Multivitamin, calcium and folic acid supplements and the risk of colorectal cancer in Lynch syndrome

Abstract

Introduction

Colorectal cancer (CRC) is the third most common cancer in both men and women and the third leading cause of cancer-related deaths in the USA. ¹ The lifetime risk of CRC is 5% in men and 4.7% in women. ¹ Approximately 2–4% of all colorectal cancers, ^2–9 but 10–15% of colorectal cancers diagnosed before age 50 years, ^2,3,10 are attributable to Lynch syndrome. Lynch syndrome is an autosomal dominant disorder of cancer predisposition caused by a germline mutation in one of the DNA mismatch repair (MMR) genes MLH1, MSH2, MSH6 and PMS2.¹¹ Approximately 1 in 3100 people in the general population carry a mutation in an MMR gene. ¹² MMR gene mutation carriers have a substantially increased risk of colorectal cancer (CRC) to age 70 years, estimated to be between 20% and 70% depending on sex and the mutated MMR gene. ^13–16

Personal and lifestyle factors may modify the risk of CRC for MMR gene mutation carriers. Identifying modifiers of disease risk is important for understanding carcinogenesis, as it may indicate potential initiators or promoters of the disease. In addition, identifying potentially protective factors, or conversely, harmful and avoidable risk factors might provide opportunities for mutation carriers to reduce their risk of cancer. Previous studies have reported that increased body mass index (BMI), ^17–19 alcohol consumption ²⁰ and cigarette smoking ^18,20–23 are associated with an increased risk of CRC for mutation carriers, whereas aspirin intake, ^24,25 and fruit and dietary fibre intake ¹⁸ are associated with a decreased risk.

For the general population, intake of multivitamin, ²⁶ calcium ²⁷ and dietary folate or folic acid supplements ²⁸ have been reported to be associated with a decreased risk of CRC, although individual study findings have been inconsistent. In the current study, we report associations between intake of multivitamin, calcium and folic acid supplements and CRC risk for people with Lynch syndrome.

Materials and Methods

Study sample

The study sample comprised carriers of a germline pathogenic mutation in an MMR gene, who had been recruited by the Colon Cancer Family Registry. Detailed description of the study design and recruitment strategies have been previously published. ²⁹ Between 1997 and 2012, families were recruited either via population-based probands who were recently diagnosed with CRC and identified from state or regional population cancer registries in the USA (Washington, California, Arizona, Minnesota, Colorado, New Hampshire, North Carolina and Hawaii), Australia (Victoria) and Canada (Ontario), or via probands who were members of, multiple-case cancer families referred to family cancer clinics in the USA (Mayo Clinic, Rochester, Minnesota and Cleveland Clinic, Cleveland, Ohio), Canada (Ontario), Australia (Melbourne, Adelaide, Perth, Brisbane, Sydney) and New Zealand (Auckland). Probands were asked for permission to contact their relatives to seek their enrolment in the Colon Cancer Family Registry. Informed consent was obtained from all study participants and the study protocol was approved by the institutional research ethics review board at each recruitment centre.

Data Collection

Information on demographics, personal characteristics, personal and family history of cancer, CRC screening, colorectal polyps and polypectomy and other surgical procedures, as well as lifestyle factors including intake of multivitamin, calcium and folic acid supplements, were obtained from all probands and participating relatives, at the time of recruitment. The information was collected using standardized questionnaires via personal interviews, telephone interviews or mails. The questionnaires are available at: [ http://coloncfr.org/questionnaires ]. We attempted to request and obtain blood samples from all participants and tumour tissues from all CRC-affected participants.

MMR gene mutation testing

Testing for germline mutations in MLH1, MSH2, MSH6 and PMS2 was performed for all population-based probands who had a colorectal tumour displaying evidence of MMR-deficiency as evidenced by either tumour microsatellite instability (MSI) and/or loss of MMR-protein expression by immunohistochemistry, and for the youngest-onset CRC case from each clinic-based family regardless of tumour MMR-deficiency status. Details of germline testing methods have been described elsewhere. ³⁰ Relatives of the probands with a pathogenic MMR germline mutation, who provided a blood sample, underwent testing for the specific mutation identified in the proband. All probands and relatives with a pathogenic MMR gene mutation were eligible for the current study.

Exposure definitions

Intakes of multivitamin, calcium and folic acid supplements were the three main exposure variables. Ever users of supplements were defined as those who answered ‘yes’ to ‘Have you ever taken [supplement] at least twice a week for 1 month or longer?’ Never users were defined as those who answered ‘no’ to ‘Have you ever taken [supplement] at least twice a week for 1 month or longer?’ Duration of supplement intake was defined as self-reported total number of years of supplement intake before age at CRC diagnosis or censored age. Given that the age at first intake of these supplements was not reported, we estimated the age at first intake by subtracting the reported duration of intake from the age at the baseline questionnaire, assuming that the duration was continuous and recent. Frequency of supplement intake was defined as self-reported weekly frequency of supplement intake.

