Concordance of Non–Low-Risk Disease Among Pairs of Brothers With Prostate Cancer
Prostate cancer among first-degree relatives is a strong risk factor for diagnosis of prostate cancer, and the contribution of heritable factors in prostate cancer etiology is high. We investigated how the concordance of non–low-risk prostate cancer among brothers is affected by their genetic relation.
We identified 4,262 pairs of brothers with prostate cancer in the Prostate Cancer Database Sweden. Their cancers were categorized as low risk (Gleason score ≤ 6; clinical stage T1-2, Nx/N0, Mx/M0; and prostate-specific antigen ≤ 10 ng/mL) or non–low risk. The odds ratio (OR) for concordance of non–low-risk cancer was calculated with logistic regression for the different types of fraternity (monozygotic twins, dizygotic twins, full brothers, and half-brothers)
Among monozygotic twins who both were diagnosed with prostate cancer, the OR for both brothers being in the non–low-risk category was 3.82 (95% CI, 0.99 to 16.72) after adjusting for age and year of diagnosis. Among full brothers, the corresponding adjusted OR was 1.21 (95% CI, 1.04 to 1.39). When the analysis was restricted to brothers who both were diagnosed within 4 years, the results were similar.
Non–low-risk prostate cancer has a heritable pattern suggesting shared genetic factors, with the highest concordance among monozygotic twins. Our results suggest that a man whose brother has been diagnosed with a non–low-risk prostate cancer is at a clinically relevant increased risk of developing an aggressive prostate cancer himself.
Familial occurrence aggregation of prostate cancer is one of the strongest risk factors for prostate cancer.1 Men with one or two first-degree relatives with prostate cancer have a two- to five-fold increased risk of being diagnosed with prostate cancer compared with men with no family history of prostate cancer,2,3 and the risk is also increased among men with second- and third-degree relatives with prostate cancer.4 In twin studies, prostate cancer had the highest heritability.5,6 A few rare high-penetrance prostate cancer susceptibility alleles have been identified, but in general, prostate cancer is genetically multifactorial, and most risk alleles individually contribute with little effect.7 As a result of the risk of overtreatment of low-risk prostate cancer, public health organizations advocate, with two exceptions,8,9 against public screening programs. However, men with a family history of prostate cancer are recommended to undergo prostate-specific antigen (PSA) testing by many professional organizations.10,11
Previous twin studies have assessed the concordance of a prostate cancer diagnosis but not the concordance of the type of prostate cancer.6,12 If there is prognostic concordance within families, then knowledge of tumor characteristics and disease development among relatives may be helpful when counseling men with suspected or newly diagnosed prostate cancer. We have previously reported estimates for concordance in Gleason score among pairs of full brothers13 and on how family history influences the age-specific probabilities for different prostate cancer risk categories.14 In the current study, we linked the National Prostate Cancer Register, the Swedish Twin Register, and the Swedish Multi-Generation Register to investigate how the concordance of non–low-risk prostate cancer within brother pairs is affected by their genetic relation (monozygotic twins, dizygotic twins, full brothers, or half-brothers).
The National Prostate Cancer Register registers clinical data on all patients with prostate cancer in Sweden. It started as a collaboration among six regions in Sweden in 1996 and became nationwide in 1998. The coverage is approximately 97% of all prostate cancers.15 In total, > 400 variables related to diagnosis, tumor characteristics, treatment, and follow-up are registered.
The Multi-Generation Register provides information on parents of all individuals born in Sweden from 1932 onward. Before 1991, deceased parents were under-reported because dead persons were not registered as parents, but the register has complete parental information for close to 100% of the index participants in the current study.16
The Swedish Twin Register was founded at the end of the 1950s and holds records of twins born in Sweden since 1886.17 The register includes data on approximately 85,000 twin pairs with information on whether they were considered monozygotic or dizygotic twins. For living individuals, information is collected through surveys and automatic updates from population registers.
The data in the current study are provided by Prostate Cancer Database Sweden (PCBaSe), which is a linkage between the National Prostate Cancer Register and several other national databases, including those mentioned earlier.18 The most recent version of PCBaSe (version 3.0) includes data from approximately 150,000 patients with prostate cancer. In PCBaSe, the personal identity number has been replaced by a code, and the code key is kept at the Swedish National Board of Health and Welfare. The Research Ethical Board of Umeå University (EPN 2013/153-31) approved the study.
