PSA-based end point is not a reliable surrogate for survival in prostate cancer

Biochemical recurrence (BCR) in patients with prostate cancer was shown to be prognostic for overall survival (OS) but failed to meet the criteria for surrogacy, suggesting that BCR may not be a reliable primary end point in clinical trials for localized disease.^1,2

The investigators analyzed data from 11 randomized controlled trials that evaluated treatments to reduce the risk of BCR.

The data were published in the Journal of Clinical Oncology.

For the study, the investigators analyzed data from 11 randomized controlled trials that evaluated treatments to reduce the risk of BCR, including radiotherapy dose escalation (n = 3639), short-term androgen deprivation therapy (ADT) (n = 3930), and ADT prolongation (n = 3772). The median follow-up among all patients was 9.2 years. The validity of BCR as a surrogate intermediate clinical end point was measured using the Prentice criteria with landmark analyses and estimating Kendall’s tau (with surrogacy as ≥ .7) and R2 (with surrogacy as ≥ .7).

Significant improvements in BCR were seen with dose escalation (HR, 0.71; 95% CI, 0.63 to 0.79), short-term ADT (HR, 0.53; 95% CI, 0.48 to 0.59), and prolongation of ADT (HR, 0.54; 95% CI, 0.48 to 0.61). Significant improvements in OS were seen with the addition of short-term ADT (HR, 0.91; 95% CI, 0.84 to 0.99) and the prolongation of ADT (HR, 0.86; 95% CI, 0.78 to 0.94), but not with dose escalation (HR, 0.98; 95% CI, 0.87 to 1.11).

For the landmark analyses at 48 months, BCR was associated with worse OS in all 3 treatment groups ((HR, 2.46 [95% CI, 2.08 to 2.92]; HR, 1.51 [95% CI, 1.35 to 1.70]; and HR, 2.31 [95% CI, 2.04 to 2.61], respectively). However, when an adjustment was made for BCR at landmark time points, OS was not significantly affected by treatment (HR, 1.10 [95% CI, 0.96 to 1.27]; HR, 0.96 [95% CI, 0.87 to 1.06] and HR, 1.00 [95% CI, 0.90 to 1.12], respectively).

The Kendall’s tau for BCR-free survival and OS ranged from 0.59 to 0.69 among the 3 treatment groups, and the Kendall’s tau for time to BCR and OS ranged from 0.23 to 0.41. The R2 values for the correlation of treatment effect on BCR-free survival with that on OS was 0.563, and the R2 value for the correlation of treatment effect on time to BCR with that on OS was 0.162. None of the values met the threshold to determine a strong correlation between the treatment effect on surrogate and true end points.

“One reason for our finding could be that many patients in the study died from causes unrelated to prostate cancer,” said senior author Amar U. Kishan, MD, in a news release on the findings.² Kishan is an associate professor of radiation oncology at the David Geffen School of Medicine at the University of California, Los Angeles (UCLA) and a researcher at the UCLA Health Jonsson Comprehensive Cancer Center.

He added, “The strength of the correlation between biochemical recurrence and overall survival varied depending on how deaths from non-cancer-related causes were accounted for. However, it is important to note that we looked specifically at death, and not at quality of life. Certainly, biochemical recurrence could impact quality of life. Unfortunately, high-level data tracking this are lacking, and it is something our group, and others, are hoping to explore further.”

The authors suggest that metastasis-free survival may serve as more suitable end point for future clinical trials of radiation therapy in localized prostate cancer.

References

1. Roy S, Romero T, Michalski JM, et al. Biochemical recurrence surrogacy for clinical outcomes after radiotherapy for adenocarcinoma of the prostate. J Clin Oncol. Published online August 28, 2023. Accessed August 30, 2023. doi:10.1200/JCO.23.00617

2. PSA levels after treatment may not be a reliable predictor of survival for patients with prostate cancer. News release. University of California, Los Angeles Health. August 28, 2023. Accessed August 30, 2023. https://www.uclahealth.org/news/psa-levels-after-treatment-may-not-be-reliable-predictor