Genetic polymorphisms in fatty acid metabolism modify the association between dietary n3:n6 intake and risk of ulcerative colitis: A prospective cohort study

Study Cohort

The participants in our study were women who were enrolled in the Nurses' Health Study (NHS) or Nurses' Health Study II (NHS II). Details of these cohorts have been described previously^{13, 23-27}. In brief, the NHS enrolled 121,700 female registered nurses who were between the ages of 30 and 55 years in 1976 while the NHS II included 116,430 female nurses between the ages of 25 and 42 years in 1989. Women in both cohorts were followed through biennial questionnaires detailing new medical diagnoses and environmental exposures including diet. The rate of completion of the biennial questionnaires consistently exceeds 95%. In each of the biennial questionnaires, women self-reported a diagnosis of CD and UC. Each diagnosis was followed by a supplemental questionnaire that ascertained further details regarding the disease and requested permission to review medical records. Chart review was performed independently by two board-certified gastroenterologists blinded to exposure to confirm the diagnosis of CD or UC. Disagreements were infrequent and resolved through consensus. A final diagnosis of CD was established based on the presence of typical symptoms for at least 4 weeks duration accompanied by supporting endoscopic, radiologic, and histologic features suggesting small bowel involvement, skip lesions, transmural inflammation or granulomas^{2, 28}. A final diagnosis of UC similarly required at least 4 weeks of symptoms and typical contiguous inflammation beginning from the rectum extending proximally³.

Selection of Cases and Controls

In each of these cohorts (1989-90 for NHS I and 1996-99 for NHS II), a subset of women provided blood samples stored on ice packs through overnight courier. This included 32,826 women in NHS and 29,611 women in NHS II. Upon receipt, the blood samples were processed immediately and buffy coat extracted for genotyping and stored in a continuously monitored liquid nitrogen freezer. In 2001-04, an additional 29,684 women in NHS and 29,859 women in NHS II who had not previously provided blood provided buccal cells using a swish-and-spit method. DNA was extracted from the cheek cells and stored for genotyping as noted above.

This nested case-control study included women with a confirmed diagnosis of CD and UC who provided either blood or cheek cells for genotyping. For each case, 2 controls matched for age, menopausal status, month of blood sample collection, and use of menopausal hormone therapy were selected from the cohort. Premenopausal blood samples were matched for the day of luteal phase.

Assessment of Dietary Fat

Details about the assessment of dietary intake in these cohorts have been published previously^{13, 24, 29-32}. Briefly, in 1980, women in NHS were administered a 61-item semi-quantitative food frequency questionnaire that was expanded to a 121-item questionnaire in 1984 and 136-item questionnaire in 1986 that was repeated every 4 years. Initial dietary assessment was performed in 1991 in NHS II using a 131-item questionnaire and similarly repeated every 4 years. Total intake of dietary fat, saturated fatty acids (SFA), n-3 PUFA, n-6 PUFA, and ratio of n-3/n-6 PUFA was calculated based on U.S. Department of Agriculture food composition data incorporating fats used in cooking and baking. Intake of specific fatty acids including EPA, DHA, linoleic acid, and arachidonic acid were also calculated. Prior studies have demonstrated the assessment of fat intake to be valid and reproducible, correlating well with 1-week dietary records, red blood cell measurements, and composition from subcutaneous fat aspirates^{29, 33-36}. Dietary intake was energy-adjusted. To minimize potential for bias, we used prospectively assessed diet four years prior to diagnosis of CD or UC.

Covariates

Information was obtained on other covariates demonstrated previously to be associated with CD or UC in these cohorts including smoking³⁷, oral contraceptive use²⁵, menopausal hormone therapy use³⁸, regular use of non-steroidal anti-inflammatory drugs defined as use 2 or more times per week³⁹, and dietary intake of fiber²⁴ and vitamin D²³. Body mass index was calculated using self-reported height and weight and expressed as kilograms per square meter (kg/m²)⁴⁰.

Genotyping

We selected eight single nucleotide polymorphisms (SNPs), four in CYP4F3, and two each in fatty acid desaturase (FADS) 1 and FADS2 for genotyping based on previous studies on their possible role in modifying the association between n-3/n-6 PUFA intake and risk of pediatric CD^{22, 41}. Extraction of genomic DNA was performed using conventional methods and direct genotyping was performed by the 5′ nuclease assay (TaqMan®) and the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). Primers and probes were designed using the Primer Express® Oligo Design software v2.0 (ABI PRISM). Genotyping was performed by personnel blinded to exposure and outcome. A randomly selected 10% sample was genotyped in duplicate and demonstrated 100% concordance. All SNPs were in Hardy-Weinberg equilibrium (p > 0.25).

