Whole exome germline sequencing in early‐onset prostate cancer patients: Genomic findings and clinical outcomes - Siegelmann‐Danieli - The Prostate - Wiley Online Library
1 INTRODUCTION
Prostate cancer is the second most common solid tumor diagnosed in men worldwide. In 2020, there were over 1.4 million estimated new prostate cancer cases worldwide.1 In the United States, the majority of prostate cancer patients (62%) are diagnosed at 65 years or older, and the median age at diagnosis is 67 years.2
In recent years, significant advances in somatic testing using whole genome and whole exome sequencing (WES) of tumor samples derived from primary, advanced, and metastatic disease led to in-depth genome-based characterization of prostate tumors.3 In addition, the role of germline mutations in cancer susceptibility genes and their association with prostate cancer risk has also been investigated (reviewed by Nombela et al.4). Understanding the role of germline mutations in prostate cancer is clinically important since it affects the screening strategy of family members, and as emerging data on germline oncogenetic genes in various tumors may affect the prognosis and treatment approach of diagnosed patients.4, 5
In prostate cancer, the most common genes in which germline mutations are found include DNA damage repair (DDR) genes. Cohort studies in prostate cancer patients who were unselected for family history of cancer or age at diagnosis have shown that germline mutations in DDR genes (e.g., BRCA1/2, ATM, CHEK2, RAD51, and PALB2) are present in 4%–5% of patients with localized prostate cancer and 12%–16% in patients with metastatic disease.6-8 Information on the frequency of germline mutations in early-onset prostate cancer patients and their association with clinical outcomes is limited.
Herein, we present a cohort study in which 134 consecutive unselected patients with early-onset prostate cancer (diagnosed at ≤65 years of age) were assessed with respect to patient/disease characteristics at diagnosis (localized vs. metastatic disease), clinical outcomes (biochemical and distant recurrence), and whole genome sequencing which was used to detect all known likely pathogenic/pathogenic (LP/P) germline variants. The associations between the clinical and genomic characteristics were then examined.
2 MATERIALS AND METHODS
2.1 Study design and patients
This retrospective cohort study included adult prostate cancer patients who were members of Maccabi Healthcare Services (the second largest health maintenance organization in Israel with over 2.3 million members), were first diagnosed with prostate cancer at ≤65 years, provided a blood sample to the biobank of Kahn-Sagol-Maccabi (KSM) the Maccabi Research and Innovation Center, and signed an informed consent to use their sample for research purposes. Except for age at diagnosis, no exclusion criteria were applied with respect to stage at diagnosis, self- or family cancer history.
Patients who met the eligibility criteria were included in the cohort based on their consecutive consent to donate blood to the Biobank, regardless of the date of their prostate cancer diagnosis. For each patient, the prostate cancer diagnosis was confirmed by a senior medical oncologist (V.N.) who reviewed the computerized medical records and approved genomic WES of coded blood samples by the Genetic Lab of Maccabi Healthcare Services (Rehovot, Israel). Clinical information was abstracted (blinded to genomic data) from the computerized medical records of all patients. It included demographics, biopsy dates, prostate cancer-specific pharmacy purchasing information, and hospital claims for surgeries and irradiation. This information identified the date of diagnosis, stage at diagnosis (local/metastatic), and if applicable, date of biochemical recurrence, date of distant recurrence, date of secondary primary cancers, and date of death. Only KSM-prespecified data analysts could link coded genomic results with clinical information, and the final table was further anonymized for summary and analysis. Table S1 includes the definitions used for clinical data extraction.
2.2 WES and bioinformatics
Blood samples (peripheral blood or buffy coat) provided to the KSM biobank were frozen and kept at −80C° until DNA extraction. DNA was extracted using the QIAsymphony DSP DNA Kit (Qiagen LLC). Whole exome libraries were constructed from the extracted DNA (500–1000 ng per sample) using the DNA prep with enrichment kit (Illumina), and DNA enrichment was performed using IDT for Illumina DNA/RNA Kit (Illumina). All procedures were performed according to the respective manufacturer's instructions.
WES was performed using the xGen Exome Research Panel v2 (Integrated DNA Technologies Inc.), which includes 19,433 genes and covers 24 Mb, on the Illumina NovaSeq 6000 next-generation sequencing platform with the S1 200 cycle kit (Illumina) to an average coverage depth of >20× for >95% of the bases.
