Objectives To evaluate efficacy and safety of gonadotropin-releasing hormone (GnRH) antagonists compared to standard androgen suppression therapy for advanced prostate cancer.
Setting The international review team included methodologists of the German Cochrane Centre and clinical experts.
Participants We searched CENTRAL, MEDLINE, Web of Science, EMBASE, trial registries and conference books for randomised controlled trials (RCT) for effectiveness data analysis, and randomised or non-randomised controlled studies (non-RCT) for safety data analysis (March 2015). Two authors independently screened identified articles, extracted data, evaluated risk of bias and rated quality of evidence according to GRADE.
Results 13 studies (10 RCTs, 3 non-RCTs) were included. No study reported cancer-specific survival or clinical progression. There were no differences in overall mortality (RR 1.35, 95% CI 0.63 to 2.93), treatment failure (RR 0.91, 95% CI 0.70 to 1.17) or prostate-specific antigen progression (RR 0.83, 95% CI 0.64 to 1.06). While there was no difference in quality of life related to urinary symptoms, improved quality of life regarding prostate symptoms, measured with the International Prostate Symptom Score (IPSS), with the use of GnRH antagonists compared with the use of standard androgen suppression therapy (mean score difference −0.40, 95% CI −0.94 to 0.14, and −1.84, 95% CI −3.00 to −0.69, respectively) was found. Quality of evidence for all assessed outcomes was rated low according to GRADE. The risk for injection-site events was increased, but cardiovascular events may occur less often by using GnRH antagonist. Available evidence is hampered by risk of bias, selective reporting and limited follow-up.
Conclusions There is currently insufficient evidence to make firm conclusive statements on the efficacy of GnRH antagonist compared to standard androgen suppression therapy for advanced prostate cancer. There is need for further high-quality research on GnRH antagonists with long-term follow-up.
Trial registration number CRD42012002751.
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Strengths and limitations of this study
We searched CENTRAL, MEDLINE, Web of Science, EMBASE, trial registries and conference books. Two authors independently screened identified articles, extracted data, evaluated risk of bias and rated quality of evidence according to GRADE.
There were no statistically significant differences in overall mortality, treatment failure, or prostate-specific antigen progression and no study reported cancer-specific survival or clinical progression.
Quality of evidence for all assessed outcomes was rated low according to GRADE. Available evidence is hampered by risk of bias, selective reporting and limited follow-up.
The question that was addressed by this systematic review was in some points different from the available evidence. There is currently insufficient evidence to make firm conclusive statements on the efficacy of GnRH antagonist compared to standard androgen suppression therapy for advanced prostate cancer and there is a need for further high quality research on GnRH antagonists with long-term follow-up.
Gonadotropin-releasing hormone (GnRH) antagonists, such as abarelix or degarelix, are new agents for androgen suppression therapy in advanced prostate cancer. They act by competitively binding to receptors in the pituitary gland, leading to reduced amounts of luteinising hormone and follicle-stimulating hormone. GnRH antagonists are, thereby, able to decrease the level of testosterone immediately to castration levels without flare.1–3 Testosterone is important for the growth of prostate cells and its suppression slows down the disease progression and leads to a decrease in prostate-specific antigen (PSA).
Data from published randomised controlled trials support the use of degarelix as an alternative to standard androgen suppression therapies.4 ,5 Abarelix appears to be equally effective.2 ,6 Androgen suppression therapy with degarelix may also be more cost-effective in patients with locally advanced prostate cancer7–9 and may increase PSA progression-free and overall survival.5 ,10 Additionally, degarelix might also have beneficial effects on lower urinary tract symptoms.11 Furthermore, GnRH antagonists might provide an alternative to castration in symptomatic patients with advanced prostate cancer because there is no risk for testosterone flare associated with GnRH agonists that might aggravate clinical symptoms.12 Despite these positive findings, the current European guideline indicate that there is no definitive evidence that GnRH antagonists have advantages over GnRH agonists.13
An analysis of pooled individual patient data of five randomised clinical trials found clinical benefits with degarelix compared with GnRH agonists.10 However, no systematic review based on a comprehensive literature search using predefined methodology have yet evaluated the efficacy and tolerability of GnRH antagonists in comparison with standard androgen suppression therapy for advanced prostate cancer. Therefore, the objectives of this systematic review are to determine the efficacy and safety of GnRH antagonists compared with standard androgen suppression therapy for advanced prostate cancer treatment.
