Vitamin D and Cancer Incidence—Response to Grant

We thank Grant (1) for his thoughtful reading and commenting of our article (2). First, Grant mentions that the long follow-up time in the study may be the reason for not finding statistically significant associations between vitamin D and cancer risk. In additional analyses not included in our article, we therefore included a maximum of 5 years of follow-up only (5.0 year median follow-up time). Hazard ratios [HR; number of events (n)/total number (N); 95% confidence intervals (CI)] per 10 nmol/L higher baseline vitamin D level were now (only categories with >20 events are shown): for all cancers (n = 371/N = 10,709; HR, 1.03; 95% CI, 0.98–1.07), all cancers excluding non-melanoma skin cancer (NMSC; n = 297/N = 10,866; HR, 1.01; 95% CI, 0.96–1.06), colorectal cancer (n = 43/N = 11,119; HR, 0.94; 95% CI, 0.82–1.08), cancer of bronchus and lung (n = 32/N = 11,133; HR, 1.04; 95% CI, 0.91–1.19), breast cancer (n = 72/N = 5,606; HR, 1.02; 95% CI, 0.93–1.12), prostate cancer (n = 29/N = 5,451; HR, 1.03; 95% CI, 0.88–1.21), NMSC (n = 98/N = 10,972; HR, 1.09; 95% CI, 1.01–1.18), and malignant melanoma (n = 22/N = 11,100; HR, 1.00; 95% CI, 0.84–1.19). Thus, consistent with our published results, our data do not support the hypothesis that there is a beneficial effect of a higher vitamin D status on cancer risk when using shorter follow-up time.

Another concern raised is the different levels of vitamin D in the three studies. To take this into account, we performed meta-analyses of the study-specific estimates (2, 3). Despite signs of heterogeneity across the cohorts, the estimates were quite similar to the results from the pooled data analyses. In addition, the limitation of the assumption of a linear association between vitamin D and risk of cancer is mentioned. In fact, we analyzed the data both with vitamin D status in season-specific quartiles and as a continuous variable where we found the quadratic term of vitamin D status nonsignificant in all models (2). In this way, there were no signs of nonlinear relationships between vitamin D status and cancer risk.

See the original Letter to the Editor, p. 1950