Clinical and functional outcomes for risk‐appropriate treatments for prostate cancer

Abstract Objectives To describe real‐world clinical and functional outcomes in an Australian cohort of men with localised prostate cancer according to treatment type and risk category. Subjects and methods Men diagnosed from 2008 to 2018 who were enrolled in South Australian Prostate Cancer Clinical Outcomes Collaborative registry—a multi‐institutional prospective clinical registry—were studied. The main outcome measures were overall survival, cancer‐specific survival, decline in functional outcomes, biochemical recurrence and transition to active treatment following active surveillance. Multivariable adjusted models were applied to estimate outcomes. Results Of the 8513 eligible men, majority of men (46%) underwent radical prostatectomy (RP) followed by external beam radiation therapy with or without androgen deprivation therapy (EBRT +/− ADT) in 22% of the cohort. Five‐year overall survival was above 91%, and 5‐year prostate cancer‐specific survival was above 97% in the low‐ and intermediate‐risk categories across all treatments. Five‐year prostate cancer‐specific survival in the active surveillance group was 100%. About 37% of men with high‐risk disease treated with RP and 17% of men treated with EBRT +/− ADT experienced biochemical recurrence within 5 years of treatment. Of men on active surveillance, 15% of those with low risk and 20% with intermediate risk converted to active treatment within 2 years. The decline in urinary continence and sexual function 12 months after treatment was greatest among men who underwent RP while the decline in bowel function was greatest for men who received EBRT +/− ADT. Conclusion This contemporary real‐world evidence on risk‐appropriate treatment outcomes helps inform treatment decision‐making for clinicians and patients.

active surveillance, 15% of those with low risk and 20% with intermediate risk converted to active treatment within 2 years.The decline in urinary continence and sexual function 12 months after treatment was greatest among men who underwent RP while the decline in bowel function was greatest for men who received EBRT +/À ADT.

Conclusion:
This contemporary real-world evidence on risk-appropriate treatment outcomes helps inform treatment decision-making for clinicians and patients.
biochemical recurrence, health outcomes, prostate cancer, quality of life, survival

| INTRODUCTION
In 2021, prostate cancer accounted for 23% of all new cancer cases and 12% of all cancer deaths among Australian males, imposing a significant burden on the healthcare system. 1 While prognosis is generally very favourable for men with localised disease who undergo radical treatments (i.e.prostatectomy or radiotherapy), the impact on physical functioning can be substantial. 2Side effects following radical prostatectomy (RP) include erection and ejaculation problems, urinary incontinence, reduced penis length and loss of fertility, while radiation therapies may lead to urinary problems such as frequency, urgency, burning/discomfort and haematuria; bowel problems such as increased frequency, urgency and rectal bleeding; erection and ejaculation problems; and fatigue. 3Active surveillance is regarded as a safe alternative for men with low-risk (indolent) disease to reduce morbidity associated with radical therapies. 4For men on active surveillance, anxiety about potential disease progression may impact their quality of life or lead them to unnecessarily opt for active treatment. 5Men who require androgen deprivation therapy (ADT) will invariably experience hormonal imbalances and be at increased risk of developing additional comorbidities such as cardiovascular disease and diabetes. 6sk categorisation, based on biopsy grade, prostate-specific antigen (PSA) levels and clinical stage at diagnosis, is the major determinant of appropriate treatment options for men with prostate cancer. 7However, most treatment guidelines specify a range of risk-appropriate treatment options, with choice being guided by patient and physician preferences on survival benefits and impacts on physical functioning. 8Evidence from 'real-world data', reflecting actual clinical practice, can be a useful guide for decision-making. 9is study describes clinical and functional outcomes following different primary treatment modalities in a contemporary Australian cohort using outcomes data from the South Australian Prostate Cancer Clinical Outcome Collaborative (SA-PCCOC) registry.We evaluated survival, biochemical recurrence, conversion to active treatment during active surveillance and impact on sexual, urinary, bowel and hormonal function.Information on date and cause of death was obtained from SA-PCCOC registry via the SA births, deaths and marriages registry and SA Cancer Registry, which undertakes routine linkages with state and national death indexes.Biochemical recurrence was derived from follow-up PSA data, and defined according to treatmentspecific criteria for RP (two consecutive PSA measures >0.2 ng/dL 14 ) and radiotherapy (2.0 ng/dL above the PSA nadir following radiotherapy 15 ).Functional outcomes were measured at baseline and 12-month post-treatment using Expanded Prostate Cancer Index Composite (EPIC)-26 domain scores. 16EPIC-26 provides scores for five functional domains that range from 0 to 100, with higher scores indicating better function.We applied Skolarus et al.'s suggested values (10 points for sexual function, 6 points for urinary continence, 5 points for urinary obstruction and 4 points for both bowel and hormonal function) to define clinically significant declines in function at 12-month post-treatment. 17