For details on our predefined methodology and outcomes see the prospective registry entry in the ‘International Prospective Register of Systematic Reviews’ (http://www.crd.york.ac.uk/PROSPERO;CRD42012002751).
We included studies that compared GnRH antagonists (abarelix and degarelix) with standard androgen suppression therapy in patients with advanced prostate cancer. Included studies had to be randomised controlled trials (that were used for efficacy and safety analysis) or prospective non-randomised controlled studies (that were used for adverse events and quality of life analysis). If randomised controlled trials were identified with cross-over design, we only included the data just before cross-over started. We did not exclude studies because of publication status or language of publication, nor did we make restrictions based on age or ethnicity of patients.
We included all patients with advanced prostate cancer. Advanced disease was defined as either locally advanced (T3-4, N0, M0), local to regionally advanced (T1-4, N1, M0), disseminated disease (T1-4, N0-1, M1) or PSA relapse after local therapy.
Included studies had to compare GnRH antagonists (abarelix or degarelix) with standard androgen suppression. The standard androgen suppression therapy included monotherapy with surgical or medical castration, antiandrogen monotherapy or maximal androgen blockade (combination of either surgical or medical castration with antiandrogens).
Our prospectively defined primary outcomes were overall survival and adverse events. We defined cancer-specific survival, clinical or PSA progression, treatment failure and quality of life as secondary outcomes. No study was excluded solely because the outcome of interest was not reported.
Unit of analysis was the study rather than publications, and we named the studies according to their study identification numbers assigned by the sponsors. We used the sponsors identification numbers for differentiation because several authors were involved in more than one study, publications were identified reporting information on several studies (pooled analyses of individual patient data of five randomised controlled trials: CS21, CS28, CS30, CS31, CS35), and as there were several publications available for some studies (eg, different follow-up time or reporting different outcomes).
We searched the Cochrane Library (CENTRAL, Issue 3, 2015), MEDLINE (via Ovid; 1946 to March 2015), Web of Science (Thomson Reuters Web of Knowledge; 1970 to March 2015), and EMBASE (via DIMDI; 1947 to March 2015) databases. For details on the search strategy, see table 1.
Additionally, we searched three trial registries: Current Controlled Trials (ISRCTN; http://www.controlled-trials.com/; last search in March 2015), ClinicalTrials.gov (http://www.clinicaltrials.gov/; last search in March 2015), and the WHO International Clinical Trials Registry Platform Search Portal (WHO ICTRP Search Portal; http://www.who.int/ictrp/en/; last search in March 2015). We used the following keywords for this search: ‘abarelix’, ‘degarelix’, ‘plenaxis’, ‘firmagon’.
We also searched the electronically available abstract books from three major conferences: American Society of Clinical Oncology (ASCO; jco.ascopubs.org; 2004 to March 2015), European Association of Urology (EAU; http://www.uroweb.org; 2004 to March 2015), and American Urological Association (AUA; http://www.jurology.com/; 2008 to March 2015). We used the following keywords for this search: ‘abarelix’, ‘degarelix’, ‘plenaxis’, ‘firmagon’.
Furthermore, reference lists of retrieved articles were also searched manually. We also used the safety data analyses from the websites of the Food and Drug Administration (FDA), and the European Medicines Agency (EMEA) to obtain additional information on studies that included patients treated with GnRH antagonists.
The search of all databases was initially conducted in March 2014 and was updated in March 2015. The search update included only studies that were published since our initial search (studies published from March 2014 to March 2015). No language restrictions were applied.
Two authors independently screened retrieved references for inclusion (FK, HB), and two authors (FK, AB) independently extracted data using standardised data extraction forms and assessed each study's risk of bias. We resolved any disagreements through double-checking the respective articles, or through discussion with a third review author (JM). One review author performed the search update (FK). Randomised studies’ risk of bias was assessed following the recommendations of the Cochrane Handbook by Higgins et al.14 We used the checklist recommended by Reeves et al15 for data collection and study assessment for non-randomised studies.
We used the Cochrane RevMan V.5.2 for statistical data analyses (http://tech.cochrane.org/revman/), and the GRADE working group's software GRADEpro to develop the GRADE evidence table (http://www.gradeworkinggroup.org/).16 ,17 We identified no studies evaluating time-to-event outcomes. Therefore, no HRs were extracted.
