Endocrine Therapy-Based Strategies for Metastatic Breast Cancer with Different Endocrine Sensitivity Statuses: A Systematic Review and Network Meta-Analysis

Simple Summary Over the past few years, a great deal of considerable research has been conducted to develop more effective targeted agents for metastatic breast cancer (mBC) patients. This systematic review and meta-analysis of 47 randomized clinical trials identified CDK4/6i + Fulvestrant 500 as the best treatment option in improving efficacy outcomes in endocrine therapy-sensitive (ETS) patients. Chemotherapy had a higher likelihood of therapeutic success in endocrine therapy-resistant (ETR) patients. In ETR settings, with visceral and bone metastases, CDK4/6i +Fulvestrant 500 was the best regimen, while in ETS patients, CDK4/6i + Fulvestrant 500 was the best for bone metastases and CDK4/6i + aromatase inhibitors (AI) for visceral metastases. This study assessed the relative efficacies and provided a rank order of ET-based regimens in the key endpoints PFS and OS for HR+/HER2- mBC patients with different endocrine sensitivity statuses. Different endocrine sensitivity statuses required various optimal treatment strategies, which may provide guidance for clinical practice. Abstract Background: Novel endocrine therapies (ETs) and targeted therapeutic regimens have been developed to dramatically improve the outcome of hormone receptor-positive (HR+)/HER2-negative (HER2-) metastatic breast cancer (mBC). Methods: We performed a systematic search with a predefined search strategy in PubMed, Embase and Cochrane CENTRAL databases to perform a network meta-analysis and evaluate the relative efficacies of ET-based treatment regimens in HR+/HER2- mBC patients with different endocrine sensitivity statuses. The study was registered in the PROSPERO database (CRD42021235570). Results: A total of 47 trials (20,267 patients) were included. Analysis of progression-free survival (PFS) in endocrine therapy-sensitive (ETS) patients revealed cyclin-dependent kinases 4/6 inhibitors (CDK4/6i) + fulvestrant 500 mg (Ful 500) (random effect (RE): hazard ratio (HR), 0.46; 95% credibility interval (CrI), 0.27–0.78; surface under the cumulative ranking curve (SUCRA), 0.93; fixed effect (FE): HR, 0.48; 95% CrI, 0.40–0.58; SUCRA, 0.99) to be the best therapy followed by CDK4/6i + aromatase inhibitors (AIs) (RE: HR, 0.53; 95% CrI, 0.40–0.72; SUCRA, 0.86; FE: HR, 0.54; 95% CrI, 0.48–0.61; SUCRA, 0.91). Chemotherapy followed by CDK4/6i + Ful 500 appears to be the most effective option for the endocrine therapy-resistant (ETR) group. Analysis of overall survival revealed CDK4/6i + Ful 500 (SUCRA: 0.99) and AKTi + Ful 500 (SUCRA: 0.87) to be the first-rank regimen for the ETS group and ETR groups, respectively. Conclusion: Our comprehensive analysis suggests that CDK4/6i combined with ETs may be the best treatment option in terms of PFS for ETS patients and chemotherapy for ETR patients with HR+/HER2- mBC. Different endocrine sensitivity statuses required various optimal treatment strategies, which may provide guidance for clinical practice.


Introduction
Breast cancer is currently the most common malignancy in women worldwide, accounting for approximately 11.7% of all new malignancies [1]. Hormone-receptor-positive (HR+)/HER2-negative (HER2-) breast cancer accounts for approximately 65-70% of mBC. Existing evidence substantiated that the estrogen receptor signaling pathway plays a pivotal role in cancer cell growth and survival in these tumors, and, hence, endocrine-based therapies (ETs) are considered to be one of the most effective systemic treatment cornerstones [2] for these patients, even in visceral metastatic settings. Despite the perceived significant therapeutic effect of ETs, a major therapeutic challenge in treating HR+ breast cancer is to delay or reverse endocrine resistance (ER). A subset of patients (30-40%) eventually acquire resistance to ETs due to prolonged use in the adjuvant setting [3].
Over the past few years, a great deal of considerable research has been made to develop more effective targeted agents. In patients with mBC, the main drivers of resistance include ESR1 gene fusion which may require a combination treatment with targeted agents to overcome the resistance and also to minimize the probability of the emergence of resistance in endocrine-sensitive patients [4,5]. Among the other mechanisms of endocrine resistance, targeting high expression of EGFR or HER2 [6,7], phosphoinositide 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) or RAS/RAF/MEK/ERK pathways could also minimize the emergence of ER in patients with mBC [8]. Furthermore, addition of cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) has also been shown to extend the duration of progression-free survival (PFS) and overall survival (OS) in comparison to ET alone in patients with mBC. Given the absence of direct head-to-head comparisons for all available treatment regimens, decision-making guidelines are urgently needed for multiple therapeutic options.
Previous meta-analyses have taken into account the efficacy of CDK4/6i and ET combinations but were limited in many aspects, such as discovery of appropriate targeted therapies, use of heterogeneity model, non-inclusion of data related to OS, menopausal women and heterogeneity of disease [9,10]. Furthermore, the studies did not perform an extensive efficacy analysis of all possible subgroups of patients with different endocrine sensitivity statuses. Thus, the objective of this network meta-analysis is to perform an extensive analysis and rank the available ET-based strategies (with/without targeted agents) based on their relative efficacies to guide the clinical decision making for patients with HR+/HER2-mBC with different endocrine sensitivity statuses.