| Statistical analyses
Flexible survival models were executed to estimate all-cause and disease-specific survival probabilities for all treatments, the probability of developing biochemical recurrence following radical treatments and the probability of transitioning to active treatment at 2 and 5 years for men on active surveillance, as described by Lambert et al. 18 Separate subgroup analyses were undertaken for each risk category, with adjustment for age, socio-economic advantage and year of diagnosis, followed by post-estimation prediction to estimate survival probabilities for each treatment in each relevant risk category.Follow-up time was calculated from the date of primary treatment to the date of the event, date of death or censoring date (14 June 2021).
Men who died from other causes were censored from prostate cancer-specific survival estimation.Active surveillance was excluded from modelling outcomes for high-risk disease, while ADT alone was excluded from models for low-risk disease.
For functional outcomes, we used logistic regression models to predict likelihood of experiencing clinically significant decline.Functional outcomes data were not available for all men enrolled in SA-PCCOC registry.This is due to late notification of their diagnosis (so that baseline surveys were not able to be sent before treatment) or survey non-response (response rates at each round were $60%).
Inverse probability propensity weighting was used to adjust for any variability between respondents and non-respondents.The use of propensity weighting to account for missingness has been suggested in previous literature. 19nal models were fitted adjusting for inverse probability weight, risk category, age at diagnoses, year of diagnosis and socio-economic advantage.In these analyses, no adjustments were made for salvage treatments or conversion to active treatment during the 12-month follow-up period.Sensitivity analyses in which men who had EBRT within 12 months of RP were excluded (n = 76) did not affect the results substantially; hence, we reported outcomes collectively for all men who underwent RP.
For greater equivalence across treatment groups, we reported probabilities of each outcome measure for a 68-year-old man (i.e. the average age at diagnosis of the study cohort).All analyses were performed using Stata version 15.0 (StataCorp, College Station, TX, USA).

| RESULTS
In total, 8513 men were included.The mean age of participants at diagnosis was 68 years (SD ± 9).Forty six per cent had undergone RP, 22% had EBRT +/À ADT, 16% initially underwent observant management (active surveillance or watchful waiting), 9% received ADT alone, and 6% had brachytherapy.Summaries of patient characteristics are presented in Table 1.
In our study, only 38% of men with low-risk disease were on active surveillance (Table S1).In the low-risk group, proportion of men who underwent RP decreased overtime, while uptake of active surveillance increased from 7% in 2008 to 75% in 2018.Throughout the years, about half of men in the intermediate-risk group underwent RP, peaking at 68% in 2016, whereas all other treatments remained relatively stable.Most men with high-risk disease had either EBRT or ADT till 2013, but from 2013 onwards, RP became more popular.A decline in ADT was observed (from 42% in 2008 to 18% in 2017) among the high-risk group (Figure 1).

| Overall survival
Median follow-up time was 7 years (interquartile range, 4-10), during which time 1298 patients had died including 430 from prostate cancer.Across all treatment options, the 5-year overall survival probability for a 68-year-old man with low-or intermediate-risk disease exceeded 91%.For high-risk disease, the 5-year survival probability ranged from 61% (95% CI: 46-78) among those on watchful waiting to 96% (95% CI: 95-98) following RP.Ten-year overall survival probabilities were above 85% for low-and intermediate-risk disease across all treatment options but varied more widely by treatment type for high-risk disease (92%, 80%, 73% and 45% for RP, EBRT +/À ADT, ADT alone and watchful waiting, respectively) (Table 2).

