Mortality Risk for Docetaxel-Treated, High-Grade Prostate Cancer With Low PSA Levels

Key Points Question Is the addition of docetaxel to standard of care (SOC) therapy with radiation and testosterone suppression or radical prostatectomy associated with decreased risk of prostate cancer–specific and all-cause deaths in patients with high-grade, nonmetastatic prostate cancer with low prostate-specific antigen levels? Findings In this meta-analysis of 4 randomized clinical trials (145 participants with prostate cancer), adding docetaxel to SOC treatment in patients who were in otherwise good health was associated with a significant reduction in death due to prostate cancer. Meaning The poor prognosis of patients with high-grade prostate cancer and low levels of prostate-specific antigen levels may be improved with the addition of docetaxel to SOC treatment.


Introduction
2][3][4][5][6] The reason for this observation is supported by the genomic characterization of these high-grade prostate cancers with low levels of PSA (<4 ng/mL; to convert to μg/L, multiply by 1.0) revealing genomic signatures that are associated with both resistance to testosterone suppression and a higher propensity to harbor micrometastatic disease at presentation. 7Therefore, for these patients, an improved treatment paradigm is needed.][10][11] However, 7 prospective randomized clinical trials (RCTs) [12][13][14][15][16][17][18] that examined the role of adding docetaxel to National Cancer Center Network-defined standard of care (SOC) treatments 19 have been completed and reported.The totality of evidence for use of radiotherapy and androgen deprivation therapy (ADT) with testosterone suppression alone or radical prostatectomy (RP) in patients with high-risk localized or locally advanced nonmetastatic prostate cancer suggests that treatment intensification with docetaxel does not uniformly improve treatment outcomes in these disease states.
In 1 of the 7 RCTs, 18 a hypothesis-generating postrandomization analysis of OS stratified by PSA level was performed and yielded a hazard ratio (HR) of 0.33 among men with high-grade prostate cancer with low PSA levels when docetaxel was added to testosterone suppression for 6 months and primary radiotherapy.Furthermore, in a prior study of men with intermediate or high-risk nonmetastatic prostate cancer, 20 only men with no or minimal comorbidity appeared to benefit from the addition of ADT to radiotherapy with respect to OS; those with moderate to severe comorbidity did not.Therefore, using individual patient data from the available RCTs from the Intermediate Clinical Endpoints in Cancer of the Prostate (ICECaP) Consortium database, we evaluated whether adding docetaxel to testosterone suppression plus curative local treatment (radiotherapy or RP) can reduce prostate cancer-specific mortality (PCSM) and all-cause mortality (ACM) in patients with localized or locally advanced nonmetastatic prostate cancer who have both a PSA level of less than 4 ng/mL and a biopsy Gleason score of 8 to 10 and are in otherwise good health with performance status defined as 0.

Methods
This meta-analysis was not registered but is within the ICECaP collaboration with an institutional review board-approved protocol and a charter to regulate all secondary analyses; the statistical analysis plan (outlined below) was prespecified and followed for the current study.We used the checklist specifications for reporting of meta-analyses following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline.

Patient Population and Treatment
Five RCTs that included 2597 patients treated with testosterone suppression plus radiotherapy or RP (SOC treatment) with or without docetaxel in the ICECaP database 13,14,[16][17][18] were eligible for study inclusion; the data in these studies were updated as reflected in the most recent date of publication.Two RCTs 12,15 could have been helpful to answer the study question, but their data were not available.
Treatment consisted of either RP alone or radiotherapy and ADT for 6 months to 2 years in the SOC group.For patients treated with radiotherapy, 3-dimensional conformal radiotherapy was primarily used, and inclusion of the pelvic lymph nodes in the radiation volume was generally performed for patients with Gleason scores of 8 to 10.In the docetaxel group of the RCT, 17 where RP was the SOC treatment, patients also received 18 to 24 weeks of ADT.Patients with localized or locally advanced nonmetastatic prostate cancer who had a serum PSA level of less than 4 ng/mL and a biopsy Gleason score of 8 to 10 were further identified, totaling 150.The 5 eligible patients of the JAMA Network Open | Oncology 413 enrolled in the Urogenital Tumor Study Group (GETUG)-12 RCT 16 were excluded due to the use of estramustine with docetaxel, given that estramustine is no longer used in the treatment of prostate cancer after being shown to significantly increase the risk of thromboembolic events 21 and therefore increase ACM.This led to a final study cohort of 145 patients as shown in Figure 1.

