Garrido-Castro et al. Breast Cancer Research (2020) 22:120 https://doi.org/10.1186/s13058-020-01354-y RESEARCH ARTICLE Open Access Phase 2 study of buparlisib (BKM120), a pan-class I PI3K inhibitor, in patients with metastatic triple-negative breast cancer Ana C. Garrido-Castro1†, Cristina Saura2,3,4†, Romualdo Barroso-Sousa1,5, Hao Guo6, Eva Ciruelos7, Begoña Bermejo8, Joaquin Gavilá9, Violeta Serra10, Aleix Prat11, Laia Paré4, Pamela Céliz4, Patricia Villagrasa4, Yisheng Li12, Jennifer Savoie1, Zhan Xu13, Carlos L. Arteaga14, Ian E. Krop1, David B. Solit15, Gordon B. Mills16,17, Lewis C. Cantley18, Eric P. Winer1, Nancy U. Lin1*† and Jordi Rodon2,4,19† Abstract Background: Treatment options for triple-negative breast cancer remain limited. Activation of the PI3K pathway via loss of PTEN and/or INPP4B is common. Buparlisib is an orally bioavailable, pan-class I PI3K inhibitor. We evaluated the safety and efficacy of buparlisib in patients with metastatic triple-negative breast cancer. Methods: This was a single-arm phase 2 study enrolling patients with triple-negative metastatic breast cancer. Patients were treated with buparlisib at a starting dose of 100 mg daily. The primary endpoint was clinical benefit, defined as confirmed complete response (CR), partial response (PR), or stable disease (SD) for ≥ 4 months, per RECI ST 1.1. Secondary endpoints included progression-free survival (PFS), overall survival (OS), and toxicity. A subset of patients underwent pre- and on-treatment tumor tissue biopsies for correlative studies. Results: Fifty patients were enrolled. Median number of cycles was 2 (range 1–10). The clinical benefit rate was 12% (6 patients, all SD ≥ 4 months). Median PFS was 1.8 months (95% confidence interval [CI] 1.6–2.3). Median OS was 11.2 months (95% CI 6.2–25). The most frequent adverse events were fatigue (58% all grades, 8% grade 3), nausea (34% all grades, none grade 3), hyperglycemia (34% all grades, 4% grade 3), and anorexia (30% all grades, 2% grade 3). Eighteen percent of patients experienced depression (12% grade 1, 6% grade 2) and anxiety (10% grade 1, 8% grade 2). Alterations in PIK3CA/AKT1/PTEN were present in 6/27 patients with available targeted DNA sequencing (MSK-IMPACT), 3 of whom achieved SD as best overall response though none with clinical benefit ≥ 4 months. Of five patients with paired baseline and on-treatment biopsies, reverse phase protein arrays (RPPA) analysis demonstrated reduction of S6 phosphorylation in 2 of 3 patients who achieved SD, and in none of the patients with progressive disease. (Continued on next page) * Correspondence: nlin@partners.org † Ana C. Garrido-Castro and Cristina Saura contributed equally to this work. † Nancy U. Lin and Jordi Rodon are co-senior authors. 1 Department of Medical Oncology, Susan F. Smith Center for Women’s Cancers, Dana-Farber Cancer Institute, Harvard Medical School, 450 Brookline Avenue, Boston, MA 02215, USA Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 2 of 13 (Continued from previous page) Conclusions: Buparlisib was associated with prolonged SD in a very small subset of patients with triple-negative breast cancer; however, no confirmed objective responses were observed. Downmodulation of key nodes in the PI3K pathway was observed in patients who achieved SD. PI3K pathway inhibition alone may be insufficient as a therapeutic strategy for triple-negative breast cancer. Trial registration: NCT01790932. Registered on 13 February 2013; NCT01629615. Registered on 27 June 2012. Keywords: Buparlisib, BKM120, Triple-negative breast cancer, PI3K pathway, Phase 1 Introduction potential expansion cohorts were to be performed in Triple-negative breast cancer (estrogen receptor [ER]- order to explore activity in a potential biomarker- negative, progesterone receptor [PgR]-negative, and hu- selected patient population. man epidermal growth factor receptor 2 [HER2]-nega- tive) accounts for approximately 12–17% of all breast Patients and methods cancers and is associated with poor outcomes, with an Patients increase in the risk of recurrence compared to other The study enrolled patients with metastatic breast cancer breast cancer subtypes and more frequent spread to vis- who met the following key criteria: pathologically con- ceral organs and the central nervous system (CNS) [1– firmed invasive breast cancer, ER- and PgR-negative (de- 4]. Median survival after a diagnosis of metastatic triple- fined as < 1% staining by immunohistochemistry [IHC]) negative breast cancer is only approximately 1 year with [12], and HER2-negative by IHC or FISH/CISH per local current chemotherapy regimens [3, 5]. At the time this assessment [13]. If a patient had more than one histo- study was designed, cytotoxic chemotherapy was the logical result, the most recent one was considered for in- mainstay of treatment, and no targeted or immunothera- clusion. Eligibility criteria included age 18 years or older; peutic agents had yet gained an indication in this setting. Eastern Cooperative Oncology Group performance status The phosphatidylinositol-3-kinase (PI3K) pathway is in- 0–2; at least one measurable lesion according to RECIST volved in diverse cellular functions, including cell growth 1.1; adequate bone marrow, hepatic, and renal function; and metabolism, migration, proliferation, and survival [6]. and fasting plasma glucose ≤ 140 mg/dL. Availability