External Validation of Association of Baseline Circulating Tumor Cell Counts with Survival Outcomes in Men with Metastatic Castration-Sensitive Prostate Cancer

Intensification of androgen-deprivation therapy (ADT) with novel hormonal therapies (NHTs; abiraterone, apalutamide, or enzalutamide) or docetaxel has shown to improve survival outcomes for patients with newly diagnosed metastatic castration-sensitive prostate cancer (mCSPC) in randomized controlled trials and is considered standard of care (1–5). However, recent studies show an alarming underutilization of intensified ADT in the real world, with < 30% of patients receiving intensified ADT in three distinct and large cohorts of patients in the United States (6–9). Furthermore, even with intensified ADT ∼20% of patients with mCSPC experience disease progression within a year of starting treatment, and experience inferior outcomes (1–5). Up front identification of these men at the highest risk of progression and/or death will not only improve patient counseling and prognostication, but also drive clinicians to offer intensified ADT to the patients or preferentially enroll them in clinical trials.

Yet, there is no biomarker available in clinic use to identify these patients with a new diagnosis of mCSPC. Genomic alterations such as in AR and cell-cycle genes have been associated with poor overall survival (OS), while alterations in SPOP and WNT pathway have been associated with better OS (10, 11). However, comprehensive genomic profiling (CGP) of the tumor tissue may not be possible in up to one-third of patients in the real world, needs a few weeks to obtain the results, and is expensive (12).

Circulating tumor cells (CTCs) have been established as a reliable predictive and prognostic marker in the setting of metastatic castration-resistant prostate cancer (mCRPC; ref. 13, 14). CellSearch is the only FDA-approved method for CTC enumeration and is readily available in the clinic (13, 14). Recently, in a prospective randomized phase III trial of ADT combined with orteronel or bicalutamide in patients with mCSPC (SWOG S1216), higher baseline CTCs (≥ 5 vs. 0/7.5 mL) predicted a lower likelihood of 7-month PSA responses [≤ 0.2 ng/mL; an intermediate endpoint for OS and inferior 2-year progression-free survival (PFS); ref. 15].

Herein, we aimed to validate these findings in men with mCSPC in the real world and interrogate tumor genomic profiles with respect to the CTC level.

After institutional review board approval, we identified consecutive patients who received ADT with or without intensification with docetaxel or NHTs for a new diagnosis of mCSPC and had enumeration of baseline CTCs by FDA cleared CellSearch CTC assay at metastatic diagnosis before treatment initiation (13, 14). CTC counts were categorized as 0, 1–4, and ≥ 5/7.5 mL. Gene alterations were determined by CGP of baseline primary or metastatic tumor tissue (Foundation Medicine, Cambridge, MA; Clinical Laboratory Improvement Amendments—certified), which interrogates all exons from 324 cancer-related genes and has been described previously (16). All variant types were analyzed, including small mutations, copy-number alterations, and gene rearrangements. Variants of unknown significance were excluded (16). The primary endpoint was to investigate the PFS and OS by CTC enumeration using the same cutoffs described in the post hoc analysis of the SWOG 1216 trial. The secondary endpoints included assessing the correlation between number of genes altered, number of mutations per patient, or proportion of specific altered genes with the separate CTC enumeration brackets.

OS was defined from the time of ADT initiation to the time of death from any cause or loss to follow-up. PFS was defined from the time of ADT initiation to the time of either PSA, radiographic, or clinical progression (per Prostate Cancer Working Group 2 criteria), whichever occurred first, or death from any cause. The relationships between CTC counts and both PFS and OS were assessed by Cox proportional hazards models, both unadjusted and adjusted for age, Gleason score, PSA at ADT initiation, de novo versus non–de novo metastatic disease at diagnosis, presence or absence of ADT intensification therapy, volume of disease (high or low per CHAARTED criteria; ref. 5), number of genes altered, and categorized CTC counts (0, 1–4, and ≥ 5/7.5 mL). Relationships between CTC counts, and the number of genes altered and individual gene alterations (after excluding variants of unknown significance) were assessed via Kruskal–Wallis and χ² tests, respectively. Throughout, results are only presented for individual genes where they show suggestive evidence (P < 0.10) of an association with either CTC counts, PFS, or OS as appropriate.