Clinical Outcome Measures
All-cause mortality and PCSM were coprimary end points.All-cause mortality was measured from the date of randomization to death due to any cause, censored at the date of last follow-up in surviving patients.Prostate cancer-specific mortality was defined analogously to ACM, but death not due to prostate cancer was considered a competing risk.Cause of death was based on trial-defined events.

Statistical Analysis
Data were analyzed on December 16, 2022.The distributions of the patient, cancer, and clinical factors including self-reported race as per institutional guidelines were enumerated according to randomized treatment group (SOC with or without docetaxel).For the purpose of illustration, the unadjusted distributions of ACM and PCSM were calculated by randomized treatment group using the 1 − Kaplan-Meier method 22 and cumulative incidence function accounting for competing risks, 23 respectively.These distributions were compared across randomized treatment groups using a log-rank 24 P value for the end point of ACM and Gray test 23 P value for the end point of PCSM.
To evaluate the treatment effect on the coprimary end points, an HR (with its 95% CI) was estimated for ACM using multivariable Cox proportional hazards regression, 25 and a subdistribution HR (sHR) (with its 95% CI) was estimated for PCSM using multivariable Fine and Gray 26 competing risk regression, with adjustment for performance status (1 vs 0 [reference]), biopsy Gleason score (9 or 10 vs 8), clinical T category (T3-T4 vs T1-T2 or TX), and duration of ADT (2 years [long-term] vs 4 to 6 months [short-term]).Duration of ADT was added as a covariate to the model given that longterm ADT has been shown to be superior to short-term ADT in the management of high-grade prostate cancer with respect to the end point of OS. 27 Performance status of 0 (or otherwise good health) was defined as Eastern Cooperative Oncology Group, Zubrod, or World Health Organization  performance status scores of 0. Performance status data were not available for patients enrolled in CALGB (Cancer and Leukemia Group B) 90203, 17 and they were assigned a performance status of 0, given that they needed to be considered medically able to undergo an RP.A subgroup analysis was performed in the 139 patients with performance status of 0 to investigate whether the treatment effect of adding docetaxel to SOC treatment on the risk of PCSM and ACM was consistent with the results in the overall cohort.Given that only 6 patients had a performance status of 1, an interaction test between performance status and the docetaxel treatment effect was not possible.
A test for heterogeneity was not performed given that 1 clinical trial 14 contributed only 9 patients to the study cohort.However, the study cohort was homogenous, with all patients having a PSA of less than 4 ng/mL and biopsy Gleason score of 8 to 10, and the randomized treatment groups were similar, consisting of SOC plus docetaxel vs SOC.These factors combined should minimize heterogeneity of the treatment effect on the end points of ACM and PCSM.All statistical tests were 2-sided, and the Bonferroni method 28 was used for multiplicity adjustment of the 2 comparisons performed (ie, the entire study cohort and patients with performance status of 0) when calculating the unadjusted estimates of PCSM and ACM such that P < .025(.05/2) was considered statistically significant.Median follow-up and associated IQR were calculated using the reverse Kaplan-Meier method. 29We used Stata/SE, version 16.1 (StataCorp LLC) for all statistical analyses.

Patient Population, Treatment, PCSM and ACM Event Rates
As shown in Figure 1

Unadjusted Estimates of ACM and PCSM
To illustrate the results of the models for patients with performance status of 0, Figure 2 depicts the unadjusted estimates of ACM and PCSM for patients randomized to SOC plus docetaxel vs SOC.After adjustment for multiple comparisons (ie, all 145 patients and the 139 patients with performance status of 0; 2-sided P < .025was considered statistically significant), the comparisons of the estimates between randomized treatment groups was not significant for ACM (P = .03)and was significant for PCSM (P = .007).Specifically, among the 139 patients with performance status of 0, the unadjusted 8-year ACM estimate (Figure 2A) was 30% (95% CI, 18%-48%) for SOC vs 14% (95% CI, 6%-28%) for SOC plus docetaxel; the unadjusted 8-year PCSM estimate (Figure 2B) was 23% (95% CI, 11%-38%) for SOC vs 8% (95% CI, 2%-18%) for SOC plus docetaxel.

