If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Most patients with nmCRPC have PSADT >12 months and a long natural history
However, for those with shorter PSADT, the risk of metastasis and death, as well as healthcare costs, increased
These data can help select patients for novel hormonal therapy and conversely those who can safely delay such treatments for nmCRPC
In patients with nonmetastatic castration-resistant prostate cancer (nmCRPC), prostate-specific antigen doubling time (PSADT) is associated with risk of metastasis and survival. This study evaluated the association of PSADT with clinical and economic outcomes in a real-world setting among patients with nmCRPC not receiving novel hormonal therapy (NHT), using 2-month PSADT thresholds.
Patients and Methods
We retrospectively identified Veterans Health Administration patients with nonmetastatic prostate cancer and ≥2 PSA increases after medical/surgical castration (2012–2016). The third measurement was the index (CRPC) date. Patients with ≥3 post-index PSA measurements, including index, were followed until death or ≥12 months until disenrollment, study end, or death and grouped into 2-month cohorts based on post-index PSADT. Cox regression models assessed association between PSADT, time to metastasis, and death. Healthcare resource utilization and costs were evaluated.
Among 2800 evaluable patients, median follow-up was 30 months and median PSADT was 17 months. Relative to the reference cohort (PSADT >12 months), all cohorts had significantly higher metastasis risk. PSADT ≤10-month cohorts had significantly greater mortality risk than the reference; hazard ratios (95% confidence intervals) ranged from 12.3 (9.2, 16.4) in the PSADT ≤2-month cohort to 1.3 (0.9, 2.0) in the >10–≤12-month cohort. Total costs were significantly higher for cohorts up to and including the PSADT >8–≤10-month cohort, than for the reference cohort. Mean per patient per month all-cause medical plus pharmacy costs were $6623, $4768, and $4049 in the PSADT ≤2-month, >2–≤4-month cohort, and >4–≤6-month cohorts, respectively, vs. $1911 in the PSADT >12-month cohort (P <0.05).
Most patients with nmCRPC have PSADT >12 months and a long natural history. For those with shorter PSADT, the risk of metastasis, death, and costs increased. These data can help select patients for NHT and conversely those who can safely delay NHT for nmCRPC.
In nonmetastatic castration-resistant prostate cancer (nmCRPC), the assessment of shorter prostate-specific antigen doubling time (PSADT) intervals than those previously studied may help inform clinical risk and economic outcomes. This analysis of 2800 patients in the Veterans Health Administration found that relative to those with PSADT >12 months, cohorts with PSADT ≤2, >2–≤4, >4–≤6, >6–≤8, and >8–≤10 months had significantly higher metastasis and mortality risk and healthcare costs. These results can help select patients for novel hormonal therapy and conversely those who can safely delay such treatments for nmCRPC.
Absent metastases (i.e., rising PSA only), it is classified as nonmetastatic CRPC (nmCRPC). Patients with nmCRPC are at high risk of developing metastatic CRPC (mCRPC), for which real-world data have shown that median overall survival (OS) is < 3 years with best available pharmacotherapy.
In nmCRPC, higher PSA velocity or shorter PSA doubling time (PSADT) in the control arms of clinical trials and in real-world practice is associated with greater risk of metastasis, all-cause mortality (ACM), and PC-specific mortality (PCSM).
In clinical trials, ADT plus NHTs enzalutamide, apalutamide, and darolutamide significantly prolonged metastasis-free survival (MFS) by approximately 22–24 months and OS by approximately 10–14 months, and increased time to PSA progression vs. placebo (ADT alone).
Given the availability of NHT but also considering clinical burden and costs associated with prolonged treatment, it is critical to identify and treat the patients most likely to progress to metastatic disease, while identifying patients who may be best served by conservative strategies to balance risks and costs. Recognizing the predictive power of PSADT, stratification into shorter time intervals represents one way to better resolve clinical risk and economic outcomes in nmCRPC. Several studies have examined the impact of post-treatment PSADT on outcomes such as metastasis risk and ACM.
