Moderate hypofractionated radiotherapy is more effective and safe for localized prostate cancer patients: a meta-analysis

To compare the efficacy and safety of moderate hypofractionated radiotherapy (H-RT) with those of conventional radiotherapy (C-RT) in patients with localized prostate cancer, we conducted extensive literature searches of The Web of Science, Embase, Pubmed and Cochrane Library databases. We identified nine studies with 5969 patients for a meta-analysis. We calculated pooled risk ratios (RRs) and the 95% confidence intervals (CIs) for multiple parameters and performed statistical analysis using RevMan 5.3 software. Our analysis showed that the H-RT group obtained greater improvements in the 5-year biochemical or clinical failure-free survival (RR = 1.04, 95% CI:1.01–1.08; P = 0.01) and 5-year disease-free survival(RR = 1.04, 95% CI: 1.01–1.07, P = 0.02)than the C-RT group. However, the 5-year overall survival rates were comparable in the two groups (RR = 1.02, 95% CI: 0.99–1.04; P = 0.18). Comparison of multiple secondary parameters, including grade 2-4 acute/late gastrointestinal toxicity, grade 2–4 acute/late genitourinary toxicity, biochemical failure, local failure, distant failure and prostate cancer-specific mortality between the H-RT and the C-RT groups showed no statistical differences. This meta-analysis thus indicates that in patients with localized prostate cancer, moderate H-RT exerts a great beneficial effect on the primary parameters than C-RT without enhancing adverse events.


INTRODUCTION
Prostate cancer is the second most common solid tumor diagnosed in older men in the United States, Britain, and Western Europe [1]. The recommended treatment for localized prostate cancer patients is external beam radiotherapy [2]. Although increasing the radiation dose as proposed by the National Comprehensive Cancer Network (NCCN) can improve the control of the disease, increasing the radiation dose using traditional methods can also extend the time of treatment and increase the risk of radiotherapy toxicity [3].
Moderate hypofractionated radiotherapy (H-RT) that uses larger dose radiation treatments (2.2-4 Gy/ fraction) has garnered increasing attention compared to conventional radiotherapy (C-RT) (1.8-2.2 Gy/fraction) due to its low α/β value for prostate cancer(approximately 1.4 (0.9-2.2) Gy) [4,5]. Based on the radiation biology model of prostate cancer, H-RT can improve the treatment without increasing toxicity [6]. Moreover, since H-RT is implemented over a shorter period of time, it is more convenient and cheaper for the patients [7].
In recent years, many RCTs (randomized controlled trials) have focused on using H-RT to treat localized prostate cancer. Some of these studies have shown that moderate H-RT can be more effective for patients with localized prostate cancer without increasing the acute or late toxicities [8]. However, all these trials have not reached consensus in regard to efficacy and safety because they lacked comprehensive evidence [9]. Although two previous studies systematically reviewed and recommended H-RT as the primary management therapy

Research Paper
Oncotarget 2648 www.impactjournals.com/oncotarget for prostate cancer, their data was not comprehensive as they did not include several phase III randomized prospective trials from recent years [10,11]. Therefore, the aim of our study was to ascertain the efficacy and toxicity of moderate H-RT in the treatment of localized prostate cancer by conducting an updated systematic review and meta-analysis.

Quality analysis of selected studies
Each of the nine RCTs with 5969 patients that were included in this meta-analysis underwent quality evaluation based on the Handbook of Cochrane for Systematic Reviews of Interventions: "random" was mentioned in all the studies, all the RCTs reported the basic features of patients, and approaches of allocation concealment were reported in all the trails. However, one of the studies did not describe the method of randomized sequence generation and their reasons for incomplete outcome data was interpreted as selective report bias [16][17][18]. The qualities of the assessed trials in the metaanalysis are shown in Figure 2.

Analysis of 5-year biochemical or clinical failure free (BCFF)
Five of the nine studies reported the 5-year BCFF rate and included 3763 localized prostate cancer patients in total. Compared to the C-RT group, the moderate H-RT group had increased 5-year BCFF rate (RR = 1.04, 95% CI: 1.01-1.08; P = 0.01). Also, fixed-effect model analysis was performed as no statistical heterogeneity was found (Chi² = 3.83, I² = 0%; P = 0.43).This benefit was not different between the two subgroups analyzed (conventional group dose < 70 Gy or ≥ 70 Gy), with a Oncotarget 2649 www.impactjournals.com/oncotarget negative test for subgroup differences (P = 0.17). The complete pooled analysis is shown on Figure 3A.

