SLFN11 Expression in Advanced Prostate Cancer and Response to Platinum-based Chemotherapy

This article is featured in Highlights of This Issue, p. 1089

Prostate cancer is the second leading cause of male cancer-related death worldwide (1). Patients with metastatic prostate cancer typically respond well to androgen deprivation therapy, often administered together with docetaxel or second-generation hormonal drugs, but invariably progression to castration-resistant prostate cancer (CRPC) occurs within a median of 16 months (2). Therapies for metastatic CRPC include potent androgen receptor (AR)-targeted therapies, taxane chemotherapies, sipuleucel-T, radium-223, and a number of investigational drugs. With a growing armamentarium of approved and investigational agents, the identification of predictive biomarkers to help guide individual therapy and the optimal sequence of therapy for men with advanced prostate cancer is needed now more than ever. Even unapproved but widely available agents, such as platinum chemotherapy, may provide significant clinical benefit for select patients.

Prior clinical studies investigating platinum chemotherapy for patients with molecularly unselected advanced prostate cancer showed that the traditional platinum compounds (i.e., cisplatin, carboplatin, oxaliplatin), administered alone or in combination with other chemotherapies, resulted in moderate antitumor activity for the overall population (3). In a phase III trial of the oral platinum compound satraplatin for patients with metastatic CRPC, there was a delay in progression of disease, but no significant overall survival benefit (4). Nonetheless, exceptional responders to platinum have been reported (5–7), and platinum is sometimes still used clinically for patients with metastatic CRPC, particularly those with aggressive disease and/or following resistance to approved drugs.

DNA repair alterations involving homologous recombination genes such as BRCA1, BRCA2, and ATM, are present in up to 20% of patients with CRPC (at either germline or somatic level) and have been found to mediate both platinum sensitivity (5, 6, 8) as well as PARP inhibitor (PARPi) response in metastatic CRPC (9–11). The phase III PROfound clinical trial was recently reported as positive, demonstrating a rPFS benefit for the PARPi olaparib versus potent AR therapies for patients with CRPC with DNA repair aberrations (12). These observations have reinvigorated the field's interest and a number of additional clinical studies testing either platinum or PARPi for molecularly enriched patients with CRPC with DNA repair alterations at various stages of prostate cancer progression are underway.

In addition to those with genomic alterations involving DNA repair genes, there may be other patients that might preferentially benefit from receiving platinum. For instance, patients that develop treatment-related small-cell/neuroendocrine prostate cancer or those with aggressive variant prostate cancer (AVPC) clinical features respond to platinum-based chemotherapy regimens with similar response rates as seen as patients with small-cell lung cancer (13–17).

To explore other potential biomarkers of platinum sensitivity in prostate cancer, we investigated the role of Schlafen Family Member 11 (SLFN11). SLFN11 is a DNA/RNA helicase, first described for its role in normal thymocyte development and differentiation (18). SLFN11 is only expressed in vertebrates and especially in mammals, playing a critical role not only for thymocyte development, but also for immune response and cell proliferation (19). Later, SLFN11 expression was assessed in several cancer cell lines, including Ewing sarcoma, small-cell lung, ovarian, and colon cancers, and SLFN11 expression was found to be associated with response to DNA-damaging agents including platinum compounds, topoisomerase I and II inhibitors, and alkylator chemotherapies (20). In addition, SLFN11 expression has been associated with sensitivity to PARPi in small-cell lung cancer patient-derived xenograft models (21). In a recent randomized phase II trial of temozolomide in combination with either the PARPi veliparib or placebo in patients with relapsed-sensitive or refractory small-cell lung cancer (22), SLFN11 expression determined in tumor tissue by IHC was associated with a significantly improvement in clinical outcomes for patients with SLFN11-expressing tumors treated with veliparib compared with those that were SLFN11-negative.

To date, clinical studies investigating the role of SLFN11 as a potential mediator of sensitivity to DNA-damaging agents has been largely investigated in lung and ovarian cancers. In this study, we sought to identify SLFN11 expression patterns in both tissue and blood from advanced prostate cancer patients and to elucidate its potential predictive and/or prognostic role in platinum-treated patients.

