Advertisement

Multiple Bayesian network meta-analyses to establish therapeutic algorithms for metastatic triple negative breast cancer

Open AccessPublished:September 27, 2022DOI:https://doi.org/10.1016/j.ctrv.2022.102468

      Highlights

      • We compared 118 1st- and 33 ≥ 2nd-line regimens for advanced triple negative (TN)BC.
      • Paclitaxel ± bevacizumab is a valuable 1st-line based on efficacy, activity and safety.
      • PARP-inhibitors(i) are an effective 1st-line in germline-BRCA1/2-mutant(mut) TNBC.
      • Immunotherapy + chemotherapy is an effective 1st-line option for PD-L1-positive TNBC.
      • Sacituzumab govitecan is the best 2nd-line, then T-DXd in HER2-low and PARPi in mut TN.

      Abstract

      Metastatic triple-negative breast cancer (mTNBC) is a poor prognostic disease with limited treatments and uncertain therapeutic algorithms. We performed a systematic review and multiple Bayesian network meta-analyses according to treatment line to establish an optimal therapeutic sequencing strategy for this lethal disease. We included 125 first-line trials (37,812 patients) and 33 s/further-lines trials (11,321 patients). The primary endpoint was progression-free survival (PFS). Secondary endpoints included overall response rates (ORR), overall survival (OS) and safety, for first and further lines, separately. We also estimated separate treatment rankings for the first and subsequent lines according to each endpoint, based on (surface under the cumulative ranking curve) SUCRA values. No first-line treatment was associated with superior PFS and OS than paclitaxel ± bevacizumab. Platinum-based polychemotherapies were generally superior in terms of ORR, at the cost of higher toxicity.. PARP-inhibitors in germline-BRCA1/2-mutant patients, and immunotherapy + chemotherapy in PD-L1-positive mTNBC, performed similar to paclitaxel ± bevacizumab. In PD-L1-positive mTNBC, pembrolizumab + chemotherapy was better than atezolizumab + nab-paclitaxel in terms of OS according to SUCRA values. In second/further-lines, sacituzumab govitecan outperformed all other treatments on all endpoints, followed by PARP-inhibitors in germline-BRCA1/2-mutant tumors. Trastuzumab deruxtecan in HER2-low mTNBC performed similarly and was the best advanced-line treatment in terms of PFS and OS after sacituzumab govitecan, according to SUCRA values. Moreover, comparisons with sacituzumab govitecan, talazoparib and olaparib were not statistically significant. The most effective alternatives or candidates for subsequent lines were represented by nab-paclitaxel (in ORR), capecitabine (in PFS) and eribulin (in PFS and OS).

      Keywords

      Introduction

      Triple negative breast cancer (TNBC) represents approximately 10–15 % of all breast tumors and is defined by the absence of endocrine receptors (ER) and HER2 gene overexpression/amplification [
      • Tutt A.
      • Tovey H.
      • Cheang M.C.U.
      • Kernaghan S.
      • Kilburn L.
      • Gazinska P.
      • et al.
      A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
      ]. Metastatic TNBC (mTNBC) remains an incurable disease with unfavorable prognosis [
      • Gong Y.
      • Liu Y.-R.
      • Ji P.
      • Hu X.
      • Shao Z.-M.
      Impact of molecular subtypes on metastatic breast cancer patients: a SEER population-based study.
      ]. However, the scenario is quickly changing. 10.6–41.9 % TNBC harbor a germline mutation in the homologous recombination repair (HRR) genes BRCA1/2 [
      • Hartman A.-R.
      • Kaldate R.R.
      • Sailer L.M.
      • Painter L.
      • Grier C.E.
      • Endsley R.R.
      • et al.
      Prevalence of BRCA mutations in an unselected population of triple-negative breast cancer.
      ]. In these cases, following positive results from respective pivotal trials, it is now possible to administer the PARP-inhibitors (PARPi) olaparib or talazoparib in first or further lines [
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ]. Additionally, the TNT phase III first-line trial recently demonstrated that carboplatin was more effective than the standard of care docetaxel in germline BRCA1/2-mutant (gBRCA-mut) mTNBC [
      • Tutt A.
      • Tovey H.
      • Cheang M.C.U.
      • Kernaghan S.
      • Kilburn L.
      • Gazinska P.
      • et al.
      A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
      ]. Intriguingly, in the same trial, in a broader condition of HRR deficiency (HRD), none of the 2 drugs was superior to the other [
      • Tutt A.
      • Tovey H.
      • Cheang M.C.U.
      • Kernaghan S.
      • Kilburn L.
      • Gazinska P.
      • et al.
      A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
      ]. Recently, two randomized phase III studies showed that PD-L1 positive (+) mTNBC (though selected with different methodologies) derive benefit from the addition of an anti-PD-L1/PD-1 immune-checkpoint inhibitor (ICI) to upfront chemotherapy (CT) [
      • Cortes J.
      • Cescon D.W.
      • Rugo H.S.
      • Nowecki Z.
      • Im S.-A.
      • Yusof M.M.
      • et al.
      KEYNOTE-355: Randomized, double-blind, phase III study of pembrolizumab + chemotherapy versus placebo + chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer.
      ,
      • Schmid P.
      • Adams S.
      • Rugo H.S.
      • Schneeweiss A.
      • Barrios C.H.
      • Iwata H.
      • et al.
      Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer.
      ]. However, the majority of mTNBC are PD-L1-negative (40–80 %) and germline BRCA1/2-wild type (gBRCA-wt) (60–90 %) [
      • Hartman A.-R.
      • Kaldate R.R.
      • Sailer L.M.
      • Painter L.
      • Grier C.E.
      • Endsley R.R.
      • et al.
      Prevalence of BRCA mutations in an unselected population of triple-negative breast cancer.
      ,
      • Rozenblit M.
      • Huang R.
      • Danziger N.
      • Hegde P.
      • Alexander B.
      • Ramkissoon S.
      • et al.
      Comparison of PD-L1 protein expression between primary tumors and metastatic lesions in triple negative breast cancers.
      ,

      Mittendorf EA, Philips AV, Meric-Bernstam F, Qiao N, Wu Y, Harrington S, et al. PD-L1 expression in triple-negative breast cancer. Cancer Immunol Res 2014;2:361–70. https://doi.org/10.1158/2326-6066.CIR-13-0127.

      ], thus being only manageable with mono- or poly-CT, with the possibility (only in Europe) to use a first-line combination of bevacizumab with either paclitaxel or capecitabine [

      National Comprehensive Cancer Network. NCCN Guidelines for Breast Cancer, vers.4.2022 n.d. https://www.nccn.org/professionals/physician_gls/pdf/breast.pdf (accessed June 28, 2022).

      ,
      • Gennari A.
      • André F.
      • Barrios C.H.
      • Cortés J.
      • de Azambuja E.
      • DeMichele A.
      • et al.
      ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
      ].
      Importantly, the novel anti-TROP2 antibody-drug conjugate (ADC) sacituzumab govitecan provided unprecedented overall response rates (ORR), progression-free survival (PFS) and overall survival (OS) benefit in heavily pretreated mTNBC [
      • Bardia A.
      • Mayer I.A.
      • Vahdat L.T.
      • Tolaney S.M.
      • Isakoff S.J.
      • Diamond J.R.
      • et al.
      Sacituzumab Govitecan-hziy in Refractory Metastatic Triple-Negative Breast Cancer.
      ,
      • Bardia A.
      • Hurvitz S.A.
      • Tolaney S.M.
      • Loirat D.
      • Punie K.
      • Oliveira M.
      • et al.
      Sacituzumab Govitecan in Metastatic Triple-Negative Breast Cancer.
      ], quickly leading to a US Food and Drug Administration (FDA) approval in this subset. Although preliminary, similar findings have been recently observed with the novel ADC trastuzumab deruxtecan (T-DXd) in the subset of TNBC with low expression levels of HER2 [
      • Schettini F.
      • Chic N.
      • Brasó-Maristany F.
      • Paré L.
      • Pascual T.
      • Conte B.
      • et al.
      Clinical, pathological, and PAM50 gene expression features of HER2-low breast cancer.
      ,
      • Modi S.
      • Jacot W.
      • Yamashita T.
      • Sohn J.
      • Vidal M.
      • Tokunaga E.
      • et al.
      Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer.
      ].
      In this complex and changing scenario, uncertainties exist on whether and when to prefer single-agent over multi-agent treatments and most regimens have not undergone head-to-head comparisons. Also, the role of platinum-based regimens is unclear and therapeutic sequencies uncertain [
      • Gennari A.
      • André F.
      • Barrios C.H.
      • Cortés J.
      • de Azambuja E.
      • DeMichele A.
      • et al.
      ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
      ,

      Egger SJ, Chan MMK, Luo Q, Wilcken N. Platinum-containing regimens for triple-negative metastatic breast cancer. Cochrane Database Syst Rev 2020;10:CD013750. https://doi.org/10.1002/14651858.CD013750.

      ,
      • Moy B.
      • Rumble R.B.
      • Come S.E.
      • Davidson N.E.
      • Di Leo A.
      • Gralow J.R.
      • et al.
      Chemotherapy and Targeted Therapy for Patients With Human Epidermal Growth Factor Receptor 2–Negative Metastatic Breast Cancer That is Either Endocrine-Pretreated or Hormone Receptor–Negative: ASCO Guideline Update.
      ]. To date, only network meta-analyses (NMA) provides a methodologically reliable statistical framework to indirectly compare treatments that have never been confronted in head-to-head studies, given that such therapies have been compared to at least one common comparator [
      • Jansen J.P.
      • Fleurence R.
      • Devine B.
      • Itzler R.
      • Barrett A.
      • Hawkins N.
      • et al.
      Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1.
      ,
      • Hoaglin D.C.
      • Hawkins N.
      • Jansen J.P.
      • Scott D.A.
      • Itzler R.
      • Cappelleri J.C.
      • et al.
      Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2.
      ]. NMA also allows to determine the amount of agreement between the results obtained when different linking treatments are used; it can incorporate results from direct comparisons, to account for both direct and indirect evidences at the same time, and it can provide a rank ordering of the interventions [
      • Jansen J.P.
      • Fleurence R.
      • Devine B.
      • Itzler R.
      • Barrett A.
      • Hawkins N.
      • et al.
      Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1.
      ,
      • Hoaglin D.C.
      • Hawkins N.
      • Jansen J.P.
      • Scott D.A.
      • Itzler R.
      • Cappelleri J.C.
      • et al.
      Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2.
      ].
      Therefore, we conducted a systematic literature search to identify all phase II/III randomized controlled trials (RCT) published in the last 20 years comparing all CT and target therapies (TT) in metastatic HER2-negative breast cancer (BC) and carried out multiple Bayesian NMA to define the best therapeutic options for mTNBC according to treatment line.

      Methods

      Search strategy and selection criteria

      We performed a systematic literature search on Pubmed® and CENTRAL to identify all phase II/III RCT published between 01/01/2000 and 30/06/2020 comparing all CT and TT in metastatic HER2-negative BC. The time span was selected to both retrieve studies more uniformly conceived and presented, and include all currently available and most promising therapeutic options. The full search query is reported in the Supplementary Methods. Articles relevant to the topic published between July 2020 and June 2022 were manually included in the networks before conducting final analyses. Online archives of the San Antonio Breast Cancer Symposium, European Society for Medical Oncology (ESMO)’s Congress, ESMO Breast Congress and American Society of Clinical Oncology (ASCO)’s Annual Meeting were also consulted. No language restrictions were adopted. Records had to be preferably full papers, however, for each trial where only an abstract was available and results were adequately provided, the study was included in our analyses. In case of more publications for the same study, the most updated one was considered. Two independent reviewers (FS and MG) carried out the systematic revision of the literature and a third one (DG) was consulted in case of controversy. Some cross-references from main international guidelines were also included.

      Endpoints

      We aimed at identifying the best therapeutic option for the first-line and second/further (advanced) lines, separately, according to efficacy/activity and safety. PFS was the primary endpoint whereas ORR, OS and safety were secondary endpoints.

