Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Nov 7.
Published in final edited form as: Cancer Epidemiol Biomarkers Prev. 2015 Aug 25;24(11):1796–1800. doi: 10.1158/1055-9965.EPI-15-0543

Genome-wide association study of prostate cancer-specific survival

Robert Szulkin 1,2,*, Robert Karlsson 1, Thomas Whitington 1, Markus Aly 1, Henrik Gronberg 1, Rosalind A Eeles 3,4, Douglas F Easton 5, Zsofia Kote-Jarai 3, Ali Amin Al Olama 5, Sara Benlloch 5, Kenneth Muir 6,7, Graham G Giles 8,9, Melissa C Southey 10, Liesel M FitzGerald 8, Brian E Henderson 11, Fredrick R Schumacher 11, Christopher A Haiman 11, Csilla Sipeky 12, Teuvo LJ Tammela 14, Børge G Nordestgaard 15,16, Timothy J Key 17, Ruth C Travis 17, David E Neal 18,19, Jenny L Donovan 20, Freddie C Hamdy 21, Paul DP Pharoah 22, Nora Pashayan 23,22, Kay-Tee Khaw 24, Janet L Stanford 25,26, Stephen N Thibodeau 27, Shannon K McDonnell 27, Daniel J Schaid 27, Christiane Maier 28, Walther Vogel 29, Manuel Luedeke 28, Kathleen Herkommer 30, Adam S Kibel 31, Cezary Cybulski 32, Jan Lubiński 32, Wojciech Kluźniak 32, Lisa Cannon-Albright 33,34, Hermann Brenner 35,36,37, Volker Herrmann 35, Bernd Holleczek 38, Jong Y Park 39, Thomas A Sellers 39, Hui-Yi Lim 40, Chavdar Slavov 41, Radka P Kaneva 42, Vanio I Mitev 42, Amanda Spurdle 43, Manuel R Teixeira 44,45, Paula Paulo 44, Sofia Maia 44, Hardev Pandha 46, Agnieszka Michael 46, Andrzej Kierzek, on behalf of the PRACTICAL consortium46,**, Jyotsna Batra 47, Judith A Clements, on behalf of The Australian Prostate Cancer BioResource47,47,48, Demetrius Albanes 49, Gerald L Andriole 50, Sonja I Berndt 49, Stephen Chanock 49,51, Susan M Gapstur 52, Edward L Giovannucci 53,54, David J Hunter 55, Peter Kraft 55, Loic Le Marchand 56, Jing Ma 54,57,58, Alison M Mondul 49, Kathryn L Penney 54,57,58, Meir J Stampfer 58, Victoria L Stevens 52, Stephanie J Weinstein 49, Antonia Trichopoulou 59,60,61, Bas H Bueno-de-Mesquita 62,63,64,65, Anne Tjonneland 66, David G Cox, on behalf of the BPC3 consortium67,68, Lovise Maehle 69, Johanna Schleutker 12,13, Sara Lindström 55, Fredrik Wiklund 1
PMCID: PMC5674990  NIHMSID: NIHMS910841  PMID: 26307654

Abstract

Background

Unnecessary intervention and overtreatment of indolent disease are common challenges in clinical management of prostate cancer. Improved tools to distinguish lethal from indolent disease are critical.

Methods

We performed a genome-wide survival analysis of cause-specific death in 24,023 prostate cancer patients (3,513 disease-specific deaths) from the PRACTICAL and BPC3 consortia. Top findings were assessed for replication in a Norwegian cohort (CONOR).

Results

We observed no significant association between genetic variants and prostate cancer survival.

Conclusions

Common genetic variants with large impact on prostate cancer survival were not observed in this study.

Impact

Future studies should be designed for identification of rare variants with large effect sizes or common variants with small effect sizes.

Introduction

Prostate cancer is the second leading cause of cancer death among men in the developed world. Randomized trials have shown that PSA-based screening can reduce prostate cancer mortality up to 40%, though at the cost of considerable over-diagnosis and over-treatment of indolent disease(1). Thus, improved tools to distinguish lethal from indolent disease to guide clinicians in treatment decisions are critical. Epidemiological studies support the existence of a genetic component to prostate cancer prognosis(2). The purpose of this study was to identify Single Nucleotide Polymorphisms (SNPs) associated with prostate cancer specific survival. We performed a genome-wide search among individuals from two large prostate cancer genetics consortia (PRACTICAL(3) and BPC3(4) with replication of top findings in a Norwegian prostate cancer cohort (CONOR)(5).

