PI3K indicates phosphoinositide 3-kinase. Dotted line indicates 30% decrease in size of target lesions, consistent with partial response according to Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1. Asterisks indicate patients with progression in nontarget or new lesions designated as having 20% progression. Plus sign indicates patient with a subcentimeter lesion that was evaluable but not measurable by RECIST. This lesion decreased by a maximum dimension of 40% with therapy and was counted as a partial response.
A, Incidence of phosphoinositide 3-kinase (PI3K) pathway aberrations in patients with metaplastic breast cancer. Only mutations predicted to be pathway activating are depicted. Some tumors had more than 1 aberration detected. B, Spectrum of aberrations seen in patients with metaplastic breast cancer, detected by means of polymerase chain reaction–based DNA sequencing of hot spots in select genes of interest, next-generation sequencing through Foundation Medicine, and immunohistochemical analysis for PTEN loss. Mutations predicted to be germline variants are excluded. Response for each patient is also depicted except one; this patient died due to pneumonia after 2 cycles of DAT and was not evaluated for response.
eTable 1. Baseline Patient Characteristics
eTable 2. Grade 3-4 Toxicity with DAT/DAE per Dose Level
eFigure. Progression-Free Survival in Metaplastic TNBC Patients with and without PI3K Pathway Aberrations
Customize your JAMA Network experience by selecting one or more topics from the list below.
Basho RK, Gilcrease M, Murthy RK, et al. Targeting the PI3K/AKT/mTOR Pathway for the Treatment of Mesenchymal Triple-Negative Breast CancerEvidence From a Phase 1 Trial of mTOR Inhibition in Combination With Liposomal Doxorubicin and Bevacizumab. JAMA Oncol. 2017;3(4):509–515. doi:10.1001/jamaoncol.2016.5281
Copyright 2016 American Medical Association. All Rights Reserved.
Using metaplastic breast cancer as a surrogate, is phosphoinositide 3-kinase (PI3K) pathway–directed therapy effective in the treatment of mesenchymal triple-negative breast cancer (TNBC)?
In this cohort of patients with metaplastic TNBC, the combination of liposomal doxorubicin, bevacizumab, and the mammalian target of rapamycin inhibitors temsirolimus or everolimus had a higher objective response rate than has historically been seen in this chemorefractory subset of tumors. There was a high incidence of PI3K pathway aberrations, and patients whose tumors had pathway aberrations had a significantly higher objective response rate compared with patients with tumors lacking pathway aberrations.
Further investigation of PI3K pathway–directed therapy is warranted in the treatment of metaplastic TNBC, as well as mesenchymal TNBC as a whole, once diagnostic assays are available.
Triple-negative breast cancer (TNBC) classified by transcriptional profiling as the mesenchymal subtype frequently harbors aberrations in the phosphoinositide 3-kinase (PI3K) pathway, raising the possibility of targeting this pathway to enhance chemotherapy response. Up to 30% of mesenchymal TNBC can be classified histologically as metaplastic breast cancer, a chemorefractory group of tumors with a mixture of epithelial and mesenchymal components identifiable by light microscopy. While assays to identify mesenchymal TNBC are under development, metaplastic breast cancer serves as a clinically identifiable surrogate to evaluate potential regimens for mesenchymal TNBC.
To assess safety and efficacy of mammalian target of rapamycin (mTOR) inhibition in combination with liposomal doxorubicin and bevacizumab in patients with advanced metaplastic TNBC.
Design, Setting, and Participants
Phase 1 study with dose escalation and dose expansion at the University of Texas MD Anderson Cancer Center of patients with advanced metaplastic TNBC. Patients were enrolled from April 16, 2009, to November 4, 2014, and followed for outcomes with a cutoff date of November 1, 2015, for data analysis.
Liposomal doxorubicin, bevacizumab, and the mTOR inhibitors temsirolimus or everolimus using 21-day cycles.
Main Outcomes and Measures
Safety and response. When available, archived tissue was evaluated for aberrations in the PI3K pathway.
