 1 1 1.5% 0.01
Atopic 80 20 0 10.0% 0.14

In the four three-generation atopy pedigrees (all ascertained through a highly atopic single proband), there was significant support for association between the G237 allele and atopy via various transmission- disequilibrium tests. The strongest individual result was for sensitisation to indoor allergens: G237 allele frequency was 36% in affecteds and 6% in unaffecteds (unified transmission-disequilibrium test P=0.002). Paternal transmission was as skewed as maternal transmission for indoor allergen sensitisation: 20/26 transmissions of G237 overall, 9/11 paternal, 10/13 maternal.

4.11 Thomas et al [2000]

This is an update on the Southampton group's study. There are three panels of families: the "random" sample of 131 nuclear families (685 individuals) unselected for asthma; 60 "multiplex" asthma families (354 individuals) containing two or more affected members; 49 "simplex" families ascertained via a single affected child.

Results for 6 markers around FCER1B, including the coding polymorphism FCER1B*E237G, were presented. The best single-point linkage was between wheezing and FCER1B*E237G (lod=1.52), and the best multipoint result (lod=1.4) with a quantitative asthma score was close to D11S480. There was no association with any marker.

4.12 Nagata et al [2001]

These authors chose allergic rhinitis as the phenotype to test for association with FCER1B E237G. A total of 233 cases (outpatients mainly from Chiba University Hospital, rhinitis plus positive RAST score, mean total sIgE of 641 IU/ml) and 100 controls (chronic sinusitis, hypertrophic rhinitis, parosmia etc, with negative RAST screen, mean total sIgE 56 IU/ml).

The G237 allele was more common among the cases (P=0.015), and most strikingly (30%) among the 45 cases with multiple sensitization. There was Hardy-Weinberg disequilibrium in the controls (a dearth of heterozygotes).

Table 8. Frequency of the FCER1B-E237G genotypes among Japanese allergic rhinitics and controls, from Nagata et al [2001].
Group E/EE/GG/GG237 FrequencyHWE P-val
Controls77 18 5 14.0% 0.012
Rhinitis15076 7 19.3% 0.468

4.13 Hakonarson et al [2001]

Among 94 Icelandic atopic asthmatics and 94 unrelated controls the L181 and L183 alleles were not detected at all. The G237 allele was uncommon, and not increased in cases (Table 9).

Table 9. Frequency of the FCER1B-E237G genotypes among Icelandic allergic asthmatics and controls, from Hakonarson et al [2001].
Group E/EE/GG/GG237 FrequencyHWE P-val
Controls 93 1 0 1.1% 0.942
Asthmatics 92 2 0 1.7% 0.883

4.14 Furumoto et al [2000]

These authors describe functional studies of I181L, V183L and E237G and wild type Fc-epsilon beta chains in mouse-derived mast cells. There were no detected differences between all four variant chains in cytokine and LTC4 production, intracellular calcium mobilization and beta-hexosaminidase release after activation of the receptor. This is another line of evidence suggesting the functional variant may be in LD with E237G.

4.15 Hage-Hamsten et al [2002]

This is a survey of 461 Swedish farmers performed by a collaboration including Bill Cookson and Miriam Moffatt. There were 83 asthmatics in the sample, and 147 atopics. The 237G allele was present in only 7 subjects (0.8% allele frequency). The RsaI-in2 and RsaI-ex7 "B" allele frequencies were 62% and 39% respectively. There was no association of these polymorphisms with asthma, total sIgE level or atopy. In the case of specific IgE, there was some evidence of association (P=0.005-0.03) of RsaI-ex7 to mites, notably L. destructor, T. putrescentiae, and mixed pollens. The numbers of subjects included in these analyses are smaller (eg "all" mites uses 73 subjects).

4.16 Leung et al [2002]

These authors compared allele frequencies in 76 Hong-Kong Chinese asthmatic children and 70 controls at six candidate SNPs. The allele frequencies for the Intron 2 RsaI (18% and 21%) and Exon 7 UTR RsaI (4% and 4%) polymorphisms in FCERB1 did not differ between the two groups.

4.17 Asthma and uteroglobin (CC16, CC10)

Human uteroglobin (gene symbol UGB, 11q13) is a possible candidate for the 11q13 atopy gene, as it has cytokine-like anti-inflammatory activities, and is expressed in almost all epithelial cells, most notably in this context by Clara cells (the nonciliated secretory cells present in respiratory epithelia), where it is known as the 16 kD (or 10 kD, presumably because it is a homodimer of two 8 kD subunits) Clara cell protein.

4.18 Kim et al [2007]

In a cohort study of rural Korean adolescents (N=2055), the E237G polymorphism was found to predict BHR to methacholine (G/G genotype RR=2.43, logistic regression P=0.01; crude association P=0.051), but not atopy (sIgE or SPT). The prevalence of BHR was 26% in the sample

Table 10. Frequency of the FCER1B-E237G genotypes among Korean adolescents versus BHR to methacholine, from Park et al [2007].
Group E/E E/GG/GG237 FrequencyHWE P-val
BHR- 111627416 10.9% 0.858
BHR+ 372 10313 13.2% 0.094

3.5.1 3.5.1 Zhang et al [1997]

Zhang et al [1997] used SSCP analysis to screen 26 sib pairs affected with asthma and their parents through the 5' promotor region and the three exons of the gene. This failed to detect any variation, although a similar screen of probands with Best's disease did detect a common C to T substitution at position +33.

3.5.2 3.5.2 Laing et al [1998]

Laing et al [1998] reported a mutation screening and case-control study of 67 children with asthma and 46 controls. Cases either attended a respiratory outpatient clinic or were sampled from a cohort of 76 families (266 subjects) unselected for asthma. Nonasthmatic controls came solely from the latter source, had no personal or family history of asthma, atopy, and no BHR on histamine challenge.

The authors detected a common A to G substitution at +38 in the 5' UTR of exon 1. A Sau96I RFLP was then used to type the entire sample. The A>G variant was present at a frequency of 67% in the family cohort (N=266), and was in HWE (P=0.5). Among asthma cases, the allele frequency was 57.7%, and was 73.9% among the 46 unaffected controls (P=0.008). If an additional 7 children from the cohort study were included as cases, using a definition of past doctor diagnosis of asthma plus BHR,

3.5.3 3.5.3 Mao et al [1998] and Gao et al [1998]

Mao et al [1998] and Gao et al [1998] describe attempted replications using the Osaka case-control sample and a British sample using both the Sau96I RFLP and a first intron STRP. There were no significant differences in allele frequency between any case groups and the controls (see Table; tests for HWE all nonsignificant). The frequency of the A>G polymorphism is quite similar across all three studies.

Table 11. Frequency of the +38A>G variant among British and Japanese asthmatics and controls, from Gao et al [1998].
GroupNA>G frequency
JapaneseControls1000.610
Allergic asthma1000.655
Nonallergic asthma1000.610
British Controls1500.620
Atopic asthmatics1250.676

3.5.4 3.5.4 Choi et al [2000]

This abstract (also from Mukherjee's group) describes a study of 22 asthma families. The +38A>G variant was more common in asthmatics. A similar claim was made for a novel tetranucleotide SSR (-3100).

4.19 Other association studies of 11q13

Weaker evidence for association in the region was described by Holgate et al [1996]. As part of the Southampton study described above, allelic effects of two markers were detected on different phenotypes: allele 168 of D11S527 with BHR (P=0.0003, Bonferroni corrected for 13 alleles P=0.004), and allele 235 of D11S534 with total IgE (P=0.007, Bonferroni corrected for 14 alleles P=0.09).

Martinati et al [1996] used the published AS-PCR assay to look for the Leu181 mutation in all 213 subjects. As in Shirakawa et al [1996b], no mutations were detected. Similar findings were reported by Hill [1996], a sample of 65 asthmatics, Kofler et al [1996] in 40 atopic subjects (30 asthmatics), and by the present author [Duffy et al 1996], where no Leu181 mutations were present in the total 939 subjects tested (MZ and DZ twins, available parents of DZ twins and unrelated controls). In the latter three studies, the AS-PCR results were confirmed by sequencing in some or all of the atopic subjects.

4.20 An association "scan" of 11q13

Because of these replication difficulties, Adra et al [1998] "stepped back" and looked at allelic association at 9 candidate loci across 12 cM of 11q13 using 125 atopic asthmatics and 150 controls taken from a British sample. Three variants within FCER1B and one with HTm4 (the CD20L (CD20 like) gene, a hematopoeitic cell specific surface marker cloned by Adra in 1994) were associated with asthma, positive HDM RAST, and atopy. There was strong disequilibrium between FCER1B and HTm4 polymorphisms.

4.21 Conclusions

We can conclude that Leu181 and Gly237 may be functional mutations in FCER1B that lead to atopy. It seems just as likely that they are in disequilibrium with the trait polymorphism, although the FcERI beta-subunit remains the most likely site for this. The gene frequencies of only 3% and 5% respectively in the general population sample from Busselton, and even in the enriched sample of families from Oxford, only 10% carried Leu181. The maternal effect seems to hold up, though the data still does not clearly point to the mechanism for this.

Table 12. Studies of linkage of atopic disease to chromosome region 11q. If multiple publications including the same families have been published, only the most recent results are included.
StudyNo. subjects (families)lod Score and comment
Cookson et al [1989](7)6.4 to D11S97 (10 cM)
Young et al [1992]281 (64)5.2 to D11S97 (M 18 cM, F 0.1 cM)
Inacio et al [1991]83 (17)No linkage to D11S97
Shirakawa et al [1991,1994](4)4.88 to D11S97 Selected pedigrees: used 4 of 136.
Amelung et al [1992]117 (20)< -2.0 within 12 cM of FGF3 Families linked to chr 5.
Hizawa et al [1992]60 (4)< -2.0 within 4 cM of D11S97
Lympany et al [1992]89 (9)< -2.0 within 4 cM of FGF3
Rich et al [1992]67 (3)< -2.0 within 5 cM of D11S97
Coleman et al [1992]407 (95)-7.88 multipoint D11S97, PGYM, CD20 12 families with affected mothers and unaffected fathers gave lod 0.8
Collee et al [1993] 52 (26)increased ASP sharing D11S97
Brereton et al [1994](12)negative lod
van Herwerden et al [1995]246(123)increased ASP sharing FCER1B-ca
Duffy et al [1994]424 (212)no increase ASP sharing FCER1B-ca Unable to detect Leu181 mutation
Watson et al [1995]560 (131)negative lods for 3 markers Combined segregation-linkage
Martinati et al [1996]213 (45)no increase ASP sharing FCER1B-ca Unable to detect Leu181 mutation
Noguchi et al [1997]306 (68)weak H-E result for sIgE and FCER1B-ca Negative ASP analysis
Neely et al [1996]218 (12)increased ASP sharing (and positive TDT) INT2, D11S1369
Trabetti et al [1998]659 (168)increased ASP sharing FCER1B i2 RsaI poly
Folster-Holst et al [1998]98 (12)3.55 to FCER1B (0 cM) under two locus model
Palmer et al [1998]576 (121)H-E linkage to total sIgE
Laprise et al [2001]4 (106)2.3 for atopy to E237G


5 Linkage studies of the chromosome 5 atopy gene

There are numerous candidate genes in and around the lymphokine cluster on 5q31. This has made it hard to definitely identify the particular genes involved in atopy and asthma linked to the region.

Table 13. Sequence map positions of candidate genes and allergy-linked markers on chromosome 5q.
Gene or Marker Location (Mbp) Remarks
SLC22A4 131.658-131.708 RA and Crohn's
IL5 131.905-131.907 Especially eosinophilia
IL13 132.022-131.025 IL13 KO mice hypoallergic
IL4 132.037-132.046 Gene methylation important
D5S393 135.729
IL9 135.256-135.259 T-cell growth factor
CD14 139.993 Binds LPS; phagocytosis
NR3C1 142.763 Glucocorticoid receptor
D5S436 145.184
SCGB3A2 (UGRP1) 147.238 Uteroglobin related protein
SPINK5 147.424
ADRB2 152.638 Beta-adrenergic receptor
D5S487 155.601
HAVCR1 156.419 Hepatitis-A receptor
CYFIP2 156.629 Cytoplasmic FMR1 interacting protein 2
IL12B 158.674 IL-12 and IL-23 p40 beta-subunit
D5S422 162.086
LTC4S 187.999

5.1 Marsh et al [1994]

In the same 11 Amish families discussed above, sib-pair linkage analysis of total sIgE, mite- and multiallergen-specific (chemiluminescent immunoassay) sIgE was performed using markers from the 5q (5q31.1-33) lymphokine gene cluster. Significant evidence of linkage between total, but not specific, sIgE and polymorphisms in the IL-4, IL-9, and Interferon Regulatory Factor-1 genes were detected in SIBPAL analyses. However, IL-3, IL-5, CD14, CSF-1 receptor, and a lymphocyte-specific glucocorticoid receptor are also close neighbours, and as such, good candidates for disease genes. If the analysis was restricted to the 128 pairs without any detectable allergen-specific IgE (from a total of 349), probability of linkage increased, while age and sex adjustment made minimal difference.

5.2 Meyers et al [1994]

The study of asthma families from Groningen has also confirmed the presence of linkage to the 5q region of total IgE level as well as nonspecific bronchial hyperresponsiveness [1994]. There were 92 families containing 538 individuals studied, all ascertained through a hospital diagnosed asthmatic parent (in the case of nuclear families) originally studied 1962-70. Histamine challenge testing, skin testing, and measurement of total, mite- and grass-specific serum IgE levels were performed on all family members.

A preliminary Class D regressive segregation analysis of logIgE level was consistent with the presence of a common (increasing allele frequency 59%) autosomal recessive gene with a residual sib-sib correlation for IgE levels of 0.27 (general model log likelihood, -520.35; polygenic with one distribution, -534.6; environmental with three distributions, -530.8; codominant, -523.3; recessive with mean(BB,AB)=41.7 and mean(AA)=436.5, -523.8). An analysis with PAP reached similar parameter estimates P=60%, mean(BB,AB)=22.9 and mean(AA)=239.9.

A sib-pair analysis (SIBPAL) of logIgE level confirmed the presence of linkage between the 5q interleukin cluster region and log IgE level (see Table). This closely paralleled findings for BHR treated as a discrete trait (PC20>32 mg/ml histamine versus PC20<32 mg/ml - see Table). When BHR was modelled as a continuous trait with log sIgE as a covariate, the evidence for linkage persisted (P<0.002). I would interpret this last finding as a comment on the repeatability/validity of each measure in isolation as a measure of atopy. A parametric linkage analysis in PAP under the preferred model gave a lod score of 3.56 at r=0.10.

Table 14. Sib-pair analysis for log-transformed serum IgE level versus 5q region markers.
Markerdf (pairs)P
IL91310.07
D5S3932480.01
D5S4362820.0003
CSF-1R2260.03
D5S410239---
D5S412252---
D5S415230---

Table 15. Sib-pair analysis of bronchial hyperresponsiveness to histamine versus 5q region markers.
MarkerdfP
IL9160.14
D5S393220.04
D5S500350.09
D5S658350.03
D5S436350.009
FGF1250.02
D5S434340.08
D5S470350.10
CSF-1R200.05
D5S410200.31
D5S412330.05
D5S415221.00

5.3 Blumenthal et al [1996]

This group [1996] reported an absence of linkage of chromosome 5q markers to sIgE level in their four large atopic families (110 typed individuals). Maximum likelihood linkage analysis (under a common dominant gene model), and Haseman-Elston analysis using SIBPAL, were performed.

The maximum lod score among 12 markers spanning chromosome 5 was 0.06 (r=0.23), with LOD scores of -2.3 at r=0 for IL-9, and less damningly, -0.4 at r=0 for IL4-R1. Similarly, the best P-value from SIBPAL was 0.29.

5.4 Laitinen et al [1997]

This study [1997] found neither association and linkage to 5q in asthma families from Finland. Families containing at least one self-reported asthmatic were recruited by advertising within one province (Kainuu). There were 51 multiplex families and 105 uniplex families (used for TDT) where probands met the study diagnostic criteria for asthma. These required agreement between two study respiratory physicians based on medical records, and bronchial challenge. The authors found 87% of volunteers met these criteria. A total of 487 subjects underwent total and specific (eight allergens) sIgE determination and genotyping at 17 markers spanning D5S404, IL4, IL9, to D5S413.

Mapmaker-Sibs analysis of sIgE (73 pairs) did not find strong evidence for linkage to the region (t-values running from 0.9 to 1.6), and Genehunter NPL scores for asthma were consistently negative. No specific haplotypes were increased among subjects with sIgE>100 kU/ml. An IL9 polymorphism (T113M) was found in 38/253 atopics (high sIgE), 43/323 asthmatics, and 80/542 of the entire sample. The authors also present a refined (radiation hybrid based) physical map of the region.

5.5 Noguchi et al [1997]

As noted above, these authors [1997] report evidence for linkage between atopic asthma, atopy and sIgE level and several 5q31-33 markers in a relatively small sib-pair based study. The Haseman-Elston analysis (single-point, SIBPAL) of sIgE gave a best t=-2.83 (df=91, P=0.003) for IL4 (IL9 t=-1.89, df=79, P=0.03; D5S393 t=-2.28, df=89, P=0.01). The affected sib-pair (and discordant sib-pair) results for asthma and atopy were similar (Table).

Table 16. Mean ibd sharing in sib-pairs from Noguchi et al [1997]. There were 60-73 ASPs (27-46 DSPs) for atopy, and 32-42 ASPs (44-68 DSPs) for asthma.
MarkerMean ASP ibdP-valueMean DSP ibd
Atopy
IL40.580.0060.41
IL90.550.010.43
D5S3930.61<0.000010.52
D5S4360.510.420.41
Asthma
IL40.620.00130.49
IL90.600.0180.48
D5S3930.600.0070.57
D5S4360.491.000.51

This study was extended in Yokouki et al [2000]. This used the subset of families also described in Kimura et al [1999] (see below). These were 47 families containing two siblings with mite-sensitive asthma (65 affected sib pairs), typed at 398 markers. The best genome wide affected sib pair multipoint MLS (using Mapmaker-Sibs) was 4.52, peaking over D5S820, 20 cM telomeric of IL4, and between ADRB2 and IL12B. Further fine mapping increased the MLS to 5.3 between D5S487 and D5S422 [Noguchi et al 2005].

5.6 Kamitani et al [1997]

The group from Melbourne [1997] also looked at D5S399 sharing among the 119 sib pairs. The biggest subset of affected sib pairs was 105 pairs concordant for atopy. There was no increase in IBS sharing for asthma, atopy, or BHR.

5.7 Palmer et al [1998]

As noted above, this sample of largely unselected sib-pairs (up to 226 pairs) provided evidence of linkage between chromosome 5q markers (D5S393, D5S399) and total sIgE levels (adjusted for specific IgE level, following Marsh et al 1995, P=0.004), specific sIgE level, and eosinophil counts.

5.8 Martinez et al [1998] on linkage between eosinophilia and 5q

This study [1998] obtained surprisingly strong results from an affected sib pair study of peripheral blood eosinophil count. Extremely concordant and discordant sib pairs were defined using "low" (eosinophil count of 2% or less) and "high" (5% or more) thresholds, selected from a total of 123 nuclear families from within the Tucson Children's Respiratory Study. Only the 35 concordant low-low pairs exhibited linkage to chromosome 5q markers between IL9 and IL13.

Extending this sample to a total of 53 low-low pairs, a multipoint analysis obtained a lod of 2.4, with a peak around D5S658. A single-point Haseman-Elston analysis of eosinophil count (using 70-82 pairs) obtained a peak slope of -0.41 (P=0.002) at D5S500.

5.9 Dizier et al [1999] on association and linkage of total sIgE to IL4

The full sample of Busselton families were analysed for linkage and association to the IL4 variants described by Walley & Cookson [1996]. The phenotypes chosen were those of Dizier et al [1995]: age-sex-generation adjusted log total sIgE, with and without adjustment for specific sensitization (defined as any positive RAST).

Haseman-Elston analyses were weakly suggestive for linkage to the RAST adjusted total sIgE level (IL4, P=0.01; -590C>T IL4 promotor variant, P=0.02). A lod score analysis (combined segregation-linkage using REGRESS) under the recessive model gave similar results (lods=1.17, 1.25). Adding in an association model found no evidence for allelic association between either the IL4 STR or -590C>T.

5.10 Shek et al [2001]

This is a Singaporean Chinese affected sib pair study of asthma and atopy linkage to 5q. There were 88 families containing 125 asthma ASPs. The multipoint NPL analysis gave a peak lod (equivalent) of 1.7, but a peak singlepoint ASP lods of 5.3 (using GAS with 89 informative pairs) was obtained for D5S412 (close to IL12B), and 5.1 for D5S2110 (96 pairs, close to IL9). Singlepoint (old) Haseman-Elston analysis obtained a best linkage of D5S2001 to total sIgE (P=0.01).


6 Association studies of the chromosome 5 atopy gene

Given the number of good candidate genes in this region, there have been a large number of pure association studies.

Table 17. Frequency of the -590C>T IL4 promotor variant (rs2243250) in asthmatics and controls from several studies.
StudySubjectsNFrequency
Whalley & Cookson [1996] Busselton Asthma 2110.30
Busselton Control7930.26
Oxfordshire Asthma 1240.33
Oxfordshire Control 590.27
Basehore et al [2004] US CaucasianAsthma 2330.19
US CaucasianControl2450.14
US Hispanic Asthma 1160.39
US Hispanic Control2300.35
African-Am Asthma 1680.64
African-Am Control2690.67
Nagarkatti et al [2004] North IndianAsthma 1710.00
North IndianControl1280.00
Korzycka-Zaborowska et al [2004]Polish Atopy 980.01
Polish Control 870.02

6.1 Borish, Rosenwasser and coworkers [1994]

This abstract [1994a] describes polymorphisms (SSCPs) in the IL-3, IL-4 and IL-9 genes [Borish et al 1994b] on chromosome 5q. Rosenwasser et al [1995] enlarge on this. In 20 families ascertained through an asthmatic, a -590C>T IL4 promotor polymorphism (rs2243250) was associated with high total sIgE level. In an enlarged sample [Borish et al 1996; Rosenwasser & Borish 1997], association was also found between a -571C>A variant in the IL10 promotor, PBMC production of IL-10 and high total sIgE level.

6.2 Walley & Cookson [1996]

This is another analysis of the two Oxford panels (Busselton and UK), this time seeking to replicate an association between the -590C>T IL4 promotor polymorphism and atopy. There was weak evidence of association found in the Busselton sample only to wheeze and presence of anti-HDM specific IgE (P=0.03 and P=0.01, uncorrected for relatedness of subjects or for multiple comparisons).

6.3 Ulbrecht et al [1997]

This case-control study was mentioned earlier. It also failed to find any evidence of association to D5S436, D5S393, D5S210, IL4, and IL9.

6.4 Mansur et al [1998] on associations between atopy and 5q

Britton's group [1998] genotyped 89 asthmatics (symptoms plus positive methacholine challenge) and 92 controls at 11 markers spanning 5q31. This sample came from the 2633 Nottingham residents they had previously surveyed [Britton et al 1994]. There was association to BHR and sIgE for D5S404, IRF1, and D5S210 with uncorrected P values of 0.02 to 0.04 (22 tests).

6.5 Hijazi & Haider [2000] on association of asthma to IL4

This paper describes a case-control study of the -590C>T polymorphism in 84 Kuwaiti asthmatics, 53 unaffected first-degree relatives, and 47 unrelated unaffected controls. The C>T allele frequency was 79% in cases, 75% in the unaffected relatives and 76% in the other controls.

6.6 Yang et al [2004] on IL4 -590C>T

In a survey of 1333 Taiwanese undergoing health screening, there was no association between questionnaire diagnosed atopic disease or specific or total sIgE level and -590C>T.

6.7 Basehore et al [2004] on IL4

Basehore and coworkers [2004] looked at 11 SNPs across the IL4 gene, obtaining a best individual SNP P-value of 0.0003 within their US white group for IgE level, but only P=0.01 for asthma. The commonest haplotype (74% in whites) was associated with low sIgE level (P=0.00008) in Caucasians, but not in the other two groups. There were large differences in SNP and haplotype frequency across the three ethnic groups in this study.

6.8 Wellcome Trust Consortium [2007] on IL4

The 1958 UK birth cohort contributed to the control panel for the Wellcome Trust Case Control Consortium, and association results for phenotypes such as serum IgE level and FEV1 are available on the web at http://www.b58cgene.sgul.ac.uk/.

Although well short of genome-wide significance, there are effects of several SNPs on log serum IgE level.
Association between log sIgE level and IL4 SNPs for the 1958 UK Birth Cohort
PositionSNPNBeta trendSE P-value
132036543rs224324814360.04650.04790.331804
132036543rs224324813790.04060.04900.407382
132037053rs224325016080.05140.03240.112387
132040624rs2227284 4460.10350.04950.036872
132040624rs222728414360.08530.02680.001467
132041078rs222728214360.08530.02680.001467
132042008rs224327014360.09960.03180.001758
132045843rs224328813680.07730.03350.021338
132046068rs224329014360.08930.03310.007141
132046068rs224329013810.07710.03400.023790

6.9 IL13 variants and asthma: Heinzmann et al [2000]

IL-13 has been implicated in asthma by several studies and variants in CD124, a subunit of both the IL-4 and IL-13 receptors, may be associated with asthma. The British/Japanese collaboration (that includes Taro Shirakawa and Julian Hopkin) therefore screened IL13 for variants, detecting a single functional polymorphism R130Q in exon 4. They then performed case-control studies in their usual UK and Japanese panels (see Table). The Q allele was significantly more common aong asthmatics than controls in both populations (for Japanese atopic asthma OR=1.85, P=0.03; for the British, OR=1.83, P=0.01). In the cohort of 290 Japanese schoolchildren previously examined for a relationship between tuberculin reactivity and atopy [Shirakawa et al 1997], there was a similar association, and that serum IL-13 levels were higher in R130Q carriers (data not shown).

Via molecular modelling, the authors went on to show that the Gln at position 130 should alter receptor binding. In a subsequent paper [Arima et al 2002], it was shown that the Q130 variant does not affect binding to the high affinity IL-13 receptor in vitro, but slightly lowers binding to the lower affinity IL-13R-alpha-2 molecule, and seems to have a longer plasma half-life.

Table 18. Frequency of the R130Q variant among British and Japanese asthmatics and controls, from Heinzmann et al [2000].
GroupNR130Q frequency
JapaneseControls1000.430
Allergic asthma1000.630
Nonallergic asthma1000.620
British Controls1500.267
Atopic asthmatics1500.400

6.10 IL13 variants and atopy: Liu et al [2000]

Mutation screening of the German MAS-90 cohort also highlighted variation in the fourth exon of IL13, this time R130E. The E allele was present in 21% of 604 children, and was found to be increased in subjects with a high total sIgE level (>85th age-sex percentile, P=0.003, see Table). A weaker effect was seen for atopic dermatitis.

Table 19. IL13 R130Q variant and high IgE level [Liu et al 2000].
IgE PercentileGenotype countsE allele frequency
Q/QQ/RR/R
>85%642 520.27
<15%130 800.14

6.11 IL13 variants and asthma: Hoffjan al [2002,2004]

This paper describes results from the Childhood Origins of Asthma (COAST) study. A total of 207 intensively phenotyped infants were genotyped at 61 SNPs in candidate genes including FCER1B, NOS2A, IL5, IL10, IL13 and IFNG. The strongest results were for the FCER1B*E237G and a NOS2A*D346D variant, associated with low IL-13 levels in cord blood following PHA stimulation (P=0.0025, 0.0062). The TGFB1*-509T allele was a risk factor for respiratory syncytial virus-related wheezing in the first year (P=0.0005).