Materials and Methods

Study populations and genotyping

In total, 24,023 prostate cancer patients with follow-up on cause specific death from the PRACTICAL (n = 21,241) and BPC3 (n = 2,782) consortia were included in the present study (Table 1). All men from BPC3 have an aggressive disease, defined by a tumor Gleason score of eight or above. Participants had either been genotyped on a custom designed SNP chip (iCOGS) with 211,155 markers or on standard genome-wide arrays (Table 1). Imputation was performed using a cosmopolitan panel from the 1000 Genomes Project (March 2012) to increase the genetic coverage. Only SNPs that had an imputation quality above 0.75 and minor allele frequency (MAF) above 1% were assessed (1.2–9.5 million SNPs in each separate study, Table 1). Detailed information regarding study populations, genotyping and imputation is found in (3) and (4).

Table 1.

Patient characteristics of included study populations.

Study N No of
prostate
cancer
deaths
Total person-
years
Person-years at
risk
median (min-
max)
Number of
SNPs
with
imputation
quality>=0.75

PRACTICAL
CAPS 412 49 3 476,6 9.3 (0.4–11.8) 5,752,274Ɨ
CAPS1 492 214 3 120,2 7.1 (0.1–11.8) 8,933,855$
CAPS2 1493 331 11 644,6 9.1 (0–11.8) 8,715,366$
CPCS 925 97 1 772,1 1.3 (0.1–18.1) 5,550,954Ɨ
EPIC 404 35 467,4 0 (0–13.2) 5,503,395Ɨ
ESTHER 300 22 2 179,9 7.7 (0.1–9.8) 5,495,692Ɨ
FHCRC 760 46 7 716,6 8.1 (0.2–18.1) 5,532,283Ɨ
MAYO 737 40 4 612,8 6.8 (0.1–13.9) 5,555,791Ɨ
MCCS_PCFS 1663 139 41 656,4 22.6 (0–71.2) 5,425,082Ɨ
MEC 581 15 3 531,5 5.8 (0.1–13.1) 5,462,203Ɨ
PCMUS 57 7 68,9 1.1 (0.1–4.0) 5,359,978Ɨ
SEARCH 1369 70 4 966,3 3.7 (0.1–4.5) 5,510,831Ɨ
STHM1 2199 71 8 247,2 3.9 (0–4.3) 5,724,947Ɨ
TAMPERE 2463 248 21 037,6 7.8 (0.8–20.9) 6,455,082Ɨ
UKGPCS 4344 826 22 906,2 4.3 (0–27.3) 5,485,041Ɨ
UKGPCS1 1783 457 13 689,0 7.0 (0.1–30.0) 9,536,409Ɣ
UKGPCS2 772 189 6 961,8 9.1 (0–24.9) 1,235,003Ϯ
ULM 365 32 3 151,9 9.0 (0.6–22.0) 5,457,321Ɨ
UTAH 122 27 603,7 4.0 (0.1–26.9) 5,641,408Ɨ

BPC3
ATBC 245 133 1 426,2 5.8 (0–19.9) 8,232,459§
CPSII 636 79 5 859,4 9.2 (0.3–16.3) 7,448,367§
EPIC 431 159 2 197,3 5.2 (0–14.3) 7,612,553§
HPFS 214 37 1 616,6 7.6 (0.1–14.4) 7,539,277§
MEC 244 23 1 868,3 7.7 (0.9–15.4) 7,571,269§
PHS 298 97 2 811,8 9.4 (0–24.7) 7,569,352§
PLCO 714 70 4 664,5 6.7 (0.1–12.9) 7,526,690§

Total 24023 3513 182 254,8

CONOR 1496 791 8741.4 5.0 (0.08–20.8)
Ɨ

Genotyped on a custum Illumina SNP infimum chip (iCOGS) with 211,155 SNPs, enriched in regions associated with incidence of prostate, breast and ovarian cancer.

Ɣ

Genotyped on Illumina Infinium HumanHap 550 Array.

