Panomic assessment reveals DNA repair alterations are common in prostate cancer (PC) and predicts potential therapeutic response to taxane-platinum combination therapy

Nancy Ann Dawson1, Elisabeth I. Heath2, Rebecca Feldman3, Sandeep K. Reddy3, David Spetzler3, George H Poste3,4, Derek Raghavan5 1Lombardi Comprehensive Cancer Center, Washington, DC; 2Karmanos Cancer Institute, Wayne State University, Detroit, MI; 3Caris Life Sciences, Phoenix, AZ; 4Complex Adaptive Systems Initiative, Arizona State University, Scottsdale, AZ; 5Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC

Background Results Results, contd. **Final analysis included an additional 81 patients Comparison of DNA Repair Deficient (DRD) and Intact (DRI) Prostate Cancers Abstract (# 5040 ) Table 1. DNA Damage Repair Genes assessed in this study • 100% There are few therapy options for advanced refractory prostate cancer [highlighted genes are regulated by or direct target genes of AR signaling; *DNA repair status panel] DRD (n=50) 87% 80% • Recent data have shown metastatic prostate cancer patients: Gene Platform DNA Damage Repair (DDR) Gene Platform DNA Damage Repair (DDR) 75% DRI (n=14) Background: Patients with PC have limited treatment options after 1 damage signaling; DSB repair; oxidative 60% 60% 60%  with DNA repair deficiencies (BRCA, ATM) respond to PARP inhibitors (olaparib) PTEN NGS; IHC damage signaling (indirct); DSB; NER MRE11A* NGS 58% 57% 54% stress 52% 50% 50% TP53 NGS damage signaling; DSB MUTYH* NGS MMR failure of hormonal and taxane therapy. Androgen receptor (AR)  have longer overall survival when treated with chemo[docetaxel]-hormonal 50% 40% NGS; IHC 36% 33%

ERCC1*/2*/3*/4*/5 NER POLE* NGS damage signaling (%) Frequency 2 (ERCC1) 27% 28% signaling may exert therapeutic effects on the DNA repair pathway in therapy vs. androgen-deprivation therapy (ADT) alone 25% 20% 20% BRCA1*/2* NGS HR; damage signaling RAD51*/51B*/50 NGS HR; damage signaling; DSB repair 16% 13% MLH1*/MSH2*/MSH6*P 7% 5% 7% PC. We have assessed the proteomic/genomic DNA repair aberrations NGS MMR WRN* NGS HR, BER 0% 0% 0% 0% 0% MS1*/PMS2* 0% • AR signaling evolves during the progression of prostate cancer and has been shown to HR; damage signaling; DSB; oxidative in primary (P) and metastatic (M) PC and explored the therapeutic ATM*/ATR NGS XPA NGS NER AR (high) AR mut. PTEN ↓ PTEN mut. MGMT ↓ ERCC1 ↓ PD1 + TILs PDL1 cMET EGFR HER2 TOP2A TOPO1 Optimal directly regulate genetic activities and pathways that influence DNA damage repair stress Taxane ATRX NGS NHEJ; DSB PALB2 NGS HR; damage signaling 100% Sensitivity implications of these mutations using panomic next generation mechanisms3,4 DDB2 NGS damage signaling ATRX NGS NHEJ Profile

BLM* NGS HR NBN* NGS DSB 76% sequencing (NGS). We hypothesized that there is a differential in CHK1*/2* NGS HR; damage signaling BARD1 NGS HR 75% 71% 70% Figure 4. Comparison of DNA Repair FANCA*/C*/D2*/E*/F*G* 61% NGS HR; ICL repair BRIP1 NGS HR; damage signaling Deficient (DRD) and Intact (DRI) • We sought to explore complex relationships between DNA repair genes, biomarkers /L* gene expression and mutation between P and M tumors. 44% prostate cancers. Using a panel of 50% 40% predictive of chemotherapy and androgen receptor status in primary vs. metastatic PRKDC* NGS NHEJ 30 DNA repair genes, deficiency was MMR (mismatch repair); DSB (double strand break); NHEJ (non-homologous end-joining); (%) Frequency 29.00% 28% 29% 30% 21% defined as a mutation in at least 1 prostate cancers to identify novel therapy approaches, and potentially new HR (homologous repair); BER (base excision repair); NER (nucleotide excision repair) 25% 16% 12% 14% gene. DNA repair deficiency was Methods: Molecular profiles of 437 PC tumor samples were defined. 7% combination strategies 0% detected in 78% of PC. Chi-squared 0% test was used to determine Protein expression (IHC), gene amplification (ISH) and sequencing 32-50 51-60 61-70 71-80 81-90 Metastatic Visceral Non-Visceral AR Histoscore [intensity x % staining] significant differences (yellow- Age Groups Disease Location of Metastasis highlighted groups) between DRD (NGS) were performed. A panel of 30 DNA repair genes was used to 250 Status and DRI. 100