Fifty-two women with metaplastic TNBC (median age, 58 years; range, 37-79 years) were treated with liposomal doxorubicin, bevacizumab, and temsirolimus (DAT) (N = 39) or liposomal doxorubicin, bevacizumab, and everolimus (DAE) (N = 13). The objective response rate was 21% (complete response = 4 [8%]; partial response = 7 [13%]) and 10 (19%) patients had stable disease for at least 6 months, for a clinical benefit rate of 40%. Tissue was available for testing in 43 patients, and 32 (74%) had a PI3K pathway aberration. Presence of PI3K pathway aberration was associated with a significant improvement in objective response rate (31% vs 0%; P = .04) but not clinical benefit rate (44% vs 45%; P > .99).
Conclusions and Relevance
Using metaplastic TNBC as a surrogate for mesenchymal TNBC, DAT and DAE had notable activity in mesenchymal TNBC. Objective response was limited to patients with PI3K pathway aberration. A randomized trial should be performed to test DAT and DAE for metaplastic TNBC, as well as nonmetaplastic, mesenchymal TNBC, especially when PI3K pathway aberrations are identified.
Triple-negative breast cancer (TNBC) is defined as invasive breast carcinoma that lacks estrogen receptor expression, progesterone receptor expression, and human epidermal growth factor receptor 2 overexpression. It is a heterogeneous group of tumors with varying prognoses.1 Because of the lack of well-defined molecular targets, cytotoxic chemotherapy remains the mainstay of treatment, and prognosis is poor in patients whose disease is resistant to standard chemotherapy.2-4 Recently, gene expression profiling has identified subgroups of TNBC with unique molecular aberrations that may be actionable.1,5 Mesenchymal TNBC is one such group that comprises approximately 30% of TNBCs.1,5 This subgroup has been labeled by various investigators as “claudin-low,” “mesenchymal,” or “mesenchymal stem–like” and has been associated with a worse prognosis.1,5-9 Mesenchymal TNBCs are enriched in epithelial-to-mesenchymal transition (EMT) and cancer stem cell (CSC) features and contain frequent aberrations in the phosphoinositide 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) pathway, raising the possibility of targeting this axis for treatment.1,5,7,8 Mesenchymal stem–like tumors are also enriched in genes involved in angiogenesis including VEGFR2, EPAS1, TEK, and TIE1.5
Although Clinical Laboratory Improvement Amendment–certified diagnostic assays are under development, at this time, the clinical identification of mesenchymal TNBC remains a challenge, making patient selection for therapeutic trials of targeted therapy difficult. Metaplastic breast cancer is a rare subtype of TNBC composed of a diverse group of histologic subtypes in which epithelial and mesenchymal components are admixed.10-16 Approximately 10% to 30% of tumors profiled as mesenchymal TNBCs are metaplastic on the basis of morphologic features identified by light microscopy.7,17 Similar to mesenchymal TNBCs, metaplastic TNBCs are enriched in CSC and EMT features, are chemorefractory, and carry a poor prognosis.8,18-21 Furthermore, metaplastic TNBCs also have frequent aberrations in the PI3K/AKT/mTOR pathway and prominent expression of vascular endothelial growth factor (VEGF).8 As a result, metaplastic TNBC is a subset of mesenchymal TNBC that can be recognized by light microscopy and used as a clinically identifiable population to validate targeted regimens for mesenchymal TNBC.
