Background
About 127,106 patients worldwide are diagnosed with prostate cancer (PCa) annually, accounting for 7.1% of all cancers diagnosed [
1]; and it is the most common malignant tumor in the USA [
2]. Among men diagnosed with PCa, approximately 20%–30% of patients are grouped as high-risk PCa [
3], which is more likely to progress and relapse [
4]. To date, radiotherapy (RT) plus androgen deprivation therapy (ADT) has still been the standard treatment for high-risk PCa. In several randomized controlled trials (RCT), RT plus ADT showed better survival benefit than single treatment (RT or ADT alone) [
5‐
10]. Although the level of evidence is low, increasing population-based evidence in recent years has suggested that radical prostatectomy (RP) could provide similar or better survival benefit than RT-based systemic therapy [
11‐
15].
Now, both RT and RP are recommended by current guidelines for patients with high-risk PCa [
16]. However, as no large RCT has directly compared the two treatments in high-risk PCa settings, the optimal treatment for this population remains a debate, and selection of patients to receive proper therapy is still an unsettled question. Previous meta-analyses have tried to compare the efficacy of RP and RT in patients with high-risk PCa [
17,
18]; however, they failed to perform detailed subgroup analyses for patients with high-risk PCa, such as when patients had different levels of Gleason score (GS) and T stage, or when patients received different types of RT. In fact, it is also unclear whether these differences would affect the comparison between RP and RT. Additionally, limited information was provided by these previous meta-analyses due to inappropriate statistical methods and rough analyses.
Thus, with increasing literature on this topic, we updated this systematic review and meta-analysis to compare the effectiveness of RP and RT in patients with high-risk PCa and select candidates for optimal treatment.
Materials and methods
Protocol and searching strategy
This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines [
19]. A protocol was developed to define the search strategy and the review was registered on the PROSPERO of the Centre for Review and Dissemination (CRD42019132967). EMBASE (1947 to July 2019), PubMed (1966 to July 2019), and the Cochrane Library database (1948 to July 2019) were searched for relevant studies. We also searched relevant journals and reviews for additional articles. Detailed search strategies and keywords can be found in the protocol.
Inclusion and exclusion criteria
Inclusion criteria included (a) patients with high-risk PCa: National Comprehensive Cancer Network (NCCN) criteria (≥ T3 or GS 8–10 or PSA > 20), D’Amico criteria (≥ T2c or GS 8–10 or PSA > 20) or the other criteria; (b) patients who received RP or RT as primary treatment, and RT including external beam radiotherapy (EBRT), brachytherapy (BT) or combined RT (EBRT+BT); (c) articles which reported survival outcomes or disease control using hazard ratios (HRs) to present the results of comparison, or articles which reported quality of life (QoL); and (d) studies published in English.
Exclusion criteria include (a) patients with metastatic disease; (b) patients with any disease incompatible with the planned treatment; (c) review, editorial or case report; and (d) studies published not in English.
Study selection and data extraction
Two researchers (W.Z.P. and N.Y.C.) reviewed titles, abstracts, and then full texts to determine the final included studies. The two reviewers independently collected and checked the data from those included studies. For each included study, we extracted information on the first author, publication year, median age, sample size, study design, characteristics of high-risk PCa, comparison of treatments, median follow-up, RT dose, and end-points. Any disagreements or discrepancies were resolved by consultation with a third researcher (C.J.R.).
Quality assessment and publication bias
Two researchers (W.Z.P. and N.Y.C.) independently evaluated the methodological quality of the included cohort studies according to the Newcastle-Ottawa Scale (NOS) [
20]; scores ≥ 7 points were considered as high quality. Patient selection, comparability, and outcomes were assessed to evaluate the quality. RCT was evaluated according to the criteria outlined in chapter 8 of the
Cochrane Handbook for Systematic Reviews of Interventions. Publication bias was evaluated by a funnel plot.
Outcomes
The primary outcomes were survival outcomes, including cancer-specific survival (CSS) and overall survival (OS). CSS was defined as the time from RP/RT until death from PCa. OS was defined as the time from RP/RT until death from any cause.
The secondary outcomes were disease control, including biochemical recurrence-free survival (BRFS), metastasis-free survival (MFS), and clinical recurrence-free survival (CRFS). BRFS was defined as the time from RP/RT until biochemical failure. MFS was defined as the time from RP/RT to metastasis. CRFS was defined as the time from RP/RT until metastasis identified via imaging or biopsy-proven local recurrence. And we chose urinary, sexual, and bowel function as the main evaluation indicators of QoL.
Statistical analysis
The meta-analysis was conducted using Review Manager 5.3 software. The hazard ratio (HR) and corresponding 95% confidence interval (95% CI) were extracted directly from the study reports. If insufficient data were available, supplementary data might be sought directly from the investigators of studies. A fixed-effect model or random-effect model was used for analyses based on heterogeneity among studies. We used the Chi-square and the I-square tests to assess the heterogeneity among the studies. Chi-squared with a P < 0.10 or I-square > 50% was considered as significant heterogeneity. Subgroup analyses were performed according to RT types, GS, and clinical T stage. What we need to pay attention to is that we did not conduct a meta-analysis with secondary outcomes in some subgroups since we were unable to extract enough data from these studies.
