rPFS and cPFS May Act as Surrogate OS End Points in Phase 3 mHSPC Clinical Trials

Susan Halabi, PhD

Both radiographic progression-free survival (rPFS) and clinical PFS (cPFS) appear to be valid surrogates for overall survival (OS) in phase 3 clinical trials evaluating testosterone suppression–based treatment strategies for patients with metastatic hormone-sensitive prostate cancer (mHSPC), according to findings from a study that were published in the Journal of Clinical Oncology.¹

This trial pooled individual patient data (IPD) from 6390 patients who were randomly assigned to phase 2 or phase 3 prostate cancer clinical trials between June 1994 and July 2013. The median OS among these patients was 4.3 years (95% CI, 4.2-4.5). In comparison, the median rPFS and cPFS were 2.4 years (95% CI, 2.3-2.5) and 2.3 years (95% CI, 2.2-2.4), respectively. The respective surrogate threshold effects (STEs) for rPFS and cPFS were 0.80 and 0.81, indicating that respective risk reductions of 20% and 19% would have a nonzero effect on OS.

“In this surrogacy analysis, we observed a strong correlation between the intermediate clinical end points [ICEs] and OS,” lead study author Susan Halabi, PhD, and coauthors, wrote in the paper. Halabi is the James B. Duke Distinguished Professor, professor of biostatistics and bioinformatics, chief of the Division of Biostatistics, and member of the Duke Cancer Institute in Durham, North Carolina.

The current standard treatment paradigm for patients with mHSPC is informed by findings from trials such as the phase 3 CHAARTED (NCT00309985) and phase 2/3 STAMPEDE (NCT00268476) trials. In CHAARTED, patients who received concurrent androgen deprivation therapy (ADT) plus docetaxel achieved an OS benefit vs ADT alone, with an HR for OS of 0.72 (95% CI, 0.59-0.89; P < .0018).² Similarly, in STAMPEDE, patients who received ADT plus abiraterone acetate (Zytiga) and prednisolone achieved an OS benefit compared with those who received ADT alone, with an HR of 0.63 (95% CI, 0.52-0.76; P < .001).³

“Although OS remains the gold standard end point in phase 3 mHSPC trials, given the OS outcomes with current therapy of ADT plus newer hormonal therapies, it will take close to a decade to complete new mHSPC studies with OS as the primary end point,” the authors emphasized in the paper.¹ They noted that although OS is a clearly defined end point that is simple to measure and translates into clinical benefit, its disadvantages include its requirement of large clinical trial sizes and extended follow-up periods, which lead to long mHSPC trial durations. However, ICEs in well-powered comparative trials can help investigators determine the efficacy of certain therapies. Compared with OS, ICEs can be readily available earlier in a cancer’s natural history, are less costly, and are measured more frequently.

“We hypothesized that rPFS and cPFS are valid surrogates for OS in patients with mHSPC and could potentially be used to expedite phase 3 clinical trials,” the authors explained.

To evaluate the validity of rPFS and cPFS as surrogates for OS in mHSPC clinical trials, investigators gathered IPD from 9 eligible randomized clinical trials comprising 13 arms comparing treatment regimens in patients with mHSPC. Treatment regimens in both the control and investigative arms of these trials included different ADT strategies or ADT plus docetaxel. Trials with primary end points of safety, feasibility, and quality of life were excluded.

In these trials, the established end point of OS was defined as the time from random assignment to death from any cause. Patients who had not died were censored at the date of the last follow-up. rPFS was defined as the time from random assignment to radiographic progression (defined per protocol) or death from any cause, whichever occurred first. cPFS was defined as the time from random assignment to the date of radiographic progression, initiation of a new treatment, symptoms, or death, whichever occurred first.