The study included 4,262 pairs of brothers who were both diagnosed with prostate cancer (Table 1). The brother pairs were identified in PCBaSe. Brothers sharing both parents were categorized as full brothers, and those sharing either their father or mother only were categorized as half-brothers. The brother first diagnosed with prostate cancer within a family was considered the index case. In families with more than two brothers with prostate cancer, only the index case and the second brother diagnosed were included. This restriction was to limit the potential bias from large age differences between brothers, especially between paternal half-brothers. In families with concordant twin pairs, the first twin diagnosed was always considered the index case and his twin brother the second case, to make sure that no cases of concordant twins were lost as a result of the restriction described earlier. The 102 concordant twin pairs were identified using the Swedish Twin Register instead of the Multi-Generation Register. There may be a few more concordant twin pairs of unknown zygosity in the National Prostate Cancer Register.
The fraternity status as full brother or half-brother could not be determined for 330 pairs of brothers, and these pairs were excluded from the analysis. We also excluded six pairs of twins concordant for prostate cancer but with unknown zygosity. We categorized the cancers as either low risk (TNM stage T1-2, N0/Nx, M0/Mx; Gleason score ≤ 6; and PSA < 10 ng/mL) or non–low risk (all others).
Using logistic regression, we calculated crude and adjusted odds ratios (ORs) for the association of within-pair concordance for low- or non–low-risk prostate cancer at diagnosis. The resulting point estimates reflect the interaction term that describes ORs for sharing non–low-risk tumor characteristics by different types of fraternity, given that both brothers were diagnosed with prostate cancer. The point estimates were adjusted for calendar year of diagnosis and age at diagnosis.
Heritability is defined as the proportion of variance in phenotype explained by variance in genotype. The contribution of variance to a specific trait can be purely genetic, environmental, or an interaction between environmental and genetic factors. We used a standard quantitative genetic model to assess heritability, defined as the proportion of phenotypic variance caused by additive genetic effects.19,20 The model assumed that the concordance between brothers was related to genetic factors that only acted additively on their probability to develop non–low-risk prostate cancer. If the concordance correlation is influenced by shared nongenetic factors or if the genetic effects are not additive, then the model will overestimate the heritability. All patients were put into a contingency table by risk group (low risk or non–low risk). Polychoric correlations were calculated with SEs. The polychoric correlations are by themselves not intuitively interpretable but are used to estimate heritability. Crude estimates on heritability are obtained as follows:
where k = 1 for monozygotic twins, k = 0.5 for full brothers and dizygotic twins, and k = 0.25 for half-brothers, on the basis of the average proportion of the genome shared by descent. If the calculated value of the heritability was < 0 or > 1, which is biologically impossible, the value was replaced with 0 or 1.
Missing data in the variables used to assign the risk category (PSA, Gleason score, and stage) were imputed using multiple imputation by chained equation.21-23 The technique assumes that data are missing at random. Each variable with missing data was regressed as a dependent variable with all of the other variables acting as predictors. For each cycle of regression, the data set was updated with imputed values. The multiple imputation by chained equation algorithm was repeated five times to produce five data sets. The data sets were then pooled to create the final imputed data set.
The median of within-pair time differences between pairs of different fraternity type was tested using the Wilcoxon rank sum test. To address possible bias stemming from a potential change in clinical and histopathologic assessment of prostate cancer over time, we performed a sensitivity analysis in brother pairs in which the second brother was diagnosed < 4 years after the first. All statistical analyses were computed using R software.24
Of 4,262 concordant brother pairs, assignment of risk category was possible for 3,556 pairs of full brothers, 57 monozygotic twin pairs, 34 dizygotic twin pairs, 38 pairs of paternal half-brothers, and 45 pairs of maternal half-brothers. Table 2 lists numbers of concordant pairs of brothers, with their cancers categorized as low or non–low risk. Figure 1 presents a forest plot of the ORs for sharing non–low-risk status among brothers who were both diagnosed with prostate cancer. The adjusted OR for full brothers was 1.21 (95% CI, 1.05 to 1.40). Among monozygotic twins, the OR for both being in the non–low-risk category was 3.82 (95% CI, 0.99 to 16.72). The association remained after adjustment for T stage and PSA in separate analyses (data not shown).
The ORs were similar in the analysis restricted to brothers who were diagnosed within 4 years. This analysis included 77% of the monozygotic twin pairs, 68% of the dizygotic twin pairs, and 57% of the full brother pairs. Because imputation did not change the estimates, the nonimputed results are presented. The imputed results can be found in Appendix Table A1 (online only).