Search Strategy and Eligibility Criteria
A systematic search with a predefined search strategy (Supplementary Table S1) consisting of HR status, advanced breast cancer, interested therapeutic regimens and desired endpoints with relevant study design was performed in PubMed, Embase and Cochrane CENTRAL databases. The search was performed from January 2000 till April 2022 and included only studies published in English. We also performed a supplementary search of major conferences in the American Society of Clinical Oncology Annual Meeting, European Society for Medical Oncology Congress and San Antonio Breast Cancer Symposium to retrieve updated survival data from unpublished landmark trials.
Only phase II/III randomized controlled trials (RCTs) evaluating the efficacy of at least one ET-based regimen for the treatment of patients with advanced or metastatic HR+/HER2-breast cancer were considered for this analysis. This systematic review and network meta-analysis was based on the Cochrane Collaboration Handbook (version 6.1, https://training.cochrane.org/handbook, accessed on 5 October 2022) and registered in the PROSPERO database (CRD42021235570). The PRISMA (Preferred Reporting Items for Systematic reviews) and Meta-Analyses extension for network meta-analysis guidelines were followed to report the present study. The information was recorded by one investigator (H. Yiqun) and is concurrently collected by two investigators (X. Binghe and W. Jiani). Three investigators independently extracted data. The classification criteria used for defining endocrine sensitivity statuses followed the current international consensus guidelines. Characteristics of endocrine therapy-sensitive (ETS) population include: (1) completion of adjuvant ETs and recurrence after one year of drug discontinuation; (2) no ETs background, including those mBC without any ETs and recurrence after surgery without adjuvant ETs. Characteristics of endocrine therapyresistant (ETR) population include: (1) Primary ETR: relapse after less than 2 years of adjuvant ETs, or disease progression after less than 6 months of advanced first-line ETs; (2) Secondary ETR: patients with adjuvant ETs for more than 2 years and relapsed within 1 year of discontinuation, or progress on late first-line endocrine therapy for more than 6 months. If not explicitly mentioned in the clinical trial, other information, such as enrollment and subgroup analysis, was used to determine endocrine status.

Study Objectives and Endpoints
The primary objective of the study was to compare the relative efficacies of the available therapeutic options including monotherapy and combination therapy for the treatment of HR+/HER2-mBC. The primary efficacy endpoint was PFS, defined as the time from randomization to progressive disease or death, and the secondary endpoint was OS, defined as the time from randomization until death with any cause. The efficacy analysis was performed based on endocrine status (ETS and ETR). Subgroup analysis based on specific targeted regimens, site of metastasis (visceral and bone-only) and menopausal status was also performed. Multiple reports collection and data extraction were performed by one author and verified by a second author. A third author resolved disagreements.

Statistical Analysis
Evaluation of quality of the selected RCTs was performed by the investigators using Cochrane Collaboration's risk-of-bias tool. For each study, the hazard ratio (HR) for PFS and OS along with their 95% credibility intervals (CrI) were extracted from baseline demographics. A Bayesian network meta-analysis was performed using the "Gemtc" 4.2.1 package from R software (version 4.0.4). Within the framework, heterogeneity across studies was tested by I 2 and the Q statistics along with a forest plot; statistically significant heterogeneity was defined as a p-value < 0.1 or an I 2 statistic > 50%. A random effect (RE) model was performed when I 2 ≥ 50%; otherwise, a fixed effect (FE) model was applied to quantitate effect size. The surface under the cumulative ranking curve (SUCRA) value was used to rank the effective treatment regimens. Thus, treatments were ranked in a table from best to worst along the leading diagonal [11] with high odds in SUCRA suggesting better ranking [12]. PFS and OS, as defined above for efficacy analysis of treatments based on endocrine status, will be ranked based on the SUCRA value to suggest the most efficacious treatment group.

