PCa: Risk-adapted early detection

Evidence for genetic predisposition
There is evidence for genetic predisposition to PCa from twin studies and epidemiological data from case control studies. These show that about 58% of the predisposition to PCa is due to genetic factors. Studies have shown that this predisposition is composed of rare genetic variation (about 7% of the component), predominantly mutations in the genes BRCA2 and ATM and the mismatch repair genes, and 44% is common genetic variation. This means that just under a half of the genetic predisposition has not yet been discovered.

This is the first challenge in this field; it is not known whether this ‘missing heritability’ is due to (i) further common genetic variation (but extensive studies have found the majority of these variants, at least in those of European origin), (ii) further rare variation and extensive genetic sequencing of those with PCa at a younger age or those with a family history of PCa may  yield further alterations in other genes not yet identified as being associated with PCa or (iii) other types of genetic variability which are more difficult to detect such as large deletions or other types of inherited alterations. There is also a possibility that individual families may harbour private mutations restricted to one or only a few families and these will be difficult to confirm that they are associated with disease.

Genetic variation in the coding region of genes is rarer than common variation which gives a risk score relative to the average of a population. The former is therefore most useful for identifying individuals at moderate/high risk and has implications for their family as these variants are inherited with a 1 in 2 risk by offspring whereas common variation gives an individual a relative risk which is relative to the average of the population but can be different in their offspring as common variation only applies to the individual in whom it is being assessed.

Rare variation
Germline pathogenic mutations, particularly in DNA repair genes, are associated with localised PCa in about 7% of cases and with metastatic disease in 11-13% of cases.

The main challenge in rare variation detection is thelimitation of testing guidelines in some healthcare systems; for example, in the UK, a 10 gene germline panel test is undertaken if an individual has PCa diagnosed at <50 years if localised or <60 years if metastatic. However, our group and others have shown that this is likely to be too limited since older cases (at least those diagnosed at 67 years or less) harbour mutations in the genes in this panel and other genes which are not on the panel are also associated with PCa. Of interest, the majority of the genes which harbour pathogenic mutations associated with PCa are in the DNA repair pathway with the exception of HOXB13, which is associated with the androgen receptor.

The relative risk of PCa from a pathogenic variant in BRCA2 is 6-8 fold and BRCA1 is 1,8-3 fold. There are data which show that these are associated with a higher Gleason score, and, particularly for BRCA2, with a poorer prognosis with earlier progression to metastasis and lower overall PCa specific survival. Recent data also show increased relapse post PSMA lutetium therapy and a shorter response time to hormone therapy.

Alterations in these genes associated with disease do not all behave equally. For example, we have shown that in CHEK2, the common genetic alteration (1100delC) in Europeans increases the risk of PCa about 2-2.5 fold but other mutations in the gene have been associated with decreased survival. The ATM gene confers an increased relative risk of PCa, but the degree of risk depends on mutation type, for example pathogenic truncating mutations confer a 4-fold risk but missense mutations a 1.5-fold risk. In BRCA2, mutations in the ovarian cancer cluster region in the centre of the gene are associated with a lower PCa risk than in other parts of the gene, but as yet this is not considered in genetic counselling. The other caveat when considering risk is the context of the mutation and the ancestry of the individual. For example, BRCA mutations in the Ashkenazim are associated with a lower PCa risk than is reported in other populations.

When offering genetic sequencing to look for mutations, other variants can be found in the genetic sequence which have not to date been associated with disease (so-called variants of uncertain significance or VUS). These are more common in diverse populations as less genetic sequencing has been conducted in those of Asian or African ancestry and it therefore can be more difficult to distinguish normal variants from pathogenic ones due to paucity of data. VUS are not acted upon for clinical care due to the uncertainty of their clinical effect.

Common variation
Genome wide association studies have shown that numerous genetic variants (most of which are single nucleotide changes or SNPs) are associated with PCa risk. Each variant has a small per allele odds ratio of risk, but the variants are numerous, and the overall risk is calculated by multiplying these ratios for each variant to get an overall risk score (the so-called polygenic risk score or PRS). The overall risk can therefore be substantial and equal (at least in those in the top 80% of risk strata) to those risks from germline rarer pathogenic variants in DNA repair genes.