Patients with metastatic CRPC and evaluable metastatic tissue and/or circulating tumor cells (CTC) treated with platinum-containing chemotherapy at a single institution were retrospectively identified. All patients had clinical features of castration-resistance defined according to Prostate Cancer Clinical Trials Working Group 3 (PCWG3) criteria (23) with a histology review confirming adenocarcinoma or small-cell/neuroendocrine prostate cancer using published morphologic criteria (24).

Previous systemic therapies for CRPC included AR-targeted therapies (e.g., enzalutamide, abiraterone), taxanes (e.g., docetaxel, cabazitaxel), radium-223, and investigational agents. Platinum-based treatments (monotherapy or combination therapy) were administered for a planned number of cycles according to regime-specific treatment protocols. The study was conducted in accordance with the Declaration of Helsinki and the Good Clinical Practice guidelines. The protocol was approved by the WCM Institutional Review Board (IRB#1305013903 and IRB#0905010441) and patients provided written informed consent.

Clinical and demographic information was collected by medical record review. PCWG3 criteria (23) was used to assess clinical, biochemical and radiographic response.

In patients with metastatic tumor biopsies and RNA sequencing (RNA-seq) data available, mRNA expression of SLFN11 was assessed (n = 27). These RNA-seq data were previously published (26) and are included in dbGaP phs000909.v.p1. Patients were stratified into two categories (high or low) according to the median value of SLFN11 expression. We also mined our published datasets (7, 25–27) to assess SFLN11 expression across stages of prostate cancer progression [31 benign, 66 localized prostate cancer, 74 CRPC-Adeno, 35 NEPC] and in other cancer types. Reads were mapped to the human genome reference sequence (hg19/GRC37) using STAR v2.3.0e. For each sample, HTSeq and Cufflinks were utilized to produce read counts and FPKM values, respectively (28). In addition, when available, metastatic tumor genomic status of select genes (i.e., AR, TP53, RB1, PTEN, BRCA2, BRCA1, ATM) was collected from whole-exome sequencing (WES) data obtained through a clinical CLEP/CLIA compliant tumor/normal WES assay (EXaCT-1; IPM-Exome-pipeline, version 0.9; ref. 29).

Whole blood (10 mL) from a subset of patients (n = 20) was collected using Streck tubes and shipped to Epic Sciences for processing using protocols described previously (30). CTC enumeration was performed, and slides were stained for DNA (DAPI), whole blood cell linage marker (CD45), and epithelial cells marker (CK). SLFN11 protein expression in CTCs was assessed by IF. The images of nucleated cells were visualized using a multiparametric digital pathology algorithm (28). The threshold for categorizing SLFN11 expression into “positive” (SLFN11+) and “negative” (SLFN11-) was locked at 6.0 and established following an analytic feasibility of the SLFN11 assay employing laboratory-derived samples (LDS). LDS consisted of healthy donor blood spiked with cell lines with varying levels of SLFN11 expression to mimic physiologically appropriate ranges of SLFN11 expression (Supplementary Methods).

An organoid model (WCMO1388) derived from fresh metastatic tumor tissue using methods described in Puca and colleagues (31) was assessed for SLFN11 expression by Western blot analysis. To stably knockout SLFN11 in organoids, a single-guide RNA was cloned into lentiCRISPRv2 vector [from Feng Zhang, the Broad Institute of MIT and Harvard (Addgene plasmid #52961; http://n2t.net/addgene:52961; RRID: Addgene_52961)] and transfected into 293FT cells using Lipofectamine 3000 (Thermo Fisher Scientific) to produce lentiviral supernatants. Organoids were dissociated into single cells using TrypLE (Thermo Fisher Scientific) and infected with lentivirus harboring either CRISPRv2-sgGFP or CRISPR-sgSLFN11 (MOI = 1) for 23 hours. After infection, cells were replated and cultured in media with 1 μg/mL of puromycin for three days. After puromycin selection, protein lysates were collected and analyzed by immunoblot using anti-SLFN11 (Santa Cruz Biotechnology, 1:250) and anti-GAPDH (Cell Signaling Technology) antibodies. To test the sensitivity to cisplatin and olaparib (both purchased from Selleckchem), organoids were dissociated to single cells, and 3,000 cells were plated in a 96-well plate coated with 1% collagen I; cisplatin or olaparib was added on the following day. Cell viability was measured using CellTiter-Glo assay (Promega) following the manufacturer's protocol after 6 days of treatment. The sequences of sgRNAs were sgGFP:GGGCGAGGAGCTGTTCACC; sgSLFN11-1:CCCAATTCATGGATAGTGG; sgSLFN11-6:AGCCTGACAACCGAGAAAT.

