Re: IGF-1 gene polymorphism and risk for hereditary nonpolyposis colorectal cancer.

Recently, Zecevic et al. ( 1 ) reported an association between the size of the CArepeat sequence residing in the 5 ′ untranslated promoter region upstream of the start site of the IGF-1 gene and age of disease onset in 121 hereditary nonpolyposis colorectal cancer patients who harbored germline mutations in the mismatch repair (MMR) genes hMLH1 or hMSH2 ( 1 ). In their study, an association between the length of the polymorphism and age of disease onset in patients harboring hMSH2 germline mutations was observed. To determine if this relationship was applicable to other populations, to only hMSH2 mutation carriers, or to only men or women, we investigated the IGF-1 CA-repeat polymorphism in a total of 220 MLH1 and MSH2 mutation – positive patients from 36 families, including 123 probands/single family members with con fi rmed hMLH1 or hMSH2 germline mutations. Polymerase chain reaction con ditions and CA-repeat analyses were as previously described ( 1 , 2 ). Allele sizes were categorized as reported by Zecevic et al. ( 1 ) such that patients were grouped as having one allele with 17 or fewer CA repeats ( ≤ 17 CA) or both alleles with 18 or more CA repeats ( ≥ 18 CA). A clear relationship was observed for early-onset disease in the patient group with 17 or fewer CA repeats. Overall, patients with 17 or fewer CA repeats were more likely than patients with 18 or more repeats to have early onset of colorectal cancer using the log-rank (LR) test (LR [df = 1] = 4.71, P = .03). Kaplan – Meier analysis also revealed a 15-year difference in the age of colorectal cancer onset between patients in the two groups ( ≤ 17 CA, median age = 48 years, 95% confi dence interval [CI] = 44.5 to 51.3 years, range 21 – 84 years versus ≥ 18 CA, median age = 63 years, 95% CI = 51.8 to 74.2 years, range 21 – 95 years). When patients with hMSH2 and hMLH1 mutations were analyzed separately, a statistically signifi cant difference in the age of disease onset was observed only among the hMLH1 mutation carriers ( ≤ 17 CA versus ≥ 18 CA, LR [df = 1] = 5.05, P =. 025). For proportional hazard Cox regression modeling, two models were tested including hMLH1 – hMSH2 mutation group, IGF-1 CA-repeat group alone, and IGF-1 CA-repeat group plus sex with or without family clustering. No association between colorectal cancer risk and hMLH1 – hMSH2 mutation or interaction between hMLH1 – hMSH2 mutation group and the two CA-repeat IGF-1 groups was observed, which contradicted the Kaplan – Meier results based on MMR mutation type. The Cox model interpretation is preferred over the individual Kaplan – Meier survival curves, and it appears from the interaction test that it is the IGF-1 CA-repeat group alone, which is associated with risk of colorectal cancer and not the MMR mutation type. The fi nal model from Cox regression analysis included the IGF-1 group plus sex and family as a cluster variable. The smaller number of IGF-1 CA repeats had the strongest association with colorectal cancer risk compared with the 18 or more CA group, ( ≤ 17 CA group, hazard ratio = 1.5, 95% CI = 1.02 to 2.16; P = .044). The association be tween sex and risk of colorectal cancer approached borderline statistical significance (compared with females, for males, hazard ratio = 1.40, 95% CI = 0.97 to 2.03; P = .071). Our results indicate that the increased risk for colorectal cancer is equal in both hMLH1 and hMSH2 carriers. Although the association was not statistically signifi cant in this study, the risk of colorectal cancer may be slightly more profound in males.