IBD5 is associated with an extensive complicated Crohn's disease feature: implications from genotype–phenotype analysis

Currently, decisions regarding patient care in Crohn's disease are largely dictated by clinical phenotypes, incorporating disease localisation and behaviour. In this context, a molecular classification based on genetic susceptibility can provide a far more meaningful stratification for biologically relevant genotype–phenotype associations, and ultimately, the individualisation of patient treatment.¹

The IBD5 risk haplotype (IBD5^risk) located within the 5q31 cytokine gene cluster has been unequivocally associated with Crohn's disease.² Specific clinical phenotypes have not been firmly established for IBD5^risk, although an association with perianal and with fistulising Crohn's disease has been proposed.³,⁴,⁵,⁶,⁷,⁸ We performed genotype–phenotype correlations for IBD5^risk and Crohn's disease in a cohort of 325 German patients, described elsewhere,³ to determine whether IBD5^risk is associated with specific localisation phenotypic features and, in particular, with an extensive disease feature. Disease characteristics and age of onset were also analysed. Phenotyping procedures have been tested for interobserver and intraobserver stability, and quality criteria for phenotype information have been developed and applied. The IBD5^risk was defined using the risk allele at SNP IGR298a_1 as a proxy, given its central location within this region. We adopted two different approaches to fully explore disease localisation data for IBD5 (ie specific site of involvement and extent of involvement): (1) the classical approach of studying ileal Crohn's disease, colonic Crohn's disease and perianal Crohn's disease regardless of the presence of inflammation in the other two sites; (2) entailing the construction of mutually exclusive categories that allow us to focus on the number of sites (and which ones) in which the disease is present. We compared IBD5^risk frequencies using the Pearson's χ² test. Table 1 shows the results of the test.

Table 1 Allele and genotype frequencies by phenotype*.

Phenotypes	All			IBD5risk
	n	%	Allele frequency (%)	n	%
Localisation
Ileal (1+4+5+7)	271	83.4	48.0	197	72.7
Perianal (2+4+6+7)	149	45.8	46.6	109	73.2
Colon (3+5+6+7)	160	49.2	47.5	116	72.5
				†
One location
1 Ileal only	101	31.1	44.1	67	66.3
2 Perianal only	10	3.1	20.0	4	40.0
3 Colon only	27	8.3	46.3	20	74.1
Two locations
4 Ileal+perianal	54	16.6	50.9	43	79.6
5 Ileal+colon	48	14.8	49.0	34	70.8
6 perianal+colon	17	5.2	32.4	9	52.9
Three locations
7 Ileal+perianal+colon	68	20.9	50.7	53	77.9
				p = 0.06
Behaviour
Inflammatory	71	21.8	38.0	44	62.0
Fistulising only	47	14.5	57.4	42	89.4
Stenotic only	63	19.4	50.8	46	73.0
Fistulising/stenotic	144	44.3	44.4	98	68.1
				p = 0.01
Sex
Female	230	70.8	45.2	160	69.6
Male	95	29.2	48.4	70	73.7
				p = 0.46
Smoking
Ever‐smokers	167	51.4	43.4	111	66.5
Never‐smokers	158	48.6	49.1	119	75.3
				p = 0.08
Age of onset
<17	55	16.9	51.2	41	74.6
17–40	259	79.7	45.6	184	71.0
>40	11	3.4	31.8	5	45.4
				p = 0.15
Smoking
Age at onset, mean (SD)	23.4 (7.8)			22.8 (7.5)
All	325	100.0	46.2	230	70.8

*The p values refer to Pearson's χ² test comparing counts within boxes.

†These categories are not mutually exclusive and therefore were not compared.

From the first approach, we did not find any difference in the percentage of patients with IBD5^risk in the three sites. From the mutually exclusive categories, it emerges that, even with smaller numbers, patients with at least one risk allele (IBD5^risk) are more frequent in categories including the ileal site with something else (especially categories 4 and 7).

Given the results in table 1, we looked for an association of IBD5^risk with the “ileal+other” category using a logistic regression model adjusting for age of onset, smoking status, sex and CARD15 status. We found that being an IBD5^risk carrier almost doubles the probability of developing Crohn's disease in the ileum and somewhere else at the same time (odds ratio (OR) 1.77 (95% confidence interval (CI) 1.1 to 2.9)). No difference was found between homozygotic and heterozygotic status, suggesting that one copy of the risk allele is enough to confer risk of developing an “extended” Crohn's disease phenotype. No significant association was found for the ileal only category. CARD15 and smoking status were not associated with the “ileal+other” category.

For Crohn's disease behaviour, we found a significantly lower percentage of patients with IBD5^risk with pure inflammatory Crohn's disease compared with those with fistulising or stricturing Crohn's disease (p = 0.01). In the multivariate analyses, fistulising behaviour was seen to be associated with IBD5^risk (OR 5.61 (95% CI 1.9 to 16.4)) and there seems to be a dose haplotype effect given that the OR for homozygosity is higher than that for heterozygosity (OR 6.5 (95% CI 1.6 to 26.5) v 4.8 (1.5 to 14.8), respectively). Stricturing behaviour was also associated with IBD5, but only in homozygotic patients (OR 3.45 (95% CI 1.2 to 9.5)).

The probability of being an IBD5^risk carrier decreases with age of onset but the OR did not reach statistical significance (OR = 0.97 per year of age (0.94–1.00)).

In summary, adopting an alternative approach to genotype–phenotype correlations—that is, looking at mutually exclusive categories based on the number of sites involved, we have defined an association of IBD5^risk with an “extensive complicated disease” feature in Crohn's disease in a rigorously phenotyped European population. These findings, if replicated in a larger independent cohort, would implicate that IBD5^risk could act as a modifier rather than as a true susceptibility gene, leading to the development of more complex multi‐site disease with coexistent ileal involvement and non‐purely inflammatory behaviour, especially for patients who carry two copies of the risk allele.