Discussion
In this meta-analysis, we found that patients with localized or locally advanced nonmetastatic, highgrade prostate cancer with low PSA levels (<4 ng/mL) had a nonsignificant reduction in the risk of both PCSM and ACM when docetaxel was added to the SOC therapies of testosterone suppression plus radiotherapy or RP.This observation was also valid but was more pronounced among patients with performance status of 0, for whom the reduction in PCSM risk was significant.The clinical importance of these findings is that in this patient cohort with progression to a castrate-resistant state early in the disease course, 7 both PCSM and ACM risk may be reduced with the addition of docetaxel to testosterone suppression plus radiotherapy or RP.
Several points require further discussion.First, the performance status was associated with both PCSM and ACM.While the latter can be explained by competing risks in patients with comorbidity, the former can be explained by comorbidity in patients with performance status of 1, which can lead to a decreased ability to tolerate and complete the prescribed course of PC treatment.
Moreover, while both PCSM and ACM were reduced among all men in the study cohort, these associations were more pronounced among men in otherwise good health (performance status of 0), highlighting the importance of performance status on outcomes following docetaxel use.Further study is warranted to assess the effect of docetaxel use on PCSM and ACM among patients with performance status of 1, given that only 6 men in the current study had a performance status of 1.In addition, with only 5 eligible patients in the GETUG-12 RCT, 16 a sensitivity analysis to assess whether these patients benefited from the addition of docetaxel despite concurrent estramustine use was not possible.Second, while the study cohort included only 6.6% of all patients enrolled in the 4 eligible RCTs, death due to prostate cancer in this patient subgroup was the leading cause of death, accounting for 71.0% of all observed deaths; this finding illustrates the importance of identifying

SOC plus docetaxel
Includes the 139 patients with performance status of 0, comparing patients in the standard of care (SOC) group with those in the SOC plus docetaxel group.For ACM, log-rank P = .03.

JAMA Network Open | Oncology
Mortality Risk for Docetaxel-Treated, High-Grade Prostate Cancer With Low PSA Levels more effective therapy for reducing PCSM that can in turn reduce the risk of ACM.Specifically, by adding docetaxel to the SOC treatment, the absolute PCSM rate decreased more than 2-fold (from 15 to 7) among all patients and nearly 3-fold (from 14 to 5) in patients with a performance status of 0.
Third, while the 95% CI included 1.00, numerical reduction occurred in PCSM and ACM among patients who received long-term (ie, 2 years) compared with short-term (ie, Յ6 months) ADT with testosterone suppression therapy.This observation is consistent with findings of a prior RCT 27 where long-term and short-term ADT were compared among patients with locally advanced prostate cancer.Given that in the RP RCT, 17 men in the docetaxel group also received a short course (18-24 weeks) of ADT, they would have been included in the 4-to 6-month ADT cohort; given the docetaxel use, this could have reduced PSCM and ACM, thereby decreasing power to assess a reduction in PCSM and ACM risk in patients receiving long-term compared with short-term ADT.Therefore, we believe it is prudent to consider long-term ADT as part of the testosterone suppression treatment plan when using radiotherapy as the local treatment in patients with Gleason scores of 8 to 10 and PSA levels of less than 4 ng/mL.
Finally, a meta-analysis of 2 phase 3 RCTs 30 has shown that among patients treated with primary radiotherapy, adjuvant hormonal therapy using testosterone plus adrenal androgen suppression with abiraterone acetate has been associated with both a metastasis-free survival (MFS) and OS benefit in very high-risk prostate cancer when compared with testosterone suppression alone.Some patients enrolled in that study are included in the current study; namely those with Gleason scores of 8 to 10 and T3 to T4 prostate cancer.Therefore, it would be of interest to know whether the patient subgroup defined by a PSA level of less than 4 ng/mL, Gleason score of 8 to 10, and T category of T3 to T4 benefited from the addition of abiraterone to testosterone suppression therapy and radiotherapy.If true, then 2 options would exist to improve outcomes in this patient population.Further study would then be indicated in this patient cohort to ascertain whether triple therapy with testosterone suppression, abiraterone, and docetaxel is superior with respect to OS to the doublet of testosterone suppression and abiraterone.2][33] Multiple investigations studying the treatment effect on OS of adding these agents to testosterone suppression treatment with radiotherapy or RP in men with high-risk prostate cancer are actively recruiting or are in follow-up. 34,35The treatment effect of these more potent hormonal therapies in patients with high-risk prostate cancer and a PSA level of less than 4 ng/mL and Gleason score of 8 to 10 will be of major interest.If an OS benefit is observed with addition of the more potent hormonal therapy in this patient subgroup, then multiple options for treatment escalation will exist.Further study would be needed to determine the optimal specific combinations and sequencing of these agents for patients with a PSA level of less than 4 ng/mL and Gleason score of 8 to 10 and what effect docetaxel has on OS when added to these agents.It is also possible that a luminal proliferating gene expression profile or other genomic classifiers may be able identify which patients are most likely to benefit from docetaxel. 36,37

Limitations
Limitations of this study exist.First, 94 of the 145 patients (64.8%) had local treatment with RP, and these patients could represent a distinct subset, for example with less underlying comorbidity, compared with those who underwent radiotherapy.Second, 38 patients (26.2%) were in trials where 2 years of ADT was used.While we adjusted for ADT duration in the model and docetaxel use was associated with a 70% reduction in PCSM risk among patients with performance status of 0, the question remains as to what that result would have been if all patients received 2 years of ADT.Third, with 31 deaths representing only 21.4% of the study cohort, the data are still maturing.In addition, it is unlikely that an RCT addressing this question will be performed in this patient group with a poor prognosis.As a result, the current data are all that are available to address this question.Therefore, as the data mature, more precise estimates of the possible benefit of adding docetaxel to SOC in patients with a performance status of 0 will be needed before routine adoption into clinical practice.

