Discovery of a Recurrent Frameshift Ashkenazi Jewish Founder Mutation (F722fs) in the PARP Inhibitor-sensitive MMS22L Gene Associated with Higher Risk of Prostate Cancer
What does this study add?
An inherited Ashkenazi Jewish founder mutation (F722fs) in MMS22L, a gene involved in DNA repair and response to PARP inhibitor therapy, was significantly associated with the risk of prostate cancer, an incredibly more aggressive disease. Genetic testing for this and other MMS22L variants could help in identifying men at higher risk of aggressive prostate cancer and a favorable response to PARP-targeted therapy.
Clinical Relevance
The future of cancer care is undoubtedly personalized. Personalization will occur for treatment selection (both for efficacy and adverse event prediction), surveillance planning (type and frequency), and for answering the Why me? The question that most cancer patients ask. Genetics will be the mode of personalization in the majority of cases. Many prostate cancer-causing gene mutations have been discovered but explain only a fraction of observed cases. This paper describes a new gene, MMS22L, that when mutated not only causes prostate cancer but that might be responsive to PARP inhibition. This gene should be added to standard germline testing panels for prostate cancer. Associate Editor: Derya Tilki.
Patient Summary
We analyzed genetic data for patients of Ashkenazi Jewish ancestry with prostate cancer. A mutation in MMS22L, which is a gene involved in DNA repair, was significantly associated with prostate cancer risk, especially for patients with more aggressive disease. Testing for this and other variants of the MMS22L gene may help to identify men at higher risk of prostate cancer.
Introduction
Prostate cancer (PCa) has a strong genetic susceptibility, with an estimated heritability rate of 57% [1]. To date, hundreds of common risk-associated genetic variants have been identified from genome-wide association studies [2,3]. In addition, it has been shown that rare pathogenic mutations in several major genes such as HOXB13, BRCA2, ATM, CHEK2, and PALB2 are associated with higher PCa risk, particularly for more aggressive disease [4–8]. These genetic variants have important clinical utility in PCa risk stratification for the development of personalized cancer screening strategies [9–11].
Most of the genes for which an association with PCa susceptibility has been established are involved in DNA repair [5–8]. Inherited mutations in DNA repair genes reduce the capacity to correct DNA damage, resulting in the accumulation of somatic mutations that can lead to cancer. Furthermore, PCa cells with DNA repair deficiency, particularly in the homologous recombination pathway, often exhibit greater reliance on the PARP1-mediated base excision repair pathway, making them highly sensitive to PARP inhibitor (PARPi) therapy owing to the ‘‘synthetic lethality’’ principle [12]. A therapeutic synthetic lethality effect was observed after PARPi therapy among PCa patients whose tumors harbored pathogenic variants of BRCA2 or other DNA repair genes [13]. However, few other genes for which inactivation leads to a synthetic lethality effect similar to BRCA2 have been identified.
Tsujino et al [14] recently identified a novel set of 65 genes for which experimental inactivation via CRISPR – conferred PARPi sensitivity in PCa cell lines. Their findings suggest that inactivation of these genes could influence susceptibility to PCa and create a synthetic lethality effect with potential therapeutic benefits [12]. In the present study, we tested the association of this set of genes with PCa risk, which may predict not only predisposition to PCa development but also sensitivity to PARPi therapy.
Results
We first tested the association of PCa risk with the 65 PARPi sensitivity genes by comparing the proportion of aggregated LoF mutations in each gene between 3716 PCa patients from Hopkins and 103 221 male population controls from gnomAD using Fisher’s exact test. As expected, the LoF mutation rate for each gene was low in both the case and control groups (Supplementary Table 1). However, ancestry-specific analyses revealed a significantly higher rate in the case group than in the control group for three genes (Fisher’s test, nominal p < 0.05). Among 2641 cases versus 93 944 controls of NFE ancestry, the LoF mutation carrier rate was 0.19% versus 0.06% for MUS81 (odds ratio [OR] 3.0, 95% confidence interval [CI] 0.9–7.5; p = 0.03), and 0.19% versus 0.07% for SWSAP1 (OR 2.9, 95% CI 0.9– 7.1; p = 0.04). Among 469 Hopkins ASJ cases versus 5573 controls of ASJ ancestry from gnomAD, the LoF mutation carrier rate for MMS22L was 1.3% versus 0.38% (OR 3.4, 95% CI 1.1–8.8; p = 0.02). There was no significant difference for any gene in the AFR ancestry analysis (606 cases and 3704 controls).