of an The molecular alterations leading to activation of the PI3K archival tumor specimen was also required from primary pathway in cancer are varied and include mutations in the or metastatic disease. Key exclusion criteria were previous genes encoding the PI3K alpha (PIK3CA) and beta (PIK3CB) treatment with any PI3K inhibitor; symptomatic CNS me- catalytic subunits, AKT1 mutations, and loss of expression tastases; concurrent use of moderate to strong modifiers of the phosphatidylinositol-3,4,5 trisphosphate (PIP3) phos- of CYP3A4; or pre-existing significant mood disorder, in- phatases PTEN and INPP4B. Mutations in PIK3CA and cluding active major depressive episode, bipolar disorder, AKT1 are relatively uncommon in triple-negative breast schizophrenia; Common Terminology Criteria for Ad- cancer [7]; however, loss of PTEN and/or INPP4B has been verse Events (CTCAE) grade ≥ 3 anxiety, or a history of reported in 30–60% of basal-like tumors [7–9]. suicide attempt or homicidal ideation. When the trial first Buparlisib (BKM120, Novartis) is an orally bioavail- opened, there was no limit on the number of prior lines of able, potent, pan-class I PI3K inhibitor. In the phase 1 chemotherapy for metastatic breast cancer. After 30 pa- dose-escalation study enrolling patients with advanced tients were enrolled, the trial was amended to restrict the solid tumors, 100 mg daily was identified as the max- number of prior lines to 0–2 to include a less heavily pre- imum tolerated dose, and one confirmed partial re- treated population. sponse was observed in a patient with triple-negative breast cancer [10, 11]. Dose-limiting toxicities were Study design mood alteration, rash, and hyperglycemia. This was a single-arm, phase 2 study. Two parallel pro- The primary objective of this phase 2 study was to tocols enrolled patients in the USA (Dana-Farber Cancer evaluate the clinical activity of buparlisib monotherapy Institute, Boston, MA) or Spain (coordinating center, in patients with metastatic triple-negative breast cancer. Vall d’Hebron, Barcelona) with a pre-specified combined Secondary objectives included progression-free survival analysis plan. All patients received buparlisib at a start- (PFS), overall survival (OS), and further characterization ing dose of 100 mg once daily. The planned duration of of the toxicity profile. A subset of patients underwent one cycle of therapy was 4 weeks. Buparlisib dose modi- pre- and on-treatment biopsies for correlative studies. If fications were per protocol according to a pre-defined sufficient clinical activity was observed in the first stage algorithm. Up to two dose reductions were allowed (to of the study, planned biomarker analyses to define 80 mg daily and 60 mg daily). Safety assessments, Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 3 of 13 including fasting plasma glucose, were conducted every the mean expression values in log2. Lysates were probed 2 weeks for the first 2 cycles, and then every 4 weeks. with antibodies validated for RPPA that are enriched for Mood alterations were further assessed using the Patient components of the PI3K pathway. The RPPA data were Health Questionnaire (PHQ)-9 and Generalized Anxiety median-centered and scaled to one standard deviation Disorder Scale (GAD)-7 on the same schedule [14, 15]. before performing analyses. RPPA level 4 was down- Adverse events were assessed according to the National loaded from the Cancer Proteome Atlas (TCPA) data Cancer Institute CTCAE version 4.0. Tumor assessments portal (www.TCPAportal.org) and samples with PAM50 occurred every 2 cycles. Confirmation of response at intrinsic subtype classification were selected. The cluster least 4 weeks later was required to deem a confirmed analyses were displayed using Java Treeview version 3.0. complete response (CR) or partial response (PR). Paired Centroid linkage hierarchical clustering was performed tumor biopsies at baseline and end of cycle 1 were re- using Gene Cluster v3.0. Biomarker data were analyzed quested in patients with accessible disease. A minimum using R (version 3.5.1). of 10 patients with paired biopsies enrolled within the The PI3K RPPA proteomic signature (PI3K score) was first stage (prior to the interim analysis) was targeted. calculated as a composite biomarker PI3K score = [pAkt + Patients were treated until disease progression, un- pmTOR + pGSK3 + pS6K + pS6 + p4EBP1] – [INPP4b + acceptable toxicity, or withdrawal of consent. PTEN]10. v3.5.1 (http://cran.r-project.org). This trial was registered with the US National Institutes of Health (ClinicalTrials.gov identifier: NCT01790932, Statistical analysis NCT01629615). All patients provided written informed The study was designed in two parts. In part 1, up to 50 consent before the initiation of any study-related proce- patients were to be enrolled in order to describe the effi- dures or treatments. cacy of buparlisib in a molecularly unselected popula- tion. An interim analysis was planned after the first 29 Study endpoints patients were enrolled (first stage). Assuming a prior dis- The primary study endpoint was clinical benefit, defined tribution of beta (0.5, 9.5) for the response rate (p), the as confirmed CR, confirmed PR or stable disease (SD) ≥ early stopping criterion of prob (p < 0.05 | data) > 0.95 4 months, as assessed by RECIST 1.1 [16]. Secondary would lead to a stopping