Prospective Genomic Profiling of Prostate Cancer Across Disease States Reveals Germline and Somatic Alterations That May Affect Clinical Decision Making

Abstract

Purpose: A long natural history and a predominant osseous pattern of metastatic spread are impediments to the adoption of precision medicine in patients with prostate cancer. To establish the feasibility of clinical genomic profiling in the disease, we performed targeted deep sequencing of tumor and normal DNA from patients with locoregional, metastatic non-castrate, and metastatic castration-resistant prostate cancer (CRPC).

Methods: Patients consented to genomic analysis of their tumor and germline DNA. A hybridization capture-based clinical assay was employed to identify single nucleotide variations, small insertions and deletions, copy number alterations and structural rearrangements in over 300 cancer-related genes in tumors and matched normal blood.

Results: We successfully sequenced 504 tumors from 451 patients with prostate cancer. Potentially actionable alterations were identified in DNA damage repair (DDR), PI3K, and MAP kinase pathways. 27% of patients harbored a germline or a somatic alteration in a DDR gene that may predict for response to PARP inhibition. Profiling of matched tumors from individual patients revealed that somatic TP53 and BRCA2 alterations arose early in tumors from patients who eventually developed metastatic disease. In contrast, comparative analysis across disease states revealed that APC alterations were enriched in metastatic tumors, while ATM alterations were specifically enriched in CRPC.

Conclusion: Through genomic profiling of prostate tumors representing the disease clinical spectrum, we identified a high frequency of potentially actionable alterations and possible drivers of disease initiation, metastasis and castration-resistance. Our findings support the routine use of tumor and germline DNA profiling for patients with advanced prostate cancer, for the purpose of guiding enrollment in targeted clinical trials and counseling families at increased risk of malignancy.

Fig 1.

Clinical sequencing of tumors and germline for patients with prostate cancer. (A) MSK-IMPACT assay workflow. (B) Four hundred fifty-one patients underwent tumor profiling in the clinic. Their last known disease state when seen in the clinic is represented at the top. Their disease state at the time of tissue collection for the 504 tumors that were profiled is represented at bottom. Tumors that were profiled represented all three prostate cancer clinical disease states: locoregional, metastatic noncastrate, and metastatic castration resistant. Locoregional disease indicates disease without distant clinical or pathologic spread, including lymph node (LN) involvement in the pelvis only (TxN0/1). Sample type (prostate v metastasis) is represented at bottom. (C) Site of disease for metastatic tumors successfully profiled by MSK-IMPACT. CNV, copy number variation; EMR, electronic medical record, MSK-IMPACT, Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets; SNV, single-nucleotide variation.

Fig 2.

Selected genomic alterations across disease states in the MSK-IMPACT data set. Each column represents an individual patient with prostate cancer whose tumor was acquired in the disease state indicated at the top. Total number of nonsynonymous somatic mutations in the tumor is represented in histogram form. Sample type (prostate v metastasis) is also represented. Alterations in commonly affected genes in prostate cancer (eg, AR, PTEN, TP53, FOXA1) are shown in addition to genes in potentially actionable or biologically relevant pathways. The type of alteration (eg, copy number variation, rearrangement, mutation) is indicated in the bottom row. MAPK, mitogen-activated protein kinase; MSK-IMPACT, Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets; PI3K, phosphatidylinositol-3-kinase.

Fig 3.

Somatic and germline alterations in DNA damage repair genes. (A) Oncoprint of somatic alterations in genes that are involved in DNA repair by homologous recombination (HR) for 451 patients. Twenty-two percent of patients harbor a tumor alteration in one of the listed genes. (B) Germline alterations for patients who consented to germline analysis (n = 221). In total, 19% of patients were found to have a germline pathogenic alteration. (C) Frequency of combined somatic and germline alterations in DNA repair genes BRCA2, BRCA1, ATM, and CHEK2 for patients who consented to germline analysis. Overall, 27% of patients had either a germline or a somatic-only alteration in one of these genes. (D) Oncoprint of somatic alterations in genes that are involved in DNA mismatch repair (MMR; top). Three percent of patients harbor a tumor alteration in one of these genes. The majority of tumors that display more than 20 somatic mutations (bottom) harbor a somatic alteration in an MMR gene.