Analysis of 5-year disease free survival (DFS)
Four trials with 4252 patients were included in this analysis as they reported the 5-year DFS data. Compared to the C-RT group, the H-RT group showed enhanced 5-year DFS rate (RR = 1.04, 95% CI: 1.01-1.07, P = 0.02).The fixed-effect model was used since no statistical heterogeneity was observed (Chi² = 0.02, I² = 0%; P = 1.00). Similar to 5 year BCFF, the benefit was similar between the two subgroups analyzed (conventional group dose < 70 Gy or ≥ 70 Gy), with a negative test for subgroup differences (P = 0.95). Subgroup analysis indicated that the group with conventional dose ≥ 70 Gy was more beneficial to the patients (P = 0.02; Figure 3B)

Analysis of 5-year overall survival (OS)
Five studies that evaluated 5188 localized prostate cancer patients for OS were included in the analysis. The results of the five trials showed that the 5-year OS did not differ significantly between the H-RT and the C-RT groups (RR = 1.02, 95% CI: 0.99-1.04; P = 0.18). This comparison had no statistical heterogeneity (I 2 = 0%, P = 0.94) ( Figure 3C).

Analysis of biochemical failure
Six of the included studies involving 3520 patients with localized prostate cancer reported the results of biochemical failure and were included in the metaanalysis. The C-RT and the H-RT group of patients showed no substantial differences in biochemical failure analysis (RR = 0.87, 95% CI: 0.72-1.07, P = 0.18) and showed a high level of heterogeneity based on the random effect model (P = 0.07, I² = 51%; Figure 4A).

Analysis of local recurrence
The outcomes of local recurrence were reported in 4 studies. Our analysis showed that there was no significant benefit for the H-RT group in comparison to the C-RT group (RR = 0.77, 95% CI: 0.41-1.45, P =0.42) and they showed a low-level of heterogeneity (P = 0.33, I² = 12%; Figure 4B).

Analysis of distant failure
Three trials that reported distant metastasis outcomes were analyzed by the meta-analysis. No significant difference was detected between the C-RT group and the H-RT group (RR = 1.19, 95% CI: 0.70-2.04, P = 0.52).

Analysis of grade 2-4 acute gastrointestinal (GI) toxicity
Six trials with 4529 patients that reported the grade 2-4 acute GI toxicity were included in the meta-analysis. The grade 2-4 acute GI toxicity data were similar between the H-RT and C-RT groups (RR = 1.26, 95% CI: 0.99-1.61; P = 0.06) ( Figure 5A).The random-effect

Analysis of the grade 2-4 late GU toxicity
Six studies reported that included 5049 patients reported the grade 2-4 late GU toxicity results and were analyzed by the meta-analysis. Analysis by the fixed-effect model showed that H-RT did not significantly increase the grade 2-4 late GU toxicity in comparison to the C-RT (RR = 1.10, 95% CI: 0.97-1.24; P = 0.14) ( Figure 6A).

Analysis of prostate cancer-specific mortality
Four studies with 3049 patients that reported prostate cancer-specific mortality were included in the meta-analysis. The fixed-effect model was used to  Figure 6B).

Meta-regression analysis
To investigate the effects of various study characteristics on RR estimates, a meta-regression analysis was evaluated for 5-year BCFF and 5-year DFS rates by grouping the studies according to specific characteristics like the trial year, the mode of radiotherapy, clinical stage, and androgen deprivation therapy (ADT). However, the univariate meta-regression analyses did not detect any association between5-year BCFF, 5-year DFS rates and other characteristics (Tables 2 and 3).