The primary aim of this study was to assess the impact of SLFN11 expression on outcomes in patients with advanced prostate cancer treated with platinum-based chemotherapy. Radiographic progression-free survival (rPFS) was calculated from the first day of platinum-based therapy to the date of progression disease or death. Progression disease was defined using PCWG3 criteria (23). Overall survival (OS) was calculated from initiation of therapy to death from any cause. Patients still alive at time of last follow-up were censored. A multivariate Cox regression model was used to investigate potential predictors of rPFS and OS and to estimate HR and their 95% confidence intervals (95%CI).

Forty-one patients with metastatic CRPC treated with platinum-based chemotherapy with metastatic tumor tissue sufficient for RNA-seq and/or CTCs for analysis were identified. Median age was 67.1 years (range 50.6–90.7). Metastatic tumor pathology review revealed adenocarcinoma (n = 21) or small-cell/neuroendocrine (n = 20) histology. Median lines of prior systemic therapy for CRPC was two (range 1–7). All patients underwent platinum-based chemotherapy treatment [38 in combination with etoposide (n = 21) or taxanes (n = 17)]. Data regarding indication or reasons for administering platinum chemotherapy were not available. Specimens were collected with a median of 85 days (range 0–707 days) between the time of specimen collection and platinum treatment. Clinical features are summarized in Table 1.

We first evaluated SLFN11 mRNA expression by RNA-seq across multiple solid tumors at our institution (7, 25) and confirmed overexpression of SLFN11 in a subset of lung and ovarian cancers (Supplementary Fig. S1) consistent with what has been reported previously (18–22). We also identified SLFN11 overexpression in a subset of prostate cancers. When we looked across additional prostate cancer cohorts (n = 197), SLFN11 was overexpressed in approximately 45% of metastatic CRPC compared with 25% of primary prostate cancers (Fig. 1). SLFN11 expression was lower in neuroendocrine prostate cancer compared with CRPC with adenocarcinoma histology; Wilcoxon test neuroendocrine (NE) versus castration-resistant adenocarcinoma P = 0.009, NE versus localized prostate adenocarcinoma P = 0.019, NE versus Benign P = 0.002 (Fig. 1). In the 27 of 41 patients treated with platinum chemotherapy with metastatic tumor tissue RNA-seq data, 58.5% expressed high levels of SLFN11 in metastatic biopsies (with a cutoff determined according to the median value of SLFN11 mRNA expression; Fig. 2A). Of these 27 patients, we observed a longer rPFS when treated with platinum chemotherapy in patients with SLFN11 overexpression compared with those without SLFN11 overexpression (6.9 vs. 2.8 months, HR = 3.72, 95% CI 1.56–8.87, P < 0.001; Fig. 2B). No significant association was observed between SLFN11 expression and OS (Fig. 2C). A greater than 50% decline in serum PSA was observed in all patients with SLFN11 overexpression (Fig. 2F).

We evaluated 23 blood samples from 20 patients with CRPC treated with platinum chemotherapy. Seven of these patients also had matched metastatic tissue biopsies with RNA-seq data. The median number of CTCs/mL was 23 (range 1–136). Detectable SLFN11 expression in CTCs (SLFN11+) by IF was identified in 16 of 23 samples (70%). Patients with SLFN11⁺ CTCs exhibited a dynamic range in their proportion of SLFN11⁺ CTCs from 4% to 100% (Fig. 3). SLFN11 expression in CTCs did not correlate with cytokeratin (CK) expression in CTCs, and SLFN11⁺ cells had a range of CK expression (Supplementary Fig. S2).

SLFN11 expression in CTCs of patients was concordant with SLFN11 mRNA expression by RNA-seq of matched metastatic tumor biopsies in 6 of 7 patients (85.7%). Only one discordant case was identified (with low SLFN11 expression in tissue but positive CTCs); this case harbored heterogeneity of SLFN11 expression across individual CTCs (Supplementary Fig. S3).