We extracted outcomes data relevant to this review, as needed, for calculation of summary statistics and measures of variance. For dichotomous outcomes, we attempted to obtain numbers of events and totals to calculate pooled risk ratios (RRs) with their 95% CIs using Mantel-Haenszel method. Continuous outcomes were analysed using the inverse variance method and were expressed as mean differences (MD) with 95% CI. We defined p<0.05 as statistically significant. We assessed statistical heterogeneity among studies (χ2, I2) and employed a fixed effects model for I2≤50% and additionally, a random effects model for I2>50% for use in a sensitivity analysis.
We performed subgroup analyses for the different doses of androgen suppression therapy and for the different GnRH antagonists (abarelix and degarelix). Initially, we also planned to perform subgroup analyses for non-metastatic versus metastatic disease. However, results were not reported for these subgroups in the included studies.
We identified 15 studies but only 13 (10 randomised and 3 non-randomised controlled trials) were included in this review. Two of the three non-randomised studies were cross-over studies (Zuckerman 2013, Garnick 2011). See figures 1 and 2 for details regarding the literature search.
Abarelix depot 100 mg intramuscularly administered on day 0, day 15 and every 4 weeks thereafter was evaluated in six studies:
Seven studies evaluated degarelix 240 mg subcutaneously administered as a starting dose, and 80 or 160 mg subcutaneous maintenance doses every 4 weeks thereafter:
The two excluded studies were retrieved from the FDA website (149-01-03 and 149-01-05). We identified no publications regarding these studies and were, therefore, not able to include the studies in our analyses because we found no further methodological information or study results. Study 149-01-03 was an open-label trial that compared neoadjuvant hormonal therapy with abarelix depot 100 mg intramuscularly with leuprolide depot 7.5 mg intramuscularly in patients with prostate cancer who planned to undergo brachytherapy or external-beam radiation therapy.23 Study 149-01-05 was an open-label cross-over study to evaluate the feasibility of switching to treatment with a GnRH agonist following 12 weeks of treatment with abarelix in patients with prostate cancer.23
The 13 included studies resulted in 55 citations (16 full journal publications, 34 abstracts, and 5 other data sources). Two studies were published as conference abstracts or only within the meta-analysis of several studies (CS35, CS37), one in conference abstracts (149-99-03), and one study as a conference abstract, in FDA safety data publications or within narrative reviews (ABACS1). We did not identify journal publications that reported details of the methodology for any of these studies.
We did not identify any active controlled study with follow-up beyond 1 year. There are publications available for an extension of study CS21, which reports on outcomes with longer follow-up.4 ,73–76 However, randomisation was rescinded in study CS21 after 1 year of follow-up because all patients were switched from GnRH agonist intervention to GnRH antagonist treatment. Thus, after 1 year of follow-up, this study became an observational study without a control group, and results from this extension phase were not included in this systematic review. Study characteristics of the included studies are presented in tables 2 and 3.
Risk of bias
Two trials were terminated early (CS28, CS35). Regarding randomised controlled trials, there was adequate information on random sequence generation in only one study (CS21), and on allocation concealment in four studies (CS21, 149-98-02, 149-98-03, 149-99-03). All studies included were open-label trials. Study results for adverse events, treatment failure and quality of life are, therefore, likely to be influenced by lack of blinding. Two studies did not report the dose of GnRH agonist and the number of patients per group included (CS35, CS37). In six studies (CS28, CS31, CS35, CS37, 149-99-03, ABACS1), there was insufficient reporting of attrition and exclusions to permit judgement on incomplete outcome data. One study did not report Gleason score (149-99-03), and four studies did not report either Gleason score or disease stage (ABACS1, 149-97-04, CS35, CS35).
All of the 10 randomised and 3 non-randomised controlled trials provided data on adverse events. However, in five studies, several adverse events were reported incompletely and therefore, could not be entered into our meta-analysis (CS28, CS35, CS37, ABACS1, Zuckerman 2013). There was no wash-out period between the different interventions of the two included cross-over studies (Zuckerman 2013, Garnick 2011).
Information on mortality presented as time-to-event data was not provided by a single study. Therefore, we could not, as initially planned, analyse these data with HRs, but had to report number of deaths during the study duration. After screening the available entries of the study protocols in the registries, mortality was not predefined as primary or secondary outcome in any of the included studies, but was only assessed as an adverse event outcome.
