Radiation therapy compared to radical prostatectomy as first-line definitive therapy for patients with high-risk localised prostate cancer: An updated systematic review and meta-analysis

ABSTRACT Objective To present an update of the available literature on external beam radiation therapy (EBRT) with or without brachytherapy (BT) compared to radical prostatectomy (RP) for patients with high-risk localised prostate cancer (PCa). Methods We conducted a systematic review and meta-analysis of the literature assessing the survival outcomes in patients with high-risk PCa who received EBRT with or without BT compared to RP as the first-line therapy with curative intent. We queried PubMed and Web of Science database in January 2021. Moreover, we used random or fixed-effects meta-analytical models in the presence or absence of heterogeneity per the I2 statistic, respectively. We performed six meta-analyses for overall survival (OS) and cancer-specific survival (CSS). Results A total of 27 studies were selected with 23 studies being eligible for both OS and CSS. EBRT alone had a significantly worse OS and CSS compared to RP (hazard ratio [HR] 1.38, 95% confidence interval [CI] 1.16–1.65; and HR 1.55, 95% CI 1.25–1.93). However, there was no difference in OS (HR 1.1, 95% CI 0.76–1.34) and CSS (HR 0.69, 95% CI 0.45–1.06) between EBRT plus BT compared to RP. Conclusion While cancer control affected by EBRT alone seems inferior to RP in patients with high-risk PCa, BT additive to EBRT was not different from RP. These data support the need for BT in addition to EBRT as part of multimodal RT for high-risk PCa. Abbreviations: ADT: androgen-deprivation therapy; BT: brachytherapy; CSS: cancer-specific survival; HR: hazard ratio; MFS, metastatic-free survival; MOOSE: Meta-analyses of Observational Studies in Epidemiology; OR: odds ratio; OS: overall survival; PCa: prostate cancer; RR: relative risk; RP: radical prostatectomy; RCT: randomised controlled trials; (EB)RT: (external beam) radiation therapy


Introduction
High-risk non-metastatic disease (i.e. PSA level >20 ng/ mL, Gleason score of 8, and/or a clinical stage of T2c-3a) accounts for ~30% of newly diagnosed prostate cancer (PCa) [1,2], depending on its definition [3,4]. The optimal primary treatment for these patients remains unresolved with standard local therapeutic options including radical prostatectomy (RP) and radiation therapy (RT) consisting of external beam RT (EBRT) with or without brachytherapy (BT). Owing to the heterogeneous nature of the published cohort studies and the lack of prospective randomised controlled trials (RCTs), it remains unclear which single and/or multimodal therapeutic strategy is optimal for each patient with high-risk PCa.
In the current and likely future, the absence of propensity designed RCTs comparing RP to RT for patients with high-risk PCa, meta-analyses may help a framework for clinical decision-making and patient counselling. To this end, we performed a systematic review and meta-analysis to summarise the results of available studies including the latest literature on this subject. We focussed on the additive value of BT in addition to EBRT.

Methods
In this meta-analysis, we followed the Meta-analyses of Observational Studies in Epidemiology (MOOSE) statement guidelines that propose a checklist of items which provides a RCTs checklist [5]. Moreover, the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) was used.

Eligibility criteria
The question of this study was, 'Do patients with highrisk PCa who receive EBRT with or without BT have a better survival outcome compared to those who receive RP?'. All current articles covering the study question were eligible for this systematic review. We selected studies that perform quantitative synthesis according to the similarity in Population, Interest, Context (PICO) elements to decrease the selection bias and heterogeneity. The inclusion criteria for the quantitative meta-analysis were original research articles that assessed survival outcomes and reported an estimated risk effect (hazard ratio [HR], odds ratio [OR], relative risk [RR]) for both patient and control groups. Exclusion criteria were BT usage alone as definitive therapy and lack of definition of high-risk PCa in the main or subgroup analysis.
Consequently, the more comparable cohort studies according to the MOOSE guidelines were included in the analyses. Furthermore, the heterogeneity of the population was explored by detecting the source and country of databases. According to OS, we categorised studies' outcomes, CSS, biochemical recurrence survival, and metastatic-free survival (MFS).

