2.1

Literature Search Strategy

OVID was used to systematically search MEDLINE and EMBASE databases for articles evaluating risk of PCa detection in association with race or ethnicity and genetic factors. Databases were searched for studies published from January 2013 through September 2023 (Week 36). In addition to the MEDLINE and EMBASE databases search, reference lists of included systematic reviews and primary literature were scanned for potentially useful studies.

2.2

Study Selection Criteria and Process

All hits from the OVID literature search were input into reference management software (EndNote 21), where duplicate citations were removed. Abstracts were reviewed by the methodologist to determine if the study addressed the PICO (Patient, Intervention, Comparison and Outcome; Table S1 ) criteria. Randomized controlled trials, observational studies, modelling studies, and case-control studies with at least 30 patients per study arm were considered for inclusion in the evidence base. Case series, letters, editorials, in vitro studies, studies conducted in animal models, and studies not published in English language were excluded from the evidence base a priori. Full-text review was conducted on studies that passed the abstract screening phase by all authors. Studies were compared to the PICO criteria by one author and inclusion criteria was verified by the methodologist. The results are limited by numerous manuscripts providing results without distinction between clinically significant versus indolent PCa. When specified in the original manuscripts, more specific terminology was used. Low-risk PCa was defined as GG1 while csPCa or high-risk PCa was defined as GG≥2.

2.3

Data Extraction

Data were extracted from all studies that passed full-text review by the methodologist.

2.4

Quality Assessment: Individual Study Quality and Potential for Bias

Quality assessment for retained studies was conducted. Studies deemed to be low quality would not be excluded from the systematic review, but would be retained, and their methodological strengths and weaknesses discussed where relevant. To define an overall study quality rating for each included study, risk of bias as determined by validated study-type specific tools was paired with additional important quality features. To evaluate the risk of bias within the identified studies, the Assessment of Multiple Systematic Reviews (AMSTAR) [ ] tool was used for systematic reviews and a Risk of Bias in Non-Randomized Studies – of Intervention (ROBINS-I) [ ] was used for observational studies, and Prediction model Risk of Bias aSsessment Tool (PROBAST) was used for model studies [ ]. Summary risk of bias assessment figures were created using robvis [ ]. Additional important quality features, such as study design, comparison type, power of statistical analysis, and sources of funding were extracted for each study. We note that nearly all included studies have moderate or high risks of bias.

3.1

Family History

Family history (FH) of PCa is associated with being diagnosed with the disease. There were four studies included for FH [ ] ( Table S2 ). All included studies were limited by an overall high risk of bias (summarized in Figures S1-2 ). FH-positive status was based on first-degree relatives with PCa (defined as ≥GG1) in all studies, while one study [ ] also included second-degree relatives. A positive FH is associated with a 1.2-8.4-times odds of detecting csPCa on subsequent biopsy [ , ]. Prostate cancer detection was higher in FH-positive men, with risk increasing with additional first-degree relatives with the disease [ ]. Family history has also been used to improve predictive risk models [ , , ] ( Table S2 ).

3.2

Race/Ethnicity

Ten studies evaluating the association between race or ethnicity and risk of PCa detection or outcomes were included in the evidence base [ , ]. All studies carried an overall critical risk of bias based upon critical risk of bias in patient selection across all studies, plus several studies were also limited by moderate risk in other bias domains (summarized in Figure S1 ).

3.3

Genetic Risk: Pathogenic Variants (PV)

Ten studies evaluated the association between pathogenic germline variants and PCa [ , ]. They included one systematic review with meta-analysis [ ] limited by an overall moderate risk of bias, seven cohort studies carrying an overall moderate [ , ] or critical risk of bias [ ] (summarized in Figure S1 ), and two modeling studies limited by an overall high risk of bias [ , ] (summarized in Figure S2 ).

3.4

Polygenic Risk Score (PRS)

Thirteen observational cohort studies [ , , , ] and eleven modeling studies [ , , ] examined the role of PRS in predicting PCa. Of the total 13 observational studies, one [ ] carried an overall moderate risk of bias, six [ , , , , , ] were at serious risk, and six [ , , , ] were at critical risk of bias (summarized in Figures S1-2 ). While one modeling study [ ] carried low risk of bias, the remaining ten [ , , , , ] were at high risk of bias (summarized in Figure S2).

