Methods
A systematic literature search was performed in May 2014 to identify published RCTs evaluating the clinical efficacy of everolimus (EVE) plus exemestane (EXE) versus chemotherapy (CT) or chemotherapy plus biological (CT + BIO) in hormone receptor (ER) positive and human epidermal growth factor 2 (HER2) negative MBC in first or second line of treatment.
PFS has been selected as the primary endpoint based on the common definition that PFS was the time from randomization to either death or disease progression, whichever occurs first. The lack of follow-up information in most of the papers was the main reason not to consider the overall survival (OS) as the primary endpoint. As reported in several studies, the use of PFS is mixed up with Time To Progression (TTP) [16], thus it was decided to use PFS and TTP as the same variable for the same endpoint. However, because PFS/TTP are surrogate endpoints, an improvement in PFS/TTP alone was not considered as sufficient evidence of a patients’ benefit; thus, the Tumor Response Rate (RR), defined accordingly to the RECIST Criteria, was added into the analysis as a variable for the evaluation of patients’ benefit.
Search strategy
The search following the recommendations of The Cochrane Collaboration was performed in order to identify all relevant, published and unpublished trials. Due to the relatively recent availability of studies involving everolimus-based therapies for metastatic breast cancer and the new chemotherapeutical schedule and/or compounds used in clinical routine, a literature search starting from year 2000 was considered sufficient for the purpose of this review. The search strategy was based by using a combination of subject headings and text words relating to everolimus, chemotherapy, biological/target therapies and metastatic breast cancer. In particular, the combination of disease, treatment, and study design terms in both studies’ titles or abstracts was:
- disease terms: breast or mammary and disease descriptors (cancer, neoplasm, oncology, tumor, malignancy, carcinoma, adenocarcinoma, or sarcoma) as well as metastasis, advanced, secondary, recurrent, inoperable, disseminated, or incurable.
- treatment terms: chemotherapy, endocrine therapy, everolimus (as well as Afinitor, SDZ-RAD, rad001, or 159351-69-6), cyclophosphamide, methotrexate, fluorouracil, doxorubicin, mitoxantrone, epirubicin, paclitaxel, docetaxel, liposomal doxorubicin, nab-paclitaxel, pegylated, liposomal doxorubicin, eribulin, capecitabine, vinorelbine.
- RCT terms: British Medical Journal filter for RCTs.
The electronic databases Pubmed®, EMBASE®, Cochrane Central Register of Clinical Trials and Web of Science were used for the search.
Three reviewers independently evaluated each identified single study against the established predetermined criteria. Details were extracted on study design, study population characteristics, and interventions. The hazard ratios and associated 95% confidence intervals were extracted for PFS/TTP where reported.
Statistical Analysis
A Bayesian Network Meta Analysis (NMA) framework was used for each endpoints considered: the tumor response rate(RR), the hazard ratio (HR) of PFS/TTP and the whole PFS/TTP curve [17]. In Figure S1, the scanned survival proportions are presented by study (Panel A) and by treatment (Panel B). Additionally, based on the PFS/TTP analysis, a meta-analysis of the Number Needed to Treat (NNT) for EXE/EVE versus each one of the other chemotherapy-based regimes was also performed [18].
All models have been implemented with both fixed and random effects to identify the best fit to the data. The parameters of the different models were estimated using a Markov Chain Monte Carlo (MCMC) method as implemented in the WinBUGS software package [19]. For all the analyses, the WinBUGS sampler, using two chains, was run for 500,000 iterations that were discarded as ‘burn-in’, and the model was run for a further 2,500,000 iterations on which inferences were based. A thinning rate of 100 iterations was used to reduce autocorrelation of the sampled values [20], thus leaving 25,000 iterations per chain to use for estimation and inference. Convergence of the chains was confirmed by the Gelman-Rubin statistic and by inspection of the trace plots. The Deviance Information Criterion was used to compare the goodness-of-fit of different models [22]. DIC provides a measure of model fit that penalizes model complexity. The model with the lowest DIC was considered the model providing the “best” fit to the data.
Quality assessments
The evaluation of the presence of bias in the studies included into the analysis is therefore an essential component of a review or meta-analysis. Internal validity of eligible studies was assessed according to the Cochrane Collaboration’s ’Risk of Bias’ tool in Review Manager (RevMan 5, The risk of bias and judged bias were described in the following specific domains: sequence generation; allocation concealment; blinding of participants and personnel and outcome assessors; blinding of outcome assessment: PFS/TTP and response; blinding of outcome assessment: Toxicity; incomplete outcome data: PFS and OS; incomplete outcome data: Response; incomplete outcome data: Toxicities; selective outcome reporting.