6.12 IL13 variants and asthma: DeMeo et al [2002a]

The transmission-disequilibrium test (FBAT) found some evidence of association in 640 asthma families in the Childhood Asthma Management Program (CAMP). The R130Q did not increase in frequency with AHR, asthma diagnosis or severity, but was correlated with overall allergy (combining eosinophil count, total sIgE and number of positive SPTs, P=0.007), as well as with eosinophil count alone (P=0.01).

6.13 IL13 variants and COPD: DeMeo et al [2002b]

These authors found no association or linkage of the R130Q variant to IgE level or lung function in families ascertained through 72 COPD patients.

6.14 IL13 variants and atopy: He et al [2003]

This study also found weak evidence of association using a family-based design. A sample of 368 The R130Q allele increased risk of atopy and atopic dermatitis, most strongly in white children (best P=0.014).

6.15 IL13 variants and atopy: Maier et al [2006]

Up to 3615 participants in the UK 1958 Birth Cohort sample were genotyped at a set of candidate genes including IL13. All four IL13 SNPs tested were associated with serum tIgE level (best P=0.00003), with the increasing homozygotes at the promoter SNP rs1881457 having a geometric mean tIgE of 46.8, and the decreasing homozygotes, 26.9. The R110Q (rs20541) gave essentially the same point estimates (with slightly smaller P=0.0003), but dropped out of multiple regression analysis if rs1881457 was included, so the authors concluded that the coding variant was not causative. The same SNPs were found to be not associated with Type 1 diabetes.

6.16 IL13 variants and atopy: Bottema et al [2008]

In a combined analysis of three European birth cohorts (N=3062 children), IL13 and CD14 (see below) SNPs were tested for association to total and specific IgE level. The best association (P<1e-4) was for rs1295685; evidence for R110Q (rs20541) was weaker. There was no intersample heterogeneity.

6.17 CD14 and atopy

CD14 is a 55 kD glycoprotein expressed on the surface of predominantly monocytes, and binds bacterial lipopolysaccharide (LPS, thus explaining its involvement in toxic shock syndrome) and peptidoglycan. It is central to phagocytosis of apoptotic cells. It is located within the 5q31.1 cluster. Three posters at the 1999 ASHG meeting described evidence for association between CD14 and atopy. Reijmerink et al genotyped 159 asthmatics and 159 spouse controls for a -159C>T promotor variant (probably rs778594). Although cases and controls did not differ in allele frequencies, the C/C homozygotes tended to have higher total IgE levels and more positive skin test results. Among the Hutterites [Schneider et al 1999], there was an association with (any) positive skin test (P=0.0009). In Iceland, Hakonarson et al [1999] found no association with moderate to severe atopic asthma (C frequency of 94 cases, 58.5%; 94 controls, 54.5%).

Robertson et al [2002] failed to find any evidence of association to asthma in the Williams and McNicholl 1964 Melbourne asthma cohort. For -159C/T, the genotype frequencies were C/C 0.262, C/T 0.488 and T/T 0.250 (N=257). In another TSANZ poster, Savarimuthu et al [2003] reported genotypes for 463 asthmatics: proportions were C/C 0.28, C/T 0.49 and T/T 0.23.

Heinzmann et al [2003] carried out a case-control study (182:270) of German child asthmatics. The -159C>T and ­1359G>T SNPs were not associated with asthma (P=0.876), with genotype frequencies for -159 of C/C 0.293, C/T 0.481 and T/T 0.226 (N=443).

Rupp et al [2004] link the CD14-159 polymorphism with susceptibility to C. pneumoniae infection (detectable in circulating monocytes in 13% of their coronary artery disease patients). Stimulation by C. pneumoniae increases CD14 expression more in T/T homozygotes [Eng et al 2004].

John et al [2004] examined CD14-159 genotype in 442 children from a birth cohort. Overall, allergic sensitisation was not associated with CD14-159 genotype (P=0.4), but there was a strong interaction with endotoxin level in house dust samples from the childrens' homes: endotoxin exposure reduced sensitisation in C/C carriers (OR=0.70, 95%CI=0.55-0.89).

Choudry et al [2005] reported results from the GALA study. CD14-159 genotype was found to be associated with total sIgE level in Puerto Rican and Mexican asthmatics, but only in those exposed to cigarette smoke (best P-value 0.00008 in Puerto-Ricans). Low baseline FEV1 in smoke-exposed asthmatics was associated with CD14+1437 in family-based association analyses.

Yanamandra et al [2004], at the same meeting, found no association in African-American asthmatics compared to controls, with genotype frequencies for -159 of C/C 0.494, C/T 0.425 and T/T 0.081 (N=160 controls). These genotype frequencies are quite different to those reported for samples of European descent.

Gern et al [2004] describe results from the COAST study "at risk" birth cohort. There were 285 probands with a parental history of asthma or hayfever. Atopic dermatitis and food allergy in the first 12 months of life were assessed, and it was found that AD was less common in children exposed to dogs (30% versus 50%), with no protection evident for cat exposure. CD14 genotype effects interacted with dog ownership (see Table).

Exposed Unexposed
Table 20. IL10*-159C>T variant and infantile AD [Gern et al 2004].
Exposed to dogsGroupGenotype countsT allele frequency
C/CC/TT/T
AD 714 1 0.364
Control 1125 16 0.548
AD 1538 13 0.485
Control 2319 17 0.449

Eder et al [2005] describe gene by environment interactions in the Allergy and Endotoxin Study (Austria and Germany). Among 624 rural children, there was no overall association between the CD14*-260C>T polymorphism and sIgE level or atopy. However, both mattress endotoxin levels and contact with farm animals were found to interact with genotype in a complex fashion (see Table).

Table 21. Adjusted odds ratios (95% CI) for the association between CD14*-260C>T variant and specific sIgE (>=3.5 IU/mL) versus exposure to animals [Eder et al 2005].
Animal contactC/CC/TT/T
None 10.52 (0.15-1.84)0.69 (0.15-3.11)
Dogs/cats only10.23 (0.07-0.80)0.25 (0.05-1.15)
Stable animals11.92 (0.88-4.19)2.56 (1.07-6.15)

6.18 IL9 variants and asthma: Levitt et al [2000]

This US patent describes association between a T117M variant in exon 5 of IL9 (rs2069885) and sIgE level. In an initial sample of Italian asthma families and unselected subsequent samples, total sIgE was lower in M/M homozygotes than the other genotypes. The estimated M allele frequency was 12-13%, so homozygotes are relatively infrequent.

The University of Washington-Fred Hutchison CRC Variation Discovery Resource (SeattleSNPs) found the M117 allele frequency to be 13% in African-Americans and 12% in their CEPH panel.

6.19 SPINK5 and Netherton's disease

SPINK5 (5q32) encodes a serine protease inhibitor ("lymphoepithelial Kazal-type-related inhibitor"). Mutations in SPINK5 lead to Netherton's syndrome, a rare autosomal recessive disorder characterized by congenital ichthyosis, "bamboo" hair and severe atopy (with high total sIgE levels). It may modulate Th1/Th2 skewing via effects on NFKB (a mechanism shared by other serpins).

Moffat et al [2000] tested for association between several SNPs in the gene and total serum IgE level in the Busselton sample (1004 individuals in 230 families). Several of these were associated with IgE level, the strongest result with a silent polymorphism (118C>T), but common haplotypes combining the variants did not exhibit association.

Cookson's group [Walley et al 2001] further describe association to atopy in 148 families containing a child with atopic dermatitis (Greater Ormond Street dermatology outpatients panel) and the UK asthma family panel (73 families). A total of 32 SNPs through the gene were typed, and a significant TDT was exhibited for 2 cSNPs in exons 13 and 14 (best P-value, 0.005 for atopy versus Q420L, uncorrected for multiple testing). There were parent of origin effects, with maternal overtransmission in the both samples, and weaker paternal overtransmission in the asthma panel alone.

Kabesch et al [2004] looked at Q420L in the Munich children (1161 individuals). There was weak evidence of association with asthma (OR=1.8, P=0.04), and especially if accompanied by atopic dermatitis (OR=4.6, P=0.007). There was no assocation with atopic dermatitis per se, skin atopy or total sIgE level.

Rihs et al [2005] developed a high throughput assay for Q420L, and applied it to a case-control study of latex allergy (63 asthma, 172 other symptoms; 80 nonatopic controls). No association was found (for any allergy, P=0.46).

6.20 HAVCR1 and BHR

McIntire et al [2001] described mapping airway hyperesponsiveness in congenics from a BALB/c background (high BHR) with DBA/2 (low BHR, low TH2 responsiveness) derived chromosomal intervals. One line, carrying a small (20 cM) DBA/2 region homologous to human 5q23-35, was hyporesponsive to keyhole limpet hemocyanin (increases IL-4 production in BALB/c mice) or ovalbumin inhalation post-immunisation. This segment acted recessively in backcrosses to BALB/c.

Further (2000) backcrosses were used to perform high resolution linkage mapping. The BHR and IL4 responsiveness locus Tapr was found to be separate from the cytokine gene cluster and tightly linked to a marker within the mouse homologue to the rat Kim1 locus, homologue itself to the human Hepatitis A Virus receptor gene HAVCR1.

Three members of the newly designated TIM gene family were cloned (30-45% amino acid sequence identity with human HAVcr-1 protein), and multiple polymorphisms described in TIM1 and TIM3 that cosegregated with BHR. The HAVCR1 homologue TIM1 was shown to be expressed on CD4+ T cells and transcribed during primary antigen stimulation. Furthermore, the author cited two studies by Matricardi and coworkers showing atopic individuals being less likely to show seropositivity to Hepatitis A Virus, toxoplasma and H. pyloridis [Matricardi et al 1999, 2000], as evidence that the relationship between the polymorphisms and BHR was likely to be causative rather than due to linkage disequilibrium. HAV infection has not protected against asthma in other studies, however.

Monney et al [2002] showed that the TIM3 protein is only present on activated Th1 cells and macrophages, that such cells are common in the CNS at the onset of experimental autoimmune encephalomyelitis (EAE, a well known model for MS), and that anti-TIM3 antibodies exacerbated the course of the EAE.

Graves et al [2005] tested for association of the TIM family genes with sathma and atopy in the Tucson Children's Respiratory Study. There were 564 individuals with available DNA, SPT, and four rounds of questionnaire data (birth to age 16 years), of whom one-third gave a history of asthma. At 8 of 21 SNPs genotyped, there was significant heterogeneity in frequency between Hispanic and European descended participants. While no SNPs were associated to asthma, association to atopy (SPT) and to atopic dermatitis was observed (best P=0.002 for rs3087616 in TIM3). TIM3 haplotypes were similarly associated; for the rs3087616-rs1036199-rs470853 GGT haplotype, the relative risk for atopy was 1.24 (P=0.0007).

6.21 CYFIP2 and asthma

Noguchi et al [2005] describe association mapping in 26 genes (90 SNPs) under their linkage peak [Noguchi et al 1997; Yoyouki et al 2000] centred roughly on HAVCR1. HAVCR1 and IL12B were excluded, but there were strongly positive TDT results for CYFIP2. An additional 15 SNPs in CYFIP2 were therefore genotyped (a total of 30 SNPs in 6 LD blocks across the gene).

There were 6 SNPs (in complete LD) where 28/33 transmissions to cases were observed (individual P=7e-5, lod=3.46), including rs12654973. All were noncoding, but homozygotes haplotype carriers had 25% greater expression of CYFIP2 in lymphocytes than heterozygotes (P=0.03).

CYFIP2 is the gene coding for cytoplasmic fragile X mental retardation protein (FMRP) interacting protein 2 (also called p53 inducible protein, PIR121). The protein colocalizes with FMRP and ribosomes in the cytoplasm, is probably involved in p53-dependent apoptosis, but also in T cell fibronectin-mediated adhesion [Mayne et al 2004] in vitro as well as in multiple sclerosis. In keeping with the involvement in apoptosis and adhesion, it is also regulates the actin cytoskeleton. Its' close (88% homology) relative CYFIP1 is on 15q11.2.

6.22 UGRP1 and asthma

Uteroglobin-related protein 1 is, as the name might suggest, similar in amino acid sequence to uteroglobin/CC16, and maps to 5q31 [Niimi et al 2002]. It is 93 residues long. The gene (secretoglobin, family 3a, number 2; SCGB3A2) spans 3 kbp (3 exons). There are no known coding variants.

Niimi et al [2002] also performed a case-control analysis of association of a -112G>A promotor variant, that they had shown to affect nuclear protein binding. The -112A allele was more common among 85 asthmatics (Hokkaido), compared to 85 controls (P=0.002).

Table 22. SCGB3A2 -112G>A promotor variant and asthma [Niimi et al 2002].
GroupGenotype counts-112A allele frequency
G/GA/GA/A
Control7013 20.10
Asthma 5031 30.22

6.23 IL12B and asthma

Interleukin-12 is a heterodimer, the p35 35 kD alpha chain coded by IL12A (3p), and the p40 (40 kD) beta chain by IL12B (5q13). The p40 chain is also shared with IL-23. The IL12B gene contains 7 exons and spans 12 kbp.

IL-12 is another cytokine with a complex pattern of action. It is secreted by macrophages and other antigen presenting cells. Administration of IL-12 reduces circulating eosinophil count (it induces eosinophil apoptosis via stimulation of IL-12 receptors on that cell), while treatment of RSV-infected mice with anti-IL-12 results in BHR, mucus production, and eosinophilia [Tekkanat et al 2001]. However, it failed to reduce the bronchoconstriction or the late phase response to allergen challenge in human asthmatics [Bryan et al 2000]. Natural Killer cells exposed to IL-12 tend to secrete gamma-interferon, and in individuals with IL-12 deficiency due to IL12B deletion, impairment of gamma-interferon production and bacterial immunity have been observed. In the skin IL-12 induces DNA repair and so reduces UV-induced apoptosis.

Morahan et al [2001] reported an association between a 3'UTR SNP in IL12B and IDDM. This has not replicated in a number of studies.

Using SSCP, Noguchi et al [2001] identified four variants in IL12B, two rare (1/100, 1/200). The other SNPs were tested for association with asthma and rhinitis via a TDT using the families described in Yokouki et al [2000]. Neither the S226N (allele frequency 4%) nor 1188A>C (50%) variants were overtransmitted.

More recently, Morahan et al [2002] report association between asthma and an IL12B promotor polymorphism, that they note is not in strong linkage disequilibrium with the 3'UTR polymorphism. A total of 844 children from the West Australian Pregnancy Cohort Study were genotyped at IL12B SNPs. This included 203 asthmatic children, of whom 39 were classified as severely affected.

In the first 411 children genotyped, there was a weakly significant overall chi-square for association, and Hardy-Weinberg disequilibrium in some of the asthmatics (Table). The authors interpreted this as suggesting overdominance in the severely affected subgroup.

Table 23. Frequency of the IL12B promotor variant in asthmatics and controls [Morahan et al 2002].
GroupGenotype counts 1 allele frequencyHWE P-value
1/11/22/2
Nonatopic Control26 67 310.480.358
Atopic Control 43 81 450.490.591
Mild Asthma 10 20 120.480.768
Moderate Asthma 20 17 180.520.004
Severe Asthma 2 16 3 0.520.013

This specific hypothesis was then tested in the remaining 85 asthmatics, where again there was H-W disequilbrium in the "severe" group due to an excess number of heterozygotes (16/18). Functional studies of the three genotypes found suggestive decreases in IL12B mRNA expression and in IL-12 (p70) production in heterozygote compared to either type of homozygote cultured PBMCs.

There was no association with the 3'UTR polymorphism. Data on haplotypes was not presented: according to Huang et al [2000] and Morahan et al [2001], the distance between these polymorphisms is 20 kbp.

Khoo et al [2004] looked at a subsample of 244 subjects from the 1957 UK birth cohort enriched for a history of childhood asthma followed over 7 examinations. There was no association between IL12B promoter genotype and asthma or atopy, with a possible association with low (standardised) FEV1 at age 7 only. Noguchi et al [2001] also failed to find association within their panel of 47 families linked to the region.

6.24 Glucocorticoid receptors and asthma

This seems a natural candidate for steroid resistant asthma.

6.25 Kurz et al [2004] and PTGER4

In the Hutterite study, linkage of BHR to D5S1470 (5p13.3) was observed, with a lod=2.1. In an ASHG meeting abstract, Kurz et al describe overtransmission of the 177 bp allele at this marker, which was followed up SNP genotyping in the neighbouring genes: the prostaglandin E4 receptor (PTGER4, deletion of which impairs contact hypersensitivity and skin APC migration in mice [Kabashima et al 2003]), leukemia inhibitory factor receptor (LIFR, Stuve-Wiedemann syndrome), atrial natriuretic peptide clearance receptor (NPR3), and ADAMTS12. Multiple SNPs exhibited overtransmission, but most of the linkage signal could be abolished by excluding individuals carrying variants at LIFR and PTGER4.

The association was replicated in a German case-control panel of 231 asthmatics, 196 atopics and 270 controls. PTGER4 was associated with asthma (P=0.008), while LIFR and NPR SNPs were associated with atopy.

The prostaglandin D2 receptor gene (PTGDR) has also exhibited association to asthma (see below). The knock-out phenotypes for both PTGDR and PTGER4 seem quite similar.

6.26 Beta adrenergic receptor and asthma

As noted earlier, Szentivanyi [Szentivanyi 1968] proposed the unifying hypothesis that asthma represents an underresponsiveness of the lungs to sympathetic neurotransmitters, suggesting the beta adrenergic receptor as the most likely site for this. A number of recent studies have looked at functional mutations in the beta-2-adrenergic receptor gene ADRB2 (5q31-32).

Table 24. Studies of association between asthma and beta-2-adrenergic receptor (ADRB2).
PolymorphismStudyMutation frequency (No. subjects)
AsthmaticsControls
RFLPPotter et al [1993](42)(30)
R16GReishaus et al [1993]71.6% (51)73.2% (56)
Turki et al [1995]66.7% (45)---
Kowalski et al [1997]*31.7% (60)41.4% (47)
Martinez et al [1997]**66.7% (78)59.7% (191)
Dewar et al [1997]***67% (321)
Hakonarson et al [1999]55.8% (94)55.2% (94)
Wang et al [2001]38.3% (128)48.5% (136)
Leung et al [2002]42.1% (76)42.1% (70)
Hermann et al [2002]--- 38.0% (2010)
Horvath et al [2004]39.4% ---
Ng et al [2004] China--- 62.1% (207)
Ng et al [2004] France--- 37.9% (858)
Ng et al [2004] Spain--- 39.7% (395)
Cho et al [2005] Korea51.3% (37)47.8% (157)
Q27EReishaus et al [1993]49.0% (51)49.1% (56)
Hall et al [1995]55.0% (65)---
Turki et al [1995]51.1% (45)---
Kowalski et al [1997]*43.3% (60)51.1% (47)
Martinez et al [1997]**38.5% (78)35.3% (191)
Dewar et al [1997]***48% (322)
Hakonarson et al [1999]39.5% (94)37.5% (94)
Wang et al [2001]8.2% (128)8.8% (136)
Leung et al [2002]7.9% (76)10.7% (70)
Hermann et al [2002]--- 42.0% (2010)
Horvath et al [2004]35.5% ---
Ng et al [2004] China--- 11.8% (207)
Ng et al [2004] France--- 39.6% (858)
Ng et al [2004] Spain--- 38.7% (395)
Cho et al [2005] Korea10.5% (38)8.6% (157)

* Seasonal asthma and/or rhinitis.
** Wheeze in the previous 12 months.
*** All members of family ascertained through multiple asthmatic probands.

6.27 Potter et al [1993]

The frequency of an RFLP in the beta-adrenoceptor was measured in 42 atopic children and 30 controls by these authors [1993]. Two alleles were recognised, but were equally frequent in both both groups. There was apparently little polymorphism in this receptor, as over 50 restriction enzymes were screened to obtain this RFLP.

6.28 Reihaus et al [1993]

By contrast, nine point mutations of the beta-2-adrenoceptor among 107 subjects (51 asthmatic). No excess of a particular polymorphism was found in the asthma group compared to controls, but an R16G mutation (56% of the overall sample) was found more commonly in steroid-dependent asthmatics compared to less severe asthmatics.

6.29 Hall et al [1995]

This paper [1995] describes a similar study of the beta-2- adrenoceptor polymorphism in 65 asthmatics. These authors concentrated on the R16G (Arg16Gly, rs1042713) and another, Q27E (Gln27Glu, rs1042714), mutation found to be common by Reihaus et al [1993], although the frequency of the latter did not differ between asthmatics and controls. However, in vitro studies have suggested that the Q27E mutation might be associated with diminished downregulation after agonist exposure [Green et al 1994].

The frequency of the E27 mutation was 55% in this sample. E27/E27 homozygotes were more likely to be atopic (trend X2(1 df)=3.1, P=0.08), and exhibited a significantly higher PD20 to a methacholine Yan challenge (P=0.03) than Q27/Q27 homozygotes. No effects of R16G were detectable. The absence of controls and small sample means replication is needed.

This report [Tan et al 1997] reanalysed three previous studies (total N=22) including Hall et al [1995] to conclude that the G16/G16 carriers became more desensitised to formoterol than asthmatics carrying R16/R16.

6.30 Turki et al [1995]

This study [1995] compared 23 nocturnal asthmatics (who tended to have a lower mean diurnal FEV1, and to be more likely to be steroid dependent) to 23 "normal" asthmatics. The R16G frequency was 80.4 percent in the nocturnal group, and 50.0 percent in the nonnocturnal group (P=0.004).

6.31 Kowalski et al [1997]

This abstract [1997] compares the frequencies of the two polymorphisms in 60 atopics ("seasonal asthma and/or rhinitis") and 47 history and SPT negative controls. The table they give is hard to interpret, but suggests a severe dearth of Q27/Q27 heterozygotes among the cases (HWE X2(1 df)=37.4; cases versus controls X2(2 df)=27.4).

6.32 Dewar et al [1997]

The Southhampton group [1997] compared the frequencies of the two polymorphisms in 324 members of 60 families ascertained for asthma. In the combined segregation-association analysis using COMDS, asthma and atopy were not correlated with either polymorphism. However, sIgE was (P=0.009). There was also weak evidence for linkage between sIgE and the position 27 ADRB2 polymorphism (with NOPAR, P=0.04).

6.33 Martinez et al [1997]

A subset (N=269) of participants in the Tucson Children's Respiratory Study were typed at these two ADRB2 polymorphisms. As before, there was no relationship between genotype and presence of a history of asthma (N=37) or wheeze in the previous 12 months (N=78). However, there was evidence of a relationship with a significant bronchodilator response (here defined as a 15% improvement in FEV1 as a fraction of predicted FEV1). The direction of the relationship was in keeping with Hall et al [1995] and Tan et al [1997] in that the G16/G16 homozygotes were less likely to exhibit a bronchodilator response.

6.34 Weir et al [1998]

These authors [1998] report more evidence of an association between ADRB2 polymorphisms and asthma. They examined 78 cases of fatal or near-fatal asthma, 82 cases of nonfatal asthma, and 84 controls, all presumably ascertained through the University of British Columbia. They also performed subgroup analyses among the "nonfatal" asthmatics, dividing them into moderate and mild severity groups based on therapy (all but 9 of 41 the mild group did not use inhaled corticosteroids). The sample was also stratified on ethnicity.

Importantly, the authors found significant differences in allele frequencies across Whites, Blacks and Asians. For example, while the frequency of the Gly16 allele was 61% in both Causasian asthmatics and Caucasian controls, it was 40% in the Asian asthmatics.

Although no differences were found between asthmatics and nonasthmatics as a whole, or between fatal asthma and controls for either G16 or Q27, the moderate asthmatics (N=33) had a higher frequency of the Q27 than mild asthmatics. The lack of an obvious dose-response relationship leaves open the possibility of a Type I error.

Table 25. ADRB2 polymorphism frequencies among Caucasian asthmatics and controls studied by Weir et al [1998].
SubgroupNQ27 frequency
Severe Asthma690.54
Mod Asthma330.67
Mild Asthma410.48
Nonasthmatics840.60

6.35 D'Amato et al [1998]

This Italian study documents a relationship between nonspecific BHR and the G16/Q27 haplotype that persisted after including total and specific IgE as a covariate in the analysis.

6.36 Hakonarson et al [1999]

This abstract includes an author from deCode and describes a case-control study of 94 Icelandic "moderate to severe" asthmatics and 94 controls. There were no association with either the R16G or Q27E variant. This study was presented in more in detail in Hakonarson et al [2001].

6.37 Moore et al [2000]

The relationship between response of isolated airway smooth muscle preparations to isoproterenol was examined in 15 individuals of differing ADRB2 genotype. The G16/Q27 haplotype was associated with greater desensitisations. They comment that E27 was in strong LD with the R19 allele in the 5' leader peptide, which was more strongly associated with the phenotype.

6.38 Wang et al [2001]

Anqing (China) is the site of a large genetic epidemiological study of asthma and lung diseases. Over 2000 families containing two or more doctor-diagnosed asthmatics have been ascertained, and 7100 of the 10000 subjects have undergone methacholine inhalation challenge.

Wang et al [2001] describe a nested case-control study within these families: 128 BHR+ asthmatics and 136 BHR- nonasthmatic controls. Cases were twice as likely to have ever smoked (30% versus 17%). The R16 allele was more common among the asthmatics (OR=1.04, P=0.04), but since the effect was stronger among the ever smokers (see Table), the authors interpreted the data as suggestive of a recessive gene by smoking interaction.

Table 26. Frequency of ADRB2*R16 variant in asthmatics and controls versus smoking status [Wang et al 2001].
SmokingAsthmaG/GR/GR/RR16 FrequencyHWE P-val
Ever Yes 6 11 22 70.5% 0.045
Ever No 6 12 5 47.8% 0.827
Never Yes 16 43 30 57.9% 0.931
Never No 28 52 33 52.2% 0.408
I note that these data are also compatible with a simple multiplicative no-interaction model (for each R16 allele OR=1.4, ever smoking OR=2.1). Including age, sex, FEV1 and occupation did not alter the relationship, while plotting risk versus pack-years of smoking shows a steep dose-response slope for R16/R16 homozygotes only.

6.39 Leung et al [2002]

This Hong-Kong study contrasted 76 asthmatic children attending outpatients and 70 controls. The allele frequencies did not differ between the groups and are close to those from Wang et al [2001].

6.40 Santillan et al [2003]

This is a Mexican case-control study (N=907). The E27 and G16/E27 haplotype were found to be protective against asthma, especially in women. The G16 allele was associated with nocturnal asthma.

6.41 Silverman et al [2003]; Horvath et al [2004]

These papers describes one of the first outings for the haplotypic FBAT. The CAMP study provided 652 nuclear families (707 asthmatics, 2011 individuals) who were genotyped at seven polymorphic SNPs in ADRB2. There were no significant associations between any single SNP and measures of bronchodilator response or asthma symptom score. The 20 asthmatics with an FEV1 below 80% predicted did give significant results for 4 SNPs including R16G and Q27E.

By contrast, highly significant results were obtained for 7-SNP haplotypes and asthma (X29=35.6, P=4.10-5). These results do seem to be driven by less common haplotypes, so I do worry about breakdown of the chi-square approximation to the score statistic.

