C-reactive protein polymorphism and Crohn's disease: authors' reply
This article is corrected by:
- Errata: Errata / Corrigenda Volume 25, Issue 8, 1002, Article first published online: 30 March 2007
AP&T correspondence columns are restricted to letters discussing papers that have been published in the journal.A letter must have a maximum of 300 words, may contain one table or figure, and should have no more than 10 references. It should be submitted electronically to the Editors via http://mc.manuscriptcentral.com/apt.
Sir, Thank you for commenting on our study published in AP&T (1). However, we disagree on the comments made by Szalai and Alarcón (2):
- (i) We never stated that the +1059G/C CRP polymorphism is associated with Crohn's disease (CD). Instead, we pointed out in this article: ‘Our results demonstrate that the CRP +1059G/C polymorphism is not over-represented in CD patients compared with the normal population…’ (see page 1111 of Ref. 1). In particular, none of the P-values comparing the genotypes in patients with CD with healthy controls demonstrated significance in our analysis (Table 3). Therefore, our conclusion was not – as claimed by Szalai and Alarcón – that this polymorphism is associated with CD, but that this polymorphism is associated with involvement of the terminal ileum and decreased serum CRP levels in patients with CD. This conclusion was not solely based on 5 CC homozygous patients but on all carriers of C alleles, which form the combined group of CC homozygous (n = 5) and GC heterozygous carriers (n = 17), compared with all wild-type carriers (n = 219).
- (ii) All P-values given in Table 3 are >0.05 (and not <0.05 as assumed by Szalai and Alarcón) and therefore not significant, which is consistent with the result of the chi-squared test as performed by Szalai and Alarcón.2
- (iii) Given our P-values in Table 2 and our explanations in the figure legend and the corresponding text in the Results and Discussion sections, it should have become clear that the conclusions drawn were not based on only five homozygous patients with the CC genotype but rather on the combined group of GC (n = 17) and CC genotypes (n = 5). Performing the analysis with the combined L1 phenotype (ileal involvement) and L3 phenotype (ileocolonic involvement), as suggested by Szalai and Alarcón, the prevalences were as follows: 58.4% among GG wild-type carriers (128/219), 88.2% among GC heterozygotes (15/17), and 80% among CC homozygotes (4/5). Analysing the resulting 3-by-2 Table (three genotypes vs. ileal involvement yes/no), as suggested by Szalai and Alarcón, yields a P-value below α = 0.05, namely P = 0.029 (Fisher's exact test). Given that GG wild-type patients are by far the largest genotype group (90.9%) of all patients analysed, it is not unexpected that they also are the largest patient group with ileal involvement, as they are in all other phenotype categories: L1 (72.7%), L2 (97.9%), L3 (89.6%), L4 (95.7%). Nonetheless, GG wild-type patients are relatively underrepresented in the L1 group (ileal involvement only). In view of the exploratory, i.e. rather hypothesis-generating than confirmatory, character of our analyses, we agree with Szalai and Alarcón that large studies are necessary to confirm our results. This point was discussed in great detail in the Discussion section of our article (for details see page 1112/1113 of ref. 1), which also included a power calculation. Depending on the prevalence of the CC genotype in different populations, between 1042 and 25000 patients would be necessary to demonstrate with 80% power an association of CC homozygosity with CD, which clearly can only be achieved in a large multicentre study, but also demonstrates that this polymorphism is not a major contributor to overall disease susceptibility for CD.