Circulating Vitamin D and Risk of Prostate Cancer—Response Free

Our study was designed to examine the main effect between circulating 25-hydroxyvitamin D [25(OH)D] and risk of developing prostate cancer over a 20-year period (1). At the same time, we conducted exploratory, hypothesis-generating subgroup analyses that included, for example, stratification by total calcium intake (as pointed out by Dr. Schwartz), as well as by several other biologically relevant factors including total vitamin D intake, level of physical activity, vitamin biochemical status (e.g., vitamin E), vitamin supplementation, and stage and grade of the prostate cancers at diagnosis. These analyses provide additional insight into the potential biology underlying the main association; for example, the interaction with leisure physical activity was statistically most significant (P = 0.03) and is consistent with a synergistic influence of higher UVB exposure and higher vitamin D serologic status on prostate cancer risk. The 25(OH)D–calcium intake interaction (P = 0.06) did show a stronger positive vitamin D–prostate cancer association in the higher calcium subgroup (as did several of the other subgroups we tested, it should be noted), yet the 95% confidence interval (CI) in the lower calcium stratum has an upper bound for the OR estimate of 1.75. [Stratification for dietary calcium intake yielded results very similar to those for subgroups of total calcium intake, with highest (vs. lowest) 25(OH)D quintile OR estimates of 1.18 and 1.75 for the low and high dietary calcium groups, respectively.] The relatively high calcium intake of the ATBC Study cohort was accounted for by high dairy product consumption, with calcium supplementation having been reported by only 11% of the men.

Dr. Schwartz also raises the issue of residual confounding by calcium intake possibly accounting for the positive vitamin D–prostate cancer association we observed. We had tested a wide range of potential confounders (including calcium intake) during our risk modeling analyses and found virtually no change to the prostate cancer risk estimates. For example, the OR for the highest quintile category of season-specific 25(OH)D status adjusted for calcium was 1.50 (95% CI, 1.09–2.07; P_trend = 0.02), as compared with the data we reported in Table 3 (i.e., OR = 1.56; 95% CI, 1.15–2.12; P_trend = 0.01; ref. 1).

Our conclusion that “men with higher vitamin D blood levels are at increased risk of developing prostate cancer” is supported by our study data as well as by several other prospective investigations that we discussed and depicted in our Fig. 2 (1). Determining whether the vitamin D–calcium–prostate cancer interaction suggested by our data is true will require replication in other studies.

See the original Letter to the Editor, p. 246