Information sources
We searched PubMed and the Web of Science for studies published before 1 January 2021. The search queries line and search strategies were '((Prostate OR prostatic) AND (cancer OR carcinoma) OR (Prostatic Neoplasms[Mesh])) AND external beam radiotherapy OR brachytherapy OR radiotherapy [Mesh]) AND (radical prostatectomy [Mesh])' in PubMed and '((Prostate OR prostatic) AND (cancer OR carcinoma)) AND (external beam radiotherapy OR brachytherapy OR radiotherapy) AND (radical prostatectomy)' in the Web of Science. The search results were restricted to English language articles. Two reviewers screened titles and abstracts independently; any disagreement about the articles' eligibility was resolved by Delphi consensus with the co-authors. A data extraction sheet was developed based on the Cochrane Consumers and the Communication Review Group's (http://cccrg.cochrane.org/author-resources). We extracted the following data: first-author, type of article, year of publication, dates of the data collection or enrolment, cohort type, sample size, number of individuals on treatment, outcome, how the outcome was measured, type of effect statistic, effect statistic error measures, and effect statistic P value. There were no limitations in the articles' data, so we did not need to contact any authors for additional details. Modified Newcastle-Ottawa Scale criteria were used to assess the quality of the included studies [6]. Subsequently, the HRs and 95% CIs with OS, CSS outcomes were retrieved, all discrepancies regarding data extraction were resolved by Delphi consensus with co-authors.

Statistical analysis
Forest plots were used to assess the multivariable HRs. We summarised them to depict the relationship of our outcomes with the type of treatment (i.e. RP, EBRT alone and EBRT plus BT). When HRs and P value only were reported, we calculated the corresponding 95% CIs. We utilised multivariable adjusted or propensity score matched analyses in the quantitative meta-analyses. Studies included in performing the meta-analyses were adjusted for the effects of age, clinical T stage, Gleason grade, and PSA. The primary meta-analysis was performed for all studies that reported OS as an outcome. A secondary meta-analysis was conducted using studies that reported CSS as an outcome. The next four metaanalyses were conducted among studies that reported the risk of OS and CSS according to the type of RT (i.e. EBRT alone or EBRT plus BT). Heterogeneity across the studies was appraised using P values, Q and I 2 statistics [7]. In the presence of statistically significant heterogeneity (˃50%), random effect meta-analysis was used. When there was no significant heterogeneity observed, the fixed-effect model was used. Funnel plots was used to detect the risk of publication bias. Statistical analyses were considered significant if the P value was <0.05. All analyses were carried out using Stata version 14 (Stata Corp., College Station, TX, USA).

OS and CSS for patients with high-risk PCa who received RT compared to RP
In the first meta-analysis of OS, 23 were included. We found that patients with high-risk PCa who received RT (regardless of the type of RT) had significantly worse OS than those treated with RP with a HR of 1.27 (95% CI 1.11-1.45; Figure 2(a)). The 23 studies included in the meta-analysis demonstrated a high heterogeneity (I 2 = 84.5%, P < 0.001), so a random-effect model was used. The funnel plot was asymmetrical (Figure1S-A in supplemental data). In the second meta-analysis, 23 studies were again included. We found that patients with high-risk PCa who received RT (regardless of the type of RT) had significantly worse CSS than those treated with RP with a HR of 1.37 (95% CI 1.15-1.65; Figure 2(b)). The 23 studies included in the meta-analysis showed a high heterogeneity (I 2 = 84.6%, P < 0.001), so a random-effect model was used. The funnel plot was asymmetrical (Figure1S-B in supplemental data).

OS and CSS for patients with high-risk PCa who received only EBRT compared to RP
In the first subgroup meta-analysis, 15 studies were included. We found that patients with high-risk PCa who received EBRT alone (i.e. without a concomitant BT) had significantly worse OS than those treated with RP with a HR of 1.38 (95% CI 1.16-1.65; Figure 3(a)). The 15 studies included in the meta-analysis demonstrated a high heterogeneity (I 2 = 81.2%, P < 0.001), so a random-effect model was used. The funnel plot was slightly asymmetrical (Figure1S-C in supplemental data). In the second subgroup meta-analysis, 18 studies were included. We found that patients with high-risk PCa who received EBRT alone had significantly worse CSS than those treated with RP with a HR of 1.55 (95% CI 1.25-1.93; Figure 3(b)). The 18 studies included in the meta-analysis showed a high heterogeneity (I 2 = 85.6%, P < 0.001), so a random-effect model was used. The funnel plot was slightly asymmetrical (Figure1S-D in supplemental data).