DISCUSSION
Higher doses of radiotherapy have been shown to be more effective in controlling localized prostate cancer [32,33]. This was confirmed by a meta-analysis that included seven RCTs with a total of 2812 patients [34].This study showed that higher doses of radiation were superior to the conventional dose in low-,intermediate-, and highrisk prostate cancer patients and proposed that higher doses of radiotherapy should be offered as a treatment option for all prostate cancer patients regardless of their risk classification. The 2014 NCCN Guidelines recommended a minimum radiation dose of 75.6-79.2 Gy for low-risk patients, and 81 Gy for intermediate-and high-risk patients [35]. However, increasing the external radiation dose without enhancing the single radiation dose would result in increased frequency of therapy that would increase the duration of treatment and cost to the patients [8].
Recently, moderate hypofractionated external beam radiotherapy has gained popularity in the treatment of prostate cancer. Evidence has shown that prostate cancer has a lower α/β ratio compared to the surrounding organs at risk and hence, there is a potential clinical benefit in using moderate hypofractionated radiotherapy [4]. In addition, moderate H-RT has numerous additional advantages, such as reducing treatment time, medical resources, and improving the patient's convenience [5]. Although several trials have demonstrated the safety and efficacy of moderate H-RT compared to C-RT in regard to localized prostate cancer, the outcomes have been inconsistent [9]. Therefore, we undertook this metaanalysis to further evaluate the therapeutic efficacy and safety of moderate H-RT.
We found that the moderate H-RT groups had significantly increased 5-year BCFF (RR = 1.04, 95% CI:1.01-1.08; P = 0.01) and 5-year DFS rates (RR = 1.04,  [27]. Lee and others showed that adverse late grade 2 and 3 GI and GU events increased (HR:1.31 to 1.59) in patients who were treated with H-RT [31]. However, our pooled analysis showed no significant increase in the grade 2-4 acute/late gastrointestinal toxicity and grade 2-4 acute/late genitourinary toxicity. Therefore, we concluded that moderate H-RT in localized prostate cancer was superior to conventional radiotherapy and did not elicit increased toxicity. In 2013, a meta-analysis study found that freedom from biochemical failure was similar in patients receiving either moderate hypofractionated or conventional radiotherapy [10]. Although the incidence of acute adverse gastrointestinal events was higher in the hypofractionated group, acute genitourinary toxicity and other late adverse events were similar in both groups. In 2015, another meta-analysis performed by Koontz and colleagues found similar biochemical control and late grade 2 genitourinary and gastrointestinal toxicities between moderate hypofractionated and conventional radiotherapy [11]. The outcomes of our study are not Oncotarget 2654 www.impactjournals.com/oncotarget consistent with these two previous reports because these studies had limited data and did not include the results from several recently published phase III clinical trials, including the CHHiP [24], HYPRO [30] and RTOG-0415 [31] trials. Moreover, besides the aforementioned outcome measures, our meta-analysis included a total of eleven indicators and provided a thorough comparison between the moderate H-RT group and C-RT group regarding differences in efficacy and toxicities, thus providing better evidence for the clinicians regarding the choice of radiotherapy. However, since the studies analyzed by this meta-study had varying hypofractionation schemes, our study could not determine the optimal hypofractionation scheme, thus limiting our findings.
In conclusion, we present systematic review and meta-analysis showing that moderate hypofractionated radiotherapy could significantly improve the therapeutic effect in patients suffering from localized prostate

Study selection criteria
To compare conventional radiotherapy (with doses per fraction from 1.8 to 2.2 Gy) versus moderate hypofractionated (2.2 to 4 Gy), RCTs with a parallel design with patients suffering from localized prostate cancer without metastases were included. Studies with quasi-randomized, single-arm phase II or non-original, non-randomized trials were excluded from the analysis.

Literature search
Relevant articles for the meta-analysis were identified through searches in the Embase, Pubmed, Cochrane Library and Web of Science databases until June16, 2016. There was no limitation in the electronic literature search regarding the publication year, status or language in the electronic search. The searches were conducted using either Emtree or MeSH terms and were appropriately adjusted in various electronic records. In order to check the potentially qualifying articles, abstracts from the academic meetings mentioned below were included. Besides the electronic search for original papers, a review of the references was conducted.

Data extraction and assessment of the risk of bias
Two independent investigators conducted the literature search and tested its quality. The bias risk of the included studies was evaluated according to the Handbook of the Cochrane for Systematic Reviews of Interventions [36]. A third reviewer was responsible for addressing any disagreements between the original two independent investigators. The studies were classified into unclear, low or high bias risk groups based on the evaluation on the general sequence allocation, allocation concealment, blinding of personnel and participants(performance bias), outcome evaluation blinding (detection bias), partial addressing of the data, presence of biases in the reports and other bias sources that could influence the validity of the research.

Statistical analysis
Statistical analysis was conducted using the RevMan 5.3 software (Nordic Cochran Centre in Copenhagen of Denmark, 2014).Meta-regression analysis was performed with R version 3.3.1 (2016-06-21)and "meta" package. The risk ratio (RR) was assessed and the 95% confidence intervals (CI) were calculated to evaluate the various parameters. Shared heterogeneity among the studies was evaluated using the I² and Chi² tests. We used the fixedeffect model when the data lacked heterogeneity (P > 0.10, I 2 < 50%). Otherwise, the random-effect model was used. Three potential sources of heterogeneity, including statistical, clinical and methodological, were studied.