A retrospective database analysis of the Veterans Health Administration (VHA) including 441 patients at 5 hospitals used PSADT thresholds of < 3, 3–8.9, 9–14.9, and ≥ 15 months to stratify risk of metastasis, OS, and PCSM.
Another VHA database analysis used PSADT thresholds of unknown, < 3, 3–9, 9–15, and ≥ 15 months to evaluate MFS and OS, and separately assessed the cost associated with nmCRPC and mCRPC diagnosis; however, cost was not assessed by PSADT cohorts.
The objective of our study was to investigate the predictive ability of PSADT by evaluating the association of 2-month PSADT thresholds with time to metastasis (TTM), OS, and all-cause and PC-related HCRU and costs in patients with nmCRPC in a real-world setting. Our study aimed to add to the literature by assessing cost by PSADT cohorts. We hypothesized an inverse relationship between PSADT and outcomes and that a difference would be observed between the 2-month PSADT cohorts and the reference cohort of PSADT > 12 months. The study covers a large patient population treated at VHA clinics across the US and is the first to assess HCRU and costs as outcomes associated with PSADT differences.
Patients and Methods
Study Design and Patient Eligibility
This was a retrospective analysis of claims data from the VHA from January 1, 2012 through December 31, 2016, with 12 months of pre- and post-index data availability. All PC-related diagnoses and procedures were identified by medical claims with the pertinent codes, and death status and date were obtained from the Vital Status File. In accordance with guidelines in 45 CFR 164.514(a) and (b) (Code of Federal Regulations Title 45, Public Welfare), Institutional Review Board approval was not required.
Eligible patients were males aged ≥ 18 years with ≥ 1 claim with a PC diagnosis code and evidence of surgical or medical castration (Figure 1). CRPC was defined as ≥ 2 PSA increases from the first measurement (PSA1) after castration (PSA1 < PSA2 < PSA3). Each PSA measurement was ≥ 14 days apart. PSA3 had to be ≥ 2 ng/mL and 25% higher than PSA1. For surgically castrated patients, PSA1 was measured ≥ 2 days post-surgery; for medically castrated patients, PSA1 was measured ≥ 14 days post-initiation of ADT and while the patients were still using luteinizing hormone-releasing hormone (LHRH) agonists. PSA3 was defined as the index date. Only patients with ≥ 3 post-index PSA measurements (including PSA3) qualified for analysis. Those with metastasis prior to index (PSA3) were excluded from the final population.
Sociodemographic and clinical characteristics were evaluated on the index date (PSA3), while individual comorbidities (hypertension, stroke, angina pectoris, arrhythmia, congestive heart failure, hyperlipidemia, type II diabetes, liver abnormality/damage, and myocardial infarction) and the modified comorbidity index by Quan
All PSA values on and after the index date but before development of metastasis were used to calculate PSADT. This calculation required ≥ 2 PSA values over ≥ 3 months. Post-index PSA values were used, as many PSA values were expected to be of undetectable concentration prior to the index date and increase exponentially over time thereafter.
PSADT was measured in months and patients were assigned to cohorts based upon post-index 2-month cutoffs (≤ 2, > 2–≤ 4, > 4–≤ 6, > 6–≤ 8, > 8–≤ 10, > 10–≤ 12 and > 12).
TTM was defined as the time from index to metastasis based on the presence of secondary malignant neoplasm diagnosis codes (ICD-9-CM: 197-199.1) or a prescription claim for enzalutamide (approved for mCRPC but not nmCRPC during the study period), abiraterone acetate, or docetaxel following the index date. Time to ACM was defined as the period between the index date and death. Outcome events were captured from diagnostic and prescription codes in approximately equal proportions.