Table S1: Deviance information criterion (DIC) values for the models of the whole PFS/TTP curves.
Fixed effects model / Random effects modelWeibull model
(Ouwens et al, 2011) / 7,330.92 / 7,331.01
Second-order fractional polynomial with and
(Jansen, 2011) / 7,164.68 / 7,168.05
Table S2A:NMA results for the PFS/TTP curves. Values reported are differences in the expected PFS/TTP (in months) for the combination of everolimus plus exemestane versus each one of the chemotherapy-based treatments. It provides the posterior means and medians, together with the 80 and 95% CrIs. If the 95% CrI does not include the value 0, this suggests that there is evidence of a difference between the treatments (in gray when significantly superior).
Comparator / Number of trials / Posterior mean / Posterior median / 95% credible intervalCapecitabine+Sunitinib / 1 / 9.79 / 9.49 / 3.79 / 17.74
CMF / 4 / 8.97 / 8.87 / 0.96 / 17.25
Megestrol Acetate / 2 / 6.91 / 6.31 / 3.2 / 14.4
Capecitabine / 5 / 8.59 / 8.72 / -1.44 / 17.13
Mitoxantrone / 2 / 7.71 / 7.75 / -1.78 / 16.24
Tamoxifen / 2 / 6.04 / 5.58 / 1.37 / 13.87
BMF / 1 / 7.81 / 8.06 / -3.99 / 16.47
FEC / 4 / 7.2 / 7.25 / -3.63 / 15.98
Docetaxel+Epirubicin / 2 / 6.1 / 6.08 / -4.86 / 14.81
Gemcitabine+Epirubicin+Paclitaxel / 1 / 5.95 / 6.22 / -5.97 / 14.63
Bevacizumab 15 mg/kg+Docetaxel / 1 / 6.45 / 6.83 / -8.07 / 15.81
Sunitinib+Paclitaxel / 1 / 6.28 / 7.19 / -12.22 / 16.6
Exemestane / 3 / 3.89 / 3.91 / -5.24 / 13.8
Vinorelbine+Capecitabine / 1 / 6.82 / 9.2 / -15.23 / 18.05
Paclitaxel / 5 / 5.58 / 7.35 / -14.49 / 17.09
Capecitabine+Sorafenib / 1 / 4.03 / 5.44 / -14.67 / 15.42
Liposomal Doxorubicin / 1 / 4.33 / 5.95 / -19.02 / 16.39
Gemcitabine+Paclitaxel / 2 / 3.86 / 5.67 / -17.88 / 16.1
Everolimus+Tamoxifen / 1 / -3.1 / -3 / -13.93 / 7.55
Docetaxel+Sunitinib / 1 / 3.07 / 5.25 / -20.89 / 16.16
Paclitaxel+Sorafenib / 1 / 3.03 / 5.41 / -19.94 / 16.29
Fluorouracil+Vinorelbine / 1 / 3.21 / 5.69 / -20.65 / 16.38
Doxorubicin / 1 / 2.6 / 4.71 / -21.71 / 15.81
Placebo / 1 / 2.88 / 5.53 / -22.16 / 16.54
Docetaxel+Gemcitabine / 4 / 2.33 / 4.75 / -21.02 / 15.93
Docetaxel / 7 / 1.95 / 4.58 / -22.13 / 16.18
Bevacizumab+Paclitaxel / 4 / 1.43 / 4.1 / -22.61 / 16.06
Capecitabine+Docetaxel / 4 / -0.15 / 2.93 / -26.08 / 15.65
Bevacizumab+Capecitabine / 1 / -2.59 / 0.68 / -28.54 / 14.94
Paclitaxel+Bevacizumab+Gemcitabine / 1 / -2.08 / 1.38 / -30.27 / 15.72
Nab-Paclitaxel 150 mg/m2 / 1 / -4.68 / -1.1 / -32.38 / 14.61
BMF: BENDAMUSTINE HYDROCHLORIDE, METHOTREXATE, 5 FLUOROURACIL
CMF: CYCLOPHOSPHAMIDE, METHOTREXATE, 5 FLUOROURACIL
FEC: 5 FLUOROURACIL, EPIRUBICIN, CYCLOPHOSPHAMIDE
1
Table S2B:NMA results for the RRs. Numbers reported are odds ratios of each chemotherapy-based treatment versus the combination everolimus plus exemestane (in gray when significantly superior).