On the basis of these data, patients with metastatic metaplastic breast cancer were enrolled in a phase 1 trial of doxorubicin, bevacizumab, and temsirolimus (DAT) or liposomal doxorubicin, bevacizumab, and everolimus (DAE) at MD Anderson. Toxicity and maximum tolerated dose have been previously reported for DAT; so herein, we report the efficacy results for patients with metaplastic TNBC treated with DAT or DAE.22
A phase 1 trial with dose escalation of DAT conducted in patients with advanced solid malignant tumors at MD Anderson has been previously reported (NCT00761644).22 Expansion cohorts were undertaken in specific tumor types including metaplastic TNBC due to an observed response signal during dose escalation. The MD Anderson institutional review board approved the study, and written informed consent was obtained from all patients. Treatment was conducted at MD Anderson. Later in the trial, the protocol was amended to allow treatment with DAE because oral administration of everolimus was less cumbersome compared with weekly intravenous temsirolimus. Additional patients with metaplastic TNBC were also treated with these regimens off study but per protocol guidelines using a second institutional review board–approved protocol to observe patients for documentation of toxic effects and response. Patients were enrolled from April 16, 2009, to November 4, 2014, and followed for outcomes with a cutoff date of November 1, 2015, for data analysis. Treatment was administered on an outpatient basis with intravenous liposomal doxorubicin and bevacizumab on the first day of every cycle with weekly intravenous temsirolimus (DAT) or daily oral everolimus (DAE). Each cycle was 21 days.
Herein, we report the results for 52 patients with metaplastic TNBC treated with DAT or DAE. The diagnosis of metaplastic TNBC was confirmed by pathology review at MD Anderson. Of note, 2 patients were found to have invasive ductal carcinoma on biopsy of localized disease and squamous metaplasia on pathology review of subsequent metastatic lesions. These patients were considered to have metaplastic TNBC and were included in the study. All patients who received 1 dose of any of the study agents were considered evaluable for safety and efficacy. The severity of adverse events was graded according to the Common Terminology Criteria for Adverse Events, version 3.0.23 Patients with prior cumulative doxorubicin dose more than 300 mg/m2 were excluded, and all patients underwent close cardiac monitoring as defined in the phase 1 study.22 Radiographic imaging studies were repeated after every 2 cycles (6 weeks) of therapy to assess response using Response Evaluation Criteria in Solid Tumors.24
When archived tissue was available, DNA was analyzed by a polymerase chain reaction–based DNA sequencing method for hot spot mutations in select genes of interest.25,26 After 2011, mass spectrometric detection (Sequenom MassARRAY) was used, and after 2012, a 46- or 50-gene Ampliseq Ion Torrent Assay was used.27,28 When possible, loss of PTEN was assessed using immunohistochemical analysis (Dako). Loss of PTEN was defined as the complete absence of PTEN staining in tumor cells. All of this molecular testing was conducted in a Clinical Laboratory Improvement Amendment–certified molecular diagnostic laboratory at MD Anderson. Additionally, tumors from some patients underwent next-generation sequencing through Foundation Medicine (182- and 236-gene panel). The numbers of patients who underwent various types of molecular testing are presented in the Table. Only patients who had some form of testing for mutations in PIK3CA were included in analyses based on mutation status.
Descriptive summary statistics were used to assess demographic characteristics, safety, and antitumor activity. Categorical data were summarized using frequencies and relative frequencies (ie, proportions). Confidence intervals for proportions were computed using the exact binomial method. Continuous data were summarized by median and range. Differences in categorical variables were assessed using Fisher exact tests. Progression-free survival was estimated using the Kaplan-Meier method and compared between groups using Cox proportional hazards regression analysis. Statistical inferences were based on 2-sided tests at a significance level of P < .05.
A total of 52 female patients with metaplastic TNBC (median age, 58 years; range, 37-79 years) were treated with DAT (N = 39; 2 dose levels) or DAE (N = 13; 3 dose levels). Bevacizumab was omitted in 1 patient treated with DAT due to a history of gastrointestinal bleeding. Baseline patient characteristics are presented in eTable 1 in the Supplement. The median number of prior regimens for metastatic disease was 1 (range, 0-5). Most patients had received prior anthracycline-based (39 [75%]) and/or taxane-based (42 [81%]) therapy, either in the localized or metastatic setting.