Network Geometry
Based on the availability of evidence, networks were constructed for ETS and ETR patients separately. Since the purpose of the network meta-analysis is to provide the relative efficacies of all the available regimens, more than one evidence network was provided for some of the evaluated endpoints and subgroups. In case of two regimens not being accommodated within the major network of evidence, those regimens were excluded from the analyses. For the construction of the evidence networks, comparison was made at the level of drug class. However, additional subgroup analyses at the level of individual drugs were performed for specific drugs as stated above.

Risk Bias in Individual Studies
The quality of each RCT included was evaluated by two independent authors using a Cochrane Risk of Bias 2 (RoB 2.0) tool [13,14]. The tool includes (1) randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of  [13]. Biases are assessed either as "Low" or "High" risk of bias, or can express "Some concerns".

Publication Bias and Sensitivity Analysis
Egger's test and Begg's test were used to assess the potential publication bias. Sensitivity analysis was performed for those which showed publication bias by step-wise removal of single studies (Supplementary Figures S1-S3).

Systematic Review and Characteristics
A total of 1032 publications and 152 abstract reviews were retrieved through the initial literature and conference search, and 1083 studies remained after duplications were excluded. A total of 47 studies with a cumulative population of 20,267 patients were identified for the detailed analysis. Among the 47 studies, 35 studies were included for the analysis of ETR and 34 studies were included for the analysis of ETS patients ( Figure 1). Since the widespread standardized testing of HER2 status was implemented in 2007, articles published from 2009 to 2022 were screened based on the eligibility criteria of this study. The baseline factors of the included studies are provided in Table 1. individual drugs were performed for specific drugs as stated above.

Risk Bias in Individual Studies
The quality of each RCT included was evaluated by two independent authors using a Cochrane Risk of Bias 2 (RoB 2.0) tool [13,14]. The tool includes (1) randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome and (5) selection of the reported result [13]. Biases are assessed either as "Low" or "High" risk of bias, or can express "Some concerns".

Publication Bias and Sensitivity Analysis
Egger's test and Begg's test were used to assess the potential publication bias. Sensitivity analysis was performed for those which showed publication bias by step-wise removal of single studies (Supplementary Figures S1-S3).

Systematic Review and Characteristics
A total of 1032 publications and 152 abstract reviews were retrieved through the initial literature and conference search, and 1083 studies remained after duplications were excluded. A total of 47 studies with a cumulative population of 20,267 patients were identified for the detailed analysis. Among the 47 studies, 35 studies were included for the analysis of ETR and 34 studies were included for the analysis of ETS patients ( Figure 1). Since the widespread standardized testing of HER2 status was implemented in 2007, articles published from 2009 to 2022 were screened based on the eligibility criteria of this study. The baseline factors of the included studies are provided in Table 1.

Quality of the Evidence
The overall risk of bias of included studies is presented in Supplementary Figure S4. Of the 47 trials included, 33 studies (70%) presented an overall low risk in the five areas of potential bias. A high risk of bias was present in seven (15%) trials while the remaining trials (15%) had "some concerns", mainly due to open label design and assessment of outcomes by the investigator. The risk of bias was frequently high/unclear in trials with results published only in the form of meeting abstracts (n = 3).
The methodological quality showed high risk of bias in three trials [29,53,56]. The final results of the ETR OS and ETR of CDK4/6i in combination with other drugs were not affected by the presence of these trials. After removing studies by Jeresaleum et al. [53] and Martin et al. [56], treatment regimens remained the same, confirming that the results were reliable. Similarly, in cases of CDK4/6i in combination with other drugs in ETR and ETS patients, the best treatment regimens remained unchanged after removal of studies [29,56]. There was, however, a noticeable change in the treatment regimens in cases of ETR-PFS and post-menopausal status in ETR patients after exclusion of the trials [53,56].
The hazards ratio and the respective 95% CrI are provided in online Supplementary Tables S2-S37.