During the 12-month post-index period, per patient per month (PPPM) all-cause and PC-related HCRU and direct costs were evaluated. This included outpatient visits, inpatient stays, and pharmacy utilization, and would have included any costs for treating complications related to local progression. PC-related HCRU and costs were identified by medical claims (outpatient and inpatient) with primary or secondary diagnoses of PC, and abiraterone, enzalutamide, LHRH agonists, or bicalutamide pharmacy claims. All costs were adjusted for inflation to 2019 US dollars using the medical care component of the Consumer Price Index.
Continuous variables were presented as means (standard deviation [SD]) or medians (interquartile range [IQR]). Categorical variables were presented as counts (n) and percentages (%). Kaplan–Meier analysis, log-rank tests, and Cox proportional hazards models were used to compare TTM and time to all-cause death between the reference (PSADT > 12 months) and the remaining cohorts. Age, race, and individual comorbidities were included as covariates in the Cox models. Adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) supported the strength and significance of these findings, respectively.
HCRU and costs were compared between the reference and the other cohorts. P-values for comparison of HCRU and costs between cohorts were obtained from the t-test for continuous variables and chi-square test for categorical variables. All statistical analyses were performed using SAS (v9.4; SAS Institute Inc., Cary, NC).
A total of 3579 patients ≥ 18 years of age with PC and ≥ 2 PSA increases after castration, and continuous VHA enrollment for ≥ 12 months pre- and post-index date were identified (Figure 1). A total of 2800 eligible patients had ≥ 3 PSA measurements after the index period. Mean age was 72.6 years and 59% were White. The most prevalent comorbidities were hypertension (73%), hyperlipidemia (60%), and type II diabetes (33%). Median Quan-Charlson comorbidity index (CCI) score was 3 (Table 1). Median PSA concentration (index plus all post-index PSA) was 3.3 ng/mL (IQR, 12.8; Q1, 0.75; Q3, 13.51). All values after index (third PSA value) but before observed metastasis were included. The average number of PSA measurements for the study population was 7.4. During a median follow-up of 30 months (range, 3.7–48), median PSADT was 17 months. Patients were assigned to the following cohorts based on post-index PSADT: ≤ 2 months (n = 94, 3%), > 2–≤ 4 months (n = 274, 10%), > 4–≤ 6 months (n = 264, 9%), > 6–≤ 8 months (n = 186, 7%), > 8–≤ 10 months (n = 172, 6%), > 10–≤ 12 months (n = 140, 5%), and > 12 months (n = 1670, 60%). During the study follow-up period, 672 patients developed metastases and 1400 patients died.
Table 1Baseline demographic and clinical characteristics
Association of Post-index PSADT and Clinical Outcomes
Cohorts with PSADT ≤ 12 months experienced significantly shorter TTM relative to the reference cohort (Figure 2a). In multivariable analysis, all cohorts had a significantly higher risk of metastasis than the reference cohort in an inverse relationship; i.e., the shorter the PSADT, the higher the risk of metastasis (Figure 2b). Compared with PSADT > 12 months, the adjusted HRs (95% CIs) for TTM ranged from 33.77 (25.93, 43.96) in the ≤ 2-month cohort to 1.82 (1.32, 2.53) in the > 10–≤ 12 month-cohort (Figure 2b).
Cohorts with PSADT ≤ 12 months had significantly shorter time to all-cause death relative to the reference cohort (Figure 3a). All cohorts with PSADT ≤ 10 months had significantly higher mortality risk than the reference cohort (Figure 3b); adjusted HRs (95% CIs) ranged from 12.27 (9.20, 16.35) in the PSADT ≤ 2-month cohort to 1.34 (0.90, 2.00) in the > 10–≤ 12-month cohort (Figure 3b).