Table 27. Haplotypes and FBAT association score from Horvath et al [2004]
Haplotype16/27ProportionZ-score
ATACGCCRQ0.357+0.54
GCGGGCCGE0.355+0.63
GTGCACAGQ0.180-0.75
GTACGCCRQ0.036+2.46
GTGCACCGQ0.029+0.41
Other --0.024--

6.42 Burchard et al [2004]

The Genetics of Asthma in Latino Americans (GALA) study includes both families and case-control samples from the US, Mexico and Puerto Rico. Among Puerto Ricans but not Mexicans, the R16 allele strongly predicted bronchodilator response, especially in subjects with a baseline FEV1 < 80% predicted. Burchard et al [2004] note that in the US, Puerto Ricans and Mexicans have the highest and lowest prevalences of asthma respectively, and Puerto Rican asthmatics are characterised by the poorest bronchodilator response.

6.43 Turner et al [2004]

In a West Australian cohort of children (N=253), the R16/R16 homozygotes demonstrated persistent differences compared to the G16/G16 group: higher AHR at 1 month old, a lower FEV1 at 11 years old, and a higher admission hospital admission rate for asthma (OR=3.2, 0.9-12.6), though not for the diagnosis of asthma per se.

6.44 Litonjua et al [2004]

A nested case-control study within the Normative Aging Study cohort looked at AHR cases (methacholine challenge, N=152) versus nonresponsive nonasthmatic controls (N=391). There was weak association of AHR risk to R16 (P=0.10; test for smoking by genotype interaction, P=0.37), and more so to the G16-Q27 haplotype (P=0.02), especially in the relatively small nonsmoking group.

Table 28. Frequency of ADRB2*R16 variant in BHR+ cases and BHR- controls versus smoking status [Litonjua et al 2004].
SmokingAHR G/GR/GR/RR16 FrequencyHWE P-val
Ever Yes 42 59 17 39.4% 0.702
Ever No 10411436 36.6% 0.591
Never Yes 9 17 8 48.5% 1.000
Never No 58 62 17 35.0% 1.000
6.45 Israel et al [2004]

This is a double-blind randomized controlled trial of regular salbutamol in R16/R16 and G16/G16 carriers matched on genotype and baseline FEV1. After 16 weeks on active therapy, morning peak expiratory flow rate had increased by 14 l/min in G16/G16 homozygotes, while by only 2 l/min in R16/R16 homozygotes. Most interestingly, placebo had no effect on G16/G16, but lead to a 12 l/min rise in PEFR in R16/R16. In the run-in for the study, when all subjects were salbutamol-free, R16/R16 had risen 23 l/min, and G16/G16 only 2 l/min (pretrial salbutamol use in these mild asthmatics was 1 puff/day).

6.46 Cho et al [2005]

1159 children from the Korean island of Jeju completed the ISAAC questionnaire and underwent methacholine inhalation challenge. A history of wheezing was reported by 15% of the sample, and 18% exhibited AHR. Genotyping at the codon 16 and 27 polymorphisms was undertaken in 195 AHR+ subjects, none of whom had taken bronchodilators in the previouis two weeks.

There was no association between genotype and history of wheeze. The R16 homozygotes did exhibit a greater response to bronchodilator.
Bronchodilator response post methacholine challenge in AHR+ Korean children [Cho et al 2005] versus codon 16/27 ADRB2 genotype.
GenotypeNMean BD Response (95%CI)
RQ/RQ5262.8 (53.8-71.8)
RQ/GE2161.6 (47.3-75.9)
RQ/GQ7653.3 (46.9-59.7)
GQ/GE1352.1 (31.0-73.2)
GQ/GQ3243.2 (33.0-53.4)

6.47 PCDH1 and asthma

Koppelman et al [2009] carried out fine mapping under their 5q13 linkage peak in 200 Dutch families and picked out the protocadherin 1 gene (PCDH1) for replication studies. Association to variants in the gene was detected in multiple samples.

7 Chromosome 12 atopy genes

Gamma interferon (IFNG, 12q13) plays an important role in T-cell regulation that makes it another good candidate gene for involvement in allergic disease. The first published examination of this region was that of Watson and coworkers [Watson 1995]. The maximal lod score for IGF1 (tightly linked to D12S318 and PAH and approximately 30 cM distal to IFNG) was 0.09 at 25 percent recombination distance.

Another candidate that has been examined is NOS1 (12q24.2, 116.1 Mbp from 12pter). The mouse knockouts for NOS1 exhibit less nonspecific BHR after OVA sensitization [DeSanctis et al 1999]. Variation in this gene has been tested for association to a huge variety of phenotypes, ranging from lupus nephritis, cluster headache, schizophrenia, COPD and asthma.

More recently, Dillon et al [2004] described the IL31 gene (12q24.31, 121.1 Mbp from 12pter), that codes for a novel four-helix bundle cytokine. Il-31 is largely produced in activated CD4+ T-cells, and especially Th2 cells. The IL-31 receptor is upregulated in IFN-gamma treated monocytes. IL-31 levels rise in the lung after allergen challenge in the ovalbumin BHR mouse model, and more so in BALB/c mice. A transgenic mouse overexpressing IL-31 developed pruritic alopecia and conjunctivitis, with histopathology reminiscent of atopic dermatitis.

The closest microsatellite markers to IL31 are D12S1349 and D12S1578. At approximately 142 cM on the Decode map of chromosome 12, this is 25-40 cM distal to the region showing strongest evidence for linkage to asthma (see below), though some linkage peaks have included this locus.

Variation in the Vitamin D receptor (VDR, 12q12) is suspected to modulate a variety of immune processes, and association with asthma was examined in three datasets in 2004. Wittke et al [2004] have shown that VDR knockout mice do not develop airway inflammation or BHR following allergen sensitisation.

Table 29. Ensemble physical map positions of candidate genes and allergy-linked markers on chromosome 12q.
Gene or Marker Location (Mbp)Remarks
D12S1048 39.313
IRAK4 42.45 IL1R-assoc kinase
D12S85 45.623 "IBD2"
VDR 46.523
D12S398 51.483 "RA21"
ITGB7 51.86 Integrin beta 7
STAT6 55.775 IL-4 Stat
IFNG 66.835
IL22 66.928
D12S379 83.461
KITLG 87.393 c-Kit ligand
IGF1 101.292 Somatomedin C
PAH 101.735
NOS1 116.111
IL31 121.1

7.1 Barnes et al [1996]

Barnes and coworkers [Barnes 1996] have reported evidence of linkage and association between asthma and tIgE over a region stretching from close to IGF1 up to IFNG in two different populations -- the Amish families originally used to detect linkage to 5q, and a sample of 29 Barbadian pedigrees (693 individuals) ascertained through a proband with a history of asthma.

Asthmatic sibpairs exhibited increased ibd sharing at D12S379 (61.8 percent), D12S95 (67.2 percent), PAH (58.5 percent) and D12S360 (58.6 percent). Similarly, Haseman-Elston regression analyses were significant for PAH and D12S360. In the Amish sample, where a diagnosis of asthma was not available on any family members, there were significant Haseman-Elston results for D12S360, IGF1, as well as PLA2. Several replications of this finding have been recently presented.

7.2 Collins et al [1996]

The Southampton group performed further analysis of their 240 families for linkage to 12q markers under their nonparametric BETA model [Morton 1996]. The best single point lod score was 3.3 for wheezing and D12S366, and the multipoint peak of 2.3 was near D12S97.

7.3 Nickel et al [1997]

The German Multicentre Atopy Study has followed a cohort of 1314 children born in 1990. In this paper [1997], the authors describe results from the TDT for chromosome 12 markers in a subset of children whose total sIgE level exceeds the 85th percentile in the cohort - 214 IU/ml (N=52).

Significant global TDT results were obtained for D12S379 and D12S351 (P=0.001). The most extreme transmission imbalance was for D12S351*161, which was transmitted to the proband 31/38 times.

7.4 IFNG mutation screening by Hayden et al [1997]

These authors have screened two groups ascertained to have high rates of asthma: a subset of families from the Busselton study, and individuals from a genetic semi-isolate (a Venezualuan island population). They performed SSCP on all 265 subjects, and heteroduplex analysis on a random 10% subsample. Sequencing of the first exon in a small number of subjects found no variants from that previously reported. The authors concede that intronic or upstream polymorphisms, possibly of (transcriptional) regulatory importance, might remain to be discovered.

7.5 Association of NOS1 variants [Grasemann et al 1999, 2000]

Neuronal nitric oxide synthase gene variants were tested in a case-control study (410 asthmatics, 1481 controls). The best result was for a 3'UTR CA-repeat polymorphism (10 alleles), with the 17-repeat allele being more common in cases than in either of the two control samples (Physicians Health Study and convenience sample of Boston college students).
Table 30. NOS1 CA repeat polymorphism versus asthma from Grasemann et al [2000]
AlleleCasesStudentsPHS
17 0.8280.763 0.774
18 0.0610.119 0.101
Other 0.1110.118 0.125

7.6 Malerba et al [2000]

This abstract describes fine mapping around several peaks in their scan of 117 Italian families (also see below). On chromosome 12q, their best P-value was 6x10-4 was for D12S270 using Genehunter's NPL statistic (equivalent lod=2.1). This is approximately 30 Mb proximal to IFNG.

7.7 Demenais et al [2002]

The 14 repeat allele at the IFNG CA-repeat was found via combined segregation-linkage analysis to be associated with low sIgE level in EGEA families.

7.8 Raby et al [2002, 2003]

This paper describes linkage analysis of 12q markers in 112 asthmatic children from 55 families in the CAMP study. There was weak evidence of linkage to asthma (lod=0.6) and airway responsiveness (lod=1.91).

7.9 Shao et al [2004]

Shao and coworkers describe a small linkage study (39 asthmatic children in 19 nuclear families) and a case-control association study of 115 asthmatics and 184 controls, using chromosome 12 markers. All subjects came from Osaka and Sendai. The best multipoint lod score was 2.9, for asthma at 150 cM. Of the candidate polymorphisms tested (2 in STAT6; 1 in NOS1, IFNG, AICDA respectively), the best results were a lod of 1.9 for a STAT6 exon 1 GT repeat polymorphism, and 2.1 for the NOS1 intron 2 GT repeat.

7.10 Poon et al [2004] and VDR

Using the collection of Quebecois asthma families described earlier, Poon and coworkers tested for association of 12 Vitamin D Receptor SNPs with asthma and allergy. There were 223 families (N=1139 individuals) containing 570 asthmatics.

There was transmission distortion for the SNPs to both asthmatics, and those with high tIgE levels (FBAT test), with the best P-value being for atopy and rs2239185 at P=0.002. One common haplotype involving this SNP (45% frequency) was overtransmitted, giving a uncorrected P=0.0004.

7.11 Raby et al [2004] and VDR

The CAMP study looked at 7 candidate genes across 12q, IFNG, STAT6, CPM, KITLG, IL22, IRAK3 and VDR. They genotyped at 7 VDR SNPs, including rs2239185. There was undertransmission of rs2239185 to asthmatic children, and the best single SNP result was for the neighbouring (5.7 kbp distant) rs7975232 at P=0.01.

Replication was looked for using an asthma case-control panel (517 cases, 519 matched controls) from the Nurses Health Study. In this case, the minor (C) allele at rs2239185 was increased in cases (P=0.02), in keeping with the Canadian results, and contrary to those from the CAMP transmission tests.

7.12 Vollmert et al [2004] and VDR/ITGB7

7.13 Nieters et al [2004] and IL2

From the European Prospective Investigation into Cancer and Nutrition-Heidelberg study sample, Nieter and coworkers assembled a nested case-control study of hayfever versus 13 cytokine genes (322 cases, 322 controls). The IL2 -330T>G homozygotes were at a 2.6 fold increased risk of hayfever. Other results were for IL6 -174C>G.

7.14 Pykalainen et al [2005] and STAT6

In the set of 120 Finnish asthmatic families [Laitinen et al 1997, 2000, 2004] previously screened at a number of candidate genes, STAT6 SNPs were tested for association. HPLC screening of STAT6 rediscovered 5 noncoding SNPs, none of which exhibited association with asthma or high tIgE level.

7.15 Balaci et al [2007] and IRAK3

Using Sardinian families ascertained to contain an asthmatic sib pair (100 families, 121 ASPs), this study obtained linkage to a 11 cM region of chromosome 12q centred on D12S75 (ASP lod=3.6 for the 60 families with onset before 13 years). Sequencing of linked cases excluded IFNG, IL22 and IL26. Variants in IRAK3 (encoding IRAK-M, involved in TLR signal transduction) did show association. Using the initial pedigrees supplemented by 194 additional asthma families, the authors detected association to multiple variants in a TDT analysis of 22 SNPs spanning IRAK3 (best uncorrected SNP P=3x10-5 for rs1177578 in the childhood onset subgroup). Association to this gene was confirmed by a case-control analysis comparing 139 early-onset probands to a panel of 460 healthy controls. A haplotype comprising 3 intragenic tagging SNPs (rs11465955, rs1624395 and rs1370128) was sufficent to explain most or all of the association and linkage signal.

In a separate Italian case-control panel, the best 6 SNPs were tested for association, again confirming rs1624395 and rs1370128 to be associated (uncorrected P=0.004 and 0.002). The risk haplotype frequencies were similar in the Sardinian and Italian panels.

Raby et al [2004] screened 7 SNPs in IRAK3, finding no association with asthma or related phenotypes, but unfortunately not presenting any results. Nagashima et al [2006] report association results from a Japanese case-control study of childhod asthma (391 cases, 639 healthy adult controls) that included genotyping at 12 IRAK3 SNPs. None exhibited any evidence of association with asthma. Instead of directly genotyping rs11465955, they relied on LD with a tagging SNP (rs2293657), which was significantly associated with asthma in the Sardinian dataset.

Comparison of results from Nagashima et al [2006] and Balaci et al [2007] for rs2293657 in IRAK3. In Sardinian families, this SNP gave a TDT P=0.0007, and a case-control P=0.0008.
Group N T/T A/T A/AMAF
Sardinian Child Asthma 139 - - - 0.460
Sardinian Controls 460 - - - 0.341
Japanese Control 639 0.76 0.22 0.02 0.100
Japanese Child Asthma 391 0.72 0.26 0.02 0.150
Japanese Adult Asthma 462 0.78 0.21 0.01 0.115
Yusuke Japan Controls 735 - - - 0.128
AFD European 24 0.50 0.42 0.08 0.292
AFD Asian 24 0.75 0.21 0.04 0.146
AFD African 23 0.87 0.13 0.00 0.065


8 Chromosome 13 atopy genes

Several genome scans (see below) have detected linkage of asthma and atopy with chromosome 13q. For example, Pilia [1999] obtained their best result, a lod score of 3.1 (single point) for D13S164.

8.1 Kimura et al [1999] on chromosome 13

Two genome scans (see below) have detected linkage to chromosome 13q. The Tsukuba group who have previously reported on 5q and 11q markers [Noguchi et al 1997], typed their 86 families (172 atopic asthmatic children, 325 individuals) at 15 markers spanning chromosome 13. Affected sib pair analyses obtained two peak lods of 2.4 and 2.0 that were 50 cM apart, the latter centred between D13S153 and D13S156. A multiallelic TDT P-value of 0.007 was obtained for D13S153, which is the marker found linked to atopy by Daniels et al [1996]. A candidate gene with 1.5 cM of D13S153 in STAT5a. As noted earlier, Esterase D (ESD) has been linked to total sIgE, and is also only 1.4 cM from D13S153.

8.2 Pandit et al [1999] on the Endothelin B Receptor (EDNRB)

The gene for this receptor has been located on 13q14. In a case-control panel (40 "random" asthmatics, 51 asthmatics from families linked to 13q, 40 CEPH controls), eight variants (five coding) were found to be uncorrelated with disease.

8.3 Leaves et al [1999] on chromosome 13

The Oxford group report confirmation of linkage to 13q14 using a total of 604 affected sib pairs giving a MLS of over 5. Fine mapping using the TDT had reduced their region of interest to 600 kb.

8.4 Cysteinyl leukotriene 2 receptor (CYSLTR2) and atopy

Takasaki et al [2000] mapped the CYSLTR2 gene to 13q14, "near a marker for atopic asthma". In an ASHG abstract, Thompson et al [2002] describe screening this gene for variants in the Tristan da Cunha population. One coding polymorphism (M202V), of the five detected, was increased in atopic individuals (21% versus 7%, P=0.003). The variant receptor was less activated by LTD4.

8.5 Zhang et al [2003] on PHF11

Anderson et al [2002] described further fine mapping by the Oxford workers. There was significant association to a new microsatellite marker USAT24GF1 in two panels of families.

Using 49 SNPs (4 indels and one repeat marker) spanning 62 kbp centred on USAT24G1, Zhang et al [2003] screened for association to IgE level using the first 80 Busselton families. A quantitative trait TDT, implemented using QTDT, found maximum association to 3 SNPs within PHF11 (best SNP association lod 2.7). This was replicated in the other three panels of families, most strongly in the eczema families. Association to asthma was also observed, with a dose-response effect evidenced in a British asthma case-control panel.

PHF11 is a widely-expressed gene carrying a zinc-finger motif. Splice variants are more tissue specific, one variant being expressed largely in immune-related tissues. It is 1.2 Mbp distal to D13S153, and 2 Mbp from ESD (Esterase D).

Jang et al [2005] report testing of this association using 111 families of children with atopic dermatitis. Seven SNPs were chosen (4 in PHF11 and 3 in SETDB2 and the intergenic gap), with two of those in PHF11 exhibiting a significant TDT (best SNP rs1046295, Bonferroni corrected P=0.034)

McClenaghan et al [2009] genotyped 20 SNPs across PHF11 in two West Australian samples: 230 twin families, and a population-based nested case-control panel (617 case, 706 controls). No SNPs survived Bonferroni corrected testing at P<0.05 for multiple phenotypes.

Gao et al [2009] genotyped 6 PHF11 SNPs in 408 Chinese asthmatics and 288 controls, obtaining a best P=0.0026 for rs16659.


9 Chromosome 17 atopy genes

There are a number of good candidate genes for asthma on chromosome 17 (various chemokines), as well as evidence for linkage to 17q from genome scans, see below.

Table 31. Ensemble physical map positions of candidate genes and allergy-linked markers on chromosome 17.
Gene or Marker Location (Mbp)Remarks
CCL2 29.606-29.608 MCP1
CCL11 29.637-29.639 Eotaxin
CCL5 31.231-31.223 RANTES
D17S250 34.406 (Dizier et al 2000)
ITGB3 42.686-42.745 beta-3 integrin
TBX21 43.166-43.178 T-bet (T-box 21)

9.1 Eotaxin (CCL11, CCL24, CCL26) and asthma

There are three human eotaxins, encoded for by genes on 17q21 (eotaxin-1, CCL11) and 7q11 (2 and 3, CCL24, CCL26), where most CC chemokine genes are clustered. All three act via the CCR3 chemokine receptor, expressed on eosinophils, basophils and Th2 cells.

Shin et al [2003] performed a case-control study of 37 SNPs in the eotaxin genes. There were 550 asthmatics recruited via 4 Seoul hospitals, and 171 controls (spouses and "general population"), all of whom underwent methacholine challenge, SPT to 24 allergens and total and specific IgE determinations.

The best single association was asthma with CCL24*+1265A>G (14% frequency in cases, 23% in controls; P=0.002). This apparently was driven by the atopic controls (Table), in that the main effect on asthma is highly significant, but the interaction term just significant, in a log-linear model treating penetrances as multiplicative (see Table).

Table 32. Frequency of CCL24*+1265A>G variant in asthmatics and controls versus atopy [Shin et al 2003].
Atopy AsthmaA/AA/GG/G1265G FrequencyHWE P-val
Yes Yes 23074 7 14.1% 0.645
Yes No 39 32 7 29.5% 1.000
No Yes 10434 5 15.4% 0.329
No No 57 24 3 17.9% 0.718
Table 33.Analysis of deviance table for Shin et al [2003] data on atopy, asthma and CCL24*+1265A>G.
TermdfDevianceResid. dfResid. DevP-value
NULL -- 11690.50-
Gtp 2463.67 9226.832.070e-101
Asthma 1144.13 8 82.703.327e-33
Atopy 1 43.11 7 39.595.177e-11
Atopy:Gtp 1 0.17 6 39.430.68
Asthma:Gtp 1 12.23 5 27.194.698e-04
Asthma:Atopy 1 21.84 4 5.362.971e-06
Asthma:Atopy:Gtp1 4.86 3 0.500.03

9.2 RANTES (CCL5, SCY5A) and asthma

The CCL5 gene (17q11) encodes the pro-inflammatory chemokine RANTES. RANTES is increased in lavage fluids following allergen challenge in asthma, and is also implicated in autoimmune diseases (eg rheumatoid arthritis) and infection, notably HIV.

Nickel et al [2000] reported a -403 A>G promoter polymorphism in the RANTES gene that was increased among German AD cases (OR=1.84). There was weak linkage of this polymorphism to asthma in Caribbean and African-American asthmatic sib pairs, but no association.

Fryer et al [2000] tested the same -403A>G polymorphism for association with asthma and atopy in a case-control study of 201 individuals. The overall test for association was mildly significant (Fisher P=0.06), and the comparison of all atopic cases to nonatopic controls under an multiplicative penetrance model slightly more so (P=0.01).

Table 34. Frequency of -403A>G RANTES promotor variant in asthmatics and controls [Fryer et al 2000].
GroupGenotype counts A allele frequencyHWE P-value
G/GA/GA/A
Nonatopic Control35 6 1 0.100.335
Atopic Control 16 15 1 0.260.227
Atopic Asthma 66 34 5 0.210.819
Nonatopic Asthma 9 5 1 0.230.795

9.3 MCP1 (CCL2, SCYA2) and asthma

Close to CCL5 (17q11, approx 1.6 Mbp proximal) is the gene encoding Monocyte Chemoattractant Protein, another proinflammatory chemokine. Szalai et al [2001] reported an association between a (-2518A>G) "gene regulatory region" SNP and asthma (OR=2.0, 95%CI=1.4-2.6) and atopy (OR=2.0, 1.4-2.9) in a Hungarian case-control panel (160 asthmatics, 151 atopic nonasthmatics, 303 controls). This SNP has previously been implicated in a variety of diseases (rheumatoid arthitis, HCV infection, atherosclerosis), and affects transcription and serum MCP-1 levels (G/G is increasing genotype [Rovin et al 1999; Tabara et al 2003]). Presence of the G allele predicted increased disease severity and eosinophilia in the asthmatics.

In a subsequent paper [Kozma et al 2002], this group reported no association of the -2518 polymorphism to atopic dermatitis using the same controls and 128 child cases. Yao et al [2004] failed to confirm the asthma association in a Taiwanese case-control panel. There was no evidence of an effect on eosinophil count or sIgE in either of these studies.

Tucci et al [2004] examined the same SNP in SLE. The G allele frequency was 32% in cases, and 17% in controls. They confirm serum MCP-1 levels were 655 pg/ml for A/A, 3470 pg/ml for A/G, 5149 pg/ml for G/G in cases (being uniformly low in 4 controls assayed).

9.4 TBX21 and asthma

The T-box family of genes code for transcription factors, and T-box 21 (T-bet) specifically controls gamma-interferon production by Th1 cells. The mouse knockout spontaneously develops AHR. TBX21 is at 43.166 Mbp on 17q.

Chung et al [2003] undertook SNP discovery in TBX21. They confirmed a single coding SNP (rs2240017, H33Q), which was not significantly increased in asthma.

Table 35. TBX21 SNP frequencies in Korean asthmatics and controls
GroupNH33Q
AfrAm Control 500.020
Cauc Control 500.030
Korean Control1630.107
Korean Asthma 5220.115

Tantisira et al [2004] genotyped 701 children from the CAMP study at the coding SNP rs2240017 (allele frequency 4.5%). Only 5 of 139 Caucasian children in the corticosteroid treatment arm of CAMP carried the Q allele. Over a 4 year period, the treated Q carriers experienced a 3.5 fold greater improvement of AHR (Mann-Whitney P=0.003) compared to treated H/H homozygotes.

Akahoshi et al [2005] also resequenced TBX21, finding 5 novel SNPs. The H33Q allele was not significantly increased among Japanese asthmatics, but a -1993T>C variant was overrepresented in aspirin intolerant asthmatics (uncorrected P=0.004).

Table 36. TBX21 SNP frequencies in Japanese asthmatics and controls
GroupNH33Q-1993T>C
Japanese controls 6400.1140.102
Pediatric asthmatics 3610.1140.109
Atopic asthmatics 3130.1490.097
Nonatopic asthmatics 880.1170.064
Aspirin intolerant asthmatics 720.1470.181

9.5 ITGB3 and asthma

The beta-3 integrin gene (ITGB3) is located on 17q21 (42.7 Mbp from the pter). It is also commonly known as platelet glycoprotein IIIa, and is the beta-subunit of the platelet membrane adhesive receptor complex, the fibronectin receptor and the vitronectin receptor. Mutations lead to Glanzmann thrombasthenia (with impaired platelet aggregation and clot retraction), immune thrombocytopenia, and ischemic heart disease (notably L33P (Pl(A), rs5918) which has a 20% allele frequency in Caucasians).

Weiss et al [2005] carried out association analysis using 19 SNPs in ITGB3 in the Hutterites, CSGA Chicago families and the Childhood Onset of Asthma Study (parent-child triads). The best uncorrected P-value observed were for positive SPT to mould in the Hutterites (P=0.0005), but these noncoding SNPs were not strongly associated with this or other phenotypes in the other samples.


10 Other asthma/atopy loci

10.1 Brereton, Turner and Ruffin [1990]

This abstract [1990] reported early results of a linkage study of atopic asthma in two kindreds. The Jeffries 33.15 "fingerprint" probe was used, and a unique fingerprint element appeared to be linked to asthma and allergy (family 1 Zmax=1.3, recombination frequency=0.1; family 2 Zmax=0.9, recombination frequency=0.0). It should be noted that Cookson et al [1989] mention that they had examined this particular probe for linkage - presumably without success. Indeed, on the extension of this study [Brereton et al 1994] to a total of 12 extended pedigrees, the linkage, now localised to 8pter-p22, failed to be replicated in the other 11 families.

10.2 Walter et al [1990]

These authors [1990] examined 5 highly atopic subjects for DNA structural abnormalities around the immunoglobulin heavy chain gene cluster on 14q32.3. One subject was found to have a deletion of the region containing the IGHA1 through to IGHE genes. This was not present in the parents. Since the population rate of heterozygous deletions in this region is 3%, this can only explain a small proportion of cases of atopy, but may be a modifier of disease effect in these patients [Fujii et al 1995].

10.3 Moffat et al [1994]

Linkage between the T-cell receptor (TCR) alpha-delta subunit genes and atopy was reported by Moffat and coworkers [Moffat et al 1994]. Families containing at least two atopic siblings were ascertained in the Busselton study (413 subjects), as well as from the Oxford samples (410 subjects). Mean ibd sharing was increased for TCR alpha-subunit alleles in sib pairs concordant for several different phenotypes including high IgE level (P=0.002), house dust mite (P=2x10-4), cat (P=6x10-5) and timothy grass (P=0.02) allergen sensitization. No such linkage was seen for a TCR beta-subunit polymorphism.

A number of twin studies have looked at (peripheral blood) T-cell receptor gene usage. These show that MZ twins are correlated in the proportions of each gene family used, more strongly for the V-beta than V-alpha, and in CD4 rather than CD8 cells [Hawes et al 1993; Davey et al 1994]. These correlations are only partly mediated via HLA [Akolkar et al 1995; Troye-Blomberg et al 1997].

