Diabetes mellitus as a potential risk factor for severe late toxicity after stereotactic body radiotherapy for prostate cancer: a large single-institution cohort study

Abstract

This study aimed to evaluate the incidence of severe late gastrointestinal and genitourinary toxicities after prostate stereotactic body radiotherapy (SBRT) and to explore potential clinical risk factors, with a particular focus on diabetes mellitus. We retrospectively analyzed patients with clinically non-nodal and non-metastatic prostate cancer treated with SBRT between 2016 and 2024. A total of 940 patients were included. The prescribed doses ranged from 36.25 to 47.5 Gy in five fractions. A total of 29 grade ≥ 3 composite toxicity events (20 gastrointestinal and 14 genitourinary events) were observed. The cumulative incidence of late grade ≥ 3 composite toxicity was 2.5% at 2 years (95% CI, 1.6–3.9%) and 4.6% at 5 years (95% CI, 3.1–6.8%). In univariate analysis, diabetes mellitus was significantly associated with an increased risk of severe toxicity (HR 2.6, 95% CI 1.2–5.4, P = 0.014). Subgroup analyses demonstrated generally consistent associations across clinically relevant subgroups. Although severe late toxicities were infrequent overall, clinically meaningful and potentially devastating events were observed. Exploratory analyses suggested that patient-related clinical factors, including diabetes mellitus, may contribute to increased vulnerability to severe late radiation injury following prostate SBRT and should be considered when interpreting toxicity risk.

INTRODUCTION

Radiotherapy plays a central role in the curative treatment of localized prostate cancer. Advances in radiation delivery over the past two decades have enabled both dose escalation and improved sparing of organs at risk, facilitating the development of prostate stereotactic body radiotherapy (SBRT), which delivers high radiation doses in a limited number of fractions [1, 2]. Randomized phase III trials such as PACE-B and HYPO-RT-PC have demonstrated that ultra-hypofractionated SBRT provides biochemical control comparable to conventional fractionation with acceptable toxicity [3, 4]. In addition, systematic reviews summarizing multiple studies have reported favorable long-term disease control with low rates of severe late gastrointestinal (GI) and genitourinary (GU) toxicity using SBRT regimens mainly delivering 36.25–40 Gy in five fractions [5]. Contemporary guidelines including the National Comprehensive Cancer Network (NCCN) now endorse or permit 5-fraction SBRT across prostate cancer risk groups [6].

In conventionally fractionated radiotherapy for localized prostate cancer, dose escalation has consistently been associated with improved biochemical disease control, but at the cost of increased toxicity [7–10]. These findings have raised interest in exploring dose escalation strategies in prostate SBRT. However, unlike conventional regimens, robust evidence supporting an oncologic benefit of escalating SBRT doses—particularly when prescribed to the whole prostate gland beyond 40 Gy in five fractions—remains limited. Most SBRT dose-escalation studies evaluating 42.5–50 Gy have primarily focused on dose tolerance and late GI and GU toxicity rather than on improvements in cancer control [11–14].

Given the potential for increased rectal dose at higher prescription levels, rectal spacers have been introduced to reduce rectal irradiation in prostate radiotherapy. Previous studies in conventionally fractionated and SBRT settings have demonstrated consistent dosimetric benefits and suggested a reduction in GI toxicity with spacer use [15–18].

While early SBRT experience has demonstrated that total doses of 36.25–40 Gy to the whole gland in five fractions can be delivered safely with favorable oncologic outcomes, the toxicity profile of dose-escalated SBRT beyond this range remains less clearly defined. Prior dose-escalation studies evaluating 42.5–50 Gy in five fractions have yielded mixed results: some reports have described acceptable late GI and GU toxicity, whereas others have observed severe late events at higher dose levels [11–14]. Moreover, the determinants of severe late toxicity in SBRT are unlikely to be explained by dosimetric factors or spacer use alone. Previous studies mainly from conventionally fractionated radiotherapy have suggested that patient- and treatment-related characteristics—including prior local surgery history and comorbidities such as antithrombotic therapy or diabetes mellitus —may contribute to individual susceptibility to late radiation injury [19–22]. Diabetes mellitus, in particular, has long been proposed as a potential modifier of radiation-related injury, potentially through impaired tissue repair, microvascular dysfunction, and altered inflammatory responses [19]. However, whether these associations are applicable to prostate SBRT remains unclear. Robust clinical evidence addressing patient-related risk factors for severe late toxicity specifically in the prostate SBRT setting is still limited.