Comparator / Number of trials / Posterior mean / Posterior median / 95% credible intervalMitoxantrone / 2 / 0.08 / 0.05 / 0.01 / 0.28
Tamoxifen / 2 / 0.08 / 0.07 / 0.02 / 0.19
CMF / 2 / 0.09 / 0.06 / 0.01 / 0.35
Capecitabine / 4 / 0.1 / 0.06 / 0.01 / 0.43
Capecitabine+Sunitinib / 1 / 0.11 / 0.07 / 0.01 / 0.49
Megestrol Acetate / 2 / 0.12 / 0.11 / 0.04 / 0.29
FEC / 4 / 0.13 / 0.09 / 0.01 / 0.51
Capecitabine+Sorafenib / 1 / 0.15 / 0.09 / 0.01 / 0.66
Exemestane / 3 / 0.15 / 0.14 / 0.05 / 0.33
Everolimus+Tamoxifen / 1 / 0.19 / 0.16 / 0.04 / 0.55
Bevacizumab+Capecitabine / 1 / 0.21 / 0.13 / 0.02 / 0.88
Gemcitabine+Epirubicin+Paclitaxel / 1 / 0.21 / 0.14 / 0.02 / 0.84
Epirubicin+Cyclophosphamide / 1 / 0.24 / 0.12 / 0.01 / 1.25
Paclitaxel+Carboplatin / 2 / 0.27 / 0.13 / 0.01 / 1.37
Epirubicin+Paclitaxel / 4 / 0.36 / 0.17 / 0.02 / 1.86
Paclitaxel / 6 / 0.41 / 0.21 / 0.02 / 2.03
Docetaxel+Epirubicin / 2 / 0.42 / 0.26 / 0.03 / 1.76
Capecitabine+Paclitaxel / 1 / 0.45 / 0.22 / 0.02 / 2.42
Fluorouracil+Vinorelbine / 1 / 0.51 / 0.25 / 0.02 / 2.62
Epirubicin+Paclitaxel+Capecitabine / 1 / 0.53 / 0.25 / 0.02 / 2.83
Ixabepilone (16 mg/m2)+Bevacizumab / 1 / 0.55 / 0.25 / 0.02 / 2.91
Docetaxel / 7 / 0.58 / 0.3 / 0.03 / 2.89
Placebo / 1 / 0.63 / 0.31 / 0.03 / 3.21
Paclitaxel+Sorafenib / 1 / 0.74 / 0.37 / 0.03 / 3.74
Gemcitabine+Paclitaxel / 2 / 0.8 / 0.42 / 0.04 / 3.93
Doxorubicin / 1 / 0.87 / 0.44 / 0.04 / 4.41
Bevacizumab+Paclitaxel / 5 / 0.9 / 0.5 / 0.05 / 4.18
Capecitabine+Docetaxel / 2 / 0.9 / 0.46 / 0.04 / 4.52
Docetaxel+Gemcitabine / 3 / 0.91 / 0.49 / 0.05 / 4.38
Sunitinib+Paclitaxel / 1 / 0.92 / 0.46 / 0.04 / 4.7
Docetaxel+Sunitinib / 1 / 0.99 / 0.5 / 0.05 / 5.01
Nab-Paclitaxel 150 mg/m2 / 1 / 1.08 / 0.52 / 0.05 / 5.65
Bevacizumab 15 mg/kg+Docetaxel / 1 / 1.2 / 0.61 / 0.06 / 6.12
Paclitaxel+Bevacizumab+Gemcitabine / 1 / 1.43 / 0.7 / 0.06 / 7.48
Vandetanib+Docetaxel / 1 / 2.43 / 1.01 / 0.07 / 13.78
CMF: CYCLOPHOSPHAMIDE, METHOTREXATE, 5 FLUOROURACIL
FEC: 5 FLUOROURACIL, EPIRUBICIN, CYCLOPHOSPHAMIDE
1
Table S2C:NMA results for the HRs of PFS/TTP. Numbers reported are hazard ratios of the combination everolimus plus exemestane versus each one of chemotherapy-based treatments (in grey when significantly superior).