boundary of 0 patient with endpoints included PFS, OS, and toxicity profile. PFS clinical benefit out of the first 29 patients. The chosen was defined as the time from date of enrollment until prior distribution yields a prior mean response rate of date of disease progression or death of any cause, which- 0.05 based on a prior (hypothetical) sample size of 10 ever occurred first. If patients were alive without disease patients. If any patient experienced clinical benefit, part progression at the end of the study, their PFS time was 1 of the study was to proceed to full accrual up to 50 pa- censored at the date of last radiographic tumor assess- tients (second stage). After 50 patients were enrolled, ment. OS was defined as time from date of enrollment tumor analysis was to be conducted, with correlation to until date of death from any cause. If patients were alive clinical benefit. If clinical benefit was observed in one or at the end of the study, their OS time was censored at more molecular subgroups (e.g., loss of PTEN or the date of last known alive. INPP4B), then part 2 of the study would open to expand specific subgroups. Patient disposition, efficacy, and Targeted DNA sequencing data safety analyses were performed using SAS. DNA was extracted from formalin-fixed paraffin- embedded (FFPE) tumor tissue samples, either archival Results tumor or research tumor tissue biopsies, for targeted Patients and treatment panel sequencing. Targeted panel next-generation se- Fifty women were enrolled between June 2012 and Sep- quencing was performed using MSK-IMPACT testing, tember 2014 (Table 1). Median age was 53 years (range following previously described methods [17], using panel 29–79). Forty-four (44/50, 88%) patients had received version IMPACT410. neoadjuvant or adjuvant chemotherapy. Patients received a median of 1.5 lines of prior chemotherapy for meta- Proteomic data static disease (range 0–7). Proteomic data were generated by reverse phase protein Follow-up information is available through September arrays (RPPA) in 5 patients with available tissue from 30, 2015, for a median follow-up time of 13.8 months. paired baseline and on-treatment samples. RPPA was The reasons for discontinuation of protocol therapy performed in the MD Anderson Cancer Center Func- were progressive disease (n = 41; 82%), unacceptable tox- tional Proteomics Core Facility as previously described icity (n = 7; 14%), physician discretion (n = 1; 2%), and [18, 19]. Following sonication, samples were quantitated withdrawal of consent (n = 1; 2%). Toxicities leading to with BCA Assay Kit and protein levels were expressed as discontinuation were grade 2 or 3 rash (n = 3), persistent Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 4 of 13 Table 1 Baseline characteristics Table 1 Baseline characteristics (Continued) Characteristic No. of patients (%) Characteristic No. of patients (%) Age, years Platinum 18 (36) Median (range) 53 (29–79) Capecitabine 23 (46) Sex Eribulin 8 (16) Female 50 (100) Other 26 (52) Male 0 (0) CNS central nervous system, ECOG PS Eastern Cooperative Oncology Group Performance Status Race White 46 (92) Black 1 (2) grade 2 or higher transaminitis (n = 1), grade 3 hypergly- cemia (n = 1), grade 3 fatigue and grade 2 behavior alter- Asian 1 (2) ation (n = 1), and intercurrent illness (n = 1). At the time Other 2 (4) of data-lock, all patients had discontinued protocol ther- Ethnicity apy, 29 patients had died, 17 patients were still alive, and Hispanic or Latino 12 (24) 4 patients were lost to follow-up. Non-Hispanic 37 (74) Unknown 1 (2) Efficacy Of 50 patients enrolled, 37 patients were evaluable for best ECOG PS at baseline response, 12 patients were taken off treatment before the 0 32 (64) first restaging evaluation due to clinical progression, and one 1 18 (36) patient was taken off treatment early due to toxicity (Table 2). Disease-free interval (from primary diagnosis to metastatic The first stage of accrual had a protocol-specified target ac- diagnosis) crual of 29 patients; however, one more patient was enrolled De novo metastatic disease 5 (10) in this stage (with prior approval of a deviation) due to the ≤ 2 years 25 (50) protocol requirement to also enroll at least 10 patients with > 2 years 20 (40) paired metastatic biopsies. Out of the first 30 patients, 3 ex- perienced SD ≥ 4 months, meeting the study pre-defined pri- Sites of disease at inclusion mary endpoint of clinical benefit and failing to cross the CNS 2 (4) early stopping boundary. Accrual thus continued toward a Lung or pleural effusion 13 (26) total of 50 patients, among which six (6/50, 12%) experi- Liver 5 (10) enced clinical benefit of SD ≥ 4 months. One patient had an Bone 8 (16) unconfirmed PR; however, there were no confirmed CR or Breast or chest wall 15 (30) PR. Median PFS was 1.8 months (95% CI 1.6–2.3) (Fig. 1a). Median OS was 11.2 months (95% CI 6.2–25) (Fig. 1b). Lymph nodes 20 (40) Soft tissue 2 (4) Safety Other 4 (8) A total of 143 cycles of buparlisib were administered Previous treatment (Supplemental Table 1). The median number of cycles Adjuvant or neoadjuvant chemotherapy 44 (88) per patient was 2 (range 1–10). Dosing was held in 48% Adjuvant or neoadjuvant anthracycline 36 (72) of patients; however, only 7 (14%) patients required a dose reduction. The most common reason for dose delay Adjuvant or neoadjuvant taxane 