Analysis of the WES results was performed on the Franklin platform (Genoox) using the Burrows-Wheeler Aligner tool (v0.7.16) with GRCh37/hg19 as the reference human genome. Base calling, first filtration of reads, and identifying variants (single-nucleotide polymorphism, insertions/deletions, etc.) and artifacts were performed using the Genome Analysis Toolkit (v4.0.7.0; the Broad Institute) as well as Freebayes (v1.1.0).9 Variants were classified as P, LP, or variant of uncertain significance (VUS) on the Franklin platform using several databases, including ClinVar, VarSome, ClinGen, and the Leiden Open-source Variation Database.10-13
The genes evaluated in the analysis were grouped into four panels (Table S2). The oncogenetic panel (121 genes), which combined genes from multiple available panels such as the University of Washington BROCA Cancer Risk Panel, the University of Chicago Genetic Services (UCGS) Comprehensive hereditary cancer panel, the hereditary cancer panel by Color Health Inc., full focus cancer panel by Fulgent Genetics, and the multi-cancer panel by Invitae, as well as additional genes which were of interest due to their suggested association with tumor pathogenesis.14-18 Two immune response-related panels, including the comprehensive primary immunodeficiency panel (468 genes) and the Monogenic autoimmunity panel plus the autoimmune lymphoproliferative syndrome panel (76 genes) (Fulgent Genetics). The last panel used was the secondary findings minimum list, which includes 59 genes that are recommended by the American College of Medical Genetics and Genomics (ACMG) as the minimum list to be reported as secondary findings (25 of the genes in this panel are also on the oncogenetic panel, and the non-oncogenetic genes are referred to hereafter as the non-oncogenetic ACMG panel).19
2.3 Statistical analysis
Descriptive statistics were used to summarize clinicopathological characteristics and variant information. The chi-square test was used to compare biochemical recurrence and distant recurrence rates between LP/P variant carriers and the remaining patients. p < 0.05 was considered statistically significant. Data calculation and statistical analysis were performed using MedCalc® version 20.111 (MedCalc Software Ltd.).
3 RESULTS
3.1 Patients' characteristics
The final analysis included 134 patients with successful genomic results (WES failed in two patients). The patients were diagnosed with prostate cancer between 1996 and 2021. Their median age at diagnosis was 61 (range 46–65) years, and little over half (56%) were diagnosed at age 61–65 years. Close to two-thirds (62%) were born in Israel, and the vast majority (98%) had local disease, with only three patients (2%) presenting with metastatic disease (Table 1).
| Characteristic | N = 134 |
|---|---|
| Age at diagnosis | |
| Median (IQR), years | 61 (58–64) |
| Age category, n (%) | |
| ≤50 years | 3 (2) |
| 51-55 years | 18 (13) |
| 56-60 years | 38 (28) |
| 61-65 year | 75 (56) |
| Country of birth, n (%) | |
| Israel | 83 (62) |
| Europe | 25 (19) |
| Middle East/North Africa | 12 (9) |
| Central/South America | 5 (4) |
| USA | 3 (2) |
| Other/unknown | 6 (4) |
| Extent of disease at diagnosis | |
| Local | 131 (98) |
| Metastatic | 3 (2) |
- Abbreviations: IQR, interquartile range; N/A, not available.
3.2 Whole exome sequencing
Of the 134 patients, WES analysis revealed 20 patients (15%) with 21 variants in the oncogenetic panel: 8 patients (6%) carried LP/P variants and 12 (9%) had 13 VUS. No pathogenic variants in the two immunogenic panels and the non-oncogenetic ACMG panel were identified. None of the patients with LP/P variants had metastatic disease at diagnosis. The median (range) age at prostate cancer diagnosis of patients with LP/P variants (59 [54–64] years) was overall similar to that of patients with VUS (62 [50–65] years) as well as to that of patients for whom no variants were identified (61 [46–65] years). The cases with LP/P variants are summarized in Table 2. Of the identified LP/P variants, two pathogenic variants were in BRCA1, and two LP variants were in BRCA2, and both BRCA1 and BRCA2 mutations were known founder mutations in the Ashkenazi Jewish population.20, 21 Other pathogenic variants involved the ATM, HOXB13, MUTYH, and MYH7 genes.
