Introduction

Prostate cancer is the second most common form of cancer affecting men worldwide [1]. The majority (94%) of patients with prostate cancer have curable localised disease, for which the treatment options include active surveillance, surgery or radiotherapy [2]. Radiotherapy is a common treatment for localised prostate cancer [3, 4]. However, the incidence of late genitourinary toxicity (GUT) and its associated burden of treatment across a variety of practice settings remains poorly understood. Radiotherapy injuries often present late due to progressive fibrosis and the difficulties in accurately recording these long-term adverse effects are reported in the literature frequently [5,6,7]. The majority of studies on the incidence of genitourinary toxicity after radiotherapy and its associated burden of treatment are studies from specialised high-volume single centres [7,8,9]. There are few multi-institutional studies [10,11,12] and the randomised trials often involve a disproportionately younger and healthier patient demographic when compared to a typical population [13, 14]. An improved understanding of the incidence of late treatment-related genitourinary toxicity following prostate radiotherapy would enhance patient-centred decision making [6].

The primary aim of this study was to determine the cumulative incidence of treatment-related genitourinary toxicity following external beam prostatic radiotherapy in patients with localised prostate cancer at a population level. The secondary aims were to determine clinical factors predictive of genitourinary toxicity and the volume of admissions and procedures required.

Materials and methods

Participants

A population-based prospective cohort study of all patients with localised (T1–T3, according to the American Joint Committee on Cancer) biopsy-proven prostate cancer who underwent primary external beam radiotherapy (EBRT) was performed between January 1, 1998, and January 31, 2019, in South Australia. We excluded patients with metastatic prostate cancer and those without a histological tissue diagnosis of prostate cancer. We excluded patients who underwent adjuvant radiotherapy following either radical prostatectomy, or prior radiotherapy treatment (Fig. 1).

Fig. 1
figure 1

HYPERLINK "sps:id::fig1||locator::gr1||MediaObject::0" Patient selection flow chart. SA-PCCOC, South Australian Prostate Cancer Clinical Outcomes Collaborative; ISAAC, Integrated South Australian Activity Collection; RP, radical prostatectomy; BT, brachytherapy; GUT, genitourinary toxicity

Full size image

The South Australian Prostate Cancer Clinical Outcome Collaborative (SA-PCOCC) registry prospectively recruits > 90% of patients who are diagnosed with prostate cancer in South Australia. We linked patient records from the SA-PCCOC registry with the Integrated South Australian Activity Collection (ISAAC) Hospital Administrative Database to identify patients who presented to any major hospital in South Australia with treatment-related genitourinary toxicity, as defined by a pre-selected list of International Classification Disease 10th Edition (ICD-10-AM)/Australian Classification of Health Interventions (ACHI). Data linkage was performed by matching patient identifiers within Envido (Adelaide, South Australia). The list of admission and procedures codes were selected based on the literature [6], and recommendations from a multidisciplinary panel, including a urologist, radiation oncologist, general surgeon and a clinical epidemiologist (Appendix 1). Baseline characteristics including age, Charlson Comorbidity Index, anticoagulant medication use, and oncological characteristics, including T-stage, Gleason score and baseline Prostate-specific antigen (PSA) level were extracted. Treatment-related factors including dose (Gray), fractionation and date of treatment completion were also extracted.

Primary outcomes

The treatment-related complication categories used were hospital admission and urological procedures associated with genitourinary toxicity. Genitourinary toxicity-related hospital admission or procedures required for each patient were identified using the ISAAC Database (using the relevant hospital admission or procedure code based on the ICD-10 or ACHI codes). The time to the first genitourinary toxicity-related hospital admission, death or censor were analysed to determine the cumulative incidence of genitourinary toxicity. Patients were censored at the last date of the last admission in the ISAAC electronic hospital database.