42 (84) or hold was hepatic toxicity (6 patients). Lines of chemotherapy for metastatic or recurrent disease Adverse events are summarized in Table 3. The most fre- Median (range) 1.5 (0–7) quently reported adverse events (all relatedness) were fa- None 7 (14) tigue (58% all grades, 8% grade 3), nausea (34% all grades, 1 line 18 (36) none grade 3), hyperglycemia (34% all grades, 4% grade 3), 2 lines 9 (18) and anorexia (30% all grades, 2% grade 3). Any grade de- pression and anxiety were each reported in 9 patients, none 3 or more lines 16 (32) of which were grade 3. QTc prolongation was noted in 13 Prior metastatic chemotherapy patients (10 [20%] grade 1; 3 [6%] grade 2; no grade 3 or 4). Anthracycline 4 (8) As expected, hyperglycemia was noted, with one (2%) pa- Taxane 14 (28) tient experiencing grade 2 and two (4%) patients experien- cing grade 3 hyperglycemia. Three (6%) patients required a Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 5 of 13 Table 2 Best response by RECIST 1.1 grade 3 depressive and anxiety symptoms according to these Response No. of patients (%) scales. Supplemental Table 2 collates data from patients with Confirmed CR 0 (0) at least one neuropsychiatric assessment by PHQ-9 or GAD- Confirmed PR 0 (0) 7 showing a score ≥ 10, the corresponding reported psychi- atric adverse events by CTCAE 4.0 by investigator assess- SD ≥ 4 months § 6 (12) ment, and medical management of psychiatric adverse SD < 4 months 11 (22) events. Of note, standard investigator-reported CTCAE psy- Progressive disease 20 (40) chiatric adverse events (e.g., depression, anxiety) underesti- Non-evaluable* 13 (26) mated psychiatric symptoms as compared to patient- CR complete response, PR partial response, SD stable disease reported neuropsychiatric outcomes as assessed with the § One patient had an unconfirmed PR PHQ-9 and GAD-7 scales. *12 patients discontinued treatment before first post-baseline assessment due to clinical progression. One patient discontinued treatment before the first post-baseline assessment due to unacceptable toxicity Correlative studies Targeted DNA sequencing was performed using MSK- change or addition of medication for glucose control (n = 1, IMPACT testing on archival or research tumor tissue bi- addition of metformin; n = 1, increase in dose of metfor- opsies. Gene single-nucleotide variants (SNV) and copy min, addition of glimepiride; n = 1, increase in dose of met- number variation (CNV) results were available in tumor formin, addition of insulin). samples from 27 patients, including 20 archival FFPE Results of neuropsychiatric assessments using the PHQ-9 samples and 7 research core biopsies (baseline, n = 5; and GAD-7 instruments are displayed in Table 4. Comple- C1D28, n = 2). Of these 27 samples, 6 (22.2%) harbored tion rates of the instruments exceeded 70% at all timepoints. an alteration in PIK3CA (E542K mutation, n = 1), AKT1 Moderate (grade 2) depressive symptoms were reported by (E17K mutation, n = 1; AKT1 amplification, n = 1), or 11 (22%) patients on study, and in 2 (4%) patients at baseline. PTEN (D24H mutation, n = 1; K237fs mutation, n = 1; Moderate anxiety symptoms were reported by 8 (16%) pa- PTEN deep deletion, n = 1). In the sample with PTEN tients on study, in contrast to 2 (4%) patients at baseline. deletion (log ratio − 2.87), the TP53 variant allele fre- Three patients reported treatment-emergent grade 3 depres- quency was 0.57, supporting likely homozygous loss of sive symptoms, one patient reported grade 3 anxiety symp- PTEN. Additional alterations in genes in the PI3K path- toms, and one patient reported both treatment-emergent way [20] are displayed in Fig. 2. Fig. 1 Kaplan-Meier curves for progression-free survival (a) and overall survival (b) in patients with metastatic, triple-negative breast cancer treated with buparlisib. Median PFS was 1.8 months (95% confidence interval [CI] 1.6–2.3) and median OS was 11.2 months (95% CI 6.2–25) Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 6 of 13 Table 3 All grade adverse events reported in ≥10% of patients or any reported grade 3 events (all relatedness), n = 50 AE description Number (%) of AE by CTCAE grading Frequency G1 G2 G3 Total* Fatigue 15 (30) 10 (20) 4 (8) 29 (58) Nausea 12 (24) 5 (10) 0 (0) 17 (34) Hyperglycemia 14 (28) 1 (2) 2 (4) 17 (34) Anorexia 12 (24) 2 (4) 1 (2) 15 (30) QTc prolongation 10 (20) 3 (6) 0 (0) 13 (26) Alanine aminotransferase increased 3 (6) 2 (4) 5 (10) 10 (20) Anxiety 5 (10) 4 (8) 0 (0) 9 (18) Depression 6 (12) 3 (6) 0 (0) 9 (18) Diarrhea 8 (16) 1 (2) 0 (0) 9 (18) Aspartate aminotransferase increased 3 (6) 1 (2) 4 (8) 8 (16) Psychiatric disorders—other 4 (8) 3 (6) 0 (0) 7 (14) Gastrointestinal disorders—other 5 (10) 0 (0) 1 (2) 6 (12) Rash acneiform 5 (10) 0 (0) 1 (2) 6 (12) Constipation 6 (12) 0 (0) 0 (0) 6 (12) Cough 5 (10) 1 (2) 0 (0) 6 (12) Insomnia 6 (12) 0 (0) 0 (0) 6 (12) Pain 1 (2) 5 (10) 0 (0) 6 (12) Dyspnea 2 (4) 3 (6) 0 (0) 5 (10) Mucositis oral 4 (8) 1 (2) 0 (0) 5 (10) Rash maculo-papular 0 (0) 0 (0) 2 (4) 2 (4) Alkalosis 0 (0) 0 (0) 1 (2) 1 (2) Dry skin 0 (0) 0 (0) 1 (2) 1 (2) Hepatic failure 0 (0) 0 (0) 1 (2) 1 (2) Hyponatremia 0 (0) 0 (0) 1 (2) 1 (2) Nervous system disorders—other 0 (0) 0 (0) 1 (2) 1 (2) Pain in extremity 0 (0) 0 (0) 1 (2) 1 (2) Papulopustular rash 0 (0) 0 (0) 1 (2) 1 (2) Skin and subcutaneous tissue disorders—other 0 (0) 0 (0) 1 (2) 1 (2) AE