Circulating Tumor Cell Genomic Evolution and Hormone Therapy Outcomes in Men with Metastatic Castration-Resistant Prostate Cancer

Abstract

Men with circulating tumor cell (CTC) AR-V7-positive metastatic castration-resistant prostate cancer (mCRPC) have worse outcomes when treated with enzalutamide/abiraterone. However, most men lack CTC AR-V7 detection, and additional predictive biomarkers are needed. We conducted a retrospective secondary analysis of the prospective PROPHECY trial (NCT02269982) of men with mCRPC undergoing treatment with enzalutamide/abiraterone, analyzing pooled CTC and germline DNA for whole-genome copy-number alterations (CNA) in 73 samples from 48 men over time along with pooled CTC and germline whole-exome sequencing on 22 paired samples before and following progression on androgen receptor (AR) inhibitor therapy to identify somatic genomic alterations associated with acquired resistance. We observed broad interpatient and longitudinal CTC genomic heterogeneity from AR-V7-negative men with mCRPC, including common gains of KDM6A, MYCN, and AR, and loss of ZFHX3, BRCA1, and PTEN. Men who had progression-free survival of ≤3 months despite enzalutamide/abiraterone treatment were more likely to have baseline CTC genomic loss of CHD1, PTEN, PHLPP1, and ZFHX3 and gains of BRCA2, KDM5D, MYCN, and SPARC. After progression on abiraterone/enzalutamide, we observed clonal evolution of CTCs harboring TP53 mutations and gain of ATM, KDM6A, and MYC, and loss of NCOR1, PTEN, RB1, and RUNX2. CTC genomic findings were independently confirmed in a separate cohort of mCRPC men who progressed despite prior treatment with abiraterone/enzalutamide (NCT02204943). IMPLICATIONS: We identified common and reproducible genomic alterations in CTCs from AR-V7-negative mCRPC men associated with poor outcomes during enzalutamide/abiraterone treatment, including CNAs in genes linked to lineage plasticity and epigenetic signaling, DNA repair, AR, TP53/RB1, PTEN, and WNT pathways.

Conflict of interest statement

Conflicts of interest

1. SH sits on DSMB for Ferring, Bayer, and Eisai and receives funding via ASCO as a statistician on the TAPUR trial.

2. GK, MA, PG, DMN and SG have no conflicts of interest.

3. DJG has research support (to Duke) from Acerta, Astellas, BMS, Bayer, Calithera, Exelixis, Janssen, Myovant, Pfizer, Novartis, Sanofi Aventis. Consulting income from Vizuri health sciences, UroToday, Sanofi, Pfizer, Nektar, Myovant, Modra, Merck, Ipsen, Flatiron, Exelixis, Capio, EMD Serono, BMS, Bayer, AstraZeneca, Astellas.

4. SGG is a paid consultant for Pfizer and receives research funding to the institution from Janssen Pharmaceuticals.

5. AJA is a paid consultant with Pfizer, Astellas, Janssen, Bayer, AstraZeneca, and Merck and receives research funding (to his institution) from Pfizer, Astellas, Janssen, Bayer, Dendreon, Novartis, Genentech/Roche, Merck, BMS, AstraZeneca, Constellation, Beigene.

©2021 American Association for Cancer Research.

Figures

Figure 1:. Summary of clinical specimens for genome-wide copy number alterations and mutations study.

Overview of the PROPHECY mCRPC patients cohorts utilized for the biomarker analyses.

Figure 2:. Overall genomic-wide copy number alterations analysis in CTCs DNA and independent clinical validations.

In this prospective study, we analyzed a panel of sixty genes identified earlier in prostate cancer progression and drug resistance across the spectrum of all baseline and progression groups. (A) The overall prevalence of genomic somatic alterations detected in CTCs DNA; copy gain and loss in 45 baseline CTCs DNA from mCRPC men, and ranked by their prevalence, as manifested in the bar charts. (B) Showing the overall prevalence of somatic alterations in 28 CTCs DNA from 28 mCRPC men. We detected widespread tumor heterogeneity between patients, and common CNAs, such as copy gain in KDM6A, FOXA1, MYCN, and AR, and loss in BRCA1, PTEN, and ERG.

Figure 3:. Genomic-wide CNAs analysis in CTCs over time.

(A) Violin plots showing the number of copy number variant counts were called by Cytogenomics software and compared between baseline and paired progression CTCs DNA samples (N=25 mCRPC cases). (B) Violin plot displaying CellsSearch CTCs enrichment between baseline and progression (N=18 cases). (C) CellSearch CTC number comparison case by case; baseline versus progression (N=18 cases). (D) and (E) displaying genomic alterations over time; 25 paired baseline vs. progression CTCs DNA.

Figure 4:. CTCs clonal evolution in response to enzalutamide or abiraterone therapy.

(A) Progression-free survival (PFS); sensitive vs. resistant (N=23 vs. 14). (B) Violin plot showing the number of copy number variants at genome level was called by Cytogenomics software and compared in between sensitive vs. resistant CTCs DNA samples (N=23 vs. 14 cases), and (C) CellSearch CTCs enrichment between sensitive vs. resistant (N=22 vs. 14 cases). (D) Genomic alterations for targeted sixty genes; sensitive vs. resistant. (E) and (F) displaying the top-most genomic gain and loss in sensitive vs. resistant CTC-DNA, respectively, and illustrating the difference between sensitive and resistant CTC-DNA CNAs prevalence. (G) Overall heatmap of DNA damage and repair genes alteration associated with DNA damage repair pathways (ATM, BRCA2, APC, PALB2, or BRCA1) were validated in metastatic prostate cancer datasets in cBioPortal.

Figure 5:. Comprehensive CTCs DNA analysis approach identifies mutations in sensitive and resistant mCRPC cases over time.

(A) The line graph illustrates top COSMIC somatic mutations detected by whole-exome sequencing analysis, compared between baseline and paired post-treatment CTC-DNA (N=11) in a set of sixty genes. (B) The mutation hotspot for the TP53 gene is visualized in cBioPortal Mutation-Mapper, where read coverage is shown in the IGV browser, including matched WBC, baseline-CTC, and progression CTC DNA from the same patients, for genes SPOP (D144G) and TP53 (T111K) in patients 809–5 and 809–7, respectively. (C) Genes with copy gain, loss, and COSMIC mutations detected in 11 common baselines and progression CTCs DNA were combined and compared to each other, are shown in stacked bars. (D) The line graph exemplifies the top somatic mutations validated in the COSMIC database were compared between baseline sensitive (N=5) and resistant CTC-DNA (N=5). One out of eleven patients who had a PFS in between 3 and 6 months was excluded from the analysis in order to enrich for outlier responses at either extreme of clinical benefit as determined by PFS times. (E) Somatic mutation in the PTEN gene detected in baseline sensitive and resistant are presented in cBioPortal by Mutation-Mapper software. For example, PTEN (D178G) and CDK12 (E131K) mutations are visualized in the IGV browser. (F) Genes with copy gain, loss, and COSMIC mutations identified in five sensitive and five resistant baselines CTCs DNA were combined and compared, are shown as a bar graph.

Figure 6:

Conceptual model of CTCs evolution in mCRPC: sensitive versus non-responder mCRPC patients.