      Data extraction

      To be included, a publication had to provide at least the data for one of the three endpoints of this study (detailed response rates and/or hazard ratios [HR] of PFS and/or OS). In case HR for Time-to-Progression (TTP) instead of PFS were provided, TTP was considered for the analysis. TTP and PFS are very similar endpoints, which have been frequently mixed-up in other previously published NMA [
      • Giuliano M.
      • Schettini F.
      • Rognoni C.
      • Milani M.
      • Jerusalem G.
      • Bachelot T.
      • et al.
      Endocrine treatment versus chemotherapy in postmenopausal women with hormone receptor-positive, HER2-negative, metastatic breast cancer: a systematic review and network meta-analysis.
      ,
      • Li L.
      • Pan Z.
      Progression-Free Survival and Time to Progression as Real Surrogate End Points for Overall Survival in Advanced Breast Cancer: A Meta-Analysis of 37 Trials.
      ,
      • Generali D.
      • Venturini S.
      • Rognoni C.
      • Ciani O.
      • Pusztai L.
      • Loi S.
      • et al.
      A network meta-analysis of everolimus plus exemestane versus chemotherapy in the first- and second-line treatment of estrogen receptor-positive metastatic breast cancer.
      ]. The HR and associated 95 % confidence intervals (CI) for OS and PFS/TTP were extracted from each paper. The number of patients achieving a partial or complete response as their best response according to treatment arm were extracted to calculate the ORR.
      Data concerning the following variables were also extracted from all the studies: full publication reference, publication year, line of treatment, phase of the trial, investigated treatments, single center vs multicenter studies, follow-up period (months), total number of patients, proportion of patients with ER + BC, median age, age range, proportion of patients with PD-L1+, gBRCA-mut and altered PI3K/PTEN pathway, as well as proportion of visceral, lung, liver and bone metastases, and main grade (G)3–5 adverse events (AEs).
      Several adjustments regarding data extraction need to be disclosed. Firstly, when a study set in first and further lines presented with a result for the entire population enrolled and separate results for the first line and for the following, results according to treatment lines were extracted. Secondly, three trials specifically enrolled patients with gBRCA-mut tumors [
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ,
      • Diéras Véronique
      • Han H.S.
      • Kaufman B.
      • Wildiers H.
      • Friedlander M.
      • Ayoub J.-P.
      • et al.
      Veliparib with carboplatin and paclitaxel in BRCA-mutated advanced breast cancer (BROCADE3): a randomised, double-blind, placebo-controlled, phase 3 trial.
      ]. Therefore, our results concerning the therapeutic agents administered within these studies have to be intended only for the specific subgroup of gBRCA-mut TNBC. Third, the same three studies presented separate aggregate data for TNBC and ER + tumors, although such results were not provided according to treatment line. To better estimate the therapeutic efficacy in the first and subsequent lines, we preferred to include results according to treatment line, although a proportion of ER + BC was comprised in these estimates. Fourth, the studies of pembrolizumab and atezolizumab included both patients with and without PD-L1+ tumors [
      • Schmid P.
      • Adams S.
      • Rugo H.S.
      • Schneeweiss A.
      • Barrios C.H.
      • Iwata H.
      • et al.
      Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer.
      ,
      • Miles D.
      • Gligorov J.
      • André F.
      • Cameron D.
      • Schneeweiss A.
      • Barrios C.
      • et al.
      Primary results from IMpassion131, a double-blind, placebo-controlled, randomised phase III trial of first-line paclitaxel with or without atezolizumab for unresectable locally advanced/metastatic triple-negative breast cancer.
      ,
      • Cortes J.
      • Cescon D.W.
      • Rugo H.S.
      • Nowecki Z.
      • Im S.-A.
      • Yusof M.M.
      • et al.
      Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355): a randomised, placebo-controlled, double-blind, phase 3 clinical trial.
      ,
      • Winer E.P.
      • Lipatov O.
      • Im S.-A.
      • Goncalves A.
      • Muñoz-Couselo E.
      • Lee K.S.
      • et al.
      Pembrolizumab versus investigator-choice chemotherapy for metastatic triple-negative breast cancer (KEYNOTE-119): a randomised, open-label, phase 3 trial.
      ]. Considering that such drugs have been FDA and/or European Medicine Agency (EMA)-approved only in mTNBC with a PD-L1 combined positive score (CPS) ≥ 10 % or PD-L1 levels ≥ 1 % according to the Ventana SP142 assay, respectively, only results for these subpopulations have been included. As a consequence, our results concerning the therapeutic agents administered within these studies have to be intended only for the specific subgroup of PD-L1+ mTNBC, with PD-L1 positivity defined according to the relative trials’ assays. Finally, AKT-inhibitors ipatasertib in Kim et al. 2017 and capivasertib in Schmid et al. 2019 were tested in TNBC with and without alterations in PIK3CA, AKT1 or PTEN, whilst the former was tested in Dent et al. 2021 only in PIK3CA/PTEN/AKT-altered mTNBC [
      • Kim S.-B.
      • Dent R.
      • Im S.-A.
      • Espié M.
      • Blau S.
      • Tan A.R.
      • et al.
      Ipatasertib plus paclitaxel versus placebo plus paclitaxel as first-line therapy for metastatic triple-negative breast cancer (LOTUS): a multicentre, randomised, double-blind, placebo-controlled, phase 2 trial.
      ,

      Dent R, Kim S-B, Oliveira M, Barrios C, O’Shaughnessy J, Isakoff SJ, et al. Abstract GS3-04: Double-blind placebo (PBO)-controlled randomized phase III trial evaluating first-line ipatasertib (IPAT) combined with paclitaxel (PAC) for PIK3CA/AKT1/PTEN-altered locally advanced unresectable or metastatic triple-negative breast cancer (aTNBC): primary results from IPATunity130 Cohort A. Cancer Res 2021;81:GS3-04. https://doi.org/10.1158/1538-7445.SABCS20-GS3-04.

      ,
      • Schmid P.
      • Abraham J.
      • Chan S.
      • Wheatley D.
      • Brunt A.M.
      • Nemsadze G.
      • et al.
      Capivasertib Plus Paclitaxel Versus Placebo Plus Paclitaxel As First-Line Therapy for Metastatic Triple-Negative Breast Cancer: The PAKT Trial.
      ]. For ipatasertib, which failed to prove a significant benefit transferrable to the clinical practice, the results for the intention-to-treat (ITT) trials’ population were included. Conversely, capivasertib was included for the obtained results in the PIK3CA/PTEN/AKT-altered population, a subset with a potential approval; although results from the phase III trial are required to draw definitive conclusions. Finally, in order to include some trials in the networks, some links had to be forced. More specifically, the capecitabine-containing treatment of physician’s choice (TPC) arms from Litton et al. and Robson et al. were linked to the capecitabine arm of Harbeck et al. to include olaparib and talazoparib in the first-line networks [
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ,
      • Harbeck N.
      • Saupe S.
      • Jäger E.
      • Schmidt M.
      • Kreienberg R.
      • Müller L.
      • et al.
      A randomized phase III study evaluating pegylated liposomal doxorubicin versus capecitabine as first-line therapy for metastatic breast cancer: results of the PELICAN study.
      ]. The CT arm of Cortes et al., which included also paclitaxel, and the taxane-containing arm of Takashima et al. were linked to weekly paclitaxel and the CT arm of Von Minckwitz et al. was considered as a TPC arm, allowing its inclusion in the second/further-lines networks [
      • Cortes J.
      • Cescon D.W.
      • Rugo H.S.
      • Nowecki Z.
      • Im S.-A.
      • Yusof M.M.
      • et al.
      Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355): a randomised, placebo-controlled, double-blind, phase 3 clinical trial.
      ,
      • von Minckwitz G.
      • Puglisi F.
      • Cortes J.
      • Vrdoljak E.
      • Marschner N.
      • Zielinski C.
      • et al.
      Bevacizumab plus chemotherapy versus chemotherapy alone as second-line treatment for patients with HER2-negative locally recurrent or metastatic breast cancer after first-line treatment with bevacizumab plus chemotherapy (TANIA): an open-label, randomised phase 3 trial.
      ,
      • Takashima T.
      • Mukai H.
      • Hara F.
      • Matsubara N.
      • Saito T.
      • Takano T.
      • et al.
      Taxanes versus S-1 as the first-line chemotherapy for metastatic breast cancer (SELECT BC): an open-label, non-inferiority, randomised phase 3 trial.
      ]. Additionally, TPC arms did not differ too much among different trials, therefore they were always linked together, whenever possible.

      Data analysis

      A Bayesian NMA framework was used for each endpoint and for each treatment line for a total of 6 networks, i.e. first-line and advanced lines networks of PFS/TTP, ORR and OS, respectively [
      • Jansen J.P.
      • Fleurence R.
      • Devine B.
      • Itzler R.
      • Barrett A.
      • Hawkins N.
      • et al.
      Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1.
      ,
      • Hoaglin D.C.
      • Hawkins N.
      • Jansen J.P.
      • Scott D.A.
      • Itzler R.
      • Cappelleri J.C.
      • et al.
      Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2.
      ,
      • Lunn D.J.
      • Thomas A.
      • Best N.
      • Spiegelhalter D.
      WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility.
      ].
      We also provided a ranking of treatments based on the surface under the cumulative ranking curve (SUCRA). The SUCRA values range from 0 to 100 %. The higher the SUCRA value, and the closer to 100 %, the higher the likelihood that a therapy is in the top rank; the closer to 0 the SUCRA value, the more likely that a therapy is in the bottom rank [
      • Mbuagbaw L.
      • Rochwerg B.
      • Jaeschke R.
      • Heels-Andsell D.
      • Alhazzani W.
      • Thabane L.
      • et al.
      Approaches to interpreting and choosing the best treatments in network meta-analyses.
      ].
      The parameters of the different models (HR of PFS/TTP and OS with 95 % credible intervals [CrI] and OR for ORR with 95 %CrI) were estimated using a Markov Chain Monte Carlo method as implemented in the WinBUGS software package [
      • Lunn D.J.
      • Thomas A.
      • Best N.
      • Spiegelhalter D.
      WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility.
      ]. For all the analyses, the WinBUGS sampler, using three chains, was run for 1,000,000 iterations that were discarded as ‘burn-in’, and the model was run for a further 2,000,000 iterations on which inferences were based. A thinning rate of 100 iterations was used to reduce autocorrelation of the sampled values, thus leaving 20,000 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 [
      • Gelman A.
      • Rubin D.
      Inference from iterative simulation using multiple sequences.
      ,
      • Lynch S.M.
      Introduction to Applied Bayesian Statistics and Estimation for Social Scientists.
      ]. For each NMA, the model providing the best fit to the data between random- and fixed-effect was chosen based on the Deviance Information Criterion (DIC). The 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, otherwise, wherever DIC values were similar (difference of < 5) a fixed-effect model was preferred [
      • Lunn D.J.
      • Thomas A.
      • Best N.
      • Spiegelhalter D.
      WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility.
      ]. For the NMA of the HR, we assumed that the logHR were normally distributed with the logHR mean equaling the true logHR observed in each study and the variance equaling the observed variability in each study.
      We used a vague flat (i.e., uniform) prior distribution for between-study standard deviation τ. Moreover, as for the correlations between the random effects for each trial, we adopted the standard approach to set this correlation equal to 0.5 [

      Dias S, Ades A, Welton N, Jansen J, Sutton A. Network Meta-Analysis for Decision-Making | Wiley. 1st Edition. John Wiley and Sons; 2018.

      ]. We used a common between-study variance parameter τ2 for all studies.
      The PRISMA guidelines for NMA were followed [
      • Hutton B.
      • Salanti G.
      • Caldwell D.M.
      • Chaimani A.
      • Schmid C.H.
      • Cameron C.
      • et al.
      The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations.
      ]. Inconsistency of the results was explored, as recommended [
      • Hutton B.
      • Salanti G.
      • Caldwell D.M.
      • Chaimani A.
      • Schmid C.H.
      • Cameron C.
      • et al.
      The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations.
      ,
      • Higgins J.P.T.
      • Jackson D.
      • Barrett J.K.
      • Lu G.
      • Ades A.E.
      • White I.R.
      Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies.
      ,
      • Dias S.
      • Welton N.J.
      • Caldwell D.M.
      • Ades A.E.
      Checking consistency in mixed treatment comparison meta-analysis.
      ]. For all treatment line networks according to each endpoint, an inconsistency model was obtained by omitting the consistency equations. Then, for each endpoint, the consistency and inconsistency models were compared in terms of goodness of fit by using their relative DIC [
      • Higgins J.P.T.
      • Jackson D.
      • Barrett J.K.
      • Lu G.
      • Ades A.E.
      • White I.R.
      Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies.
      ,
      • Dias S.
      • Welton N.J.
      • Caldwell D.M.
      • Ades A.E.
      Checking consistency in mixed treatment comparison meta-analysis.
      ]. A difference of less than 5 points was considered to be not significant.
      All the analyses were performed with WinBUGS version 1.4.3 and the results were processed using R version 4.2.0 [
      • Lunn D.J.
      • Thomas A.
      • Best N.
      • Spiegelhalter D.
      WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility.
      ,

      R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. n.d.

      ]. All the equations adopted had been published elsewhere and adapted for our analyses [

      Dias S, Ades A, Welton N, Jansen J, Sutton A. Network Meta-Analysis for Decision-Making | Wiley. 1st Edition. John Wiley and Sons; 2018.

      ]. Internal validity of eligible studies was assessed with Review Manager version 5.4 according to the Cochrane guidelines [

      Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration; n.d.

      ].
      The study protocol was registered on PROSPERO (ID: CRD42020211971).

      Results

      Overall, 139 studies were included in our networks for a total of 118 different therapeutic regimens for the first-line and 33 for the following lines (all references in Supplementary Methods). The study selection process is summarized in Supplementary Fig. 1. First-line networks included 125 RCT whilst advanced-line networks included 33 RCT (Supplementary Tables 1–2 and Supplementary Figs. 2–7). The first-line PFS, OS and ORR networks were based on trial-level data deriving from 31,203, 22,848 and 36,015 patients, respectively. Advanced-line PFS, OS and ORR networks were based on trial-level data deriving from 7,185, 6,467 and 11,321 patients, respectively. Eighteen (56.2 %) advanced-line studies included a proportion of first-line patients and were also incorporated in first-line networks. Ninety-two (73.6 %) first-line studies also included ER + metastatic BC (MBC) (minimum–maximum [min–max] range of ER + MBC patients in the included studies: 14 % − 91 %). Twenty-one (63.6 %) advanced-line studies involved ER + MBC (min–max range: 26.1 % − 83.3 %), as well. Considering all included studies, the median proportion of patients with ER + MBC per treatment arm was 59.5 % (interquartile range [IQR] of 35.3 % – 74.0 %) and endocrine therapy (ET) for the metastatic disease had been administered before study treatments in 37 % of the cases in at least a proportion of patients per treatment arm. Four (2.9 %) trials explicitly included a proportion of PD-L1+ tumors [
      • Cortes J.
      • Cescon D.W.
      • Rugo H.S.
      • Nowecki Z.
      • Im S.-A.
      • Yusof M.M.
      • et al.
      KEYNOTE-355: Randomized, double-blind, phase III study of pembrolizumab + chemotherapy versus placebo + chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer.
      ,
      • Schmid P.
      • Adams S.
      • Rugo H.S.
      • Schneeweiss A.
      • Barrios C.H.
      • Iwata H.
      • et al.
      Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer.
      ,
      • Winer E.P.
      • Lipatov O.
      • Im S.-A.
      • Goncalves A.
      • Muñoz-Couselo E.
      • Lee K.S.
      • et al.
      Pembrolizumab versus investigator-choice chemotherapy for metastatic triple-negative breast cancer (KEYNOTE-119): a randomised, open-label, phase 3 trial.
      ,
      • Miles D.
      • Gligorov J.
      • André F.
      • Cameron D.
      • Schneeweiss A.
      • Barrios C.
      • et al.
      Primary results from IMpassion131, a double-blind, placebo-controlled, randomised phase III trial of first-line paclitaxel with or without atezolizumab for unresectable locally advanced/metastatic triple-negative breast cancer.
      ], 5 (3.6 %) trials included, exclusively or in part, gBRCA-mut MBC [
      • Tutt A.
      • Tovey H.
      • Cheang M.C.U.
      • Kernaghan S.
      • Kilburn L.
      • Gazinska P.
      • et al.
      A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
      ,
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ,
      • Bardia A.
      • Hurvitz S.A.
      • Tolaney S.M.
      • Loirat D.
      • Punie K.
      • Oliveira M.
      • et al.
      Sacituzumab Govitecan in Metastatic Triple-Negative Breast Cancer.
      ,
      • Diéras Véronique
      • Han H.S.
      • Kaufman B.
      • Wildiers H.
      • Friedlander M.
      • Ayoub J.-P.
      • et al.
      Veliparib with carboplatin and paclitaxel in BRCA-mutated advanced breast cancer (BROCADE3): a randomised, double-blind, placebo-controlled, phase 3 trial.
      ], and 3 (2.2 %) trials included patients with PIK3CA/PTEN/AKT-altered tumors [
      • Kim S.-B.
      • Dent R.
      • Im S.-A.
      • Espié M.
      • Blau S.
      • Tan A.R.
      • et al.
      Ipatasertib plus paclitaxel versus placebo plus paclitaxel as first-line therapy for metastatic triple-negative breast cancer (LOTUS): a multicentre, randomised, double-blind, placebo-controlled, phase 2 trial.
      ,

      Dent R, Kim S-B, Oliveira M, Barrios C, O’Shaughnessy J, Isakoff SJ, et al. Abstract GS3-04: Double-blind placebo (PBO)-controlled randomized phase III trial evaluating first-line ipatasertib (IPAT) combined with paclitaxel (PAC) for PIK3CA/AKT1/PTEN-altered locally advanced unresectable or metastatic triple-negative breast cancer (aTNBC): primary results from IPATunity130 Cohort A. Cancer Res 2021;81:GS3-04. https://doi.org/10.1158/1538-7445.SABCS20-GS3-04.