adverse event, CTCAE Common Terminology Criteria for Adverse Events, G grade *No grade 4 or 5 events were reported As specified in the “Patients and methods” section, the trial response was SD ≥ 4 months (n = 1), SD < 4 months (n = 4), was originally designed to evaluate predictive biomarkers of progressive disease (n = 7), and non-evaluable (n = 9). In the response to buparlisib (with particular emphasis on alter- single patient who achieved clinical benefit ≥ 4 months from ations in the PI3K/PTEN pathway), with potential expansion buparlisib with available SNV/CNV results, splicing muta- cohorts enriched for the identified biomarkers. However, we tions in TP53 and RB1 were identified in the baseline meta- did not observe sufficient clinical activity to move on to ex- static research biopsy. pansion cohorts or to test for potential predictors of clinical Seventeen patients underwent a baseline biopsy, 6 of benefit. Thus, the ability to correlate genomic findings with them had sufficient tissue for proteomic analysis and 5 response to buparlisib was limited. In the six patients with patients had available paired baseline and on-treatment tumor alterations in PIK3CA, AKT1, or PTEN, the best over- biopsies. Baseline samples (n = 6) were analyzed by all response was SD < 4 months (n = 3, including the two pa- RPPA and compared with the TCGA dataset (Fig. 3, tients with AKT1-altered tumors and one patient with PTEN SU2C samples in black are located in close proximity to K237fs mutation), progressive disease (n = 2), and non- triple-negative breast cancer samples from TCGA in evaluable (n = 1). Of the remaining 21 patients without iden- red). All samples from this study clustered among the tified alterations in PIK3CA/AKT1/PTEN, the best overall TCGA triple-negative breast tumors. As shown in Fig. 3, Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 7 of 13 Table 4 Neuropsychiatric assessment Pt02), albeit there was evidence of concomitant feedback Completion rate of PHQ-9 and GAD-7 scales by timepoint activation on AKT T308 and p70S6K T389. With regards PHQ-9 No. patients required No. patients to treatment response, these two patients experienced dis- completed (% of ease stabilization, along with Pt04, for whom there was no required) pathway modulation observed. Pt03 and Pt05 experienced Baseline 50 49 (98) progressive disease as the best overall response and their Cycle 1 day 45 44 (98) tumor samples showed no change in S6 phosphorylation. 15 As previously suggested by Elkabets et al. for the PI3Kα Cycle 2 day 1 40 38 (95) inhibitor alpelisib, reduction of S6 phosphorylation ap- Cycle 2 day 26 26 (100) peared to be a key component of pathway modulation as- 15 sociated to treatment response [21]. Cycle 3 day 1 19 15 (79) Off treatment 50 36 (72) Discussion Given the observation that the majority of triple-negative GAD-7 No. patients required No. patients completed (% of breast tumors demonstrate activation of the PI3K path- required) way, we tested the pan-class I PI3K inhibitor buparlisib in Baseline 50 49 (98) this phase 2 trial. In a small subset of patients in whom Cycle 1 day 45 44 (98) paired baseline and on-treatment tumor biopsies were 15 available, we observed evidence of target inhibition in 2 of Cycle 2 day 1 40 38 (95) 3 patients with SD as best response, and none of 2 patients Cycle 2 day 26 26 (100) with progressive disease as best response. Overall, only 15 12% of patients achieved prolonged SD, and we did not Cycle 3 day 1 19 15 (79) observe any confirmed objective responses. In those with Off treatment 50 36 (72) available targeted panel DNA sequencing results, genomic alterations in PIK3CA, AKT1, or PTEN were present in 3 Assessment of 8 (37.5%) patients who achieved SD, and in 2 of 9 Overall Normal, Grade 1, Grade 2, Grade 3, (22.2%) patients with progressive disease as best overall re- (%) n (%) n (%) n (%) n (%) sponse. Median PFS for the overall study population was PHQ-9 less than 2 months. Given the lack of a strong clinical sig- Baseline 49 33 (67) 14 (29) 2 (4) 0 (0) nal, we chose not to open expansion cohorts in molecu- Highest grade 50 11 (22) 24 (48) 11 (22) 4 (8) larly selected subsets of triple-negative breast cancer. on study Other trials have demonstrated that treatment with GAD-7 single-agent inhibitors of the PI3K pathway has only lim- Baseline 49 33 (67) 14 (29) 2 (4) 0 (0) ited efficacy in patients with metastatic breast cancer, Highest grade 50 21 (42) 19 (38) 8 (16) 2 (4) whether unselected [10] or selected for the presence of on study PIK3CA/AKT/PTEN alterations in the tumor [22]. Our Change to medications for mood disorder study adds to the previous reports by demonstrating that Total (n = 50), n (%) buparlisib has essentially no single-agent activity in un- selected triple-negative breast cancer, and the develop- No - 45 (90) ment of buparlisib has been discontinued in breast Yes - 5 (10) cancer. The toxicity profile we observed is in keeping PHQ-9 and GAD-7 Score Severity CTCAE grading: 0–4 normal; 5–9 grade 1; 10– with other reports of buparlisib at the studied dose and 14 grade 2; ≥ 15 grade 3 GAD-7 Generalized Anxiety Disorder Scale, PHQ-9 Patient schedule. Notably, any grade anxiety and any grade de- Health Questionnaire-9 pression were reported in 18% of patients. Buparlisib is one of the PI3K inhibitors with highest blood-brain bar- five out of six exhibited