Familial aggregation of diabetes, hypertension and cardiovascular conditions in a case-control study of colorectal neoplasia

Past studies have suggested an association between insulin resistance and colorectal cancer (CRC). The aim of this thesis was to examine whether these associations have familial determinants. Familial aggregation of diseases potentially associated with insulin resistance (diabetes mellitus, hypertension, cardiovascular diseases, breast and prostate cancer) was assessed in a case-control study of colorectal neoplasia via self-administered questionnaire. The generalized estimating equations method, which accounted for possible confounding by family size, was used to estimate regression coefficients and standard errors. Case probands had either adenomatous polyps (AP) (n = 172), current CRC (n = 25), or previously diagnosed CRC (n = 132), while control probands (n = 282) had a negative colonoscopy and no history of CRC, AP or hyperplastic polyps. Only prostate cancer [odds ratio (95%Cl): 2.4 (1.1–5.2)]( p < 0.05) was associated with proband case-control status in models adjusted for age, sex and year of birth of relatives and age and sex of probands. Significant effect modification was evident in the data, however, especially for diabetes. In models stratified by proband age at diagnosis (<55 vs. ≥55 years of age), reason for referral due to symptoms (abdominal pain, rectal bleeding or altered bowel habit) and generation (parents vs. siblings and children), diabetes was significantly inversely associated with case-control status among parents of younger probands with no symptoms [0.3 (0.1–0.9)]( p < 0.05), and strongly positively associated among siblings and children of older probands with symptoms [7.4 (2.3–23.7)](p = 0.0007). Stratified analyses showed significant positive associations between familial prostate cancer, diabetes, hypertension, stroke and angina and proband case-control status only among older probands with symptoms. Similar associations were seen in AP and CRC separately. Referral, recall and information biases (assessed using Monte Carlo simulation) did not account for the findings. The results support an insulin resistance hypothesis in a subgroup of patients with AP or CRC. Further study is warranted to confirm these results and determine the roles of environmental and genetic factors in accounting for these familial associations.