Data on number of biopsies are available from 2008. In a subset analysis of 872 pairs of full brothers adjusted for the number of biopsies, the OR for non–low-risk prostate cancer was 1.11 (95% CI, 0.8 to 1.53).
Crude estimates of heritability were 45% (approximate 95% CI, 7% to 82%) for monozygotic twins and 16% (approximate 95% CI, 6% to 26%) for full brothers (Table 3). Estimates were also calculated for the individual factors used for the risk category classification (Gleason score, PSA, and T stage). For full brothers, positive correlations were also found for the individual factors (Gleason score, PSA, and T stage) at about the same level as for risk category.
The median time between the diagnoses was significantly shorter for monozygotic twins (1.1 years) than for full brothers (3.2 years; P < .001), dizygotic twins (2.8 years; P < .014), paternal half-brothers (4.1 years; P < .001), and maternal half-brothers (3.0 years; P < .003).
We found that among brothers both diagnosed with prostate cancer, if the first brother was diagnosed with non–low-risk disease, the second brother’s disease was also likely to be non–low risk (OR, 1.21). The OR for monozygotic twins was higher (3.82), although with a wide CI (95% CI, 0.99 to 16.72). This indicates that shared genes cause shared tumor characteristics and prognosis in families afflicted by prostate cancer. This is consistent with our previous finding of Gleason score concordance within affected brother pairs.13 The current study adds detailed information on the brothers’ fraternity. By linking the nationwide Swedish National Prostate Cancer Register data with the Swedish Twin Register and the Multi-Generation Register, we were able to rank brother pairs by genetic similarity and to analyze the concordance of non–low-risk prostate cancer by type of fraternity.
The ORs presented here should not be confused with relative risks of incident cancer for men with a family history of prostate cancer. The OR estimates correspond to the additional relative risk for non–low-risk prostate cancer, given that both brothers are diagnosed. Therefore, to obtain the overall relative risk compared with an individual without an affected brother, these odds ratios should be multiplied by the increased baseline relative risk25 associated with familial prostate cancer occurrence.
PSA testing has been reported to be more common among men with relatives with prostate cancer.25 The shorter latency between diagnoses among monozygotic twins in our data suggests that their entirely shared genome predisposes them to develop prostate cancer at the same age or that monozygotic twin brothers are likely to obtain a PSA test shortly after a prostate cancer diagnosis in their twin. Because PSA testing predominantly leads to the detection of low-risk cancer, the ORs for concordance of non–low-risk cancer calculated in the current study might have been falsely low.
Because of the limited sample size, we categorized prostate cancer as low risk or non–low risk only. Although this dichotomy roughly divides the cancers into those that usually need to be treated and those that may be managed with active surveillance or watchful waiting, the non–low-risk group is clinically heterogeneous. This means that there may be higher concordance for some specific cancer types within the non–low-risk group. We have previously reported that brothers of men with high-grade prostate cancer are at a particularly high risk of a diagnosis of a similar high-grade cancer, compared with men in the general population.13 Our findings in the current study are in agreement with these previous findings, but the outcome measure was different, the study population was considerably larger, and the brothers’ fraternity was characterized.
Two previous studies with data from Nordic twin registers have estimated heritability and the influence of shared environment for prostate cancer.6,12 The twin study by Lichtenstein et al12 investigated the concordance of many cancer types with combined data from the Swedish, Finnish, and Danish twin registers. The heritability for prostate cancer was estimated to be 42% on the basis of 40 monozygotic and 20 dizygotic concordant twin pairs, which was greater than for any other type of cancer. Hjelmborg et al6 further investigated cancer heritability by adding the Norwegian twin cohort and another 10 years of follow-up. In this study, the heritability for prostate cancer was estimated to be 58% on the basis of 194 monozygotic and 146 dizygotic twin pairs. In our study, the number of affected twin pairs and, thus, the statistical precision were lower. We used a less precise method for calculating heritability because we assumed that the similarities between the brother pairs were attributable to additive gene effects only. This is a strongly reductionist assumption that likely overestimated the genetic effect and disregarded genetic interactions, which are much more common in monozygotic twins than in other brother pairs. Given the wide CIs, however, we chose to not refine the model further and conclude that our data did not suggest a strong heritability for non–low-risk prostate cancer.