4.1

Family History

A positive PCa FH is a well-established risk factor for developing PCa. In this systematic review, we found evidence that a positive FH increases odds of csPCa diagnosis by 1.3-8.4-fold [ ]. This risk increases further with ≥2 first-degree relatives with PCa [ , ]. Patients with second-degree FH of PCa also have increased risk of PCa, though this risk is lower than in first-degree relatives [ ]. Interestingly, Castro et al. (2016) found 52% of FH-positive patients diagnosed with PCa had PSA level <3.0 ng/mL and otherwise would not have undergone a prostate biopsy, thus arguing that prostate biopsy is a feasible screening tool for patients with FH of PCa regardless of PSA [ ]. The four selected FH studies were limited by FH serving as a secondary outcome, patient reporting bias, and lack of granularity regarding GG, pathology, and age at diagnosis of PCa-positive family members.

While family history may be viewed as a proxy for genetic factors, in many cases it also comprises shared environmental factors, such as diet, stress, socioeconomic status (SES), neighborhood deprivation, and resource access amongst other factors [ ]. Documenting PCa FH is important when counseling patients on their risk—it is oftentimes more readily available and comes without the financial costs of genetic testing. However, in instances where FH is incomplete and/or unknown, genetic testing may close the gap.

4.2

Race/Ethnicity

The association between race/ethnicity and PCa has been extensively studied, demonstrating higher rates of PCa incidence, aggressiveness, younger age at diagnosis, and worse outcomes in Black patients and lower risk of PCa in Asian patients [ , , , ]. While Giri et al. (2022) found no differences in GS between Black and White races, this could be explained by significantly greater rates of PCa FH in White men in this cohort [ ]. This has prompted PCa screening guideline adjustments and community efforts to increase PCa screening education and participation in racial and ethnic minority populations to eliminate related disparities [ , ].

Nevertheless, race is a social-construct and it’s important to recognize the limitations of race-based medicine. Race and ethnicity are limited proxies for genetic contribution to disease development as they lack nuance regarding true contributions of ancestral genetics, whether it is bias related to self-identification or misunderstanding or lack of knowledge regarding ancestry. Additionally, race and ethnicity act as proxies for socioeconomic, environmental, and structural contributions to disease development, especially in the United States where racial and ethnic minorities disproportionately experience lower SES and face both environmental and structural bias as it pertains to hazardous exposure, stressful environments, and worse access to healthcare, among other factors. On the other hand, race and ethnicity may be more useful in investigating how individuals interact with the healthcare system. For instance, PCa screening behaviors amongst racial and ethnic minorities may be influenced by differences in patient-provider cultural approaches as well as mistrust in the healthcare system based on its historical mistreatment of racial and ethnic minority populations [ ]. In one study, non-Hispanic Black and Hispanic patients were less likely to experience shared decision-making regarding PSA screening and were therefore less likely to undergo PSA screening. Among patients who underwent shared decision-making, there were no racial disparities observed in PSA screening [ ]. While these studies did not fulfill our PICO criteria, it’s important to recognize these factors and how they influence our understanding of racial and ethnic disparities in PCa and PCa screening.

To eliminate or decrease confounding, some studies investigating racial/ethnic disparities in PCa adjusted for clinical factors. When adjusting for age, facility type, and diagnosis year, non-Hispanic Black, non-Hispanic other, and Hispanic-White, -Black, and -other patients had greater rates of Stage IV PCa at presentation compared to non-Hispanic White patients [ ]. When adjusting for SES factors, racial disparities in Stage IV PCa at presentation were reduced, though nearly all comparisons remained statistically significantly different, with the exception of Hispanic-Blacks. This suggests that both genetic and SES factors may contribute to advanced PCa presentation and racial/ethnic disparities. To highlight this, Hispanic-Black patient disparities in presentation of mPCa disappeared after adjusting for key SES factors [ ]. Furthermore, Chandra et al. (2021) reported racial differences in healthcare access and utility by demonstrating that Black patients across all insurance categories were less likely to receive mpMRI vs. commercially-insured White patients, which could help explain why Black patients tend to be diagnosed with PCa at more advanced stages because 12-core template prostate biopsies may miss mpMRI-detectable disease that could’ve been appropriately sampled [ ]. Even in some equal-access health care systems, which in theory may limit SES disparities, African-American men had increased risk of PCa detection and higher stage at presentation compared to Caucasian men [ , ]. However, these disparities were not noted in men aged 40-54 years, suggesting that early PSA testing in African-American men may allow earlier detection of PCa before the disparities noted in the older cohort arise [ ].

Evidence suggests that germline genetic testing is not equitable among races [ ]. Most studies do not engage in targeted recruitment to increase under-represented minorities in study populations, resulting in under-representation of non-White populations [ ]. Moreover, non-White populations are also under-represented in clinically-indicated genetic germline testing and in PCa clinical trials, which may further exacerbate racial and ethnic disparities [ , ].