Discussion
Over the past decade, several RCTs have led to the introduction of innovative targeted therapeutic strategies (mTOR, HDAC, CDK4/6, PI3K inhibitors, etc.) in combination with ETs into clinical practice. Furthermore, pivotal RCTs have proven the efficacy of these combinations as first and subsequent lines of treatment for HR+/HER2-mBC with substantial improvement in patient outcomes [1,2].
This network meta-analysis assessed the relative efficacies and provided a rank order of ET-based regimens in key endpoints PFS and OS for HR+/HER2-mBC patients with different endocrine sensitivity statuses. Based on the systematic review, the endocrine therapies were used in combination with the regimens targeting the CDK4/6, PI3K and HDAC pathways, which led to significant improvement in therapeutic outcomes. The main strength of this current network meta-analysis is the extensive analysis of all possible subgroups of patients and the number of studies included for the analysis. Given the geometry of the networks, the SUCRA values may guide clinical decision making by indicating ranking probabilities for each treatment.
The results of this study suggest that CDK4/6i in combination with Ful 500 is likely to be the best treatment option in terms of both PFS and OS benefit among ETS patients. The second-best options for treatment with significant differences for ETS patients were CDK4/6i + AI for improving PFS outcomes, and Ful 500 for improving OS outcomes. This was consistent with a recent network meta-analysis by Brandao et al. which reported CDK4/6i + Ful 500 to be the most effective treatment option for ETS patients [9]. Similar results were also reported in another network meta-analysis where addition of CDK4/6i to ET significantly improved median PFS (pooled HR = 0.547, p < 0.001) and OS (pooled HR = 0.755, p < 0.001). In addition, the PI3Ki + Ful therapy provided meaningful improvement in PFS (pooled HR = 0.686, p < 0.001), which was also noticed in our analysis [70].
In patients with ETR, chemotherapy has a high likelihood of therapeutic success, followed by CDK4/6i and AKTi in combination with fulvestrant. However, the network meta-analysis of 29 trials in ETR patients reported CDK4/6i + Ful 500 to be the best among 20 compared regimens, followed by capivasertib + Ful 500 as the second-best regimen [9]. However, in the current study, a total of 35 studies were included and all chemotherapeutic drugs were combined into a single drug class for comparison. Especially in cases of ETR-PFS and post-menopausal status in ETR patients containing chemotherapy arms, sensitivity analysis revealed a change in the treatment regimens after exclusion of Jeresaleum et al. [53] and Martin et al. [56]. The difference in the included trials and classification of regimens could be the reason for observed discordance. Nevertheless, the previous study also reported a lack of statistical significance among the top regimens [9].
Patients with visceral metastases have poor prognosis compared to those without visceral metastases among women with HR+/HER2-breast cancer. International guidelines recommend ET over chemotherapy as the frontline treatment option for mBC with HR+/HER2-type. Previous studies also revealed that treating patients with CDK4/6i + ET provided the advantage of PFS over ET alone in patients with or without visceral metastasis in both ETS and ETR settings [10]. Our study also revealed concordant findings, with CDK4/6i + FUL500 being the best regimen for ETS patients and CDK4/6i + Ful 500 for ETR patients. In patients with bone-only mBC, the treatment course is varied. However, bone-only disease is uncommon, and often observed in combination with other non-bone or non-visceral sites in clinical trials. A meta-analysis revealed that CDK4/6i + ET in combination improved PFS (HR, 0.54; 95% CI, 0.39-0.75; p < 0.001) for first-line treatment in bone-only disease settings, which is similar to our study findings [70].
We observed that chemotherapy was the best regimen for post-menopausal women with breast cancer. This is inconsistent with several published studies that reported hormone therapies with or without targeted therapies in post-menopausal women with HR+/HER2-mBC as the best treatment option. However, it must be noted that chemotherapy as an upfront treatment remains common even in the absence of visceral crisis and, hence, these results are to be interpreted with caution. This was also observed in a trial by Giuliano et al. [71] where chemotherapy or hormone therapy regimen was non-inferior to palbociclib plus letrozole for improving PFS. Though CDK4/6i + Ful500 was a better regimen in pre-menopausal women and consistent with previous studies [10,72], the number of studies were insufficient to derive to a meaningful conclusion. Combination of targeted therapies with ETs also revealed interesting results. In two of the meta-analyses, HDACi + ET was associated with prolonged PFS and OS compared to ET [73] or AI alone [74]. Accordingly, HDACi + ET was found to be a suitable treatment option in patients with HDAC modifications, with Tuc + AI and Ent + AI as the best regimen for patients with ETS and ETR status, respectively.
The classification of endocrine sensitivity statuses was as per the information provided in the included trials. However, if the trial did not categorically mention the endocrine status, then alternative information from the trial was used to determine this. For instance, the S0226 trial did not mention the endocrine status of the included patients in the initial publication with PFS data. Hence, patients with a previous history of tamoxifen administration were considered as endocrine-resistant patients [27]. However, in the subsequent publication with OS data, HR for patients stratified into endocrine-sensitive and -resistant were provided and the same were used for analysis [28].