Lindström et al26 reported that mortality among sons of fathers who died of prostate cancer within 10 years after diagnosis was two-fold greater than that of men with fathers with a good prognosis. The present data are not yet mature enough for analyses of mortality.
In addition to sharing genes, twins share intrauterine and early-life environment to a larger extent than nontwin brothers. Monozygotic twins also share placenta, with resulting similar effects of nutrition and hormonal milieu during early organ development. On the basis of several studies of in utero factors, high birth weight has been associated with increased risk of prostate cancer.27-29 In an intriguing analysis by Cnattingius et al,30 high birth weight was associated with increased prostate cancer risk among dizygotic but not monozygotic twins, suggesting that intrauterine growth influences the prostate cancer risk more in individuals who share less genetic material or who have not shared placenta. The mechanisms behind these associations are unclear, but in the context of this study, it should be noted that monozygotic twins are more similar than nontwin brother pairs in many other ways than just by having identical genes.
In conclusion, in this study, we hypothesized a priori a concordance of non–low-risk prostate among brothers in general and among monozygotic twins in particular. Our data support this hypothesis, although with wide CIs and a borderline significant estimate for monozygotic twins. Our results suggest that a man whose brother has been diagnosed with a non–low-risk prostate cancer is at a clinically relevant increased risk of developing an aggressive prostate cancer himself.
Supported by Strategic Research Programme in Cancer (StratCan) at Karolinska Institutet, the Swedish Cancer Society (Grant No. CAN 2011/825), and the Foundation Johanna Hagstrand och Sigfried Linnérs Minne (Grant No. D110458013).
Conception and design: Fredrik Jansson, Thomas Frisell, Pär Stattin, Ola Bratt, Olof Akre
Financial support: Pär Stattin, Olof Akre
Administrative support: Pär Stattin
Provision of study materials or patients: Pär Stattin
Collection and assembly of data: Fredrik Jansson, Linda Drevin, Pär Stattin
Data analysis and interpretation: All authors
Manuscript writing: All authors
Final approval of manuscript: All authors
Accountable for all aspects of the work: All authors
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.
No relationship to disclose
No relationship to disclose
No relationship to disclose
No relationship to disclose
No relationship to disclose
Consulting or Advisory Role: Astellas Pharma, Bayer
This project was made possible by the continuous work of the National Prostate Cancer Register of Sweden Steering Group, whose members include Pär Stattin (chairman), Anders Widmark, Camilla Thellenberg, Ove Andrén, Ann-Sofi Fransson, Magnus Törnblom, Stefan Carlsson, Marie Hjälm-Eriksson, David Robinson, Mats Andén, Jonas Hugosson, Ingela Franck Lissbrant, Maria Nyberg, Ola Bratt, René Blom, Lars Egevad, Calle Walller, Olof Akre, Per Fransson, Eva Johansson, Fredrik Sandin, and Karin Hellström.
|1.||Damber J-E, Aus G: Prostate cancer. Lancet 371:1710-1721, 2008 Crossref, Medline, Google Scholar|
|2.||Hemminki K: Familial risk and familial survival in prostate cancer. World J Urol 30:143-148, 2012 Google Scholar|
|3.||Kiciński M, Vangronsveld J, Nawrot TS: An epidemiological reappraisal of the familial aggregation of prostate cancer: A meta-analysis. PLoS One 6:e27130, 2011 Google Scholar|
|4.||Albright F, Stephenson RA, Agarwal N, et al: Prostate cancer risk prediction based on complete prostate cancer family history. Prostate 75:390-398, 2015 Crossref, Medline, Google Scholar|
|5.||Lichtenstein P, De Faire U, Floderus B, et al: The Swedish Twin Registry: A unique resource for clinical, epidemiological and genetic studies. J Intern Med 252:184-205, 2002 Google Scholar|
|6.||Hjelmborg JB, Scheike T, Holst K, et al: The heritability of prostate cancer in the Nordic Twin Study of Cancer. Cancer Epidemiol Biomarkers Prev 23:2303-2310, 2014 Crossref, Medline, Google Scholar|