Post-Index PSADT and HCRU
Throughout the 12-month follow-up, shorter PSADT was associated with higher all-cause and PC-related HCRU in the first year after nmCRPC diagnosis (Figure 4). For example, the PSADT ≤ 2-month cohort experienced significantly greater all-cause inpatient (0.08 vs. 0.02), outpatient (3.03 vs. 1.90), and pharmacy visits (2.95 vs. 2.24) PPPM than the reference cohort. Significantly greater HCRU was reported for the PSADT > 2–≤ 4 and > 4–≤ 6-month cohorts vs. the reference cohort. Similar relationships were observed for PC-related HCRU among all cohorts up to and including the PSADT > 8–≤ 10-month cohort.
PSADT and 12-Month Follow-up Healthcare Costs
Throughout the 12-month follow-up, shorter PSADT was associated with greater all-cause and PC-related costs (Figure 5). PPPM costs were significantly higher for cohorts up to and including the PSADT > 8–≤ 10-month cohort than for the reference cohort, except for inpatient costs in the > 6–≤ 8 and > 8–≤ 10 cohorts. As PSADT shortened, more costs were related to PC. For example, average PPPM all-cause medical plus pharmacy costs were $6623, $4768, and $4049 in the PSADT ≤ 2-month, > 2–≤ 4-month cohort, and > 4–≤ 6-month cohorts, respectively, vs. $1911 in the PSADT > 12-month cohort (P < 0.05 for all comparisons).
This real-world retrospective study assessed the association of PSADT simultaneously with clinical and economic outcomes in patients with nmCRPC not receiving NHT. In the single-payer VHA, using a stratification of PSADT by 2-month cohorts, an inverse relationship was observed between PSADT and outcomes. Shorter PSADT was associated with higher risk of metastasis and death as well as higher all-cause and PC-related HCRU and costs. Based on the stratification of PSADT by 2-month cohorts, there was an inverse relationship between PSADT and clinical and economic outcomes. Of note, 32.5% of patients in this analysis were Black, providing applicability of the study findings to this patient population with a disproportionately high prostate cancer incidence and mortality.
For patients with nmCRPC, the increased risk of metastasis for those with PSADT ≤ 12 months and risk of death for those with PSADT ≤ 10 months versus the PSADT > 12-month reference cohort are consistent with results from previous clinical trials and real-world studies demonstrating that the shorter the PSADT, the greater the predisposition to metastasis or death.
In the placebo (treatment with ADT only) arms of nmCRPC trials, the median PSADT was < 6 months, and our study shows increased risk of metastasis and death among those with a PSADT under 6 months. Importantly, by examining the outcomes associated with multiple, discrete PSADT cohorts, we demonstrated that a rolling, real-time estimation of PSADT might be adopted in clinical practice upon nmCRPC diagnosis. This may assist healthcare providers to elect NHT for appropriate patients with nmCRPC or to determine eligibility for clinical trials.
PSADT ≤ 6 months was associated with higher HCRU, and PSADT ≤ 10 months was associated with higher healthcare expenditure. A simple explanation for these associations is that the shorter PSADT in patients with nmCRPC substantially increases metastasis risk, resulting in increased HCRU and costs.
it follows that delaying disease progression by appropriately allocating treatment might mitigate or delay some of these costs through savings that increase with longer PSADT. However, while clinical trials show NHTs increase survival (by 10–14 months) and do not adversely affect quality of life in nmCRPC patients,
any cost savings could be offset by increased therapy costs. That being said, the addition of NHT to ADT in the nmCRPC setting has been found to be within accepted ranges of cost-effectiveness in recent analyses by the US Institute for Clinical and Economic Review and the UK National Institute for Health and Care Excellence.
Our study (N = 2800) collected inpatient and outpatient information from VHA providers nationwide during the 5-year interval of 2012–2016 and assessed both clinical and economic outcomes according to 2-month PSADT cohort. We found that for most patients, PSADT was > 12 months. Our use of 2-month PSADT thresholds allowed for a granular analysis of these relationships. Unlike our analysis, the smaller Howard et al. study (2000–2015; N = 441) explored the association between PSADT and survival but not economic outcomes
looked at economic outcomes attributable to mCRPC, not by PSADT cohort. Both previous analyses used 6-month PSADT thresholds from < 3 months to ≥ 15 months. Despite methodological differences, shorter PSADT was consistently associated with worse outcomes, reinforcing the importance of monitoring PSADT in nmCRPC.