Methods 194 define DNA repair intact (DRI) and DNA repair deficient (DRD) 200 189 50 subgroups. Unclassified variants were included for analysis. 150 518 advanced prostate cancer patients were included in this analysis and tested centrally 0 Biomarker Differences in Primary and Metastatic Prostate Cancers 0 50 100 150 200 250 300 Pearson’s chi-squared test was used to test for significant differences. at a CLIA laboratory (Caris Life Sciences, Phoenix, AZ). Tests included one or more of the 100 (n) Frequency 80% Histoscore metastatic 63% Frequency (n) Frequency 60% 54% following: gene sequencing (MiSeq and NextSeq Illumina platforms), copy number 50 25 21 43% 20 12 primary 39% 8 7 4 3 3 2 36% variation (NextSeq Illumina) and protein expression (immunohistochemistry [IHC]). 0 40% 32% 28% Results: Biopsies from 437 PCs (median age 67) were studied. 0 27% Cutoff for PTEN, ERCC1 and MGMT : >10% is positive. AR High is defined as 3+ 100% Intensity: 0+ 1+ 2+ 3+ 1+ 2+ 3+ 1+ 2+ 3+ 1+ 2+ 3+ 20% Frequency (%) Frequency 3% % cell staining: 100% 1-10% 11-50% 51-75% 76-100% 2% 2% Specimens submitted for profiling included 158 P PCs (36%) and 279 staining. Additional cutoffs and antibodies are available upon request. 0% ATM_SEQ MGMT ↓ ERCC1 ↓ AR High cMET cMET_SEQ TP53_SEQ TOP2A TS TUBB3 RRM1 M PCs (64% [18% bone; 37% visceral; 24% lymph nodes; 21% other Figure 2- Distribution of AR staining in prostate cancer patients (n=488) and DRD (n=54) DRI (n=14) Figure 7. Biomarker differences between primary and metastatic PC. sites]). The most frequently mutated DNA repair genes included conversion of immunohistochemistry results into histoscore (inset). Measure of Association: p=0.047 Biomarkers include those tested by IHC (MGMT, ERCC1, cMET, TOP2A, TS, TP53 (31%), ERCC5 (19%), FANCG (16%), MSH6 (13%), POLE (10%), Results Tumor and Patient Parameters Figure 6 – Box Plot comparing AR TUBB3 and RRM1) and sequencing (ATM, cMET, TP53). Only biomarkers PMS1 (13%), PTEN (9%) and BRCA2 (6%). Functional protein loss as Frequency (%) of Mutations Histoscore to DNA Repair Status showing statistical differences (p<0.05) are shown. A 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 measured by IHC was seen in ERCC1 (44%), MGMT (39%), and PTEN TP53 (126/367) ERCC5 (13/64) (43%). In a limited cohort of patients tested using a 592-gene hybrid- Visceral PMS2 (7/46) Conclusions PMS1 (9/63) • Panomic assessment reveals frequent alterations in DNA repair genes in prostate cancer capture NGS, 26/31 (84%) had alterations in at least 1 DNA repair POLE (9/63) Frequency (%) of Protein Loss 36% 14% Bones & Joints PTEN (32/328) BARD1 (6/63) 0 20 40 60 6% • DRD tumors associate with higher rates of androgen receptor staining, which supports previous gene. DRD PC exhibited higher expression rates of AR (57% vs. 20%; Primary MSH6 (5/61) 38% in vitro findings3,4 regarding the important role of AR/androgen signaling in DNA repair n=185 Lymph Nodes BRCA2 (14/172) PTEN (n=464) 42.2 p=.048) and TOPO1 (88% vs. 40%; p=.02) than DRI PC. An optimal FANCG (5/64) mechanisms. DRD tumors peak in the 61-70 age group and associate with higher rates of 64% 23% PRKDC (5/64) ERCC1 (n=210) 48.1 Metastatic FANCA (4/62) TOPO1. A taxane sensitivity profile is present in 20% of DRD tumors, suggesting a potential role taxane therapeutic response profile was observed in 20% of DRD Connective Tissues n=333 19% FANCF (4/64) MGMT (n=366) 38.80 for platinum-taxane combination in a subset of patients with DNA repair defects. RAD50 (3/56) tumors. Significant differences between P and M tumors were seen in WRN (3/58) Other (e.g. peritoneum, CHK2 (3/60) • PTEN loss and mutations associated with DRI tumors. Loss of PTEN has been implicated in ERCC1, AR, ATM and TP53. M tumors had significantly increased omentum, mediastinum) ATM (18/371) genomic instability, whereby loss of function leads to lowered DNA repair rates (through FANCL (3/61) expression of TOP2A, TS and TUBB3. BLM (3/63) interactions with DNA repair genes like p53 and Chk1) suggesting an alternate mechanism of B C ERCC4 (3/63 3+3 Figure 1 – reducing DNA Repair efficiency. 50% MRE11A (3/62) 43% Gleason 6 Distribution of (A) RAD51B (3/62) Figure 3 – Frequency of mutations detected by Illumina MiSeq