| No. | Age at diagnosis | Country/region of birth | Extent of disease at diagnosis | Variant type | Gene involved | Nucleotide change | Protein effect | Location | Frequencies (exome, genome)a | Follow-up time, years | Clinical characteristics including recurrence and secondary cancersb |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 61 | Israel | Local disease | Pathogenic | ATM | c.8022_8023delAG | p.Gly2675fs | 11-108205706 | 0, 0 | 15.1 |
|
| 2 | 63 | Europe | Local disease | Pathogenic | BRCA1 | c5266dupCc | p.Glu1756fs | 17-41209079 | 0.000183, 0.0000526 | 1.3 |
|
| 3 | 57 | Israel | Local disease | Pathogenic | BRCA1 | c.68_69delAGc | p.Glu23fs | 17-41276044 | 0.000227, 0.000145 | 9.1 |
|
| 4 | 54 | Central/South America | Local disease | Likely pathogenic | BRCA2 | c.5946delTd | p.Ser1982fs | 13-32914437 | 0.000291, 0.000171 | 3.5 |
|
| 5 | 64 | Israel | Local disease | Likely pathogenic | BRCA2 | c.5946delTd | p.Ser1982fs | 13-32914437 | 0.000291, 0.000171 | 7.2 |
|
| 6 | 54 | Israel | Local disease | Pathogenic | HOXB13 | c.251G>A | p.Gly84Glu | 17-46805705 | 0.0184, 0.00148 | 0.6 |
|
| 7 | 57 | Middle East/North Africa | Local disease | Pathogenic | MUTYHe | c.1187G>A | p.Gly396Asp | 1-45797228 | 0.00303, 0.00332 | 6.4 |
|
| 8 | 61 | Middle East/North Africa | Local disease | Pathogenic | MYH7 | c.728G>A | p.Arg243His | 14-23900798 | 0.00000795, 0 | 14.0 |
|
The 12 cases with the 13 VUS are summarized in Table 3. Overall, six VUS involved the CHEK2 gene, two the ATR gene, two the BRCA2 gene, and the remaining three VUS were in the AR, PMS2, and RAD51C genes.
| No. | Age at diagnosis | Country/region of birth | Extent of disease at diagnosis | Gene involved | Nucleotide change | Protein effect | Location | Frequencies (exome, genome)a | Follow-up time, years | Clinical characteristics including recurrence secondary cancers and deathb |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 64 | Israel | Local disease | AR | c.808G>C | p.Ala270Pro | X-66765796 | 0.0000327, 0.00000888 | 14.3 |
|
| 2 | 65 | Europe | Local disease | ATR | c.4102C>T | p.Arg1368* | 3-142242885 | 0.00000398, 0 | 8.9 |
|
| 3 | 60 | Israel | Local disease | ATR | c.4877A>G | p.Tyr1626Cys | 3-142226927 | 0, 0 | 14.9 |
|
| 4 | 50 | Europe | Local disease | BRCA2 | c.6587A>G | p.Lys2196Arg | 13-32915079 | 0, 0 | 17.3 |
|
| 5 | 62 | Other | Local disease | BRCA2 CHEK2 |
c.5126A>C c.1556G>T |
p.Asp1709Ala Arg519Leu |
13-32913618 22-29083961 |
0.00000441, 0 0.000223, 0.000118 |
13.8 |
|
| 6 | 65 | Israel | Local disease | CHEK2 | c.1561C>T | p.Arg521Trp | 22-29083956 | 0.0000515, 0.0000329 | 10.3 |
|
| 7 | 54 | Israel | Local disease | CHEK2 | c.1513T>C | p.Ser505Pro | 22-29085152 | 0.0000128, 0.00000657 | 20.4 |
|
| 8 | 62 | Israel | Local disease | CHEK2 | c.1270T>C | p.Tyr424His | 22-29091220 | 0.000287, 0.000191 | 12.9 |
|
| 9 | 65 | Israel | Local disease | CHEK2 | c.1427C>T | p.Thr476Met | 22-29090054 | 0.000329, 0.000309 | 8.3 |
|
| 10 | 53 | Israel | Local disease | CHEK2 | c.1180G>A | p.Glu394Lys | 22-29091777 | 0.0000199, 0.00000657 | 18.5 |
|
| 11 | 63 | Israel | Local disease | PMS2 | c.625G>A | p.Gly209Arg | 7-6038819 | 0, 0 | 17.4 |
|
| 12 | 61 | Israel | Local disease | RAD51C | c.492T>G | p.Phe164Leu | 17-56774141 | 0.0000398, 0.0000657 | 14.4 |
|
3.3 Clinical outcomes
The median follow-up time from diagnosis for the entire cohort was 14 (range <1–25) years. Of the 131 patients with local disease at diagnosis, 32 (24%) experienced biochemical recurrence after a median (range) of 4 (0.5–18) years from the diagnosis, and 10 (8%) experienced distant recurrence after a median (range) of 10 (0.9–17) years from the diagnosis. Seven patients (5%) died. These deaths occurred after a median (range) of 16 (9–21) years from the prostate cancer diagnoses. Only two patients had metastatic prostate cancer at the time of death; for one of them, it was the cause of death, whereas, for the other patient, the cause of death was metastatic renal cell carcinoma. Of the remaining five deaths, two patients died of another cancer (glioblastoma, metastatic colon cancer), and for three patients, the cause of death was unavailable.