Fig 4.

Cross-cohort comparative analysis of the pattern and degree of copy number alterations across the genome. Represented here are regions of amplification (red) or deletion (blue) with chromosomes listed horizontally (top) in the MSK-IMPACT, SU2C-PCF (metastatic castration-resistant prostate cancer [CRPC]), and TCGA (primary localized prostate cancer) data sets. MSK-IMPACT tumors are sorted by disease state at time of tissue acquisition, from locoregional (bottom, gray) to metastatic CRPC (top, blue). Whereas sequencing platforms differ between studies, the degree and pattern of copy number alteration were similar for tumors that were acquired in the same disease state. MSK-IMPACT, Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets; SCNA, somatic copy number alterations; SU2C-PCF, Stand Up to Cancer-Prostate Cancer Foundation; TCGA, The Cancer Genome Atlas.

Fig 5.

Enrichment of genomic alterations between disease states. (A) Enrichment of genomic alterations in tumors from patients with metastatic castration-resistant prostate cancer (mCRPC) versus locoregional disease. The level of enrichment is represented as difference in frequency between the two indicated classes (x-axis) and its significance (P value, y-axis). The type of alteration is represented by color. (B) Enrichment of genomic alterations in tumors from patients with metastatic CRPC versus metastatic noncastrate disease. (C) Enrichment of genomic alterations in tumors from patients with metastatic noncastrate disease versus locoregional disease. (D) Frequencies of alterations in select genes across disease states in the MSK-IMPACT data set. P values are represented (Fisher’s exact test). amp, amplification; homdel, homozygous deletion; mut, mutation.

Fig 6.

Somatic alterations identified in matched tumors from the same patients. (A) Somatic mutation count in pairs of matched tumors. The latter tumor (Tumor 2) tends to have a higher mutation burden. A notable exception (*) involves a bladder metastasis and a bone metastasis acquired 5 months later, where the patient had received salvage radiation to the pelvis, possibly explaining the higher mutation count in the earlier bladder tumor. (B) Somatic alterations in TP53 (red), BRCA2 (blue), AR (gold), and PIK3CA (green) in matched tumors in the data set, including localized primaries and later metastases (green box) and other matched tumors from the same patients. (C) Evolutionary analysis representing the acquisition of genomic alterations in sequential tumors obtained from two patients. Filled circles represent tumors sequenced by MSK-IMPACT, labeled with their sites, disease state, and relative dates of acquisition. CRPC, castration-resistant prostate cancer.

Fig A1.

Tumor sequencing success rates. Sequencing success rate for prostate and metastatic samples (includes archived samples acquired for sequencing as well as fresh biopsies). Highest overall success rates were observed for prostate samples, with failures occurring primarily among older archived tumors (median age of failed primary sample: 41.2 months, median age of successful primary sample: 13.9 months). Lowest overall success rates were for bone and lung (42-52%).

Fig A2.

MSK-IMPACT versus SU2C-PCF dataset comparison. Frequencies of alterations in select genes in metastatic CRPC tumors from the MSK-IMPACT dataset (orange) versus the SU2C-PCF dataset (red). Overall, the frequencies of alterations in these genes of interest are similar in the two datasets. (*) Does not include germline alterations.

Fig A3.

Comparison of primary localized samples from the MSK-IMPACT dataset to TCGA. (A) Gleason score comparison for TCGA and MSK-IMPACT primary localized tumors. More than 50% of primary localized tumors from the MSK-IMPACT dataset are Gleason 8-10 tumors, compared with approximately 25% in the TCGA dataset. (B) Frequencies of alterations in select genes in primary localized tumors from the MSK-IMPACT dataset (dark blue) versus the TCGA dataset (light blue). P-values are represented (Fisher’s exact test).

Fig A4.

