Vitamin D Supplementation and Total Cancer Incidence and Mortality

A Meta-Analysis of Randomized Controlled Trials

N. Keum; D. H. Lee; D. C. Greenwood; J. E. Manson; E. Giovannucci

Disclosures

Ann Oncol. 2019;30(5):733-743. 

In This Article

Results

Characteristics of Included RCTs

After screening 1861 articles, a total of 10 RCTs were included in the meta-analysis of total cancer incidence (6537 cases),[27–29,33,34,37–41] of which five RCTs were included in the meta-analysis of total cancer mortality (1591 deaths) (Table 1).[28,29,34,37,41] Five RCTs were conducted in the United States,[27,28,33,34,40] three RCTs in Europe,[37,39,41] and one in Australia and one in New Zealand.[29,38] Six RCTs provided vitamin D3 daily (400–2000 IU/day).[27,28,33,34,40,41] Four RCTs provided a large bolus of vitamin D3 nondaily (20 000 IU/week to 500 000 IU/year).[29,37–39] Circulating levels of 25(OH)D of the included trials ranged approximately between 38 and 83 nmol/l at baseline, and the range of the intervention group reached between 54 and 135 nmol/l at a point during the follow-up. The contrast in the attained blood 25(OH)D level between intervention and control group ranged between 12 and 50 nmol/l. The durations of follow-up period, including intervention and post-intervention follow-up, were approximately between 3 and 10 years.

Primary Meta-analysis: Vitamin D Supplementation and Total Cancer Incidence

The summary RR for intervention versus control group was 0.98 (95% CI, 0.93–1.03; P = 0.42) with no evidence of heterogeneity (I2 = 0%) (Figure 2A). In sensitivity analyses, the summary RR was 0.97 (95% CI = 0.91–1.04, P = 0.43, I2 = 18%) after replacing the results[27,29,33] that included all cases with the results that excluded cases during the first year of follow-up when provided; and 0.99 (95% CI, 0.92–1.06; P = 0.77; I2= 0%) after excluding two studies that tested the combined effect of vitamin D3 and calcium against placebo.[27,34] Small study effects, such as publication bias, were not indicated in either primary (P Egger = 0.53) or sensitivity analyses (P Egger = 0.32 and 0.84, respectively).

Figure 2.

Meta-analyses of vitamin D supplementation and (A) total cancer incidence; and (B) total cancer mortality.

The association between vitamin D supplementation and total cancer incidence was not statistically significantly heterogeneous with respect to regimen of intake (P heterogeneity = 0.37, Figure 3A), attained 25(OH)D level (P heterogeneity = 0.55, Figure 4A), and other stratifying factors (Supplementary Table S2, available at Annals of Oncology online).

Figure 3.

Subgroup meta-analyses of vitamin D supplementation and (A) total cancer incidence; and (B) total cancer mortality by regimen of vitamin D intake.

Figure 4.

Subgroup meta-analyses of vitamin D supplementation and (A) total cancer incidence; and (B) total cancer mortality by attained 25(OH)D level.

Primary Meta-analysis: Vitamin D Supplementation and Total Cancer Mortality

The summary RR for intervention versus control groups was 0.87 (95% CI, 0.79–0.96; P = 0.005) with no evidence of heterogeneity (I2 = 0%) (Figure 2B). An inverse association became stronger in sensitivity analyses accounting for potential latent effect (utilizing studies that excluded the first year of follow-up) (RR, 0.86; 95% CI, 0.78–0.95; P = 0.004; I2=0%);[28] and excluding a study that compared concomitant supplementation of vitamin D and calcium with placebo[34] (RR, 0.85; 95% CI, 0.75–0.97; P = 0.02, I 2 = 0%). There was no evidence of small study effects, such as publication bias, in either primary (P Egger = 0.76) or sensitivity analyses (P Egger = 0.91 and 0.08, respectively).

Given the small number of trials, the meta-analysis of total cancer mortality had limited statistical power to explore potential sources of heterogeneity. Indeed, the association between vitamin D supplementation and total cancer mortality did not vary statistically significantly by any of the stratifying variables tested (Pheterogeneity ≥ 0.64 for all) (Figures 3B and 4B, Supplementary Table S2, available at Annals of Oncology online). Yet, it is notable that a statistically significantly reduced cancer mortality was observed in response to daily dose (RR, 0.87; 95% CI, 0.78–0.96; P = 0.007; I2 = 0%) but not to infrequent bolus dose (Figure 3B). Additionally, the protective benefit of vitamin D supplementation against total cancer mortality was evident even when attained levels of circulating 25(OH)D were ≤100 nmol/l (RR, 0.88; 95% CI, 0.78–0.98; P = 0.02; I2 = 0%) (Figure 4B). Yet, although not statistically significant, a similar magnitude reduction was also observed in trials where the attained 25(OH)D levels were >100 nmol/l (RR, 0.85; 95% CI, 0.70–1.03; P = 0.11; I2 = 0%). An inverse association was manifest among trials with >5 years of follow-up (RR, 0.87; 95% CI, 0.78–0.96; P = 0.005; I2 = 0%) but not in a trial with ≤5 years (RR, 0.99; 95% CI, 0.60–1.64; P= 0.97; I2 = not applicable) (Supplementary Table S2, available at Annals of Oncology online) adds to evidence for a requirement of a latency period for an effect of vitamin D to emerge as suggested by previous trials.[27,29,33]

Secondary Meta-analysis: Vitamin D Supplementation and Total Mortality

Based on eight trials[27–29,35,37,38,40,41] including 5002 total deaths, the summary RR for intervention versus control groups was 0.93 (95% CI, 0.88–0.98; P = 0.009) with no evidence of heterogeneity (I2 = 0%) and small study effects (PEgger > 0.99) (Supplementary Figure S1, available at Annals of Oncology online). When the analysis was restricted to five trials (4872 total deaths)[28,29,35,37,41] included in total cancer mortality, an inverse association with total mortality remained virtually unchanged (RR, 0.93; 95% CI, 0.88–0.98; P = 0.01; I 2 = 0%), which is weaker than an association with total cancer mortality.

Supplementary Figure 1.

Meta-analysis of Vitamin D Supplementation and Total Mortality.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....