Comparator / Number of trials / Posterior mean / Posterior median / 95% credible intervalCapecitabine+Sunitinib / 1 / 7.21 / 5.81 / 1.67 / 20.70
Capecitabine / 4 / 5.86 / 4.78 / 1.41 / 16.55
Epirubicin+Paclitaxel+Capecitabine / 1 / 5.21 / 3.86 / 0.86 / 17.71
CMF / 2 / 5.00 / 4.12 / 1.24 / 13.89
Capecitabine+Paclitaxel / 1 / 4.43 / 3.27 / 0.72 / 15.02
Epirubicin+Paclitaxel / 4 / 4.34 / 3.24 / 0.73 / 14.66
Epirubicin+Cyclophosphamide / 1 / 4.07 / 3.02 / 0.68 / 13.70
Bevacizumab+Capecitabine / 1 / 4.06 / 3.30 / 0.95 / 11.58
Mitoxantrone / 2 / 3.77 / 3.17 / 1.00 / 10.18
Capecitabine+Sorafenib / 1 / 3.71 / 2.96 / 0.81 / 10.98
FEC / 4 / 3.63 / 3.01 / 0.93 / 9.98
Gemcitabine+Epirubicin+Paclitaxel / 1 / 3.30 / 2.70 / 0.80 / 9.26
Paclitaxel+Carboplatin / 2 / 3.28 / 2.48 / 0.57 / 10.95
Megestrol Acetate / 2 / 3.17 / 3.14 / 2.29 / 4.29
Tamoxifen / 2 / 3.17 / 3.14 / 2.39 / 4.13
Sunitinib+Paclitaxel / 1 / 3.16 / 2.24 / 0.46 / 11.29
Paclitaxel / 6 / 3.05 / 2.31 / 0.55 / 10.14
Docetaxel+Epirubicin / 2 / 2.98 / 2.38 / 0.65 / 8.89
Exemestane / 3 / 2.65 / 2.63 / 2.11 / 3.28
Placebo / 1 / 2.47 / 1.74 / 0.35 / 8.85
Paclitaxel+Sorafenib / 1 / 2.44 / 1.82 / 0.41 / 8.34
Ixabepilone (16 mg/m2)+Bevacizumab / 1 / 2.41 / 1.69 / 0.34 / 8.71
Fluorouracil+Vinorelbine / 1 / 2.32 / 1.73 / 0.39 / 7.85
Docetaxel+Gemcitabine / 3 / 2.21 / 1.69 / 0.41 / 7.15
Gemcitabine+Paclitaxel / 2 / 2.06 / 1.57 / 0.37 / 6.81
Docetaxel / 7 / 1.94 / 1.47 / 0.35 / 6.45
Bevacizumab+Paclitaxel / 5 / 1.91 / 1.38 / 0.30 / 6.69
Capecitabine+Docetaxel / 2 / 1.83 / 1.41 / 0.35 / 5.86
Doxorubicin / 1 / 1.81 / 1.36 / 0.32 / 6.08
Docetaxel+Sunitinib / 1 / 1.80 / 1.35 / 0.31 / 6.01
Everolimus+Tamoxifen / 1 / 1.75 / 1.70 / 1.01 / 2.83
Paclitaxel+Bevacizumab+Gemcitabine / 1 / 1.59 / 1.13 / 0.23 / 5.69
Bevacizumab 15 mg/kg+Docetaxel / 1 / 1.50 / 1.13 / 0.26 / 5.05
Vandetanib+Docetaxel / 1 / 1.37 / 1.01 / 0.22 / 4.62
Nab-Paclitaxel 150 mg/m2 / 1 / 0.99 / 0.72 / 0.16 / 3.46
CMF: CYCLOPHOSPHAMIDE, METHOTREXATE, 5 FLUOROURACIL
FEC: 5 FLUOROURACIL, EPIRUBICIN, CYCLOPHOSPHAMIDE
Figure S1A: Progression-free survival as observed in individual studies by trial.
1
Figure S1B: Progression-free survival as observed in individual studies by treatment.
1
Figure S2: Risk of bias summary: reviewers’ judgement about each risk of bias item for each included study. Both the 80 and 95% CrIs are provided; posterior means are also reported using triangles for those treatment whose 95% CrI does not include the value 1, and using crosses for those including the value 1. If the 95% CrI does not include the value 1, this suggests that there is evidence of a difference between the treatments.
1
Figure S3: Risk of bias graph: review authors’ judgement about each risk of bias item presented as percentage of all included studies. Both the 80 and 95% CrIs are provided; posterior means are also reported using triangles for those treatment whose 95% CrI does not include the value 1, and using crosses for those including the value 1. If the 95% CrI does not include the value 1, this suggests that there is evidence of a difference between the treatments.
1