Investigators used a current, standard, state-of-the-art 2-stage meta-analytic validation model in which conditions of trial-level and patient-level surrogacy needed to be met. In the first stage, investigators tested for individual-level association between OS and the ICEs and calculated the trial-specific treatment effects of the end points. In the second stage, investigators used weighted linear models (WLRs) to test for correlations between treatment effects on both OS and the ICEs. The clinically relevant surrogacy of R2 value was defined as over 0.75, a standard oncology surrogacy assessment. In the secondary analyses, trials were excluded if they had insufficient follow-up, defined as a median follow-up of less than 4.3 years. From there, the STE was calculated.

The validity of rPFS and cPFS as surrogate OS end points was also evaluated in patient subgroups delineated by high-volume disease, treatment received, and de novo metastatic diagnosis.

Patients had a median age of 67 years (interquartile range, 61.0-73.1) at random assignment, 69.7% had an ECOG performance status of 0, and 76.5% had de novo disease at diagnosis. Furthermore, 35.9% of patients had high-volume disease, and 36.1% had unknown disease volume status. In total, 41.1% of patients received ADT plus docetaxel, and 70.2% received ADT alone. Most patients (53.3%) had a Gleason score of 8 to 10.

A total of 71% of patients had died of prostate cancer, and the median follow-up in the 2529 surviving patients was 6.1 years (range, 0.0-17.7). Overall, 4501 rPFS events occurred, 55% and 45% of which were radiographic progression and death, respectively. Additionally, 4574 cPFS events occurred.

Overall Surrogacy Analysis: Condition 1

Results from condition 1 of the overall surrogacy analysis showed that at the patient level, Kendall’s tau was 0.83 (95% CI, 0.82-0.84) for OS and rPFS and 0.85 (95% CI, 0.85-0.86) for OS and cPFS. From WLR, the R2 between the 3-year rPFS and 5-year OS rates was 0.62 (95% CI, 0.29-0.89). The R2 between the 3-year cPFS and 5-year OS rates was 0.74 (95% CI, 0.49-0.90). When investigators limited the analysis to patients with sufficient follow-up, the R2 between the 3-year rPFS and 5-year OS rates was 0.74 (95% CI, 0.40-0.96).

Overall Surrogacy Analysis: Condition 2

Results from condition 2 of the overall surrogacy analysis showed that at the study level, the R2 for log(HR)-OS vs log(HR)-rPFS was 0.83 (95% CI, 0.64-0.98), and the R2 for log(HR)-OS vs log(HR)-cPFS was 0.84 (95% CI, 0.61-0.99). Furthermore, the median R2 from the LOOCV was 0.78 for both rPFS and cPFS, and the HRs fell within the 95% prediction intervals in 12 of the 13 trial arms, indicating robust models.

Subgroup Analysis: Docetaxel Trials

The Kendall’s tau between OS and rPFS were 0.82 (95% CI, 0.81-0.83) and 0.71 (95% CI, 0.70-0.73), in the trials evaluating ADT and ADT plus docetaxel, respectively. The R2 from the WLR between the 5-year OS and 3-year rPFS rates was 0.77 (95% CI, 0.40-0.96) for the ADT alone trials and 0.49 (95% CI, 0.13-0.98) for the ADT plus docetaxel trials. Additionally, the R2 from the WLR between the 5-year OS and 3-year cPFS rates was 0.89 (95% CI, 0.73-0.98) for the ADT alone trials and 0.50 (95% CI, 0.05-0.99) for the ADT plus docetaxel trials.

For condition 2, the R2 from the WLR of log(HR)-OS vs log(HR)-rPFS were 0.86 (95% CI, 0.53-0.99) for the ADT alone trials and 0.73 (95% CI, 0-1) for the ADT plus docetaxel trials. The R2 from the WLR of log(HR)-OS vs log(HR)-cPFS was 0.84 (95% CI, 0.53-0.99) for the ADT alone trials and 0.88 (95% CI, 0-1.00) for the ADT plus docetaxel trials. The respective STEs on OS for HR(rPFS) and HR (cPFS) for patients who received ADT were 0.87 and 0.86. The STEs on OS for HR(rPFS) and HR (cPFS) for patients who received ADT plus docetaxel were not evaluable.