Alterations in the PI3K pathway. (A) Oncoprint of somatic alterations in PI3K pathway genes. 24% of patients harbor a somatic alteration in one of the genes listed. (B) Mutations in PIK3CA, AKT1 and AKT3. Known activating hotspot mutations are labeled.

Fig A5.

Alterations in the MAP kinase pathway. (A) Oncoprint of somatic alterations in MAP kinase pathway genes. 5% of patients harbor a somatic alteration in one of the genes listed. (B) Missense mutations in MAP kinase pathway genes. Known activating hotspot mutations are labeled.

Fig A6.

Alterations in the Wnt-β catenin pathway. (A) Oncoprint of somatic alterations in Wnt-β catenin pathway genes. Fifteen percent of patients harbor a somatic alteration in one of the genes listed. (B) Mutations in APC, CTNNB1 and RNF43. APC alterations are primarily deletions and truncating mutations predicted to inactivate the protein. CTNNB1 alterations are primarily hotspot N-terminal missense mutations reported to prevent phosphorylation and degradation of the protein product β catenin.

Fig A7.

Somatic mutations identified in DNA damage repair genes. Shown here are missense, truncating and in-frame mutations in BRCA2, BRCA1, ATM and CDK12. The majority of somatic alterations in BRCA2 are predicted to result in a truncated version of the protein product or in loss of expression of the gene, while somatic alterations in ATM were primarily missense mutations occurring across ATM coding regions. Alterations in CDK12 were primarily truncation mutations, as previously reported in ovarian cancer.

FigA 8.

MMR/MSI mutation signatures in hypermutated tumors. K-means clustering of mutations/Mb (mut/Mb) across all 504 tumors identified two distinct clusters: cluster1 with Mutations/Mb10. The latter represents hypermutated tumors. Tumors in the hypermutated group correspond to the 8 patients with > 20 MSK-IMPACT mutations represented in Figure A5B. Mutational signature decomposition analysis for hypermutated tumors and control non-hypermutated tumors with >8mutation/Mb revealed a high contribution of MMR/MSI signatures to all hypermutated tumors, as measured by proportion of mutations attributed to a specific signature across 30 different mutational signatures.^,

Fig A9.

Fraction of patients with actionable and oncogenic alterations per OncoKB database. The dataset was queried for actionable and oncogenic alterations using the OncoKB genomic alteration annotation tool (www.OncoKB.org), and frequencies are represented for the 451 patients in the study. Actionable alterations are ranked by level of evidence (Level 2B: Standard of care biomarker predictive of response to an FDA-approved drug in another indication, but not standard of care for this indication; Level 3A: Compelling clinical evidence supports the biomarker as being predictive of response to a drug in this indication, but neither biomarker nor drug are standard of care; Level 3B: Compelling clinical evidence supports the biomarker as being predictive of response to a drug in another indication, but neither biomarker nor drug are standard of care; Level 4: Compelling biological evidence supports the biomarker as being predictive of response to a drug, but neither biomarker nor drug are standard of care). Only the highest level actionable alteration is represented per patient.

Fig A10.

Copy number variations represented as fraction of the genome altered across disease states. Copy number alterations increase from tumors in the locoregional disease state to tumors in the metastatic non-castrate state to tumors in metastatic castrationresistant disease.

Fig A11.

Mutations identified in AR. Missense mutations are represented as green dots, and an in-frame insertion as the single black dot. Well-characterized mutations in the ligand binding domain (blue box) are labeled.

Fig A12.

Frequency of alterations in select genes in metastatic castration-resistant prostate cancer (mCRPC) versus metastatic noncastrate tumors by site of biopsy. The genes listed showed a statistically significant enrichment in mCRPC when comparing all samples in the two sets (Figs. 5B and 5D). We compared alterations in these genes among (A) metastatic samples only (excluding prostate samples) and (B) lymph node samples only. Trends for enrichment in mCRPC still hold in most cases; however, statistical significance was often lost as a result of the lower number of samples included in these subsets. NS, not significant.