Similar to what was observed in tumor tissues, PSA declines of ≥50% on platinum chemotherapy occurred in all patients with SLFN11⁺ CTCs. A longer rPFS was also observed in patients with SLFN11⁺ CTCs compared with those without (6.0 vs. 2.2 months, HR = 4.02, 95%CI 0.77–20.86, P = 0.002; Fig. 2D), and this was independent of the number of CTCs present. There was no association between SLFN11 expression in CTCs and OS (Fig. 2E).

Representative images (Fig. 4) showing CTC characteristics of one patient analyzed at two different time-points suggested a change of SLFN11 expression status. This patient transitioned from SLFN11 negative to SLFN11⁺ CTCs after approximately 24 months from the first SLFN11 assessment and four therapeutic lines (abiraterone, radium-223, docetaxel, enzalutamide) in between. SLFN11⁺ CTCs on the second blood collection was detected 10 days before starting cisplatin and etoposide as fifth-line therapy. He had a PSA decline of 16.7 ng/mL to 4.5 ng/mL and a radiographic partial response after six cycles of cisplatin and etoposide, which was durable for 6 months. He did not have somatic DNA repair alterations, neuroendocrine histology, or clinical features of AVPC.

We performed a multivariate analysis which included SLFN11 overexpression (by metastatic tumor or CTCs), age, histology (adenocarcinoma versus neuroendocrine), presence of visceral metastasis, serum PSA, serum chromogranin A, somatic alterations of AR, TP53, RB1, PTEN, DNA repair genes (BRCA1/2 or ATM; Fig. 2G). Frequencies of aberrations are listed in Table 1; 13 of 32 assessable patients had alterations in AR, 21 had loss of one tumor suppressor (TP53, PTEN, RB1), 15 had loss of two or more tumors suppressors, 11 patients had DNA repair gene aberrations. SLFN11 overexpression was independently associated with rPFS (HR = 0.14, 95% CI 0.03–0.70, P = 0.017). There was no association between SLFN11 expression and OS. Conversely, multivariable analysis identified neuroendocrine histology and genomic alterations involving AR and TP53 as independent predictors of inferior OS (HR = 0.11, 95% CI 0.02–0.57, P = 0.009, and HR = 0.02, 95% CI 0.01–0.24, P = 0.002, and HR = 0.09, 95% CI 0.01–0.79, P = 0.030, respectively), but not PFS on platinum-based chemotherapy.

We utilized CRISPR-Cas9 to knockout SLFN11 in a human prostate cancer organoid, WCMO1388, and examine its effect on cisplatin sensitivity. This organoid demonstrated loss of multiple tumor suppressors (i.e., PTEN, TP53, RB1), and also expressed SLFN11; it did not harbor BRCA1, BRCA2, or ATM genomic alterations (Supplementary Table S1). Despite other possible mediators of platinum sensitivity, SLFN11 knockout with two independent sgRNAs resulted in a 3-fold increase in IC₅₀ with cisplatin compared with control (sgGFP; Fig. 2H). Moreover, SLFN11 knockout also resulted in enhanced resistance to cisplatin at lethal concentrations. Given the reported association between SLFN11 and PARPi sensitivity (21), we also performed the same experiments using the PARPi olaparib. We found that SLFN11 knockout also reduced response to olaparib with an approximate 1.5-fold increase in IC₅₀ (Supplementary Fig. S4).

Platinum-based chemotherapy has demonstrated moderate antitumor activity in unselected patients with metastatic CRPC but has been associated with significant clinical benefit and even exceptional responses in select patients particularly those with DNA repair alterations (5–8) or aggressive clinical features (15, 16).

Here we report a cohort of 41 men with metastatic CRPC who received platinum chemotherapy after progression on standard therapies. We discovered that overexpression of SLFN11 was associated with longer rPFS on platinum compared with those without SLFN11 overexpression as well as significant PSA responses. This was regardless of the presence of DNA repair gene alterations or small cell/neuroendocrine carcinoma morphology.

We found that SLFN11 is overexpressed a subset of CRPC tumors but less so in primary prostate cancers and may be acquired in some cases. As repeat metastatic biopsies are invasive to perform in patients, we investigated the feasibility of using CTCs to detect SLFN11 expression. CTCs have been shown to capture disease heterogeneity in prostate cancer as well as identify the expression of key pathways and targets through in in situ analyses or single-cell profiling (28, 32–35). In the subset of cases where we had matched tumor tissue and CTCs, we demonstrated high concordance of SLFN11 expression.

