Prostate cancer patients can benefit from 5-alpha-reductase inhibitor treatment: a meta-analysis

PICOS question

Patient, intervention, comparison, outcome, and study design (PICOS) guided the main question of this review: does 5ARI treatment (Intervention) improve the prognosis (PSA progression, total progression, progression-free survival time and side-effects) (Outcome) of PCa patients (Patient) compared with those who did not receive these treatments (Comparison)?

Search strategy

Potential comparative trials studying treatment effects of 5ARIs for PCa were searched comprehensively using electronic databases, such as the Pubmed, Embase, Medline and Cochrane Library. The last search was in October, 2019. All references of included studies were also carefully checked. Language was restricted to English on eligible studies. We applied the following search terms to obtain original studies: “5alpha-reductase inhibitors” or “5-alpha reductase inhibitors” or “5ARIs” or “finasteride” or “dutasteride” in combination with “prostate cancer” or “prostate carcinoma”. Two authors (Tuo Deng and Xueming Lin) independently performed literature search and screening, quality assessment of eligible studies and data extraction. Disagreements were resolved by discussing with the third author (Guohua Zeng).

Inclusion and exclusion criteria

Studies were included on the basis of the following criteria: (1) Studies should compare the effects of 5ARI and non-5ARI treatments for PCa patients; (2) 5ARIs should be applied after PCa diagnosis; (3) At least one prognostic outcome could be achieved, such as progression, recurrence, metastasis, progression-free survival and so on. Accordingly, studies with participants receiving 5ARIs before PCa diagnosis were excluded from this analysis. Meanwhile, abstracts, case reports, conference proceedings, non-published materials, editorials, reviews, animal experiments and repeated publications were also excluded.

Quality assessment and data extraction

The level of evidence (LOE) of all included studies was assessed by the criteria from the Oxford Centre for Evidence-based Medicine. The quality of RCTs was evaluated using the Cochrane risk of bias tool, and the Newcastle-Ottawa Scale (NOS) was applied for the quality of non-RCTs.

Data from original studies was attentively extracted as follows: first author, year of publication, study country, population, period of recruitment, research methodology, type of PCa, tumor stage, Gleason score, previous therapy, type of 5ARIs, characteristics of participants, detailed 5ARIs treating strategy, follow-up time, and related outcomes. When incomplete data existed, we contacted authors of these studies.

Outcome and statistical analysis

Primary outcomes of this analysis were PSA progression, total progression, progression-free survival time and side-effects. Secondary outcomes included recurrence, metastasis, biopsy reclassification during treatments and therapeutic response. PSA progression was defined as PSA value >20 ng/ml (for subjects who received no previous therapies or underwent primary RT) or >10 ng/ml (for subjects who underwent radical prostatectomy (RP) with or without salvage RT) associated with a 50% increase in PSA from baseline. Total progression referred to patients experiencing any PCa progression, including PSA progression, pathologic progression (having at least one of the predefined criteria: cancer involving ≥3 cores, Gleason pattern score ≥4, or ≥50% of any one core involved), clinical symptoms of progression, and therapeutic progression (defined as necessary post-baseline rescue therapies or medications).

Summarized odds ratios (ORs) with 95% confidence intervals (CIs) were calculated to compare dichotomous variables. Multivariable adjusted hazard ratios (HRs) of 5ARIs were also pooled for predicting progression-free survival in eligible studies. The chi-square test-based Q- and I²- statistic was used to test heterogeneity among studies. If heterogeneity was not significant with a P value > 0.10, the fixed-effect model was used; otherwise, the random-effect was applied. Significant results were considered with a two-sided P value < 0.05. For possible studies, we performed subgroup analyses according to study population, study design, tumor type, Gleason score, previous therapy, and 5ARI type. The publication bias was assessed through the inverted funnel plot visual inspection and Egger’s test. All statistical analyses were done by RevMan (version 5.3; Cochrane Collaboration, Oxford, UK) and STATA (version 13.0; StataCorp., College Station, TX, USA).

Characteristics and quality assessment of eligible studies

Ten studies (Andriole et al., 1995; Banez et al., 2009; Chu et al., 2015; Dai et al., 2018; Dutkiewicz, 2012; Finelli et al., 2011; Fleshner et al., 2012; Ozkan et al., 2018; Ross et al., 2012; Schroder et al., 2013) containing 2,277 PCa patients were involved in this analysis (Fig. 1). Basic characteristics were shown in Table 1. Most studies were conducted in America or Europe. All literature was published between 1995 to 2018. Among them, five were RCTs (Andriole et al., 1995; Chu et al., 2015; Dutkiewicz, 2012; Fleshner et al., 2012; Schroder et al., 2013), one was prospective cohort study (Banez et al., 2009), and other four were retrospective cohort studies (Dai et al., 2018; Finelli et al., 2011; Ozkan et al., 2018; Ross et al., 2012). Most studies focused on local PCa, while metastatic PCa (Dutkiewicz, 2012) and non-metastatic CRPC (Chu et al., 2015) were investigated in only one study respectively. Therapies that PCa patients received prior to recruitments included RP, RT, AS or no treatments. 5ARIs used in these studies were finasteride or dutasteride. In total, 694 patients were allocated to 5ARI group, and 1,583 patients belonged to the control group. Detailed treatment strategies in 5ARI and control groups for each study were also listed in Table 1.