To validate the three ancestry-specific associations, we performed a confirmation analysis in the UKB cohort, which included 14 660 PCa cases and 172 669 controls of NFE ancestry, and 107 PCa cases and 1236 controls of ASJ ancestry. While an association in the opposite direction was observed for MUS81 and SWSAP1 (lower LoF mutation carrier rate for cases than for controls; p > 0.05, Firth’s test), a trend for a higher MMS22L LoF mutation carrier rate for cases (0.93%) than for controls (0.16%) was confirmed in the ASJ population, although the difference was not significant (p = 0.11, Firth test).
Analysis of MMS22L LoF mutations in ASJ from Hopkins and UKB revealed that the vast majority of the carriers from Hopkins and UKB (26 of 28) had the same variant, a 5-bp deletion of GAAAA (NM198468:c.2164_2168del) in exon 15 leading to a frameshift mutation (F722fs), prematurely terminating after 54 amino acids (Fig. 1). Among subjects with individual-level SNP data (Hopkins and UKB), all nine F722fs carriers shared the same 4.98-Mb haplotype (H1), suggesting an ASJ founder mutation (Supplementary Table 2).
To confirm this association, we analyzed data for 167 additional ASJ PCa patients from three independent cohorts (Michigan, NorthShore, and GoPath), which revealed four more carriers of the F722fs variant. In a combined analysis of the discovery and confirmation cohorts, the F722fs carrier rate was 1.5% for ASJ PCa cases versus 0.31% for ASJ controls (OR 4.9, 95% CI 2.1–10.6; p = 1.44 10 4, Fisher’s test; Table 1).
Four more F722fs carriers were found among PCa patients whose genetic probability for ASJ ancestry was <75% (range 1–17%; Supplementary Table 3), three of whom shared the same H1 haplotype. One patient (ASJ genetic probability 1%) had a different haplotype (H2), suggesting a different origin of the variant.
To examine the relationship of MMS22L F722fs with PCa characteristics, we performed a case-case association analysis among 469 ASJ PCa patients from Hopkins for whom clinical variables were obtained in a more consistent fashion. Among the six F722fs carriers, five (83%) had more aggressive disease (pathologic Gleason score 8, preoperative PSA 20 ng/ml, metastasis, or death from PCa), which is significantly higher than the 27% of noncarriers in the cohort with aggressive disease (OR 12.3, 95% CI 2.2–132.5; p = 0.003, Firth’s test; Table 2). Specifically, the carrier group had higher proportions of patients than the noncarrier group with Gleason score 8 (50% vs 20%), preoperative PSA 20 ng/ml (33% vs 8.5%), and metastasis (33% vs 9.5%). However, none of the six carriers died of PCa, whereas 2.2% of the noncarriers died of PCa.
Beyond the ASJ founder mutation, 12 additional PCa patients carried seven other LoF variants, all of whom were of NFE ancestry (Fig. 1 and Supplementary Table 3). One variant (c.340+1G>A, a splicing variant predicted to affect function according to in silico metrics) was recurrent and found in six cases, five of whom shared the same 5.20-Mb haplotype (Supplementary Table 4). We further tested the association of this variant with PCa risk in the group of men of NFE ancestry in the UKB cohort (Table 3). The PCa rate was significantly higher in the carrier subgroup than in the noncarrier subgroup (35% vs 7.8%; OR 7.7, 95% CI 2.6–21.0; p = 5.1 10 4, Firth test). Although the rates for several other solid cancer types were also higher in the carrier subgroup, the differences were not statistically significant (p > 0.05).
Finally, we tested the association of all MMS22L LoF mutations with the risk of PCa among individuals of NFE ancestry in the UKB cohort. The carrier rate for aggregated MMS22L LoF mutations was 0.08% in the NFE PCa group versus 0.03% in the NFE control group. The PCa rate was significantly higher in the subgroup of 59 carriers (19%) than in the subgroup of 187 270 noncarriers (7.8%; OR 3.2, 95% CI 1.6–6.1; p = 0.002, Firth test). The higher PCa rate in the carrier subgroup remained after removing the c.340+1G>A variant, although the association was no longer statistically significant (OR 2.0, 95% CI 0.7–4.6; p = 0.2, Firth test).
Conclusion
In summary, our results demonstrate the presence of rare LoF germline variants in MMS22L that are associated with PCa risk, especially aggressive PCa, in populations with ASJ or NFE ancestry. If validated, our findings could be applied in risk stratification of PCa.