      ,
      • Schmid P.
      • Abraham J.
      • Chan S.
      • Wheatley D.
      • Brunt A.M.
      • Nemsadze G.
      • et al.
      Capivasertib Plus Paclitaxel Versus Placebo Plus Paclitaxel As First-Line Therapy for Metastatic Triple-Negative Breast Cancer: The PAKT Trial.
      ]. Overall, 49 (35.3 %) studies evaluated TT, with or without CT while the remaining 91 (64.7 %) only compared different CT regimens or schedules. Study characteristics are extensively reported in Supplementary Tables 1–2.
      To graphically visualize the results, paclitaxel + bevacizumab was chosen as the reference treatment for the first-line, being a potentially good compromise between mono-CT and poly-CT, while capecitabine was the reference for the advanced lines, as in numerous RCT.
      Inconsistency and consistency models’ DIC for each network are reported in the Supplementary Methods. A marginally significant inconsistency was observed only for the first-line ORR network (a difference between the DIC of consistency and inconsistency models of 6.49).

      First-line networks

      A total of 92 different regimens were compared for PFS/TTP. None was deemed to be significantly superior to paclitaxel + bevacizumab, including poly-CT regimens such as anthracycline/taxane combinations or other anthracycline-based regimens and taxanes + platinum agents (Fig. 1). At the same time, paclitaxel + bevacizumab was not likely to be superior to any first-line mono-CT, such as weekly paclitaxel, pegylated liposomal doxorubicin (PLD) or nab-paclitaxel (Fig. 1). Sixty-four regimens included in the PFS/TTP analysis were also comparable in terms of OS with consistent results (Fig. 2). In contrast, when compared to other 117 regimens, paclitaxel + bevacizumab was likely to be significantly associated with superior ORR than several poly-CT regimens like cyclophosphamide + methotrexate + 5-fluorouracil (CMF) (odds ratio [OR]: 6.57, 95 % credible intervals [CrI]: 2.05–21.63), FEC (OR: 4.44, 95 %CrI: 1.33–15.23), ixabepilone + capecitabine (OR: 3.45, 95 %CrI: 1.02–12.03) or capecitabine + bevacizumab (OR: 2.47, 95 %CrI: 1.08–5.73) (Supplementary Fig. 8). The reference regimen was associated with lower ORR only when compared to docetaxel + cisplatin (OR: 0.13, 95 %CrI: 0.02–0.82). Additionally, paclitaxel + bevacizumab was likely to be significantly superior to several mono-CT, like PLD (OR: 6.89, 95 %CrI: 1.30–36.67) or capecitabine (OR: 5.47, 95 %CrI: 1.89–16.17) (Supplementary Fig. 8).
      Figure thumbnail gr1
      Fig. 1Forest plot of all first-line regimens compared to paclitaxel + bevacizumab in terms of PFS/TTP. The figure has been split in two panels (A and B) to improve its readability. The forest plot includes the log hazard ratios (HR) of each treatment versus paclitaxel + bevacizumab. Central dots represent posterior medians; thin lines represent 95 % credible intervals (CrI), while thicker ones represent 80 % CrI. Log scale was adopted to graphically represent the 95 % CrI. The first column of values on the right reports the log HR with 95 % CrI, the second column reports HR with 95 % CrI. Statistically significant results according to Bayesian posterior medians and 95 % credible intervals are highlighted by asterisks. 5-FU: 5-fluorouracil; AC: doxorubicin + cyclophosphamide; ATEZO: atezolizumab; AXI: axitinib; BEVA: bevacizumab 10 mg/kg IV q2w; BEVA 7.5: bevacizumab 7.5 mg/kg IV q3w; BEVA 15: bevacizumab 15 mg/kg IV q3w; BMF: bendamustine + methotrexate + 5-FU; CAPE: capecitabine; CAPE LD: capecitabine low dose/metronomic; CAPI: capivasertib; CYC: cyclophosphamide; CARBO: carboplatin; CIS: cisplatin; CME: cyclophosphamide + mitoxantrone + etoposide; CMF: cyclophosphamide + methotrexate + 5-FU; CT: chemotherapy; DOC: docetaxel; DOXO: doxorubicin 75 mg/m2 q3w; DOXO 60: doxorubicin 60 mg/m2 q3w; EC: epirubicin + cyclophosphamide; EPI: epirubicin; ERI: eribulin; EVE: everolimus; FAC: 5-FU + doxorubicin + cyclophosphamide; FEC: 5-FU + epirubicin 75 mg/m2 + cyclophosphamide q3w; FEC 100: 5-FU + epirubicin 100 mg/m2 + cyclophosphamide q3w; FEC 50: 5-FU + epirubicin 50 mg/m2 + cyclophosphamide q3w; GEM: gemcitabine; INI: iniparib; IPA: ipatasertib; IXA: ixabepilone; LONI: lonidamine; MITO C: mitomycin C; MITOX: mitoxantrone; MMM: mitoxantrone + mitomycin C + methotrexate; MOT: motesanib; NAB-PAC: nab-paclitaxel 260/300 mg/m2 q3w; NAB-PAC 100 qw: nab-paclitaxel 100 mg/m2 q3/4w; NAB-PAC qw: nab-paclitaxel 125/130/150 mg/m2 q3/4w; NPLD: non-pegylated liposomal doxorubicin; OLA: olaparib; PAC within schedules and PAC q3w: paclitaxel 175 mg/m2 q3w; PAC 210: paclitaxel 210 mg/m2 q3w; PAC 250: paclitaxel 250 mg/m2 q3w; PAC qw: paclitaxel 80/90 mg/m2 q3/4w; PEMBRO: pembrolizumab; PEMBRO + CT: pembrolizumab + nab-paclitaxel/paclitaxel/carboplatin + gemcitabine; PLD: pegylated liposomal doxorubicin 50 mg/m2 IV q4w; PLD 40 mg: peghylated liposomal doxorubicin 40 mg/m2 IV q4w; PLD 60 mg: pegylated liposomal doxorubicin 60 mg/m2 IV q4w; RAMU: ramucirumab; SCT: stem cell transplant; SOR: sorafenib; SUN: sunitinib; TALAZO: talazoparib; TMZ: temozolomide; TREB3: trebananib 3 mg/kg qw; TREB10: trebananib 10 mg/kg qw; TRILA: trilaciclib 240 mg/m2 d1,8 IV q3w; TRILA 1,2,8,9: Trilaciclib 240 mg/m2 d1,2,8,9 IV q3w; VELI: veliparib; VINFLU: vinflunine; VNR: vinorelbine; IV: intravenous; d: day; qw: weekly schedule; q2w: biweekly schedule; q3w: threeweekly schedule; q3/4w: 3 weeks out of 4 schedule; q4w: every-4-weeks schedule; →: followed by; DD: dose dense; HD: high dose; TI: time intensive; *: statistically significant results.
      Figure thumbnail gr2
      Fig. 2Forest plot of all first-line regimens compared to paclitaxel + bevacizumab in terms of OS. The forest plot includes the log hazard ratios (HR) of each treatment versus paclitaxel + bevacizumab. Central dots represent posterior medians; thin lines represent 95 % credible intervals (CrI), while thicker ones represent 80 % CrI. Log scale was adopted to graphically represent the 95 % credible intervals. The first column of values on the right reports the log HR with 95 % credible intervals, the second column reports HR with 95 % credible intervals. Statistically significant results according to Bayesian posterior medians and 95 % credible intervals are highlighted by asterisks. 5-FU: 5-fluorouracil; ATEZO: atezolizumab; BEVA: bevacizumab 10 mg/kg IV q2w; BEVA 7.5: bevacizumab 7.5 mg/kg IV q3w; BEVA 15: bevacizumab 15 mg/kg IV q3w; CAPE: capecitabine; CAPE LD: capecitabine low dose/metronomic; CAPI: capivasertib; CYC: cyclophosphamide; CARBO: carboplatin; CIS: cisplatin; CMF: cyclophosphamide + methotrexate + 5-FU; DOC: docetaxel; DOXO: doxorubicin 75 mg/m2 q3w; DOXO 60: doxorubicin 60 mg/m2 q3w; ERI: eribulin; EVE: everolimus; FEC: 5-FU + epirubicin 75 mg/m2 + cyclophosphamide q3w; FEC 100: 5-FU + epirubicin 100 mg/m2 + cyclophosphamide q3w; FEC 50: 5-FU + epirubicin 50 mg/m2 + cyclophosphamide q3w; GEM: gemcitabine; INI: iniparib; IPA: ipatasertib; IXA: ixabepilone; NAB-PAC: nab-paclitaxel 260/300 mg/m2 q3w; NAB-PAC 100 qw: nab-paclitaxel 100 mg/m2 q3/4w; NAB-PAC qw: nab-paclitaxel 125/130/150 mg/m2 q3/4w; NPLD: non-peghylated liposomal doxorubicin; OLA: olaparib; PAC within schedules and PAC q3w: paclitaxel 175 mg/m2 q3w; PAC qw: paclitaxel 80/90 mg/m2 q3/4w; PLD: peghylated liposomal doxorubicin 50 mg/m2 q4w; SOR: sorafenib; SUN: sunitinib; TALAZO: talazoparib; TMZ: temozolomide; TRILA: Trilaciclib 240 mg/m2 d1,8 IV q3w; TRILA 1,2,8,9: Trilaciclib 240 mg/m2 d1,2,8,9 IV q3w; VELI: veliparib; VINFLU: vinflunine; VNR: vinorelbine; qw: weekly schedule; q2w: biweekly schedule; q3w: threeweekly schedule; q3/4w: 3 weeks out of 4 schedule; q4w: every-4-weeks schedule; →: followed by; DD: dose dense; HD: high dose.
      Notably, similar results were observed with weekly paclitaxel, as well as with atezolizumab + nab-paclitaxel or pembrolizumab + CT in PD-L1+ TNBC, and with olaparib or talazoparib in gBRCA-mut TNBC (data not shown). Direct comparisons based on indirect evidence of atezolizumab + nab-paclitaxel vs pembrolizumab + CT or talazoparib vs olaparib did not show significant differences, respectively (Supplementary Table 3).

      Advanced lines networks

      Seventeen regimens entered the advanced-line network of PFS/TTP. Compared to capecitabine, sacituzumab govitecan (HR: 0.51, 95 %CrI: 0.36–0.72), ixabepilone + capecitabine (HR: 0.75, 95 %CrI: 0.64–0.88), and talazoparib (HR: 0.67, 95 %CrI: 0.46–0.97) were likely to be superior. Conversely, commonly used alternatives like eribulin and vinorelbine, or olaparib in gBRCA-mut tumors, showed comparable results (Fig. 3). T-DXd in HER2-low TNBC showed a numerically similar result to sacituzumab govitecan, although not significant (Fig. 3).
      Figure thumbnail gr3
      Fig. 3Forest plot of all advanced lines regimens compared to capecitabine in terms of PFS/TTP. The forest plot includes the log hazard ratios (HR) of each treatment versus capecitabine. Central dots represent posterior medians; thin lines represent 95 % credible intervals (CrI), while thicker ones represent 80 % CrI. Log scale was adopted to graphically represent the 95 % CrI. The first column of values on the right reports the log HR with 95 % CrI, the second column reports HR with 95 % CrI. Statistically significant results according to Bayesian posterior medians and 95 % credible intervals are highlighted by asterisks. BEVA: bevacizumab 10 mg/kg q2w; CAPE: capecitabine; ENZA: enzastaurin; ERI: eribulin; GEM: gemcitabine; IXA: ixabepilone; OLA: olaparib; PEMBRO: pembrolizumab; RAMU: ramucirumab; T-DXd: trastuzumab deruxtecan; SAC GOV: sacituzumab govitecan; SOR: sorafenib; SUN: sunitinib; TALAZO: talazoparib; TPC: treatment of physician’s choice (mostly CAPE, ERI, or VNR); VINFLU: vinflunine; VNR: vinorelbine; qw: weekly schedule; q2w: biweekly schedule; q3w: three weekly schedule; *: statistically significant results.
      Fifteen of the 16 previous regimens entered the network of OS. Only sacituzumab govitecan was significantly superior to capecitabine (HR: 0.52, 95 %CrI: 0.38–0.72) (Fig. 4).
      Figure thumbnail gr4
      Fig. 4Forest plot of all second/further-lines regimens compared to capecitabine in terms of OS. The forest plot includes the log hazard ratios (HR) of each treatment versus capecitabine. Central dots represent posterior medians; thin lines represent 95 % credible intervals (CrI), while thicker ones represent 80 % CrI. Log scale was adopted to graphically represent the 95 % credible intervals. The first column of values on the right reports the log HR with 95 % credible intervals, the second column reports HR with 95 % credible intervals. Statistically significant results according to Bayesian posterior medians and 95 % credible intervals are highlighted by asterisks. BEVA: bevacizumab 10 mg/kg q2w; CAPE: capecitabine; ENZA: enzastaurin; ERI: eribulin; IXA: ixabepilone; OLA: olaparib; PEMBRO: pembrolizumab; RAMU: ramucirumab; SAC GOV: sacituzumab govitecan; SOR: sorafenib; SUN: sunitinib; TALAZO: talazoparib; TPC: treatment of physician’s choice (mostly CAPE, ERI, or VNR); VINFLU: vinflunine; VNR: vinorelbine; qw: weekly schedule; q2w: biweekly schedule; q3w: threeweekly schedule; *: statistically significant result.
      When considering ORR, 33 treatments were comparable. Among clinically relevant regimens, only sacituzumab govitecan (OR: 3.34, 95 %CrI: 1.16–9.45) and ixabepilone + capecitabine (OR: 3.19, 95 %CrI: 2.24–4.61) were associated with better ORR than capecitabine (Supplementary Fig. 9). Importantly, sacituzumab govitecan showed the best survival results, being significantly superior to many common therapies, including eribulin and ixabepilone + capecitabine (in PFS and OS), vinorelbine (in PFS), pembrolizumab in PD-L1+ mTNBC (in PFS and OS), olaparib and talazoparib in gBRCA-mut tumors (in OS) (not shown). No significant difference was observed in PFS/TTP and OS between sacituzumab govitecan and T-DXd in HER2-low tumors, as well as between olaparib and talazoparib in gBRCA-mut patients (Supplementary Table 3).