OS and CSS for patients with high-risk PCa who received EBRT plus BT compared to RP
In the third subgroup meta-analysis, eight studies were included. The HR of 1.1 (95% CI 0.76-1.34) suggested that there was no difference in OS between patients with high-risk PCa who received EBRT plus BT compared to those who underwent RP (Figure 4(a)). The eight studies included in the meta-analysis demonstrated a high heterogeneity (I 2 = 92.8%, P < 0.001), so a random-effect model was used. The funnel plot was asymmetrical (Figure1S-E in supplemental data). In the fourth subgroup meta-analysis, seven studies were included. The HR of 0.69 (95% CI 0.45-1.06) suggested no difference in CSS between patients with high-risk PCa who received EBRT plus BT and those who underwent RP; however, statistical significance was not reached (Figure 4(b)). The seven studies included in the meta-analysis showed moderate heterogeneity (I 2 = 84.9%, P = 0.096), so a random-effect model was used. The funnel plot was slightly asymmetrical (Figure1S-F in supplemental data). Table 3 [9,10,[12][13][14]16,17,19,29] shows the reported data regarding RT and RP components among studies included in this systematic review.

Discussion
The present systematic review and meta-analyses assessed the comparative survival effectiveness of RT and RP as a definitive therapy with curative intent of PCa patients with high-risk features. Although we found that RT irrespective of RT type (i.e. EBRT alone or plus BT) resulted in significantly worse OS and CSS compared to RP, patients who underwent a EBRT plus BT combination had OS and CSS that were not inferior to RP. However, EBRT alone (i.e. without concomitant BT) was inferior to RP with regards to OS and CSS. While there is no direct, well designed comparison of RP vs RT as the first step in a multimodal therapeutic concept in concordance with our metaanalyses, cumulative data support the concept of combining maximal RT consisting of EBRT plus BT with androgen-deprivation therapy (ADT) as the firstline multimodal strategy for therapy of high-risk PCa [10,12,16,17,19,29]. Indeed, major guidelines recommend ADT combined with EBRT plus BT, based on the OS benefits shown in several RCTs that compared it to EBRT plus ADT [35]. There is to date no comparison between the different multimodal therapies (i.e. EBRT plus BT and ADT) vs RP with RT vs RP with ADT, etc. Some single institution and small cohorts included in this systematic review tried to conduct a fair comparison between those strategies (i.e. RP plus adjuvant/ salvage RT and ADT) [17,19,36,37]; however, the inherent selection bias limits any fair comparisons (Table 2), this is especially true for in multiinstitutional and population-based datasets and registries, while they are at least likely to suffer from systematic bias based on their geographic and specialty representativeness.
Today, RT as adjuvant or salvage strategy is supported as a part of a multimodal therapy after RP by major guidelines for patients with high-risk PCa based on data recruiting from several RCTs [38][39][40]. However, more studies that compared EBRT plus BT to RP have not reported postoperative RT usage, therefore making fair comparisons impossible [9,10,12,14,16]. Limitations for optionally combining BT with EBRT include dose distribution and prostate size, reflecting a selection bias that may reflect the local tumour burden. In contrast to most cohort studies that assessed EBRT (i.e. without a concomitant BT) vs RP for high-risk disease, we found a significantly worse OS and CSS [9,11,18,22]. Finally, until well-designed RCTs assess survival outcomes between MaxRT and MaxRP, a multidisciplinary approach should be considered in treating patients with high-risk PCa beyond the results of survival outcomes of cohort studies.
The main limitation of the present systematic review and meta-analysis was the lack of well-designed controlled trials. However, because of this lack, we believe that this systematic review and meta-analysis might help frame and equipoise decisions to guide patients' counselling as part of the shared decision process [5]. Another limitation was the heterogeneity across studies regarding the age of included patients, usage of ADT and its ADT duration, usage of adjuvant or salvage therapies such as postoperative RT after RP. Moreover, there was no precise data regarding the number of ADT patients, making subgroup analysis impossible. Indeed, designing a cohort study considering and adjusting for all variables' effects is unlikely to be possible/to be performed. On the other hand, it is clear that the included multi-institutional cohorts suffer from a significant selection bias that limits the validity of the findings assessing from the former. Moreover, the effect of delayed definitive therapy due to the coronavirus disease 2019 (COVID-19) pandemic should be considered in future studies [41,42].

Conclusions
According to this systematic review and metaanalyses, patients with high-risk PCa who received EBRT alone (without a concomitant BT) as a first-line definitive therapy had worse OS and CSS than those who underwent RP. However, EBRT plus BT as a multimodal RT was not inferior to RP in high-risk PCa. These data support the need for a multimodal strategy to achieve optimal therapy in high-risk PCa.
RT as a primary definitive strategy could be better when combined with EBRT and BT to achieve maximal radiation dose combined with ADT. The role, timing, indication of postoperative RT after RP, and the type and duration of ADT for each patient need assessment and evidence. Until then, we would postulate based on the findings of our present study an equipoise of EBRT with BT for local control as a part of multimodal flexible and dynamic treatment strategy tailored to each tumour in each patient.

Disclosure statement
No potential conflict of interest was reported by the author(s).