Limitations inherent to claims-based studies include reliance on diagnostic coding, use of pharmacy claims as a surrogate for actual medication administration, and the possibility of treatment outside the studied healthcare system. Requiring ≥ 3 post-index PSA measurements may exclude patients with a very short PSADT, because they would likely develop metastases by that time and therefore be ineligible. The use of bicalutamide was not captured and therefore its impact on PSADT was not accounted for; however, a less accurate PSADT should bias the results toward the null and underestimate associations with outcomes. Restricting the identification of metastasis to secondary malignant neoplasm diagnostic codes and enzalutamide, abiraterone acetate, or docetaxel claims following the index date would have overlooked a small number of patients who received sipuleucel-T or radium-223 dichloride for mCRPC. While type of metastatic imaging was not captured, given the time period of the study (2012–2016), we anticipate nearly all imaging would have been conventional bone scans or computerized tomography (CT) scans. Additionally, the population could include patients with suspected metastatic disease who were started on treatment prior to being officially diagnosed with mCRPC. Gleason scores were not available in the dataset and could not be accounted for, although this may not be an important predictor when PSA is consistently monitored after nmCRPC diagnosis.
HCRU and cost analyses were not adjusted for differences in baseline characteristics. Additionally, we evaluated all-cause deaths, not specifically PC deaths, though our study population included only patients with PC. Finally, this study does not examine the details behind the increased all-cause and PC-related costs; we hope future studies building upon our work can identify the specific drivers of the increased costs we identified.
The findings of this study align with existing literature to confirm that PSADT is a strong predictor of metastasis and death and is moreover highly associated with HCRU and costs in patients with nmCRPC (see the graphical abstract summarizing the analysis in the online publication). Importantly, the majority of patients had long PSADT (> 12 months) and were at low risk of progression or significant HCRU. For those with shorter PSADT, however, the clinical and economic sequalae of a short PSADT in nmCRPC provides rationale to treat with approved therapies, although cost-effectiveness and potential adverse effects of therapy must be considered.
The current information can help to better determine high-risk patients potentially eligible for NHT and conversely low-risk patients best served with conservative strategies to balance risks and costs. It can also help identify patients for clinical trials. Future studies in different patient populations and other healthcare settings are needed to generalize these results.
Video: A video summary of this article is available here: [please see document entitled “PSADT nmCRPC VHA video summary for journal review”]
SJF, KR, RS, NMS, and DJG contributed to study conception and design, and AH, JM, and NJ were responsible for data acquisition and statistical analysis. All authors contributed to the interpretation of data. All authors contributed to the drafting of the manuscript and critical revision of scientific content and approved the final manuscript.
Data sharing statement
Upon request, and subject to review, Pfizer will provide the data that support the findings of this study. Subject to certain criteria, conditions and exceptions, Pfizer may also provide access to the related individual de-identified participant data. See https://www.pfizer.com/science/clinical-trials/trial-data-and-results for more information.
Clinical Practice Points
PSADT is known to be predictive of outcomes in nmCRPC. This study evaluated the association of shorter (2-month) PSADT intervals with both clinical and economic outcomes. Among 2800 evaluable patients in the VHA, the median PSADT was 17 months. Relative to the reference cohort (PSADT > 12 months), all cohorts had significantly higher metastasis risk. PSADT ≤ 10-month cohorts had significantly greater mortality risk than the reference; HRs (95% CIs) range from 12.3 (9.2, 16.4) in the PSADT ≤ 2-month cohort to 1.3 (0.9, 2.0) in the > 10–≤ 12-month cohort. Total healthcare costs were also higher among those cohorts with PSADT up to 10 months.