Schlafen (from the word schlafen, which in German means sleeping) family genes were originally identified during screening for growth regulatory genes during lymphocyte development. In cancer, SLFN11 has been identified as mediator of platinum sensitivity, including in clinical studies of non–small cell and small-cell lung cancer and ovarian cancer (22, 36, 37). In addition to platinum chemotherapy, SLFN11 expression has also been associated with response to PARPi in lung cancer (21, 22). Our patient-derived organoid data with olaparib suggests the same may also be the case for prostate cancer. There are several PARPi trials currently underway in advanced prostate cancer that could further investigate this question.

The mechanisms by which SLFN11 is regulated in CRPC and how this contributes to platinum and/or PARPi sensitivity in prostate cancer require further study. SLFN11 is a DNA/RNA helicase that is actively recruited to sites of DNA damage and regulates replication stress. SLFN11 has been found to directly interact with replication protein A1 (RPA1) to destabilize RAP-single strand DNA complexes and inhibit homologous recombination (38). This may explain how overexpression of SLFN11 contributes to tumor cell response to DNA-damaging agents. SLFN11 has also been found to extend cell-cycle arrest at S phase, resulting in DNA double strand breaks and replication stalling in the presence of PARP inhibitors (39).

In small-cell lung cancer, studies using patient-derived xenografts models (40) have identified an EZH2-SLFN11 axis leading to a gene silencing of SLFN11 with marked deposition of H3K27me3, a histone modification placed by EZH2, contributing to platinum resistance. EZH2 inhibition also contributes to PARPi response in breast cancer through PARP-mediated poly-ADP ribosylation (41). It has also been shown that SLFN11 gene promoter hypermethylation correlates with resistance to platinum agents, particularly in ovarian and lung cancers (37). These data suggest that the expression levels of SLFN11 may be epigenetically regulated. Targeting epigenetic modifications might serve as a means for reexpressing SLFN11 and potentially sensitizing tumors to DNA-damaging agents (40).

Limitations of this study include a small sample size, retrospective design, and variability across patients in the clinical cohort including their number of prior and subsequent treatments. Additional mechanistic studies in prostate cancer are also warranted to better understand SLFN11 regulation and its impact on DNA repair processes and therapy response. Nonetheless, these data identify SLFN11 as a potential marker of platinum sensitivity in CRPC and pave the way for future studies. Additional studies are needed to translate these findings into biomarker-informed clinical decision making for men with advanced prostate cancer.

V. Conteduca has received speakers bureau honoraria from Bayer, Sanofi, Astellas, and Janssen. L. Puca is an associate director, biology at Petra Pharma Corporation. R. Dittamore is chief of medical innovation at Epic Sciences, Inc. S.T. Tagawa is a consultant at Pfizer, Clovis, Amgen, Janssen, Bayer, and Novartis and reports receiving a commercial research grant from Sanofi, Astellas, Janssen, Bayer, Clovis, Merck, Amgen, Novartis, and Abbvie. No potential conflicts of interest were disclosed by the other authors.

Conception and design: V. Conteduca, R. Dittamore, H. Beltran

Development of methodology: S.-Y. Ku, M. Slade, L. Fernandez, J.M. Mosquera, H. Beltran

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.-Y. Ku, L. Fernandez, J. Hess, P.J. Vlachostergios, M. Sigouros, J.M. Mosquera, D.M. Nanus, R. Dittamore, S.T. Tagawa, H. Beltran

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Conteduca, S.-Y. Ku, L. Puca, M. Slade, L. Fernandez, R. Bareja, P.J. Vlachostergios, J.M. Mosquera, A. Sboner, O. Elemento, H. Beltran

Writing, review, and/or revision of the manuscript: V. Conteduca, S.-Y. Ku, M. Slade, L. Fernandez, P.J. Vlachostergios, M. Sigouros, J.M. Mosquera, D.M. Nanus, R. Dittamore, S.T. Tagawa, H. Beltran

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.-Y. Ku, L. Puca, L. Fernandez, S.T. Tagawa

Study supervision: S.T. Tagawa, H. Beltran

This work was supported by the Prostate Cancer Foundation (to H. Beltran), Department of Defense PC121341 (to H. Beltran.), NIH/NCI SPORE in Prostate Cancer P50-CA211024 (to O. Elemento, J.M. Mosquera, H. Beltran).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.