These twin studies have attempted to measure "baseline" usage, and have concluded that they are under strong genetic control. This means that it is difficult to invoke non-genetic differences in TCR gene usage as an discordance between MZ twins for atopy based on VDJ recombination mechanism.

One possibility is that usage in sensitised individuals are skewed by clonal proliferation, stimulated by exposure to allergen. In one set of MZ twins discordant for asthma, Davey and coworkers [1994] found no differences in VB usage in CD4 cells, but significant differences in CD8 cells persisting over a one year period. Another is that just as the same peptide can be presented successfully by many different Class II molecules, VB usage patterns may bear little relationship to the presence or intensity of the allergic response to even purified peptide.

11.3.1 11.3.1 Mansur et al [1999]

The Southhampton study [Mansur et al 1999] has also uncovered evidence of linkage to this region. The 60 families (355 individuals, 148 sib pairs) described here underwent SPT, histamine inhalation challenge and IgE determination. There was increased sharing (mean ibd=0.597) at the TCRA microsatellite marker for skin atopy (P=0.04), but not to specific allergens, total serum IgE level. The TDTs using TCRA were not significant. Multipoint Haseman-Elston regression for sIgE obtained a peak lod of 1.9 at D14S63

11.3.2 11.3.2 Soriano et al [2000]

Soriano et al [2000] performed a case-control analysis (see above) of association of asthma to a number of microsatellites, including one in TCRA. The allele frequencies at the TCRA microsatellite were significantly different (P=0.0015) in cases compared to controls. Such a difference was not found by Moffat et al [1994] or Mansur et al [1999].

10.4 Shirakawa et al [1998] on BCL6

BCL6 is a protooncogene (3q27) mutated in Hodgkin's disease. Knock-out mice seem to be skewed to a Th2 hyperresponsiveness, with eosinophilia etc. These authors performed a case-control study within their British panel. There was no significant association with asthma or atopy per se, but subgroups with severe atopy (RAST to HDM and grass mixture) gave a hint of a relationship (P=0.01, uncorrected).

10.5 Hobbs et al [1998] on IL10

The IL10 gene (1q31) codes for cytokine synthesis inhibitory factor, which acts to suppress inflammation. IL-10 levels are increased in rheumatoid arthritis, SLE, obesity and the metabolic syndrome, and psoriasis. Levels are under genetic control, notably by variants in the IL10 promoter. The promoter variants have been reported to be associated to alcoholic liver disease, HIV infection and AIDS,

Borish's group report here [Hobbs et al 1998] on an association study looking at variants in the IL10 (1q) and TGFB (19q) genes detected via SSCP/heteroduplex analysis. In 144 individuals, from 30 families ascertained through an asthmatic proband, a C>A polymorphism (21% frequency) at position 571 in the IL10 promotor, and four TGFB promotor polymorphisms (3 in TGFB1, and 1 in TGFB2; the frequency of the critical TGFB1 509 C>T was 38%) were recognised. There was severe Hardy-Weinberg disequilibrium for the IL10 polymorphism (and one TGFB polymorphism).

The authors then tested for association in a subsample of one individual per family, and supplementing this with 17 other subjects (with and without asthma). Both the IL10 571 C>A (P=0.009) and TGFB1 509 C>T (P=0.01) alleles were associated with increased total sIgE level, and seemed to interact in an additive fashion on the log scale (though the analysis was performed on the untransformed scale).

It would be nice to see the analysis using all the family data, looking for both confirmatory linkage and association to the two loci. The lack of HWE might reflect a heterogenous ethnic background for the families, genotyping probems, or perhaps an effect of ascertainment on a locus strongly associated with asthma.

Lyon et al [2004] report on TDT and GEE-regression analyses using 518 asthmatic children and their parents: 441 triads, 34 ASPs and 3 affected sib-trios from the CAMP (Genetics Ancillary) Study. There were no significant associations with bronchodilator responsiveness. A 3'UTR SNP 4299T>C was strongly associated with the post-BD FEV1 observed/predicted (P=0.0002) and sIgE level (P=0.005), as were three promoter polymorphisms to a lesser extent. There were similar less impressive haplotype associations.

10.6 Malerba et al [1999] on chromosome 14

This Italian study (presumably a subsample of that described in Pignatti et al 1998, see above) obtained weak evidence of linkage of BHR to D14S617, close to PI and AACT using 116 families ascertained through an affected sib pair. They did not detect linkage or association to PI, but weakly positive TDTs were obtained for the AACT signal peptide polymorphism T-15A: best P-value 0.014 for total sIgE level.

10.7 FCER2 and atopy

The CD23 gene FCER2 (19p13.3) was chosen by Laitinen et al [2000] as a potential asthma/atopy candidate for association and linkage analysis using Finnish and Catalan population samples -- populations chosen to maximize genetic trait homogeneity and linkage disequilibrium between putative trait loci and neighbouring markers.

Using 124 pedigrees (488 subjects) segregating asthma, high total or specific sIgE level, linkage was excluded to the 19p region for all 3 phenotypes. However, a particular three marker haplotype spanning approximately 5 Mb across the region containing FCER2 was found to be associated with elevated total sIgE level (permutation test based P=0.04). In an analysis of haplotypes seen only when transmitted from affected parents to affected children, this effect was even more evident.

When 4 subjects carrying this haplotype were sequenced at FCER2, no coding variants were detected. The authors suggest that there may be an atopy locus in this region other than FCER2 -- obviously this could be in the promotor region or nearby regulatory gene.

10.8 TLR4 and bronchial responsiveness

A number of studies have implicated inhalation of endotoxin (lipopolysaccharides, LPS) with the development or exacerbation of asthma, especially occupational asthma. Bronchial responsiveness to LPS is correlated to nonspecific (histamine) BHR, but possibly not to atopy [Michel et al 1992]. This is complicated by the observation of an inverse association between asthma (but not atopy) and levels of endotoxin in house dust from the home, for example in the ISAAC 2 study [Gehring et al 2008].

Kline et al [1999] examined bronchial responsiveness to endotoxin in a sample of 72 healthy volunteers and confirmed individual differences, with 8 subjects developing an early 20% drop in FEV1 during the challenge protocol, and a total of 61 by the end of the protocol (cumulative dose 41 mcg of LPS). In followup papers, [Arbour et al 2000; Schwartz et al 2001], they tried to explain hypersensitivity in terms of variants in the TLR4 gene (9q32). The toll-like receptor-4 is the receptor through which LPS initiate their inflammatory effects, and disruption has been shown in mice to lead to hyporesponsiveness.

Two common variants, D299G (G/G 1, D/G 9, D/D 73) and T399I (in complete disequilbrium with the D299G variant), were found to be associated with decreased response to endotoxin (D299G in 3/52 responders and 7/31 nonresponders, P1-tail=0.029). In vitro functional studies showed that IL-1a secretion and NFkB activity were decreased by R299G, and transfection with the wild type gene lead to a return to normal activity of cultured respiratory epithelial cells and macrophages. In two population panels the D299G allele frequency was 6.6% (Iowans) and 3.3% (CEPH founders).

Raby et al [2002] screened TLR4 for additional SNPs. They identified 17 novel SNPs, and 12 previously known SNPs. There was no association between 5 common SNPs (notably D259G and T359I) and asthma or related phenotypes in 756 families (Quebec and US), .

Kiechl et al [2002] studied 810 Italians. The D299G allele was carried by 55 subjects (frequency 3.4%), and was associated with decreased carotid atheroma, and increased "severe bacterial infection".

Schmitt et al [2002] developed assays for several TLR4 SNPs. They confirmed monocytes from D299G and T399I homozygotes exhibit a decreased LPS response (fourfold higher EC50). The allele frequencies in 115 healthy Germans were: D299G 5.6%, T399I 6.0%.

Eder et al [2004] looked at atopic disease and TLR4 and TLR3 in rural children. Only a TLR2 SNP was associated with asthma (P=0.004), and only if the proband lived on a farm.

Berghofer et al [2005} looked at 5 SNPs in TLR9 (3p21, important in recognition/adjuvant action of CpG DNA, and presumably stimulating T-bet levels, TBX21 see below). Neither total sIgE or allergen specific sIgE level was found to be associated with TLR9 haplotypes in 303 individuals. Importantly, the haplotypes did not predict stimulated dendritic cell IFN-alpha production in a subset of 220 samples.

10.9 CCR5 delta32 and asthma

Hall et al [1999] suggested that the CCR5 32 bp deletion polymorphism (Caucasian population frequency 10%) might be protective against asthma. In 98 asthmatic children the deletion frequency was 5.6%, and in 317 nonasthmatics, 14.0% (OR=0.36, 0.18-0.66).

This hypothesis was tested by Sprankle et al [2000] using 286 nuclear asthma families and a case-control panel of 200 asthmatics and 100 controls. No association was found via TDT or case-control analysis. At the same ASHG meeting, Romano-Spica et al [2000] reported a case-control study of children with asthma (19) and recurrent lower respiratory tract infection (33). No significant differences at CCR5del32 were found, but the CCR2 264I allele was increased in the cases (asthma 24%, LRTI 20%, versus 7% in the 130 controls). Similarly, Mitchell et al [2000], in the Busselton and Oxford panels, found neither linkage or association of the polymorphism with atopy. The overall mutant allele frequency was 11% (five of the Busselton parents were homozygotes), and the odds ratios for atopy (1.02, 95%CI=0.69-1.50), and asthma (0.88, 0.58-1.36), excluded any large effects.

McGinnis et al [2002] examined CCR5del32 frequencies in English and German asthma families. The TDT did not find significant evidence of an effect, but a analysis comparing asthmatic and nonasthmatic parents found the del32 frequency decreased in the cases (English, P=0.009; German, P=0.265).

10.10 IL1RN and asthma

Gohlke et al [2004] looked at SNPs in the genes of the IL1 gene cluster (2q12) using German and Italian families. Haplotypes involving the IL1RN gene (that encodes the IL1 receptor antagonist were associated with asthma (best P=0.002).

10.11 KCNS3 and asthma

Hao et al [2005] describe association fine mapping using extreme cases and controls for AHR from the Anqing asthma genetics sample (see below). From under a linkage peak on 2p, they selected KCNS3 (a potassium channel gene) as a candidate, and successfully genotyped at 3 SNPs. There was weak evidence for association (permutation P=0.006 for best SNP, rs1031771).


11 Genome scans for asthma and atopy

11.1 Daniels et al [1996]

The first genome scan to be published [1996] was of two subsamples from the Busselton (80 families, 364 subjects) and Oxford (77 families) study families, with the Oxford families being used to replicate any positive findings in the first panel. The chromosome 11 findings were of course replicated. Different regions of the genome gave positive results for differing phenotypes: BHR was linked to chromsomes 4, 7 and 16; tIgE to chromosomes 6, 7, and 16; eosinophil count to 6 and 7; atopy (combining skin tests and IgE) to 6 and 13. Replication in the second panel was of atopy to chromosomes 13, and of asthma to chromosome 16. Interestingly, maternal linkage was stronger than paternal linkage for, in addition to FCER1B, markers on chromosome 4 and 16.

11.2 CSGA

This is the first genome scan specifically targetting asthma [CSGA 1997]. The Collaborative Study for the Genetics of Asthma is a multicentre study funded by the NHLBI and NIAID. This first paper reported on 140 families were ascertained to contain an affected sib-pair where both suffered "definite" asthma (current symptoms, consistent history or medical diagnosis, and BHR to methacholine). These were recruited from seven geographical locations in the US, selected to represent populations of Caucasian, Hispanic and African descent. As noted earlier, black Americans have higher rates of asthma than whites.

A multipoint affected-sib pair analysis was performed within each ethnic subgroup. The highest overall Maximum Lod Score (MLS) was 2.30, for 2q33 in Hispanics. Interestingly, there was little replication across the different ethnic groups, though the size of each stratum was relatively small (see Table).

Table 37. Chromosomal regions suggestive of linkage to asthma in the CSGA genome scan [1997].
African-AmericanCausasianHispanic
Number Affecteds12520046
Number SPT Positive70%75%87%
Geometric Mean sIgE (IU/ml)275149277
Regions with MLS>15p15,17p11-q11(5q)*,6p,11p15,12q14-24,14p,19q132q33,21q21

* MLS=0.9.

In a meeting abstract, Cox et al [1998] looked for correlations in nonparametric linkage score (GENEHUNTER NPL score) between different regions, arguing these will only occur in the presence of epistasis. This analysis suggested 1p, 2q, 10q as likely "true" positives.

Colilla et al [2003] describe an analysis stratifying on exposure to tobacco smoke. The sample now includes an additional 15 Minnesotan families. There are no striking differences when compared to the original scan -- for example, the linkage lod for asthma to 5q was 0.34 overall, but 1.23 in ETS exposed asthmatics. Overall, 6 regions showed an increase in linkage evidence (four significant increases at the 5% level on pointwise permutation testing).

11.3 Ober et al [1998]

There are several papers describing the subgroups studied within CSGA. Ober et al [1998] describes genome scan results specific to the Hutterites. The population studied comprises a primary panel of 361 subjects (246 "definite" non-asthmatic, 50 "definite" asthmatics using the CSGA criteria) recruited from the Schmiedeleut group of Hutterite colonies. These individuals were embedded within a 1101 pedigree spanning six generations back to 64 founders (mean coefficient of inbreeding F=0.0327). The second panel comprised an additional 292 individuals (177 non-asthmatics, 26 asthmatics).

Testing for linkage was performed by maximizing over linkage models (genetic models and strictness of diagnosis) as in SPLINK. There were 12/292 (4.1%) autosomal markers in the primary analysis meeting the threshold of P=0.01 (best five: D19S178, D9S925, D3S2432, D3S1768, D21S1262). Of these twelve, four gave a TDT P-value of better than 0.10 (presumably best allele versus all others). In the replication sample, a TDT for that "best" allele in the primary sample was significant at the 5% level in only one case (D12S375, P=0.033).

Ober et al [2000] gives more information about an extended scan (N=693) of the Hutterites. A web table gives marker by marker results for linkage (SPLINK) for asthma and/or BHR: peaks include 1p (D1S648), 5 (D5S1462)

Ober et al [2001] describe a genome scan for allelic association between STR markers and eosinophil count and total sIgE level. Two methods based marker homozygosity were tested. The best associations (D1S547, D2S1790 and D3S1764, D5S1505 respectively) were associated with genome-wide corrected P-values greater than 0.5.

A fine mapping study of linkage peaks on 6p [Nicolae et al 2005] is discussed below in the section on HLA.

11.4 Mathias et al [2001]

This paper describes a variance components QTL linkage analysis of sIgE level in the CSGA families. The best lod score in the different ethnic samples was 1.80 for D4S2431 and age-sex adjusted log sIgE level in the African American group. When Allergy Index was added as a covariate however, a peak of 2.75 appeared in the Caucasian group near D18S844.

11.5 Blumenthal et al [2004]

Here, the CSGA families' genome scan was utilised for linkage to D. Pteronyssinus (and D. farinae) sensitization. The best overall lod of 2.4 was on chromosome 19 (D19S591, D19S1037), with a lesser peak on chromosome 20 (1.4 for D20S473 and D20S604). The African-American families gave independent linkage for D12S373 (lod 1.5), and Hispanic families D11S1984 (lod 1.6) and D13S787 (lod 1.3).

11.6 Koppelman et al [1998]

An extended sample of 141 families from the Groningen asthma study (nuclear families ascertained via an affected parent) underwent a genome scan (Weber set 8). This confirmed the linkage to 5q31. Total sIgE level was linked to Chromosome 12, but not 11 or 13. A multipoint lod of 2.3 was obtained for BHR (as a dichotomous trait).

11.7 Dizier et al [2000]

The French EGEA study included a genome scan of 107 nuclear families containing two or more asthmatic siblings. Linkage of total sIgE was strongest to 11p13 (P=0.005), asthma to 17q12 (P=0.001) and eosinophil count to 12q24 (P=0.0008). In addition, replication of several previous findings were found: 11q13 to total sIgE (P=0.006), asthma to 1p (P=0.005), and eosinophil count to 13q (P=0.001).

11.8 Wjst et al [1999]

The German Asthma Genetics Group carried out a 10 cM resolution scan of 97 nuclear families ascertained via hospital outpatient clinics to contain at least two siblings with asthma (200 affected children, median age 10 years).

The strongest evidence for linkage to asthma was found for markers in the 2pter, 6p21 and 9q chromosomal regions (best lod approximately 1.5, back transforming P-values). These were also linked to total and specific serum IgE level, and in the case of 6p to eosinophil count as well (lod 2.6 for D6S426, approximately 2 Mb distal to TNF). There was a scattering of weak linkages for other traits, but none of these exceeded a lod of 1.5.

Wjst et al [2000] and Immervoll et al [2001] describe followup fine mapping (50 extra markers). The eosinophilia linkage to 6p was strongest now for D6S1641 (P=5x10-5, equivalent lod score 3.55). Of 6 candidate SNPs, 4 gave positive results, the best being a 3391C>T variant in NOS1 (P=0.009).

Altmuller et al [2005] expanded this sample to a total of 201 families ascertained via an asthmatic sib pair (N=867), containing a total of 506 asthmatics. Best lod scores were on chromosomes 1q (lod 3.3), 7, 11p and 11q for high tIgE; on 12q for HDM sensitization. Asthma did not exceed a lod of 1.9 anywhere: the best P-value on chromosome 6 was only 0.01. The authors do not present eosinophil count results in this paper.

11.9 Pilia et al [1999]

This (ASHG) abstract describes a scan of 8 candidate regions using 100 asthma affected sib pairs from 80 Sardinian families . The best lod score was 3.1 (single point) for D13S164, but weakly positive results for 4q, 5q, 11p were also obtained.

11.10 Sequana Therapeutics

This gene discovery company carried out a 14 cM resolution genome scan of the population of the island of Tristan de Cunha [Zamel 1995; Zamel et al 1996; Brooks-Wilson et al 2000]. This population was known to have high rates of asthma (a prevalence of 45% in the years 1942-44 [Wooley 1963], and 23% in the present sample [Brooks-Wilson et al 2000]), and this was thought due either to a strong founder effect, endemic parasitosis, or the humid climate. Most of the population are related through 29 founder colonists in the 19th century, with inbreeding to the level of first cousins.

The best peak out of 274 markers was for D11S907: a P-value of 10-4 for asthma. I will describe the subsequent work on 11p below.

11.11 Xu et al [2000]

The Groningen study in this paper has been extended to a total of 200 families (1171 individuals), including 66 three generation pedigrees. A variance components linkage scan (Weber Version 8 set of markers) using SOLAR was performed. The biggest peak was around D7S554 (a lod of 3.65, and a drop-one-lod interval of 30 cM), but other peaks were around D2S1391, D3S1311, D5S1505, D6S2427 and PAH.

11.12 Yokouchi et al [2000]

The Tsukuba group [Noguchi et al 1997; Kimura et al 1999] performed a genome scan of their families with mite-sensitive asthma (47 families containing 65 affected sib pairs). Their best peak (MLS of 4.8) was obtained for 5q close to D5S280 and IL12B. There were suggestive peaks on 13p (MLS 2.8) and 4q (MLS 2.7).

11.13 A meta-analysis of genome scans [Becker et al 1998]

This interesting paper looks at clustering of "hits" in genome scans for atopy, IDDM, inflammatory bowel disease, psoriasis and multiple sclerosis, comparing them to those obtained for control non-immune conditions such as hypertension and NIDDM. Assessment of the reality of any clusters was performed by randomization tests.

They finished with a list of 18 clusters. Those including currently described asthma were on 4q, 7p, 11p and 11q, 12q, 13q, 16q, and 19q. A similar analysis using solely IDDM as the target trait, looking for matches to atopy, also finds significant overlap.

11.14 A genome scan for atopic dermatitis [Lee et al 2000]

A total of 199 nuclear families (839 individuals) containing at least two children suffering from moderate to severe atopic dermatitis were recruited through seven European centres to take part in this study. The probands' mean age was 7.5 years, and 24% were asthmatic.

Genehunter's Zall statistic and genome-wide simulation-based empirical P-values were used. It should be noted that uniform marker allele frequencies were assumed. Only the 3q21 region stood out, with an empirical P-value of 0.0009 (Zall=4.3). Subsequent lod score linkage analysis (PD=0.01, Pen(DD)=Pen(Dd)=0.9, Pen(dd)=0.001) obtained a HLOD=3.65 (39% families linked). Restricting analysis to "unlinked" families uncovered two more peaks on 1p (D1S2876, Allegro's Zlr=2.7) and 19p (D19S209, Zlr=2.7).

Evidence of linkage of the 3q21 region to atopy and high total sIgE was weak unless a paternal imprinting model was fitted (HLOD=3.7, genome-wide empirical P=0.016). However, no such effect for atopic dermatitis was detected.

11.15 A genome scan for asthma [Xu et al 2001]

Anqing (China) is the site of a large genetic epidemiological study of asthma and lung diseases. Over 2000 families containing two or more doctor-diagnosed asthmatics have been ascertained, and 7100 of the 10000 subjects have undergone methacholine inhalation challenge.

11.16 A genome scan for atopic dermatitis [Cookson et al 2001]

In this study 148 nuclear families (383 children) were ascertained through one or more children with atopic dermatitis attending a dermatology clinic. Analysis was via affecteds-only and affected-unaffected sib pair methods. Linkage peaks for AD were found around D17S784 (ASPs only, lod=2.4) and D1S498 (analysing affected and unaffecteds with equal weighting, lod=2.4). Total sIgE was linked to D16S520 (lod=2.2), and asthma to D20S115 (lod=2.3). The authors were impressed by the fact that these peaks are close to previously reported linkage regions for psoriasis, but it is I think disappointing that the chromosome 1 AD peak has a peak lod of 0.75 for tIgE (and none shown for chromosome 20), given the high levels of tIgE seen in AD.

11.17 An association study of 24 candidate genes [Hakonarson et al 2001]

Results from this paper are also described under the heading of various candidate genes. These authors (from DeCODE Genetics) studied extended families of patients attending outpatient allergy and respiratory clinics at a university hospital. A total of 94 families (269 atopic asthmatics) were selected and compared to 94 unrelated controls. The 269 cases and 230 unaffected relatives were genotyped at STR markers in 12 candidate regions (4 cM resolution scan), and genotyping at 42 SNPs within 24 candidate genes in these regions was performed on 94 index cases and the 94 controls.

No evidence of allelic association was found for any of the 42 SNPs. Furthermore, no evidence for linkage to any of the regions was found.

11.18 Hakonarson et al [2002]

DeCODE Genetics also performed a linkage genome scan. Index cases were a subset of 7000 patients attending National University Hospital who were related to at least one other asthma patient "within six meiotic events", as recorded in the Iceland wide genealogy database. A total of 596 asthmatics and 538 nonasthmatic relatives in 174 families were genotyped at 976 markers (3-4 cM resolution scan).

Linkage peaks for asthma were found on Chromosomes 14 (NPL SPairs lod=2.7), and 10 (lod=2.1). On increasing the density of markers around the Chromosome 14 peak, the peak increased to 4.0, maximal around D14S588 and D14S603. The "drop 1 lod" region was only 4 cM wide.

Other peaks were on 3p, 11q, and for glucocorticoid resistant asthma on 18q (lod 2.5).

11.19 Speer et al [2002]

The Genetics of Asthma International Network is another multicentre collaborative study: eleven groups each contributing at least 100 ASP families, including a West Australian group (Peter Sly) and Glaxo-Smith-Kline. Speer et al [2002] in an ASHG abstract report an interim (7 cM resolution) scan of 389 families. The best lod score was 2.8, linking total serum IgE to the 20 cM wide region D10S1221--D10S1432. Other areas with lod scores above 2.0 included near D8S1136, D15S87 and chromosome 21.

11.20 Haagerup et al [2002, 2004]

Haagerup and coworkers genome scanned 100 Danish atopic nuclear families (424 individuals). These contained 39 "allergic asthmatic" affected sib pairs and 57 high IgE ASPs. In view of the sample size, it is not surprising that no lod score exceeded 3.0. In subsequent fine mapping of 11 regions [Haagerup et al 2004], the best lod was 2.69 on chromosome 3q. Lods over 2 for 6p and 12qter may be regarded as replications of existing findings.

Brasch-Andersen et al [2006] extends this sample to a total of 167 Danish asthma families. The phenotyping in the additional families included a methacholine inhalation challenge test. These were genotyped at a panel of 12q24 markers. There was consistent evidence for linkage to this region in the followup families, with the best MLS=3.27 between D12S63 and D12S392 from the combined sample.

SFRS8 was selected as the best candidate gene under the peak, and 8 SNPS through the gene were tested for association, with the best P-value of 0.018 (using FBAT).

11.21 Malhotra et al [2003]

These authors have studied lung function in 26 of the Utah CEPH pedigrees (264 individuals). These families have been extensively genotyped, and this paper uses 1324 markers. Preliminary segregation analyses failed to find evidence of major loci for FEV1 or FVC, but a single major locus explaining 40% of the phenotypic variance was supported for FEV1/FVC ratio.

Variance components and parametric linkage analyses (Genehunter) were performed for FEV1/FVC. Peak parametric Hlods of 2.3 and 2.2 were observed on 2q (a drop-1 lod region from 216-251 cM) and 5q (with a two-point lod of 2.64 for D5S422, see above) -- the VC linkage lods for the same locations were 1.1.

11.22 Tamari et al [2004]

This ASHG meeting poster describes a Genome Wide Association Study of 94 asthmatics and 658 controls, followed by replication in a second panel of 940 asthmatics and 658 controls. A PCR-Invader assay was used to genotype 86100 SNPs. A set of 17 SNPs out of 2026 selected in the first round were ultimately judged to be significantly associated, best SNP association lod 4.1 (Risch and Merikangis critical lod is 5.5).

11.23 Kurz et al [2000, 2005]

These authors report on a repeat genome scan (density increased to 6 cM) of 82 nuclear families (N=366) containing two or affected offspring (17 families lacked sufficient DNA). Subjects were from Germany, Portugal and the British Isles. The highest new lods (MOD score analysis) were 3.70 for D16S2620, 3.57 at D3S3606, and 3.04 for D2S1777. It should be noted that the NPL Z scores were 1.31, 1.52 and 2.15 respectively for those three markers.


12 Genome wide association scans for asthma and atopy

12.1 Moffatt et al [2007]

This is the first published Genome Wide Association Scan (GWAS) for asthma. Genotyping of 994 UK and German asthmatic children and 1243 controls was carried out on the Illumina Sentrix HumanHap300 BeadChip (giving 307 000 usable SNPs) and for the UK subset of sib pairs, on the Illumina Sentrix Human-1 (an additional 91000 gene centred SNPs).

There were 16 SNPs associated to asthma where P-values of less than 5 x 10-6 (5% false discovery rate) were obtained. Eight SNPs were within a 200 kbp region of chromosome 17q21, and the remainder scattered across chromosome 1, 2, 8, 10, 21, and the X.

The chromosome 17 SNPs spanned the ORMDL3 gene. A set of four most strongly associated SNPs was selected by stepwide logistic regression: rs7216389 (intron 7), rs11650680, rs3859192 (intron 6) and rs486512 (in next gene STAC2), but other SNPs such as rs2305479, rs9303281, rs7219923, some of which are nonsynonymous coding SNPs, and exhibit fairly similar strength of association. All these SNPs were common, and at similar frequencies in the African and Asian HapMap samples.

When transcript levels of ORMDL3 were tested for association to SNP genotype, rs7216389 and other asthma associated SNPs were found to account for 30% of the variance.