The majority of patients had long PSADT (> 12 months) and were at low risk of progression or significant healthcare resource utilization and costs. For those with shorter PSADT, however, the clinical and economic sequalae of a short PSADT in nmCRPC provides rationale to treat with approved therapies, although cost-effectiveness and potential adverse effects of therapy must be considered. By examining the outcomes associated with multiple, discrete PSADT cohorts, we demonstrated that a rolling, real-time estimation of PSADT might be adopted in clinical practice upon nmCRPC diagnosis. This may assist healthcare providers to choose NHT for appropriate patients with nmCRPC and, conversely, to identify low-risk patients best served with conservative strategies to balance risks and costs.
This study was sponsored by Pfizer (New York, NY, USA) and Astellas Pharma (Northbrook, IL, USA), the co-developers of enzalutamide. STATinMED received compensation from the sponsors for the overall conduct of the study and development of this manuscript. The sponsors were involved in the study design, collection, analysis, and interpretation of data, and review of the manuscript; employees of the sponsor companies are coauthors. Ultimate responsibility for opinions, conclusions, and data interpretation lies with the authors.
CRediT authorship contribution statement
Stephen J. Freedland: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision. Krishnan Ramaswamy: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Ahong Huang: Methodology, Software, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization. Rickard Sandin: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Jack Mardekian: Methodology, Software, Formal analysis, Investigation, Data curation, Writing – original draft, Writing – review & editing, Visualization. Neil M. Schultz: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization, Funding acquisition. Nora Janjan: Methodology, Software, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization. Daniel J. George: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Visualization.
Conflicts of Interest
SJF reports consulting and honoraria to Astellas Pharma, AstraZeneca, Bayer, Exact Sciences, Janssen Biotech, Merck, Myovant Sciences, Pfizer, and Sanofi. KR is an employee of Pfizer and owns stocks or stock options. AH was employed by STATinMED during the time of the study; STATinMED received compensation from the sponsors for the overall conduct of the study and development of this manuscript. RS is an employee of Pfizer and owns stocks or stock options. JM was employed by Pfizer during the time of the study and owns stock or stock options. NMS is an employee of Astellas Pharma and owns stocks or stock options. NJ was employed by STATinMED during the time of the study; STATinMED received compensation from the sponsors for the overall conduct of the study and development of this manuscript. DJG reports grants from Acerta Pharmaceuticals, other from American Association for Cancer Research, grants and personal fees from Astellas, personal fees from AstraZeneca, personal fees from Axess Oncology, grants, personal fees and non-financial support from Bayer H/C Pharmaceuticals, grants and personal fees from BMS, grants from Calithera, grants from Capio Biosciences, grants from EMD Serona, grants, personal fees, and non-financial support from Exelixis, Inc., personal fees from Flatiron, personal fees from Ipsen, grants and personal fees from Janssen Pharmaceuticals, personal fees from Merck, Sharp & Dohme, personal fees from Michael J Hennessey Assoc, personal fees from Millennium Medical Publishing, personal fees from Modra Pharmaceuticals B.V., personal fees from Myovant Sciences, Inc., personal fees from NCI Genitourinary, personal fees from Nektar Therapeutics, grants and personal fees from Novartis, personal fees from Physician Education Resource, grants and personal fees from Pfizer, grants, personal fees and non-financial support from Sanofi, personal fees from UroGPO, personal fees and non-financial support from UroToday, personal fees from Vizuri Health Sciences, personal fees from Platform Q, outside the submitted work.
We thank Stanislav Lechpammer, MD, PhD, of Pfizer and John Tonrey, Keshia Maughn, and Virginia Noxon of STATinMED for their contributions to the analysis and manuscript. STATinMED received compensation from the sponsors for the conduct of the study and development of this manuscript. Medical writing and editorial support funded by the sponsors was provided by Malcolm Darkes, Stephanie Vadasz, and Dena McWain of Ashfield Healthcare Communication and Kirstie Anderson and Matthieu Larochelle of Onyx (a Prime Global agency).