|7.||Eeles RA, Olama AA, Benlloch S, et al: Identification of 23 new prostate cancer susceptibility loci using the iCOGS custom genotyping array. Nat Genet 45:385-391, 2013 Google Scholar|
|8.||Gondos A, Krilaviciute A, Smailyte G, et al: Cancer surveillance using registry data: Results and recommendations for the Lithuanian national prostate cancer early detection programme. Eur J Cancer 51:1630-1637, 2015 Crossref, Medline, Google Scholar|
|9.||Ishkinin Y, Zhylkaidarova A, Nurgaliyev N, et al: Population-based prostate cancer screening in Kazakhstan. Iran J Public Health 46:917-922, 2017 Medline, Google Scholar|
|10.||European Association of Urology: Uroweb. http://uroweb.org/ Google Scholar|
|11.||National Comprehensive Cancer Network: NCCN homepage. https://www.nccn.org/ Google Scholar|
|12.||Lichtenstein P, Holm NV, Verkasalo PK, et al: Environmental and heritable factors in the causation of cancer: Analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med 343:78-85, 2000 Crossref, Medline, Google Scholar|
|13.||Jansson KF, Akre O, Garmo H, et al: Concordance of tumor differentiation among brothers with prostate cancer. Eur Urol 62:656-661, 2012 Crossref, Medline, Google Scholar|
|14.||Bratt O, Drevin L, Akre O, et al: Family history and probability of prostate cancer, differentiated by risk category: A nationwide population-based study. J Natl Cancer Inst 108:djw110, 2016 Crossref, Medline, Google Scholar|
|15.||Nationella Prostatacancerregistret: Homepage. http://npcr.se/ Google Scholar|
|16.||Ekbom A: The Swedish Multi-generation Register. Methods Mol Biol 675:215-220, 2011 Google Scholar|
|17.||Karolinska Institutet: Svenska Tvillingregistret. http://ki.se/forskning/svenska-tvillingregistret Google Scholar|
|18.||Van Hemelrijck M, Wigertz A, Sandin F, et al: Cohort profile: The National Prostate Cancer Register of Sweden and Prostate Cancer data Base Sweden 2.0. Int J Epidemiol 42:956-967, 2013 Crossref, Medline, Google Scholar|
|19.||Falconer DS: The inheritance of liability to certain diseases, estimated from the incidence among relatives. Ann Hum Genet 29:51-76, 1965 Crossref, Google Scholar|
|20.||Falconer DS, Mackay T: Introduction to Quantitative Genetics (ed 4). New York, NY, Pearson, 1996 Google Scholar|
|21.||Azur MJ, Stuart EA, Frangakis C, et al: Multiple imputation by chained equations: What is it and how does it work? Int J Methods Psychiatr Res 20:40-49, 2011 Crossref, Medline, Google Scholar|
|22.||van Buuren S, Groothuis-Oudshoorn K: mice: Multivariate imputation by chained equations in R. J Stat Soft 45:1-67, 2011 Google Scholar|
|23.||Morisot A, Bessaoud F, Landais P, et al: Prostate cancer: Net survival and cause-specific survival rates after multiple imputation. BMC Med Res Methodol 15:54, 2015 Crossref, Medline, Google Scholar|
|24.||R: The R Project for Statistical Computing. https://www.r-project.org/ Google Scholar|
|25.||Bratt O, Garmo H, Adolfsson J, et al: Effects of prostate-specific antigen testing on familial prostate cancer risk estimates. J Natl Cancer Inst 102:1336-1343, 2010 Crossref, Medline, Google Scholar|
|26.||Lindström LS, Hall P, Hartman M, et al: Familial concordance in cancer survival: A Swedish population-based study. Lancet Oncol 8:1001-1006, 2007 Crossref, Medline, Google Scholar|
|27.||Ekbom A, Hsieh CC, Lipworth L, et al: Perinatal characteristics in relation to incidence of and mortality from prostate cancer. BMJ 313:337-341, 1996 Crossref, Medline, Google Scholar|
|28.||Tibblin G, Eriksson M, Cnattingius S, et al: High birthweight as a predictor of prostate cancer risk. Epidemiology 6:423-424, 1995 Crossref, Medline, Google Scholar|
|29.||Lope V, García-Esquinas E, Ruiz-Dominguez JM, et al: Perinatal and childhood factors and risk of prostate cancer in adulthood: MCC-Spain case-control study. Cancer Epidemiol 43:49-55, 2016 Crossref, Medline, Google Scholar|
|30.||Cnattingius S, Lundberg F, Sandin S, et al: Birth characteristics and risk of prostate cancer: The contribution of genetic factors. Cancer Epidemiol Biomarkers Prev 18:2422-2426, 2009 Crossref, Medline, Google Scholar|