Statistical Analysis

Continuous variables were summarized using means and standard deviations while categorical variables were expressed as proportions and compared using the chi-square test. Each dietary constituent including total n-3 and n-6 PUFA intake and ratio of n-3/n-6 PUFA were dichotomized based on intake above or below the median for the cohort. Conditional multivariable logistic regression was performed to identify the independent association between total n-3 and n-6 PUFA intake and ratio of n-3/n-6 PUFA and risk of CD and UC separately, adjusting for other covariates that had a univariate p-value < 0.10 or had previously been shown to modify risk of CD or UC. The p-value for interaction between dietary n-3, n-6 or n-3/n-6 PUFA intake and genotype was performed by introducing a cross-product term with genotype. If we identified a statistically significant interaction between a particular genotype and ratio of n-3/n-6 PUFA intake and CD or UC or if the association was statistically significant in individuals with specific genotypes alone, we performed further analyses for each individual n-3 or n-6 fatty acid. A two-sided p-value < 0.05 indicated independent statistical significance. Institutional review board approval for the study was obtained from Partners Healthcare.

Study Population

Our final study population comprised of 101 and 139 incident cases of CD and UC respectively, matched to 495 healthy controls. Table 1 compares the characteristics of the three groups. There was no difference in the mean age between the three groups though women with CD tended to be slightly older than the other groups (p=0.07). There was no difference in the distribution of body mass index and physical activity between the three groups. Regular use of NSAIDs was infrequent and similar between those with CD, UC, and healthy controls. The daily mean caloric intake [CD (1733 kCal/day), UC (1750 kCal/day) and healthy controls (1756 kCal/day)] as well as total consumption of carbohydrates, protein, and fat were similar in those with CD, UC, or healthy controls. With respect to individual fatty acids, the three groups were similar in their total intake of n-6 PUFA and n-3 PUFA, with similar ratio of n-3/n-6 PUFA. None of the 8 SNPs genotyped were independently associated with CD or UC (Supplemental Table 1).

Ulcerative colitis

In the nested case-control study population, neither total intake of n-3 PUFA (Odds ratio (OR) 0.92, 95% confidence interval (CI) 0.62 – 1.37) nor n-6 PUFA (OR 1.19, 95% CI 0.80 – 1.76) was associated with risk of UC while a higher ratio of n-3/n-6 PUFA intake demonstrated a non-statistically significant trend (OR 0.71, 95% CI 0.47 – 1.08, p=0.11) upon adjustment for other covariates. The magnitude of the association was consistent with our prior prospective analysis of this cohort¹³. However, the association of the ratio of n-3/n-6 PUFA appeared to differ according to genetic variation at the CYP4F3 locus. Among women with a GG genotype at rs4646904 at the CYP4F3 locus (84 cases, 282 controls), a significant reduction in risk of UC was observed when the intake was greater than the median for the cohort when compared to intake less than the median (adjusted OR 0.57, 95% CI 0.32 – 0.99) (Table 2). However, among individuals with the AA or AG genotype, there was no association between n-3/n-6 PUFA intake and risk of UC (OR 0.95, 95% CI 0.47 – 1.93) (p-interaction=0.049). At rs1290617 at the CYP4F3 locus, a reduction in risk of UC was only noted in those with a GT or TT genotype (OR 0.47, 95% CI 0.26 – 0.84) but not in those with the GG genotype (1.03, 95% CI 0.49 – 2.17). However, a formal test for heterogeneity at this SNP was not statistically significant (p-interaction=0.32). Similarly, at rs379497, a reduced risk of UC was noted only in those with the GG genotype (OR 0.56, 95% CI 0.33 – 0.96) but not in those with GA/AA genotypes (OR 0.75, 95% CI 0.31 – 1.81). No statistically significant interactions were noted with any of the other SNPs.

Table 3 presents the results of the interaction across levels of intake of n-3 PUFA, n-6 PUFA, DHA, EPA, and long-chain n-3 PUFA spectively. Non-statistically significant trends were noted with a reduction in risk of UC with high intake of total n-3 PUFA (p-interaction=0.20) and EPA (p-interaction=0.20) among those with GG genotype of rs4646904 at CYP4F3 but not among those with AG/AA genotypes. At the rs1290617 SNP, a higher intake of total n-3 PUFA was associated with lower risk of CD in individuals with GT/GT genotype (OR 0.48, 95% CI 0.27 – 0.85) but not in those with the GG genotype (OR 1.51, 95% CI 0.76 – 3.00) (p-interaction=0.02).

Crohn's disease

Total n-3/n-6 PUFA ratio was not associated with risk of CD in our nested case-control cohort (OR 0.82, 95% CI 0.44 – 1.55) (Table 4). Stratifying by genotypes at CYP4F3, FADS1, and FADS2 loci also did not yield any significant associations.