Prostate SBRT was initiated in May 2016 with a prescribed dose of 36.25 Gy in five fractions, which was subsequently increased to 40 Gy as clinical experience accumulated. Rectal spacers were introduced during this early period. A phase I dose-escalation study was later conducted to evaluate higher prescribed doses ranging from 42.5 to 47.5 Gy in five fractions [14]. In that study, the maximal tolerated dose without spacer use was 42.5 Gy. Higher prescribed doses were subsequently introduced in selected patients treated with rectal spacers based on clinical experience and safety considerations. Ultimately, 45 Gy in five fractions was adopted for selected spacer-treated patients. To date, nearly 1000 patients have been treated with SBRT at our institution.

Given our large institutional cohort and substantial experience with dose-escalated prostate SBRT, a comprehensive evaluation of severe late GI and GU toxicities is warranted. The aims of this study were to describe the incidence of grade ≥ 3 late GI and GU adverse events and to explore potential clinical risk factors associated with severe late toxicity, with particular attention to diabetes mellitus.

MATERIALS AND METHODS

Study design and patient selection

This was a single-institution retrospective study. We identified consecutive patients who initiated SBRT at our institution between May 2016 and May 2024. Patients were eligible if they received SBRT delivered in five fractions. There were no restrictions on age, initial prostate-specific antigen (iPSA) level, Gleason score, or NCCN risk classification. Clinical stage N0M0 was required based on contemporary staging evaluations. Use of androgen deprivation therapy (ADT) and placement of a rectal spacer were not used as inclusion or exclusion criteria. Patient background, disease characteristics, treatment characteristics, and toxicity outcomes were obtained from the medical records. Collected variables included age, clinical T stage, grade group, iPSA, NCCN risk classification, ADT use, rectal spacer use, and prescribed dose. Comorbidities (diabetes mellitus, hypertension, dyslipidemia, and hemodialysis), medication use (antithrombotic agents), surgical history (rectal surgery, transurethral resection of the prostate, and bladder surgery), and smoking status were also recorded. Diabetes mellitus was defined as the presence of a documented clinical diagnosis in the medical records (including referral letters or patient self-report) and/or laboratory evidence of diabetes, defined as a hemoglobin A1c level ≥ 6.5%.

Radiotherapy

All patients underwent CT-based simulation in the supine position. A planning CT scan with 1 mm slice thickness was acquired, and MRI fusion was routinely performed for target delineation. Bowel preparation (enema) and maintenance of a comfortably full bladder were instructed at simulation and for each treatment fraction. SBRT was delivered at total doses ranging from 36.25 to 47.5 Gy in five fractions. The clinical target volume (CTV) included the prostate with or without the seminal vesicles according to NCCN risk group. A planning target volume (PTV) margin of 5 mm was applied to the CTV except posteriorly, where a 3-mm margin was used. Treatment was prescribed to PTV D_95%. Treatments were delivered using X-ray–based VMAT SBRT with cone-beam CT image guidance before each fraction. Rectal spacer placement was routinely performed unless contraindicated (e.g. suspected extracapsular extension into the perirectal space or history of pelvic surgery). ADT was administered according to NCCN risk group, typically as short-term (4–6 months) or long-term (more than one year) therapy.

Follow-up and toxicity assessment

Although follow-up schedules were not strictly standardized, most patients were evaluated every 3–6 months during the first year and every 6–12 months thereafter. For the present study, late grade ≥ 3 late GI and GU toxicities were identified and individually reviewed. GI and GU toxicities were assessed and graded according to the Common Terminology Criteria for Adverse Events version 5.0. Late events were defined as toxicities occurring 3 months or later after the initiation of radiotherapy. When multiple grade ≥ 3 events occurred within the same organ system, including concurrent manifestations of a single clinical episode, the event judged to best represent the overall clinical course was selected as the representative toxicity.

