The left half of Table 1 shows the baseline characteristics of all patients (n = 860) before PSM. There were significant differences between the RARP and VMAT groups for all variables that were assessed. The median follow-up duration was 79 months (interquartile range [IQR]: 62–98 months) and those in the RARP and VMAT groups were 76 months (IQR: 63–89 months) and 87 months (IQR: 59–108 months), respectively (Student’s t-test, P < 0.001). All VMAT patients who underwent ADT received neoadjuvant ADT, which was continued if indicated: 88 of 360 (24.4%) VMAT patients underwent long-term (≥ 2 years) ADT. Detailed descriptions of neoadjuvant and/or adjuvant treatments in the RARP group are provided in our previous article [12].

Supplementary Fig. 1 shows the Kaplan–Meier curves of the RARP vs. VMAT patients for OS, CSS, rRFS, and bRFS in the original cohort before PSM (n = 860). In the RARP group, 111 patients developed biochemical recurrence, 18 developed radiological recurrence, 2 died of PC, and 11 died from other causes. In the VMAT group, 37 patients developed biochemical recurrence, 15 developed radiological recurrence, 3 died of PC, and 21 died from other causes. Accordingly, RARP patients had significantly longer OS compared with VMAT patients (log-rank test, P = 0.027; Supplementary Fig. 1A), whereas there were no significant differences between the groups for CSS (P = 0.54; Supplementary Fig. 1B) and rRFS (P = 0.14; Supplementary Fig. 1C). VMAT patients had significantly longer bRFS compared with RARP patients, albeit based on different definitions of biochemical recurrence (P < 0.001; Supplementary Fig. 1D). In addition, there was no difference in bRFS between VMAT with < 2-year ADT vs. VMAT with ≥ 2-year ADT (Supplementary Fig. 2A). For reference, 6‑year OS, CSS, rRFS, and bRFS rates in the RARP group were 97.6%, 99.6%, 95.3%, and 76.8%, while those in the VMAT group were 94.9%, 99.3%, 92.3%, and 89.5%, respectively.

Univariate and multivariate Cox proportional hazard regression analyses in all patients (n = 860) before PSM were performed for OS, rRFS, and bRFS, but not for CSS owing to the small number of events (n = 5) (Supplementary Table 1). In the multivariate analyses, D’Amico risk classification (low vs. intermediate vs. high) and age-adjusted CCI (continuous) were identified as independent prognostic factors for both OS (Supplementary Table 1A) and rRFS (Supplementary Table 1B), whereas treatment modality (RARP vs. VMAT) and concomitant ADT (yes vs. no) were not. In contrast, treatment modality was identified as an independent prognostic factor for bRFS, along with D’Amico risk classification and concomitant ADT (Supplementary Table 1C).

Table 1 shows the baseline characteristics of patients before and after PSM. For PSM, all seven variables listed in Table 1 were matched: age, initial PSA, biopsy Gleason score, clinical T stage, D’Amico risk classification, age-adjusted CCI, and concomitant ADT. The right half of Table 1 shows the characteristics of 260 patients after PSM. Baseline characteristics between the two groups (RARP vs. VMAT) were balanced after PSM, and differences between all variables were nonsignificant. Eight of 130 (6.2%) VMAT patients underwent long-term (≥ 2 years) ADT.

Figure 1 shows the Kaplan–Meier curves of the RARP vs. VMAT patients for OS, CSS, rRFS, and bRFS in the matched cohort (n = 260). After PSM, 30 patients developed biochemical recurrence, nine developed radiological recurrence, two died of PC, and two died from other causes in the RARP group, whereas 17 patients developed biochemical recurrence, four developed radiological recurrence, one died of PC, and six died from other causes in the VMAT group. Accordingly, RARP and VMAT patients had equivalent outcomes for OS (log-rank test, P = 0.65; Fig. 1a), CSS (P = 0.57; Fig. 1b), and rRFS (P = 0.40; Fig. 1c), whereas VMAT patients had significantly longer bRFS compared with RARP patients based on different definitions of biochemical recurrence (log-rank test, P = 0.003; Fig. 1d). In addition, there was no difference in bRFS between VMAT with < 2-year ADT vs. VMAT with ≥ 2-year ADT (Supplementary Fig. 2B). For reference, 6‑year OS, CSS, rRFS, and bRFS rates in the RARP group were 96.4%, 98.3%, 91.6%, and 75.4%, while those in the VMAT group were 96.5%, 99.1%, 94.7%, and 90.4%, respectively.

Kaplan–Meier curves of the RARP vs. VMAT patients for a OS, b CSS, c rRFS, and d bRFS in the matched cohort (n = 260). (bRFS biochemical recurrence-free survival, CSS cancer-specific survival, OS overall survival, RARP robot-assisted radical prostatectomy, rRFS radiological recurrence-free survival, VMAT volumetric modulated arc therapy)

Univariate and multivariate Cox proportional hazard regression analyses after PSM (n = 260) were performed for OS, rRFS, and bRFS, but not for CSS owing to the small number of events (n = 3) (Supplementary Table 2). No variable, including treatment modality (RARP vs. VMAT), was an independent prognostic factor on multivariate analysis for both OS (Supplementary Table 2A) and rRFS (Supplementary Table 2B), whereas treatment modality and concomitant ADT were independent prognostic factors for bRFS (Supplementary Table 2C).

In addition to oncological outcomes, we assessed complication outcomes of RARP and VMAT. Supplementary Table 3 shows the Clavien–Dindo grade ≥ 3 perioperative complications in the RARP group (n = 500). Thirteen of 500 (2.6%) patients experienced grade ≥ 3 complications, such as postoperative hemorrhage (n = 3), rectal injury (n = 2), small bowel injury (n = 2), and abdominal wall hernia (n = 2). One patient developed nonocclusive mesenteric ischemia and died 35 days after surgery (Clavien–Dindo grade 5). Figure 2a, b show the cumulative proportions of continence recovery in the RARP group (n = 500); 458 (91.6%) and 344 (68.8%) patients achieved ≤ 1 pad/day and pad-free, respectively, for urinary continence. In contrast, Fig. 2c, d illustrate the cumulative proportions of grade ≥ 2 GU and GI complications in the VMAT group (n = 360); 72 (20.0%) and 16 (4.4%) patients experienced grade ≥ 2 GU and GI complications, respectively. Furthermore, 8 (2.2%) patients developed grade ≥ 2 rectal bleeding (Supplementary Fig. 3). Regarding grade ≥ 3 complications (ten cases in total), one patient had diverticular bleeding (grade 3 GI), whereas five had vesical bleeding and four had urinary retention (all grade 3 GU).

Cumulative proportions of continence recovery in the RARP group (n = 500) and those of grade ≥ 2 GU and GI complications in the VMAT group (n = 360). a Proportion of patients achieving ≤ 1 pad/day for urinary continence and b that of patients achieving pad-free status for urinary continence in the RARP group (n = 500). c Proportion of patients with grade ≥ 2 GU complications and d that of patients with grade ≥ 2 GI complications in the VMAT group (n = 360). (GI gastrointestinal, GU genitourinary, RARP robot-assisted radical prostatectomy, VMAT volumetric modulated arc therapy)