Ϯ

Genotyped on Illumina iSELECT in 43,671 SNPs.

$

Genotyped on Affymetrix GeneChip 5.0K or 500K.

§

Genotyped on Illumina Human 610 or 610K.

Statistical analysis

Within each study, SNPs were assessed for association with disease survival, assuming an additive genetic effect, in a Cox regression model allowing for left truncation and right censoring of observational times. Results were combined in fixed-effects meta-analysis. In the discovery stage, we considered an association to be genome-wide significant if the overall meta-analysis achieved p < 5E-08 and the test for heterogeneity across studies was non-significant (p > 0.05). We also adjusted the most associated SNPs for population structure (principal components), age at diagnosis, diagnostic PSA and Gleason score but we did not observe any confounding (data not shown).

Replication

Genome-wide significant SNPs in the discovery stage were directly genotyped in 1,783 individuals from the UKGPCS1 study (Table 1) using TaqMan assays to verify imputation quality, evaluated as the concordance rate between imputed and genotyped data (percentage of individuals correctly classified by imputation). Significant SNPs from the discovery stage with satisfactory imputation qualities were assessed for replication in a Norwegian case-cohort study (CONOR(5)) comprising 1,496 prostate cancer cases of which 791 died due to prostate cancer during follow-up. Genotypes were derived through TaqMan assays and analyzed in a proportional hazards model for case-cohort designs(6) with adjustment for age at diagnosis.

Results

Among the 24,023 prostate cancer patients included in the discovery stage, we observed 3,513 deaths due to prostate cancer (Table 1). No inflation was observed in the combined meta-analysis (λ1000 = 1.02)(7). Ten SNPs reached genome-wide significance, two common variants (MAF 7–8%) and eight rare variants (MAF 1–2%, Table 2). Six of these SNPs failed genotyping in the UKGPCS1 sample (either because of unsuccessful assay design, failed clustering or monomorphism) while the remaining four SNPs (rs114997855 on chromosome 2, rs76010824 on chromosome 3, rs140659849 and rs723557 on chromosome X) had an excellent concordance rate (98–99%) between genotyped and imputed data. These four SNPs were put forward for replication in the Norwegian CONOR cohort. None of the four SNPs showed any evidence of association in the Norwegian cohort (p > 0.05) and inclusion of these results in the meta-analysis resulted in non-genome-wide significance levels for each SNP (Table 2).

Table 2.

Genome-wide assessment of prostate cancer survival.

PRACTICAL and BPC3
CONOR
All studies§
SNP
CHR:BP
AllelesϮ
MAF
Total number
No of
PC/deaths
HR (95% CI)
P-value
HR (95% CI)
P-value
HR (95% CI)
P-value
rs190087062 G/A 2,416/704 2.83 (1.99–4.02)
1:115063785 0.02 6,5E-09
rs114997855 A/G 20,051/2,729 1.75 (1.44–2.13) 0.88 (0.42–1.85) 1.67 (1.38–2.03)
2:30622824 0.02 2,6E-08 0,73 1,20E-07
rs76010824 A/G 23,251/3,324 1.29 (1.18–1.41) 1.01 (0.76–1.35) 1.26 (1.16–1.38)
3:67442642 0.07 2,8E-08 0,94 1,10E-07
rs184342703 T/C 6,812/832 2.36 (1.73–3.20)
4:135989066 0.02 4,2E-08
rs192864713 G/A 1,738/464 3.54 (2.31–5.43)
5:27429220 0.01 7,3E-09
rs111414857 G/A 17,146/2,236 1.98 (1.56–2.50)
7:126639415 0.01 1,7E-08
rs149470135 A/T 4,725/599 3.09 (2.09–4.59)
8:86472701 0.01 2,0E-08
rs117643112 C/A 6,306/1,577 1.93 (1.53–2.43)
12:81746712 0.02 3,1E-08
rs140659849ʔ A/G 2,702/271 3.00 (2.06–4.36) 0.75 (0.24–2.33) 2.61 (1.83–3.73)
X:50194937 0.01 9,6E-09 0,62 1,20E-07
rs723557ɣ G/T 23,251/3,324 1.17 (1.10–1.24) 1.00 (0.84–1.19) 1.15 (1.09–1.22)
X:126653357 0.08 1,5E-07 0,98 6,10E-07

Abbreviations: CHR=Chromosome, BP=Base position (Genome build 37), MAF=Minor allele frequency, HR=Hazard ratio, 95% CI=95% confidence interval.