| Patient-reported outcome measures
Risk-adjusted probabilities for having a clinically significant decline in physical functioning at 12 months after initial treatment, compared with baseline levels, are presented in Table 4 for each treatment option.
For the average 68-year-old man, the probability of having

| DISCUSSION
This study reports real-world clinical and functional outcomes following each of the main primary treatment modalities for prostate cancer for a contemporary Australian cohort.Overwhelmingly, our study findings indicated favourable survival outcomes across most risk and treatment categories.Exceptions include men with high-risk disease on either ADT or watchful waiting.
Given our study used real-world data with inherent selection biases, we did not seek to directly compare outcomes between prostate cancer treatments.Rather, we have reported the probability of each outcome for the average man facing decisions about prostate cancer treatments.As such, our findings do not imply that men should choose one treatment over another, as each man's circumstances will differ. 3Clinician and patient considerations about treatment choices will depend on a range of factors including underlying medical conditions, cancer aggressiveness and extent of spread, personal circumstances and patient preferences.As reported previously, 20 most men with localised prostate cancer die from causes other than their cancer.
The presence of other comorbid conditions is likely to contribute to both poorer overall and poorer disease-specific survival among men with high-risk disease. 21veral findings from this real-world data are consistent with results from the ProtecT trial including high prostate cancer-specific survival rates, 22 greatest decline of sexual function and urinary continence in prostatectomy group 2 and worse bowel function in the radiotherapy group. 2 Nevertheless, unlike our findings, the ProtecT trial showed less decline in functioning among the active monitoring group.management, including imaging technologies that enabled clinicians to more precisely stratify risk and recommend therapies based on cancer prognosis. 23The advancements in imaging and radical therapy techniques are likely to have led to improvements in functional and oncologic outcomes.For example, PSMA-PET imaging improves the detection or staging of prostate cancer, 24 whereas different nervesparing surgical techniques [25][26][27] and more accurate delivery of radiotherapies such as intensity-modulated radiation therapy (IMRT) and image-guided radiation therapy (IGRT) are likely to have led to an improvement in functioning after treatment. 28,29Advancements in multimodal therapies are expected to further improve survival, particularly for men with high-risk disease.
Our results relating to biochemical recurrence are in line with the findings of the systematic review by Fakhrejahani et al 30 report that about one-third of men on RP and 25%-33% of men on radiotherapy develop biochemical recurrence within 5 years.Conversely, a more recent report indicates a significantly higher risk of biochemical recurrence for patients receiving EBRT compared to RP 10 years after treatment, and the cumulative incidence of biochemical recurrence was 23.8% (95% CI: 17.9-30.2) in the RP and 43.0%(95% CI: 34.9-50.8) in the EBRT groups. 31Our findings regarding biochemical recurrence should be interpreted with caution.Differences between the RP and the EBRT +/À ADT groups are likely the higher proportion who remain on ADT after radiotherapy, which lowers PSA levels, 32 leading to an underestimate of biochemical recurrence in the EBRT +/À ADT group.In addition, it is difficult to directly compare treatment modalities due to different (standard) definitions of biochemical recurrence for RP and radiotherapy and potentially differences in the frequency of post-treatment PSA testing.Furthermore, our findings in the EBRT +/À ADT group may reflect the lower risk of biochemical recurrence associated with the current trend towards the use of dose-escalated radiotherapy, as reported by Tanaka et al. 33 While the relationship between biochemical recurrence and prostate cancer mortality remains unclear, biochemical recurrence is an important trigger for secondary treatment (e.g.salvage radiotherapy/RP or ADT) with associated morbidity.