There are several challenges with PRS determination. The first is that until recently PRS were predominantly developed in European populations and do not perform well in those from Asian or African ancestry. However, a large meta-analysis of over 100,000 samples was recently published in December 2023 which reported a 451 SNP profile that performs well in diverse populations. Over 80 new variants were reported in this report in those of African ancestry

Even with the latest score of 451 SNPs, this profile is less effective when an individual has more than two ancestries in their heritage, as admixture can reduce the usefulness of the PRS. If there is minimal admixture, PRS can provide a risk estimate of PCa. Previously, it was believed to mainly identify indolent cases, not aggressive cases (defined as PCa with a Gleason score >7).

However, this is not correct as they are associated with both indolent and aggressive disease (only a few are associated with aggressive disease per se), and modelling has suggested that PRS which can risk stratify populations should identify more cases in higher risk strata and therefore more aggressive cases in the higher risk strata also. Until recently this had not been tested in a clinical setting but now the BARCODE1 study has demonstrated that such modelling is correct.

Interaction of rare and common variation
At present in most diagnostic settings, rare variation is the only genetic variant for which testing is performed, however increasingly, particularly in the private healthcare systems, common variation is starting to be tested. However, it will be important to show that this is clinically useful. What is already known is that there is interaction between common and rare variation to modify the risk of PCa from the rare variant. This has been best reported for risks from BRCA1/2 and HOXB13 variants.

Evidence for targeted screening using genetic variation
Due to concern of overdiagnosis, there is currently no international PCa screening programme using PSA testing in the general population (except for Lithuania and Kazakhstan) even though reduction in mortality has been shown in European studies (the ERSPC).

However, there is a European guideline recommending annual PSA screening in BRCA2 carriers from age 40. This is based on results from the IMPACT study which is an international study of annual PSA in 65 centres in 20 countries conducted in the pre-MRI era, in carriers of mutations in DNA repair genes, BRCA1 and BRCA2. This study showed a higher PCa detection in BRCA2 mutation carriers versus controls, with 70% of cancers being clinically significant and needing radical treatment.

There are insufficient data to extend this guideline to BRCA1 carriers. In carriers of mutations in mismatch repair genes the baseline data show that these also have more aggressive PCa based on Gleason score but follow-up data in IMPACT are awaited and there is an ongoing study in this area which would need to be reported prior to adoption in EAU guidelines. For assessment of PRS, the BARCODE 1 study has just reported on biopsy outcome irrespective of PSA level in European men in London (GB). This reported on over 6,000 men in London who had a saliva sample for PRS; those 745 in the top 10% of the risk stratum were offered biopsy and targeted biopsy of lesion(s) with PI-RADs >3 on MRI. 187 cancers were found; 55% had Gleason score >7  and 21% needed radical treatment on international guidelines. This study showed that PRS can risk stratify and identify a high proportion of clinically significant disease. As yet this has not been undertaken in those from diverse populations (the 451 PRS score was not reported when BARCODE1 was undertaken).

The future
There is strong evidence of genetic predisposition to PCa, involving both rare and common variations, which interact to influence the overall risk score. Data are becoming available to show its utility in targeted screening but there are challenges in uptake in healthcare systems which will require more extensive genetic testing, cheaper testing and provision of education both to the population and to healthcare professionals about the nuances of such risk profiling.

The early data from IMPACT and BARCODE 1 suggest that such genetic profiling will identify higher numbers of cases in these individuals, and many have aggressive disease that would need radical treatment on international guidelines. The challenge will be when should we consider more extensive population testing for both rare and common variation and the optimal PCa screening algorithm for variation and the optimal PCa screening algorithm for individuals identified as being at higher genetic risk.

• Prof. Ros Eeles (GB) will be presenting on “Early detection and diagnostic pathways in prostate cancer revisited”, Sunday, 23 March 10:45 – 12:30