The LOE of all eligible studies was shown in Table 1. For RCTs, four (Andriole et al., 1995; Chu et al., 2015; Fleshner et al., 2012; Schroder et al., 2013) were evaluated as high-quality studies with a low risk of bias according to the Cochrane risk of bias tool. Biases of allocation concealment, incomplete outcome data and selective reporting were also at a low risk (Fig. 2). Table 2 showed the results of quality assessments of five cohort studies. All studies were relatively high in quality with seven or more stars under NOS measurements.

PSA progression

Four studies reported the PSA progression during the follow-up time. There was no heterogeneity among these four studies (I² = 0%, P = 0.54), and no significant difference was found in PSA progression between 5ARI and control groups (OR = 0.82, 95% CI [0.52–1.29], P = 0.40) (Fig. 3).

In the subgroup analyses based on previous therapy, PCa patients receiving previous RP/RT (OR = 0.76, 95% CI [0.47–1.23], P = 0.26) or no treatments (OR = 1.56, 95% CI [0.39–6.16], P = 0.53) did not show significant difference in PSA progression between 5ARI and control groups; and no patient with previous AS treatments were involved in this analysis. In addition, pooled results evaluating the tumor type also indicated no significant difference in all subgroups (local PCa: OR = 0.66, 95% CI [0.37–1.17], P = 0.16; metastatic PCa: OR = 1.56, 95% CI [0.39–6.16], P = 0.53; non-metastatic CRPC: OR = 1.06, 95% CI [0.44–2.53], P = 0.90).

Total progression

Total progression of PCa was detected in nine included studies, and no significant heterogeneity existed (I² = 20%, P = 0.26). Results of meta-analysis revealed that 5ARI treatment significantly reduced the total progression of PCa (OR = 0.61, 95% CI [0.48–0.77], P < 0.0001) (Fig. 4). No publication bias was discovered through either inverted funnel plot (Fig. 5) or Egger’s test (t = −0.52, P = 0.622).

Table 3 showed the results of subgroup analyses for total progression. It seemed that the 5ARIs were effective in decreasing PCa progression regardless of patients’ race, study design, previous therapy, and 5ARI type. However, in subgroups of tumor type and Gleason score, 5ARI treatment proved to be beneficial for patients with local PCa (OR = 0.56, 95% CI [0.44–0.73], P < 0.00001) and low Gleason score (≤7) (OR = 0.63, 95% CI [0.48–0.84], P = 0.002).

Progression-free survival

Multivariable adjusted HR of 5ARI treatment in predicting progression-free survival was available in five studies. Our results demonstrated 5ARIs could significantly prolonged the progression-free survival time in PCa patients (HR = 0.57, 95% CI [0.34–0.96], P = 0.04), with moderate heterogeneity (I² = 69%, P = 0.01) (Fig. 6). The inverted funnel plot did not find any publication bias. In the subgroup analyses based on previous therapy, however, no significant difference was found in neither PCa patients received previous RP/RT (OR = 0.40, 95% CI [0.14–1.14], P = 0.09) nor AS (OR = 0.70, 95% CI [0.32–1.52], P = 0.36) between 5ARIs and control groups. Similarly, no significant difference was found in subgroup analyses based on tumor type (local PCa: OR = 0.56, 95% CI [0.27–1.15], P = 0.11; non-metastatic CRPC: OR = 0.62, 95% CI [0.36–1.07], P = 0.08).

Side-effects

Three studies compared cases with side-effects between 5ARIs and control groups. No significant difference was found in the occurrence of side-effects between two groups (OR = 1.01, 95% CI [0.68–1.50], P = 0.95) with no heterogeneity (I² = 0%, P = 1.00) (Fig. 7). Publication bias was not reached from the inverted funnel plot.

Other outcomes

Table 4 provided summarized results of outcomes explored in less than three studies. No significant differences were found in PCa recurrence, metastasis, biopsy reclassification or partial treatment response between 5ARIs and control groups. However, 5ARIs were still found to be effective in improving the complete treatment response for PCa patients (OR = 71.13, 95% CI [4.01–1262.24], P = 0.004).