To read the full study, please visit the European Urology Focus.
What does this study add?
An inherited Ashkenazi Jewish founder mutation (F722fs) in MMS22L, a gene involved in DNA repair and response to PARP inhibitor therapy, was significantly associated with the risk of prostate cancer, an incredibly more aggressive disease. Genetic testing for this and other MMS22L variants could help in identifying men at higher risk of aggressive prostate cancer and a favorable response to PARP-targeted therapy.
Clinical Relevance
The future of cancer care is undoubtedly personalized. Personalization will occur for treatment selection (both for efficacy and adverse event prediction), surveillance planning (type and frequency), and for answering the Why me? The question that most cancer patients ask. Genetics will be the mode of personalization in the majority of cases. Many prostate cancer-causing gene mutations have been discovered but explain only a fraction of observed cases. This paper describes a new gene, MMS22L, that when mutated not only causes prostate cancer but that might be responsive to PARP inhibition. This gene should be added to standard germline testing panels for prostate cancer. Associate Editor: Derya Tilki.
Patient Summary
We analyzed genetic data for patients of Ashkenazi Jewish ancestry with prostate cancer. A mutation in MMS22L, which is a gene involved in DNA repair, was significantly associated with prostate cancer risk, especially for patients with more aggressive disease. Testing for this and other variants of the MMS22L gene may help to identify men at higher risk of prostate cancer.
Introduction
Prostate cancer (PCa) has a strong genetic susceptibility, with an estimated heritability rate of 57% [1]. To date, hundreds of common risk-associated genetic variants have been identified from genome-wide association studies [2,3]. In addition, it has been shown that rare pathogenic mutations in several major genes such as HOXB13, BRCA2, ATM, CHEK2, and PALB2 are associated with higher PCa risk, particularly for more aggressive disease [4–8]. These genetic variants have important clinical utility in PCa risk stratification for the development of personalized cancer screening strategies [9–11].
Most of the genes for which an association with PCa susceptibility has been established are involved in DNA repair [5–8]. Inherited mutations in DNA repair genes reduce the capacity to correct DNA damage, resulting in the accumulation of somatic mutations that can lead to cancer. Furthermore, PCa cells with DNA repair deficiency, particularly in the homologous recombination pathway, often exhibit greater reliance on the PARP1-mediated base excision repair pathway, making them highly sensitive to PARP inhibitor (PARPi) therapy owing to the ‘‘synthetic lethality’’ principle [12]. A therapeutic synthetic lethality effect was observed after PARPi therapy among PCa patients whose tumors harbored pathogenic variants of BRCA2 or other DNA repair genes [13]. However, few other genes for which inactivation leads to a synthetic lethality effect similar to BRCA2 have been identified.
Tsujino et al [14] recently identified a novel set of 65 genes for which experimental inactivation via CRISPR – conferred PARPi sensitivity in PCa cell lines. Their findings suggest that inactivation of these genes could influence susceptibility to PCa and create a synthetic lethality effect with potential therapeutic benefits [12]. In the present study, we tested the association of this set of genes with PCa risk, which may predict not only predisposition to PCa development but also sensitivity to PARPi therapy.
Results
We first tested the association of PCa risk with the 65 PARPi sensitivity genes by comparing the proportion of aggregated LoF mutations in each gene between 3716 PCa patients from Hopkins and 103 221 male population controls from gnomAD using Fisher’s exact test. As expected, the LoF mutation rate for each gene was low in both the case and control groups (Supplementary Table 1). However, ancestry-specific analyses revealed a significantly higher rate in the case group than in the control group for three genes (Fisher’s test, nominal p < 0.05). Among 2641 cases versus 93 944 controls of NFE ancestry, the LoF mutation carrier rate was 0.19% versus 0.06% for MUS81 (odds ratio [OR] 3.0, 95% confidence interval [CI] 0.9–7.5; p = 0.03), and 0.19% versus 0.07% for SWSAP1 (OR 2.9, 95% CI 0.9– 7.1; p = 0.04). Among 469 Hopkins ASJ cases versus 5573 controls of ASJ ancestry from gnomAD, the LoF mutation carrier rate for MMS22L was 1.3% versus 0.38% (OR 3.4, 95% CI 1.1–8.8; p = 0.02). There was no significant difference for any gene in the AFR ancestry analysis (606 cases and 3704 controls).
