Effects of Androgen Deprivation Therapy on Prostate Cancer Outcomes According to Competing Event Risk: Secondary Analysis of a Phase 3 Randomised Trial

Abstract

Background: Previous studies indicate that the benefit of short-term androgen deprivation therapy (ADT) with radiotherapy (RT) for prostate cancer depends on competing risks.

Objective: To determine whether a quantitative method to stratify patients by risk for competing events (omega score) could identify subgroups that selectively benefit from ADT.

Design, setting, and participants: An ancillary analysis of NRG/RTOG 9408 phase 3 trial (NCT00002597) involving 1945 prostate cancer patients was conducted.

Intervention: Short-term ADT.

Outcome measurements and statistical analysis: We applied generalised competing event regression models incorporating age, performance status, comorbidity, T category, Gleason score (GS), and prostate-specific antigen (PSA), to stratify patients according to relative hazards for primary cancer-related events (distant metastasis or prostate cancer death) versus competing noncancer mortality. We tested interactions between ADT and subgroups defined by standard risk criteria versus relative risk (RR) using the omega score.

Results and limitations: T2b, higher GS, and higher PSA were associated with an increased RR for cancer-related versus competing mortality events (a higher omega score); increased age and comorbidity were associated with a decreased omega score. Of 996 patients with low-risk/favourable intermediate-risk (FIR) disease, 286 (28.7%) had a high omega score (≥0.314). Of 768 patients with unfavourable intermediate-risk disease, 175 (22.8%) had a low omega score. The overall discordance in risk classification was 26.1%. Both standard criteria and omega score identified significant interactions for the effect of ADT on cancer-related events and late mortality in low- versus high-risk subgroups. Within the low-risk/FIR subgroup, a higher omega score identified patients in whom ADT significantly reduced cancer events and improved event-free survival. Limitations are the need for external/prospective validation and lower RT doses than contemporary standards.

Conclusions: Stratification based on competing event risk is useful for identifying prostate cancer patients who selectively benefit from ADT.

Patient summary: We analysed the effectiveness of androgen deprivation therapy (ADT) for localised prostate cancer among patients, defined by the relative risk (RR) for cancer versus noncancer events. Among patients with traditional low-risk/favourable intermediate-risk disease, those with a higher RR benefitted from short-term ADT.

Keywords: Generalised competing event model; Hormone therapy; Prostate cancer; Risk stratification; Short-term androgen deprivation therapy.

Fig. 1 –

Comparison of baseline cumulative incidences of cancer-related events (metastasis or cancer death) versus competing mortality for patient groups defined by omega score in the RT-alone arm: (A) all controls, (B) low risk or favourable intermediate risk, and (C) unfavourable intermediate or high risk. Solid lines represent cancer-related events and dashed lines represent competing mortality. Blue indicates low omega score (<0.314) and red indicates high omega score (≥0.314). RT = radiotherapy.

Fig. 2 –

Effects of short-term androgen deprivation therapy (ADT) on cancer-related events versus competing mortality within risk groups defined by standard criteria versus omega score. (A) Concordant: low risk/FIR, low omega score. (B) Discordant: low risk/FIR, high omega score. (C) Discordant: UIR/high risk, low omega score. (D) Concordant: UIR/high risk, high omega score. Solid lines represent cancer-related event (metastasis or cancer death) and dashed lines represent competing mortality event. FIR = favourable intermediate risk; RT = radiotherapy; UIR = unfavourable intermediate risk.

Fig. 3 –

Effects of short-term androgen deprivation therapy (ADT) on overall survival within risk groups defined by standard criteria versus omega score. (A) Concordant: low risk/FIR, low omega score. (B) Discordant: low risk/FIR, high omega score. (C) Discordant: UIR/high risk, low omega score. (D) Concordant: UIR/high risk, high omega score. FIR = favourable intermediate risk; RT = radiotherapy; UIR = unfavourable intermediate risk.