Send letter to journal:
Re: Authors' reply to Browning

Dear Editor,

We note Dr Browning's interest in our work. Unfortunately, we find his letter to be factually incorrect, and therefore scientifically misleading.

In attempting to analyse further our own data, he has made invalid assumptions regarding our data set, which render his calculations inaccurate, and his conclusions inappropriate. Moreover, the selected literature references which he quotes omit critical recent data and reviews of this area, and are not representative of the current state of knowledge.

In response to requests from the scientific referees of our paper, we included data on 294 controls, who had been typed for either the DLG variants, the OCTN1/2 variants, or both OCTN1/2 and DLG5 variants. The OCTN1/2 data (not the focus of this paper) were added, as a marker independent of the DLG5 gene, in order to show that the two sets of controls were homogeneous. Of these 294 controls, DLG5 genotyping was undertaken in 269 samples, and successful genotypes obtained in 256 (95.2%). Thus the failure rate in controls (4.8%) was similar to that in IBD patients (4.4%, p=0.78 by Chi squared analysis). The Hardy-Weinberg data which Dr Browning derives are not attributable to genotyping failure.

The contribution of the DLG5 gene is clearly complex. Dr Browning fails to quote a series of large studies in adult European populations which are directly pertinent to our work. In the same volume of Gut as our paper, Torok et al.[1] show no association of the DLG5 113A variant in the German population, with a non-significant reduction in 113A allelic frequency in Crohn's disease (625 CD patients, 1012 controls). Vermeire et al. performed studies in 2032 Belgian individuals, using strategies of case-control analysis, as well as transmission disequilibrium testing, and noted significant undertransmission of the DLG5 113A allele towards affected offspring (p=0.01, by Chi squared analysis.[2] Data presented in abstract from Cambridge[3], and Oxford[4] show no significant association - in the full paper in press, the final analysis of the Cambridge data again demonstrates non-significant reduction in 113A allelic frequency in Crohn's disease (personal communication, Miles Parkes).

It is noteworthy that unpublished data from our own Unit suggest that in childhood onset disease, the penetrance of the DLG5 113A allele may be critically influenced by gender, and environmental influences (Russell et al., submitted).

These issues have recently been the subject of thoughtful editorials, and correspondence both in Gastroenterology[5], and most recently in the European Journal of Human Genetics[6,7] - again unfortunately these manuscripts are not considered by Dr Browning. We of course feel strongly that this area is clearly of great interest, but we regretfully find the deficiencies in Dr Browning's letter to be potentially counter-productive.

References

1. Torok H, Glas J, Tonenchi L et al. Polymorphisms in the DLG5 and OCTN cation transporter genes in Crohn's disease. Gut 2005:54:1421-7.

2. Vermeire S, Pierik M, Hlavaty T et al. Association of Organic Cation Transporter Risk Haplotype with Perianal Penetrating Crohn's disease, but not with Susceptibility to IBD. Gastroenterology 2005:129:1845-53.

3. Waller S, Tremelling M, Bredin F, Greenfield S, Parkes M. Replication of association between IBD and TNF-857, but not DLG5,NFKB, Keratin 8, or TUCAN/CARD8 Gut 2005:54;suppl2 A95.

4. Cummings JRF, Herrlinger KR, Ahmad T, Jewell DP. Genotype- phenotype analyses of the IBD susceptibility gene DLG5. Gut 2005:54: suppl 2 A95.

5. Trinh TT, Rioux JD. Understanding Association and Causality in the Genetic Studies of Inflammatory Bowel Disease. Gastroenterology 2005;129: 2106 -2109.

6. Tenesa A, Noble C, Satsangi J, Dunlop M. Association of DLG5 and inflammatory bowel disease across populations. Eur Journal Hum Genet 2006 Jan 4 e-pub.

Send letter to journal:
Re: DLG5 variants and susceptibility to inflammatory bowel disease in the Scottish population

Dear Editor,

The recent study by Noble and colleagues[1] did not find an association of the DLG5 G113A polymorphism with either inflammatory bowel disease (IBD) or Crohn’s disease (CD). Previous studies in other European populations had reported an increase in 113A allele frequency in IBD and CD patients [2,3], but, in striking contrast, Noble et al.[1] reported a trend toward decreased carrier frequency for the 113A allele in the IBD patients (p=0.069) and in the CD patients (p=0.057).

We believe there is reason to re-examine the genotype data used in the Noble et al. report.

First, there are significant differences in the proportion of missing genotypes between the IBD patients and the controls (p<0.00001, Fisher’s exact test). Using the data in Table 1 and Table 3[1], the missing genotype rate is 13% in controls and 4% in IBD patients. This difference strongly suggests differences in DNA quality or genotyping process between the IBD and control samples.

Second, using the data in Table 3[1], the G113A polymorphism is not in Hardy-Weinberg equilibrium (HWE) in the IBD patients (HWE p-value=0.0056, likelihood ratio test). Hardy-Weinberg equilibrium describes the relationship of allele and genotype frequencies in a randomly mating population, and departure from HWE in controls is commonly used to test for genotyping error.[4] If the G113A polymorphism is independent of IBD disease status, then we expect the G113A polymorphism to be in HWE in IBD patients. There are fewer heterozygote IBD patients than expected under Hardy-Weinberg equilibrium, and this may be related to the differences in heterozygote frequencies between IBD patients and controls (p=0.033) and between CD patients and controls (p=0.029) reported by Noble et al.[1] One possible explanation for this difference in heterozygote frequencies is genotyping error in IBD patients caused by allelic dropout of the 113A allele.

The differences in missing genotype rates and the departure from HWE in IBD patients suggest that the quality of the genotype data that underlie the conclusions of this study should be investigated.

References

1. Noble CL, Nimmo ER, Drummond H, et al. (2005) DLG5 variants do not influence susceptibility to inflammatory bowel disease in the Scottish population. Gut 2005;54:1416-20.

2. Stoll M, Corneliussen B, Costello CM, et al.(2004) Genetic variation in DLG5 is associated with inflammatory bowel disease. Nat Genet 2005;36:476-80.

3. Daly MJ, Pearce AV, Farwell L, et al. (2005) Association of DLG5 R30Q variant with inflammatory bowel disease. Eur J Hum Genet 2005;13:835- 9.