To validate the three ancestry-specific associations, we performed a confirmation analysis in the UKB cohort, which included 14 660 PCa cases and 172 669 controls of NFE ancestry, and 107 PCa cases and 1236 controls of ASJ ancestry. While an association in the opposite direction was observed for MUS81 and SWSAP1 (lower LoF mutation carrier rate for cases than for controls; p > 0.05, Firth’s test), a trend for a higher MMS22L LoF mutation carrier rate for cases (0.93%) than for controls (0.16%) was confirmed in the ASJ population, although the difference was not significant (p = 0.11, Firth test).
Analysis of MMS22L LoF mutations in ASJ from Hopkins and UKB revealed that the vast majority of the carriers from Hopkins and UKB (26 of 28) had the same variant, a 5-bp deletion of GAAAA (NM198468:c.2164_2168del) in exon 15 leading to a frameshift mutation (F722fs), prematurely terminating after 54 amino acids (Fig. 1). Among subjects with individual-level SNP data (Hopkins and UKB), all nine F722fs carriers shared the same 4.98-Mb haplotype (H1), suggesting an ASJ founder mutation (Supplementary Table 2).
To confirm this association, we analyzed data for 167 additional ASJ PCa patients from three independent cohorts (Michigan, NorthShore, and GoPath), which revealed four more carriers of the F722fs variant. In a combined analysis of the discovery and confirmation cohorts, the F722fs carrier rate was 1.5% for ASJ PCa cases versus 0.31% for ASJ controls (OR 4.9, 95% CI 2.1–10.6; p = 1.44 10 4, Fisher’s test; Table 1).
Four more F722fs carriers were found among PCa patients whose genetic probability for ASJ ancestry was <75% (range 1–17%; Supplementary Table 3), three of whom shared the same H1 haplotype. One patient (ASJ genetic probability 1%) had a different haplotype (H2), suggesting a different origin of the variant.
To examine the relationship of MMS22L F722fs with PCa characteristics, we performed a case-case association analysis among 469 ASJ PCa patients from Hopkins for whom clinical variables were obtained in a more consistent fashion. Among the six F722fs carriers, five (83%) had more aggressive disease (pathologic Gleason score 8, preoperative PSA 20 ng/ml, metastasis, or death from PCa), which is significantly higher than the 27% of noncarriers in the cohort with aggressive disease (OR 12.3, 95% CI 2.2–132.5; p = 0.003, Firth’s test; Table 2). Specifically, the carrier group had higher proportions of patients than the noncarrier group with Gleason score 8 (50% vs 20%), preoperative PSA 20 ng/ml (33% vs 8.5%), and metastasis (33% vs 9.5%). However, none of the six carriers died of PCa, whereas 2.2% of the noncarriers died of PCa.
Beyond the ASJ founder mutation, 12 additional PCa patients carried seven other LoF variants, all of whom were of NFE ancestry (Fig. 1 and Supplementary Table 3). One variant (c.340+1G>A, a splicing variant predicted to affect function according to in silico metrics) was recurrent and found in six cases, five of whom shared the same 5.20-Mb haplotype (Supplementary Table 4). We further tested the association of this variant with PCa risk in the group of men of NFE ancestry in the UKB cohort (Table 3). The PCa rate was significantly higher in the carrier subgroup than in the noncarrier subgroup (35% vs 7.8%; OR 7.7, 95% CI 2.6–21.0; p = 5.1 10 4, Firth test). Although the rates for several other solid cancer types were also higher in the carrier subgroup, the differences were not statistically significant (p > 0.05).
Finally, we tested the association of all MMS22L LoF mutations with the risk of PCa among individuals of NFE ancestry in the UKB cohort. The carrier rate for aggregated MMS22L LoF mutations was 0.08% in the NFE PCa group versus 0.03% in the NFE control group. The PCa rate was significantly higher in the subgroup of 59 carriers (19%) than in the subgroup of 187 270 noncarriers (7.8%; OR 3.2, 95% CI 1.6–6.1; p = 0.002, Firth test). The higher PCa rate in the carrier subgroup remained after removing the c.340+1G>A variant, although the association was no longer statistically significant (OR 2.0, 95% CI 0.7–4.6; p = 0.2, Firth test).
Conclusion
In summary, our results demonstrate the presence of rare LoF germline variants in MMS22L that are associated with PCa risk, especially aggressive PCa, in populations with ASJ or NFE ancestry. If validated, our findings could be applied in risk stratification of PCa.
To read the full study, please visit the European Urology Focus.