Overall, therapy was well tolerated in all patients enrolled in the phase 1 trial that established the maximum tolerated dose of DAT, and toxicity was similar for DAE.22 The incidence of grade 3 to 4 toxic effects in patients with metaplastic TNBC treated with DAT or DAE is presented in eTable 2 in the Supplement. Generally, grade 3 to 4 toxic effects were managed by dose reductions and/or delays in treatment. One patient discontinued treatment with DAT due to adverse events including grade 3 mucositis and grade 3 pancreatitis. One patient died with pneumonia after 1 cycle of DAT, but this was not believed to be treatment related because the patient had a history of postobstructive pneumonia due to pulmonary metastases. The sole patient treated with everolimus, 10 mg, had grade 3 mucositis. Furthermore, at 1 dose level below, with everolimus dosed at 7.5 mg daily, 1 of 9 patients experienced grade 4 mucositis with bleeding gastric ulcers requiring hospitalization and urgent transfusion of blood. No other grade 3 to 4 mucositis events were observed at this dose level. Grade 3 heart failure was seen in 1 patient treated with DAT; this patient had prior exposure to anthracyclines in both the adjuvant and metastatic setting. Additionally, 1 patient treated with DAE had grade 2 pulmonary toxic effects; this patient had a history of everolimus-associated pulmonary toxic effects.
Patients who were removed from the study before the first scheduled restaging workup because of discontinuation of treatment were excluded from waterfall plots of best response. Patients who were removed from the study because of progression in nonmeasurable or new lesions were arbitrarily designated as having 20% increase in tumor size on waterfall plots showing best tumor response and were counted as progression events. All responses were confirmed on at least 1 subsequent staging evaluation. The objective response rate (ORR) was 21% (complete response [CR] = 4 [8%]; partial response [PR] = 7 [13%]; 95% CI, 11%-35%), and 10 (19%) patients had stable disease (SD) for at least 6 months, for a clinical benefit rate (CBR) of 40% (95% CI, 27%-55%). Figure 1A displays the waterfall plot of best response for all patients with metaplastic TNBC.
Tissue was available for testing in 43 patients and 32 (74%) of them had PI3K pathway aberrations that were considered to be pathway activating (Figure 2). In patients with PI3K pathway activation, the ORR was 31% (CR = 4 [13%]; PR = 6 [19%]; 95% CI, 16%-50%), and 4 patients had SD for at least 6 months, for a CBR of 44% (95% CI, 26%-62%). One patient has had a durable CR for more than 5 years, and this patient was found to have a mutation in NF2. No response was seen in the 11 patients lacking PI3K pathway aberrations (ORR, 0%; 95% CI, 0%-28%), but 5 patients had SD for at least 6 months, for a CBR of 45% (95% CI, 17%-77%). As a result, PI3K pathway activation was associated with a significant improvement in ORR (P = .04) but not CBR (P > .99). Figure 1B and 1C display the waterfall plots of best response for patients with and without PI3K pathway activation. The improvement in ORR seen in patients with PI3K pathway activation was also associated with longer progression-free survival, but this was not significant (median, 5.1 vs 2.9 months; P = .35; hazard ratio, 0.69; 95% CI, 0.33-1.46) (eFigure in the Supplement). Because mutation in PIK3CA was the most common aberration seen in the PI3K pathway, responses were further assessed by their location in the helical or kinase domain of PIK3CA. Outcomes were similar if mutations in PIK3CA were located in the helical or kinase domain (ORR, 22% vs 23%; P > .99; and CBR, 33% vs 46%; P = .67, respectively).