      Treatment rankings based on SUCRA and safety

      We estimated separate rankings for the first and subsequent lines based on SUCRA values (Supplementary Figs. 10–11). For practical purposes, we refined the rankings by including only the currently US National Comprehensive Cancer Network (NCCN)-recommended [

      National Comprehensive Cancer Network. NCCN Guidelines for Breast Cancer, vers.4.2022 n.d. https://www.nccn.org/professionals/physician_gls/pdf/breast.pdf (accessed June 28, 2022).

      ] and/or FDA/EMA-approved treatments, as of June 2022 (Table 1). Nevertheless, the ranking includes also T-DXd, which is likely to be approved in the near future for HER2-low TNBC.
      Table 1Ranking of available regimens according to treatment line.
      FIRST-LINE
      TreatmentSUCRAPFS/TTP RankingTreatmentSUCRAOS RankingTreatmentSUCRAORR Ranking
      DOC + CIS0.91351NAB-PAC q3w + CARBO0.76601DOC + CIS0.98001
      CIS + PAC q3w0.81742PEMBRO + CT0.73552CIS + PAC q3w0.88962
      NAB-PAC q3w + CARBO0.75373DOC + CIS0.72683NAB-PAC qw + ATEZO0.73263
      DOXO + DOC0.71474CARBO + GEM0.69484PAC qw + BEVA0.69804
      NAB-PAC qw + ATEZO0.66965CAPE + BEVA0.67975GEM + DOC0.68935
      CIS + GEM0.65356TALAZO0.66386NAB-PAC q3w + CARBO0.68336
      PAC qw + BEVA0.64307ERI0.64557PAC qw0.68027
      NPLD + CYC0.63698IXA q3w + CAPE0.63698CAPE + PAC q3w0.67798
      AC0.63659OLA0.62599DOXO + DOC0.61989
      DOXO + PAC q3w0.629110PAC qw + BEVA0.606010EPI + PAC q3w0.612610
      FEC 1000.618411S10.594011CAPE + DOC0.593611
      VNR + CAPE0.615812PAC qw0.568312OLA0.576612
      PEMBRO + CT0.613213CAPE0.564913NAB-PAC q3w0.572213
      TALAZO0.608014PLD0.556714DOXO + PAC q3w0.563214
      CAPE + DOC0.590715DOXO 600.516515NAB-PAC qw0.545515
      OLA0.576416FEC 1000.508916CARBO + PAC q3w0.545316
      CARBO + GEM0.561317NAB-PAC 100 qw + ATEZO0.489517GEM + PAC q3w0.525817
      FAC0.525018FEC0.452218TALAZO0.521418
      GEM + DOC0.499819CIS + GEM0.414519NPLD + CYC0.518919
      CAPE + BEVA0.494220VNR + CAPE0.406820VNR + CAPE0.511020
      GEM + PAC q3w0.485321CARBO + PAC q3w0.385221EPI + DOC0.508721
      EPI + PAC q3w0.478522CAPE LD + DOC0.381522AC0.484522
      NAB-PAC qw0.476023CMF0.364123EC0.425523
      CAPE + PAC q3w0.475524GEM + PAC q3w0.356324DOXO0.425024
      IXA q3w + CAPE0.447225NPLD0.341925DOC0.406525
      EC0.443726GEM + DOC0.332726NAB-PAC q3w0.404826
      CARBO + PAC q3w0.412427NAB-PAC qw0.312027VNR + GEM0.383527
      FEC0.409428CAPE + DOC0.289628CARBO0.377028
      NAB-PAC q3w0.407429PAC q3w0.257829CAPE + BEVA0.359329
      DOXO0.391430DOC0.180630FAC0.349930
      PLD0.385531CARBO + GEM0.332731
      PAC qw0.383732CIS + GEM0.305232
      EPI + DOC0.356933FEC 1000.298133
      PAC q3w0.324734PAC q3w0.278734
      DOC0.313435IXA + CAPE0.269635
      CAPE0.312436VNR0.247936
      CMF0.266137PLD0.197937
      FEC0.188038
      CAPE0.142039
      CMF0.107040
      SECOND/FURTHER LINES
      TreatmentSUCRAPFS/TTP RankingTreatmentSUCRAOS RankingTreatmentSUCRAORR Ranking
      SAC GOV0.9651SAC GOV0.95951CAPE + DOC0.80821
      T-DXd*0.85812T-DXd*0.91102NAB-PAC q3w0.79932
      TALAZO0.81583ERI0.64873IXA q3w + CAPE0.75183
      OLA0.75554PEMBRO0.64434SAC GOV0.75044
      IXA q3w + CAPE0.74815IXA q3w + CAPE0.59625PAC qw0.71515
      GEM + VNR0.47816TALAZO0.53616CAPE LD + DOC0.70886
      CAPE0.42637OLA0.43667DOC0.63287
      ERI0.32818CAPE0.37118T-DXd*0.55168
      VNR0.13469GEM + DOC0.53599
      PEMBRO0.115210GEM + VNR0.503810
      PAC q3w0.470911
      OLA0.451912
      DOC0.416713
      CAPE + SUN0.406014
      TALAZO0.386215
      CAPE0.370416
      ERI0.353217
      VNR0.325118
      PEMBRO0.253919
      CMF: cyclophosphamide + methotrexate + 5-FU; 5-FU: 5-fluorouracil; BEVA: bevacizumab; ATEZO: atezolizumab; EPI: epirubicin; FEC: 5-FU + epirubicin + cyclophosphamide; FAC: 5-FU + doxorubicin + cyclophosphamide; DOXO: doxorubicin; DOC: docetaxel; PAC: paclitaxel; PLD: pegylated liposomal doxorubicin; NPLD: non-pegylated liposomal doxorubicin; CYC: cyclophosphamide; TALAZO: talazoparib; OLA: olaparib; IXA: ixabepilone; CAPE: capecitabine; GEM: gemcitabine; VNR: vinorelbine; ERI: eribulin; PEMBRO: pembrolizumab; NAB-PAC: nab-paclitaxel; CT: chemotherapy; CIS: cisplatin; CARBO: carboplatin; AC: doxorubicin + cyclophosphamide; EC: epirubicin + cyclophosphamide; SAC GOV: sacituzumab govitecan; ERI: eribulin; T-DXd: trastuzumab deruxtecan; LD: low dose; q3w: three weekly schedule; qw: weekly schedule; *: although still not approved, T-DXd is the new standard of care for HER2-low breast cancer and might enter quickly the therapeutic armamentarium also for this subset of triple negative tumors.
      A formal comparison of toxicities could not be carried out due to the heterogeneity of reporting side effects across trials. However, we extensively described the proportion of G3-5 AEs observed for each regimen in ≥ 2 % of study patients (Supplementary Tables 4–5). As expected, combination CT presented with the highest proportion of moderate/severe AEs. Neverthless, G3-5 neutropenia and leucopenia were relatively frequent also with mono-CT. Doxorubicin, docetaxel, vinorelbine, paclitaxel and gemcitabine were associated with the highest rates of severe alopecia, as like capecitabine with hand-foot syndrome and doxorubicin with stomatitis and febrile neutropenia. Anthracycline-based regimens, excluding pegylated and non-pegylated liposomal doxorubicin, were typically associated with cardiotoxicity. Bevacizumab-containing regimens typically showed proteinuria and hypertension. Although ICI-based combinations (pembrolizumab + CT and atezolizumab + nab-paclitaxel) are usually associated with immune-related AEs, including thyroiditis, pneumonitis, skin alterations and colitis, G3-5 rates of such toxicities were not observed, except for pneumonitis, hypo/hyperthyroidism and skin reactions (all < 2 %). PARPi were characterized by an overall better toxicity profile than mono-CT, except for partially comparable hematologic toxicities. Finally, sacituzumab govitecan and T-DXd presented with a chemo-like toxicity profile, with diarrhea, nausea/vomiting, hematotoxicity and fatigue as most frequent moderate/severe AEs. Both drugs can frequently induce alopecia (46.0 % and 37.7 %) and T-DXd was associated with interstitial lung disease/pneumonitis (<20 % cases), although the vast majority of cases was mild or moderate [
      • Bardia A.
      • Hurvitz S.A.
      • Tolaney S.M.
      • Loirat D.
      • Punie K.
      • Oliveira M.
      • et al.
      Sacituzumab Govitecan in Metastatic Triple-Negative Breast Cancer.
      ,
      • Modi S.
      • Jacot W.
      • Yamashita T.
      • Sohn J.
      • Vidal M.
      • Tokunaga E.
      • et al.
      Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer.
      ].

      Risk of bias analysis

      Internal validity of eligible studies was assessed with a risk of bias (RoB) analysis as recommended by the Cochrane Handbook for Systematic Reviews of Interventions [

      Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration; n.d.

      ]. There were no specific concerns regarding 5/9 RoB domains (Supplementary Fig. 12). However, the RoB was high in approximately half trials included with respect to blinding of participants and personnel and outcome assessments, as well as regarding the selective reporting bias (Supplementary Figs. 12–13).