a PI3K score consistent with rier penetration [23]. However, the physiopathology of PI3K pathway activation. We further analyzed the PI3K mood disorders associated with pan-PI3K inhibitors has pathway downmodulation after treatment with buparli- not been well characterized, and it remains unclear sib in the five patients exhibiting an activated PI3K sig- whether the higher incidence of neuropsychiatric effects nature in whom a paired on-treatment biopsy was also observed with buparlisib compared to isoform-selective available (Fig. 4). Integral downmodulation of key nodes PI3K inhibitors is due to its elevated CNS penetration or in the pathway including AKT S473, S6 S235/S236, S6 to other pharmacokinetic differences. Pan-PI3K inhibi- S240/S244, 4EBP1 S65, and 4EBP1 T37/T46 was exclu- tors have broader off-target effects, in addition to on- sively observed in two paired patient samples (Pt01, target inhibition of each of the class I PI3K catalytic Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 8 of 13 Sample time point Sample type Sample site of disease Best overall response Archival Sample time point Baseline research biopsy C1D28 research biopsy Primary Sample type Metastasis Breast Sample site of disease Lymph node Skin Stable disease 4 months Best overall response Stable disease <4 months Progressive disease Non-evaluable Genetic alteration Fig. 2 Co-mutation/CNV plot of tumor samples analyzed with targeted next-generation sequencing. Co-mutation and copy number variation plot for samples analyzed with MSK-IMPACT panel testing (n = 27 samples; n = 27 patients). Included here are genes related to PI3K pathway and drug resistance with at least one identified alteration in the cohort. Best overall response is displayed for each patient. Six samples harbored an alteration in PIK3CA (E542K mutation, n = 1), AKT1 (E17K mutation, n = 1; AKT1 amplification, n = 1), or PTEN (D24H mutation, n = 1; K237fs mutation, n = 1; PTEN deep deletion, n = 1). Alterations in PIK3CA, AKT1, or PTEN were observed in 3 of 8 patients (37.5%) who achieved SD, and in 2 of 9 (22.2%) patients with progressive disease as best overall response isoforms, thus resulting in a less favorable toxicity profile reportedly had increased median PFS with the addition compared to selective p110-alpha inhibitors [24]. of taselisib (mutant-selective PI3K inhibitor) to fulves- Of interest is the role of combining PI3K inhibitors trant [28]. with other therapeutic modalities. Based on preclinical Other studies have evaluated combinatorial ap- data suggesting that adding a PI3K inhibitor to endo- proaches of PI3K pathway inhibition with chemotherapy. crine therapy may overcome resistance, two large phase In a phase 1 dose escalation study of buparlisib com- 3 clinical trials, BELLE-2 [25] and BELLE-3 [26], evalu- bined with capecitabine in patients with metastatic ated the efficacy of buparlisib plus fulvestrant versus pla- breast cancer (n = 25), one patient with triple-negative cebo plus fulvestrant in postmenopausal patients with breast cancer treated at the maximum-tolerated dose ER-positive breast cancer refractory to aromatase inhibi- (buparlisib 100 mg daily; capecitabine 1000 mg/m2 twice tors. In addition, BELLE-3 required patients to be resist- daily) achieved a CR after receipt of 11 cycles of therapy, ant to mTOR inhibitors. Both trials met their primary and ultimately discontinued treatment after 32 cycles endpoint in the full population with the combination due to mood changes [29]. Genomic testing confirmed leading to modest increases in PFS. Safety data were that the tumor was of basal-like intrinsic subtype, with- consistent with that previously seen with buparlisib- out evidence of an activating PIK3CA mutation. Four based regimens, with transaminase elevation and hyper- additional patients (two with ER-positive, one with glycemia as the most common grade 3–4 adverse events. HER2-positive, and one with triple-negative breast can- Among patients with ER-positive breast cancer whose cer) exhibited a PR to treatment, and the remaining 12 tumors harbor an activating PIK3CA mutation, results of evaluable patients experienced SD, suggesting potential the SOLAR-1 trial demonstrated a clear PFS improve- clinical activity. In contrast, the randomized phase 2 trial ment with the alpha-specific PI3K inhibitor, alpelisib, in comparing buparlisib plus paclitaxel versus placebo plus combination with fulvestrant [27]. Similarly, in the paclitaxel in patients with HER2-negative breast cancer SANDPIPER trial, patients with PIK3CA-mutant tumors was stopped early for futility, as the addition of the PI3K Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 9 of 13 HER3 IRS1 RICTOR EGFR_pY1068 HER2 HER2_pY1248 IGFBP2 CLAUDIN7 AMPKALPHA AMPKALPHA_pT172 ACC_pS79 ACC1 FASN EEF2K BIM BCL2 AR GATA3 INPP4B 5 PREX1 PDK1_pS241 JNK2 0 VEGFR2 SMAC PR ARAF −5 Basal-like RAD50 RAB25 Her2-enriched DUSP4 GAB2 Luminal A X4EBP1 X4EBP1_pS65 Luminal B X4EBP1_pT37T46 S6_pS240S244 Normal-like NFKBP65_pS536 RB_pS807S811 SU2C samples GSK3ALPHABETA_pS21S9 TUBERIN_pT1462 AKT PTEN ATM STAT5ALPHA SYK PI3KP85 EEF2 GAPDH CHK2 CYCLINE1 CYCLINB1 FOXM1 MSH2 MSH6 BETACATENIN ECADHERIN X53BP1 MTOR COG3 DVL3 EIF4G RBM15 BRD4 ADAR1 JAK2 BRAF BRAF_pS445 MYOSINIIA_pS1943 CASPASE3 TFRC EPPK1 ERCC5 IRF1 PAXILLIN YB1 TSC1 P38_pT180Y182 PDCD4 P16INK4A BAX DJ1 PEA15 X1433ZETA SRC 10 