Nine studies reported the number of patients who had died during study conduct (149-98-02, 149-98-03, ABACS1, CS21, CS28, CS30, CS31, CS35, and CS37). There were no statistically significant differences in deaths between GnRH antagonists and standard androgen suppression therapy (RR 1.35, 95% CI 0.63 to 2.93, 9 studies with 3020 patients included), nor in the subgroup analyses of abarelix or degarelix compared with standard androgen suppression therapy (abarelix 100 mg: RR 3.49, 95% CI 0.77 to 15.83, 3 studies with 697 patients included; degarelix 240/80 and 240/160 mg: RR 1.00, 95% CI 0.52 to 1.92, 6 studies with 2323 patients included; figure 3). Quality of evidence for this outcome was rated low due to study limitations and imprecision according to GRADE (table 7).
All included studies reported PSA levels, and seven studies reported PSA progression (ABACS1, CS21, CS28, CS30, CS31, CS35 and CS37). Only study CS21 was planned to evaluate time to PSA progression that was defined as two consecutive increases in PSA of 50% compared with nadir and ≥5 ng/mL on two consecutive measurements, at least 2 weeks apart.37 We did not identify a definition for PSA progression for the other studies, and the analyses for PSA progression might be of a post-hoc nature. There was no statistically significant difference in PSA progression between GnRH antagonists and standard androgen suppression therapy (RR 0.83, 95% CI 0.64 to 1.06, 7 studies with 2489 patients included; subgroup abarelix: RR 1.05, 95% CI 0.41 to 2.66, 1 study with 176 patients included; degarelix 240/80 mg and 240/160 mg: 0.81, 95% CI 0.62 to 1.05, 6 studies with 2313 patients included). We performed post-hoc subgroup analyses for patients treated with degarelix and different baseline PSA levels. There were no statistically significant differences for patients treated with different regimens of degarelix, that is, 240/80 mg or 240/160 mg and PSA ≤50 ng/mL (PSA<20 ng/mL: RR 9.10, 95% CI 0.52 to 159.00, 6 studies with 1399 patients included; PSA ≥20–50 ng/mL: RR 0.81, 95% CI 0.34 to 1.90, 6 studies with 401 patients included). GnRH antagonists decreased PSA progression in patients with baseline PSA levels >50 ng/mL compared with standard androgen suppression therapy (RR 0.74, 95% CI 0.56 to 0.98, 6 studies with 513 patients included). Quality of evidence was rated low due to study limitations and imprecision according to GRADE (table 7).
Seven studies reported treatment failure (149-98-02, 149-98-03, 149-99-03, CS21, CS28, CS30 and CS31). No statistically significant differences were observed between GnRH antagonists and standard androgen suppression therapy (RR 0.91, 95% CI 0.70 to 1.17, 7 7 studies with 2200 patients included). While subgroup analyses demonstrated a favourable effect for abarelix compared with standard androgen suppression therapy (RR 0.66, 95% CI 0.45 to 0.98, 3 studies with 1110 patients included), there was no significant difference for degarelix compared with standard therapy (degarelix 240/80 mg: RR 1.03, 95% CI 0.65 to 1.63, 4 studies with 782 patients included; degarelix 240/160 mg: RR 1.33, 95% CI 0.79 to 2.24, 1 study with 308 patients included). Quality of evidence was rated low due to study limitations and imprecision according to GRADE (table 7).
At variance with the prespecified outcomes in our protocol, we also included the outcome ‘failure to achieve or maintain castration’. Castration was defined as no testosterone value >50 ng/mL under androgen suppression therapy. Five studies provided data (149-98-02, 149-98-03, 149-99-03, ABACS1, and CS21). We identified a statistically significant difference in favour of standard androgen suppression therapy (RR 1.80, 95% CI 1.37 to 2.35, 5 studies with 1889 patients included). However, statistically significant differences did not persist after using the random effects model for heterogeneity (I2=60%; RR 1.53, 95% CI 0.95 to 2.49, 5 studies with 1889 patients included). Therefore, the overall effect on this outcome remains unclear. Subgroup analyses showed that abarelix increased the failure to achieve or maintain castration, while there was no significant difference between degarelix and standard therapy (abarelix: RR 1.88, 95% CI 1.19 to 2.97; 4 studies with 1279 patients included; degarelix 240/80 mg: RR 0.61, 95% CI 0.17 to 2.22, 1 study with 308 patients included; degarelix 240/160 mg: RR 0.50, 95% CI 0.10 to 2.41, 1 study with 302 patients included).