Our results relating to patient reported outcomes clearly show that all treatment modalities negatively impacted physical functioning, although the domain-specific effects were varied between the treatments.Compared to other studies, 34,35 we reported quite high levels of impact for most functional outcomes.This is, in part, due to choosing to report the likelihood of any 'clinically significant decline', defined as the smallest differences that patients can notice. 17Hence, our results are indicative of the impact on functional outcomes that could be problematic from a patient's perspective.
In this study, RP remained most commonly used treatment for men with intermediate-and high-risk groups (2013 onwards), while the use of radiotherapies was relatively steady across the years.
Recent evidence from analyses of the Australian Medicare Benefits Schedule data has shown that the use of RP increased between 2002 and 2009, but from 2012 to 2016, the rates of RP (15% drop) subsequently decreased, while the use of EBRT remained steady. 36 have also observed an increased trend of active surveillance uptake by men with low-risk prostate cancer.This is in line with reports from the binational registry (PCOR-ANZ) where the proportion of men on active surveillance has increased over time (54% in 2015 to 71% in 2018), while radical treatment rates have decreased. 37Active surveillance is recommended for men diagnosed with low risk and some favourable intermediate-risk localised prostate cancer, with an aim to reduce overtreatment and minimise the negative side effects of radical therapies while preserving quality of life, ensuring the option for curative treatment, without losing the window of curability. 38,39e strength of this study is its use of 'real-world data' that reflects the actual experience of men diagnosed with prostate cancer.
Our use of a decade worth of data that collects baseline PROMs is a contemporary addition to existing evidence.PROMs were reported based on 'any' clinically significant decline, which allows interpretation of the data from patient perspectives.We have used advanced statistical models standardised for risk classification and age, have accounted for missingness (imputed variables defining risk) and have applied propensity weighting to adjust for missing EPIC items.
SA-PCCOC is a community-wide registry incorporating a heterogeneous patient population including those managed in large tertiary centres as well as smaller private practices and hence can provided a population-wide picture of outcomes for prostate cancer.Populationbased prospective disease-specific registries have value in influencing clinical practice and improving outcomes overtime.They provide important information on mortality, disease characteristics at presentation, long-term oncologic and quality of life outcomes as well as quality of care. 40Two recent studies using the binational prostate cancer registry (PCOR-ANZ) showed population-level data demonstrated clinically equivalent PROMs from randomised controlled trials. 41,42Another study using PCOR-Vic revealed that positive margin rate has significantly reduced since the inception of the registry, where the authors commented that benchmarking against established aspirational real-world post-operative outcomes facilitates the detection of areas of improvement. 43Further improvements in data quality will increase the relevance of studies derived from such registries.SA-PCCOC as a main contributor to the PCOR-ANZ, evidence on riskadjusted posttreatment outcomes would leverage other registries and provide benchmarks to encourage quality improvement efforts to produce improved outcomes.
Limitations include that we did not account for secondary treatments in our assessment of outcomes, due to incomplete data on additional treatments.Adjuvant and salvage therapies were not accounted for due to practices vary widely among treating clinicians regarding salvage and adjuvant therapies, and data were not consistently collected.The receipt of adjuvant or salvage therapy with radical therapies could affect outcomes; for example, the decline in PROMs among men who had EBRT +/À ADT is likely to be worse than among men who had EBRT alone.In this study, we reported the outcomes for EBRT as the primary treatment group including ADT with EBRT, so reflects the real-world experiences among men receiving EBRT according to current best practice.Another limitation is that the type of radical treatment technique was not separately compared,