Statistical analysis

For patients who experienced both grade ≥ 3 GI and GU toxicities, only the first occurring event was used for the composite GI + GU time-to-event analyses to avoid double counting. Cumulative incidences of late grade ≥ 3 (i) composite GI + GU toxicity, (ii) GI toxicity only, and (iii) GU toxicity only for the entire cohort were estimated using the Kaplan–Meier method. Cumulative incidences of late grade ≥ 3 composite GI + GU toxicity stratified by prescribed dose (≤40 Gy vs ≥42.5 Gy) and diabetes mellitus status were also estimated using the Kaplan–Meier method, and group differences were assessed using the log-rank test. Univariable Cox proportional hazards models were constructed to evaluate potential predictors of late grade ≥ 3 composite GI + GU toxicities. In addition to diabetes mellitus, patient-related factors (e.g. prior surgeries and comorbidities), disease-related factors (e.g. NCCN risk classification), and treatment characteristics (e.g. SBRT dose and spacer use) were assessed. Given the limited number of grade ≥ 3 events, fully adjusted multivariable models were not constructed to avoid model overfitting. Instead, exploratory subgroup analyses were performed using the composite GI + GU endpoint to evaluate the consistency of observed associations across other subgroups. Based on prior evidence suggesting a potential association between diabetes mellitus and radiation-related tissue injury, diabetes mellitus was selected as the primary exposure for the subgroup analyses. All statistical analyses were performed using R version 4.5.0.

RESULTS

Patients background

A total of 940 patients were included in this study. The median follow-up duration was ~34 months. The median age at the start of SBRT was 72 years. SBRT doses ranged from 36.25 Gy to 47.5 Gy in five fractions, with 205 patients (22%) receiving >40 Gy. All six patients who received 47.5 Gy were treated as part of a dose-escalation clinical trial. Rectal spacers were used in 567 patients (60%). Diabetes mellitus was present in 181 patients (19%). Smoking status was unavailable for ~30% of patients, and only eight patients (<1%) were receiving chronic dialysis. Additional baseline clinical and treatment characteristics are summarized in Table 1.

Table 1.

Baseline patient characteristics

		Total number of patients = 940
Characteristics		Number or median (% or IQR)
Age (years), median (IQR)		72 (66–77)
Age range, n (%)	<75	596 (63%)
	≥75	344 (37%)
Median follow-up (months), median (IQR)		33.9 (16.7–56.6)
Clinical T stage, n (%)	1–2a	811 (86%)
	3–4	129 (14%)
Grade group, n (%)	1–3b	631 (67%)
	4–5	309 (33%)
Initial PSA (ng/ml), median (IQR)		8.9 (6.1–14.0)
Initial PSA (ng/ml), range, n (%)	<10	552 (59%)
	10–20	258 (27%)
	≥20	130 (14%)
NCCN riskc, n (%)	Low	45 (5%)
	Intermediate	512 (54%)
	High	383 (41%)
ADT use, n (%)	No	261 (28%)
	Short	373 (40%)
	Long	306 (33%)
Spacer use, n (%)	No	373 (40%)
	Yes	567 (60%)
Total dose (Gy), n (%)	36.25	282 (30%)
	40	453 (48%)
	42.5	67 (7%)
	45	132 (14%)
	47.5	6 (1%)
Comorbidities, n (%)	Diabetes mellitus	181 (19%)
	Hypertension	414 (44%)
	Dyslipidemia	262 (28%)
	Hemodialysis	8 (1%)
Drug use, n (%)	Antithrombotic agentsd	248 (26%)
Surgery history, n (%)	Rectal surgery history	33 (4%)
	TURP history	20 (2%)
	Bladder surgery history	35 (4%)
Smoking, n (%)	No	278 (30%)
	Yes	377 (40%)
	Unknown	285 (30%)

Abbreviations: int, intermediate; ADT, androgen deprivation therapy; NCCN, National Comprehensive Cancer Network; PSA, prostate-specific antigen; TURP, transurethral resection of the prostate

^aIncluding 10 patients with clinical Tx stage.

^bIncluding two patients with unknown grade group.

^cNCCN intermediate-risk includes both favorable and unfavorable intermediate-risk disease. High-risk includes NCCN high- and very-high-risk disease.