Subgroup Analysis: Volume of Disease

The Kendall’s tau between OS and rPFS were 0.79 (95% CI, 0.78-0.81) and 0.71 (95% CI, 0.69-0.73), for patients with high-volume and low-volume disease, respectively. The R2 from the WLR between the 5-year OS and 3-year rPFS rates was 0.77 (95% CI, 0.35-0.97) for all patients with high-volume disease and 0.85 (95% CI, 0.59-0.98) for those with sufficient follow-up. Conversely, the R2 from the WLR between the 5-year OS and 3-year rPFS rates was 0.43 (95% CI, 0.0-0.98) for patients with low-volume disease.

For condition 2, the R2 from the WLR of log(HR)-OS vs log(HR)-rPFS were 0.87 (95% CI, 0.06-1.00) and 0.85 (95% CI, 0.16-1.00) for patients with high-volume and low-volume disease, respectively. The R2 from the WLR of log(HR)-OS vs log(HR)-cPFS was 0.84 (95% CI, 0.04-0.99) for the patients with high-volume disease and 0.90 (95% CI, 0.36-1.00) for the patients with low-volume disease. The respective STEs on OS for HR(rPFS) and HR (cPFS) for patients with high-volume disease were 0.71 and 0.69. The respective STEs on OS for HR(rPFS) and HR (cPFS) for patients with low-volume disease were 0.60 and 0.68.

Subgroup Analysis: M1 Diagnosis Status

For patients with synchronous disease, the Kendall’s tau between OS and rPFS was 0.77 (95% CI, 0.76-0.78), and the Kendall’s tau between OS and cPFS was 0.78 (95% CI, 0.77-0.80). At the trial level, the R2 between the 5-year OS and 3-year rPFS rates was 0.62 (95% CI, 0.27-0.90) for patients with synchronous disease with sufficient follow-up.

For condition 2, the R2 from the WLR of log(HR)-OS vs log(HR)-rPFS was 0.82 (95% CI, 0.48-0.99) for all patients with synchronous disease and 0.89 (95% CI, 0.56-1.00) for those with sufficient follow-up. The R2 from the WLR of log(HR)-OS vs log(HR)-cPFS was 0.81 (95% CI, 0.44-0.99) for patients with synchronous disease and 0.87 (95% CI, 0.33-0.99) for those with sufficient follow-up. The STE on OS for HR (rPFS) and HR (cPFS) for patients with synchronous disease with sufficient follow-up was 0.80.

The authors noted that validation of rPFS and cPFS in trials investigating drugs with other mechanisms of action, such as androgen receptor inhibitors, is planned.

“rPFS can be justified as the primary end point and would provide potential savings in costs and study duration if it were selected as the primary end point,” the authors concluded. “Although the results were slightly higher for cPFS, our preference is to use rPFS as it is a more objective end point than cPFS.”

References

1. Halabi S, Roy A, Rydzewska L, et al. Radiographic progression-free survival and clinical progression-free survival as potential surrogates for overall survival in men with metastatic hormone-sensitive prostate cancer. J Clin Oncol. Published online January 4, 2024. doi:10.1200/JCO.23.01535
2. Kyriakopoulos CE, Chen YH, Carducci MA, et al. Chemohormonal therapy in metastatic hormone-sensitive prostate cancer: long-term survival analysis of the randomized phase III E3805 CHAARTED trial. J Clin Oncol. 2018;36(11):1080-1087. doi:10.1200/JCO.2017.75.3657
3. James ND, de Bono JS, Spears MR, et al. Abiraterone for prostate cancer not previously treated with hormone therapy. N Engl J Med. 2017;377(4):338-351. doi:10.1056/NEJMoa1702900