To address racial and ethnic PCa disparities, some have studied implementation of culturally-sensitive community outreach programs to increase education of minority populations and encourage screening and patient engagement pertaining to PCa with encouraging results, though more studies are needed to determine if these programs yield earlier PCa detection and improved outcomes [ ].

4.3

Pathogenic Variants (PV)

Given the increased risk of PCa associated with positive PCa FH, germline genetic mutations have been suspected to contribute to heritability and disease development. Numerous studies have revealed associations between rare, high-penetrance pathogenic variants of genes involved in DNA damage repair and PCa, most notably BRCA1/2, HOXB13 , and MMR pathway mutations. These variants are quite rare—approximately 2% of PCa patients may carry at least one PV [ ]. Patients with PVs are more likely to present with earlier-onset disease, higher PSA, more aggressive cancer, and increased risk of metastasis [ ].

HOXB13 variants are associated with a 3-6-fold increased risk of PCa with further increased risk in patients with FH of PCa [ , , ]. Risk of PCa amongst HOXB13 -carriers is higher in more recent birth cohorts, suggesting contribution of differing environmental and/or lifestyle across generations [ ]. Although HOXB13 mutations are associated with increased PCa incidence, evidence varies regarding its association with PCa aggressiveness. For example, Akamatsu (2022) found no association between HOXB13 mutation and high-risk PCa, though the population was likely underpowered [ ]. Nyberg et al. (2019) found HOXB13 mutations did not predict more aggressive PCa compared to non-carriers [ ]. On the other hand, others have demonstrated HOXB13 mutation-carriers are more likely to have higher PSA at diagnosis, GS ≥7 disease, positive surgical margins, and overall more aggressive disease compared to non-carriers [ ]. Others have described how HOXB13 mutation variants differ among specific races and ethnicities [ ]. To this effect, a rare African-ancestry specific variant ( HOXB13 ×285K) was found to increase risk of PCa, more aggressive, and more advanced disease [ ]. These ancestry-specific differences may explain discrepancies in PCa outcomes among HOXB13 -carriers as certain mutations may have greater impact in the HOXB13 pathway.

Significantly more evidence exists for DDR mutation-carriers and increased risk of PCa incidence and lethality [ ]. Within the HRR pathway, evidence suggests BRCA2 -carriers have the greatest risk of PCa including greater risk than BRCA1 -carriers [ ]. PCa risk is greatest at younger ages for both BRCA1/2 -carriers. BRCA2- carriers are at greater risk of PCa at any given age while increased risk with BRCA1 -carriers is limited to younger ages [ ]. Mitra et al. (2017) reported a PPV of 47.6% when offering prostate biopsy to men with FH of BRCA1/2 mutations and PSA >3. Over a 33-month follow-up, 82% of detected PCa were mutation-carriers and 73% were csPCa [ ]. BRCA1/2- carriers may also benefit from additional testing with mpMRI and/or prostate biopsy given the increased risk for csPCa and mPCa even with low PSA, but further studies are needed to elucidate optimal screening regimens in this patient population.

BRCA mutations also impact PCa-related outcomes [ ]. BRCA2 PVs have greater association with aggressive tumor behavior and higher GS compared to BRCA1 -carriers [ , , ]. BRCA2 mutations are associated with increased risk of mPCa and worse overall survival [ , ]. Notably, however, Han et al. (2022) reported no differences in stage or grade group between BRCA2 -carriers and non-carriers, though the BRCA2 -mutation group may have been underpowered [ ]. Additionally, in advanced PCa, they also observed no differences in age at diagnosis between BRCA2 -carriers and non-carriers. BRCA2 -carriers with mPCa tended to have lower PSA than in non- BRCA2 carriers with mPCa, though this was not observed for localized PCa [ ]. BRCA1/2 mutations also confer worse outcomes in mCRPC with shorter PSA-progression-free survival and overall survival [ ].

Studies evaluating MMR mutations and PCa screening are limited. Nonetheless, MSH2, MLH1 , and MSH6 are most associated with increased PCa risk. Rosty et al. (2014) found significant differences in high-risk PCa features between MMR-carriers and non-carriers [ ]. Several studies have also associated MSH2 and MSH6 mutations with more aggressive PCa and worse outcomes [ ]. Others found lower levels of functional MLH1 in node-positive vs. node-negative PCa cases [ ]. These studies, however, were limited by small sample sizes due to their rarity. Interestingly, Wu et al. (2021) found that patients with an older age at diagnosis were more likely to have MMR-mutations than other DDR mutation pathways, such as HRR [ ].