The association was able to be replicated in several datasets. One marker, rs3894194, was also available for the UK 1958 Birth Cohort, where it was shown to be associated with asthma risk.

It is possible to obtain results for a subset of the 1958 Birth Cohort for quantitative traits such as total serum IgE level and SNPs in this region [http://www.b58cgene.sgul.ac.uk] . This shows no effects on serum IgE level or FEV1 level.

12.2 Galanter et al [2008]

These authors tested 7 SNPs in ORMDL3 using Mexican and Puerto Rican parent-offspring triads (701 families), and African-American cases and controls (264 cases, 176 controls). The best results were for rs12603332 (P=0.001 in African-Americans, P=0.02 in Mexicans, and 0.080 in Puerto-Ricans).

12.3 Bisgaard et al [2008]

Association between ORMDL3 and asthma was tested for in the the Danish COPSAC study, a longitudinal childhood asthma study. There were 376 children, and a lesser number of parents genotyped. Significant association between rs7216389 and exacerbations of asthma was observed (P=0.0002). The same association was seen in parents, and in a TDT involving offspring the relative risks for various asthma phenotypes was 2.1-2.5 in TT homozygotes.

12.4 Sleiman et al [2008]

Several of the authors from the previous paper appear here as well in this letter, that described results from two case-control panels: 807 Philapdelphia white asthmatics, 2583 controls; and 1456 African-American cases, 1973 controls. The best P-value was 0.004 (rs8067378) seen in the white sample, with a similar level of association to rs7216389 (OR=1.18, P=0.007). In the African-American sample, none of the SNPs exhibited significant association (best P=0.27).

12.5 Himes et al [2009]

The CAMP study provided 359 cases, who were compared to 946 controls from the Illumina ICONdb panel. Strongest association was to chromosome 5q12, with the best SNP rs1588265 P=4.3e-7. Replication studies were carried out in 10 populations, giving a Caucasian combined P=4.1e-4, but no association in black populations. The associated SNPs lie in PDE4D, a known regulator of airway smooth muscle contractility

SNPs in PDE4D have been previously associated with risk of stroke, although a recent metaanalysis concluded effects are small [Bevan et al 2008].

Table 38. Frequencies of rs1588265 across studies of asthmatics and nonasthmatics.
Study rs1588265*G frequency (N)
Asthmatics Nonasthmatics
CAMP 0.23 (359) 0.34 (846)
1958 UK birth cohort 0.30 (821) 0.31 (2469)
CHOP (Philadelphia) 0.28 (569) 0.32 (2136)
CHOP3 0.31 (365) 0.32 (690)
Childrens Health Study0.28 (769) 0.30 (1007)
Framingham 0.26 (961) 0.29 (6516)
MRC Asthma cohort 0.29 (265) 0.32 (547)
Costa Rica 0.32 (353) 0.33 (355)
CHOP2 0.22 (1456) 0.22 (1973)
GRAAD1 Barbados 0.20 (447) 0.21 (459)
GRAAD2 Barbados 0.20 (77) 0.23 (88)
HapMap Yoruba 0.15 (60)
HapMap Asia 0.71 (90)

12.6 Hancock et al [2009]

The HumanHap 550K chip was used to genotype 492 Mexican parent-child trios in this study of asthma in children (mean age 9 years). The best result was for rs1867612 on chromosome 16 (P=1.6e-6). The best 11 SNPs were genotyped in 177 trios from the GALA study, with the strongest results being for rs2378383 (combined P=6.8e-7) near TLE4 on on chromosome 9q21. This SNP was not associated with asthma in the (Caucasian) GWAS of Himes et al [2009]. Since this SNP is less common in European populations, and the minor allele is protective, the authors argued power to replicate this association would be low in most studies.


13 HLA and atopy

Two types of relationship between the HLA region and atopy have been explored. The first is that asthma and/or atopy (ie generalised IgE hyperresponsiveness) might be controlled by a major gene in this region. I find the evidence in favour of this hypothesis largely unconvincing. The second possibility is the existence of genes controlling the response to particular allergens - immune response genes. This does seem to be the case.

13.1 Turton [1979] and others

A number of studies were undertaken in the 1970s to find any HLA association with asthma and atopic disease. Any associations with the exception of the specific one between ragweed pollinosis and HLA-DR2 (see below), have been weak and contradictory. Those studies prior to 1977 are summarised by De Weck et al [1977]. One of the largest study of this question is that of Turton et al [1979].

This group examined HLA haplotype in 122 asthmatics recruited through Brompton Hospital, and 10 families (N=68 individuals) where more than one member suffered from asthma. The asthmatics were divided into skin test and allergy history negative (N=41), extrinsic atopic asthmatics (N=40), and those with allergic bronchopulmonary aspergillosis (ABPA, N=41). All patients underwent skin prick testing for 23 common allergens, chest X-ray, spirometry with bronchodilator test, total sIGE level determination and HLA typing. Multiple X2 tests were performed on the HLA data.

The frequencies of the various HLA types among the asthmatics did not differ significantly from population control values save in four cases: HLA-B8 in intrinsic asthmatics (versus controls RR=2.7), HLA-B15 and extrinsic asthmatics (RR=3.6), a dearth of HLA-B7 in ABPA (RR=0.36), and HLA-A2 in subjects with high sIgE versus low sIgE (RR=3.5). The authors were cautious about interpreting these results in view of the large number of comparisons performed. Within the family data, there was no association between HLA haplotypes and asthma, allergic disease or elevated sIgE level.

The authors concluded that no biologically important associations between HLA type and asthma/allergy exist, and suggested that most of the previous reported associations were due to chance compounded by the usually large number of statistical comparisons made.

13.2 Caraballo et al [1990]

This group [1990] examined HLA segregation in twenty Columbian nuclear families (N=107) ascertained to contain two atopic asthmatic probands. All probands were allergic to D. farinae on SPT. No associations between HLA haplotypes and asthma were detected. In sib-pair analyses, there was evidence of linkage to asthma (sharing two haplotypes: 14; one: 5; nil: 1; X2(2 df)=21.9). This finding is most consistent with an autosomal recessive major gene (on chromosome 6).

13.3 Blumenthal et al [1992]

The results of this study [1992] may bolster that just discussed. A sample of 128 ragweed atopic individuals along with 52 relatives and 28 spouse controls were HLA haplotyped, and had total and Amb a V specific sIgE levels determined. One extended haplotype (HLA-B7, SC31, DR2) was increased in asthmatics sensitised to Amb a V, but not in ragweed sensitised rhinitics.

13.4 Marsh and others on allergen specificity

As alluded to earlier, this group has reported a HLA association specific to ragweed pollinosis. These findings were extended in a paper by Zwollo, Ansari and Marsh [1989], where the results of a linkage study using full length HLA-DR-beta, DQ-beta and DQ-alpha cDNA strands were reported. Unrelated subjects were recruited via an allergy clinic (N=16 individuals) and from the previously discussed Westinghouse Electric Corporation study (N=29), demonstrated skin atopy to short ragweed and had undergone immunisation with the ragweed allergen Amb a V (formerly Ra5) as part of another study. Subjects were divided into groups:

  1. DR2+ Dw2+ pre immunisation anti-Amb a V IgE and IgG antibodies (N=8)
  2. DR2+ Dw2+ post immunisation anti-Amb a V IgG antibodies only (N=6)
  3. DR2+ Dw2- no anti-Amb a V antibodies (N=5)
  4. DR2- Dw2- no anti-Amb a V antibodies (N=18)
  5. DR2+ Dw2+ preimmunisation anti-Amb a V antibodies (N=1)
  6. DR2+ Dw2+ no anti-Amb a V antibodies (N=1)
  7. DR2- Dw2- preimmunisation anti-Amb a V antibodies (N=2).
Hypersensitivity specific to Amb a V occurred only in those subjects possessing the DR2/Dw2 phenotype with 2 exceptions (group VII). After a screening procedure (pooling DNA from Groups I-III and Group IV), the DRChi associated RFLPs arising after digestion with EcoR1 were examined. A 6.5 kB fragment was present in all DR2+ subjects, that is 88% of the Amb a V sensitised subjects, and 25% of the nonsensitised subjects. Similarly, a 2.3 kB RFLP was found for the DQChi probe after EcoR1 restriction that was present in 82% of the sensitised patients and 21% of the nonsensitised patients. Subjects carrying both associated polymorphisms made up 14 of the 17 sensitised patients and 1 of the 24 nonsensitised subjects (Fisher exact test p2 tail=5x10-7).

In the discussion, the authors stated that these results confirm the already strong association between sensitisation to Amb a V and HLA phenotype, but did not shed any light on the responders who did not exhibit this HLA type. They draw an interesting comparison between the allergy associated 2.3 kB DQChi EcoR1 RFLP they have discovered, and the 2.2 kB (later changed to 2.75 kB) fragment reported to be negatively associated with insulin dependent diabetes mellitus. A negative correlation between asthma and diabetes has been reported repeatedly over 50 years [Helander 1958].

Marsh et al [1991] also describe work on Amb a VI and HLA-DR5. In 77 ragweed sensitive subjects, significant associations were found between polymorphisms at the variable second exon of DRB and DQB. Homologies in the allelic polymorphic 4 region of Dr5 and Dw21/22, both associated with antiAmb a VI IgE, suggested this region to be important.

13.5 Perichon et al [1991]

Another abstract presented at the 1990 meeting of the American Society of Allergy and Clinical Immunology [1991] describes investigation of the weak association between aspirin sensitivity in asthma and the HLA-DQw2 allele. They found a significant excess of a specific DQ heterodimer (0201/0301), which has been previously associated with coeliac disease.

13.6 O'Hehir et al [1991] and HDM

This study [described in O'Hehir et al 1991] claims an association between DRw12 (DR5) and restricted T-cell recognition of HDM. Other studies of this association have been more equivocal [O'Hehir et al 1988].

13.7 Young et al [1992] and HDM

The Oxford group (11q) have also performed segregation analysis of HDM allergy, following negative results of sib-pair HLA studies of this trait [1992]. They examined sixty nuclear families ascertained via asthmatic/atopic child probands as described above. Atopy was defined using skin prick tests (2mm > negative control) and RAST. No evidence for a major gene was discovered, and "as much as a third of HDM sensitive children in these families had parents who were not sensitive to HDM".

13.8 Amelung et al [1994] and HLA

The Groningen study also examined a HLA-DR marker D6S105 for linkage with atopy. A LOD<-2 for theta<0.13 was found.

13.9 Matsushita et al [1987] and cedar pollinosis

These authors [1987] examined allergic rhinitis caused by allergy to Cryptomeria japonica, which has a prevalence in the Japanese population of 10-15%. An autosomal dominant genetic model was found to fit their data, the gene being in negative linkage disequilibrium with HLA-DQw3. By contrast, a similar study in Japanese subjects with allergic rhinitis by Sadanaga and Ishikawa [unpublished, reported in O'Hehir et al 1991] suggested an autosomal recessive pattern, with an increased frequency of DQw3.

13.10 Dekker et al, HLA and aspirin sensitive asthma

Three groups of subjects were used: healthy controls, aspirin-sensitive and insensitive asthmatics. While the non-aspirin-sensitive asthmatics did not differ significantly from the healthy controls in DPB1 allele freqencies (hybrid Fisher P=0.11), the aspirin-sensitive group differed significantly from both other groups (full table hybrid Fisher P=0.0003, see Table). The differences for HLA-DRB1 were not significant.

Table 39. DPB1 allele frequencies among asthmatics and controls from Dekker et al [1997].
DPB1_alleleAS AsthmaNon-AS AsthmaControls
*01011294
*020118199
*03012355
*0401345247
*0402141912
*0501100
*0601032
Other16717
Total11811496

13.11 HLA and soybean epidemic asthma

Soriano et al [1997] report a study of 78 soybean epidemic asthma sufferers, 67 nonepidemic cases and 336 controls genotyped at HLADR and HLADQ. There was a significant difference between the three groups in frequency of different DR alleles (hybrid P=0.008), which was largely driven by an excess of DRB1*05 alleles (particularly DRB1*11) among the soybean sensitised asthmatics (19% v 9%). There was a decrease in DRB1*06 alleles in nonepidemic asthma compared to the other two groups, but the authors concentrated on the finding that the 05/05, 05/06, and 06/06 genotypes were found only in the epidemic group. No DQ differences were significant.

13.12 Stephan et al 1999

Linkage and association of specific IgE to the HLA region was tested in 64 families from the panel described by Kuehr et al [1993] containing one or more allergic members. Interestingly, although there was no association of particular DRB1, DQB1 or DPB1 alleles to sensitisation, there was strong evidence of linkage from single-point affected sib pair analyses, the best lod of 4.9 being between (Magic Lite) Der p and DPB obtained from only 12 pairs of sibs (!); grass pollen gave a lod of 4.2 from 26 concordant pairs. The Der p linkage evidence was driven by 11 sib pairs from 8 families where the DPB*0401 allele was segregating. A multiallelic TDT for Der p and DPB was weakly confirmatory (P=0.04); the TDT for *0401 versus all other allels was not reported.

13.13 Nicolae et al 2005 and HLA-G

This paper starts describing fine mapping of an asthma linkage peak around D6S1281 in the white Chicagoan families (lod 3.6) from the CSGA scan. The overall lod for Caucasians at this marker was 1.9. Of five additional STR markers, MOGc (close to HLA-F) was most strongly linked, as well as exhibiting allelic association in a TDT analysis.

Therefore, an additional 59 SNPs spanning 1 Mbp in the region were also tested via combined association-linkage analysis. Several SNPs in the HLA-G gene could independently explain away much of the linkage signal (from 3.8 to 0.9), but further resolution could not be made in that sample. The best individual association result was for rs4713210, but several pairs of markers gave much stronger evidence.

Replication was sought in a collection of parent-child triads recruited in Chicago, that group's Hutterite pedigree, as well as in the Groningen pedigrees. In the Hutterites, 3 HLA-G SNPs were associated, including a -964A>G transversion (rs1632947, P=0.046). This same SNP gave a strong signal for BHR in the Dutch pedigrees, but only where the mother did not exhibit bronchial hyperresponsiveness to methacholine (in this subgroup P=0.004).

Such maternal effects were then looked for in the other families (Table). In both Dutch and Chicagoan families, the -964A allele was overrepresented in BHR positive children of BHR negative mothers, while the -964G allele was more common in asthmatic children of BHR positive mothers.

Table 40. HLA-G -964 genotype proportions in asthmatic (Chicago) or BHR+ (Dutch) children versus maternal affection status (test for homogeneity of association across studies P=0.21).
Study Maternal status N A/A A/G G/G
ChicagoMother BHR+ 33 0.1210.5150.364
Mother BHR- 46 0.3910.4130.196
Dutch Mother BHR+ 1360.1610.4190.419
Mother BHR- 1010.2670.5540.178

Furthermore, -964 A/A genotype was associated with atopy in the Dutch families (P=0.006), but this showed no interaction with maternal BHR. Therefore, G/G children had lower risk of atopy, but higher risk of BHR if their mother was affected.

Previous work on HLA-G has focussed on its' involvement in maternal-fetal immunotolerance and spontaneous abortion [Ober et al, 2003]. Yie et al [2005], for example, found HLA-G secretion by the embryo lead to a three-fold increase in successful IVF pregnancies. An exon 8 14 bp deletion has been reported to modulate mRNA isoform levels, and similarly increase the IVF success rate [Hviid et al 2004].


14 Tumour Necrosis Factor and atopy

Moffat and Cookson [1997] performed the first association study of asthma and the Tumour Necrosis Factor alpha gene (TNF). This gene was a good candidate because of the central role of TNF-alpha in the inflammatory process, and previous associations between polymorphisms in the gene with rheumatoid arthritis, lupus and infection.

TNF is mainly produced by monocytes and macrophages, but also by T-cells and several other cell types on occasion [Bayley et al 2004]. It is expressed on the cell surface as pro-TNF, which is activated by TACE or MMPs. Once activated, it binds to its two receptors TNFRSF1A (gene locatoed at 12p13), and TNFRSF1B (13q32).

Skoog et al [1999] point out that the old published sequence for TNF is slightly incorrect, so that all the polymorphisms described below are in fact 1 base-pair too high.

An interesting case-control study from the Gambia [McGuire et al 1994] found the TNF*2 allele to be associated with a seven-fold increase in cerebral malaria. The allele frequency in controls under this strong selection is still 16% (very similar to that in Taiwan, 14% [Chen et al 1997], and in African-Americans, 10% [Perry et al 2001], but lower than that seen in European studies, 20-30%, see below). The authors conclude that strong balancing selection must exist in The Gambia. This association has not been replicated in all studies. In a recent study [Knight et al 1999], two other variants in tight linkage disequilibrium with the TNF -308 variant -- TNF -376A>G and TNF -238A>G -- were found to be associated with cerebral malaria in The Gambia and Kenya. The TNF -376 A variant was found to bind OCT-1, and was associated with increased TNF production in monocytes. The TNF -308 variant, by contrast, is probably not functional.

Apropos of asthma, animal models reviewed elsewhere have also found linkage to the region around TNF.

The GeneCanvas website gives excellent control haplotype frequencies from large European samples. In the ECTIM study [Hermann et al 1998], the frequency of -308G>A was significantly higher in Belfast 24.2% than in France 15.7%. This fact should be borne in mind in interpreting some of the studies below.

There are currently 8 TNF promoter SNPs known,

-1031T/C
-863C/A
-857C/T
-575G/A
-376G/A
-308G/A
-244G/A
-238G/A

all (almost by definition) outside the 0 to -200 region known to be highly conserved between species [Bayley et al 2004]. Allele specific transcript quantification studies (such as Knight et al 2003] seem to show no effects of these common SNPs, and so disease associations are probably being driven by disequilibrium with other genes such as NFKBIL1 or even LTA [Bayley et al 2004]. Among these SNPs there are 4-5 common haplotypes for Caucasians:

Table. TNF haplotype frequencies for three European studies [Bishop et al 2004].
Haplotype
-1031, -863, -857, -308, -238
Proportion
TCCGG0.51-0.54
TCTGG0.04-0.16
TCCAG0.13-0.16
CACGG0.11-0.15
CCCGA0.01-0.05

14.1 Moffat and Cookson [1997]

This study of the Busselton sample [1997] looked for associations between asthma and HLA region III genes. There were 92 asthmatics ("[h]ave you had an attack of asthma on more than one occasion") and 318 nonasthmatics, all of whom were typed at HLA-DR, an RFLP polymorphism in the lymphotoxin alpha gene (LT-alpha-NcoI), and a promotor variant in the TNF gene (A>G at position -308). The LT-alpha (or TNF-beta) gene LTA, and TNF (or TNF-alpha) gene TNF map adjacent to one another on 6p21.3. The polymorphisms are in strong linkage disequilibrium with each other and the HLAA 1/B8/DR3 haplotype associated with autoimmune disease, and have been previously reported to influence TNF-alpha levels.

The allele frequencies in the control sample (see Table) are similar to those from other studies. There was no deviation from HWE (LTA X2(2 df)=0.16; TNF X2(2 df)=0.47), but strong linkage disequilibrium between the two markers (D=0.58 in the nonasthmatic group and 0.70 in the asthmatics) was present. Asthma was found to be associated with LTA and TNF: the frequency of LTA*1 was 45% in asthmatics and 33% in controls (X2(1 df)=8.97); for TNF*2, they were 30% and 18% (X2(1 df)=10.3). Examining haplotypes, the frequency of LTA*1/TNF*2 was 29% in asthmatics and 16% in controls.

Table 41. Lymphotoxin-alpha and Tumour Necrosis Factor gene polymorphisms and asthma [Moffat & Cookson 1997].
AsthmaLTANcoITNF(-308)
1/11/22/21/11/22/2
Yes19422744368
No351371412118912

14.2 Huang et al [1997] and TNF

These authors [1997] tested for an association between the TNF*2 allele and chronic bronchitis in 42 Taiwanese cases (29 smokers, 13 nonsmokers, mean age 69 years), 42 age-sex-smoking history matched controls, and 99 schoolchildren. There was no significant difference in allele frequencies between the two control groups (2.4% versus 5.1%), but the cases carried significantly more TNF*2 alleles (19%, hybrid Fisher test P=0.0001, see Table). There was no difference in allele frequency between smoking and nonsmoking cases (P=0.76). Note also that the control allele frequencies are lower than in the previously cited SLE study [Chen et al 1997].

Table 42. Tumour Necrosis Factor gene polymorphisms and chronic bronchitis [Huang et al 1997].
GroupTNF(-308)*2 allele frequency
1/11/22/2
Chronic bronchitics271410.19
Matched controls40200.02
Child controls891000.05

14.3 Albuquerque et al [1998] and TNF

A group from Western Australia here [1998] describe another association study based on a cohort of 100 children (born 1987-1990) unselected for asthma, and 55 asthmatic children attending a respiratory outpatient clinic. The former subgroup contributed 19 further cases (positive histamine inhalation challenge, skin prick test to common aeroallergens, plus history) and 50 controls (BHR and SPT negative, family history negative).

The -308 TNF *2 allele was significantly less common in cases (17%) than in controls (33%). A similar nonsignificant trend was seen for LTA (*1 34% v 47%). This is the opposite of the association previously reported for Western Australian asthmatics and nonasthmatics. The authors can only suggest that there is linkage disequilibrium with a nearby trait locus.

Table 43. Tumour Necrosis Factor gene -308 polymorphism and asthma [Albuquerque et al 1998]. Numbers for the cases may be incorrect, as the published table counts sum to 124.
AsthmaTNF(-308)*2 allele frequency
1/11/22/2
Yes45245(0.23)
No241970.33

14.4 Pignatti et al [1998] and TNF

This abstract [1998] describes a study of 131 Italian families (600 individuals). A sib-pair linkage analysis found increased sharing at LTA (NcoI polymorphism), but not TNF (-308), among pairs for SPT and atopy (SPT or high IgE, P=0.015), but not BHR or asthma. This may have been driven by an association between the LTA*2/LTA*2 and high IgE levels seen in women (P=0.003), but not in men.

14.5 Skoog et al [1999] and TNF

These authors first confirm that the old published sequence for TNF is slightly incorrect, so that all the polymorphisms described earlier are in fact 1 base-pair too high. They go on to describe a study of five TNF promotor region variants in a panel of 254 healthy Swedish men (though atopy or asthma were not exclusion criteria).

A relatively common -863C>A variant (frequency 17%) was found to be associated with altered binding of nuclear proteins to TNF and decreased transcription in in vitro studies. It was also associated with a low serum TNF-alpha level (C/C mean 2.39, N=106; C/A 2.10, 44; A/A 1.94, 6; P<0.05). These relationships were not seen with the -308 polymorphism, despite a moderate level of disequilibrium between -308 and -863, D'=0.22.

14.6 Wa et al [1999] and TNF

The large survey of BHR performed by Britton and colleagues that was described earlier gives rise to a nested case-control study of TNF and BHR in this paper [Wa et al 1999]. There was an association between allele two of the -308 TNF polymorphism and bronchial hyperreactivity (OR 2.12,1.04-4.32, P=0.036).

Table 44. Tumour Necrosis Factor gene -308 polymorphism, atopy and BHR [Wa et al 1999].
PhenotypeTNF(-308)*2 Allele frequency
1/11/22/2
Atopy+ BHR+836513 0.28
Atopy+ BHR-94548 0.22
Atopy- BHR+56228 0.22
Atopy- BHR-96525 0.20

14.7 Higham et al [2000] and TNF

This study attempted to replicate the findings of Huang et al [1997], using a British sample of 86 COPD patients, and two panels of controls (63 "healthy" smokers and ex-smokers; 199 blood donors). The TNF*2 allele frequency was 15% in the first two groups, and 17% in the other blood donors. In addition, there was no relationship with COPD severity.

14.8 Louis et al [2000] and TNF

This Belgian case-control study found no relationship between the -308 TNF polymorphism and asthma. However, the allele frequencies in this sample are quite different from the other European studies reviewed above.

Table 45. Tumour Necrosis Factor gene -308 polymorphism and asthma [Louis et al 2000].
PhenotypeTNF(-308)*2 Allele frequency
1/11/22/2
Asthma (N=95)64310 0.16
Controls (N=98)692720.16

14.9 Winchester et al [2000] and TNF

This is a nested case-control study that screened 1811 antenatal clinic attendees for a history of childhood asthma. Amusingly, there are 26 cases and 1048 controls (649 with no personal or family history of asthma). These subjects were genotyped at the TNF -308 and ACE I/D polymorphisms.

The *2 allele was more common in cases than all controls (in the Irish/UK ancestry group, multiplicative model LR test P=0.06). It was not increased in the positive family history control group.

Table 46. Tumour Necrosis Factor gene -308 polymorphism and asthma [Winchester et al 2000].
PhenotypeTNF(-308)*2 Allele frequency
1/11/22/2
UK/Irish
Asthma (N=20)992 0.325
Family History (N=117) 77 3730.184
No Family History (N=416)283116170.180
South Asian
Asthma (N=6)420 0.167
Family History (N=58) 50 3730.070
No Family History (N=275)2393330.071

14.10 Winterton et al [2001] and TNF

This study looked at BHR to inhaled SO2 in 62 asthmatics versus genotype at TNF-308, LTA, IL4R and UBG. Only 21% of the asthmatics exhibited BHR, and all 12 genotyped carried the *1/*1 genotype (Fisher exact test P=0.03, no multiple testing correction).

Table 47. Tumour Necrosis Factor gene -308 polymorphism and SO2 AHR in asthmatics [Winterton et al 2001].
PhenotypeTNF(-308)*2 Allele frequency
1/11/22/2
sensitive to SO2(N=13)1200 0.00
Insensitive to SO2(N=49)281350.25

14.11 Witte et al [2002] and TNF

This case-control study compares 236 physician-diagnosed adult asthmatics and 275 controls recruited through a San Diego HMO (16% were Hispanic-American, 8% African-American and 5% Asian-American). There was no H-W disequilibrium, and the case-control difference in genotype frequencies was only small (Fisher exact P=0.088). A dominant model gave OR=1.5 (95%CI=1.01-2.24).

Table 48. Tumour Necrosis Factor gene -308 polymorphism and asthma [Witte et al 2002].
PhenotypeTNF(-308)*2 Allele frequency
1/11/22/2
Asthma (N=236)164674 0.16
Controls (N=275)2125560.12

The LTA NcoI polymorphism gave even weaker evidence for association (genotypic Fisher exact P=0.44).


15 Filaggrin

15.1 Palmer et al [2007]

The BREATHE study includes 874 asthmatics (age 3-22 years) recruited from 19 clinics in Tayside, Scotland, half of whom reported a history of atopic dermatitis. These individuals were genotyped for FLG R501X and 2282del4. Filaggrin mutation frequency was increased in the atopic dermatitis subgroup compared to controls. Filaggrin genotype was associated with asthma severity (treatment regimen level and FEV1/FVC ratio), even after adjusting for presence or absence of atopic dermatitis.

15.2 Henderson et al [2008]

In a UK study, a cohort of 7000 children born in 1990-1991 were genotyped at the FLG R501X and 2282del4 polymorphisms. Carriers were at increased risk of atopic dermatitis, asthma (RR=1.8), especially coocurring with atopid dermaittis, and sensitization to aeroallergens.

15.3 Weidinger et al [2008]

In German ISAAC II samples of children (N=4258), five FLG variants (the two common null mutations and 3 rare ones) were tested for association to atopic disease. Atopic dermatitis risk was increased 3 fold in carriers. Allergic rhinitis was also increased in carriers, independent of atopic dermatitis status of the individual, but asthma was only increased in the presence of atopic dermatitis.