Secondary outcomes

Demographic factors assessed included age (continuum), Charlson comorbidity score, diabetes (yes/no), hypertension (yes/no), use of anticoagulant (yes/no), smoking history (yes/no), bladder outlet obstruction (yes/no), Transurethral resection of the prostate (TURP) before radiotherapy (yes/no), T stage (T1 vs T2 vs T3), initial prostate-specific antigen level (continuum) and dose (continuum and > 80 Gy vs ≤ 80 Gy). Furthermore, the admission data was separated into patients who received EBRT < 2009 and ≥ 2009, to account for the use of Three-dimensional conformal radiation therapy (3DCRT) and Intensity Modulated Radiotherapy/Volumetric modulated arc therapy (IMRT/VMAT), respectively.

The overall burden of treatment, as defined by the volume of admissions as well as non-operative, minor operative and major operative procedures was determined. Non-operative procedures were defined as ACHI codes involving urethral catheterization or bladder irrigation. Minor operative procedures were defined as ACHI codes involving urethral dilation, cystoscopy, suprapubic catheter insertion, retrograde pyelogram, antegrade or retrograde ureteric stenting. Major operative procedures were defined as ACHI codes involving transurethral resection, ureteroscopy or open surgical procedure.

The outcomes were reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [15].

Statistical analysis

The cumulative incidence of hospitalisation for treatment-related genitourinary complications was determined. Patients were considered to be at risk of complications from the end date of their radiotherapy until either the date of their first admission related to genitourinary toxicity, last date of follow-up or date of death, according to the SA-PCCOC registry. The patient-related baseline characteristics and the volume of hospital admissions and procedures were summarised and compared. Categorical variables were compared using the Fischer Exact Test or Pearson’s Chi-square test. Continuous parametric and non-parametric variables were compared using one-way ANOVA or the Kruskal–Wallis Rank Sum test, respectively. p values were calculated for each variable compared and p < 0.05 was considered significant. Relationships between genitourinary toxicity-related hospital admission and patient, tumour or treatment characteristics were analysed using Cox proportional hazard regression at univariate and multivariate levels. The regression analyses’ results are presented as a hazard with a 95% confidence interval. Missing clinical data was replaced using multiple imputations by chained equations before regression analysis (Fig. 1). All statistical analysis was performed using R language, Version 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria) [16].

Results

There were 3350 patients with prostate cancer treated with primary external beam radiotherapy in this cohort. We excluded 820 patients who were initially treated surgically, with either robotic-assisted laparoscopic prostatectomy (n = 579) or open radical prostatectomy (n = 241). We also excluded 388 patients who were treated with brachytherapy before external beam radiotherapy and four patients with T4 disease (Fig. 1). All the included patients underwent primary EBRT, with a median (IQR) of 74 Gy (70–78) in 37 fractions (35–39). The median (IQR) age at diagnosis of the included patients was 71 (66–76). Most patients had Stage II (n = 914, 58%) and high-risk disease (n = 1517 [51%]), according to the National Comprehensive Cancer Network (NCCN) 2017 scoring system. Table 1 summarises the patient demographic, oncological and treatment dosimetric characteristics.

Table 1 Baseline characteristics of all included prostate cancer patients treated with primary radiotherapy
Full size table

The 5- and 10-year cumulative incidence of admission to hospital for treatment-related genitourinary toxicity were 14.8% (95% CI 13.4–16.2) and 28.4% (95% CI 26.3–30.6), respectively (Fig. 2). The 5- and 10-year cumulative incidence of patients in this cohort requiring a urological operative procedure for a treatment-related GUT were 9.9% (95% CI 8.7–11) and 18% (95% CI 16.1–19.9), respectively (Fig. 2). The 5-year cumulative incidence of treatment-related genitourinary toxicity hospital admission were 18% (95% CI 15–20%) and 12% (95 CI 11–14), amongst patients treated before and after 2010, respectively (p < 0.001; Fig. 3).