Metaplastic breast cancer is an aggressive subtype of TNBC that has historically been refractory to standard chemotherapeutic treatments.21 Although rare, metaplastic TNBCs account for several hundred new breast cancer cases in the United States annually, representing a therapeutic dilemma. To our knowledge, we report the largest series to date of prospectively treated patients with metastatic metaplastic TNBC. Response rates with the DAT or DAE regimens, which target the PI3K pathway through mTOR inhibition, were much higher than has historically been reported with chemotherapy in metaplastic TNBC.21 The use of anthracycline-based chemotherapy likely enhanced the response of these regimens due to the shared features between metaplastic TNBCs and sarcomas. In addition to DAT or DAE, various combinations of temsirolimus, bevacizumab, and taxane-, platinum-, or anthracycline-based chemotherapy have been tested in patients with metaplastic TNBC at MD Anderson.29 Six of a total of 24 patients treated achieved response, and all of these patients were treated with liposomal doxorubicin-containing regimens. Patients treated with paclitaxel- and carboplatin-containing regimens did not achieve response. Similarly, in an analysis of 100 breast cancer patients with biphasic metaplastic sarcomatoid carcinoma treated at MD Anderson, patients treated with anthracycline-containing regimens had a higher likelihood of being alive and recurrence-free at 5 years compared with patients treated with non–anthracycline-containing regimens, although this difference was not significant.18 Limited statistical power precludes definitive interpretation of these data, but they raise the hypothesis that anthracyclines are the preferred chemotherapy backbone of choice for the treatment of metaplastic TNBC.
The DAT or DAE regimens were generally well tolerated, with expected adverse events including mucositis, myelosuppression, and hand-foot syndrome. Only 1 patient discontinued treatment as a result of adverse events, and most adverse events were otherwise managed with dose reductions and/or delays. A high incidence of grade 3 to 4 mucositis was seen with higher doses of everolimus, suggesting that 10 mg of everolimus in combination with liposomal doxorubicin and bevacizumab is unlikely to be tolerated. Although 1 patient death was observed with DAT therapy, this was not believed to be treatment related. Despite prior anthracycline exposure in the majority of patients, only 1 patient treated with DAT had grade 3 congestive heart failure. Similarly, grade 2 pulmonary toxic effects with DAE therapy was only seen in 1 patient, and this patient had a history of everolimus-associated lung toxic effects.
In this cohort, PI3K pathway aberrations were present in the majority of patients, suggesting an important role of this pathway in metaplastic TNBC. More importantly, tumor responses in patients with metaplastic TNBC were limited to patients with PI3K pathway aberrations, suggesting that pathway aberrations render metaplastic tumors susceptible to pathway-directed therapy. Interestingly, a subset of patients lacking PI3K pathway aberrations achieved stable disease lasting longer than 6 months, suggesting that even patients with metaplastic TNBC who lack a detectable aberration may benefit from PI3K pathway–directed therapy. The possible benefits of these regimens in this subgroup, however, might be explained by the fact that molecular testing for PI3K pathway aberrations was frequently limited to hot spot testing in select genes of interest, permitting some pathway aberrations to remain undetected. The significantly higher response rate seen in patients with PI3K pathway aberration was associated with longer progression-free survival. However, this was not statistically significant, and larger numbers of patients are needed to confirm this observation.
Increased signaling through the PI3K pathway has been associated with primary and secondary resistance to standard therapy in breast cancer.31-36 However, pathway aberrations have not consistently been associated with improved response to pathway-directed therapy.37,38 It is interesting to note, however, that despite the association of pathway aberrations and response in this cohort of patients with metaplastic TNBC, patients with mutations in PTEN did not respond to DAT or DAE (Figure 2). In contrast, 1 patient achieved a CR that has been durable for longer than 5 years, and this patient had a mutation in NF2, which encodes a tumor suppressor that inhibits the PI3K pathway.30 Before treatment, this patient had several perihepatic abdominal tumor implants, with the largest measuring 3.8 cm in longest dimension, as well as nodular pleural changes measuring 4.1 cm and subcarinal adenopathy measuring 3.4 cm in longest dimension. Thus, the implication of specific pathway aberrations in patients with metaplastic TNBC is a subject that requires ongoing investigation.