^dAntithrombotic agents include both anticoagulant and antiplatelet agents.

Baseline characteristics differed in several respects between patients treated with ≤40 Gy and those receiving ≥42.5 Gy. In particular, follow-up duration (median 43.6 vs 20.3 months) and spacer use (53% vs 85%) differed, reflecting temporal changes in clinical practice. Detailed characteristics for each dose stratum are provided in Supplementary Table S1.

Several baseline characteristics also differed between patients with and without diabetes mellitus. Hypertension (40% vs 60%), dyslipidemia (23% vs 47%), and antithrombotic agent use (23% vs 41%) were more prevalent in patients with diabetes mellitus, likely reflecting overlapping metabolic comorbidities. Detailed characteristics are provided in Supplementary Table S1.

Toxicity

Late grade ≥ 3 GI and GU toxicities are summarized in Table 2. A total of 20 grade ≥ 3 GI events and 14 grade ≥ 3 GU events were identified. After accounting for overlap between GI and GU toxicities, 29 patients experienced at least one grade ≥ 3 late toxicity. Rectal perforation occurred in seven patients, including two grade 3 and five grade 4 events. Urinary fistula was observed in four patients (two grade 3 and two grade 4), and bladder perforation was observed in one patient; all five of these patients also experienced rectal perforation. Prostatic necrosis was identified in one patient and appeared to represent the primary pathology, subsequently leading to secondary infection. A total of eight patients developed particularly severe complications with long-term morbidity: seven patients with rectal perforation (including cases with urinary fistula or bladder perforation) and one patient with prostatic necrosis. Clinical characteristics of these patients are summarized in Supplementary Table S2.

Table 2.

Grade ≥ 3 gastrointestinal and genitourinary toxicities

Event		Grade 3, n (%)	Grade 4, n (%)	Total, n (%)
GI				20 (2.1%)
	Rectal hemorrhage	13 (1.4%)
	Rectal perforation	2 (0.2%)	5 (0.5%)
GU				14 (1.5%)
	Hematuria	5 (0.5%)
	Urinary retention	2 (0.2%)	1 (0.1%)
	Urinary fistulaa	2 (0.2%)	2 (0.2%)
	Bladder perforationa	1 (0.1%)
	Prostate infectionb		1 (0.1%)
ALL				29 (3.1%)

No grade 5 events were observed. For each patient, only the most severe grade ≥ 3 gastrointestinal (GI) or genitourinary (GU) toxicity was counted, and the event most representative of the clinical course was selected for analysis. The median time to onset of grade ≥ 3 toxicity (composite GI/GU events) was 15.9 months (interquartile range, 11.5–29.8 months).

Abbreviations: GI, gastrointestinal; GU, genitourinary

^aAll five patients with urinary fistula or bladder perforation also developed rectal perforation.

^bProstate infection was caused by prostate necrosis.

The cumulative incidences of late grade ≥ 3 (A) composite GI + GU toxicity, (B) GI toxicity, and (C) GU toxicity estimated using the Kaplan–Meier method are shown in Fig. 1. The cumulative incidence of composite GI + GU toxicity was 2.5% at 2 years (95% CI, 1.6–3.9%) and 4.6% at 5 years (95% CI, 3.1–6.8%). For GI toxicity, the cumulative incidence was 2.1% at 2 years (95% CI, 1.3–3.5%) and 2.9% at 5 years (95% CI, 1.9–4.6%). For GU toxicity, the cumulative incidence was 1.1% at 2 years (95% CI, 0.5–2.1%) and 2.4% at 5 years (95% CI, 1.3–4.3%).

Fig. 1.

Kaplan–Meier estimates of cumulative incidence of late grade ≥ 3 gastrointestinal and genitourinary toxicities in the entire cohort. (A) Composite GI + GU toxicity, (B) GI toxicity only, and (C) GU toxicity only. Numbers at risk are displayed below each panel. Abbreviations: GI, gastrointestinal; GU, genitourinary.