      Discussion

      We performed multiple Bayesian NMA to identify the best treatment for the first and subsequent lines of mTNBC, in terms of PFS/TTP, OS and ORR. We found that among the most commonly adopted first-line regimens, there were no substantial differences in terms of PFS/TTP and OS whereas paclitaxel + bevacizumab was associated with better ORR than numerous mono-CT and several poly-CT, with less G3-5 AEs compared to CT combinations. Docetaxel + cisplatin was the only regimen significantly favored in terms of ORR, at the cost of higher toxicities (Fig. 1, Fig. 2, Supplementary Fig. 8, Table 1, Supplementary Tables 4–5). These results, taken together with the available evidence, despite conflicting results on its potential OS benefit [
      • Giuliano M.
      • Schettini F.
      • Rognoni C.
      • Milani M.
      • Jerusalem G.
      • Bachelot T.
      • et al.
      Endocrine treatment versus chemotherapy in postmenopausal women with hormone receptor-positive, HER2-negative, metastatic breast cancer: a systematic review and network meta-analysis.
      ,
      • Delaloge S.
      • Pérol D.
      • Courtinard C.
      • Brain E.
      • Asselain B.
      • Bachelot T.
      • et al.
      Paclitaxel plus bevacizumab or paclitaxel as first-line treatment for HER2-negative metastatic breast cancer in a multicenter national observational study.
      ,
      • Dieras V.
      • Pop S.
      • Berger F.
      • Dujaric M.-E.
      • Beuzeboc P.
      • Escalup L.
      • et al.
      First-line Bevacizumab and Paclitaxel for HER2-negative Metastatic Breast Cancer: A French Retrospective Observational Study.
      ,
      • Smith I.E.
      • Pierga J.-Y.
      • Biganzoli L.
      • Cortés-Funes H.
      • Thomssen C.
      • Pivot X.
      • et al.
      First-line bevacizumab plus taxane-based chemotherapy for locally recurrent or metastatic breast cancer: safety and efficacy in an open-label study in 2,251 patients.
      ,
      • Bramati A.
      • Girelli S.
      • Torri V.
      • Farina G.
      • Galfrascoli E.
      • Piva S.
      • et al.
      Efficacy of biological agents in metastatic triple-negative breast cancer.
      ,
      • Miles D.W.
      • Diéras V.
      • Cortés J.
      • Duenne A.-A.
      • Yi J.
      • O’Shaughnessy J.
      First-line bevacizumab in combination with chemotherapy for HER2-negative metastatic breast cancer: pooled and subgroup analyses of data from 2447 patients.
      ], support paclitaxel + bevacizumab as a suitable first-line option. At the same time, the lack of a clear superiority in survival endpoints, especially in comparison to mono-CT options, makes the combination of paclitaxel with bevacizumab a valuable regimen mostly when a rapid and/or potent tumor shrinkage is the required therapeutic goal (e.g. high tumor burden, visceral crisis). In this case, the better safety profile of this regimen compared to those of poly-CT plays in favor of paclitaxel + bevacizumab. To note, this regimen is no longer available in the USA and, where available (e.g. Europe), its higher costs compared to mono-CTs and numerous poly-CT regimens should be properly taken into account at the moment of therapeutic decision-making.
      Interestingly, platinum-based regimens, especially in combination with taxanes, resulted in very high SUCRA values for all endpoints, but their potential benefits should be weighed against their higher toxicities (e.g. nausea/vomiting, peripheral neuropathy, hemato-toxicity) and patients’ inclination to receive the appropriate supportive care, when required [
      • Roila F.
      • Molassiotis A.
      • Herrstedt J.
      • Aapro M.
      • Gralla R.J.
      • Bruera E.
      • et al.
      2016 MASCC and ESMO guideline update for the prevention of chemotherapy- and radiotherapy-induced nausea and vomiting and of nausea and vomiting in advanced cancer patients.
      ,
      • Klastersky J.
      • de Naurois J.
      • Rolston K.
      • Rapoport B.
      • Maschmeyer G.
      • Aapro M.
      • et al.
      Management of febrile neutropaenia: ESMO Clinical Practice Guidelines †.
      ,
      • Aapro M.
      • Beguin Y.
      • Bokemeyer C.
      • Dicato M.
      • Gascón P.
      • Glaspy J.
      • et al.
      Management of anaemia and iron deficiency in patients with cancer: ESMO Clinical Practice Guidelines.
      ].
      A viable alternative in patients unfit for/unwilling to receive poly-CT, or with low tumor burden, is the use of mono-CT. The best-ranked single-agents in terms of both PFS and OS appeared to be weekly paclitaxel and PLD (Table 1). Single-agent carboplatin is another option, with many mono-CTs being not significantly superior to it in terms of ORR (data not shown), including taxanes (coherently with the TNT trial) [
      • Tutt A.
      • Tovey H.
      • Cheang M.C.U.
      • Kernaghan S.
      • Kilburn L.
      • Gazinska P.
      • et al.
      A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
      ]. Yet, it could not be included in the PFS and OS networks, limiting the evidence to support it as a clear standard. The choice among one of those drugs should be properly discussed and adapted to patients’ preferences, especially with regard to their different toxicity profile and different administration schedule.
      Notably, talazoparib and olaparib in gBRCA-mut patients were among the highest ranked available treatments, without superior mono-CTs (Table 1). Conversely, many multiagent regimens, although not selectively tested in gBRCA-mut patients, were associated to better outcomes (Table 1). Therefore, PARPi might reasonably represent an optimal first-line option in gBRCA-mut cases when no combinations are strictly required. Notably, signals of activity beyond BRCA1/2 mutant TNBC might lead to an expansion of their therapeutic indication, if further confirmed [
      • Schettini F.
      • Corona S.P.
      • Giudici F.
      • Strina C.
      • Sirico M.
      • Bernocchi O.
      • et al.
      Clinical, Radiometabolic and Immunologic Effects of Olaparib in Locally Advanced Triple Negative Breast Cancer: The OLTRE Window of Opportunity Trial.
      ,
      • Tung N.M.
      • Robson M.E.
      • Ventz S.
      • Santa-Maria C.A.
      • Nanda R.
      • Marcom P.K.
      • et al.
      TBCRC 048: Phase II Study of Olaparib for Metastatic Breast Cancer and Mutations in Homologous Recombination-Related Genes.
      ,
      • Fasching P.A.
      • Link T.
      • Hauke J.
      • Seither F.
      • Jackisch C.
      • Klare P.
      • et al.
      Neoadjuvant paclitaxel/olaparib in comparison to paclitaxel/carboplatinum in patients with HER2-negative breast cancer and homologous recombination deficiency (GeparOLA study).
      ,
      • Schettini F.
      • Giudici F.
      • Bernocchi O.
      • Sirico M.
      • Corona S.P.
      • Giuliano M.
      • et al.
      Poly (ADP-ribose) polymerase inhibitors in solid tumours: Systematic review and meta-analysis.
      ]. Importantly, recently published positive results on adjuvant olaparib in gBRCA-mut early-stage TNBC already lead to FDA approval in this subset [
      • Tutt A.N.J.
      • Garber J.E.
      • Kaufman B.
      • Viale G.
      • Fumagalli D.
      • Rastogi P.
      • et al.
      Adjuvant Olaparib for Patients with BRCA1- or BRCA2-Mutated Breast Cancer.
      ]. Whether this might affect the efficacy of first-line PARPi in case of relapse will be a matter of debate in the next years.
      Regarding PD-L1+ mTNBC, no treatment was significantly superior to atezolizumab + nab-paclitaxel for all the 3 endpoints investigated and to pembrolizumab + CT in terms of PFS/TTP and OS (data not shown). These combinations also showed a relatively favorable and manageable toxicity profile compared to numerous poly-CT regimens (Supplementary Tables 4–5). Overall, based on these results and considering that ICI + CT regimens are limited to the first-line setting, atezolizumab + nab-paclitaxel or pembrolizumab + CT might be the preferred upfront treatment for PD-L1+ mTNBC. Importantly, the comparison between the two regimens in our study was not statistically significant but pembrolizumab + CT was favored over more first-line regimens and obtained the best position in the OS SUCRA-based ranking. In addition, atezolizumab + nab-paclitaxel pivotal trial’s result was formally negative due to its hierarchical testing plan [
      • Schmid P.
      • Adams S.
      • Rugo H.S.
      • Schneeweiss A.
      • Barrios C.H.
      • Iwata H.
      • et al.
      Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer.
      ]. Furthermore, following the manufacturer’s request, the FDA (but not EMA) also recently withdrew its indication for atezolizumab in this setting [
      • Miles D.
      • Gligorov J.
      • André F.
      • Cameron D.
      • Schneeweiss A.
      • Barrios C.
      • et al.
      Primary results from IMpassion131, a double-blind, placebo-controlled, randomised phase III trial of first-line paclitaxel with or without atezolizumab for unresectable locally advanced/metastatic triple-negative breast cancer.
      ,

      Schmid P, Cortes J, Dent R, Pusztai L, McArthur H, Kümmel S, et al. VP7-2021: KEYNOTE-522: Phase III study of neoadjuvant pembrolizumab + chemotherapy vs. placebo + chemotherapy, followed by adjuvant pembrolizumab vs. placebo for early-stage TNBC. Annals of Oncology 2021;0. https://doi.org/10.1016/j.annonc.2021.06.014.

      ,

      Roche Group Media Relations. Roche provides update on Tecentriq US indication for PD-L1-positive, metastatic triple-negative breast cancer. Roche Media Releases n.d. https://www.roche.com/media/releases/med-cor-2021-08-27.htm (accessed October 18, 2021).

      ].
      In case of concomitant PD-L1+/gBRCA-mut tumors, the evidence is still too scarce to draw any meaningful conclusion. Yet, PARPi can be administered also in advanced lines, while, at present, ICI-based therapy cannot. Interestingly, an ongoing phase II/III RCT of pembrolizumab + olaparib vs pembrolizumab + CT after induction with pembrolizumab + CT in mTNBC will provide additional important insights [

      Rugo H, Llombart-Cussac A, Andre F, Robson ME, Saji S, Harbeck N, et al. Abstract OT-30-01: KEYLYNK-009: A phase 2/3, open-label, randomized study of pembrolizumab plus olaparib vs pembrolizumab plus chemotherapy after induction with first-line pembrolizumab plus chemotherapy in patients with locally recurrent inoperable or m…. Cancer Research 2021;81:OT-30-01-OT-30-01. https://doi.org/10.1158/1538-7445.SABCS20-OT-30-01.

      ].
      Another open question is whether to prefer first-line platinum agents or PARPi in PD-L1-negative/gBRCA-mut TNBC, considering that BRCA1/2 dysfunction has been associated with increased response to platinum agents [
      • Garutti M.
      • Pelizzari G.
      • Bartoletti M.
      • Malfatti M.C.
      • Gerratana L.
      • Tell G.
      • et al.
      Platinum Salts in Patients with Breast Cancer: A Focus on Predictive Factors.
      ]. Our results suggest that platinum-based regimens in unselected TNBC (potentially including also gBRCA-mut tumors) are likely better than PARPi in gBRCA-mut tumors on all 3 endpoints (Table 1). Importantly, roughly 20 % and 30 % of patients receiving talazoparib and olaparib in their pivotal trials, respectively, had been pretreated with platinum agents [
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ]. Hence, the sequencing might not impair PARPi’s efficacy, although those subpopulations had not progressed on platinums [
      • Litton J.K.
      • Rugo H.S.
      • Ettl J.
      • Hurvitz S.A.
      • Gonçalves A.
      • Lee K.-H.
      • et al.
      Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
      ,
      • Robson M.
      • Im S.-A.
      • Senkus E.
      • Xu B.
      • Domchek S.M.
      • Masuda N.
      • et al.
      Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
      ]. Nevertheless, the evidence to unequivocally support either strategy is still insufficient. The different toxicity profile among these agents may help guiding treatment choice according to individual patient characteristics and preferences.
      Finally, as observable in the full treatment rankings (Supplementary Fig. 10), paclitaxel + capivasertib, in PIK3CA/PTEN/AKT-altered TNBC, and carboplatin + gemcitabine + CDK4/6-inhibitor trilaciclib, in unselected mTNBC, proved to be particularly effective. In both cases, confirmatory phase III trials are ongoing and results eagerly awaited [
      • Schmid P.
      • Cortes J.
      • Robson M.E.
      • Iwata H.
      • Hegg R.
      • Nechaeva M.
      • et al.
      A phase III trial of capivasertib and paclitaxel in first-line treatment of patients with metastatic triple-negative breast cancer (CAPItello290).
      ,

      Therapeutics G. G1 Therapeutics Announces Initiation of Phase 3 Registrational Study of COSELATM (trilaciclib) in Triple-Negative Breast Cancer (TNBC). GlobeNewswire News Room 2021. https://www.globenewswire.com/news-release/2021/04/28/2218475/0/en/G1-Therapeutics-Announces-Initiation-of-Phase-3-Registrational-Study-of-COSELA-trilaciclib-in-Triple-Negative-Breast-Cancer-TNBC.html (accessed June 2, 2021).

      ], especially after the unexpected negative results of the IPATunity130 phase III trial of first-line paclitaxel + ipatasertib [

      Dent R, Kim S-B, Oliveira M, Barrios C, O’Shaughnessy J, Isakoff SJ, et al. Abstract GS3-04: Double-blind placebo (PBO)-controlled randomized phase III trial evaluating first-line ipatasertib (IPAT) combined with paclitaxel (PAC) for PIK3CA/AKT1/PTEN-altered locally advanced unresectable or metastatic triple-negative breast cancer (aTNBC): primary results from IPATunity130 Cohort A. Cancer Res 2021;81:GS3-04. https://doi.org/10.1158/1538-7445.SABCS20-GS3-04.