The data on adverse events are shown in table 8. We did not identify statistically significant differences for the predefined adverse events fatigue, hot flushes, infections, loss of sexual interest, sexual dysfunction, asthenia, urinary retention, diarrhoea, or constipation (table 8).
The risk of injection site pain or reaction significantly increased with GnRH antagonists compared with standard therapy (table 8).
No significant difference in urinary tract infection was observed between the different therapy groups. However, subgroup analysis showed a significant positive effect for degarelix 240/80 or 240/160 mg compared with standard androgen therapy (RR 0.57; 95% CI 0.39 to 0.83, 6 studies with 2328 patients included; table 8).
Cardiovascular events occurred less often with GnRH antagonist (degarelix 240/80 and 240/160 mg) than with standard therapy (RR 0.60, 95% CI 0.38 to 0.94, 6 studies with 2328 patients included; table 8). Given the reduced risk regarding cardiovascular events, we also evaluated further adverse events regarding the cardiovascular system. Post hoc analyses revealed no statistically significant differences regarding acute myocardial infarction or fatal cerebrovascular-related events, but showed that new diagnosis of ischaemic heart diseases occurred significantly less often in patients who were using GnRH antagonists compared with patients on standard androgen suppression therapy (RR 0.42, 95% CI 0.23 to 0.77, 1 study with 610 patients included). This was also seen for the subgroup of patients treated with degarelix 240/80 mg, but not for those treated with degarelix 240/160 mg. Therefore, the effect of GnRH antagonists on these post hoc included outcomes remains unclear. Additionally, it was also unclear if these results are also applicable for patients who already had a history of cardiovascular events because original publications did not report if this was evaluated during the study screening phase or if this was an exclusion criteria.
The risks of experiencing peripheral oedema and musculoskeletal adverse events were decreased using GnRH antagonists compared with standard androgen suppression therapy (RR 0.51, 95% CI 0.32 to 0.81, 2 studies with 520 patients included and RR 0.65, 95% CI 0.45 to 0.96, 1 study with 408 patients included, respectively).
Arthralgia and back pain also occurred less often with GnRH antagonists (table 8). However, this was only seen in the subgroup of patients treated with degarelix (RR 0.66, 95% CI 0.46 to 0.94, 6 studies with 2328 patients included, and RR 0.68, 95% CI 0.48 to 0.99, 6 studies with 2328 patients included, respectively).
Meta-analysis identified that the risk of chills was increased with GnRH antagonists (RR 9.38, 95% CI 1.26 to 69.58, 1 study with 610 patients included). Interestingly, no chills occurred with standard androgen suppression therapy (18/409 degarelix vs 0/201 standard androgen suppression therapy).
There were no statistically significant differences regarding serious adverse events (RR 0.82, 95% CI 0.62 to 1.08, 7 studies with 2179 patients included), severe/life-threatening adverse events (RR 0.76, 95% CI 0.58 to 1.00, 5 studies with 2064 patients included), or discontinuations due to adverse events (RR 0.86, 95% CI 0.57 to 1.31, 8 studies with 2290 patients included).
We identified no statistical significant differences between GnRH antagonists and standard androgen suppression therapy for immediate-onset allergic reactions (RR 2.36, 95% CI 0.55 to 10.12, 5 studies with 1694 patients included, table 8). However, this adverse event occurred in 9 of 1119 patients (0.8%) treated with abarelix but in no patient receiving standard androgen suppression therapy. We found no data for degarelix regarding this outcome.
We did not identify information about the occurrence of gynaecomastia, breast pain or sweating with the use of GnRH antagonist therapy.
Quality of life
Three studies were included for quality of life evaluation (CS28, CS20, and CS31). Further two studies (CS35 and CS37) were identified to measure quality of life outcomes through screening of protocol entries. However, we found no publications of these studies that reported this outcome. The question addressed by this systematic review was different from the results presented in included studies because we expected a measurement of quality of life related to general health, but instead found an evaluation of quality of life related to urinary or prostate symptoms only.