NBN (3/62) • Primary and metastatic PC exhibit differential protein expression and mutation rates, indicating 40% Conclusion: DNA repair defects are common in PC with a difference BRCA1 (6/171) Gleason 7 3+4 4+3 specimen sites (PTEN/TP53/BRCA1/2 only) and NextSeq next-generation ATRX (1/29) therapeutic targets may change through the progression of disease, and the need for molecular 30% 3+5 sequencing and functional protein loss in DNA Repair genes in gene expression and mutation between P and M tumors. 26% utilized for FANCC (2/63) profiling through the course of disease progression. Gleason 8 4+4 19% FANCD2 (2/63) (IHC). Mutations are classified as pathogenic or presumed Differential expression between African American and Caucasian 20% molecular profiling, ATR (2/62)

Gleason 9 4+5 5+4 (B) age of patients BRIP1 (2/62) pathogenic (dark blue bars) or variants of unknown significance Frequency (%) Frequency patients and further classification of variants are currently being 10% 5% 7% PALB2 (2/64) or unclassified variants (light blue bars). Inset chart shows the Gleason 10 5+5 and (C) Gleason FANCE (1/54) References assessed. Taxane-platinum combination chemotherapy should be 0% ERCC1 (1/64) frequency of protein loss in DNA Repair genes , PTEN, ERCC1 1. Mateo, J., J.S. de Bono, et al. (2015). “DNA-Repair Defects and Olaparib in Metastatic Prostate Cancer.” NEJM 373(18): 1697-1708. scores (available for MSH2 (1/61) 32-50 51-60 61-70 71-80 81-90 0% 25% 50% and MGMT. Variants were not detected in DDB2, ERCC2, ERCC3, 2. Sweeney, C.J., R.S. DiPaola, et al. (2015). “Chemohormonal Therapy in Metastatic Hormone-Sensitive Prostate Cancer.” NEJM 373(8): 739-746. tested specifically in DRD PC. a subset; n=122). RAD51 (1/63) 3. Polkinghorn, W.R., C.L. Sawyers, et al. (2013). “Androgen Receptor Signaling Regulates DNA Repair in Prostate Cancers.” Cancer Discov 3(11):1245-53. Age Group [median: 67] Frequency (%) XPA (1/63) CHK1, MUTYH. 4. Goodwin, J.F., K.E. Knudsen, et al. (2013). “A hormone-DNA repair circuit governs the response to genotoxic insult.” Cancer Discov 3(11): MLH1 (4/376) doi:10.1158/2159-8290.CD-13-0108.