The use of bevacizumab for the treatment of breast cancer is controversial as a result of the US Food and Drug Administration’s ruling to withdraw the breast cancer indication in 2011; however, the drug does remain available for the treatment of breast cancer in other countries. Given the design of this phase 1 study, it is impossible to determine whether the use of bevacizumab added any additional benefit to mTOR inhibition in combination with chemotherapy. In preclinical models, treatment with bevacizumab and subsequent intratumoral hypoxia results in increased production of hypoxia-induced factor 1α. Hypoxia-induced factor 1α has been shown to mediate enrichment of the CSC population, which is accompanied by increased levels of phosphorylated AKT, priming tumors for treatment with PI3K pathway–directed therapy.39-41 As such, bevacizumab may hypothetically enhance the effects of mTOR inhibition and should be further studied in randomized clinical trials to determine drug benefit in appropriately selected patients.
The results from this study may have further implications in the treatment of mesenchymal TNBC, a subset that accounts for up to 30% of TNBCs characterized by microarray profiling. Approximately 10% to 30% of tumors identified as claudin-low or mesenchymal TNBCs by gene signature are identified as metaplastic breast cancers using light microscopy. Not surprisingly, metaplastic breast cancers share molecular features with tumors characterized as claudin-low and mesenchymal TNBCs, including EMT features, enrichment of angiogenesis, and frequent PI3K/AKT/mTOR pathway aberrations.5-8 Like metaplastic TNBCs, mesenchymal TNBCs have been associated with poor response to chemotherapy and worse outcomes compared with other subtypes, including lower rates of pathologic complete response with standard anthracycline- or taxane-based chemotherapy and decreased survival, making it crucial to find novel therapeutic strategies.9 Thus, the use of metaplastic TNBC as a clinically recognizable subgroup of mesenchymal TNBC to assess response to potential therapies can provide insight into the treatment of mesenchymal TNBC, which remains identifiable only by gene expression profiling at this time.
This study has limitations due in most part to the rarity of metaplastic breast cancer. First, patients were accrued over a prolonged periodduring which substantial changes developed in the technology of molecular characterization. Thus, tumors analyzed earlier in accrual were not as extensively profiled as those analyzed later. Second, this was not a randomized trial and historical data regarding outcomes of standard chemotherapy in metastatic, metaplastic breast cancer are limited to retrospective reviews of medical records, so it is difficult to gauge the magnitude of impact for the targeted therapy regimen. As such, these limitations emphasize the need for prospective randomized studies.
In this cohort of patients with metaplastic TNBC, the DAT or DAE regimens were well tolerated and led to better outcomes than has historically been seen in this chemorefractory subset of tumors. There was a high incidence of PI3K pathway aberrations, and patients with aberrations were responsive to pathway-directed therapy with the DAT or DAE regimens, with a subset of patients achieving CR. These promising data warrant further investigation of PI3K pathway–directed therapy in the treatment of metaplastic TNBC, as well as mesenchymal TNBC as a whole.
Accepted for Publication: September 2, 2016.
Corresponding Author: Stacy L. Moulder, MD, MS, The University of Texas MD Anderson Cancer Center, Dan L. Duncan Bldg (CPB5.3542), 1515 Holcombe Blvd, Unit 1354, Houston, TX 77030 (firstname.lastname@example.org).
Published Online: November 23, 2016. doi:10.1001/jamaoncol.2016.5281
Author Contributions: Drs Basho and Moulder had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Janku and Moulder contributed equally to this work.
Study concept and design: Kurzrock, Janku, Moulder.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Basho, Valero, Chavez-MacGregor, Janku, Moulder.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Helgason, Hess, Herbrich, Chavez-MacGregor, Janku, Moulder.
Obtained funding: Janku.
Administrative, technical, or material support: Helgason, Karp, Meric-Bernstam, Valero, Subbiah, Kurzrock, Janku, Moulder.
Conflict of Interest Disclosures: Dr Moulder has received consulting fees from Novartis. No other disclosures are reported.
Previous Presentations: This work was presented at the European Cancer Congress; Vienna, Austria; September 25, 2015; the San Antonio Breast Cancer Symposium; December 10, 2015; San Antonio, Texas; and the American Association for Cancer Research Annual Meeting; April 18, 2016; New Orleans, Louisiana.