In nasal biopies, FLG was expressed only in cornified epithelium in the vestibule, but not respiratory epithelium.


16 The IL-4 receptor

The high-affinity IL-4 receptor is a heterodimer consisting of an alpha-subunit or CD124, and a gamma-subunit (CD132), and is a member of the hematopoeitin superfamily. CD124 also makes up part of the IL-13 receptor (with IL-13R beta), while CD132 is also the IL-2 receptor gamma-chain. The CD124 gene, IL4R, is located in chromosomal region 16p12.1, and codes for a 140 kD product. It binds IL-4, and carries a Stat-6 DNA-binding target site, which is one mechanism for the regulation of IL-4 levels by Stat-6. A soluble form of the IL-4 receptor may be a regulatory protein, akin to the role of soluble CD23.

The CD124 protein contains 825 residues. The extracellular domain runs from residue 26--232, the intracellular from 257--825. Binding of Stat-6, Jak and Fes tyrosine kinases is to the intracellular domain.

Given the centrality of IL-4 to so many inflammatory processes, it has been considered a candidate gene for other diseases. For example, Olavesen et al [2000] excluded it (via association analysis using four intragenic polymorphisms) as being the Crohn's disease gene IBD1 (on 16p11-12; actually CARD15, 9 MB distant).

The literature on the coding SNPs is slightly confusing in that codon numbering starts 25 lower in some publications (length of the signal peptide).

Table 49. A subset of the known polymorphisms (189 SNPs) in IL4R
Variantrs number(Alternative name)
I75V rs1805010(I50V)
E400A rs1805011(E375A)
C431R rs1805012(C406R)
S436L rs1805013
S503P rs1805015(S478P)
Q576R rs1801275(Q551R)
V579I rs3024677
P675S rs3024678
S752A rs1805016
S786P rs1805014

16.1 Hershey et al [1997]

Hershey and coworkers from the Howard Hughes Medical Institute and Washington University [1997] initially searched for mutations in a set of three patients with hyper-IgE syndrome using SSCP and subsequent sequencing of CD124 cDNAs. All were found to carry a A>G transposition at nucleotide 1902, giving rise to a Q576R transposition in the protein. Among seven subjects with severe atopic dermatitis, there were three heterozygotes and one homozygotes for the mutant R576 allele.

A sample of 50 hospital staff members underwent RAST (Alternaria, D pter, and cat dander), and total sIgE determinations. The R576 allele was significantly more frequent (P=0.001) among atopic individuals than nonatopics (Table).

Table 50. Frequency of R576 polymorphism in atopic and nonatopic subjects [Hershey et al 1997].
GroupQ/QR/QR/RR576 frequency
Atopic712135%
Nonatopic254110%

Importantly, it was shown that R576 carriers expressed twofold higher levels (2.2, SE=0.14) of CD23 after IL-4 stimulation than Q/Q homozygotes. There was no difference in Stat-6 binding between wild type and mutant 15- peptides spanning the residue 576 and its neighbouring phosphorylated tyrosine Y575. By contrast, SHP-1, a phosphotyrosine phosphatase, did bind less avidly to the mutant. SHP-1 may be involved in the termination of IL-4 signalling, and Hershey et al point out that familial erythrocytosis is due to a deletion of an SHP-1 binding site in the erythropoeitin receptor.

16.2 Deichmann et al [1998, 1999]

These papers describe linkage and association studies of 109 nuclear families ascertained via an atopic child (total sIgE > 100 IU/ml or specific IgE detectable by Magic Lite assay) from south-western Germany. This lead to a total of 144 affected sib-pairs for atopy (123 for any specific IgE). These were genotyped at four STRP markers flanking IL4R and spanning a total of 5.5 megabases:


    D16S403--D16S401--D16S313--(IL4R)--D16S298.  

Sib-pairs concordant for specific sensitization exhibited increased sharing of maternal (but not paternal) alleles at the markers closest to IL4R. This was especially noticeable comparing sharing through an affected parent (affected mother at D16S401: 31/43; affected father at D16S401: 29/61; affected mother at D16S313: 39/58; affected father at D16S313: 27/58). The results of a TDT were in keeping with these findings. For the 146bp size allele of D16S403, maternal transmission to an atopic offspring was 44/62, but paternal transmission was 25/48. This might be significant even after a Bonferroni correction for all possible tests.

In the subsequent paper [Kruse et al 1999], the same subjects were typed for the Q576R, E400A, and C431R polymorphisms, as well as a novel variant S503P. The frequency of the Q576R allele was 20.9%, and S503P, 15.5%. These were in linkage disequilibrium, and both variant alleles were associated with decreased total sIgE level (P=0.003, P=0.002), but only weakly with sensitization to aeroallergens (P=0.06, P=0.29). This result is exactly opposite to that of Hershey et al [1997]. Functional studies suggested that phosphorylation of STAT6 was increased if a variant allele was carried.

16.3 Grimbacher et al [1998]

In this letter to the editor, occurrence of the Q576R allele was described for 20 cases of hyper-IgE syndrome and 25 controls. The allele was carried by 4/20 cases and 6/25 controls. A TDT gave 0/3 cases of transmission from an unaffected parent, and one homozgous R/R parent of an affected child had a serum IgE of 13 IU/ml. Finally, linkage analysis in seven hyper-IgE pedigrees excluded linkage to four markers spanning the region containing IL4R.

16.4 Howard et al [1998,2002]

The earlier abstract [1998] reports an attempted replication of this finding using 141 families from the Groningen study. There was no linkage to asthma for 8 chromosome 16 markers. A case-control study using probands and their spouses did not detect an association between the Q576R polymorphism and high sIgE or SPTs.

The case-control study was extended [Howard et al 2002] to include 5 IL4R variants. While Q576R was not associated with asthma or atopy, the S503P allele was found to be associated with lower IgE levels (P=0.0007) and decreased risk of asthma (locus heritability/R2=0.04). Forming haplotypes did not increase the evidence for association.

Table 51. Frequency of polymorphism in atopic and nonatopic subjects [Howard et al 2002].
GroupI/II/VV/VI75 frequency
Controls25422053%
Asthma30532652%
S/SS/PP/PS503 frequency
Controls6633479%
Asthma11129487%
Q/QQ/RR/RR576 frequency
Controls7336520%
Asthma10739718%

When S503P and IL13 (-1111C>T) genotypes were combined, this gave similar levels of evidence of association (S503P: S/S versus rest, P=0.02; IL13 C/C versus rest, P=0.007; interaction, P=0.57).

16.5 Yandava et al [1998]

Another abstract from the 1998 ASHG meeting reports a case-control study (138 asthmatics, 128 controls). The Q576R allele frequency was significantly higher in the asthmatics (X2(1 df)=9.9). Furthermore, among controls, the R576 allele was associated with a sIgE>100 IU/ml (X2(2 df)=22.6).

16.6 Tan et al [1998]

A third abstract [1998] describes development of an AS-PCR assay for the Q576R polymorphism. The R allele was marginally more common in atopics in a sample of Malays (P=0.05), but not Chinese (P=0.6).

16.7 Willadsen et al [1998]

Carole Ober's Hutterite study also [1998] obtained evidence for association/linkage to IL4R via the TDT for asthma and atopy. They note that D16S401 (5 cM distant) gave a TDT P=3E-4, but that Q576R did not, P>0.1. However, other mutations within IL4R, such as Q400 did give weak evidence for association (asthma, P=0.02; atopy, P=0.04), and the haplotype Q400/R576 strengthened this slightly for asthma (P=0.004), and Q400/Q576 for atopy (P=0.04). This latter finding was considered consistent with a hypothesis of trait specific alleles, comparing this to Mitsuyasu [1998].

16.8 Mitsuyasu et al [1998, 1999]

These authors report results based on the previously described Osaka case-control panel. They found a strong association between an I75V polymorphism and allergic asthma (see Table). This acted in a largely recessive manner (assuming a population risk of asthma of 10% gives predicted penetrances of I/I 60%, I/V 6%, V/V 3%). Importantly, the I75 allele when transfected into B-cell lines was associated with increased STAT6 activation by IL4. No such effects were detectable for the Q576R polymorphism. Olaveson et al [2000] note there was linkage equilibrium between I75V and Q576R in their sample of 553 inflammatory bowel disease families. Izuhara et al [2000] describes more functional work, including a finding that PBMCs from I75 homozygotes produced three-fold more IgE on IL-4 stimulation than V75 homozygotes, while CD23 expression was also increased. All these donors were Q576 homozygotes, but it is not clear whether all were atopic or nonatopic subjects.

16.9 Noguchi et al [1999]

This paper describes case-control and TDT based analyses of the I75V polymorphism. The panel now contained 86 families ascertained through an asthmatic child (see Noguchi et al 1997), 15 singleton asthmatics and 101 nonatopic controls from the community close to the hospital. Although the control allele frequencies are very close to those found by Mitsuyasu et al [1998], the I75 allele was not increased among these atopic asthmatic children (see Table). The TDT gave similar results (I75 transmitted 51/118 times).

16.10 Tanaka et al [2001]

This case-control study is of atopic dermatitis in 302 adults versus 122 outpatient controls (normal sIgE, no history). Neither the I75V nor Q576R polymorphisms demonstrated association.

Table 52. Frequency of I75V polymorphism in atopic and nonatopic subjects from three Japanese studies [Mitsuyasu et al 1998, 1999; Noguchi et al 1999; Tanaka et al 2001].
GroupI/II/VV/VI75 frequency
Controls (Mitsuyasu et al)18463641%
Controls (Noguchi et al)16444138%
Controls (Tanaka et al)44601562%
Atopic Asthma (Mitsuyasu et al)119552673%
Atopic Asthma (Noguchi et al)10573438%
Nonatopic Asthma (Mitsuyasu et al)16503441%
Atopic Dermatitis (Tanaka et al)58432264%

16.11 Patuzzo et al [2000]

This Italian sib pair based study used 851 subjects from 192 families ascertained to contain a pair of atopic sibs. There was a total of 298 asthmatics and 522 atopics in the sample. No linkage or association to Q576R and E400A was detected (including via TDT). The R576 allele frequency was 18% in parents of cases, and 17.5% in 52 controls.

16.12 Oiso et al [1999]

These authors sequenced the IL4R gene in 27 adults with atopic dermatitis and 29 controls. The Q576R variant was present in 6 cases (11% allele frequency) and no controls.

16.13 Ziani et al [1999]

In another ASHG meeting poster, this French group reported on two known IL4R variants in 93 atopic children and 92 controls. The Q576R allele was significantly more common among cases (P=0.001) and seemed to act as a recessive.
Table 53. Frequency of Q576R polymorphism in atopic and nonatopic children from Ziani et al [1999].
GroupQ/QQ/RR/RR576 frequency
Cases44301937%
Controls5138324%

16.14 Rosa-Rosa et al [1999]

These authors found the R576 allele was more common in a sample of 149 asthmatics than in 57 controls (P=0.03). Within the asthmatic group, there was a gene dose effect on disease severity.

16.15 Ober et al [2000]

Evidence of association between asthma and IL4R haplotypes has been presented for the Hutterite and Chicago CSGA samples. These samples were haplotyped at six variants: I75V, E400A, C431R, S436L, S761P, S503P, and Q576R. In the Hutterites, the best single-locus TDT was for C431 (atopy 60/99 transmitted, P=0.005; asthma 22/30, P=0.01), but the E400-S503 haplotype gave an uncorrected P=6E-6. There were no significant results from the single locus TDT in the Chicago samples (for atopy, C431 was transmitted 25/41, P=0.2), but the two-locus haplotypes again gave more significant results (eg E400-S503 gave P=0.005).

16.16 Dupre et al [2000]

These authors compared 75 patients with atopic dermatitis or asthma and 150 controls. The Q576R allele frequency was 18% in cases and 16% in controls (P=0.59).

16.17 Sandford et al [2000]

The Q576R polymorphism was genotyped in 143 controls and 247 asthmatics, including a subset who had a fatal (N=130) or near fatal (N=27) asthma attack. There was no evidence of association (P=0.69)
Table 54. Frequency of Q576R polymorphism in asthma from Sandford et al [2000].
GroupQ/QQ/RR/RR576 frequency
Fatal/Near fatal asthma8447420%
Mild to moderate asthma6122316%
Controls9543519%

16.18 Haagerup et al [2001]

The TDT was used in 100 Danish families by these authors. The I50V and Q576R variants were not associated with five atopy phenotypes, and linkage to several multisatellite markers across the region was absent.

16.19 Tan et al [2002]

Andrew Sandford was also involved in this Singaporean study of Q576R: 241 asthmatics and 253 blood donor controls. The R576 allele frequency was 27% in Chinese controls, 28% in Indian and 55% in Malays. Only the Indian asthmatic group was higher than the ethnically matched controls (P=0.03, uncorrected).

16.20 Leung et al [2002]

This Hong-Kong study included 76 asthmatic children and 70 controls. The I50V allele frequencies were 45% and 48% respectively.

16.21 Franjkovic et al [2003]

This German study describes association between total sIgE level and IL4R haplotypes in 300 blood donors, as well as functional studies. Murine T-cell and B-cell lines transfected to carry I50V, E375A, S478P, and Q551R genotypes did not differ in terms of IL-4 induced proliferative response or CD23 expression. In the human cross-sectional study, there was association with one haplotype (P = 0.0062).

16.22 Wellcome Trust Consortium [2007]

The 1958 UK Birth Cohort contributed to the control panel for the Wellcome Trust Case Control Consortium, and association results for phenotypes such as serum IgE level and FEV1 are available on the web at http://www.b58cgene.sgul.ac.uk/.

The results for log sIgE level are fairly unimpressive, especially for rs1805010, and only slightly more encouraging for rs1801275 (R576Q).
Association between log sIgE level and IL4R SNPs for the 1958 UK Birth Cohort. Note there are multiple assays (eg Taqman, Illumina) for some SNPs.
PositiondbSNPNBetaSEP-value
27236044rs80444441436-0.02820.03060.356058
27239395rs649801113620.05290.02340.023924
27239475rs649801214360.05740.02280.011854
27248560rs478794813810.06540.02500.009140
27253855rs22835631381-0.03580.02550.160145
27258188rs302453013780.05340.02300.020078
27260214rs302453613790.00150.03230.964148
27261862rs30245471436-0.00510.03100.869298
27263704rs180501015650.01930.02270.396428
27264168rs302456013750.06740.02380.004673
27266389rs223934914360.00430.04000.913813
27267345rs302458514360.04880.02220.028313
27267345rs302458513710.04810.02290.035659
27271754rs302461314360.01990.02220.369833
27271846rs30246141436-0.04600.05330.388597
27273627rs30246301380-0.00540.03980.892636
27278659rs30246601359-0.01660.03370.623592
27279450rs30246681436-0.06680.05510.225527
27280632rs30246721368-0.01670.05020.738841
27281059rs30246761436-0.03230.02910.268232
27281373rs18050111600-0.03290.03410.336123
27281373rs18050111431-0.03110.03440.366408
27281373rs18050111364-0.03240.03580.365942
27281465rs18050121661-0.03040.03420.373972
27281465rs18050121402-0.02060.03650.572681
27281465rs18050121345-0.01650.03650.651938
27281481rs18050131638-0.02220.05120.664521
27281681rs18050151620-0.02490.02950.399932
27281681rs18050151431-0.03010.02940.306689
27281681rs18050151379-0.01750.03010.561343
27281901rs18012751656-0.05900.02660.026947
27281901rs18012752129-0.03030.02170.161439
27281901rs18012751431-0.05160.02690.054966
27281901rs18012751378-0.03710.02760.179091
27282428rs18050161357-0.08840.04990.076813
27282658rs20745701378-0.06150.05500.263474
27283043rs104963113770.0393 0.02270.084120
27283288rs8832 14360.0228 0.02220.304764


17 CARD4 and CARD15

The CARD (Caspase recruitment domain) family of proteins is involved in apoptosis, expecially in response to bacterial infection. They recognize bacterial lipopolysaccharides for example through their leucine-rich repeat (LRR) receptor domains, subsequently activating NF-kB.

CARD15 (also known as NOD2, 16q12) is expressed largely in monocytes. Common polymorphisms in CARD15 were shown by Ogura et al [2001] and Hugot et al [2001] to be strongly associated with risk of Crohn's disease (in fact to underlie the strongest linked locus for inflammatory bowel disease IBD1), as well as the granulomatous Blau syndrome. CARD15, and the mouse equivalent (Nod2), are quite polymorphic, a feature thought to reflect selective pressure.

CARD4 (NOD1, 7p15) is more widely expressed, and specifically senses gram-negative bacterial peptidoglycans. It is involved in NFKB activation. McGovern et al [2005] found a single indel (also associated with asthma, see below) to be associated with IBD.

In view of the association with atopic dermatitis suggested below, it is interesting to note a report of association of CARD15 with psoriasis. Rahman et al [2003] found that 28% of 187 Newfoundland psoriatic arthitis cases carried one of three common variants, compared to 12% of controls. However, two previous studies have not found association to uncomplicated psoriasis.

17.1 Kabesch et al [2003]

This group described an association study between the three common Crohn's disease risk variants and atopic disease, using the previously described study of Munich-Dresden schoolchildren. There were 1873 genotyped individuals. No variant was significantly associated with asthma, but one (G881R), was with hayfever (OR=3.2, P<0.001) and atopic dermatitis (OR=1.8, P<0.05)

17.2 Hysi et al [2005]

Cookson's group screened CARD4 (NOD1) for SNPs, finding 12 polymorphic sites, of which 4 were exonic. The best result was for novel indel at the intron 9/exon 9 boundary, "partially identified as rs6958571", with P-values of 0.0002 and 0.0005 in the Australian and UK panels respectively. In an additional German case-control panel (600 asthmatics, 1194 controls), the P-value was 0.01.

17.3 Weidinger et al [2005]

An unselected panel (N=1417), an atopic dermatitits case-control panel (N=227:227) and an atopic dermatitis triad sample (189 families) from Augsberg, Germany, were genotyped at 11 SNPs across CARD4. The best single SNP results were for total sIgE level with the intron 11 SNP rs2907748 (additive model for random sample P=0.006; for trios P="N.S."), and in intron 3, rs3020207 (random sample, N.S; trios, P<0.0001). Haplotype analysis found 7 common haplotypes, with a 6% frequency haplotype being protective in both the random and case-control panels against atopic dermatitis (uncorrected P=0.004, 0.003).


18 The IL-9 Receptor

The interleukin-9 receptor gene (IL9R) is located in the long-arm pseudoautosomal region (PAR2, Xq12 and Yq). It has several autosomal pseudogenes, and in the mouse is autosomal. The gene encodes a 522- amino acid type I transmembrane protein, which belongs to the hematopoietin receptor superfamily and is expressed in membrane-bound and soluble forms. It has binding sites for several Stat (1,3,5) and JAK1. There are a number of known coding variants. Grasso et al [1998] for example describe a del173 variant which abolished IL-9 binding.

18.1 Holroyd et al [1998]

Holroyd et al [1998] (Roy Levitt and Magainin Pharmaceuticals) performed a linkage analysis using 57 families (345 individuals) ascertained through an allergic asthmatic child. There was significant linkage between DXYS154 and BHR (P=6x10-5), asthma (P=6x10-4) and high sIgE level (P=0.001) using GENEHUNTER's nonparametric analysis.

Levitt et al [2003] is a patent describing mutation screening of IL9R in 52 "random" individuals, and reports a R344H variant

18.2 Kauppi et al [2002]

The Finnish Kainuu province study sample of 202 nuclear families (657 individuals) was genotyped at DXYS154 and DXS1108, 30 kbp from IL9R. There was no evidence of linkage to high total sIgE or asthma (NPL in subset of informative families), but a TDT analysis found overtransmission of the DXY1108*10 allele (30:58, Bonferroni corrected P=0.1).

18.3 Zghal et al [2002]

Slightly obliquely, an IL9 polymorphism was found to be associated with leishmania infection in a Tunisian sample.

18.4 Melen et al [2004]

The Swedish BASME birth cohort consists of 4089 children followed since 1994-6, of which 2614 gave a blood sample at age 4. A case-cohort study of 542 children with wheeze and 709 controls (542 without wheeze) was assembled, and a subset genotyped at 25 SNPs in IL9R. The four most informative SNPs (all intronic) were genotyped in the entire panel, of which two exhibited association with asthma (Table).

Table 55.
SNP Cases ControlsP-value
SNP Male FemaleMale Female
rs731477 0.17 0.21 0.22 0.18 0.40
rs731476 0.31 0.32 0.38 0.37 0.003
rs731478 0.13 0.16 0.20 0.18 0.004
rs19738800.15 0.17 0.14 0.19 0.60

A test for haplotype association was significant in males (P=0.01) but not females, but a formal test of sex heterogeneity was not presented, and the cohort analysis (N=709) combined the sexes to obtain a similar haplotype association result (P=0.05).


19 PAF acetylhydrolase

19.1 Miwa et al [1988]

Miwa et al [1988] reported the results of biochemical analyses and a small family study of the enzyme PAF acetylhydrolase (PAFAH), one of the two enzymes responsible for the degradation and resulting brief half-life of PAF in serum. An assay for acetylhydrolase was first developed and validated. The results from 816 healthy adults and 211 healthy children were found to be 1.85 and 1.25 nmol/min/50 ul serum respectively, and enzyme levels were found to be highly reproducible over an up to one year period in 15 subjects. The frequency distribution of activity in the children was definitely bimodal, while the adult results were unimodal. There were 40 healthy subjects found to exhibit no acetylhydrolase activity at all; this did not seem to be correlated with platelet aggregation or other abnormalities. Familial aggregation of the deficiency was then studied in the families of 10 deficient probands. Acetylhydrolase activity levels were divided into normal, less than 1.45 in adults and 1.2 in children, and completely absent. The pattern of segregation of these three classes was consistent with a codominant gene for low levels, with the deficiency state representing the homozygote form. The data for healthy children could also be explained in these terms (see Table); this fitted well with the expectations of the Hardy-Weinberg equilibrium (X2(1 df)=0.61, P=0.44), leading to a estimate of the frequency of the allele for low levels of 0.25.

Table 56. Trichotomisation of PAF acetylhydrolase activity in sample of healthy children from Miwa et al 1988, and Hardy-Weinberg expectations for frequencies (codominant hypothesis).
NormalLowAbsentTotal
Observed124798211
Expected126.6973.6110.69

Levels were next examined in 175 asthmatic children. Lower amounts of activity were associated with increasingly severe asthma (see Table). Complete absence of activity was found significantly more often in the severest classes of asthma (P<0.01). It seems possible that this genetically determined enzyme deficiency may modulate the severity of symptoms experienced in asthma, and the partial deficiency seems quite common in the Japanese population at least.

Table 57. PAF acetylhydrolase activity in asthmatic and healthy children (age 6-12 years) as described by Miwa et al [1988].
Enzyme ActivityControlsAsthma
No attacks of wheeze in 2 yrs <4 episodes of wheeze per year 5-9 episodes of wheeze per year >10 episodes of wheeze per year >10 episodes of dyspnoea per year
Mean1.251.261.271.401.210.99
SD0.490.450.470.490.530.59
Number deficient82/1335/42
Number2117129332418

19.2 Stafforini et al [1996]

Recently, Stafforini and coworkers [1996] have identified a Val279Phe mutation that cosegregated with PAF acetylhydrolase deficiency in four Japanese families. The PAFAH gene was mapped to 6p12-21, with gene symbol PLA2G7 (phospholipase A2, Group VII). It was present in a further 41 deficient individuals ascertained through a routine annual health screening, and absent in 84 normal controls. In a further random sample of 127 Japanese individuals the allele frequency was 0.17, agreeing with the earlier results. Interestingly, the mutation was not found in 108 US subjects. It seems possible that this genetically determined enzyme deficiency may modulate the severity of symptoms experienced in asthma, and the partial deficiency seems quite common in the Japanese population at least. An attributable risk calculation, contingent on these results being replicable, suggests this mutation could explain up to 20 percent of asthma in Japan (q=0.2, f2=0.12, f1=0.06, f0=0.04), if the heterozygote risk is in fact greater than baseline.

19.3 Kruse et al [2000]

The German and British families described earlier [Deichmann et al 1998; Kruse et al 1999; Gao et al 1998] were used to screen for PAF acetylhydrolase mutations with an intermediate phenotype that might modulate asthma risk. SSCP analysis detected three common coding variants in a panel of 50 unrelated controls (none of the Japanese deficiency variants). There was evidence of linkage disequilibrium between the different variants (as also seen by Bell et al [1997]).

Table 58. Frequency of three PLA2G7 polymorphisms in atopic and nonatopic members of German and British samples [Kruse et al 2000].
GroupNR92HI198TA379V
German controls 3324.1%0.8%10.3%
German cases 7225.4%3.7%21.9%
British controls15023.2%7.6%15.2%
British cases 15023.4%15.2%21.6%

In GEE1 linear and logistic regression analyses (allowing for the family structure), the T198 allele was found to be associated with high total sIgE level and specific sensitization (positive SPT or RAST for D pter, D far, cat dander, birch or grass pollen). The V379 was more strongly associated with specific senstization (OR=3.2, P=0.002). These associations were confirmed in the British sample for asthma (V379 OR=2.2, P=0.003). Functional studies found that the V379 recombinant enzyme to have both a higher Vmax and Km than the wild type enzyme, while the T198 Km was increased six-fold. The in vitro work also suggested that the H92 allele exhibited a cis effect, partially negating the effects of T198 and V379.

An attributable risk fraction calculation for A379V under the suggested model (q=0.14, f2=0.28, f1=0.14, f0=0.07) again gives a high estimate of 28% (drop from 9% to 7% risk of asthma if the V379 allele was eliminated from the population). Under such a model, the expected mean ibd sharing of affected sib pairs is only 52%, and the sibling recurrence risk ratio 1.1. If by contrast, the gene acts in a purely recessive fashion (q=0.14, f2=1.00, f1=0.08, f0=0.08), the attributable fraction would be 22%, but the sibling expected ibd 64%, and sib recurrence risk ratio 1.5 (much more likely to be also detectable by linkage methods).

19.4 Ito et al [2002]

In this study, association of asthma to V279F was tested via the TDT in 118 Japanese triads and in a case-control comparison versus 330 controls. The F/F genotype was significantly more common in cases (13% versus 4%) and was overtransmitted to probands.


20 PLA2G2D and asthma

An ASHG meeting poster [Basehore et al 2004] described association between asthma and SNPs in the phospholipase A2 gene cluster on 1p36. Out of 3 case-control panels (African-American, Hispanic and Causasian), the Hispanic set gave a best P-value of 0.0006 for BHR to one of 11 SNPs genotyped in PLA2G2D.


21 Drazen et al and others on 5-lipoxygenase variants

Drazen and coworkers [1999] have identified a tandem repeat polymorphism in the promotor region of the 5-lipoxygenase gene (ALOX5, 10q11.2), associated with decreased transcription. One must presume the decreasing variant is not more frequent among asthmatics, but it does seem to predict response to leukotriene antagonist therapy.

Among 221 non-steroid dependent asthmatics randomized to placebo or to ABT-761 (an ALOX-5 inhibitor, related to zileuton), the wild-type allele (5 tandem repeats) frequency was 77%. While *5 allele carriers experienced a 20% inprovement in FEV1 on active therapy, the 10 subjects carrying two mutant alleles did not (change=-1%; P<0.0001).

Kawagishi et al [2002] tested the ALOX5 promotor polymorphism for association with aspirin-intolerant asthma.