Fig. 2
figure 2

10-year cumulative incidence of admission or urological procedure (non-operative or operative) for genitourinary toxicity following prostate radiotherapy

Full size image
Fig. 3
figure 3

10-year cumulative incidence of admission for genitourinary toxicity following prostate EBRT by clinical factors

Full size image

There were 652 (19.5%) prostate cancer patients who required hospital admission for genitourinary toxicity after primary EBRT, with a total of 2545 hospital admissions, of which 1040 (41%) occurred in the emergency setting. Four-hundred and nine (63%) of these patients had multiple admissions, with a total of 1893 (74%) readmission related to genitourinary toxicity. Haematuria was the most common genitourinary toxicity (n = 386, 59%), and of these patients, 108 (28%) required blood product transfusion, 8 (2%) required HBOT and 4 (1%) required surgical urinary diversion. Table 2 summarises the treatment-related outcomes amongst patients with genitourinary toxicity following primary EBRT. Four-hundred and nine (12%) patients developed genitourinary toxicity which required management with a urological procedure, with a total of 106 (4.2%) non-operative, 1044 (41%) minor and 57 (2.2%) major operative urological procedures (Table 3). The most common procedure was diagnostic cystoscopy (701/1101 [64%] of all procedures).

Table 2 Volume of admissions and procedures for genitourinary toxicity amongst included prostate cancer patients treated with primary prostate radiotherapy
Full size table
Table 3 Cumulative Incidence of GU toxicity admission with competing risk regression including overall mortality
Full size table

Patients with BOO without TURP prior to EBRT, had the highest 10-cumulative incidence of admission for genitourinary toxicity (77% [70%, 82%] vs 20% [18%, 22%] p < 0.001; Table 3, Fig. 3). In addition, patients with BOO without TURP prior to EBRT had the most hospital admissions (178/246 [72%] vs 474/3104 [15%], p < 0.001), emergency admissions (136/246 [55%] vs 273/3104 [8.8%], p < 0.001) and readmissions (110/246 [45%] vs 282/3104 [9.1], p < 0.001), for treatment-related genitourinary toxicity (Table 1). Patients with BOO without TURP before EBRT were at the highest risk of developing genitourinary toxicity after adjustment for age, diabetes, smoking, urinary incontinence and EBRT before 2009 (HR 5.87 [95% CI 4.8–7.17], p < 0.001; Table 4).

Table 4 Cox proportional hazards regression analysis with imputation by multiple chained equations of predictive factors for genitourinary toxicity-related admission following primary prostate EBRT
Full size table

Discussion

This is one of few studies to evaluate the cumulative incidence of treatment-related genitourinary complications following radiotherapy for prostate cancer at a population level and the first in Australia. The high 10-year cumulative incidence (28.4%) of hospital admission due to treatment-related genitourinary toxicity exceeds previous estimates following primary EBRT [4, 6, 7]. The date of radiotherapy made a minimal difference in the 10-year cumulative incidence of genitourinary toxicity-related admission amongst patients in this cohort, and was not an independent predictor of genitourinary toxicity after adjustment for age, comorbidity, smoking and BOO in multivariable analysis (HR 0.87 [95% CI 0.72, 1.04], p = 0.12; Table 4). This is also the first Australian study to determine the volume of admissions and urological procedures for the management of radiotherapy treatment-related genitourinary complications at a population level. Greater than one-third of genitourinary toxicity-related hospital admissions occurred in the emergency setting. There were a significant number of admissions with a prolonged length of stay of ≥ 3 days. Whilst haematuria was the most common presentation, we are unable to confirm the diagnosis of radiation cystitis due to the limitations associated with administrative coding, we can infer the diagnosis of severe hemorrhagic radiation-induced cystitis occurred in 12/3351 (0.4%) of patients, with 8/3351 (0.2%) and 4/3351 (0.1%) patients requiring HBOT and surgical urinary diversion, respectively. A significant number of patients (18%) required an invasive urological procedure. There were significantly fewer hospital admissions and procedures amongst patients treated with EBRT after 2009, which may reflect improvements in radiotherapy techniques or the shorter follow-up in this group, which likely underestimated late toxicity.