The eight patients with LP/P variants were followed for a median (range) of 7 (<1–15) years. Two patients experienced biochemical recurrence (2 and 6 years from the prostate cancer diagnosis), and one was diagnosed with metastatic disease (within a year from diagnosis). One patient had chronic lymphocytic leukemia 5.5 years following the diagnosis. No deaths were reported (Table 2).
The 12 patients with identified VUS were followed for a median (range) of 14 (8–20) years. Within the follow-up period, no biochemical recurrences were reported. In one patient, distant recurrence was reported 9 years after the prostate cancer diagnosis. Another patient had renal cell carcinoma 8 years after his initial prostate cancer diagnosis and a prostate cancer distant recurrence, which involved the bones 16 years after the diagnosis. The patient died 17 years after his prostate cancer diagnosis, and the cause of death was metastatic renal carcinoma. No other deaths were reported (Table 3).
Among patients with local disease at diagnosis, the rate of biochemical recurrence was similar in patients with LP/P variants (2/8 [25%]) and the remaining patients (30/123 [24%]). The rate of distant recurrence was numerically but not statistically significantly higher in the LP/P variant carriers (1/8 [12.5%]) than among the remaining patients (9/123 [7%], p = 0.59).
3.4 Additional tumors
Overall, 25 patients (19%) were diagnosed with at least one additional type of cancer (excluding non-melanoma skin cancer). Over a third of these patients (10/25, 40%) were diagnosed with another cancer before the prostate cancer diagnosis, and the rest (15/25, 60%) during follow-up. The total number of additional cancer diagnoses was 30, and an additional 1, 2, or 3 tumors occurred in 21, 3, and 1 patients, respectively. Additional cancer diagnoses included six bladder cancer cases, five each of colorectal and lymphoma, three melanoma cases, two each of primary bone/connective tissue, brain/nervous system, leukemia, and kidney cancer, and one each of larynx, lung, and thyroid cancer. No association was found between having germline variants (LP/P) in cancer susceptibility genes and the occurrence of secondary tumors.
4 DISCUSSION
The current study performed WES on a cohort of 134 early-onset prostate cancer patients diagnosed at ≤65 years of age, 97% presenting with localized tumors, with analysis focusing on extended oncogenetic panel (121 genes), two immunogenic panels and a non-oncogenetic ACMG panel. We identified LP/P variants in 6% of patients, all exclusively in the oncogenetic panel. Additional tumors (excluding non-melanoma skin cancers) were diagnosed in 19% of cohort participants. With a median follow-up of 14 years for the entire cohort and 7 years in those with LP/P variants in cancer susceptibility genes, no statistically significant association was found between having LP/P germline variants in such genes and clinical outcomes, although numerically, the distant recurrence rate was higher in patients with LP/P variants versus those without such variants.