BAK RAD51 5 CMYC ETS1 PEA15_pS116 0 PI3K score CKIT AKT_pS473 AKT_pT308 −5 NDRG1_pT346 LCK −10 MEK1 EGFR P70S6K_pT389 CAVEOLIN1 COLLAGENVI RAB11 BID NCADHERIN HSP70 BCL2A1 TAZ STAT3_pY705 MAPK_pT202Y204 MEK1_pS217S221 SRC_pY527 SHP2_pY542 PKCALPHA PKCALPHA_pS657 YAP YAP_pS127 ANNEXIN1 CD26 PAI1 FIBRONECTIN P21 DIRAS3 JAB1 CD49B SRC_pY416 CHK2_pT68 P53 PARPCLEAVED SNAIL SCD G6PD TRANSGLUTAMINASE Fig. 3 Unsupervised clustering and PI3K signature score based on the RBN dataset. Heatmap depicting protein levels after unsupervised hierarchical clustering of 826 TCGA samples plus 6 baseline SU2C samples and 128 antibodies. Protein levels are indicated on a low-to-high scale (green-black-red). Annotation bars include the PAM50 intrinsic subtype classification. The PI3K score is shown in the lower panel. SU2C samples (black) are located in close proximity to triple-negative breast cancer samples from TCGA (red) inhibitor failed to demonstrate a significant PFS im- treat population (6.2 vs. 4.9 months, hazard ratio [HR] provement in either the overall study population or in 0.60) and in patients with PIK3CA/AKT/PTEN-altered the prespecified cohort of patients with PI3K-activated tumors (9.0 vs. 4.9 months, HR 0.44), leading to the tumors [30]. In the hormone receptor-negative popula- design of the ongoing randomized phase 3 trial explor- tion enrolled on the study (99/416 patients, 24%), me- ing the combination in preselected triple-negative breast dian PFS was numerically inferior with buparlisib cancer with activation of the PI3K pathway compared to placebo (5.5 vs. 9.3 months, respectively). (NCT03337724). The addition of capivasertib, another Notably, ipatasertib, a selective AKT inhibitor, improved selective AKT inhibitor, to paclitaxel also significantly PFS in combination with paclitaxel (compared to pla- increased PFS, with a trend toward longer OS, in pa- cebo plus paclitaxel) as first-line therapy for metastatic tients with prespecified alterations in PI3KCA, AKT, or triple-negative breast cancer in the LOTUS phase 2 trial PTEN; however, this benefit was not seen in patients [31]. This benefit was observed in both the intent-to- without aberrant pathway changes [32]. Confirmation of Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 10 of 13 Fig. 4 Unsupervised clustering based on the delta values of the paired samples. Heatmap depicting the delta values in the paired samples of the PI3K signature-related proteins ordered from upstream (top) to downstream (bottom). Protein levels are indicated on a low-to-high scale (green-black-red) activity will be explored in an ongoing randomized phase alterations [7, 8], it would be important to define 3 study which will enroll patients with untreated locally whether such alterations could also be responsible for de advanced or metastatic triple-negative breast cancer novo resistance to PI3K inhibitors as well. Furthermore, (NCT03997123). Similarly, the alpha-specific PI3K in- given the variable efficacy of targeting the PI3K pathway hibitor alpelisib showed promising activity in combin- to date in unselected or PI3K-activated triple-negative ation with albumin-bound paclitaxel in patients with breast cancer, additional biomarker analyses are needed metastatic HER2-negative breast cancer, most (30/43, to better understand the putative oncogenic role of these 70%) of whom had received at most one prior line of genes in this breast cancer subtype. chemotherapy in the metastatic setting [33]. In the triple- Our study had several limitations. First, we tested only negative breast cancer subgroup with available response single-agent buparlisib and cannot rule out that combination data, unselected for PI3K status, the objective response therapies might provide additional efficacy. Since we initiated rate was 58% (7/12). In the overall study population, nu- this study, the treatment landscape for patients with meta- merically greater objective response rate (74% vs. 48%), static triple-negative breast cancer has evolved considerably, clinical benefit rate (100% vs. 61%), and median PFS (13 and two targeted approaches were recently approved: the vs. 7 months, HR 0.40) were observed among patients with anti-PD-L1 inhibitor atezolizumab in combination with nab- PI3K pathway-activation compared to those without paclitaxel in PD-L1-positive triple-negative breast cancer PIK3CA-activating or PTEN-inactivating mutations, lead- [35], and the PARP inhibitors olaparib or talazoparib for ing to the design of an ongoing phase 3 randomized study germline BRCA1/BRCA2 mutation carriers [36, 37]. There of alpelisib or placebo plus albumin-bound paclitaxel as are now preliminary data supporting the molecular rationale first- or second-line therapy for patients with PIK3CA- to test combinations of PI3K inhibitors with these agents mutant or PTEN loss (without PIK3CA mutation) ad- [38]. For example, based on a strong preclinical synergistic vanced triple-negative breast cancer (NCT04251533). signal combining buparlisib and olaparib [39], a phase 1 clin- It is unclear whether the lack of effectiveness of treat- ical trial evaluating the combination was launched in ad- ment with single-agent pan-PI3K inhibitors in triple- vanced triple-negative breast cancer and ovarian carcinoma negative breast cancer is related to adaptive activation of [40]. Results of the study demonstrated feasibility and