Three large population-based studies have been published in this area with patients from the USA [17], Canada [6], and England [7]. A total of 307,252 patients were described [4, 6, 7]. However, like several other studies [18, 19], these studies did not include patient baseline oncological characteristics [6], or important treatment-related factors, including the dose and fractionation use in the radiation treatment used [6, 7, 17]. The study by Sheets et al., was the first study to demonstrate an increased risk of patients developing genitourinary toxicity following IMRT as compared to conformal radiation therapy, (absolute risk, 5.9 vs 503 per 100 person-years; relative risk, 1.12; 95% CI 1.03–1.20) [17]. Only one of these studies reported 5-year cumulative incidence of treatment-related genitourinary toxicity, which was determined to be 10.7 (95% CI 10.1–11.3) [7]. The estimate determined by the latter study was limited by missing values for the prostate cancer risk group (n = 5753) and radiotherapy treatment region (n = 3793) [7]. The other study reported a 22.2% (95% CI 21.7–22.7) 5-year cumulative incidence of admission for either genitourinary or gastrointestinal treatment-related complication and a 32.0% (95% CI 31.4–32.5) 5-year cumulative incidence of needing a urological procedure [6]. All three studies lacked a 60-month endpoint and this may have led to an underestimation of the late genitourinary toxicity events, as is the case with many other studies [19, 20]. The majority of studies of > 5-year genitourinary toxicity are not population-based, tend to focus on a narrower range of toxicity and have a shorter follow-up duration [9, 21].

Patients with bladder outlet obstruction without TURP before EBRT were at the highest risk of developing genitourinary toxicity after adjustment for age, diabetes, smoking, urinary incontinence and EBRT before 2009 (HR 5.87 [95% CI 4.8–7.17], p < 0.001; Table 4). Similarly, many other studies have also shown that pre-existing urinary symptoms can influence radiotherapy-related genitourinary toxicity [22,23,24]. TURP before radiotherapy demonstrated a protective effect against genitourinary toxicity amongst patients with bladder outlet obstruction in our study (HR 3.6, 95% CI 3.01–4.46, p < 0.001); however, other studies have shown TURP might deteriorate late urinary symptoms [25, 26]. Similarly, several other studies [23, 24, 27] have supported our finding that diabetes is an independent predictor of genitourinary toxicity in patients with prostate cancer treated with radiotherapy (HR 1.25, 95% CI 1.08–1.53, p < 0.004). Furthermore, the role of diabetes may be increasingly important in the era of dose-escalated (≥ 74 Gy) IMRT, as shown by Kalakota et al., who reported diabetes to be an independent predictor of late grade 3 genitourinary toxicity (RR 2.74, p = 0.004) in their multivariate analysis [28]. However, a few studies did not support the impact of diabetes on treatment-related genitourinary toxicity [29,30,31].

Less known is the impact of age on radiation-induced genitourinary toxicity, which may reflect physiological changes and altered clinical decision-making. Whilst we found that increased age was associated with significant lower cumulative 5-, 10- and 15-year EFS rates (p = 0.041, Table 4) in univariate analysis (HR 1.02 95% CI 1.01–1.03, p < 0.001), this did not retain significance in multivariable regression (p = 0.6). However, other studies have shown increased age to be an independent predictor of treatment-related genitourinary toxicity [6, 23, 32], including the study by Nam et al., which reported a higher incidence of hospital admission due to genitourinary toxicity (HR 1.007, 95% CI 1.003–1.010, p < 0.0001) amongst patients with prostate cancer treated with radiotherapy (n = 16,595) in a multivariable analysis performed in Cox proportional hazard modelling, adjusted for age and comorbidity treatment [6].

Similarly, whilst we found an increased risk of genitourinary toxicity amongst patients with a history of anticoagulation medication use on univariable analysis (HR 2.03 95% CI 1.67–2.49, p < 0.001), the significance was not retained in multivariable analysis (p = 0.3). However, in multivariable analysis, other studies have shown an increased risk of haematuria associated with anticoagulant use (RR 2.89, p = 0.01) [23].