Discussion
Currently, both RT and RP are first-line treatments for clinically high-risk PCa patients, and the optimal treatment remains a debate. One small RCT has compared the survival outcomes for patients with T2b-3N0M0 PCa treated with surgery or radiotherapy [
21]. Except for 2 reviews and meta-analysis focused on localized PCa [
43,
44], there were two meta-analyses about high-risk prostate cancer published in 2014 and 2015 [
17,
18], while limited information was provided due to inappropriate statistical methods and rough analyses. Petrelli and colleagues reported a meta-analysis comparing the efficacy of RP and RT in patients with high-risk PCa and demonstrated the superiority of RP [
17]. However, Petrelli and colleagues used odds ratios to present the comparison results, which inevitably ignored the time-to-event outcomes. More recently, a meta-analysis was conducted by Lei and colleagues; they reported that RP brought lower CSM than RT [
18], while this meta-analysis was conducted only based on 3 studies. It was particularly noteworthy that none of these previous meta-analyses did a subgroup analysis according to T stage, GS, or RT types, and thus, no detailed data were available for clinicians to optimize treatment strategies.
In the present study, the most up-to-date data were comprehensively analyzed and we found better survival outcomes for patients treated with RP compared to those who received RT. However, RT was associated with better disease control. Subgroup analyses furtherly showed that similar or even better survival outcomes were associated with RT in patients with high GS, high T stage, or received EBRT + BT.
The better disease control for patients treated with RT was likely due to the wider scope of radiotherapy than that of surgery, which made it possible to eliminate micro-metastases outside the prostate and resulted in improved BRFS and MFS. Moreover, adjuvant ADT in addition to RT could further help control the micro-metastases and delay biochemical relapse. However, the improvement of BRFS and MFS by RT did not convert into superior survival benefit compared to RP. Several potential reasons might explain this phenomenon. First, patients treated with RT were older and harbored more adverse clinicopathological features than those with RP. Thus, it was not surprising that patients with RT had a worse prognosis than men with RP. Secondly, to a large extent, the efficacy of RT was determined by the type and dosage. The modality of RT varied in the included studies and the dosage of RT in several studies was lower than what was recommended by current guidelines. Then, patients could choose salvage RT after receiving RP firstly but those patients who chose RT firstly rarely received salvage RP. Last, RT and ADT have greater toxicity than RP, which might lead to the worse OS.
According to our analysis, the types of RT could affect the survival and progression of patients. In fact, several RCT demonstrated a BRFS benefit to EBRT + BT over EBRT [
45‐
47] and several retrospective studies have reported that EBRT + BT brought better outcome than RP on BRFS [
39,
48,
49] and MFS [
35,
36,
38,
50]. It was not difficult to find that EBRT + BT had strong control over disease progression, which might lead to better CSS for patients treated with EBRT + BT than patients treated with RP. Although EBRT+BT had a better benefit of CSS than RP, this benefit might be neutralized by the increasing other-causes mortality caused by radiotherapy. From our data, we should believe that both RP and EBRT + BT were prior choice than EBRT for patients with high-risk PCa.
As is known, GS is one of the most important prognostic factors [
51] and some researches had shown that patients with GS 9–10 had a particularly aggressive disease [
41,
42]. Patients with GS 9–10 were at higher risk of disease progression. As mentioned above, RT might have superiority over RP on eliminating micro-metastases and ultimately resulted in better disease control. So these results might be combined to explain the improvement of CSS in patients treated with RT. Similar outcomes were also observed in subgroup analyses according to the T stage. Due to data limitations, we can only use ratios to separate two relatively high and low T stage groups. While we might infer that RT might bring better survival benefits than RP in patients with higher T stage. The inspiration for our clinicians was that RP might be less appropriate and RT might be the first choice for patients with high T stage or high GS.
With the development of treatment modalities, more patients with high-risk PCa can maintain a stable condition for a long term or even be cured. However, making an optimal treatment decision is not only pursuing maximal survival benefit but also a better health-related quality of life. Thus, the assessment of treatments on QoL is also critical for decision making. RP had better performance when considering QoL in the bowel domain, while RT was associated with better QoL in urinary and sexual domains. In clinical practice, younger patients with high-risk prostate cancer who have a greater need for retention of sexual and urinary function after treatment could be recommended with RT. Moreover, RP might be more suitable for patients who need a better bowel function.
Although our research was the most up to date and we did a lot of subgroup analysis, this study still had some limitations. First, heterogeneity was relatively high due to inconsistent inclusion criteria and different treatment modalities. Second, because of the limited data, the population ratio was used to subdivide studies for analysis. Third, in some subgroups, the number of studies and patients is relatively small. Fourth, we failed to perform a meta-analysis on QoL and only presented the results of these studies. This area requires more research and data. Finally, some people included in the studies would inevitably be partially duplicated as the same database was used. Therefore, those results obtained in these subgroups cannot be strong evidence, but can only be used as a reference for interpreting the results.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.