The median age was 72 years (range 20–91). The median prostate-specific antigen (PSA) level at diagnosis was 10 ng/mL (range 2.3–324 ng/mL), and the median Gleason score (GS) was 7 (range 6–10). Table 1 presents the detailed baseline characteristics of the patients.

Neoadjuvant ADT was used in 48 patients (39.3%), with a median duration of 4 months (range 1–60 months). Eight patients received neoadjuvant ADT for more than 6 months, and these patients were referred to our clinic after ADT had been initiated. A total of 98 patients (80.3%) used concurrent adjuvant ADT, with a median duration of 9 months (3–30 months).

In follow-ups, the median PSA nadir was 0.01 ng/mL (range 0.001–0.795 ng/mL). The median follow-up was 24.7 months (0.8–94.4 months), and the median PSA at the last follow-up was 0.09 ng/mL (0–3.23 ng/mL). At last follow-up, 109 patients were alive without disease (89.3%), and 8 patients had recurrence (6.6%). Five patients (4%) died due to reasons other than the disease. Two of the 8 patients who developed recurrence experienced distant metastasis, while others had locoregional recurrence. Patients had a 2- and 5‑year progression-free survival (PFS) of 95% and 82%, respectively.

Acute GU toxicity was observed in 81 patients (66.4%). Sixty-five (53.2%) patients experienced grade 1–2 toxicity, and 16 (13.1%) patients experienced grade 3–4 toxicity. The most common GU side effects were dysuria in 67 patients (54.9%), nocturia in 32 (26.2%) patients, and hematuria in 4 patients (3.2%). Acute GI toxicity was observed in 52 patients (42.6%). Grade 1–2 toxicity was observed in 39 (31.9%) patients, and grade 3–4 toxicity was observed in 13 (10.6%). The most common GI side effects were proctitis in 43 patients (35.2%), hematochezia in 7 patients (5.7%), and diarrhea in 4 patients (3.2%).

Late GU toxicity was observed in 33 patients (27%). Twenty-nine patients (23.7%) had grade 1–2 toxicity, and 4 patients (3.2%) experienced grade 3–4 toxicity. The most common GU late side effects were dysuria in 18 patients (14.7%), incontinence in 10 patients (8.1%), nocturia in 8 patients (6.5%), hematuria in 5 patients (4%), and urinary stricture in 2 patients (1.6%). Of the 10 patients who developed incontinence, 4 experienced persistent incontinence at the 24-month mark, whereas the other 6 patients’ incontinence was resolved (see Supplementary Fig. 1). Late GI toxicity was observed in 16 patients (13.1%). Fourteen patients (11.4%) experienced grade 1–2 toxicity, while 2 patients (1.6%) experienced grade 3–4 toxicity. The most common late GI side effects were hematochezia in 9 (7%) patients, chronic proctitis in 6 patients (4.9%), and chronic diarrhea in 1 patient (0.8%). Of the patients who had hematochezia, 2 had grade 3–4 hematochezia. We referred these patients to argon laser treatment, and at 12 months, one patient’s hematochezia had resolved (see Supplementary Fig. 1).

The median (range; min–max) values of the dose–volume constraints for the rectum and bladder are presented in Supplementary Table 2.

For acute GU toxicity, dose received by 15 cc of the bladder volume (D15cc) was a significant predictor for grade 3–4 toxicity in MVA (p = 0.04). Furthermore, dose received by 10% and 1% of the bladder volume (D10% and D1%) showed a trend towards significance and were evaluated as clinically relevant predictors for grade 1–2 and grade 3–4 GU toxicity, respectively (p = 0.07 and p = 0.05; Table 2). We further entered these predictors into ROC analysis to determine the optimal threshold (see Supplementary Fig. 2). The optimum cut-off for D15cc was 26.3 Gy, with 62% sensitivity and 58% specificity (p = 0.04, area under curve [AUC]: 0.657, 95% confidence interval [CI]: 0.541–0.774). The optimum cut-off for D10% was 19.08 Gy, with 61% sensitivity and specificity (p = 0.02, AUC: 0.623, 95% CI: 0.523–0.0723) and the optimal cut-off for D1% was 35.4 Gy, with 56% sensitivity and 56% specificity (p = 0.07, AUC: 0.640, 95% CI: 0.518–0.763).

For acute GI toxicity, none of the dosimetric parameters were predictive for grad 1–2 toxicity in UVA and MVA (Table 3). For acute grade 3–4 GI toxicity, on the other hand, rectum volume receiving a dose of ≥18 Gy  (V18Gy) was a significant predictor, and rectum V18.12Gy was also clinically relevant (p = 0.04 and p = 0.05, respectively). The optimum threshold for V18Gy was 23.02%, with 61% sensitivity and 57% specificity (p = 0.1, AUC: 0.636, 95% CI: 0.449–0.822) and for V18.12Gy, the optimum threshold was 22.8%, with 61% sensitivity and 57% specificity (p = 0.1, AUC: 0.634, 95% CI: 0.447–0.821).

For late GU toxicity, the only predictor for grade 1–2 toxicity that was marginally significant was bladder D1% (p = 0.06) and for grade 3–4 toxicity, bladder D15cc (p = 0.09) in UVA. Therefore, we were unable to conduct MVA (Table 4). The optimal cut-off for D1% was 35.3 Gy, with 51% sensitivity and 51% specificity (p = 0.6, AUC: 0.525, 95% CI: 0.393–0.657) and for D15 cc, the optimal cut-off was 28.9 Gy, with 75% sensitivity and 77% specificity (p = 0.1, AUC: 0.738, 95% CI: 0.464–1.000).

For late-grade 1–2 GI toxicity, rectum V28Gy was the only predictor that approached statistical significance (p = 0.06). None of the rectum dosimetric parameters were significant for late grade 3–4 GI toxicity (Table 5). This could be attributed to a lack of occurrences. The optimal cut-off for V28Gy was found to be 9.6%, with 57% sensitivity and 85% specificity (p = 0.2, AUC: 0.592, 95% CI: 0.376–0.808).

The median prostate volume was 51.89 cc (range 19.21–124.76), and the median PTV volume was 89.97 cc (range 44.61–185.69). No relationship was found between prostate volume and overall acute and late GU toxicity (p = 0.1 and p = 0.8, respectively). When evaluated separately for acute grade 1–2 and 3–4 toxicity, the volume was shown to be statistically significant only for the latter group (p = 0.6 and p = 0.003, respectively). The prostate volume threshold that most accurately predicted acute grade 3–4 GU toxicity was 61.7 cc (p = 0.003, AUC: 0.729, 95% CI: 0.609–0.850), with a sensitivity of 68% and a specificity of 67%. There was no correlation between late grade 1–2 and grade 3–4 GU toxicities (p = 0.8 and p = 0.3, respectively).

There was also no correlation between prostate volume and overall acute and late GI toxicity (p = 0.7 and p = 0.7, respectively). Similarly, prostate volume did not correlate with acute grade 1–2 or grade 3–4 toxicity (p = 0.4 and p = 0.2, respectively) or with late grade 1–2 or grade 3–4 toxicity (p = 0.8 and p = 0.2, respectively).