Patients treated with doses >40 Gy showed a significantly higher cumulative incidence of late grade ≥ 3 composite GI + GU toxicity compared with those treated with ≤40 Gy (Supplementary Fig. S1, log-rank P = 0.015). The 2-year cumulative incidence was 1.8% (95% CI, 1.0–3.2%) in the ≤40 Gy group and 5.3% (95% CI, 2.7–10.5%) in the >40 Gy group. Five-year cumulative incidence could not be reliably estimated in the >40 Gy group because of insufficient long-term follow-up. When GI and GU toxicities were analysed separately, the 2-year cumulative incidence of grade ≥ 3 GI toxicity was 1.6% (95% CI, 0.9–3.0%) in the ≤40 Gy group and 4.1% (95% CI, 1.8–9.1%) in the >40 Gy group. For GU toxicity, the corresponding 2-year cumulative incidences were 0.5% (95% CI, 0.2–1.5%) and 3.3% (95% CI, 1.4–7.9%), respectively.

Kaplan–Meier analysis demonstrated a significantly higher cumulative incidence of late grade ≥ 3 composite GI + GU toxicity in patients with diabetes mellitus compared with those without diabetes mellitus (Fig. 2). The 2-year cumulative incidence was 5.6% (95% CI, 2.8–10.9%) in patients with diabetes mellitus and 1.8% (95% CI, 1.0–3.2%) in those without diabetes mellitus. At 5 years, the cumulative incidence increased to 9.9% (95% CI, 5.3–18.1%) and 3.3% (95% CI, 2.0–5.3%), respectively. The difference was statistically significant by the log-rank test (P = 0.010).

Fig. 2.

Kaplan–Meier estimates of cumulative incidence of late grade ≥ 3 composite gastrointestinal and genitourinary toxicity stratified by diabetes mellitus status. Group differences were assessed using the log-rank test. Numbers at risk are shown below the curves. Abbreviations: DM, diabetes mellitus.

Risk factors

In univariable Cox proportional hazards analysis for late grade ≥ 3 composite GI + GU toxicity, several factors were significantly associated with increased risk (Table 3). Diabetes mellitus was associated with a higher risk of severe toxicity (HR 2.6, 95% CI 1.2–5.4, P = 0.014). Hypertension (HR 2.7, 95% CI 1.2–6.0, P = 0.013) and antithrombotic agents use (HR 2.4, 95% CI 1.2–5.0, P = 0.017) were also significantly associated with toxicity. Disease- and treatment-related factors, including NCCN high-risk classification (HR 3.1, 95% CI 1.4–6.9, P = 0.004), total prescribed dose >40 Gy (HR 2.8, 95% CI 1.2–6.7, P = 0.019), and long-term androgen deprivation therapy (HR 2.2, 95% CI 1.0–4.5, P = 0.037), were likewise significantly associated with severe toxicity. In contrast, spacer use was associated with a significantly lower risk of toxicity (HR 0.3, 95% CI 0.1–0.7, P = 0.006). Rectal surgery history showed a strong association with toxicity (HR 4.2, 95% CI 1.5–12.0, P = 0.008), although the number of events was limited.

Table 3.

Univariate Cox proportional hazards analysis for grade ≥ 3 composite gastrointestinal and genitourinary toxicities

Variable	n	N	HR (95% CI)	P
Age ≥ 75	13	344	1.4 (0.7–2.9)	0.388
Diabetes mellitus	11	181	2.6 (1.2–5.4)	0.014
Hypertension	20	414	2.7 (1.2–6.0)	0.013
Dyslipidemia	12	262	1.8 (0.9–3.8)	0.116
Antithrombotic agents	14	248	2.4 (1.2–5.0)	0.017
NCCN high risk	20	383	3.1 (1.4–6.9)	0.004
Spacer	7	567	0.3 (0.1–0.7)	0.006
Total dose > 40 Gy	8	205	2.8 (1.2–6.7)	0.019
Long term ADT	15	306	2.2 (1.0–4.5)	0.037
Rectal surgery history	4	33	4.2 (1.5–12.0)	0.008
TURP history	2	20	3.4 (0.8–14.2)	0.098
Bladder surgery history	1	35	0.8 (0.1–5.9)	0.830

Hazard ratios (HRs) were estimated using univariate Cox proportional hazards models for the composite endpoint of grade ≥ 3 gastrointestinal (GI) and genitourinary (GU) toxicities.

n/N indicates the number of events and the total number of patients in the exposed group.