While there was no statistically significant difference for quality of life related to urinary symptoms, improved quality of life regarding prostate symptoms measured with the International Prostate Symptom Score (IPSS) with the use of GnRH antagonists (degarelix 240/80 mg) compared with the use of standard androgen suppression therapy (mean score difference −0.40, 95% CI −0.94 to 0.14, 3 studies with 461 patients included, and −1.84, 95% CI −3.00 to −0.69, 3 studies with 459 patients included, respectively) was found. Quality of evidence was rated low according to GRADE (table 7).
Based on the assessed evidence, including trials not published as journal articles, the effects on efficacy of GnRH antagonist compared to standard androgen therapy are still unclear since no long-term follow-up data (>364 days) are available for any of the evaluated outcomes and as evidence is hampered by selective reporting of results, risk of bias, and insufficient reporting of methodology. Fifteen studies were identified, but only 13 could be included. No study reported cancer-specific survival or clinical progression. There were no statistically significant differences in overall mortality, treatment failure, PSA progression or quality of life. However, quality of evidence according to GRADE was rated low for these outcomes.
The question addressed by this systematic review could partly not be answered with the available evidence. We aimed to assess efficacy and safety of GnRH antagonists compared with standard androgen suppression therapy for advanced prostate cancer treatment. However, most of the studies available were not intended to provide, as their primary end point, safety and efficacy data. The majority of studies included were performed or sponsored by the manufacturing companies to gain regulatory approval for marketing authorisation. The studies aimed to assess the pharmacodynamic metrics of obtaining a level of testosterone </=50 ng/dL by day 28, and maintaining that level through 365 days. The primary outcome of two studies (CS30 and CS31) was the evaluation of prostate volume reduction and relief from lower urinary tract symptoms. In one study (CS21), many patients had localised disease or PSA relapse only. The majority of patients treated with androgen suppression therapy for prostate cancer had non-metastatic disease (range 58–96%), and the number of patients with Gleason score <7 ranged between 18% (CS31) and 57% (149-98-03). Future studies, therefore, should focus on patient-relevant outcomes to inform decision-making in clinical practice.
The FDA required a black-box warning on the packaging and the patient instruction sheet of abarelix in USA because immediate-onset systemic allergic reactions occurred after administration of this drug. We found no statistically significant differences in immediate-onset allergic reactions between GnRH antagonists and standard androgen suppression therapy. However, it should be mentioned that 1.1% of patients included in FDA safety data analysis, treated with abarelix, discontinued therapy because of immediate onset of allergic-type adverse events, and 0.4–0.5% had serious anaphylactic-like reactions. There were no such events in the control groups treated with standard androgen suppression therapy.23 Additionally, the risk for injection-site events was increased using GnRH antagonists. This result is consistent with the FDA safety data analysis, where 25% of patients treated with degarelix had injection-site reactions (grade 3 or 4 events in 1% of patients).51
Fewer cardiovascular events occurred among patients using GnRH antagonists than among patients using standard androgen suppression therapy. This has been noted in the literature previously.60 ,77–79 However, there is evidence for both medications that in patients with a pre-existing cardiovascular disease and/or corresponding risk factors, these drugs may increase the risk to suffer from cardiovascular events in the long term and therefore, these subgroup of patients may need careful clinical follow-up.78–81
Evidence is hampered by risk of bias, selective reporting, and limited follow-up. Quality of evidence for all assessed outcomes was rated low according to GRADE. There is currently insufficient evidence to make firm conclusive statements on the efficacy of GnRH antagonist compared to standard androgen suppression therapy for advanced prostate cancer. The risk for injection-site events was increased, but cardiovascular events may occur less often using GnRH antagonist. Further high-quality research on GnRH antagonists with long-term follow-up is required.
Twitter Follow Joerg Meerpohl at @meerpohl
Contributors FK, HB, AB, BK, BW, CS, DS, CR, SS, AW and JJM were involved in the conception or design of the work, or the acquisition, analysis or interpretation of data; and involved in the drafting the work or revising it critically for important intellectual content and gave final approval of the version. All of the authors have read and approved the manuscript. All of the authors had full access to all study data, and take full responsibility for the integrity of the data and the accuracy of the data analysis.
Funding This systematic review was supported by a grant from the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF, 01KG1218).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional data can be accessed via the Dryad data repository at http://datadryad.org/ with the doi:10.5061/dryad.11k3m.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.