Ϯ

Minor allele/Major allele. Minor allele used as effect allele (major as reference) in analysis.

§

Meta-analysis between PRACTICAL, BPC3 and CONOR.

ɣ

Proxy for rs13440791 (p=2.7E-08 in PRACTICAL and BPC3).

ʔ

Proxy for rs190977150 (p=9.5E-09 in PRACTICAL and BPC3).

Discussion

We performed a genome-wide search for SNPs associated with prostate cancer survival by combining data from the PRACTICAL and BPC3 consortia. Our null finding is in line with previous smaller studies(8) and implicates that the existence of common genetic variants with large effect sizes is unlikely. We would however like to stress that our analysis was based on imputed data and some areas of the genome were not well represented due to a low number of SNPs with good imputation quality.

Despite a reasonably large replication sample we saw no evidence of association among the four SNPs that were initially found to be genome-wide significant (p<5E-08). Two of these SNPs were rare, in which spurious associations occur more easily. It is however more surprising that the two common SNPs (MAF=7–8%) were false positives. This underlines the importance of independent replication in genetic association studies.

From this study, we conclude that the search for SNPs that are associated with prostate cancer survival should focus on the identification of rare variants with large effect sizes or common variants with small effect sizes. Large study populations with complete follow-up information regarding survival are warranted to successfully achieve this task.

Supplementary Material

Supplementary acknowledgements

Acknowledgments

Acknowledgments are found in the supplementary notes.

Financial support:

Fredrik Wiklund was recipient of the Swedish Cancer Society grant 2012/823 and Swedish Research Council grant 2014/2269.

Footnotes

Conflict of Interest: There is no conflict of interest.

References

  • 1.Hugosson J, Carlsson S, Aus G, Bergdahl S, Khatami A, Lodding P, et al. Mortality results from the Goteborg randomised population-based prostate-cancer screening trial. Lancet Oncol. 2010;11:725–32. doi: 10.1016/S1470-2045(10)70146-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Lindstrom LS, Hall P, Hartman M, Wiklund F, Gronberg H, Czene K. Familial concordance in cancer survival: a Swedish population-based study. Lancet Oncol. 2007;8:1001–6. doi: 10.1016/S1470-2045(07)70282-6. [DOI] [PubMed] [Google Scholar]
  • 3.Al Olama AA, Kote-Jarai Z, Berndt SI, Conti DV, Schumacher F, Han Y, et al. A meta-analysis of 87,040 individuals identifies 23 new susceptibility loci for prostate cancer. Nat Genet. 2014;46:1103–9. doi: 10.1038/ng.3094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Schumacher FR, Berndt SI, Siddiq A, Jacobs KB, Wang Z, Lindstrom S, et al. Genome-wide association study identifies new prostate cancer susceptibility loci. Hum Mol Genet. 2011;20:3867–75. doi: 10.1093/hmg/ddr295. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Naess O, Sogaard AJ, Arnesen E, Beckstrom AC, Bjertness E, Engeland A, et al. Cohort profile: cohort of Norway (CONOR) International journal of epidemiology. 2008;37:481–5. doi: 10.1093/ije/dym217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Therneau TM, Li H. Computing the Cox model for case cohort designs. Lifetime data analysis. 1999;5:99–112. doi: 10.1023/a:1009691327335. [DOI] [PubMed] [Google Scholar]
  • 7.de Bakker PI, Ferreira MA, Jia X, Neale BM, Raychaudhuri S, Voight BF. Practical aspects of imputation-driven meta-analysis of genome-wide association studies. Hum Mol Genet. 2008;17:R122–8. doi: 10.1093/hmg/ddn288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Penney KL, Pyne S, Schumacher FR, Sinnott JA, Mucci LA, Kraft PL, et al. Genome-wide association study of prostate cancer mortality. Cancer Epidemiol Biomarkers Prev. 2010;19:2869–76. doi: 10.1158/1055-9965.EPI-10-0601. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary acknowledgements

RESOURCES