      ].
      Concerning advanced lines, sacituzumab govitecan was associated to the best PFS, OS and ORR results (Fig. 3, Fig. 4, Supplementary Fig. 9, Table 1). These data strongly support sacituzumab govitecan being the preferred second-line in mTNBC. At the same time, olaparib and talazoparib in gBRCA-mut tumors were favored among all other treatments in terms of PFS/TTP, with no difference between the two PARPi on all endpoints (Fig. 3, Table 1, Supplementary Table 3). Hence, in the absence of sacituzumab govitecan, a PARPi might represent the preferable second-line in gBRCA-mut mTNBC, if not previously used. In the subset of HER2-low tumors, representing ∼ 37 % of all TNBC [
      • Schettini F.
      • Chic N.
      • Brasó-Maristany F.
      • Paré L.
      • Pascual T.
      • Conte B.
      • et al.
      Clinical, pathological, and PAM50 gene expression features of HER2-low breast cancer.
      ], T-DXd provided also extremely promising results, being one of the best options in terms of all 3 endpoints. The overall comparisons led to a better positioning of sacituzumab govitecan in SUCRA-based rankings. Still, a direct comparison between the two ADCs did not show significant differences. No significant difference was also observed with PARPi, but overall results, especially in OS, were in favor of T-DXd. However, considering the low number of patients treated with T-DXd, these results have to be taken carefully. T-DXd might be well positioned either after sacituzumab govitecan or before/right after PARPi in gBRCA-mut HER2-low TNBC. Direct comparisons in RCT, especially between the two ADCs should be pursued to draw more definitive conclusions.
      Subsequent most effective mono-CT were represented by nab-paclitaxel (ORR), capecitabine (PFS/TTP) and eribulin (PFS/TTP, OS) (Table 1). The former could not enter survival networks but solid efficacy data concur in supporting its use also in taxane-pretreated patients [
      • Schettini F.
      • Giuliano M.
      • De Placido S.
      • Arpino G.
      Nab-paclitaxel for the treatment of triple-negative breast cancer: Rationale, clinical data and future perspectives.
      ,
      • Puglisi F.
      • Rea D.
      • Kroes M.A.
      • Pronzato P.
      Second-line single-agent chemotherapy in human epidermal growth factor receptor 2-negative metastatic breast cancer: A systematic review.
      ]. Whether to administer one or the others is a decision that should be individualized based on toxicity profile, drug delivery method (oral for capecitabine vs intravenous for eribulin and nab-paclitaxel), previous therapies, patients’ clinical conditions and preferences. Conversely, not many poly-CT could enter our networks and the ones that could, did not show any significant advantage on all endpoints to justify their use in advanced lines compared to mono-CT, providing also their worse toxicity profile.
      Finally, single-agent pembrolizumab appeared to be another valuable option in advanced lines. However, the results hereby reported are based on PD-L1+ mTNBC, whereas it is currently FDA-approved in patients with high microsatellite instability/mismatch repair deficiency or high tumor mutational burden [
      • Marabelle A.
      • Le D.T.
      • Ascierto P.A.
      • Di Giacomo A.M.
      • De Jesus-Acosta A.
      • Delord J.-P.
      • et al.
      Efficacy of Pembrolizumab in Patients With Noncolorectal High Microsatellite Instability/Mismatch Repair-Deficient Cancer: Results From the Phase II KEYNOTE-158 Study.
      ,
      • Marabelle A.
      • Fakih M.
      • Lopez J.
      • Shah M.
      • Shapira-Frommer R.
      • Nakagawa K.
      • et al.
      Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study.
      ].
      This study has several limitations to consider. First, several assumptions and simplifications fully disclosed in the Methods section (including forcing some links), were required to include the majority of the most relevant therapeutic options and should be carefully taken into account when evaluating this study results. Second, we did not report publication bias since the approaches developed to assess this type of bias in NMA are challenging and still present limitations that make their effectiveness often debated [
      • Chaimani A.
      • Higgins J.P.T.
      • Mavridis D.
      • Spyridonos P.
      • Salanti G.
      • Haibe-Kains B.
      Graphical Tools for Network Meta-Analysis in STATA.
      ]. However, our analysis includes most of the available literature on the topic, somewhat mitigating the impact of publication bias. Third, some results are based on specific biomarker-defined subpopulations (i.e. PD-L1+, gBRCA-mut, HER2-low or PIK3CA/PTEN/AKT-altered), hence they cannot be generalized to the totality of mTNBC. However, considering that these subgroups represent a limited proportion of TNBC and their concomitant presence is uncommon, it is likely that no RCT will ever be conducted comparing the most appropriate biomarker-based treatments within such nested subgroups. Moreover, no evidences are available to conduct specific biomarker-restricted networks. In this scenario, this study provides unique results that might be valuable for clinicians. Furthermore, while sufficient evidence suggest novel biomarker-based treatments are not effective in all TNBC patients, there is no biologic rationale or published evidence supporting lack of efficacy of previous standard treatments in biomarker-based subgroups. Fourth, ∼72 % of studies included also ER+/HER2-negative MBC. Given that mTNBC are usually considered to be more CT-sensitive than ER+/HER2-negative tumors [
      • Gennari A.
      • André F.
      • Barrios C.H.
      • Cortés J.
      • de Azambuja E.
      • DeMichele A.
      • et al.
      ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
      ,
      • Giuliano M.
      • Schettini F.
      • Rognoni C.
      • Milani M.
      • Jerusalem G.
      • Bachelot T.
      • et al.
      Endocrine treatment versus chemotherapy in postmenopausal women with hormone receptor-positive, HER2-negative, metastatic breast cancer: a systematic review and network meta-analysis.
      ], it is possible that the efficacy of several regimens might have been diluted in favor of some treatments tested in TNBC-restricted trials. At the same time, in 40 % of study treatment arms, the ER positivity proportion was < 50 %, meaning that the TN population was overall highly represented. Furthermore, in such studies, ET for the metastatic disease had been administered before CT in more than 1/3 cases, further reducing the proportion of fully endocrine-sensitive ER + MBC, which are potentially less sensitive to CT than endocrine pre-treated tumors [
      • Gennari A.
      • André F.
      • Barrios C.H.
      • Cortés J.
      • de Azambuja E.
      • DeMichele A.
      • et al.
      ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
      ,
      • Moy B.
      • Rumble R.B.
      • Come S.E.
      • Davidson N.E.
      • Di Leo A.
      • Gralow J.R.
      • et al.
      Chemotherapy and Targeted Therapy for Patients With Human Epidermal Growth Factor Receptor 2–Negative Metastatic Breast Cancer That is Either Endocrine-Pretreated or Hormone Receptor–Negative: ASCO Guideline Update.
      ,
      • Burstein H.J.
      • Somerfield M.R.
      • Barton D.L.
      • Dorris A.
      • Fallowfield L.J.
      • Jain D.
      • et al.
      Endocrine Treatment and Targeted Therapy for Hormone Receptor-Positive, Human Epidermal Growth Factor Receptor 2-Negative Metastatic Breast Cancer: ASCO Guideline Update.
      ]. To note, a certain heterogeneity in the proportion of patients with visceral/non-visceral/brain metastases among included studies was also present and should be taken into account when considering results. Fifth, the highest risks of bias were observed in the domains of the performance bias, blinding of outcome assessment and selective reporting bias. Concerning the latter, since our study was focused on 3 endpoints (i.e. PFS, OS and ORR) we decided to adopt a comprehensive approach in the RoB analysis by taking into account all endpoints concomitantly. However, the limit of this approach is that at least one of the 3 endpoints was not published in many cases (usually OS), for reasons attributable to the same study design (roughly a third were phase II RCT). However, the primary endpoint (i.e. PFS) was only partially affected by this issue (∼74 % first-line and ∼ 82 % advanced-line studies reported PFS data) and results for the three different endpoints were substantially coherent among them. Regarding blinding of outcome and performance bias, the risk was high because approximately half studies were not blinded. At a closer look, this was often justified by different ways of drug administration that made it impossible, unethical or useless to apply any blinding procedure. Those issues could not be overcome and sub-analysis on the restricted casuistry of studies with no blinding bias was unfeasible because of the impossibility of closing meaningful networks. Nevertheless, most of these studies led to the approval of regimens and schedules commonly used in clinical practice. Importantly, no substantial inconsistency was observed (Supplementary Methods). Finally, we point out that NMA share the same limitations of standard pairwise meta-analyses [
      • Bailar J.C.
      The promise and problems of meta-analysis.
      ,
      • Greenland S.
      Can meta-analysis be salvaged?.
      ], with the addition of a set of specific assumptions, including transitivity and consistency, on which a lot of research is still ongoing [

      Dias S, Ades A, Welton N, Jansen J, Sutton A. Network Meta-Analysis for Decision-Making | Wiley. 1st Edition. John Wiley and Sons; 2018.

      ,

      Chaimani A, Caldwell D, Li T, Higgins J, Salanti G. Chapter 11: Undertaking network meta-analyses. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.2 (updated February 2021). Cochrane, 2021. Available from www.training.cochrane.org/handbook. n.d.

      ,
      • Dias S.
      • Sutton A.J.
      • Ades A.E.
      • Welton N.J.
      Evidence synthesis for decision making 2: a generalized linear modeling framework for pairwise and network meta-analysis of randomized controlled trials.
      ].
      In conclusion, despite limitations, this is the first study comparing all available systemic treatments for the management of mTNBC, providing timely and methodologically reliable results. In this perspective, we propose a consensus therapeutic algorithm resumed in Fig. 5, based on our original data, literature review and main international guidelines to support daily-practice therapeutic decision-making [

      National Comprehensive Cancer Network. NCCN Guidelines for Breast Cancer, vers.4.2022 n.d. https://www.nccn.org/professionals/physician_gls/pdf/breast.pdf (accessed June 28, 2022).

      ,
      • Gennari A.
      • André F.
      • Barrios C.H.
      • Cortés J.
      • de Azambuja E.
      • DeMichele A.
      • et al.
      ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
      ,
      • Moy B.
      • Rumble R.B.
      • Come S.E.
      • Davidson N.E.
      • Di Leo A.
      • Gralow J.R.
      • et al.
      Chemotherapy and Targeted Therapy for Patients With Human Epidermal Growth Factor Receptor 2–Negative Metastatic Breast Cancer That is Either Endocrine-Pretreated or Hormone Receptor–Negative: ASCO Guideline Update.
      ,
      • Schettini F.
      • Giuliano M.
      • Lambertini M.
      • Bartsch R.
      • Pinato D.J.
      • Onesti C.E.
      • et al.
      Anthracyclines Strike Back: Rediscovering Non-Pegylated Liposomal Doxorubicin in Current Therapeutic Scenarios of Breast Cancer.
      ].
      Figure thumbnail gr5
      Fig. 5Proposed therapeutic algorithms for the first and following lines. mTNBC: metastatic triple negative breast cancer; wt: wild-type; mut: mutant; gBRCA: germline BRCA1 and/or 2; CT: chemotherapy; ICI: immune-checkpoint inhibitors (atezolizumab or pembrolizumab); PLD: pegylated liposomal doxorubicin; NPLD: non– pegylated liposomal doxorubicin; neg.: negative; +: positive; PARPi: PARP inhibitor; TMB high: tumor mutational burden ≥ 10 mutations/megabase; MSI-H: high microsatellite instability (≥30 % mutations); dMMR: dysfunctional mismatch repair; 1: indicate the same advanced lines algorithm of PD-L1+/gBRCA-mut, independently from PD-L1 status; 2: indicate the same advanced lines algorithm of PD-L1+/gBRCA-wt, independently from PD-L1 status; *: pembrolizumab should be preferred over atezolizumab, where available, considering the more methodologically solid results in its pivotal trial and the better position in the SUCRA-based overall survival ranking. Moreover, atezolizumab is no longer approved in the USA for this setting; #: still not approved in Europe, but recently FDA-approved for HER2-low metastatic breast cancer, irrespective of hormone receptor status; §: if not previously administered in first-line with chemotherapy. The algorithm is based on the results published in , Fig. 1, Fig. 2, Fig. 3, Fig. 4, Supplementary Figs. 8–9, Supplementary Tables 3–5 and results not shown for space reasons. All data were then interpreted according to the available literature reviewed plus the most updated American and European Society for Clinical/Medical Oncology (ASCO and ESMO) and US National Comprehensive Cancer Network (NCCN) guidelines.

      CRediT authorship contribution statement

      Drs F. Schettini, D. Generali, M. Giuliano and prof. S. Venturini had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Dr F. Schettini and prof S. Venturini contributed equally to this work and served as co–first authors.
      Concept and design: Drs F. Schettini, D. Generali.
      Acquisition, analysis, or interpretation of data: All authors.
      Drafting of the manuscript: Drs F. Schettini, D. Generali, A. Prat, G. Curigliano, M. Giuliano, G. Jerusalem, R. Bartsch and prof S. Venturini.
      Critical revision of the manuscript for important intellectual content: All authors.
      Statistical analysis: Prof S. Venturini.
      Obtained funding: N/A.
      Administrative, technical, or material support: N/A.
      Supervision: Dr D. Generali.

      Data availability.

      All data have been retrieved from already published papers.
      Code availability.
      No proprietary codes were used for the analyses reported in this study.

      Declaration of Competing Interest

      The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Dr Giuliano, Arpino and De Placido S have declared honoraria from Roche, Pfizer, Astra-Zeneca, Novartis, Celgene, Eli Lilly, Amgen, and Eisai, outside of the submitted work.
      Dr Lambertini has declared personal fees (advisory role and/or speaker honoraria) from Roche, Takeda, Sandoz, Eli Lilly, Pfizer, AstraZeneca, Novartis Exact Sciences and Ipsen, outside of the submitted work.
      Dr Bartsch has declared honoraria from AstraZeneca, Daiichy Sankyo, Eli Lilly, Novartis, Pfizer, Pierre Fabre, Roche, consulting or advisory role for AstraZeneca, Daiichi, Eisai, Eli Lilly, MSD Oncology, Novartis, Pfizer, Pierre Fabre, Puma Biotechnology, Roche, research funding from Daiichi-Sankyo, travel, accommodations, expenses from Roche, Pfizer, outside of the submitted work.
      Dr Pinato has received lecture fees from ViiV Healthcare and Bayer Healthcare and travel expenses from BMS and Bayer Healthcare; consulting fees for Mina Therapeutics, EISAI, Roche, and Astra Zeneca; research funding (to institution) from MSD and BMS, outside of the submitted work.
      Dr Harbeck has declared stocks and other ownership interests in the West German Study Group, honoraria from Roche, Novartis, Daiichi-Sankyo, Amgen, Pfizer, Exact Sciences, AstraZeneca, Pierre Fabre, SeaGen; consulting or advisory role for Roche/Genentech, Novartis, Pfizer, Eli Lilly, Sandoz, Daiichi Sankyo, AstraZeneca, Merck Sharp & Dohme, Sandoz, SeaGen, Pierre Fabre and research funding to institution from Roche/Genentech, Eli Lilly, Merck Sharp & Dohme, outside of the submitted work.
      Dr Lüftner has received honoraria from Amgen, AstraZeneca, Celgene, Eisai, Genomic Health, Eli Lilly, Loreal, MSD, Novartis, Pierre Fabre, Pfizer, Tesaro, Teva, outside of the submitted work.
      Dr Denys has reported consulting fees (advisory role) paid to her institution by Pfizer, Roche, PharmaMar, AstraZeneca, Eli Lilly, Novartis, Amgen, Tesaro; grants from Research Funding institutional Roche; and other fees (travel, accommodations, expenses) by Pfizer, Roche, PharmaMar, Teva, AstraZeneca, outside of the submitted work.
      Dr Zaman has reported travel, accommodation and expenses from Roche, Pierre Fabri and MSD Oncology, research funding from Roche/Genentech consulting/advisory role for Roche, Eli Lilly, MSD Oncology, Mylan, Novartis, Daiichy and Pierre Fabre, and other relationships with Pierre Fabre, all outside the submitted work.
      Dr Gligorov has declared speakers’ bureau honoraria from Roche-Genentech, Novartis, Eisai, Genomic Health, Ipsen, Pfizer, Mylan, Eli Lilly, Sandoz, and Pierre-Fabre; consultant/advisory board member for Roche-Genentech, Novartis, Onxeo, Daichii Sanyo, MSD, Eisai, Genomic Health, Ipsen, Macrogenics, Pfizer, Mylan, Eli Lilly, Immunomedics, Pierre-Fabre; grant support from Roche-Genetech, Eisai, Genomic Health, Pfizer, Mylan; travel, accommodation paid by Roche-Genentech, Novartis, Eisai, Genomic Health, Pfizer, Eli Lilly, Pierre-Fabre; personal fees and non-financial support from Daichii Sanyo, MSD, outside of the submitted work.
      Dr Awada reports advisory roles, travel grants, and speaker fees from Roche, Eli Lilly, Amgen, ESAI, BMS, Pfizer, Novartis, MSD, Ipsen, and Leopharma. Dr Campone has declared honoraria from Novartis, Eli Lilly, GT1, Consulting or Advisory Role from Novartis, Servier, Menarini, Sanofi, Eli Lilly, Pfizer, AstraZeneca/MedImmune, Abbvie, Pierre Fabre, Accord Healthcare, Sandoz-Novartis, Seattle Genetics, Daiichi Sankyo Europe GmbH, Speakers' Bureau from Novartis, Amgen, Research Funding from Novartis, Travel, Accommodations, Expenses from Novartis, AstraZeneca, Pfizer, Other Relationship from Roche, outside of the submitted work.
      Dr Wildiers has declared Consulting or Advisory Role from Roche, Eli Lilly, Pfizer, Sirtex Medical, Orion Corporation, Puma Biotechnology, AstraZeneca, Biocartis, Novartis, Daiichi Sankyo, Research Funding from Roche, Novartis, Travel, Accommodations, Expenses from Pfizer, Roche, outside of the submitted work.
      Dr Gennari has declared consulting/advisory role for Roche, MSD, Eli Lilly, Pierre Fabre, EISAI, and Daichii Sankyo; speakers bureau for Eisai, Novartis, Eli Lilly, Roche, Teva, Gentili, Pfizer, Astra Zeneca, Celgene, and Daichii Sankyo; research funds from EISAI, Eli Lilly, and Roche, outside of the submitted work.
      Dr Tjan-Heijnen has declared Honoraria and Travel expenses from Novartis, Roche, Eli Lilly, Pfizer and Honoraria from Daichii Sankyo, as well as Research Funding from Roche, Eisai, Pfizer, Novartis, Eli Lilly, AstraZeneca and Daichii Sankyo, outside of the submitted work.
      Dr Cortes has declared Stock and Other Ownership Interests from MedSIR, Honoraria from Novartis, Eisai, Celgene, Pfizer, Roche, Samsung, Eli Lilly, Merck Sharp & Dohme, Daiichi Sankyo, Consulting or Advisory Role from Celgene, Cellestia Biotech, AstraZeneca, Biothera, Merus, Roche, Seattle Genetics, Daiichi Sankyo, ERYTECH Pharma, Polyphor, Athenex, Eli Lilly, Servier, Merck Sharp & Dohme, GlaxoSmithKline, Leuko, Clovis Oncology, Bioasis, Boehringer Ingelheim, Research Funding from ARIAD, Astrazeneca, Baxalta, Bayer, Eisai, Guardant Health, Merck Sharp & Dohme, Pfizer, Puma Biotechnology, Queen Mary University of London, Roche, Piqur, Travel, Accommodations, Expenses from Roche, Pfizer, Eisai, Novartis, Daiichi Sankyo, outside of the submitted work.
      Dr Del Mastro acted as a consultant for Roche, Novartis, MSD, Pfizer, Ipsen, AstraZeneca, Genomic Health, Eli Lilly, Seattle Genetics, Eisai, Pierre Fabre, and Daiichi Sankyo, received speaker honoraria from Roche, Novartis, Eli Lilly and MSD, and travel grants from Roche, Pfizer and Celgene, outside the submitted work.
      Dr Martin has declared Honoraria from Roche/Genentech, Eli Lilly, Pfizer, Novartis, Pierre Fabre, Consulting or Advisory Role from Roche/Genentech, Novartis, Pfizer, Eli Lilly, AstraZeneca, Taiho Pharmaceutical, PharmaMar, Speakers' Bureau from Eli Lilly/ImClone, Roche/Genentech, Pierre Fabre, Research Funding from Novartis, Roche, Puma Biotechnology, outside of the submitted work.
      Dr Curigliano has reported personal fees from Roche, Pfizer, Novartis, Eli Lilly, Foundation Medicine, BMS, Samsung, Astra Zeneca, Daichii Sankyo Boeringer, GSK, Seagen, non-financial support from Roche, Pfizer, grants from Merck, other from Ellipsis, outside the submitted work.
      Dr Jerusalem has reported grants, personal fees and non-financial support from Novartis, Roche and Pfizer, personal fees and non-financial support from Eli Lilly, Amgen, BMS and Astra-Zeneca, personal fees from Daiichi Sankyo and Abbvie, non-financial support from Medimmune and MerckKGaA, outside the submitted work.
      Dr Prat reports grants and personal fees from Pfizer, grants and personal fees from Eli Lilly, personal fees from Nanostring tecnologies, grants and personal fees from Amgen, grants from Roche, personal fees from Oncolytics Biotech, Daiichi Sankyo, PUMA, BMS, consulting/advisory role for NanoString Technologies, Amgen, Roche, Novartis, Pfizer and Bristol-Myers Squibb and patent PCT/EP2016/080056, outside of the submitted work. Dr Prat is also an equity stockholder, founder and consultant of Reveal Genomics and has a patent HER2DX and WO 2018/103834 licensed to Reveal Genomics, all outside of the submitted work.
      Dr Generali has declared Honoraria from Novartis and Eli Lilly, Research Funding from Novartis and Astra-Zeneca, outside of the submitted work.
      All other authors declared nothing to disclose.