Abbreviations: GI, gastrointestinal; GU, genitourinary; NCCN, National Comprehensive Cancer Network; ADT, androgen deprivation therapy; TURP, transurethral resection of the prostate; HR, hazard ratio; CI, confidence interval.

In subgroup analyses, diabetes mellitus was associated with a higher risk of late grade ≥ 3 composite GI + GU toxicity across almost all examined subgroups, with hazard ratios exceeding 1.0 (Fig. 3). Subgroup estimates were not available for patients with a history of transurethral resection of the prostate or bladder surgery because no events occurred in one of the comparison groups. A statistically significant interaction was observed for rectal surgery history (P value for interaction = 0.046); however, because of the very small number of patients and events in this subgroup, this result should be interpreted with caution. A potential interaction was also suggested for antithrombotic agents use (P value for interaction = 0.09). The hazard ratio was markedly higher in patients receiving antithrombotic agents (HR 4.1, 95% CI 1.4–12.4) compared with those not receiving antithrombotic agents (HR 0.9, 95% CI 0.2–3.8). However, given the exploratory nature of the analysis and the limited number of events, this finding should be considered hypothesis-generating.

Fig. 3.

Forest plot of subgroup analyses for grade ≥ 3 composite gastrointestinal and genitourinary toxicity with diabetes mellitus as the primary exposure. Hazard ratios (HRs) were estimated using Cox proportional hazards models. The vertical dashed line indicates HR = 1. The horizontal axis is shown on a logarithmic scale. n/N represents the number of events over the total number of patients in each subgroup. Arrowheads indicate hazard ratio estimates that extend beyond the plotting range. For subgroups with zero events, HRs and P values for interaction were not available (N/A). Abbreviations: DM, diabetes mellitus; NCCN, National Comprehensive Cancer Network; ADT, androgen deprivation therapy; TURP, transurethral resection of the prostate. CI, confidence interval.

DISCUSSION

In this large single-institution cohort, late grade ≥ 3 GI and GU toxicities after prostate SBRT were observed, including clinically meaningful severe events such as organ perforation and fistula formation. Diabetes mellitus showed a consistent association with an increased risk of severe toxicity in univariable and subgroup analyses.

The overall incidence of grade ≥ 3 toxicity in our cohort was broadly comparable to that reported in previous SBRT series using prescribed doses mainly in the range of 35–40 Gy in five fractions. Systematic reviews of prostate SBRT have generally reported severe late toxicity rates of ~1% for GI events and 2% for GU events [5]. In our cohort, the incidence of severe GU toxicity was similar, whereas severe GI toxicity was slightly higher.

Diabetes mellitus is associated with chronic microvascular dysfunction and impaired tissue repair capacity. As recovery from radiation-induced injury relies in part on sufficient tissue perfusion and oxygenation, these vascular abnormalities may contribute to increased vulnerability to late radiation toxicity [19]. Indeed, since the era of conventionally fractionated radiotherapy, diabetes mellitus has been reported to be associated with an increased risk of late radiation-related adverse events [19–21]. Reports specifically addressing diabetes mellitus as a risk factor for severe late toxicity after prostate SBRT remain scarce. Despite the limited number of events and the absence of multivariable adjustment, our univariable and subgroup analyses consistently suggested an association between diabetes mellitus and severe late toxicity following prostate SBRT. Given the large dose per fraction and high biologically effective dose delivered to surrounding normal tissues in SBRT, it is biologically plausible that diabetes-related microvascular dysfunction may exert a greater impact on late toxicity in this setting. These findings suggest that diabetes mellitus may be a clinically relevant factor when counseling patients and planning prostate SBRT.

