In the context of the current literature
Few previous studies have evaluated the association of SNPs in the CNR1 or FAAH gene with adiposity-related traits, and only one previous study has undertaken comprehensive LD mapping of the CNR1 gene. In a sample of 928 male participants of the Olivetti Prospective Heart Study, the G allele of the 3813 A/G polymorphism (rs12720071) in the CNR1 gene was nominally associated with increased subscapular skinfold thickness (P = 0.03) and increased waist circumference (P = 0.05). In two additional analyses in a subgroup (n = 360) of this cohort, the G allele of this SNP was likewise nominally associated with increased waist circumference (P = 0.007 and P = 0.04, respectively) (30). These findings were then evaluated for replication in the Wandsworth Heart and Stroke Study (n = 216). Consistently, the minor G allele of rs12720071 was associated with higher waist circumference (P = 0.006) and higher BMI (P = 0.01). In both study samples, a second variant (r2 = 0.26 between the two SNPs) in the CNR1 gene (4895 A/G variant; rs806368) was tested. This variant displayed no association with any of the quantitative phenotypes analyzed (BMI, waist circumference, subscapular skinfold thickness) in the Olivetti Prospective Heart Study, but the minor allele of rs806368 was nominally associated with higher waist circumference (P = 0.047) in the Wandsworth Heart and Stroke Study (30). Whereas these findings are intriguing, the small sample and lack of correction for multiple testing suggest that these may have been false-positive findings. In our larger sample, neither rs12720071 nor rs806368 were associated with any adiposity-related trait.
Another variant in the CNR1 gene (rs1049353) displayed evidence for association with BMI in a relatively small (total n = 419) population-based sample from Southern Italy (31). Individuals homozygous for the common G allele were overrepresented in overweight and obese individuals (P = 0.03 for trend) (31). By contrast, rs1049353 was not associated with obesity (defined as BMI ≥30 kg/m2) in a case-control study (1,064 obese vs. 251 controls) (32), although the minor allele of this genetic variant was related to abdominal obesity (increased waist circumference, P = 0.008; higher waist-to-hip ratio, P = 0.009) in a subgroup analysis (n = 455) of obese men (32). These findings are similarly limited by the small sample and presentation of nominal P values. In our sample, rs1049353 was not associated with adiposity-related traits. Differences in the study design and sample (case-control study including rather young men and premenopausal women; mean age 35 and 41 years for controls and cases; the controls being derived from the university and hospital personnel vs. community-based cohort study including pre- and postmenopausal women and men, mean age 61 years) might also account for the observed differences between the paper by Peeters and colleagues and our results (32).
Müller and associates found no evidence for association of 8 CNR1 genetic variants (including rs1049353) with obesity by conducting family-based association tests in German obesity trios and screening the coding region of the CNR1 gene for mutations in 120 Germany obese children (33). Of note, most of these SNPs were located upstream of the promoter of the CNR1 gene (33). Our findings extend this literature by presenting a comprehensive assessment of the genetic variation in the CNR1 gene and evaluation of an adequately powered sample. Despite these strengths, we were unable to confirm any significant association of variants in CNR1 with adiposity traits.
Most recently, rs806381 and rs2023239 were associated with obesity in a French case-control study (1,932 obese vs. 1,173 controls; experiment wide P = 0.002 for rs806381 and P = 0.016 for rs2023239), with BMI in a Swiss cohort of obese adults (n = 865; nominal P = 0.015 for rs806381 and P = 0.02 for rs2023239) and with BMI in a Danish population-based cohort (n = 1,780; nominal P = 0.0023 for rs806381 and P = 0.021 for rs2023239). In our sample, rs6454673, which is highly correlated with rs806381 (r2 = 0.91), was not associated with adiposity traits. SNP rs6928813 (perfect proxy of rs2023239; r2 = 1), which failed HWE, was not associated with any adiposity traits in our sample.
Previous studies relating genetic variation in the FAAH gene to obesity have focused on the Pro129Thr polymorphism (rs324420). In one study, carriers of the minor A allele of this SNP were overrepresented in obese when compared with normal weight individuals of European (n = 1,688; P = 0.004) and African (n = 614; P = 0.049) descent, but not in a smaller sample from Asia (n = 365) (34). Furthermore, the AA genotype was positively associated with BMI as a continuous trait in a combined analysis of all three populations (P < 0.0001) (34). However, these findings were not replicated in a large population-based study from Denmark including >5,000 individuals (35). Our findings are in agreement with the latter study. In our large sample, rs324420 was not associated with adiposity-related measures.
Taken together, our data provide no evidence for association of common genetic variants in the CNR1 or the FAAH genes with adiposity-related traits in a large community-based sample. Further, we found no evidence for an association of common SNPs in the CNR1 or FAAH gene with visceral abdominal or subcutaneous fat.
Strengths and limitations
The large, unselected community-based design, the careful and standardized assessment of clinical covariates and the broad spectrum of adiposity-related traits (including CT measures of subcutaneous and visceral fat in 1,023 individuals), as well as the comprehensive coverage of common genetic variation in and around the CNR1 and the FAAH genes strengthen our study design. We determined subcutaneous and visceral fat using a volume measurement, which likely reflects more accurately the true fat burden in the abdomen as compared to planimetric measurements. In a previous publication we reported differences in the relative amounts of subcutaneous and visceral fat between volumetric and planimetric measurements, likely due to heterogeneity of fat distribution along the longitudinal axis of the abdomen (36). However, some limitations merit consideration. We focused on common genetic variants; thus, we cannot rule out that rare genetic variants (minor allele frequency <5%) in these genes might influence measures of adiposity. Furthermore, we did not cover 100% of the common genetic variation in both genes; it is, therefore, possible that other SNPs with a minor allele frequency above 5% might affect adiposity-related measures. In addition, we only had adequate power to detect genetic variants that explain 0.5% of the variance in the traits; thus, variants that have even more modest effects could have been missed. The HWE disequilibrium of SNP rs6928813 might have reduced its power to detect an association with our adiposity traits. The literature is divided about whether waist circumference should be measured at the level of the umbilicus (as in our and other studies (30)) or between the iliac crest and lower rib margin (32,35) or at the iliac crest (37). Finally, our results were obtained in individuals of European ancestry. The generalizability of our finding to other ethnicities is unknown.
This comprehensive analysis of common variants within two genes encoding major determinants of endocannabinoid activity found no evidence for association with a comprehensive set of cross-sectional and longitudinal adiposity measures in a large community-based sample.