      Acknowledgements

      Dr F. Schettini is the recipient of an ESMO Fellowship – Translational and a BBVA Foundation – Hospital Clinic of Barcelona Joan Rodes – Josep Baselga Advanced Research Contract in Oncology. However, any views, opinions, findings, conclusions or recommendations expressed in this material are those solely of the authors and do not necessarily reflect those of Funders. This study was supported by MEDNotE, spin-off - University of Trieste, within the Mozart Program.

      Funding

      None.

      Appendix A. Supplementary material

      The following are the Supplementary data to this article:

      References

        • Tutt A.
        • Tovey H.
        • Cheang M.C.U.
        • Kernaghan S.
        • Kilburn L.
        • Gazinska P.
        • et al.
        A randomised phase III trial of carboplatin compared with docetaxel in BRCA1/2 mutated and pre-specified triple negative breast cancer “BRCAness” subgroups: the TNT Trial.
        Nat Med. 2018; 24: 628-637https://doi.org/10.1038/s41591-018-0009-7
        • Gong Y.
        • Liu Y.-R.
        • Ji P.
        • Hu X.
        • Shao Z.-M.
        Impact of molecular subtypes on metastatic breast cancer patients: a SEER population-based study.
        Sci Rep. 2017; 7: 45411https://doi.org/10.1038/srep45411
        • Hartman A.-R.
        • Kaldate R.R.
        • Sailer L.M.
        • Painter L.
        • Grier C.E.
        • Endsley R.R.
        • et al.
        Prevalence of BRCA mutations in an unselected population of triple-negative breast cancer.
        Cancer. 2012; 118: 2787-2795
        • Litton J.K.
        • Rugo H.S.
        • Ettl J.
        • Hurvitz S.A.
        • Gonçalves A.
        • Lee K.-H.
        • et al.
        Talazoparib in Patients with Advanced Breast Cancer and a Germline BRCA Mutation.
        N Engl J Med. 2018; 379: 753-763
        • Robson M.
        • Im S.-A.
        • Senkus E.
        • Xu B.
        • Domchek S.M.
        • Masuda N.
        • et al.
        Olaparib for Metastatic Breast Cancer in Patients with a Germline BRCA Mutation.
        N Engl J Med. 2017; 377: 523-533
        • Cortes J.
        • Cescon D.W.
        • Rugo H.S.
        • Nowecki Z.
        • Im S.-A.
        • Yusof M.M.
        • et al.
        KEYNOTE-355: Randomized, double-blind, phase III study of pembrolizumab + chemotherapy versus placebo + chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer.
        JCO. 2020; 38: 1000
        • Schmid P.
        • Adams S.
        • Rugo H.S.
        • Schneeweiss A.
        • Barrios C.H.
        • Iwata H.
        • et al.
        Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer.
        N Engl J Med. 2018; 379: 2108-2121
        • Rozenblit M.
        • Huang R.
        • Danziger N.
        • Hegde P.
        • Alexander B.
        • Ramkissoon S.
        • et al.
        Comparison of PD-L1 protein expression between primary tumors and metastatic lesions in triple negative breast cancers.
        J Immunother Cancer. 2020; 8: e001558
      1. Mittendorf EA, Philips AV, Meric-Bernstam F, Qiao N, Wu Y, Harrington S, et al. PD-L1 expression in triple-negative breast cancer. Cancer Immunol Res 2014;2:361–70. https://doi.org/10.1158/2326-6066.CIR-13-0127.

      2. National Comprehensive Cancer Network. NCCN Guidelines for Breast Cancer, vers.4.2022 n.d. https://www.nccn.org/professionals/physician_gls/pdf/breast.pdf (accessed June 28, 2022).

        • Gennari A.
        • André F.
        • Barrios C.H.
        • Cortés J.
        • de Azambuja E.
        • DeMichele A.
        • et al.
        ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer.
        Ann Oncol. 2021; 32: 1475-1495
        • Bardia A.
        • Mayer I.A.
        • Vahdat L.T.
        • Tolaney S.M.
        • Isakoff S.J.
        • Diamond J.R.
        • et al.
        Sacituzumab Govitecan-hziy in Refractory Metastatic Triple-Negative Breast Cancer.
        N Engl J Med. 2019; 380: 741-751
        • Bardia A.
        • Hurvitz S.A.
        • Tolaney S.M.
        • Loirat D.
        • Punie K.
        • Oliveira M.
        • et al.
        Sacituzumab Govitecan in Metastatic Triple-Negative Breast Cancer.
        N Engl J Med. 2021; 384: 1529-1541
        • Schettini F.
        • Chic N.
        • Brasó-Maristany F.
        • Paré L.
        • Pascual T.
        • Conte B.
        • et al.
        Clinical, pathological, and PAM50 gene expression features of HER2-low breast cancer.
        npj Breast Cancer. 2021; 7https://doi.org/10.1038/s41523-020-00208-2
        • Modi S.
        • Jacot W.
        • Yamashita T.
        • Sohn J.
        • Vidal M.
        • Tokunaga E.
        • et al.
        Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer.
        N Engl J Med. 2022; 387: 9-20
      3. Egger SJ, Chan MMK, Luo Q, Wilcken N. Platinum-containing regimens for triple-negative metastatic breast cancer. Cochrane Database Syst Rev 2020;10:CD013750. https://doi.org/10.1002/14651858.CD013750.

        • Moy B.
        • Rumble R.B.
        • Come S.E.
        • Davidson N.E.
        • Di Leo A.
        • Gralow J.R.
        • et al.
        Chemotherapy and Targeted Therapy for Patients With Human Epidermal Growth Factor Receptor 2–Negative Metastatic Breast Cancer That is Either Endocrine-Pretreated or Hormone Receptor–Negative: ASCO Guideline Update.
        JCO. 2021; 39: 3938-3958
        • Jansen J.P.
        • Fleurence R.
        • Devine B.
        • Itzler R.
        • Barrett A.
        • Hawkins N.
        • et al.
        Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1.
        Value Health. 2011; 14: 417-428
        • Hoaglin D.C.
        • Hawkins N.
        • Jansen J.P.
        • Scott D.A.
        • Itzler R.
        • Cappelleri J.C.
        • et al.
        Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2.
        Value Health. 2011; 14: 429-437
        • Giuliano M.
        • Schettini F.
        • Rognoni C.
        • Milani M.
        • Jerusalem G.
        • Bachelot T.
        • et al.
        Endocrine treatment versus chemotherapy in postmenopausal women with hormone receptor-positive, HER2-negative, metastatic breast cancer: a systematic review and network meta-analysis.
        Lancet Oncol. 2019; 20: 1360-1369
        • Li L.
        • Pan Z.
        Progression-Free Survival and Time to Progression as Real Surrogate End Points for Overall Survival in Advanced Breast Cancer: A Meta-Analysis of 37 Trials.
        Clin Breast Cancer. 2018; 18: 63-70https://doi.org/10.1016/j.clbc.2017.07.015
        • Generali D.
        • Venturini S.
        • Rognoni C.
        • Ciani O.
        • Pusztai L.
        • Loi S.
        • et al.
        A network meta-analysis of everolimus plus exemestane versus chemotherapy in the first- and second-line treatment of estrogen receptor-positive metastatic breast cancer.
        Breast Cancer Res Treat. 2015; 152: 95-117
        • Diéras Véronique
        • Han H.S.
        • Kaufman B.
        • Wildiers H.
        • Friedlander M.
        • Ayoub J.-P.
        • et al.
        Veliparib with carboplatin and paclitaxel in BRCA-mutated advanced breast cancer (BROCADE3): a randomised, double-blind, placebo-controlled, phase 3 trial.
        Lancet Oncol. 2020; 21: 1269-1282
        • Miles D.
        • Gligorov J.
        • André F.
        • Cameron D.
        • Schneeweiss A.
        • Barrios C.
        • et al.
        Primary results from IMpassion131, a double-blind, placebo-controlled, randomised phase III trial of first-line paclitaxel with or without atezolizumab for unresectable locally advanced/metastatic triple-negative breast cancer.
        Ann Oncol. 2021; 32: 994-1004
        • Cortes J.
        • Cescon D.W.
        • Rugo H.S.
        • Nowecki Z.
        • Im S.-A.
        • Yusof M.M.
        • et al.
        Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355): a randomised, placebo-controlled, double-blind, phase 3 clinical trial.
        The Lancet. 2020; 396: 1817-1828
        • Winer E.P.
        • Lipatov O.
        • Im S.-A.
        • Goncalves A.
        • Muñoz-Couselo E.
        • Lee K.S.
        • et al.
        Pembrolizumab versus investigator-choice chemotherapy for metastatic triple-negative breast cancer (KEYNOTE-119): a randomised, open-label, phase 3 trial.
        Lancet Oncol. 2021; 22: 499-511
        • Kim S.-B.
        • Dent R.
        • Im S.-A.
        • Espié M.
        • Blau S.
        • Tan A.R.
        • et al.
        Ipatasertib plus paclitaxel versus placebo plus paclitaxel as first-line therapy for metastatic triple-negative breast cancer (LOTUS): a multicentre, randomised, double-blind, placebo-controlled, phase 2 trial.
        Lancet Oncol. 2017; 18: 1360-1372
      4. Dent R, Kim S-B, Oliveira M, Barrios C, O’Shaughnessy J, Isakoff SJ, et al. Abstract GS3-04: Double-blind placebo (PBO)-controlled randomized phase III trial evaluating first-line ipatasertib (IPAT) combined with paclitaxel (PAC) for PIK3CA/AKT1/PTEN-altered locally advanced unresectable or metastatic triple-negative breast cancer (aTNBC): primary results from IPATunity130 Cohort A. Cancer Res 2021;81:GS3-04. https://doi.org/10.1158/1538-7445.SABCS20-GS3-04.