The clinical characteristics of this cohort were overall similar to those of the early-onset prostate cancer population in Western countries. For example, the proportion of early-onset patients with de novo metastatic disease in this Israeli cohort was 2.2% versus 3.2% in other Western countries series.23, 24 The identified LP/P variants in our cohort included four DDR genes: ATM, BRCA1 (identified in two patients), BRCA2 (identified in two patients), and MUTYH. LP/P variants in all of these genes have been linked to an increased prostate cancer risk (reviewed by Nielsen et al.22, 25). Another identified alteration was in the homeobox transcription factor HOXB13, an androgen receptor suppressor.26 The identified alteration (pGly84Glu) is a recurrent variant known to increase the risk for prostate cancer by up to 20-fold.27 An additional identified pathogenic variant was in the myosin heavy chain 7 (MYH7) gene. Alterations in these genes have recently been linked to prostate cancer.28, 29 The prevalence of mutations in our oncogenetic panel, mostly in DDR genes, is generally consistent with the available data on DDR mutations in patients with localized prostate cancer in Western countries regardless of age at diagnosis. A recent study that focused on metastatic prostate cancer included data on 499 localized prostate cancer patients as a control and reported a prevalence of germline mutations in DDR genes of 4.6%.7 Another study involving 1588 patients, which stratified patients by ancestry, found that the prevalence of germline mutations in DDR genes was 4.0% among those with European ancestry versus 1.4% in those with African ancestry.8 The prevalence of BRCA1 and BRCA2 germline mutations in our study was 4 of 134 (3.0%). The identified BRCA LP/P variants in our study are known founder mutations in Ashkenazi Jews. This prevalence is consistent with findings from another study, where a subanalysis focusing on young Ashkenazi Jews in the United States showed a similar prevalence of these specific BRCA1/2 mutations (0.8% and 2.0%, respectively).30
Some reports suggest an association between germline mutations in DDR genes, more aggressive prostate cancer, and poor clinical outcomes.4 The National Comprehensive Cancer Network® prostate cancer guidelines recommend germline multigene testing for patients with metastatic, very-high/high-risk localized prostate cancer for adjusting treatments and for patients with early tumors who are either of Ashkenazi Jewish ancestry or with a family history suggesting known genetic syndromes for addressing potential cancer risk in family members.31 We could not confirm this association due to the limited number of patients with LP/P variants in DDR genes (6 in a cohort of 134 patients).
The strengths of our study include the link between the clinical and genomic datasets and a long median follow-up (14 years) with longitudinal information on each patient. Long follow-up is particularly critical in prostate cancer studies due to the long natural history of the disease. Our study focused on patients diagnosed at ≤65 years, which is a patient subpopulation that has not been extensively investigated thus far.
The major limitation of our study is the relatively small sample size as a result of the WES high cost. Still, initial calculations suggested that we should explore at least 73 patients. We also noted a potential selection bias of patients donating blood to the KSM biobank. These patients tend to belong to a higher socioeconomic status (on a 1–10 scale, the average socioeconomic status of patients in the current study was 7.8 vs. 6.6 for the remaining 14,300 prostate cancer patients in the Maccabi Cancer Registry) and are probably more educated and health aware.
5 CONCLUSIONS
The current study demonstrates the feasibility of a research platform that links genomic and extensive clinical data and investigates their associations. We focused on early-onset patients with early prostate cancer and identified genetic mutations in cancer susceptibility genes in 6% of cohort participants including DDR genes (ATM, BRCA1, BRCA2, and MUTYH), the homeobox transcription factor HOXB13 gene affecting androgen receptor activity, and the MYH7 gene previously reported in association with prostate cancer. No mutations were identified in three other panels, and currently, we cannot ascertain any benefit to genomic WES as opposed to sequencing only the oncogenetic panel. The limited number of genetic carriers did not allow a formal comparison of outcomes; still, our sample size of 134 patients with prolonged 14 years median follow-up suggests a potentially higher distant recurrence rate in LP/P variant carriers compared to the rest of the cohort participants.
AUTHOR CONTRIBUTIONS
Nava Siegelmann-Danieli and Mordechai Shohat designed the study. Victoria Neiman, Hillel Alapi, Erez Tsur, Daniella Beller, Racheli Berger, Galit Rimler, and GH acquired the data. Erez Tsur, Mordechai Shohat, Nava Siegelmann-Danieli, Avital Bareket-Samish, and VS analyzed and interpreted the data. Nava Siegelmann-Danieli and Avital Bareket-Samish drafted the manuscript. All authors critically reviewed the draft of the manuscript and approved the final version.
ACKNOWLEDGMENTS
The work was supported by a research grant from KSM Maccabi Institute for Research and Innovation.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
REFERENCES
Comments
Post a Comment