Limitations
Several limitations to this study must be acknowledged. Firstly, for subgroup analysis of bone-only metastasis, some of the studies provided only non-visceral metastasis, which was considered as bone-only metastasis for comparison. The MIRACLE trial [43], which evaluated the efficacy of everolimus in combination with AI in comparison to AI alone, provided HR stratified by visceral and non-visceral metastasis. The HR for non-visceral metastasis was considered as bone-only metastasis for analysis. Secondly, the effect estimates of the OS results may not be matured enough and may require a longer follow-up time for precise evaluation. In cases of patients with ETR, studies with fulvestrant and AI were distributed into two different networks and, hence, their relative efficacies with respect to extending the OS could not be ascertained. Thirdly, due to a large degree of heterogeneity across trials, network-based meta-analysis could not be performed for defined biomarker subgroups, such as patients with PIK3CA mutations.

Conclusions
Our comprehensive network meta-analysis suggests that CDK4/6i combination with ETs may be the best treatment option in terms of PFS for ETS patients and chemotherapy may be the best regimen for ETR patients with HR+/HER2-mBC. Moreover, CDK4/6i + ETs is likely to improve bone or visceral metastases irrespective of ETR or ETS status. Further matured OS data need to be assessed to conclude the OS benefit of the different ET-based treatment regimens.  Table S12. SUCRA values of PFS for bone metastases in ETS patients. Table S13. Effect estimates of PFS for bone metastases in ETS patients: hazard ratio (95% credibility interval, CrI). Table S14. SUCRA values of PFS for CDK4/6 inhibitors in ETR patients (Network 1). Table S15. Effect estimates of PFS for CDK4/6 inhibitors in ETR patients: hazard ratio (95% credibility interval, CrI) (Network 1). Table S16. SUCRA values of PFS for CDK4/6 inhibitors in ETR patients (Network 2). Table S17. Effect estimates of PFS for CDK4/6 inhibitors in ETR patients: hazard ratio (95% credibility interval, CrI) (Network 2). Table S18. SUCRA values of PFS for PI3ki in ETR patients. Table S19. Effect estimates of PFS for PI3ki in ETR patients: hazard ratio (95% credibility interval, CrI). Table S20. SUCRA values of PFS for HDACi in ETR patients. Table S21. Effect estimates of PFS for HDACi in ETR patients: hazard ratio (95% credibility interval, CrI). Table  S22. SUCRA values of PFS for visceral metastases in ETR patients (network 1). Table S23. Effect estimates of PFS for visceral metastases in ETR patients: hazard ratio (95% credibility interval, CrI) (network 1). Table S24. SUCRA values of PFS for visceral metastases in ETR patients (network 2). Table S25. Effect estimates of PFS for visceral metastases in ETR patients: hazard ratio (95% credibility interval, CrI) (network 2). Table S26. SUCRA values of PFS for bone metastases in ETR patients (network 1). Table S27. Effect estimates of PFS for bone metastases in ETR patients: hazard ratio (95% credibility interval, CrI) (network 1). Table S28. SUCRA values of PFS for bone metastases in ETR patients (network 2). Table S29. Effect estimates of PFS for bone metastases in ETR patients: hazard ratio (95% credibility interval, CrI) (network 2). Table S30. SUCRA values of PFS for post-menopausal women in ETR patients (network 1). Table S31. Effect estimates of PFS for post-menopausal women in ETR patients: hazard ratio (95% credibility interval, CrI) (network 1). Table S32. SUCRA values of PFS for post-menopausal women in ETR patients (network 2). Table S33. Effect estimates of PFS for post-menopausal women in ETR patients: hazard ratio (95% credibility interval, CrI) (network 2). Table S34. SUCRA values of PFS for pre-menopausal women in ETR patients (network 1). Table  S35. Effect estimates of PFS for pre-menopausal women in ETR patients: hazard ratio (95% credibility interval, CrI) (network 1). Table S36. SUCRA values of PFS for pre-menopausal women in ETR patients (network 2). Table S37. Effect estimates of PFS for pre-menopausal women in ETR patients: hazard ratio (95% credibility interval, CrI) (network 2). Figure S1. Risk of bias across studies (main analysis in ETS and ETR patients). Figure S2. Risk of bias across studies (subgroup analysis for targeted regimens in ETS patients). Figure S3. Risk of bias across studies (subgroup analysis for targeted regimens and menopausal women in ETR patients). Figure S4. Quality assessment data for individual studies. Figure S5. Chemical structure of fulvestrant.
Author Contributions: B.X. has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Conception and design: B.X. and J.W.