        • Schmid P.
        • Abraham J.
        • Chan S.
        • Wheatley D.
        • Brunt A.M.
        • Nemsadze G.
        • et al.
        Capivasertib Plus Paclitaxel Versus Placebo Plus Paclitaxel As First-Line Therapy for Metastatic Triple-Negative Breast Cancer: The PAKT Trial.
        J Clin Oncol. 2020; 38: 423-433
        • Harbeck N.
        • Saupe S.
        • Jäger E.
        • Schmidt M.
        • Kreienberg R.
        • Müller L.
        • et al.
        A randomized phase III study evaluating pegylated liposomal doxorubicin versus capecitabine as first-line therapy for metastatic breast cancer: results of the PELICAN study.
        Breast Cancer Res Treat. 2017; 161: 63-72
        • von Minckwitz G.
        • Puglisi F.
        • Cortes J.
        • Vrdoljak E.
        • Marschner N.
        • Zielinski C.
        • et al.
        Bevacizumab plus chemotherapy versus chemotherapy alone as second-line treatment for patients with HER2-negative locally recurrent or metastatic breast cancer after first-line treatment with bevacizumab plus chemotherapy (TANIA): an open-label, randomised phase 3 trial.
        Lancet Oncol. 2014; 15: 1269-1278
        • Takashima T.
        • Mukai H.
        • Hara F.
        • Matsubara N.
        • Saito T.
        • Takano T.
        • et al.
        Taxanes versus S-1 as the first-line chemotherapy for metastatic breast cancer (SELECT BC): an open-label, non-inferiority, randomised phase 3 trial.
        Lancet Oncol. 2016; 17: 90-98
        • Lunn D.J.
        • Thomas A.
        • Best N.
        • Spiegelhalter D.
        WinBUGS - A Bayesian modelling framework: Concepts, structure, and extensibility.
        Statist Comput. 2000; 10: 325-337https://doi.org/10.1023/A:1008929526011
        • Mbuagbaw L.
        • Rochwerg B.
        • Jaeschke R.
        • Heels-Andsell D.
        • Alhazzani W.
        • Thabane L.
        • et al.
        Approaches to interpreting and choosing the best treatments in network meta-analyses.
        Syst Rev. 2017; 6: 79https://doi.org/10.1186/s13643-017-0473-z
        • Gelman A.
        • Rubin D.
        Inference from iterative simulation using multiple sequences.
        Statist Sci. 1992; 7: 457-511
        • Lynch S.M.
        Introduction to Applied Bayesian Statistics and Estimation for Social Scientists.
        in: Lynch S.M. Introduction to Applied Bayesian Statistics and Estimation for Social Scientists. Springer New York, New York, NY2007: 1-7
      5. Dias S, Ades A, Welton N, Jansen J, Sutton A. Network Meta-Analysis for Decision-Making | Wiley. 1st Edition. John Wiley and Sons; 2018.

        • Hutton B.
        • Salanti G.
        • Caldwell D.M.
        • Chaimani A.
        • Schmid C.H.
        • Cameron C.
        • et al.
        The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations.
        Ann Intern Med. 2015; 162: 777-784
        • Higgins J.P.T.
        • Jackson D.
        • Barrett J.K.
        • Lu G.
        • Ades A.E.
        • White I.R.
        Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies.
        Res Synth Methods. 2012; 3: 98-110https://doi.org/10.1002/jrsm.1044
        • Dias S.
        • Welton N.J.
        • Caldwell D.M.
        • Ades A.E.
        Checking consistency in mixed treatment comparison meta-analysis.
        Stat Med. 2010; 29: 932-944https://doi.org/10.1002/sim.3767
      6. R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. n.d.

      7. Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration; n.d.

        • Miles D.
        • Gligorov J.
        • André F.
        • Cameron D.
        • Schneeweiss A.
        • Barrios C.
        • et al.
        Primary results from IMpassion131, a double-blind, placebo-controlled, randomised phase III trial of first-line paclitaxel with or without atezolizumab for unresectable locally advanced/metastatic triple-negative breast cancer.
        Ann Oncol. 2021; 32: 994-1004
        • Delaloge S.
        • Pérol D.
        • Courtinard C.
        • Brain E.
        • Asselain B.
        • Bachelot T.
        • et al.
        Paclitaxel plus bevacizumab or paclitaxel as first-line treatment for HER2-negative metastatic breast cancer in a multicenter national observational study.
        Ann Oncol. 2016; 27: 1725-1732
        • Dieras V.
        • Pop S.
        • Berger F.
        • Dujaric M.-E.
        • Beuzeboc P.
        • Escalup L.
        • et al.
        First-line Bevacizumab and Paclitaxel for HER2-negative Metastatic Breast Cancer: A French Retrospective Observational Study.
        Anticancer Res. 2017; 37: 1403-1407https://doi.org/10.21873/anticanres.11462
        • Smith I.E.
        • Pierga J.-Y.
        • Biganzoli L.
        • Cortés-Funes H.
        • Thomssen C.
        • Pivot X.
        • et al.
        First-line bevacizumab plus taxane-based chemotherapy for locally recurrent or metastatic breast cancer: safety and efficacy in an open-label study in 2,251 patients.
        Ann Oncol. 2011; 22: 595-602
        • Bramati A.
        • Girelli S.
        • Torri V.
        • Farina G.
        • Galfrascoli E.
        • Piva S.
        • et al.
        Efficacy of biological agents in metastatic triple-negative breast cancer.
        Cancer Treat Rev. 2014; 40: 605-613
        • Miles D.W.
        • Diéras V.
        • Cortés J.
        • Duenne A.-A.
        • Yi J.
        • O’Shaughnessy J.
        First-line bevacizumab in combination with chemotherapy for HER2-negative metastatic breast cancer: pooled and subgroup analyses of data from 2447 patients.
        Ann Oncol. 2013; 24: 2773-2780https://doi.org/10.1093/annonc/mdt276
        • Roila F.
        • Molassiotis A.
        • Herrstedt J.
        • Aapro M.
        • Gralla R.J.
        • Bruera E.
        • et al.
        2016 MASCC and ESMO guideline update for the prevention of chemotherapy- and radiotherapy-induced nausea and vomiting and of nausea and vomiting in advanced cancer patients.
        Ann Oncol. 2016; 27: v119-v133
        • Klastersky J.
        • de Naurois J.
        • Rolston K.
        • Rapoport B.
        • Maschmeyer G.
        • Aapro M.
        • et al.
        Management of febrile neutropaenia: ESMO Clinical Practice Guidelines †.
        Ann Oncol. 2016; 27: v111-v118
        • Aapro M.
        • Beguin Y.
        • Bokemeyer C.
        • Dicato M.
        • Gascón P.
        • Glaspy J.
        • et al.
        Management of anaemia and iron deficiency in patients with cancer: ESMO Clinical Practice Guidelines.
        Ann Oncol. 2018; 29: iv96-iv110
        • Schettini F.
        • Corona S.P.
        • Giudici F.
        • Strina C.
        • Sirico M.
        • Bernocchi O.
        • et al.
        Clinical, Radiometabolic and Immunologic Effects of Olaparib in Locally Advanced Triple Negative Breast Cancer: The OLTRE Window of Opportunity Trial.
        Front Oncol. 2021; 11https://doi.org/10.3389/fonc.2021.686776
        • Tung N.M.
        • Robson M.E.
        • Ventz S.
        • Santa-Maria C.A.
        • Nanda R.
        • Marcom P.K.
        • et al.
        TBCRC 048: Phase II Study of Olaparib for Metastatic Breast Cancer and Mutations in Homologous Recombination-Related Genes.
        J Clin Oncol. 2020; 38: 4274-4282
        • Fasching P.A.
        • Link T.
        • Hauke J.
        • Seither F.
        • Jackisch C.
        • Klare P.
        • et al.
        Neoadjuvant paclitaxel/olaparib in comparison to paclitaxel/carboplatinum in patients with HER2-negative breast cancer and homologous recombination deficiency (GeparOLA study).
        Ann Oncol. 2021; 32: 49-57
        • Schettini F.
        • Giudici F.
        • Bernocchi O.
        • Sirico M.
        • Corona S.P.
        • Giuliano M.
        • et al.
        Poly (ADP-ribose) polymerase inhibitors in solid tumours: Systematic review and meta-analysis.
        Eur J Cancer. 2021; 149: 134-152
        • Tutt A.N.J.
        • Garber J.E.
        • Kaufman B.
        • Viale G.
        • Fumagalli D.
        • Rastogi P.
        • et al.
        Adjuvant Olaparib for Patients with BRCA1- or BRCA2-Mutated Breast Cancer.
        N Engl J Med. 2021; 384: 2394-2405
      8. Schmid P, Cortes J, Dent R, Pusztai L, McArthur H, Kümmel S, et al. VP7-2021: KEYNOTE-522: Phase III study of neoadjuvant pembrolizumab + chemotherapy vs. placebo + chemotherapy, followed by adjuvant pembrolizumab vs. placebo for early-stage TNBC. Annals of Oncology 2021;0. https://doi.org/10.1016/j.annonc.2021.06.014.

      9. Roche Group Media Relations. Roche provides update on Tecentriq US indication for PD-L1-positive, metastatic triple-negative breast cancer. Roche Media Releases n.d. https://www.roche.com/media/releases/med-cor-2021-08-27.htm (accessed October 18, 2021).

      10. Rugo H, Llombart-Cussac A, Andre F, Robson ME, Saji S, Harbeck N, et al. Abstract OT-30-01: KEYLYNK-009: A phase 2/3, open-label, randomized study of pembrolizumab plus olaparib vs pembrolizumab plus chemotherapy after induction with first-line pembrolizumab plus chemotherapy in patients with locally recurrent inoperable or m…. Cancer Research 2021;81:OT-30-01-OT-30-01. https://doi.org/10.1158/1538-7445.SABCS20-OT-30-01.

        • Garutti M.
        • Pelizzari G.
        • Bartoletti M.
        • Malfatti M.C.
        • Gerratana L.
        • Tell G.
        • et al.
        Platinum Salts in Patients with Breast Cancer: A Focus on Predictive Factors.
        Int J Mol Sci. 2019; 20: 3390
        • Schmid P.
        • Cortes J.
        • Robson M.E.
        • Iwata H.
        • Hegg R.
        • Nechaeva M.
        • et al.
        A phase III trial of capivasertib and paclitaxel in first-line treatment of patients with metastatic triple-negative breast cancer (CAPItello290).
        JCO. 2020; 38: TPS1109
      11. Therapeutics G. G1 Therapeutics Announces Initiation of Phase 3 Registrational Study of COSELATM (trilaciclib) in Triple-Negative Breast Cancer (TNBC). GlobeNewswire News Room 2021. https://www.globenewswire.com/news-release/2021/04/28/2218475/0/en/G1-Therapeutics-Announces-Initiation-of-Phase-3-Registrational-Study-of-COSELA-trilaciclib-in-Triple-Negative-Breast-Cancer-TNBC.html (accessed June 2, 2021).

        • Schettini F.
        • Giuliano M.
        • De Placido S.
        • Arpino G.
        Nab-paclitaxel for the treatment of triple-negative breast cancer: Rationale, clinical data and future perspectives.
        Cancer Treat Rev. 2016; 50: 129-141https://doi.org/10.1016/j.ctrv.2016.09.004
        • Puglisi F.
        • Rea D.
        • Kroes M.A.
        • Pronzato P.
        Second-line single-agent chemotherapy in human epidermal growth factor receptor 2-negative metastatic breast cancer: A systematic review.
        Cancer Treat Rev. 2016; 43: 36-49https://doi.org/10.1016/j.ctrv.2015.11.012
        • Marabelle A.
        • Le D.T.
        • Ascierto P.A.
        • Di Giacomo A.M.
        • De Jesus-Acosta A.
        • Delord J.-P.
        • et al.
        Efficacy of Pembrolizumab in Patients With Noncolorectal High Microsatellite Instability/Mismatch Repair-Deficient Cancer: Results From the Phase II KEYNOTE-158 Study.
        J Clin Oncol. 2020; 38: 1-10
        • Marabelle A.
        • Fakih M.
        • Lopez J.
        • Shah M.
        • Shapira-Frommer R.
        • Nakagawa K.
        • et al.
        Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study.
        Lancet Oncol. 2020; 21: 1353-1365
        • Chaimani A.
        • Higgins J.P.T.
        • Mavridis D.
        • Spyridonos P.
        • Salanti G.
        • Haibe-Kains B.
        Graphical Tools for Network Meta-Analysis in STATA.
        PLoS ONE. 2013; 8: e76654
        • Burstein H.J.
        • Somerfield M.R.
        • Barton D.L.
        • Dorris A.
        • Fallowfield L.J.
        • Jain D.
        • et al.
        Endocrine Treatment and Targeted Therapy for Hormone Receptor-Positive, Human Epidermal Growth Factor Receptor 2-Negative Metastatic Breast Cancer: ASCO Guideline Update.
        J Clin Oncol. 2021; 39: 3959-3977
        • Bailar J.C.
        The promise and problems of meta-analysis.
        N Engl J Med. 1997; 337: 559-561
        • Greenland S.
        Can meta-analysis be salvaged?.
        Am J Epidemiol. 1994; 140: 783-787https://doi.org/10.1093/oxfordjournals.aje.a117326
      12. Chaimani A, Caldwell D, Li T, Higgins J, Salanti G. Chapter 11: Undertaking network meta-analyses. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.2 (updated February 2021). Cochrane, 2021. Available from www.training.cochrane.org/handbook. n.d.

        • Dias S.
        • Sutton A.J.
        • Ades A.E.
        • Welton N.J.
        Evidence synthesis for decision making 2: a generalized linear modeling framework for pairwise and network meta-analysis of randomized controlled trials.
        Med Decis Making. 2013; 33: 607-617https://doi.org/10.1177/0272989X12458724
        • Schettini F.
        • Giuliano M.
        • Lambertini M.
        • Bartsch R.
        • Pinato D.J.
        • Onesti C.E.
        • et al.
        Anthracyclines Strike Back: Rediscovering Non-Pegylated Liposomal Doxorubicin in Current Therapeutic Scenarios of Breast Cancer.
        Cancers (Basel). 2021; 13: 4421