Table 59. Frequency of ALOX5 promotor polymorphism in aspirin tolerant and intolerant asthmatics [Kawagishi et al 2000].
Group N Frequency of *5 allele
Controls 550.718
Aspirin tolerant asthma 630.778
Aspirin intolerant asthma530.726


22 LTC4 synthase variants and aspirin sensitive asthma

The LTC4S gene is located on 5q, but is 35 Mbp distal to ADRB2. It is "another good candidate" given the importance of leukotrienes in allergic inflammation. Sanak et al [1997] reported a promotor polymorphism which seems associated with aspirin intolerant asthma in a smaller Polish sample. This finding was not replicated by a larger US study [Van Sambeek et al 2000]. A Japanese study [Kawagishi et al 2002] found the C allele significantly less common among aspirin tolerant asthmatics.

Table 60. Frequency of A-444C LTC4S polymorphism in aspirin-intolerant asthma from three studies [Sanak et al 1997; Van Sambeek et al 2000; Kawagishi al 2002].
GroupA/AA/CC/CC-444 frequency
Controls (Sanak et al)241710.226
Controls (Van Sambeek et al)7353110.274
Controls (Kawagishi et al)742880.200
Aspirin-intolerant asthma (Sanak et al)122960.436
Aspirin-intolerant asthma (Van Sambeek et al)351880.279
Aspirin-intolerant asthma (Kawagishi et al)401730.192
Aspirin-tolerant asthma (Sanak et al)362710.227
Aspirin-tolerant asthma (Van Sambreek et al)151440.333
Aspirin-tolerant asthma (Kawagishi et al)792010.110


23 Glutathione-S-transferase and asthma

A number of studies have examined known functional mutations in this family of genes as risk factors for cancer and inflammatory disease. There are 16 members spread across six clusters on multiple chromosomes. The enzymes detoxify a range of reactive oxidative species including carcinogens and some drugs.

The GST pi (GST P1-1) isoform of the enzyme (GSTP1, 11q13) is the commonest version in lung epithelium, and one variant I105V (rs1695) has been the commonest target of asthma association studies. The efficiency of V105 variant enzymes are 0.5-7 fold higher than I105 for conjugation of glutathione to carcinogenic epoxides, depending on the substrate [Hu et al 1997, 1998]. There are significant ethnic differences in allele frequencies, but overall, the V105 allele seems to be protective against asthma. In several studies, there are plausible interactions with exposure to tobacco smoke [Lee et al 2007, Li et al 2006, Palmer et al 2006] sports participation [Islam et al, 2009], and preexisting AHR [Imboden et al 2008]. These fit in with the idea that asthma might be precipitated by inflammatory changes due to pollution.

Table 61. Frequency of the GSTP1*V105 allele (rs1695) in studies of asthma and atopy
StudyEthnic GroupNAsthmaControl
Fryer (2001) UK 0.2840.478
Hemmingsen (2001) 0.0000.000
Spiteri (2001) 0.0000.000
Piirila (2001) Finland 0.0000.000
Hakonarson (2001)Iceland 94,94 0.3190.357
Gilliland (2002) USA 0.0000.000
Mapp (2002) Italy 0.2500.250
Safronova (2003) Russia 0.3090.349
Aynacioglu (2004)Turkey 210,265 0.2600.308
Tamer (2004) Turkey 0.4500.273
Lee (2004) Taiwan 0.0000.000
Lee (2005) Taiwan 0.1340.217
Nickel (2005) Europe 261,122 0.3160.331
Carroll (2005) Europe 0.0000.000
Li (2006) USA 0.0000.000
Palmer (2006) UK 0.362
Hanene (2007) Tunisia 105,112 0.3350.427
Lee (2007) Taiwan 0.1670.200
Kamada (2007) Japan kids391,639 0.1700.134
adults462,639 0.1370.134
Imboden (2008) Swiss 137,4289 0.2500.318
Islam (2009) USA 150,1460 0.3080.38
USA,White 50,496 0.3080.46
USA,Hisp 100,964 0.33
HapMap European 60 - 0.375
Asian 60 - 0.145
African 60 - 0.358


24 CTLA4, CD28, and IgE level

Table 62. Ensemble physical map positions of asthma/atopy candidate genes and markers on chromosome 2q33.
Gene or Marker Location (Mbp)Remarks
PPIG (CASP10) 170.266 Asthma assoc (Smith et al 2004)
D2S335 172.392
D2S364 182.860
D2S117 195.444
CFLAR (CASP8) 201.806
ALS2 202.390 Juvenile ALS
D2S2772 204.251
D2S251 204.386
CD28 204.397 Tp44
D2S1739 204.513
CTLA4 204.558 IDDM12
ICOS 204.627 Inducible costimulator (T-cell)
D2S1384 205.053
D2S325 208.096

CTLA4 (2q33, 204.558 Mbp from 2p telomere) codes for CD152 or Cytotoxic T-Lymphocyte-Associated protein 4, a member of the extensive immunoglobulin superfamily. It binds to B7-2 (CD86), acting as a costimulatory molecule to modulate ("immune brake") T-cell responsiveness. Variation at CTLA4 has been associated with IDDM (it is close to or identical to IDDM12), Graves disease, Hashimoto's thyroiditis. The strongest evidence of association to autoimmune disease is to the Ala17 variant of the exon 1 (signal peptide) Ala17Thr SNP (rs231775). Also studied is the -318C>T promotor variant (rs5742909), that increases promoter activity by 20-40% [Wang et al 2002; Anjos et al 2004, 2005], and +6230G>A (rs3087243) [Ueda et al 2003]. There have been several studies examining association with asthma and allergy.

The ICOS gene is immediately next to CTLA4. ICOS is a member of the CD28 family with a T-cell specific action, even though its ligand (B7RP-1) is found on B-cells and APCs. The ICOS knockout mouse is characterised by low levels of IgE and T-cell IL-4 production. The human ICOS deficiency state leads to hypogammaglobulinemia with decreased mature B-cells.

24.1 Hizawa et al [2001]

Hizawa et al [2001] extended their earlier case-control study (see above) to genotype 339 asthmatics and 305 controls at a CTLA4 promotor polymorphism (-319C>T) and the coding A17T polymorphism. Among the cases, but not control, the -318C/C homozygotes had a higher total sIgE level (Table). Upregulation of CTLA-4 by the -318T allele might be expected to inhibit the immune response.

24.2 Howard et al [2002]

The genome scan of the Groningen population, as noted earlier [Xu et al 2000], found linkage of total serum IgE level to D2S1391. Fine mapping lead to a peak lod of 3.2 at D2S2314 (10 cM centromeric to D2S1391, and 25 cM centromeric to CTLA4).

There was strong association of CTLA4*A17T genotype and sIgE level (Table). The T17 allele was the increaser allele (frequency 0.60 in this sample, similar to the Caucasian population frequency of 0.71 in Marron et al [1996]), with a QTL heritability of 3%.

24.3 Yang et al [2004]

A sample of 1333 Taiwanese undergoing routine health check-up were genotyped at the two CTLA4 SNPs: allergic diseases were diagnosed by brief questionnaire and total and specific sIgE levels. Atopy was found in 23% of the sample. The T17 allele increased total sIgE level, most notably in females (P=0.001) where genotypic geometric mean levels were almost identical to those found by Howard et al [2002]. The T17 allele frequency in this Chinese population and in the Japanese population studied by Hizawa et al was lower (0.35) than that in Caucasians, and this ethnic difference is noticeable in other population data [dbSNP].

Table 63. Total serum IgE level versus CTLA4*A17T genotype [Hizawa et al 2001; Howard et al 2002; Yang et al 2004].
StudyA17T-319C>T
Genotype N Geometric Mean
sIgE (IU/ml)
Genotype N Geometric Mean
sIgE (IU/ml)
Hizawa Cases T/T 40 311.2 C/C 265 298.5
A/T 178252.3 C/T 71 178.6
A/A 121285.1 T/T 3
Hizawa Controls T/T 40 68.1 C/C 238 61.9
A/T 14058.5 C/T 65 73.5
A/A 12568.2 T/T 2
Howard T/T 12869.2 C/C 305 50.1
A/T 17744.7 C/T 51 51.3
A/A 54 28.8 T/T 3 12.6
Yang Males T/T 86 66.0
A/T 31274.1
A/A 26957.5
Yang Females T/T 65 61.7
A/T 31244.7
A/A 28934.7

24.4 Shilling et al [2005]

A set of 11 SNPs across the ICOS gene were genotyped in the Hutterites. Although no SNPs were associated with asthma or AHR, two 5' promoter SNPs did give a strong signal for total sIgE and SPT positivity (best P=0.00015 for cockroach allergen SPT, allelic OR=1.9). The -1413G>A variant was found to abolish a p50 NF-kappa-beta binding site, and AA homozygotes T-cell cultures had higher levels of IL-4, IL-5, IL-13 and TNF-alpha.

Table 64. ICOS -1413G>A 5' promoter polymorphism and presence of a positive SPT to any allergen [Shilling et al 2005]. Numbers are approximate, based on Figure 1 of that paper.
Atopy G/G A/G A/A Frequency
Any SPT+ 120 103 38 0.343
Nonatopic204 130 17 0.234

24.5 Jones et al [2006]

These authors have collected nuclear families containing one or more affected children attending a pediatric dermatology outpatient clinic at Westmead Hospital in Sydney, Australia. A set of 112 parent-offspring triads was used to test association (via TDT) to two CTLA4 SNPs (A17T, 3'UTR CT60). While the single SNP TDTs were not very impressive (exact P=0.05, P=0.07), the haplotypic TDT found the AA haplotype to be undertransmitted (P=0.001).

Table 65.Transmission of A17T and CT60 allele to children with atopic dermatitis [Jones et al 2006].
SNPTransmittedOR (95%CI)
A17T 51/831.59(1.00-2.56)
CT60 53/881.51(0.97-2.39)


25 CASP10, asthma and IgE level

Caspase 10 (FLICE2) is involved in apoptosis of immune system cells. Mutations lead to Autoimmune Lymphoproliferative Syndrome, where both APCs and lymphocytes are involved, and inactivating somatic mutations are common in non-Hodgkins lymphoma.

Smith et al [2004] describe more associations from the Wake Forest case-control panel of asthma and related phenotypes. Multiple SNPs were genotyped in CASP8 and CASP10. Association was seen to multiple SNPs and haplotypes around CASP10 (best P-value=0.0004 for FEV1/FVC ratio and BHR).


26 DPP10, asthma and IgE level

Dipeptidyl peptidase 10 (DPP10, 2q14) is a member of a family of serine proteases. It is most frequently expressed in the brain and pancreas, and has no known coding variants, though splice variants do exist. Jerng et al [2004] and Zagha et al [2005] identify it as a modulator of Kv4-mediated A-type potassium channel in CNS neuron somatodendritic compartments.

26.1 Allen et al [2003]

These authors were screening the combined Busselton-Oxford family samples for association of asthma with markers around the IL1 cluster, and obtained a significant TDT with the CA-repeat marker D2S308 (P=1e-4).

Table 66. NCBI physical map positions of genes and markers on chromosome 2q14.
Gene or Marker Location (Mbp)
IL1A 113.630
IL1B 113.686
D2S308 115.308
DPP10 116.4

Screening of 85 SNPs around D2S308 found that association signal was limited to a small region, with the most strongly associated SNP, denoted WTC122P, only 1 kbp from D2S308. This was found to be within a known DPP family promotor CdxA.

Most importantly, these associations were replicated when these markers were genotyped in a separate sample of 1047 Munich children [Kabesch et al 2003] (where the same (common) D2S308*3 allele was positively associated with atopy, P=0.004) and a case-control panel of British asthmatics (where results are presented for the WTC122P*1;D2S308*3 haplotype).
Table 67.Haplotype frequencies in asthmatics and controls.
SampleWTC122P*1;D2S308*3OR (95%CI)
TDT Aus+UK109/1821.90(1.19-3.08)
Severe case-controlUK 221:3041.35(1.03-1.76)
Mild Case-control UK 92:3040.91(0.65-1.29)

DPP10 is the only nearby gene, and although not expressed in lung, was tentatively suggested to cleave cytokine signal peptides.


27 Tenascin-C and asthma

Tenascin-C (TNC, 9q33) is a 190-240 kD extracellular matrix glycoprotein with wide ranging functions.

27.1 Tenascin-C and asthma (Matsuda et al 2005)

After noting that this gene was upregulated in array experiments using bronchial epithelial cells stimulated using Th2 cytokines, Matsuda and coauthors resequenced TNC, identifying 62 polymorphisms. They selected 10 "representative" SNPs across the gene, and these were genotyped in 446 Japanese asthma cases (aged 16-70 years), and 625 healthy controls (no history of asthma, atopic dermatitis or allergic rhinitis).

The only evidence of association was found for a coding SNP, Leu1677Ile (rs2104772), with P=0.0008. There was no dose-response with asthma severity, but excluding the 105 smoking/ex-smoker asthmatics did slightly increase the odds ratios. This is a very common polymorphism (Ile (T) allele frequency 20-67% in the different HapMap population panels).

Table 68. TNC exon 17 L1677I polymorphism and diagnosis of asthma [Matsuda et al 2005].
Group A/A A/T T/T Frequency HWE-P
Asthma 125 233 79 0.447 0.12
Control 169 303 183 0.511 0.06

27.2 Tenascin-C knockout mouse and asthma (Nakahara et al 2006)

The TNC knockout mouse is known to be grossly normal in development. In the BALBC/c ovalbumin challenge model, TNC deficient mice exhibited less airway reactivity, lower bronchial lavage levels of cytokines and eosinophils.


28 GPR154 and asthma

28.1 GPRA and asthma [Laitinen et al 2004]

Given that several genome scans had detected linkage of asthma phenotypes to chromosome 7p [Laitenin et al 2001], Polvi et al [2002] carried out fine mapping of a 24 cM confidence region using asthma families recruited from the Finnish Kainuu province (a relatively genetically homogenous population) and from North Karelia, as described above [Laitinen et al 1997, 2000]. In the process of this, they produced a physical map of 7p15-14, and developed 53 novel microsatellite markers, giving an initial fine mapping set with a resolution of 210 kbp.

The TCRG gene was within the linked region, with NPL scores of 3.3, but no strong evidence for association was adduced.

Laitinen et al [2004] describe further work. The Haplotype Pattern Mining algorithm [Toivonen 2000] highlighted haplotype sharing of a 46 kbp long haplotype spanning rs324396 to a new CA-repeat marker NM51. Sequencing of one homozygote asthmatic over 132 kbp containing this haplotype revealed 187 variants (includin 72 novel SNPs and 8 indels). A sample of 131 trios were genotyped at 51 of these SNPs, revealing the risk haplotype to be actually 133 kbp long.

A subset of informative SNPs from this set was then genotyped in North Karelian and Quebecois asthma families (from Thomas Hudson and Catherine Laprise -- see above), and association to asthma was replicated in both sets. There were seven common haplotypes seen in the three populations, of which H4 and H5 were Kainuu risk haplotypes, H7 North Karelian risk haplotypes, and H2 that for the Frech Canadians (H3 and H6 were protective across all three samples). Phylogenetic analysis did show the risk haplotypes were more closely related with each other, and specifically shared the rs323922*C allele. The best P-values for association were for French-Canadian homozygote H2 carriers and asthma (P=9e-4), and high total sIgE in Kainuu (P=0.01).

Table 69. Asthma risk haplotypes from 7p15 and proportions in two populations [Laitinen et al 2004]
Haplotype SNP position515224 516174 520598 522363 529556 531632 538567 543580 546333 555608 563704 574953 585883 FinnishCanadian
SNP name rs323917SNP516174rs323920rs323922rs324377 rs324373SNP538567rs182718SNP54633rs324384rs324396 rs324957rs740347
H1CGTGCCGGGCTAG0.3280.22
H2CTCCATGGGTCGG0.1890.181
H3CGTGCCGGGCCAG0.1310.21
H4CTCCATAAATCGC0.1130.067
H5CTCCATAAGTCGC0.0860.078
H6CGTGCTGGGTCGG0.0510.087
H7GGCCATGGGTCGG0.0470.0

Two genes were found within this region. One of these (LOC402479), was denoted GPRA (G-protein coupled receptor for asthma susceptibility, now GPR154). It is an orphaned G-coupled receptor (7 transmembrane domain, rhodopsin family), and rs324981 codes a N107I (or N258I) change within it. The other gene was concluded to not code a stable protein product, and was absent in the mouse sequence.

GPR154 was found in two main isoforms. The A isoform was found to be expressed in bronchial smooth muscle and basal colonic epithelium, while the B isoform was found in bronchial and gut apical epithelial cells, as well as throughout the epidermis. Expression of the B isoform was increased in asthmatic bronchial biopsies. In the mouse ovalbulin asthma model, Gpra mRNA was upregulated. This does not of course definitively address causation, as numerous genes are upregulated in inflammed airways.

Table 70.Allele frequencies (A>T) for rs324981
Study PopulationFrequency
Laitinen et al Finnish 0.47
Canadian 0.44
Affymetrix African-American0.67
Causasian 0.55
Asian 0.60
CEPH 0.25
PDpanel 0.38

28.2 Melen et al 2005

This paper describes data from two European cohorts: the PARSIFAL and BAMSE studies. PARSIFAL recruited 14893 5-13 y.o. schoolchildren from rural areas in five countries, while BAMSE is a Swedish birth cohort (born 1994-1996) of 4089 children. DNA was available from 3113 PARSIFAL participants and 800 4 y.o.'s from BAMSE. Seven (tagging) SNPs within GPR154 were genotyped, defining the same 7 haplotypes, which were represented at frequencies similar to those in Laitinen et al's Finnish sample.

Despite the reasonably large sample size, only weak evidence for association was found (global haplotype association to asthma P=0.022), with the highest SNP or haplotype odds ratios around 1.25. The H2 and H4 haplotypes were not risk factors for asthma or allergic sensitisation, and neither was the SNP rs323922, that tags these haplotypes. .

28.3 Soderhall et al 2005

These authors also tested GPR154 in a broad collection of atopic dermatitis singleton and family panels (283 multiplex families, 1213 individuals; 222 family-history-positive infantile atopic dermatitis patients from the Early Treatment of the Atopic Child trial cohort). Again, rs323922 was not found to be associated with risk of atopy using the TDT (for atopic dermatitis, 399/797 transmitted, OR=1.00, 95%CI=0.873-1.152; for asthma 82/148, 1.24, 0.90-1.72).

28.4 Vergara et al 2009

In two Columbian samples (116 asthma families; a case-control panel of 475 cases, 394 controls), rs740347 was found to associated with asthma (case-control P=1.7e-5, familial P=0.0013) and IgE level.


29 PTGDR and asthma (Oguma et al 2004)

The prostanoid DP receptor gene (14q22.1) encodes a 389 residue G-protein-coupled receptor. It is known that prostaglandin D2 inhibits Langherhan cell migration to lymph nodes via this receptor (though PGD2 is probably also a PPARG ligand, relevant to T-cell apoptosis at least), and PTGDR blockade decreases sensitisation to ovalbumin, the local inflammatory response to allergen and alters Th1/Th2 ratios, probably via IL-10 [Angeli et al 2004]. Monneret et al [2002] found PGD2 to be an eosinophil chemoattractant in humans. The mouse KO mounts only a weak inflammatory lung response to ovalbumin challenge despite a normal rise in serum IgE level [Matsumoto et al 2000].

29.1 Oguma et al 2004

The PTGDR gene was screened using SSCP and sequencing in a set of 25 asthmatics and 25 controls, and 6 variants were detected: three in the 5' region; three exonic, one coding (L123I). The four common SNPs were genotyped in 598 asthmatic white and black American cases and 220 (US military) controls.

The -549T>C allele was more common among cases than controls for whites (P=0.04) and blacks (P=0.06). Results for haplotypes were slightly stronger (white P=0.002, black P=0.01), and these haplotypes differed in terms of transcriptional efficiency.


30 IL1RL1 and asthma (Shimizu et al 2005)

The IL1RL1 (2q11.2) codes for a soluble form (ST2) and a transmembrane receptor form (ST2L) of a molecule similar to IL-1. ST2L is expressed by Th2 but not Th1 cells, and regulates IL1R1 and TLR-4 by sequestering their signals.

30.1 Shimizu et al [2005]

These authors had already noted allelic association of SNPs around the IL1R1 gene cluster. In a case-control study of 452 Japanese atopic dermatitis patients and 636 "randomly selected population-based" unaffected controls, one of seven IL1RL1 SNPs was found to be strongly associated with atopic dermatitis (Fisher exact P=4e-5). This SNP (-26999G>A) in the distal promoter was also shown to alter expression and serum levels of ST2.

Table 71. IL1R1L promoter polymorphism (-26999G>A) and atopic dermatitis Shimizu et al 2005].
Trait G/G A/G A/A Frequency A
AD 106 240 106 0.498
Control 223 279 112 0.410

30.2 Zhu et al [2008]

30.3 Reijmerink et al [2008]

30.4 Gudbjartsson et al [2009]

The sizeable eosinophilia/asthma GWAS carried out by DeCODE, confirmed association of SNPs in IL1RL1 with both asthma and high eosinophil count. The best SNP was rs1420101, which did not associate with atopic dermatitis in Shimuzu et al [2005]. The overall odds ratio with asthma was 1.16 (Mantel-Haenszel P=5.5e-7). The overall results from all the non-Icelandic samples is 1.17 (P=8e-5), and no significant evidence of between-study heterogeneity of association, although the allele frequencies do differ significantly between Iceland and all the other studies. This may be due to the fact that the SNP is imputed in all the non-Icelandic studies.

Table 72.Results from loglinear model combining data from nine studies.
Model TermdfDevianceP-value
pop:allele 44 91.155e-06
asthma:allele 25 15.773e-07
pop:asthma:allele2.1 90.99


31 Chromosome 11p atopy genes

31.1 Brooks-Wilson et al [2000]

The gene discovery company Sequana Therapeutics (subsequently merged with AXYS Pharmaceuticals) was involved in a genome scan for asthma in the Tristan da Cunha population. They were able to replicate their initial linkage finding in a second study also carried out by their Canadian collaborators [Zamel et al 1996].

This replication panel comprised 59 nuclear families containing one or more asthmatic children (plus an additional set of 156 parent-child triads for analysis via the transmission-disequilibrium test). This panel also supported linkage to D11S907 (P=0.001), while the TDT gave evidence of association to one allele of D11S2008 (uncorrected P=0.0001).

Subsequently two nearby genes denoted ASTH1I and ASTH1J were identified, and found to be members of the ets transcription factor family (another member, ETS1, activates TNF). ASTH1J is also identified as ELF5, and its product as Epithelium-specific ETS-2 (ESE-2), and ASTH1I as ESE-3. Several polymorphisms were identified in these genes and found to be associated with the asthma phenotype.

31.2 Sreekumar et al [2000]

This ASGH annual meeting abstract reported a confirmation of the above associations. Families from Minnesota (Malcolm Blumenthal) and Denmark, and a case-control panel from North Carolina were genotyped at 13 SNPs and one SSP. Several of these loci exhibited allelic association.

31.3 Baron et al [2002]

Baron and coworkers describe a case-control study of seven SNPs in ELF5 and its neighbour EHF. They studied 311 US asthmatics (participants in a drug trial); the 177 controls were healthy US Army recruits.

None of the SNP allele frequencies differed significantly between cases and controls. LD within each gene was strong (smallest D' between adjacent SNPs 0.76), and 6 common haplotypes (>5% frequency) were inferred. The haplotype frequencies did not differ between cases and controls.

Intronic and synonymous exonic SNP allele frequencies in cases and controls from Baron et al [2002].
GeneLocationnt Minor allele frequencyUncorrected
P-value
CaseControl
ESE2i156800.1300.1010.059
i259640.2830.3030.556
i3181750.3740.3570.660
ESE3e339570.1960.2070.719
i488640.2640.2960.356
i7158450.4690.4740.907
e9161920.0860.0940.707


32 IL18 and atopy

IL18 (11q22) codes for IL-18 (interferon-gamma inducing factor, 192 residues). Activation of the IL-18 receptor ultimately leads to nuclear translocation of NF-kB and upregulation of IFNG, IL4 and IL13 [Kruse et al 2003]. In a transgenic mouse model, it causes an dermatitis similar to atopic eczema [Konishi et al 2002].

Kruse et al [2003] reported SNP discovery and association analysis in 105 children (one per family) from 109 German families described earlier [Deichmann et al 1998]. Five of the total of 8 SNPs (5 novel) were associated with high IgE level and atopic specific sensitisation (best uncorrected P=0.001).

Table 73. Frequency of IL18 promotor polymorphisms in atopy [Kruse et al 2003].
Group: sIgE<100 IU/ml sIgE>100 IU/ml
N: 40 65
Prom 1 -137G frequency*0.15 0.37
Prom 2 -133G frequency*0.15 0.40

* Approximate, estimated from published odds ratios

Harada et al [2009] report an association between genotype at rs5744247 in IL18 and severity of asthma (P=0.0034), but not with presence of asthma per se. This SNP was shown to increase IL18 expression levels, and serum levels of IL-18 (P=0.03)


33 CHI3L1, YKL-40 and asthma

YKL-40 (human cartilage glycoprotein 39) is a chitinase-like cartilage glycoprotein. Serum levels of YLK-40 are increased in asthma [Chupp et al 2007] and other inflammatory diseases, such as sarcoidosis, rheumatoid arthritis and systemic sclerosis. In the Hutterite population at least, serum levels are highly heritable (broad heritability 100%) [Ober et al 2008].

The CHI3L1 gene (chitinase 3-like 1, 1q32.1) encodes YKL-40. A functional variant in the CHI3L1 promoter (rs4950928) has been reported to be associated with risk of schizophrenia [Zhao et al 2007].

33.1 Kruit et al [2007]

These authors tested an association between polymorphisms in the gene (chitinase 3-like 1, CHI3L1) encoding YKL-40 and serum YKL-40 levels in sarcoidosis patients (N=63) and controls (N=333). In the control group, a -329G>A CHI3L1 variant was found to decrease serum levels (P < 10-4).

33.2 Ober et al [2008]

Carol Ober and colleagues carried out a GWAS for YKL-40 level in the Hutterite population sample (N=632-652). In the assessed sample, 76 were asthmatic, and a total of 156 exhibited airway hyperresponsiveness. Genome-wide, another CHI3L1 promoter SNP (-131C>G, rs4950928) was found to be the strongest predictor of increased YKL-40 level, explaining 10% of the trait variance. This SNP was also associated with airway hyperresponsiveness (P=0.002) and asthma.

In two asthma case-control samples (Freiburg N=638, Chicago N=296), an association with asthma was replicated (P=1.2x10-5), as well as with YKL-40 level. The latter was also confirmed in a third birth cohort (the COAST study), where no asthma association was detectable.

Table 74. Association between rs4950928 and asthma in four samples [Ober et al 2008; Rathke et al 2009].
Study Group GenotypeAsthmaControlOR (95%CI)
Hutterite*C/C 0.180.821.00
C/G 0.110.890.56
G/G 0.070.930.34
Freiburg C/C 227 1501.00
C/G 100 1200.55 (0.39-0.78)
G/G 17 240.46 (0.22-0.94)
Chicago C/C 69 1111.00
C/G 25 800.50 (0.28-0.89)
G/G 5 71.14 (0.27-4.39)
CopenhagenC/C 34334951.00
C/G 14418030.86 (0.69-1.06)
G/G 30 2281.50 (1.00-2.26)

* Approximate, estimated from Figure 2 of Ober et al [2008].