Whilst we found that Charlson comorbidity score was not associated with genitourinary toxicity, in univariate analysis (HR 1.06, 95% CI 0.99–1.12, p < 0.091), the study by Nam et al. found that increased comorbidity, as measured by the Johns Hopkins University ACD Case-Mix System, was associated with a higher incidence of hospital admission in multivariate analysis (HR 1·08, 95% CI 1.07–1.09, p < 0.0001) [6].

Similarly, whilst we found no statistically increased risk of toxicity for patients with a history of hypertension (HR 3.91, 95% CI 0.98–15.7, p = 0.12) on univariable analysis, other studies have shown a positive association [30, 33]. Contrastingly, other studies have reported a protective effect of hypertension, suggested to be associated with antihypertensive medication intake [34], with Barnett et al. reporting a correlation with decreased risk of a poor urinary stream (HR 0.25, 95% CI 0.09–0.71, p = 0.009) [30].

Similarly, the data on dose-related genitourinary dysfunction has been controversial, and whilst some studies suggested a correlation between dose to the bladder and genitourinary toxicity [23, 24, 35,36,37,38,39,40,41], this has generally been unconfirmed by other authors [26], including the current study in univariable regression analysis (p = 0.4). This inconsistency may be due to confounding differences in treatment scheme (target volume, position during treatment, bladder volume variation, technique, dose), patient characteristics, grading scale and the length of follow-up [42,43,44].

Similarly, patients who received radiotherapy before 2009 had a higher 10-year cumulative incidence of admission for genitourinary toxicity (29% [26%, 31%] vs 19% [16%, 21%] p < 0.001; Table 3, Fig. 3). In addition, patients with EBRT before 2009 had more hospital readmissions for genitourinary toxicity (1879 [74%] vs 1354 [77%], p < 0.001), urinary retention (757 [43%] vs 287 [38%], p = 0.032) as well as more non-operative (p < 0.001) and minor-operative procedures (p < 0.001) compared with patients who received radiotherapy ≥ 2009 (Table 1). However, date of treatment before 2009 was not an independent predictor of hospitalisation for genitourinary toxicity, after adjustment for age, comorbidity, smoking and BOO (HR 0.87 [95% CI 0.72, 1.04], p = 0.12; Table 4).

Our study has several limitations. First, whilst the use of administrative data coding based on diagnostic and admission codes has been validated in other claims-based studies assessing severe pelvic adverse effects after radiotherapy [45], the number of genitourinary complications has likely been under-reported given the retrospective data-linkage methods used. For example, we would not have captured complications that are non-life-threatening (e.g., lower urinary tract symptoms from urethral stricture or bladder neck contracture) or which do not require further procedures. Furthermore, the sampling methodology used does not account for patients who may have had complications in other states. However, the study benefits from population-level data and longer duration of follow-up. In addition, we are unable to establish a causal link between radiation treatment and the reason for admission. These potential confounding factors may lead to the incorrect attribution of radiation-related toxicity in our data set, especially for late complications given the distant temporal relationship [19, 46]. The work presented here is descriptive and may motivate further investigations focusing on causal pathways, mechanisms of action and preventive strategies. Toxicity grades were unable to be reported, as these were not coded in administrative data. The study does not include radiation-associated secondary malignancy, gastrointestinal or other pelvic treatment-related complications (e.g., rectal and pubic symphysis fistula).

Conclusions

Genitourinary complications after radiotherapy for prostate cancer are common. Although there continue to be significant advancements in radiotherapy techniques, patients and physicians should be aware of the risk of late toxicity when considering treatment options for prostate cancer. Further research is needed to identify predictive factors and develop models predicting late treatment-related genitourinary toxicity to improve pre-treatment counselling and enhance patient-centered decision making.