Among patients treated with ≤40 Gy, the observed toxicity profile appeared generally consistent with prior SBRT reports. In contrast, patients treated with ≥42.5 Gy showed a numerically higher incidence of severe toxicity. This observation may appear broadly consistent with prior dose-escalation studies suggesting a possible association between higher SBRT dose and late severe toxicity [11, 12, 14]. However, interpretation of this finding requires substantial caution. In the present cohort, prescription dose, spacer use, treatment era, and follow-up duration were closely interrelated. The protective association of rectal spacer use should also be interpreted cautiously, because spacer adoption was closely linked to treatment era and was more frequent in the higher-dose group. Thus, the higher incidence of severe toxicity in the higher-dose group and the protective association of rectal spacer use may appear difficult to reconcile at first glance. However, these findings should not be interpreted as directly contradictory, because they were derived from separate univariable comparisons rather than from a single adjusted model. In addition, the protective association of rectal spacer use was estimated across the full cohort, whereas the ≥42.5 Gy group represented a relatively small subset of the overall study population. Detailed dosimetric parameters were also not available in the present analysis, and unmeasured differences in dose distribution to surrounding organs at risk may have further contributed to the observed findings. Therefore, the present data do not allow a definitive determination of the relative contributions of dose escalation, spacer use, and other correlated factors to the observed differences in severe toxicity.

Hypertension, antithrombotic agent use, NCCN high-risk classification, and long-term ADT were also associated with severe late toxicity in univariable analysis. Some of these associations may reflect underlying vascular vulnerability or treatment-related factors. However, several of these variables are strongly correlated, and given the limited number of events and the exploratory nature of these analyses, these findings should be interpreted cautiously.

This study has several important limitations. First, its retrospective single-institution design introduces inherent selection bias and limits generalizability. Second, the limited number of grade ≥ 3 toxicity events precluded fully adjusted multivariable analyses, and the risk factor findings should therefore be interpreted as exploratory. In addition, clinical heterogeneity within the cohort may have further complicated the interpretation of potential risk factors. Third, although diabetes mellitus was consistently associated with severe toxicity, detailed information regarding diabetes severity, glycemic control, and disease duration was not systematically available in this retrospective dataset. Consequently, potential dose–response relationships between diabetes burden and radiation injury could not be evaluated. Fourth, treatment strategies evolved over time, including changes in prescribed dose levels, dose–volume constraints and spacer use, which may have influenced toxicity outcomes. Because these treatment factors were introduced sequentially over time, their individual contributions to toxicity risk could not be clearly disentangled in this retrospective analysis. Accordingly, the protective association of rectal spacer use in the overall cohort and the higher incidence of severe toxicity in the ≥42.5 Gy subgroup should not be interpreted as directly contradictory. Fifth, detailed dosimetric parameters such as rectal, bladder, and urethral dose–volume histogram metrics were not incorporated into the present analysis. Because organ-specific dose–volume parameters are important determinants of radiation toxicity, their absence limits the ability to evaluate the relative contribution of dosimetric versus clinical risk factors. Future studies integrating comprehensive dosimetric analyses are warranted to clarify the interaction between clinical risk factors and dose–volume effects. Sixth, moderate (grade 2) toxicities were not systematically evaluated. Because such events are frequently managed conservatively and inconsistently documented in retrospective clinical data, reliable assessment was not feasible. In addition, quality-of-life outcomes were beyond the scope of the present analysis. Future studies incorporating patient-reported outcomes are warranted to provide a more comprehensive evaluation of treatment-related morbidity. Finally, patients treated with higher prescribed doses had shorter follow-up durations, and additional late toxicity events may emerge with longer observation. Despite these limitations, this study represents one of the largest real-world assessments of severe late toxicity after prostate SBRT and provides clinically meaningful insights into potential patient-related risk factors.

CONCLUSION

In conclusion, this large real-world cohort provides a comprehensive characterization of severe late toxicity following prostate SBRT. Although grade ≥ 3 GI and GU toxicities were infrequent overall, clinically meaningful and potentially devastating events were observed. Exploratory analyses suggested that patient-related clinical factors, including diabetes mellitus, may contribute to increased vulnerability to severe late radiation injury. These findings highlight the importance of considering patient-related clinical factors in treatment planning. Further studies with longer follow-up are warranted to validate these observations and to refine risk-adapted approaches for prostate SBRT.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

FUNDING

None.

Presentation at a conference

This study was presented in part at the 38^th Annual Meeting of the Japanese Society for Radiation Oncology (JASTRO), 2025.

Clinical Trial Registration

Not applicable.