33.3 Rathke et al [2009]

This is a sizeable sample from the Copenhagen-based Inter99 cross-sectional study. The best results for the seven SNPs tested were from rs4950928, with an increased risk of atopic asthma being correlated with the G/G genotype (genotypic asssociation Fisher exact P=0.003). This gives rise to significant between-study heterogeneity when combining these data with the Freiburg and Chicago panels (interaction P=0.02): a random-effects combined analysis finds the heterozygote protective effect significant, but the overall G/G effect not so. There are significant differences in allele frequencies comparing Copenhagen to the other centres.


34 Van Eerdewegh et al on ADAM33 [2002]

The ADAM (A Disintegrin And Metalloproteinase Domain) family of genes code for type I integral membrane proteins that contain zinc metalloprotease and disintegrin regions. ADAM1 and ADAM2 are the fertilin genes, ADAM10 and ADAM17 exhibits TNFA convertase activity (producing the mature protein from the precursor), ADAM8 and others are involved in tissue invasion by immune and cancer cells.

34.1 Van Eerdewegh et al [2002]

Genome Therapeutics (in collaboration with Holgate's group at Southampton) genome scanned 460 UK and US families containing two asthmatic sibs. A MLS peak of 2.94 was obtained on 20p near D20S482. Using asthma plus BHR increased the peak to 3.93.

Linkage disequilibrium mapping was then used to refine the location. There were 40 genes under this peak, and an initial LD scan used 135 SNPs in 23 of these, genotyped in 150 index cases where ibd sharing within the sib pair was 100%, and 217 controls. Peak association was found for multiple SNPs in and around the gene ADAM33 (single locus P=0.03--0.006). Strength of association was greatly strengthened when two-SNP haplotypes of the ADAM SNPs were constructed, the best at P=4e-5 for the combined sample, and 3e-6 for the UK sample. Performance of the TDT using the families gave similar results. There was moderate inter-SNP LD through a region of 185 kbp around ADAM33.

Table 75. Transmission of ADAM33 SNPs (spanning approx 2 kbp) to asthmatic children, and results of parallel case-control analysis [Van Eerdewegh et al 2002].
SNP Location AlleleTransmissionP-value (TDT)P-value (case-control)
"S1" (rs3918396)Exon 19 G 37/570.030.02
"S1/T1" Exon 19/20 GT 72/1100.0030.12
"T1" (rs2280090)Exon 20 T 43/700.070.90
"T1/V-1" Exon 20/Intron 22TC 80/1260.0040.05
"V-1" (rs543749)Intron 22 C 43/700.070.01

Werner et al [2002] examined 15 SNPs in ADAM22 in two German samples (171 families and 547 case-control subjects). There was significant association, most strongly with "F+1" (P=0.002), but the pattern of association was inconsistent across the two panels.

Chae et al [2003] discovered 16 new SNPs in ADAM33 including three in the promotor.

34.2 Howard et al [2003]

This multicentre (CSGA) replication uses US (black, Hispanic, white) and Dutch asthma families to generate case-control panels (spouse controls). The 8 SNPs reported to be associated with asthma were examined.

Allele frequencies differed significantly between the ethnic subgroups, the African-American controls exhibiting most differences. Association with asthma also varied across the groups, the best single results being "ST+7" and "V4" in the Dutch. The heterogeneity of asthma association between the ethnic groups is statistically significant (for V4, X26=17.6, P=0.007), as is the overdominance exhibited by V4 (X21=8.4, P=7e-15).

Table 76. Association between ADAM33 "V4" SNP and asthma in four populations: significant population heterogeneity and overdominance.
Study Group V4 GenotypeAsthmaControlOR (95%CI)
Dutch C/C 97 941.00
C/G 75 292.51 (1.51-4.24)
G/G 7 80.85 (0.29-2.45)
AfrAm C/C 60 851.00
C/G 761380.78 (0.51-1.20)
G/G 26 400.92 (0.50-1.66)
CausAmC/C1321331.00
C/G 73 870.85 (0.57-1.25)
G/G 13 111.19 (0.51-2.80)
HispAmC/C 75 971.00
C/G 34 281.57 (0.88-2.83)
G/G 5 32.16 (0.51-10.78)
When haplotypes were formed using the 8 SNPs, differences in frequencies between populations were seen even for Dutch and US whites, though the rankings of haplotypes in terms of frequency are broadly similar:

Table 77. Chromosome counts for two populations of presumed European descent (overall difference between controls permutation P=0.002). Haplotype is at 8 ADAM33 SNPs: S1;S2;ST+4;ST+7;T1;T2;V-1;V4.
Haplotype DutchUS White
AsthmaControlAsthmaControl
GGAGTCCC139107168151
GGCGTCCC 90 67 83 81
GCAGCTCC 46 36 54 76
GGCGTCCG 55 19 60 58
ACAATCAG 28 24 38 42
GGCATCCC 15 25 23 31
GCCATCAC 8 8 14 14
GCCGTCCC 2 0 5 8
GCAGTCCC 0 0 1 14
GCCATCAG 2 3 2 6
GGAGCTCC 2 2 2 0
ACAATCAC 0 1 3 1
GCCGTCCG 4 0 0 0
AGAATCAG 0 0 0 2
GGCGTTCG 1 0 0 0
GGCGTTAC 0 0 0 1
GGCGTCAC 0 0 0 1
GGCGCTCC 1 0 0 0
GCCGCTCC 0 0 0 1
GCAGCTCG 0 0 1 0
GCAATCAG 0 0 0 1

The three significant control differences are for GCAGTCCC, GCCGTCCC and GGCGTCCG. This is relevant in view of the the fact that the best single haplotype association with asthma is in the Dutch sample (GGCGTCCG) -- the Dutch controls differed from US controls at this haplotype (6.5% versus 13.2%), as much as they did from Dutch and US cases (14.0% and 11.9%). However, log-linear modelling of the above table gives:

Table 78.
Term DeviancedfP-value
group:asthma 9.3 10.0022
hap:group 52.5200.0001
hap:asthma 49.6200.0003
hap:as:group 16.8200.6670

That is, overall tests of heterogeneity of association between asthma and haplotypes in the two groups is not significant, and a common estimate of association is significant. Given the sparseness of the table, these might be taken with a grain of salt. The strongest specific haplotype associations are with GCAGTCCC (Wald test P=0.010) and GGCATCCC (P=0.013).

34.3 Lind et al [2003]

These authors report results from the Genetics of Asthma in Latino Americans study. Six ADAM33 SNPs were genotyped in 583 asthma triad families (used for TDT) and a case-control panel of 373 asthmatic and 325 "matched" controls.

There was no evidence of association of asthma or related phenotypes to any single SNP or haplotype. The SNP allele frequencies can be compared to the Hispanic panel from Howard et al [2003]:

Table 79. V4 SNP minor allele frequencies
StudyEthnic GroupAsthmaControl
HowardHispanics 0.1930.133
Lind Mexicans 0.1530.172
Puerto.Rican0.2800.242

34.4 van Diemen et al [2005]

The effects of ADAM33 on age-related decline in lung volumes was tested using the Vlagtwedde/Vlaardingen cohort study (1390 genotyped individuals). Overall, annual decline was 18.7 and 12.7 ml/y in females and males, and this was increased 4-10 ml/y in Q_1 and S_1 homozygotes (as well as S_1 heterozygotes).

34.5 Li et al [2009]

This study tests for association between an ADAM33 SNP rs597980 and psoriasis. They obtained a P-value of 0.0057, comparing 2025 psoriasis cases and 1597 controls.


35 Superoxide dismutase and asthma

The superoxide dismutases have been tested as a possible candidate for involvement in asthma. Kinnula and coworkers [2004] looked at a noncoding SNP in SOD1, A16V in SOD2 and R213G in SOD3 in a Finnish case-control panel. None exhibited association to asthma


36 Adenosine deaminase and asthma

Adenosine as noted earlier is a mediator of inflammation and provokes bronchoconstriction on inhalant challenge in asthmatics. Activity of adenosine deaminase regulates the intracellular activity of adenosine by degrading it to inosine and ammonia [Vogel & Motulsky 1986]. The gene for the enzyme is located on chromosome 20q13; four alleles (and a null allele) are known and act in a codominant fashion. ADA*1 (Asp8) and ADA*2 (Asn8) are the commonest alleles present in European populations (p=0.95, q=0.05), and the phenotype ADA 1-1 has greater enzymatic activity than the ADA 2-1 or 2-2 phenotypes. In a mouse partial ADA deficiency model (complete deficiency leads to SCID), lung inflammation and increased BHR lead to early death. The pathology responds to treatment with theophylline [Chun et al 2001].

36.1 Mano et al [1975]

This Japanese group found no association between ADA genotype and atopic asthma in a case-control comparison of 32 cases and 360 blood donors (Fisher P=0.623).
Table 80. Frequency of ADA polymorphism in asthma from Mano et al [1975].
Group1/11/22/2ADA*2 frequency
Atopic asthma32000.0%
Controls3451502.1%

36.2 Ronchetti et al

Ronchetti et al [1984] examined the relationship between wheezy bronchitis, atopic asthma and ADA. The authors used a case-control approach, examining the phenotypes of 291 asthmatic children (aged 1 mo to 15 y) treated at the University of Rome paediatric pulmonary clinic or living in a House for Asthmatic Children (N=56), and three sets of controls from Rome and the Po delta. Cases were significantly more likely to express the ADA 1-1 phenotype than controls (89% of cases and 83% of controls, OR=1.6, 95% CI 1.45-1.7). However, children with an onset of symptoms prior to age 2, or current age <5 years did not differ significantly from the controls in terms of phenotypic frequencies. The older asthmatics also had a greater number of positive skin tests. Unfortunately the authors do not give a tabulation of ADA phenotype versus skin atopy, allowing examination of the hypothesis that ADA might modulate reaginic sensitisation. They do note that adenosine at physiological concentrations does inhibit IgE mediated histamine release from human basophils.

Three points need to be noted. Firstly, the gene frequencies in the control group do fit Hardy-Weinberg expectations on a group by group basis, but not after pooling. However, the differences in gene frequencies between control groups are not significant (X2(2 df)=0.12, p=0.95), so that the pooling for comparison with cases is acceptable. Secondly, the asthmatics over age 5 are biased towards more severe cases (especially those from the House), so that there might be a gradation of phenotypic frequencies from less severe to more severe cases. Finally, the ADA 1-1 phenotype is the most common one (80-90% of European individuals), so that the findings can be better stated as the following: 10% of the population are protected against developing chronic asthma (an approximately 42% decrease in relative risk), but given a 5% population prevalence of chronic asthma, the rate difference between ADA 1-1 and the other phenotypes for asthma will be only 2%.


37 Frohlander & Stjernberg and others on haptoglobin

Four studies to date have claimed an association between haptoglobin levels and respiratory disease. Haptoglobin (Hp, 16q22.1) is an 80-160 kD serum alpha-globulin (depending on the phenotype). It is a acute phase reactant, and has been suggested to be a modulator of immune and inflammatory processes, but its main role is the binding and transport of free haemoglobin. It is an inhibitor of protease activity. Three main genotypes exist - inheritance is autosomal codominant with the frequency of the allele Hp1 being approximately 37% in European populations. Apparently a twin study has suggested that Hp levels are partially under genetic control, and that levels are not correlated with Hp phenotype [Frohlander & Stjernberg 1989].

Mano et al [1975], as described earlier, performed a case-control analysis on the association betweeen asthma and several classical protein polymorphisms. They found no marked differences in allele frequency between 127 Japanese atopic asthmatics (Hp*1 frequency 0.217) and 1074 blood donor controls (0.247, Fisher P=0.522). Note that these frequencies differ from those seen in Europeans.

Piessens, Marien and Stevens [1984] were the first authors I can find who have reported an association between Hp and asthma. Results were gathered for two series of subjects: one of 86 randomly selected asthma and or rhinitis patients (presumably from hospital and clinic patients), along with 86 healthy blood donors as controls; the second series comprising 212 asthma/rhinitis patients selected to give "an even distribution" of confounders such as age, total sIgE and skin atopy. Various subsamples had Hp activity, Hp phenotype, serum bilirubin, reticulocyte count (excluding iron deficiency etc as a cause of haptoglobulinemia), CH50, alpha-1-antitrypsin activity, and serum immunoglobulin determinations performed.

The haptoglobin levels in the 86 patients and controls were first examined. While the control results were unimodal and close to Gaussian in distribution, the results from the asthmatics and rhinitics were fairly uniformly distributed across the entire range of Hp activity, having significantly more high and low activity results than the controls. Hp phenotype was measured for 99 patients from the second series. The phenotypic frequencies were compared to a previously reported group of French blood donors and found not to differ significantly. The authors note however that 14/19 patients with a Hp level of <65 mg% were Hp 2-2, while only 1/29 patients with a high Hp level (>175 mg%) carried this phenotype. Cases with lower levels of Hp tended to be younger than the high Hp cases, to exhibit more atopy, whether total IgE, pollen or HDM RAST, have slightly lower mean sIgA and WBC, and be on less treatment. That a number of these differences were due purely to age confounding seems evident.

Kauffman and other workers [1990, 1991] have reported on a two wave community based panel study of wheezing in a "working population" (bus drivers). At the first survey, the prevalence of questionnaire-diagnosed wheeze and level of cigarette consumption in 892 men (age 22-55 years) was compared to the subjects' Hp levels. There were 29 (3.2%) subjects who reported wheeze, excluding those with intercurrent respiratory tract infections (RTI). Smokers had increased levels of Hp (as well as the other acute phase reactants measured) and also of wheezing. However, wheezy subjects overall had a decreased level of Hp, a finding that persisted after stratifying for cigarette use (mean adjusted Hp level for wheezers 29 mg%, for nonwheezers 150 mg%).

In the second survey, 310 men from the first study underwent methacholine provocation testing, and skin atopy to three allergens and total sIgE were determined. Bronchial hyperresponsiveness (present in 15% of the sample) was positively correlated with Hp level, even after adjustment for cigarette use and intercurrent RTI. Skin atopy and total sIgE were unrelated to Hp level. Wheezing was of course strongly associated with atopy, but Hp level was similar in wheezy and non-wheezy atopics. A logistic regression with history of wheeze as the dependent variable found that skin atopy (adjusted OR=3.6), membership of the lowest Hp decile (OR=3.9), and level of cigarette use (OR=3.2 for those smoking >30 g tobacco per day) were independent risk factors. The authors concluded by stating that low level of Hp, "possibly genetically determined", was consistently associated with wheeze and bronchial hyperresponsiveness, but not with atopy.

Frohlander and Stjernberg [1989] collected serum from 148 asthmatics (69 M,79 F) recruited from a respiratory outpatient clinic. Data collected included age of onset and family history of asthma, total sIgE, RAST to 12 common allergens, and haptoglobin phenotype. Atopy (diagnosed by RAST) was associated with early onset disease (OR=10.5, 95% confidence interval 4- 27), and with family history of asthma (OR=2, 1-4). Among patients with a family history of asthma, there was significant deviation from Hardy-Weinberg proportions, and from the genotype frequencies of the control group (see Table). Further examination revealed that this difference was due only to the atopic (and therefore early-onset) asthmatics with a positive family history (FH+); atopy as a whole being not significantly associated with a difference in frequencies.

Table 81. Distribution of Haptoglobin phenotypes and Hp1 frequencies [from Frohlander & Stjernberg 1989].
GroupHp phenotypeChi-squareP Hp*1 Frequency
1-12-12-2
Controls31010918960.3729
All asthmatics2367580.570.750.3818
FH+ asthmatics1316279.100.010.3750
FH- asthmatics742223.780.510.3944
Atopic asthmatics1031331.050.590.3446
Nonatopic cases1230231.380.500.4145
FH+ atopic cases910156.780.030.4118

* Contingency X2 for group versus control phenotypic frequencies.

The difference in gene frequencies between the asthmatics of any sort and the control group was not significant. The association lies specifically in the diminished number of heterozygotes (2-1) and excess of 1-1 homozygotes found among the atopic FH+ asthmatics (odds ratio for asthma versus genotype for either= 2.2, 95% confidence interval 1.03-4.56). This shows that the strength of the association is weak and the number of key cases small (34) - suggesting a possibility of a Type I error or confounding. The results contradict the conclusions of Piessens et al reviewed above.

Deviation from Hardy-Weinberg equilibrium can be due to a number of factors such as: (1) lack of random mating; (2) small breeding population/stochastic variation; and (3) fitness advantage or disadvantage of a particular genotype [Vogel & Motulsky 1986]; proof of the presence of disequilibrium and, even more so, its cause, can be difficult to obtain. In this case, it is quite possible that cases were obtained from different geographical and population areas.


38 Srivavasta and C3 Complement type

Srivavasta, Gupta and Srivavasta [1985] examined 165 patients from a respiratory outpatient clinic (mean age 31) and compared the allelic frequencies of the C3 polymorphism from these patients to those from 327 normal adults. An increase in the gene frequency of the F gene and also the FF and FS phenotypes was found (see Table), with an odds ratio for asthma in carriers of the F gene (versus the S phenotype) of 4.0 (95%CI=1.8- 8.8). A number of studies have suggested a role for complement and/or complement deficiencies in asthma (see above), but I know of no work correlating the C3 electrophoretic phenotypes with enzyme function. In view of the studies suggesting variations in frequency of C3F with age [Sorenson & Dissing 1975], part or all of the difference in frequencies between cases and controls might be due to age confounding.

Table 82. C3 phenotypes and frequency of the C3F gene in cases and controls [from Srivavasta et al 1985].
GroupC3 PhenotypeC3F frequency
FFFSSS
Asthmatics0191430.059
Controls193010.018


39 Oxelius and Gm

Oxelius [1990a, 1990b] examined "50 consecutive atopic patients", presumably attending a paediatric allergy clinic, as there were 41 subjects under the age of 14. Atopy was defined as a total sIgE >600 kU/l (2 SD above the mean of their reference population). IgG subclasses levels were determined. Typing on Gm was performed for G1m(a), G1m(f), G2m(n), G3m(b). The Gm system represents electrophoretically recognisable different forms of IgG due to variation at the IgH C regions, and might be expected to be associated with allergic phenomena, although linkage studies using markers from Chromosome 14 have not found any evidence for an association.

In the first paper, the G2m(n) allotype was shown to be increased in the atopic subjects (38/50, 76%) compared to Swedish population controls (58%) from another study. She also claimed a significant excess (using the "binominal" test) of the Gm(f,n,b) and Gm(a,f,n,b) constellations, although a G.O.F. X2 test on the tabulated results fails to achieve significance (P=0.08). In the second paper, she goes on to find associations between particular Gm phenotypes and IgG subclass levels in atopy. IgG1 and IgG4 levels were significantly elevated in the atopic subjects (as has been reported elsewhere), and were highest in the Gm(f,f,n,n) phenotype.

In a subsequent study [1993], 34 nonatopic and 35 atopic asthmatic children were typed. Atopic asthma was characterised by an increased frequency of the Gm(f,n,b) haplotype and levels of IgG1 and IgG4); nonatopic asthma by increases in Gm(a, g/a," ",g) and Gm(a," ",g/f," ",b), with decreased IgG2 and IgG3.

I would like to see a replication of this study using age matched control data, as the frequency of various phenotypes in a number of such systems seems to vary with age, possibly due to secular changes in breeding populations or selective withdrawal from the population due to death or illness.


40 Angiotensin Converting Enzyme

I briefly reviewed the role of the kinins and kininases in lung pathophysiology earlier. Although no systematic differences in serum angiotensin converting enzyme (kininase II, peptidyl dipeptidase 1, CD134) levels (sACE) between asthmatics and controls have been documented, a recent study [Benessiano et al 1997] has claimed an association between the ACE insertion/deletion polymorphism and asthma. The I/D polymorphism has been shown to be associated to a number of diseases, notably arterial hypertension and myocardial infarction, and does control sACE levels. A sample of 79 consecutive asthma patients were recruited through hospital admission or outpatient referral, and were compared to 33 respiratory patients (14 lung cancer) and 54 hospital staff members, all without a history of atopy. The two sets of controls did not differ in I/D genotype frequencies (but were not in Hardy-Weinberg equilibrium), but the asthmatics were significantly more likely to carry the I allele (X2(4 df)=14.32, P=0.006).

Table 83. Insertion/deletion polymorphisms of the angiotensin converting enzyme gene ACE in asthmatics and controls from Benessiano et al [1997].
GroupI/I%I/DD/D
Healthy controls275815
Nonasthmatic patients276112
Asthmatic patients473815

40.1 Shirakawa et al [1998]

This study failed to replicate any association between the I/D polymorphism in asthma using both the Japanese and British association panels. The frequency of the I allele was significantly higher among the Japanese, but in stratified analyses, no differences between cases and controls were found (see Table).

Table 84. Insertion/deletion polymorphisms of the angiotensin converting enzyme gene ACE in British and Japanese asthmatics and controls [Shirakawa et al 1998].
GroupI/II/DD/DI Frequency
British controls3171480.44
British asthmatics3164550.42
Japanese controls2748250.51
Japanese asthmatics103128690.56

40.2 Winchester et al [2000]

This case-control study was mention earlier. No differences between cases and control I/D allele frequencies were detected.

Table 85. Insertion/deletion polymorphisms of the angiotensin converting enzyme gene ACE in British asthmatics and controls [Winchester et al 2000].
GroupI/II/DD/DI Frequency
UK/Eire controls891931340.45
UK/Eire asthmatics31160.43


41 The delta-F508 variant of CFTR

A number of studies have now looked for a relationship between asthma and being a heterozygote carrier of the commonest variant seen in cystic fibrosis, the delta-F508 deletion in the cystic fibrosis transmembrane regulator gene (7q37). Both positive and negative correlations have been found. Freedman et al [2004] did find the ratio of arachidonic to docosahexaenoic acid in nasal mucosal cells to be increased in obligate heterozygotes (intermediate to the two homozygote values), so this may suggest a mechanism for an association.

41.1 Warner et al [1976]

In a small sample, Warner found cystic fibrosis heterozygotes were more likely to exhibit a positive skin prick test (47%) than controls (20%). A similar trend for asthma or hayfever was not significant.

41.2 Schroeder et al [1995]

Schroeder and coworkers [1995] claimed to find a heterozygote advantage, with asthma less common in delta-508F carriers.

41.3 Dahl et al [1998, 2001]

As part of the Copenhagen Heart Study, Dahl et al genotyped 9141 individuals who had reported via questionnaire on diagnosis of asthma ("Do you suffer from asthma?"), smoking history and occupational exposures to fumes etc, and undergone spirometry. There were 250 carriers (2.7%), of whom 23 (9%) reported a history of asthma. Noncarriers reported a marginally lower rate of asthma (6%, P=0.04). No such effect was seen for chronic bronchitis ("do you bring up phlegm at least 3 months continuously every year"). Since the genotype groups did not differ significantly in lung function, the authors examined the the 256 subjects with asthma and airway obstruction (FEV1<0.8 and FEV1/FVC<0.7). Within this subgroup, heterozygote carriers had lower FEV1 and FVCs than noncarriers (P<0.001).

A longitudinal analysis within the Copenhagen Heart Study [Dahl et al 2001] found less striking associations. Overall, the age related decline in lung volume was not increased in carriers compared to the wild type homozygotes, though was in asthmatics for FEV1 (P=0.01). Carriers did have smaller volumes, and were more likely to report asthma (10% v. 7%).

41.4 Lowenfels et al [1998]

This letter to the Lancet describes a survey of 1113 obligate heterozygotes from the International Association of Cystic Fibrosis Adults and friend controls. Asthma was only slightly increased in the carriers, but "allergic disorders, sinus and upper-digestive-tract disease were significantly more common".

41.5 Lazaro et al [1999]

In this Spanish study, a mutation screen of CFTR was carried out in 144 unrelated asthmatics and 41 controls. They found 15% of asthmatics carried a missense mutation, and none among the controls.

41.6 de Cid et al [2001]

In the EGEA study, 247 asthma cases were compared to 174 controls at several CFTR variants, matching on city of residence. There were no significant differences for any of the variants, with 3.2% of the cases and 2.9% of controls being delta-F508 heterozygotes (OR=1.13, 0.36-3.52). Overall, 6.9% of cases and 9.8% of controls carried one of the four common missense variants tested (R75Q, G576A, R668C, L997F).


42 FOXP3 and the X-linked autoimmunity-allergic dysregulation syndrome (XLAAD)

A rare X-linked disorder with high mortality, XLAAD (or IPEX: immunodysregulation, polyendocrinopathy, and enteropathy) is sometimes associated with an exfoliative rash and high total sIgE levels. The FOXP3 gene codes for scurfin, overexpression of which is associated with hypoimmunity, and deficiency with elevation of multiple cytokines and lymphocyte counts. The effects of FOXP3 are so general that it would seem unlikely to harbour an atopy-specific polymorphism.


43 Animal models of asthma and bronchial responsiveness

As in other complex diseases, a number of useful animal models of asthma have been developed. The use of standard crosses and recombinant inbred lines greatly increases the power of linkage methods to detect genes of small effect, especially in the case of quantitative traits associated with asthma. The traits currently most intensely studied in mice have been BR to methacholine [De Sanctis 1995], allergic sensitization to agents such as ovalbumin and sheep erythrocytes, mediators such as PAF [Zhang 1995], and chemical irritants such as ozone [Young 1995]. Such analyses have detected the presence of at least four significant loci, the effects of some overlapping over the different challenges, the locations of which correspond to good human candidate gene regions.

De Sanctis and coworkers [De Sanctis 1995] described such a linkage analysis in 321 mice from high BR (A/J) with a low BR (C57BL/6J) strain backcrosses. Three loci were detected, explaining 26% of the genetic interstrain variance (backcross heritability was 50%). The mouse candidate genes in these regions were tumour necrosis factor alpha (human homologue TNF, 6p), interleukin (IL2, IL3) receptors (22q), and platelet derived growth factor B chain (22q). Van Eerdewegh et al [2002] suggest however that bhr1 is in fact syntenic to human 20p, and might be identical to ADAM33. The A/J strain develops increased BR and airway inflammation following ozone exposure [Young 1995], which is attenuated by anti-TNF antibody pretreatment [Ewart 1996]. Ewart and coworkers [Ewart 1996] added a fourth locus, possibly IL5 (5q), in another study of BR in the A/J strain. This latter finding may be regarded as further support for the human studies of 5q [Marsh et al 1994; Meyers et al 1994; Postma et al 1995], as may the report of another locus syntenic to 5q31.1 modifying the Th1/Th2 responses [Gorham 1996]. Candidates examined in this region include Gata3 and IL1rn [Chiaramonte et al 2002; Li et al 2002] (though the human GATA3 localizes to 10p13) and IL1RN to 2q14). McIntire et al [2001] (as described above) mapped a BHR/IL4 responsiveness locus (Tim1) using congenics from a BALB/c background (high BHR) with DBA/2 (low BHR, low TH2 responsiveness).

Choi et al [2005] mapped three QTLs controlling T-cell IL-4 secretion in a BALB/c (low) x C57BL/6 (high) backcross experiment to MMU 7, 15 and 19. Screening using RDA with recombinant inbred CXB10 v. C57BL/6 RDA pointed to two candidates: the "hypothetical" protein FLJ20274 (THUMPD1) was found to be relatively overexpressed in BALB/c mice due to a 3'UTR deletion in that strain, and was linked to the phenotype. THUMPD1 is on HSA 16p12.3, at around 20.6 Mbp.

Swinburne et al [2009] carried out a linkage scan for "heaves" using two half-sib families of Warmblood sport horses. Linkage was found to chromosomes 13 and 15.