PSA response and survival outcome
Patient demographics of full dose induction subgroups are summarized in Supplementary Table 1. Abiraterone acetate and enzalutamide was introduced as first-line ARSI for 58 and 162 patients, respectively. The median follow-up duration was 24.0 months (range: 1–85 months).
Survival outcome (PSA-PFS, TTF, CSS and OS) stratified by the level of PSA decline are shown in Supplementary Fig. 1(a-d). Levels of PSA decline correlated survival outcome significantly. (Median PSA-PFS : PSA decline < 0%: 1m (95%CI: 0–2), 0–90%: 6m (95%CI: 4–8), ≥ 90%: 11m (95%CI: 8–15), p < 0.0001. Median TFF: PSA decline < 0%: 2m (95%CI: 1–4), 0–50%: 9m (95%CI: 6–11), ≥ 90%: 25m (95%CI: 20–30), p < 0.0001, Median CSS : PSA decline < 0%: 43m (95%CI: 21–51), 0–90%: 68m (95%CI: 48-NR), ≥ 90%: not reached (NR) (95%CI: NR-NR), p < 0.0001. Median OS : PSA decline < 0%: 22m (95%CI: 15–47), 0–50%: 53m (95%CI: 45-NR), ≥ 90%: NR (95%CI: NR-NR), p < 0.0001) PSA response rate (PSA-RR) significantly correlated with PSA-PFS (p < 0.0001), TFF (p < 0.0001), CSS (p < 0.0001) and OS (p < 0.0001). (Fig. 1(a-d)) A better PSA response correlated with a better survival outcome.
Patient demographics of abiraterone acetate versus enzalutamide
Patient demographics (abiraterone acetate versus enzalutamide) are shown in Table 1. There was no significant difference in the baseline parameters at the time of diagnosis of prostate cancer and at the time of treatment initiation between the two groups, except for the visceral metastasis rate (19.0% [A] vs. 5.6% [E]; p < 0.05). Regarding sequential treatment, 2nd line treatment induction rate was 82.2% [A] and 74.6% [E] (p = 0.408) after primary ARSI failure. There were no significant differences in induction rate of 2nd line ARAT and docetaxel between the groups. And in terms of other 2nd line life prolonging agent, 2nd line 223Ra was introduced one case in abiraterone acetate group, and 2nd line 223Ra, bicalutamide, flutamide, estramustine phosphate and ethinylestradiol were introduced 2, 1, 1, 1 and 1 case in enzalutamide group, respectively.
PSA responses of first-line abiraterone acetate versus enzalutamide
In terms of PSA response rate, the median PSA responses rates were − 65.4% [A] and − 81.5% [E] (p = 0.0252) (Supplementary Fig. 2a). A waterfall plot of the best PSA response in this cohort is shown in Supplementary Fig. 2b. PSA decline of ≥ 90% were 22.4% [A] and 37.0% [E] (p = 0.0478). PSA responses were not available in 4 cases each, in both groups.
Survival outcome
We investigated PSA-PFS, TTF, CSS and OS in this study. The Kaplan–Meier curves of PSA-PFS, TTF, CSS and OS with respect to abiraterone acetate versus enzalutamide are shown in Fig. 1(a-d). The median PSA-PFS (4 months [A] and 7 months [E]; p = 0.00833), median TFF (6 months [A] and 15 months [E]; p < 0.0001), median CSS (45 months [A] and not reached (NR) [E]; p < 0.0001), median OS (34 months [A] and 80 months [E]; p < 0.001) were better in the enzalutamide group.
Analyses of survival outcome using IPTW method
IPTW method was used to minimize patients’ demographics. After IPTW adjustment, all absolute value of standardized mean difference (SMD) in weighted comparisons were < 0.10 except for LDH (SMD = 0.13(CSS) and − 0.12 (OS)) which indicated that distribution of baseline factors was similar between abiraterone acetate and enzalutamide group. (Supplementary Table. 2a, b) IPTW- adjusted Kaplan-Meier curves show that median CSS and OS was significantly longer in enzalutamide group (41 months and 80 months [E], p < 0.001 (CSS), and 34 months [A] and 80 months [E], p = 0.011 (OS)). (Figure. 2a, b)
Univariate and multivariate analyses for OS and CSS
Univariate and multivariate analyses for CSS and OS are shown in Table 2(a,b). In univariate analysis for CSS, age, PSA, Hb, ALP, ARSI (p = 0.0461, < 0.0001, 0.017, < 0.001 and < 0.001, respectively) were significant. In the multivariate analysis for CSS, ALP (p < 0.01) and ARSI (enzalutamide vs. abiraterone acetate, p < 0.01) were significant factors. (Table 2a) In univariate analysis for OS, age, PS, metastasis, PSA, Hb, ALP, LDH and ARSI (p = 0.012, 0.0024, 0.023, < 0.001, < 0.0001, 0.0045, < 0.001 and 0.00126, respectively) were significant factors. And in the multivariate analysis for OS, Hb (p < 0.01) and ARSI (enzalutamide vs. abiraterone acetate, p = 0.0155) were significant factors. (Table 2b)
2nd line treatment and survival outcome
Correlation between 2nd line treatment and survival outcome were analyzed. In ARSI-ARSI seqent cohort, overall survival was not different between the groups (median OS was 51 months [A] and 59 months [E] (p = 0.33). (Supplementary Fig. 3a) In ARSI-docetaxelcohort, although it did not reach significant difference (p = 0.0541), OS trended better in enzalutamide group (median OS was 27 months [A] and 63 months [E] (p = 0.0541). (Supplementary Fig. 3b) And in the cohort without subsequent therapy after failure of primary ARSI, although it did not reach significant difference (p = 0.0997), OS of enzalutamide group might be better (median OS was 29.5 months [A] and 61 months [E]. (Supplementary Fig. 3c)
Survival outcome of nmCRPC and mCRPC
OS of nmCRPC and mCRPC are shown in Supplementary Fig. 4 (a, b) Median OS of non-metastatic disease was 68 months (abirtaterone) and 80 months (enzalutamide) (p = 0.0733), median OS of metastatic disease was 28 months (abiraterone) and 59 months (enzalutamide) (p = 0.00577).
Subgroup analysis
Forrest plot of HR for PSA-PFS, TFF, CSS and OS are shown in Fig. 3 (a-d). HR of each subgroups are shifted to enzalutamide better in all survival outcome.