br We also performed unweighted Cox regressions for pa tient
We also performed unweighted Cox regressions for pa-tient subgroups defined by Gleason grade, clinical T-stage, NCCN risk category (intermediate, high), and androgen deprivation therapy use. For the NCCN risk category sub-group, Gleason grade, clinical T-stage, and PSA were not included in the models. For each subgroup, the significance of interaction terms between the subgroup and boost modal-ity was estimated with the Wald test. An additional subgroup analysis was also run for patients with Gleason 9 or 10 dis-ease (12, 13). Multivariate hazard ratios for boost modality are included in Forest plots for each subgroup. All statistical analysis was performed using R version 3.5.0 (The R Foun-dation for Statistical Computing). This work was performed with approval of our institutional review board.
Results
The median followup for the entire cohort was 5.0 years (interquartile range: 3.0, 7.4). Table 1 summarizes baseline
Fig. 1. Formulation of study cohort. EBRT 5 External beam radiation therapy; SBRT = stereotactic body LOXO-101 therapy.
Table 1
Baseline characteristics based on treatment (HDR boost, LDR boost, and DE-EBRT)
Unadjusted
IPTW-adjusted
Total
HDR boost
LDR boost
DE-EBRT
Factor
n
AgedMean (SD)
-
Prostate specific antigendMean (SD)
-
Gleason grade
Clinical T-stage
Race
Charlson-Deyo score
Facility location
Facility type
Educational attainment
Income
Insurance
Androgen deprivation therapy
DE-EBRT 5 dose-escalated external beam radiation therapy; HDR 5 high-dose-rate; IPTW 5 inverse probability of treatment weighting; LDR 5 low-dose-rate.
Unadjusted p-values (min) and IPTW-adjusted p-values (min) are reported.
p (min) represents the minimum p-value among the 3-group comparisons (HDR vs. LDR, HDR vs. DE-EBRT, LDR vs. DE-EBRT).
clinical and demographic factors among the three groups. Small but statistically significant differences in baseline characteristics were observed for all factors. Fig. 2a shows respective Kaplan-Meier curves before IPTW adjustment. On multivariable Cox proportional hazards regression,
Table 2
Cox proportional hazards regression model for overall survival
Fig. 2. (a) Unadjusted and (b) IPTW-adjusted Kaplan-Meier curves for HDR boost, LDR boost, and DE-EBRT. IPTW 5 inverse probability treat-ment weighting; HDR 5 high-dose-rate; LDR 5 low-dose-rate; DE-EBRT 5 dose-escalated EBRT.
As shown in Fig. 3, there was no significant difference between LDR and HDR boosts for patient subgroups divided by Gleason grade, clinical T-stage, NCCN risk category, or androgen deprivation therapy. All interaction terms between analyzed subgroups and boost modality were not significant. For the Gleason 9e10 subgroup, AHRs for LDR boost and DE-EBRT were 1.03 ([0.84, 1.25]; p 5 0.81) and 1.25 ([1.09, 1.44]; p 5 0.001), respectively.
Discussion
Previous studies have suggested that LDR and HDR bra-chytherapies in the monotherapy setting have equivalent bPFS outcomes (14, 15). A large multiinstitutional retro-spective propensity-adjusted analysis also showed no dif-ference in distant metastasis or prostate cancer specific
Factor
Hazard ratio (95% CI)
p-Value
CI 5 confidence interval; DE-EBRT 5 dose-escalated external beam radiation therapy; HDR 5 high-dose-rate; IPTW 5 inverse probability of treatment weighting; LDR 5 low-dose-rate.
mortality between LDR vs. HDR brachytherapy boost (13). Our analysis using a large National Cancer Database also suggests that both LDR and HDR brachytherapy boost provide equivalent OS outcomes for unfavorable-risk pros-tate cancer. It is reassuring that similar results were ob-tained after propensity score adjustment with IPTW, which allowed for partial mitigation of treatment