The data generated in this study are not publicly available as they could compromise patient privacy but are available upon reasonable request from the corresponding author.

Overall, 103 patients were eligible. Baseline patient characteristics and subgroup comparisons are summarized in Table 1. There was no evidence of differences in age, Gleason score, PSA at ADT initiation, percentage of patients with de novo metastatic disease, or volume of disease between patient subgroups who received or did not receive ADT intensification therapy (all P ≥ 0.158). Median time between tumor biopsy and CTC enumeration was between 1.3 months prior to, and 0.6 months after CTC enumeration. Patients who underwent ADT intensification therapy tended to have higher CTC counts at ADT initiation than patients who did not (40.9% vs. 16.9% with CTC counts ≥ 5, P = 0.025). The 2-year PFS of the cohort was 48% [95% confidence interval (CI), 39%–59%], and 2-year OS was 87% (95% CI, 80%–94%). Median PFS was 23 (95% CI, 20–39) months and median OS was not reached.

Kaplan–Meier survival estimates stratified by ADT intensification therapy, number of genes altered, and CTC counts at ADT initiation for both OS and PFS are presented in Fig. 1.

In univariate and multivariate analyses, CTC counts, and the number of genes altered were strongly associated with both PFS and OS (Table 2). The unadjusted HR for CTC ≥ 5 versus 0 with regard to PFS was 3.53 (95% CI, 1.95–6.38; P < 0.001), and for OS, it was 4.55 (95% CI, 1.67–12.40; P = 0.003). The unadjusted HR per gene altered with regards to PFS was 1.17 (95% CI, 0.99–1.38; P = 0.068), and for OS, it was 1.55 (95% CI, 1.21–1.99; P = 0.001).

In multivariate analyses, adjusted for age, Gleason, PSA (log) at ADT initiation, de novo status, volume of disease, number of genes altered, and presence or absence of ADT intensification therapy, the PFS HR for CTC ≥ 5 versus 0 was 4.52 (95% CI, 1.84–11.11; P = 0.001), and the OS HR was 3.59 (95% CI, 0.95–13.57; P = 0.060). Similarly, in multivariate analyses, the PFS HR per gene altered was 1.33 (95% CI, 1.03–1.71; P = 0.031), and OS HR per gene altered was 1.65 (95% CI, 1.16–2.35; P = 0.005). Univariate associations of PFS and OS with different genomic alterations are summarized in Supplementary Table S1.

Patients with greater CTC counts had a greater number of altered genes (P = 0.017) and a greater number of total alterations (P = 0.017; Fig. 2). Among patients harboring TP53 alterations, the proportion of patients with CTC ≥ 5 was significantly higher than those with a CTC count of 0 (P = 0.036), whereas with MYC (P = 0.072) and ATM (P = 0.093), alterations were marginal (Fig. 3A–C). No other individual genes showed suggestive or stronger evidence of a relationship to CTC counts (all P ≥ 0.10). Alterations categorized by CTC counts along with individual patient characteristics such as age, Gleason score, PSA levels, de novo status and volume of disease are summarized in Supplementary Fig. S1.

CTCs were first reported in 2008 as the most accurate and independent predictor of OS in patients with mCRPC leading to the assay approval by the FDA for evaluation in this disease setting (14). A meta-analysis of five prospective randomized phase III trials enrolling a total of 6,081 patients (COU-AA-301, AFFIRM, ELM-PC-5, ELM-PC-4, and COMET-1) showed that CTC0 (change from detectable to undetectable CTCs) and CTC conversion had a higher discriminatory power for OS as compared with PSA change supporting its use as a better response biomarker (16). Moreover, the prognostic biomarker status of CTCs is independent and complementary to androgen receptor activity/signaling pathways (17, 18).