JAMA Network Open | Oncology
Metastasis-free survival has been shown to be a surrogate for overall survival. 38However, MFS was not a primary end point in the RCTs evaluated in the current study. 13,14,17,18As a result, the MFS results may be subject to ascertainment bias and not be reliable, given that guidelines on when to look for metastasis and by what means were not prespecified.Fourth, given only 6 patients with a performance status of 1, an interaction test between performance status and the docetaxel treatment effect was not possible.Therefore, the reductions in PCSM and ACM risk may be similar among patients with performance status of 0 and 1.Finally, the inability to report a measure of heterogeneity because 1 RCT 14 only provided 9 patients to the study cohort raises the possibility of overstating the precision of results.This possibility exists because 1 of the 4 RCTs included in the meta-analysis was the source trial 18 that generated the hypothesis of potential benefit when adding docetaxel to SOC treatment.

Conclusions
In this meta-analysis of RCTs, we found that adding docetaxel to SOC treatment in patients with prostate cancer who are in otherwise good health with a PSA level of less than 4 ng/mL and a Gleason score of 8 to 10 was associated with a significant reduction in PCSM and therefore has the potential to improve prognosis.In the future, specific attention should be given to this patient subgroup with poor prognosis when reporting on studies that evaluate the effect of more potent hormonal therapies on OS.

Figure 2 .
Figure 2. Unadjusted Cumulative Incidence of All-Cause Mortality (ACM) and Cumulative Incidence Function of Prostate Cancer-Specific Mortality (PCSM) After Accounting for Competing Risk

Table 1 .
During , of the 2184 men enrolled in the 4 eligible RCTs, 145 (6.6%; median age, 63 [IQR, 46-67] years) met the selection criteria of PSA of less than 4 ng/mL, biopsy Gleason score of 8 to 10, and no estramustine use and initiated treatment between February 21, 2006, and December 31, 2015.A total of 5 patients (3.4%) were Asian, 6 (4.1%) were Black, 118 (81.4%) were White, 2 (1.4%) were of other race or ethnicity (including American Indian or Alaska Native and Hispanic or Latino), and 14 (9.7%) were of unknown race or ethnicity.A total of 67 patients were in the SOC group and 78 in the SOC plus docetaxel group; 139 (95.9%) had performance status of 0 and 6 (4.1%) had performance status of 1; 94 (64.8%) underwent RP and 38 (26.2%) had 2 years of ADT.The patient, cancer, and clinical factors were similar between the 2 randomized treatment groups as shown in a median follow-up of 7.1 (IQR, 5.4-9.9)years(6.4 [IQR, 4.7-8.2]yearsfor RP RCT; 10.1 [IQR, 8.6-11.5]yearsfor radiotherapy plus ADT RCTs) among the 145 patients in the study cohort, 31 (21.4%;19 in the SOC group and 12 in the SOC plus docetaxel group) died; of these deaths, 22 (71.0%;15 in the SOC group and 7 in the SOC plus docetaxel group) were due to prostate cancer.Among the 139 patients with performance status of 0, 27 died (19.4%; 17 in the SOC group and 10 in the SOC plus docetaxel group); of these deaths, 19 (70.4%; 14 in the SOC group and 5 in the SOC plus docetaxel group) were due to prostate cancer.

Table 1 .
Distribution of the Patient, Cancer, and Clinical Factors Stratified by Treatment Group for the 145 Patients in the Study Cohort a Unless otherwise indicated, data are expressed as No. (%) of patients.b Includes American Indian or Alaska Native and Hispanic or Latino.

Table 2 .
Covariate-Adjusted HRs for Mortality in 145 Study Patients With Performance Status of 0 or 1 a Includes 31 events among the 145 patients.bIncludes 22 events among the 145 patients.

Table 3 .
Covariate-Adjusted HRs of Mortality in the 139 Patients With Performance Status of 0 a Includes 27 events among the 139 patients.bIncludes 19 events among the 139 patients.