It’s important to consider that many PVs have been investigated in racially and ethnically-homogenous groups—predominantly in those of European ancestry. Studies involving more diverse patient populations are revealing hundreds of VUS, many of which have not previously been published [ , ]. For example, one study reported the VUS rate in Black men was 1.5-times higher than the PV rate observed in White men [ ]. VUS may play a larger role in PCa development in men of African and other non-European ancestries as these populations have been historically under-studied compared to White men/European ancestry. Future studies of racial and ethnic minorities may reveal several VUS to be PVs in PCa development. There exists a crucial need to identify novel genes, VUS, expand gene panels, and increase diverse population representation to improve understanding of germline mutation contributions to PCa and provide genetic counseling most relevant to a patient’s ancestral composition.

Overall, despite PV rarity, its presence confers a high risk for PCa and worse PCa-related outcomes. Hence, patients with germline PVs may benefit from an earlier and/or more aggressive screening regimen. Future studies should investigate optimal PSA cutoffs and timing for mpMRI and biopsy indications.

4.4

Genetic/Polygenic Risk Score (PRS)

While high-penetrance PVs are rare and account for a small percentage of PCa inheritability, evidence suggests that a combination of numerous, common, low-penetrance SNPs may account for a majority of PCa risk, oftentimes independent of FH [ ]. PRS models combine the risks from individual SNPs into a single genetic risk score, which may be used to predict a patient’s risk or odds of PCa. SNPs are identified based on genome-wide association studies and selecting for common gene variants with the greatest association with PCa. PRS models have been proposed to improve risk stratification for patients undergoing PCa screening as well as predict risk of worse PCa outcomes. While genetic testing to determine PRS is relatively low cost, it has yet to be widely adopted into clinical practice, despite reports touting its cost-effectiveness in PCa detection and reduction in over-diagnosis [ , , ].

Implementation of PRS models into clinical practice may be hampered by heterogeneity. For example, we found PRS models varied from 29 to 6.7 million incorporated SNPs. PRS percentile cutoffs for reporting risk also varied widely between usage of quartiles, quintiles, or deciles. Of course, PCa risk increased as PRS percentile increased, but the ideal number of SNPs for a given PRS model and the ideal reporting of percentiles has not been established. Further, many PRS models were created with a primary endpoint of PCa detection [ , , , , , , ], though several investigators acknowledged PRS often failed to differentiate between higher GS, limiting its clinical utility [ , , , ]. However, Chen et al. (2023) reported a 1.23-fold increased risk of aggressive PCa in the top PRS decile compared to the 40-60th percentile, suggesting some patient comparisons yield risk stratification by PCa-aggressiveness [ ]. Studies were discordant regarding associations between PRS and PCa metastasis and mortality, with some reporting no or limited association [ , ] and others reporting increased risk of PCSM with higher PRS [ , ]. Further studies are needed to elucidate whether PRS is beneficial in predicting these crucial endpoints.

To improve predictive PRS models, investigators added non-modifiable risk factors such as FH, race/ethnicity, and other clinical factors to create hybrid risk models. Combined FH-positive and PRS models performed better than PRS alone and PRS in FH-negative models [ , ]. Low-PRS and FH-negative patients had low risk of csPCa, suggesting those with lower PRS-deciles could benefit from PCa screening de-intensification. Somewhat unsurprisingly, the best predictive models incorporated PRS and multiple clinical factors [ , , , ]. Importantly, such models may identify individuals at the extremes of high or low risk, suggesting possible benefits from screening escalation or de-escalation, respectively. For patients within the “gray-zone” PSA of 4-10 ng/ml, enhanced risk stratification may improve earlier csPCa detection while limiting overdiagnosis of non-csPCa.

There is evidence PRS may identify high-risk men and worse PCa outcomes prior to their development. One group observed a stronger association between PRS and lethal PCa for men with baseline PSA <1, suggesting that PRS can help identify higher-risk men even when initial PSA levels suggest low risk; this may help identify such men earlier and possibly prior to more aggressive PCa [ ]. Similarly, Chou et al. (2022) reported improved predictive utility of PRS for men at a younger age, as OR for PCa with higher PRS decreased as age increased [ ]. This suggests that PRSs may be more beneficial in risk stratification for younger men because if germline genetics play a predominant role, then patients are more likely to present with earlier-onset disease.

Most PRS models are based on genetic variants in populations of European-ancestry. Several studies have demonstrated that these models do not perform as well in more diverse populations and perform best in those of European-ancestry [ , , , , ]. Adding ancestry-specific SNPs to PRS models may improve their performance [ ]. Multi-ancestry PRS models perform better among more diverse populations [ , , ]. Finally, PRS models concordant with the study population perform best for a given specific ancestral population [ ]. These studies highlight the two-fold importance of diversifying genome-wide studies—first, to identify novel PCa-related SNPs and second, to improve PRS predictive utility in diverse populations and reduce racial and ethnic disparities.