(
PCOR-ANZ).SA-PCCOC collects patient-reported outcome data at baseline and sequential time points and PSA follow-up data beyond the 12 months, whereas PCOR-ANZ only collect patient-reported outcome measures (PROMs) and PSA data at 12 months posttreatment.Currently, SA-PCCOC has enrolled over 19 000 men from both public and private treatment centres and captures 90% of newly diagnosed cases in South Australia.Our study cohort included men with prostate cancer diagnosed between 1 January 2008 and 31 December 2018, who consented to be in the registry.This period was chosen to ensure both currency and adequate follow-up.Men with distant metastatic prostate cancer or no primary treatment recorded were excluded.Missing data for diagnostic PSA, Gleason score and clinical stage were imputed to generate the risk categories, using multiply imputed chained equations, 10 with 10 imputed data sets.Risk category was defined according to National Clinical Cancer Network11 (low risk: International Society of Urological Pathology (ISUP) Grade group 1 and PSA < 10 ng/dL and clinical-stage T1-T2a; intermediate risk: ISUP Grade group 2-3, or PSA ≥ 10-20 ng/dL, or stage T2b-T2c; and high risk: ISUP Grade group 4-5, or PSA > 20 ng/dL, or stage T3a-T4).Primary treatment was defined as the initial or most definitive treatment within 12 months of diagnosis.Treatment groups included RP, external beam radiotherapy with or without ADT (EBRT +/-À ADT), brachytherapy (high dose and low dose), ADT alone, active surveillance and watchful waiting.While active surveillance involves close ongoing monitoring and subsequent treatment by radiation or surgery with the intent to cure, watchful waiting involves occasional monitoring and treatment for symptoms if they arise, usually by hormone therapy, without the intention to cure the cancer.Men who had (neo)adjuvant ADT (n = 488), that is, those who had EBRT within 12 months of starting ADT, were grouped with EBRT.Covariates included age at diagnosis, year of diagnosis and socioeconomic advantage.Area-level measures of socio-economic advantage were derived from the 2016 Australian Bureau of Statistics socio-economic index for areas scores, based on an individual's residential postcode. 12The following outcomes, which are endorsed by the International Committee on Health Outcome Measure Standardisation 13 were assessed: (1) overall survival; (2) prostate cancer-specific survival; (3) biochemical recurrence following radical therapies; (4) transition to active treatment following active surveillance; and (5) clinically significant changes in physical functioning in relation to sexual dysfunction, urinary incontinence, urinary obstruction, bowel symptoms and hormonal symptoms 12 months after treatment.
This study was ethically approved by University of South Australia Human Research Ethics Committee (protocol: 203716) and the Southern Adelaide Clinical Human Research Ethics Committee (LNR/22/SAC/10).Prostate cancer outcomes were assessed using data from SA-PCCOC.Established in 1998, SA-PCCOC is a diseasespecific, population-wide multi-institutional registry that prospectively collects diagnostic, treatment, clinical outcome data and patientreported functional outcomes for men with prostate cancer in South Australia.SA-PCCOC contributes core data to the binational Prostate Cancer Outcomes Registry for Australia and New Zealand Treatment method stratified by risk group from 2008 to 2018, SA-PCCOC registry.Probability of 5-year and 10-year median survival for the average 68-year-old man diagnosed with nonmetastatic prostate cancer, stratified by treatment type and risk category, SA-PCCOC registry (men diagnosed 2008-2018).
T A B L E 2Note: Separate models (subgroup analyses) were executed for each risk category adjusted for age at diagnoses, year of diagnosis and socio-economic advantage.Abbreviations: ADT, androgen deprivation therapy alone; AS, active surveillance; BT, brachytherapy (low dose and high dose combined); CI, confidence interval; EBRT+/À ADT,external beam radiation therapy with or without ADT; NA, not applicable; RP, radical prostatectomy; WW, watchful waiting.a Time from primary treatment to either death or end of follow-up.b Diagnostic PSA, Gleason score and clinical stage were imputed to generate risk category.c There were insufficient observation in the high-risk group to estimate prostate cancer-specific survival.d Probability of developing biochemical recurrence for the average 68-year-old man diagnosed with nonmetastatic prostate cancer, stratified by treatment type and risk category, SA-PCCOC registry (men diagnosed 2008-2018).Prediction models were executed for each risk category adjusted for age at diagnoses, year of diagnosis and socio-economic advantage.Abbreviations: BT, brachytherapy; CI, confidence interval; EBRT+/-ADT, external beam radiation therapy with or without ADT; NA, not applicable due to insufficient sample for meaningful analyses.Probability of having clinically significant decline in functional outcomes 1 year after treatment for the average 68-years-old man diagnosed with nonmetastatic prostate cancer, stratified by treatment type, SA-PCCOC registry (men diagnosed 2008-2018).Note: Decline in EPIC-26 functional outcome scores at 12 months (from baseline score) were set at 10-point score differences for sexual function, 6-point for urinary continence, 5-point for urinary irritation/ obstruction domain and 4-point for bowel and hormonal functions each.Prediction models were adjusted for inverse probability weighting, risk category, age at diagnoses, year of diagnosis and socio-economic advantage.The analyses were not stratified by risk category due to inadequate observations for meaningful analyses.Men on active surveillance could have been switched to secondary treatments, and this has not been accounted in the analysis.
Favourable survival outcomes in a contemporary cohort are likely due to both earlier diagnosis and advances in treatment and T A B L E 3 a Time from primary treatment to either last follow-up date or the date when the PSA test result defining the BCR was identified.b Diagnostic PSA, Gleason score and clinical stage were imputed to generate risk category.T A B L E 4 a b Decline in urinary irritation domain score implies worse urinary irritative/obstructive problems.