Statistical analysis

Proportional Cox regression, with age as the time scale, was used to estimate hazard ratios (HRs) for associations between multivitamin, calcium and folic acid supplement intake and CRC risk for MMR gene mutation carriers. Time at risk started at birth and ended at: age at diagnosis of CRC ( n = 744); or age at diagnosis of non-colorectal cancer ( n = 343); or first polypectomy ( n = 317); or interview ( n = 562), whichever occurred first. We censored at the age of first polypectomy to avoid potential bias, because polypectomy reduces CRC risk. We censored at age of diagnosis of a non-colorectal cancer because intake of supplements and CRC risk may be changed by diagnosis and treatment of a non-colorectal cancer. ^31,32

As some carriers from this study were ascertained because they had CRC, and diagnosed at a young age, the subject selection for testing for genetic mutations was not random with respect to the CRC status. To reduce bias of estimates of hazards, we adjusted for this non-random ascertainment by applying probability weights based on the approach described by Antoniou et al.³³ Weights were calculated for the carriers with CRC and without CRC for each age-stratum, so the proportion of carriers with CRC at each age-stratum equalled that expected based on the age-specific incidences of CRC for carriers. ³⁴

The proportional hazards (PH) assumption was tested using the Schoenfeld and scaled Schoenfeld residuals. ³⁵ Univariable and multivariable models were fitted separately for intake of each supplement (multivitamin, calcium and folic acid) as well as intake of two or more of the supplements. The variables that we considered as potential confounders are listed in Table 1 . The variables that did not meet the PH assumption were included in the model as time-dependent covariates. Interactions were assessed by a change in the log-likelihood ratio after the addition of a cross-product term between the exposure and potential effect modifiers identified a priori (country of recruitment, ascertainment method, mutated gene, sex and site of colorectal cancer). The overall model fit was assessed using Cox–Snell residuals as the time variable and plotting them against the Nelson–Aalen cumulative hazard function. ³⁶

For multivariable models, missing data were handled using both complete case analysis and multiple imputations. Numbers of missing values for all the variables are shown in Table 1 . Missing data were imputed using chained interactions, assuming that missingness was at random. ³⁷ Outcome status, age at the time of colorectal cancer diagnosis or censored age, year of birth, country, mutated gene, ascertainment method and whether the carrier was a proband were included in the imputation model; 50 imputed datasets were generated.

The following additional analyses were conducted: (i) analysis restricted to carriers who were diagnosed with CRC or censored within 5 years before interview, to reduce survival bias; and (ii) analysis in which outcome was defined as colorectal neoplasia, i.e. either CRC or colorectal polyps.

The 95% confidence intervals (CIs) of HRs were calculated accounting for clustering by family membership to allow for correlation of risk between relatives from the same family, using the Huber–White robust variance correction. ^38,39 All statistical tests were two-sided. All analyses were performed using Stata 13.0. ⁴⁰

Results

From the Colon Cancer Family Registry, we identified a total of 2003 carriers of a pathogenic mutation in an MMR gene. Of these carriers, we excluded 11 (0.5%) who were censored before 18 years of age and 26 (1.3%) with missing data for all main exposure variables. A total of 1966 (56% female) carriers contributed to the analysis; 719 carried a germline mutation in MLH1 , 931 in MSH2 , 211 in MSH6 and 105 in PMS2 . Carriers excluded due to missing data did not differ in main characteristics from those who were not excluded. Characteristics of the carriers included in the study are summarized in Table 1 . For carriers who were diagnosed with CRC or censored at an age younger than age at interview, the mean time interval between diagnosis or censored age and age at interview was 9.6 [standard deviation (SD) 9.8] years for carriers with CRC and 10.0 (SD 9.2) years for carriers without CRC ( P = 0.40).

A total of 744 (38%) MMR mutation carriers were reported to be diagnosed with CRC at a mean age of 42.4 (SD 10.5) years. Of these, 681 (91%) were confirmed by pathology reports or medical records or cancer registry linkage data. Overall, 18%, 6% and 5% of carriers affected with CRC reported intake of multivitamin, calcium and folic acid supplements, respectively, at least twice a week for at least 1 month, compared with 27%, 11% and 10% of unaffected carriers. Intakes of multivitamin and calcium supplements were higher in women than men and in carriers from the USA than those from Canada, Australia or New Zealand, and increased with age ( Table 2 ).

Overall, we found that intake of multivitamin and calcium supplements was associated with a decreased risk of CRC compared with never users (HR 0.55, 95% CI 0.40–0.75; and HR 0.46, 95% CI 0.30–0.71, respectively). There was a decreased risk of CRC associated with an increased duration of multivitamin intake (≥ 3 years: HR 0.47, 95% CI 0.32–0.69; per year: HR 0.95, 95% CI 0.91–0.98), and calcium intake (≥ 3 years: HR 0.42, 95% CI 0.23–0.74; per year: HR 0.90, 95% CI 0.83–0.98). There was no evidence of an association between folic acid supplement intake and CRC risk ( P = 0.82). The results from the multiple imputation analysis were similar to the complete case analysis ( Table 3 ). When we additionally adjusted for recent BMI, red meat intake and fruit and vegetable intake in the multiple imputation analysis, the results were similar to the main analysis ( Supplementary Table 1 , available as Supplementary data at IJE online). There was no evidence of interactions between each supplement intake and country of recruitment, ascertainment method, mutated gene, sex or site of colorectal cancer ( Supplementary Table 2 , available as Supplementary data at IJE online).

There was an inverse association between multivitamin supplement intake and CRC risk for both males and females (HR 0.45, 95% CI 0.28–0.72, and HR 0.64, 95% CI 0.43–0.95, respectively). There was an inverse association with calcium supplement intake in women (HR 0.63, 95% CI 0.38–1.04) but there was no evidence of such an association in men ( P = 0.29) ( Table 4 ).