Circulating 25-hydroxyvitamin D and survival outcomes of colorectal cancer: evidence from population-based prospective cohorts and Mendelian randomisation

Abstract

Background: To investigate the association between circulating 25-hydroxyvitamin D (25-OHD) and colorectal cancer (CRC) survival outcomes.

Methods: We conducted analyses among the Study of Colorectal Cancer in Scotland (SOCCS) and the UK Biobank (UKBB). Both cancer-specific survival (CSS) and overall survival (OS) outcomes were examined. The 25-OHD levels were categorised into three groups, and multi-variable Cox-proportional hazard models were applied to estimate hazard ratios (HRs). We performed individual-level Mendelian randomisation (MR) through the generated polygenic risk scores (PRS) of 25-OHD and summary-level MR using the inverse-variance weighted (IVW) method.

Results: We observed significantly poorer CSS (HR = 0.65,95%CI = 0.55-0.76,P = 1.03 × 10^-7) and OS (HR = 0.66,95%CI = 0.58-0.75,P = 8.15 × 10^-11) in patients with the lowest compared to those with the highest 25-OHD after adjusting for covariates. These associations remained across patients with varied tumour sites and stages. However, we found no significant association between 25-OHD PRS and either CSS (HR = 0.98,95%CI = 0.80-1.19,P = 0.83) or OS (HR = 1.07,95%CI = 0.91-1.25,P = 0.42). Furthermore, we found no evidence for causal effects by conducting summary-level MR analysis for either CSS (IVW:HR = 1.04,95%CI = 0.85-1.28,P = 0.70) or OS (IVW:HR = 1.10,95%CI = 0.93-1.31,P = 0.25).

Conclusion: This study supports the observed association between lower circulating 25-OHD and poorer survival outcomes for CRC patients. Whilst the genotype-specific association between better outcomes and higher 25-OHD is intriguing, we found no support for causality using MR approaches.

Fig. 1. Kaplan–Meier survival estimates of colorectal cancer-specific and overall survival in different circulating 25-OHD groups.

Survival estimates of colorectal cancer-specific survival a in SOCCS and c in UKBB, and overall survival b in SOCCS and d in UKBB, within different circulating 25-OHD groups. (Group 1: 25-OHD < 25 nmol/L, Group 2: 25-OHD = 25–50 nmol/L, Group 3: 25-OHD > 50 nmol/L). The HR and the corresponding P value were estimated in Model 1 by comparing the colorectal cancer-specific survival or overall survival difference between patients in Group 2/3 and in Group 1.

Fig. 2. Forest plots for hazard ratios of colorectal cancer-specific and overall survival.

The results of observational studies were tested in Model 1 by comparing the CCS or OS difference between patients with Group 3 and Group 1 25-OHD levels. The results of individual-level Mendelian randomisation studies were the CCS or OS difference per unit increase in 25-OHD PRS after adjusting age and sex. The results of summary-level Mendelian randomisation studies were the CCS or OS difference per standard deviation increase of rank-based inverse-normal transformed 25-OHD by adjusting age, sex and stages in SOCCS, and age and sex in UKBB.

Fig. 3. Kaplan–Meier survival estimates of colorectal cancer-specific and overall survival in the top and bottom tertiles of 25-OHD PRS.

Survival estimates of colorectal cancer-specific survival a in SOCCS and c in UKBB, and overall survival b in SOCCS and d in UKBB, in the top (Tertile 3) and bottom (Tertile 1) tertiles of 25-OHD PRS. The HR and the corresponding P value were estimated by comparing the colorectal cancer-specific survival or overall survival difference between patients in tertile 3 and in tertile 1 of 25-OHD PRS, adjusting for age and sex.