SWOG S1216 recently reported CTCs as a prognostic biomarker in mCSPC (15). Herein, we, for the first time, externally validate the previous findings from a prospective trial (15) of the association of higher CTC counts with inferior survival outcomes in an independent cohort of patients with mCSPC through CellSearch assay. We show that the prognostic nature of CTCs is independent of baseline characteristics and not a surrogate for disease volume. We also show that a higher CTC continues to be associated with a worse prognosis with both ADT and ADT with treatment intensification.

Our study also reveals that patients with mCSPC who had higher CTC counts had a significantly higher number of altered genes and total alterations. The rate of patients with CTC counts ≥ 5 was significantly higher among patients harboring TP53 alterations (Fig. 3). Alterations in TP53 have already been reported to be associated with worse prognosis in patients with mCSPC (10). Therefore, patients with higher CTC counts may have a tumor genomic landscape, which indicates aggressive disease. Limitations include a single-institution retrospective cohort and a small number of patients compared with the prospective study reported by Goldkorn and colleagues (15). In addition, the significance level of ATM and MYC were marginal and did not account for other confounding factors such as tumor purity.

To summarize, we show that a high baseline CTC count (≥5/7.5 mL) is associated with a worse prognosis in patients with mCSPC regardless of ADT intensification and is associated with a distinct tumor genomic landscape. These findings may improve prognostication and patient counseling, help identifying men at the highest risk of disease progression or death at the time of diagnosis, which may affect treatment selection such as preferential use of intensified ADT or enrollment on a clinical trial, and design of future trials in this setting.

U. Swami reports personal fees from Astellas, Exelixis, Seattle Genetics; grants from Janssen, Exelixis; and grants from Astellas/Seattle Genetics outside the submitted work. B. Haaland reports personal fees from National Kidney Foundation, Proxima Clinical Research, Value Analytics Labs; and personal fees from Guidepoint Global outside the submitted work. B.L. Maughan reports personal fees from Janssen, Astellas, Merck, Exelixis, Bayer Oncology, Pfizer; grants and personal fees from Bristol Myers Squibb; grants from Clovis, Bavarian Nordic, Genentech; and personal fees from AVEO Oncology outside the submitted work. A. Goldkorn reports a patent for US 10,261,088 B2 issued. N. Agarwal reports lifetime disclosures: Consultancy to Astellas, AstraZeneca, Aveo, Bayer, Bristol Myers Squibb, Calithera, Clovis, Eisai, Eli Lilly, EMD Serono, Exelixis, Foundation Medicine, Genentech, Gilead, Janssen, Merck, MEI Pharma, Nektar, Novartis, Pfizer, Pharmacyclics, and Seattle Genetics. Research funding to Neeraj Agarwal's institution: Astellas, AstraZeneca, Bavarian Nordic, Bayer, Bristol Myers Squibb, Calithera, Celldex, Clovis, Eisai, Eli Lilly, EMD Serono, Exelixis, Genentech, Gilead, GlaxoSmithKline, Immunomedics, Janssen, Medivation, Merck, Nektar, New Link Genetics, Novartis, Pfizer, Prometheus, Rexahn, Roche, Sanofi, Seattle Genetics, Takeda, and Tracon. No disclosures were reported by the other authors.

U. Swami: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. N. Sayegh: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. Y. Jo: Data curation, writing–review and editing. B. Haaland: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–review and editing. T.R. McFarland: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. R.H. Nussenzveig: Conceptualization, resources, data curation, formal analysis, supervision, validation, methodology, writing–original draft, project administration, writing–review and editing. D. Goel: Data curation, formal analysis, methodology, writing–original draft, writing–review and editing. D. Sirohi: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. A.W. Hahn: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. B.L. Maughan: Conceptualization, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing. A. Goldkorn: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. N. Agarwal: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing.

We thank the patients whose data made this research possible.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.