Pubertal development and prostate cancer risk: Mendelian randomization study in a population-based cohort

Carolina Bonilla, Sarah J. Lewis, Richard M. Martin, Jenny L. Donovan, Freddie C. Hamdy, David E. Neal, Rosalind Eeles, Doug Easton, Zsofia Kote-Jarai, Ali Amin Al Olama, Sara Benlloch, Kenneth Muir, Graham G. Giles, Fredrik Wiklund, Henrik Gronberg, Christopher A. Haiman, Johanna Schleutker, Børge G. Nordestgaard, Ruth C. Travis, Nora Pashayan, Kay-Tee Khaw, Janet L. Stanford, William J. Blot, Stephen Thibodeau, Christiane Maier, Adam S. Kibel, Cezary Cybulski, Lisa Cannon-Albright, Hermann Brenner, Jong Park, Radka Kaneva, Jyotsna Batra, Manuel R. Teixeira, Hardev Pandha, Mark Lathrop, George Davey Smith, The PRACTICAL consortium

Epidemiological studies have observed a positive association between an earlier age at sexual development and prostate cancer, but markers of sexual maturation in boys are imprecise and observational estimates are likely to suffer from a degree of uncontrolled confounding. To obtain causal estimates, we examined the role of pubertal development in prostate cancer using genetic polymorphisms associated with Tanner stage in adolescent boys in a Mendelian randomization (MR) approach. We derived a weighted genetic risk score for pubertal development, combining 13 SNPs associated with male Tanner stage. A higher score indicated a later puberty onset. We examined the association of this score with prostate cancer risk, stage and grade in the UK-based ProtecT case-control study (n = 2,927), and used the PRACTICAL consortium (n = 43,737) as a replication sample. In ProtecT, the puberty genetic score was inversely associated with prostate cancer grade (odds ratio (OR) of high- vs. low-grade cancer, per tertile of the score: 0.76; 95 % CI, 0.64-0.89). In an instrumental variable estimation of the causal OR, later physical development in adolescence (equivalent to a difference of one Tanner stage between pubertal boys of the same age) was associated with a 77 % (95 % CI, 43-91 %) reduced odds of high Gleason prostate cancer. In PRACTICAL, the puberty genetic score was associated with prostate cancer stage (OR of advanced vs. localized cancer, per tertile: 0.95; 95 % CI, 0.91-1.00) and prostate cancer-specific mortality (hazard ratio amongst cases, per tertile: 0.94; 95 % CI, 0.90-0.98), but not with disease grade. Older age at sexual maturation is causally linked to a reduced risk of later prostate cancer, especially aggressive disease.