This well-conducted review reliably concluded that although there were no significant differences between letrozole, anastrozole and exemestane in overall survival and progression-free survival for hormone-sensitive advanced or metastatic breast cancer in postmenopausal women, basic assumptions were not fulfilled for the indirect comparisons and network analysis and so the results must be interpreted with caution.

To examine the efficacy of the aromatase inhibitors letrozole, anastrozole and exemestane for hormone-sensitive advanced or metastatic breast cancer in postmenopausal women.

MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials (CENTRAL), DARE and HTA databases were searched from inception to January 2009 for published and unpublished studies in any language. Search terms were reported. Internet searches used Google search engine. References of retrieved articles and systematic reviews and the websites of licensing agencies and HTA agencies were checked for additional studies.

Randomised controlled trials (RCTs) of letrozole, anastrozole or exemestane for first-line treatment of postmenopausal women with hormone receptor positive (patients with oestrogen receptor-positive or unknown and/or PgR-positive or unknown breast cancer) with or without ErB2 human epidermal growth factor 2 (HER2, synonymous with ErbB2) positive metastatic breast cancer were eligible for inclusion. Eligible patients had not received previous treatment for advanced or metastatic disease. Studies had to report safety and efficacy outcomes.

The included studies compared anastrozole (1mg/day), exemestane (25mg/day) or letrozole (2.5mg) with tamoxifen (20mg/day). Median age ranged from 63 to 67 years. The proportion of patients with bone metastases ranged from 30% to 59%. The number of HER2+ patients was not reported in any of the included studies. Overall survival, progression-free survival, time to progression, response rate, quality of life and adverse events were reported.

Two reviewers independently selected studies for inclusion. Disagreements were resolved by discussion.

Methodological quality was assessed independently by two reviewers in terms of randomisation, allocation concealment, outcome reporting, withdrawals and dropouts, blinding and intention-to-treat (ITT) analysis using the Cochrane Collaboration checklist. Disagreements were resolved by consensus.

Relative risks (RRs) or odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for dichotomous data. Continuous data were extracted as mean and standard deviation. Survival data were extracted (or calculated where necessary) as hazard ratios (HR) and standard errors. Data were extracted independently by two reviewers and disagreements resolved by consensus.

Pooled relative risks or odds ratios and associated 95% CIs were calculated using the DerSimonian and Laird method and Mantel-Haenszel method respectively. Weighted mean difference (WMD) and 95% CIs were calculated for continuous outcomes. Pooled hazard ratios and standard errors were calculated. Heterogeneity was assessed using the I² statistic. Quality-adjusted survival was calculated using quality-adjusted time without symptoms or toxicity (Q-TWIST). Survival curves were partitioned into toxicity, disease progression and periods without toxicity or disease progression. Where meta-analysis was not possible, data were synthesised narratively. Indirect comparisons and network meta-analyses were undertaken.

Four RCTs were included in the review (n=2,309, range 353 to 906). The methodological quality was deemed good in all studies. Three studies reported reliable methods of randomisation and allocation concealment. Three studies reported double-blinding. None of the studies indicated selective outcome reporting. Withdrawals and dropouts and ITT analyses were reported in all four studies (one did not include all randomised participants in the ITT analysis).

Direct comparisons:

Letrozole was associated with significantly better time to progression (HR 0.70, 95% CI 0.60 to 0.82), objective response rate (RR 0.65, 95% CI 0.52 to 0.82) and quality-adjusted time without symptoms or toxicity (Q-Twist difference=1.5, p<0.001) than tamoxifen in one RCT. There was no significant difference in overall survival between letrozole and tamoxifen (one RCT).

Exemestane was associated with significantly better objective response rate (RR 0.68, 95% CI 0.53 to 0.89) and progression-free survival (HR 0.87, 95% CI 0.70 to 1.08) than tamoxifen, but there was no difference in overall survival (one RCT).

Anastrozole was associated with significantly superior time to progression than tamoxifen in one trial (HR 1.42, 95% CI 1.15 to not reported), but not in another. Overall survival was not significantly different between groups in either trial.

Tamoxifen was associated with significantly more serious adverse events compared with exemestane (OR 0.61, 95% CI 0.38 to 0.97). Exemestane was associated with significantly more arthralgia in comparison with tamoxifen (OR 2.33, 95% CI 1.07 to 5.11).

Anastrozole was associated with significantly more total adverse events (OR 1.04, 95% CI 1.00 to 1.09) and hot flushes in comparison with tamoxifen (OR 1.39, 95% CI 1.03 and 1.89; one RCT). Another trial showed no significant differences in adverse events between anastrozole and tamoxifen.

No significant differences in adverse events were found between letrozole and tamoxifen.

Indirect comparisons:

Letrozole and exemestane were associated with improved objective response rate compared with anastrozole; there was no significant difference between the three aromatase inhibitors in overall survival and progression-free survival.

There were no significant differences between aromatase inhibitors in the outcomes overall survival and progression-free survival; therefore, a class effect for all aromatase inhibitors was possible. These results were based on indirect comparisons and a network analysis for which the assumptions of homogeneity, similarity and consistency were not fulfilled; therefore, the results must be interpreted with caution.

The review question was clearly stated and supported by inclusion driteria for participants, interventions, outcomes and study design. Published and unpublished papers in any language were sought, which reduced risks of language and publication biases. The review processes of study selection, quality assessment and data extraction were performed in duplicate, which reduced the risk of reviewer error and bias. Study quality was assessed using appropriate criteria and taken into consideration. The paucity of data meant that heterogeneity could not be assessed.

Two authors worked for GlaxoSmithKline; one author received honoraria from Pfizer, Novartis and Astra Zeneca and received research funding to his institution from Pfizer, Novartis and Astra Zeneca.

This was a well-conducted review and the authors' cautious conclusions are likely to be reliable.

Practice: The authors stated that their results confirmed guidance provided by NICE that found there was insufficient evidence to conclude that any one aromatase inhibitor (used within the licenced indications) or treatment stategy was more clinically effective than another.

Research: The authors stated that head-to-head comparisons between letrozole, anastrozole and exemestane in the first-line metatstatic breast cancer setting were needed. They suggested that a greater priority was to focus on integration of established endocrine agents in combination with the contemporary generation of biological therapies, ideally using appropriately selected biologically relevant subgroups.

GlaxoSmithKline.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.