Circulating vitamin D concentration and risk of prostate cancer: a Mendelian randomization study
On October 31st 2017, we published a paper in the BMJ entitled “Circulating vitamin D concentration and risk of seven cancers: a Mendelian randomization study” (1), where we found little evidence that a multi-SNP (rs2282679, rs10741657, rs12785878, rs6013897) score of 25-hydroxyvitamin D (25(OH)D) concentrations was associated with risk of any of seven cancers (i.e., colorectal, breast, prostate, ovarian, lung and pancreatic cancer and neuroblastoma) or their subtypes. This previous paper studied 22,898 prostate cancer cases and 23,054 controls from the Prostate Cancer Association Group to Investigate Cancer Associated Alterations in the Genome (PRACTICAL).
On November 21st 2017, a new genome-wide association study for risk of prostate cancer was accepted for publication in Nature Genetics (2), which included newly genotyped data from 46,939 total prostate cases and 29,910 controls. In addition, a new genome-wide association study for serum 25(OH)D concentration is currently in press at Nature Communications (3), which included genotyped data from 79,366 participants and identified two novel SNPs (rs8018720, rs10745742). We have now updated the previous Mendelian randomization analysis for 25(OH)D concentrations and risk of prostate cancer using a multi-SNP score of 6 SNPs and a combined sample size of 69,837 prostate cancer cases and 52,964 controls. Information on the associations of the 6 SNPs with 25(OH)D concentration is shown in table 1 https://drive.google.com/open?id=1Xzw0YRNOsVTEaZJvC5neYQSCS6A6uHZT.
In agreement with the previous publication (1), the odds ratio per 1 nmol/L increase in genetically determined logarithm transformed 25(OH)D concentrations was 0.95 (95% CI, 0.80-1.13) for total prostate cancer risk (Table 2 https://drive.google.com/open?id=1Xzw0YRNOsVTEaZJvC5neYQSCS6A6uHZT ). When we evaluated presence of horizontal pleiotropy by performing the MR-Egger regression method, horizontal pleiotropy was not evident, as the p-value for the intercept was large (P, 0.40) and the pleiotropy-adjusted estimates suggested null effects (Table 2). The weighted median method also yielded little evidence of any association (Table 2). When only the original 4 SNPs were used as genetic instruments, all results were very similar (data not shown). Our updated results provide little evidence of a linear causal association between genetically determined circulating vitamin D concentration and risk of prostate cancer. Future Mendelian randomization studies are warranted to estimate associations in important prostate cancer and participant subgroups or to revise current estimates as more genetic instruments for 25(OH)D concentrations become available.
1. Dimitrakopoulou VI, Tsilidis KK, Haycock PC, et al. Circulating vitamin D concentration and risk seven cancers: a Mendelian randomization study. BMJ 2017;359:j4761.
2. Schumacher FR, Al Olama AA, Berndt SI, et al. Prostate cancer meta-analysis of more than 140,000 men identifies 63 novel prostate cancer susceptibility loci. Nat Gen 2017 [in press].
3. Jiang X, O’Reilly PF, Aschard H, et al. Genome-wide association study in 79,366 European-ancestry individuals informs the genetic architecture of 25-hydroxyvitamin D levels. Nat Commun 2018 [in press].
Table 1: Characteristics of genetic variants associated with 25(OH)D concentration and total prostate cancer in published GWAS.
Circulating Vitamin D Levels Prostate Cancer
SNP Chromosome Locus Risk Allele Beta (SE)a P-value Beta (SE)b P-value
rs3755967c 4 GC C 0.089 (0.0023) 4.74E-343 -0.0066 (0.0096) 0.495
rs12785878 11 DHCR7/NADSYN1 T 0.036 (0.0022) 3.80E-62 -0.0057 (0.0097) 0.556
rs10741657 11 CYP2R1 A 0.031 (0.0022) 2.05E-46 -0.0002 (0.0088) 0.980
rs10745742 12 AMDHD1 T 0.019 (0.0020) 2.10E-20 0.0034 (0.0088) 0.700
rs8018720 14 SEC23A G 0.019 (0.0027) 1.11E-11 0.0183 (0.0117) 0.117
rs17216707d 20 CYP24A1 T 0.026 (0.0027) 8.14E-23 -0.0069 (0.0115) 0.550
a The beta estimates for continuous 25(OH)D are reported per unit change in natural-log transformed 25(OH)D (nmol/L) per effect allele.
b The beta estimates are calculated performing a fixed-effect meta-analysis of summary association results reported in Dimitrakopoulou, et al. (data from iCOGS and BPC3) and Schumacher, et al. (data from OncoArray).
c In linkage disequilibrium with rs2282679 (r2=0.99).
d In linkage disequilibrium with rs6013897 (r2=0.70).
Table 2: Mendelian randomization estimates between multi-SNP risk scores of continuous 25(OH)D and total prostate cancer risk.
Analysis method ORa 95% CI p-value
Inverse-variance weighted 0.948 0.796, 1.130 0.553
Maximum likelihood 0.948 0.795, 1.130 0.553
MR Egger 0.845 0.581, 1.230 0.305
Weighted Median 0.929 0.766, 1.127 0.455
Abbreviations: BCAC, Breast Cancer Association Consortium; PRACTICAL, Prostate Cancer Association Group to Investigate Cancer Associated Alterations in the Genome
a The odds ratios (ORs) represent increase/decrease of risk per unit increase in the natural-log transformed 25(OH)D (nmol/L).
The authors would like to thank the Prostate Cancer Association Group to Investigate Cancer Associated Alterations in the Genome (PRACTICAL) for providing GWAS summary estimates of prostate cancer, and the SUNLIGHT consortium for providing GWAS summary estimates of circulating vitamin D concentrations. All studies and funders for the prostate cancer genome-wide association analyses are listed in Schumacher FR et al. (2017) and in Dimitrakopoulou VI et al. (2017). All studies and funders for the vitamin D genome-wide association analyses are listed in Jiang et al. (2018). KKT was supported by the World Cancer Research Fund International Regular Grant Programme (WCRF 2014/1180). SL is supported by a related research grant from World Cancer Research Fund Internation (2015/1421). RMM is supported by a Cancer Research UK Programme Grant, the Integrative Cancer Epidemiology Programme (C18281/A19169), and the National Institute for Health Research (NIHR) Bristol Biomedical Research Centre.
Competing interests: No competing interests