prom- compensatory signaling pathways or due to a failure to ising anti-cancer activity for this combination, though trans- achieve sufficient target inhibition and PI3K pathway aminase elevations and neuropsychiatric adverse events were suppression due to toxicity. PTEN loss has been found common and responsible for dose-limiting toxicities. There to lead to a convergent PTEN-null phenotype resistant are also emerging data that PTEN loss is associated with re- to PI3K-alpha inhibition [34]. Given that 60% of triple- sistance to immunotherapy, raising the question of whether negative breast cancers harbor PTEN genomic combinations of inhibitors of PI3K/AKT/mTOR signaling Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 11 of 13 with immune checkpoint inhibitors could provide additional Data analysis and interpretation/manuscript writing/final approval of the clinical efficacy [41]. Second, although we had originally de- manuscript: All authors. signed the study to evaluate predictive biomarkers of re- Funding sponse to buparlisib, with potential expansion cohorts This work was supported by NIH grants Breast SPORE P50 CA1P50CA168504 enriched for the identified biomarkers, we did not observe (PI: E.P.W.), R35 CA197588 (PI: L.C.C.), U54 U54CA210184 (PI: L.C.C.), the Gray Foundation (PI: L.C.C.), the Breast Cancer Research Foundation (PI: L.C.C.), sufficient clinical activity to move on to expansion cohorts or Program of the Entertainment Industry Foundation (SU2C-AACR-DT0209; PI: to test for potential predictors of clinical benefit. In addition, L.C.C.), and Novartis Pharmaceuticals, Inc. The funding sources for the study we could not establish if the degree of PI3K downmodula- were not involved in the collection, analysis, or interpretation of the data. Novartis was provided a draft of the manuscript prior to publication; tion was sufficient to be consistent with antitumor activity. however, they were not involved in the writing of the report or in the decision to submit the paper for publication. N.U.L. and J.R. had full access to all the data in the study and had final responsibility for the decision to Conclusions submit for publication. In summary, buparlisib, when given as a single agent in patients with triple-negative breast cancer, was not Availability of data and materials associated with a strong clinical signal of efficacy, The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. though a small subset of patients did achieve pro- longed SD. Downmodulation of key nodes in the Ethics approval and consent to participate PI3K pathway was observed in patients who achieved This trial was registered with the US National Institutes of Health (ClinicalTrials.gov identifier: NCT01790932, NCT01629615). All patients tumor stabilization. The toxicity profile of buparlisib provided written informed consent before the initiation of any study-related was consistent with previous reports, highlighting the procedures or treatments. pharmacological limitations of pan-PI3K inhibition compared to selective PI3K isoform inhibitors that Consent for publication Not applicable. may achieve improved efficacy with fewer side effects. Although our results do not support additional testing Competing interests of single-agent PI3K inhibitors in triple-negative C.S. reports the following personal financial interests: C.S. has served as consultant, participated in advisory boards or received travel grants from breast cancer, they do not preclude the potential AstraZeneca, Celgene, Daiichi Sankyo, Eisai, F. Hoffmann-La Roche Ltd., Gen- benefit for the efficacy of rational combinations in- omic Health, Merck, Sharp and Dhome España SA, Novartis, Pfizer, Philips cluding selective PI3K pathway inhibitors. Healthwork, Pierre Fabre, prIME Oncology, Puma, Synthon, and Sanofi Aven- tis. C.S. reports the following institutional financial interests, paid directly to Institution: AstraZeneca, Daiichi Sankyo, Eli Lilly and Company, Genentech, Supplementary information Immunomedics, Macrogenics, Merck, Sharp and Dhome España S.A., Novartis, Supplementary information accompanies this paper at https://doi.org/10. Pfizer, Piqur Therapeutics, Puma, Roche, Synthon and Zenith Pharma. 1186/s13058-020-01354-y. R.B-S. has served as an advisor/consultant to Eli Lilly and Roche and has received honoraria from Eli Lilly, Roche, Bristol-Myers Squib, Novartis, Pfizer, and travel, accommodations, or expenses from Roche. Additional file 1. L.C.C. is a founder and member of the SAB and holds equity in Agios Pharmaceuticals and Petra Pharmaceuticals, companies developing drugs for treating cancer. The laboratory of L.C.C also receives funding from Petra. Abbreviations D.B.S. has consulted with and received honoraria from Pfizer, Loxo Oncology, CNS: Central nervous system; CR: Complete response; CTCAE: Common Lilly Oncology, Illumina, and Vivideon Therapuetics. Terminology Criteria for Adverse Events; CNV: Copy number variation; B.B. has participated in advisory boards for MSD, Mylan, Roche, and Novartis. ER: Estrogen receptor; FFPE: Formalin-fixed paraffin-embedded; V.S. has received non-commercial support from Novartis. GAD: Generalized Anxiety Disorder Scale; HER2: Human epidermal growth C.A. serves or has served as an advisor to Novartis, Lilly, Sanofi, RADIUS, factor receptor 2; HR: Hazard ratio; IHC: Immunohistochemistry; OS: Overall ABBVIE, TAIHO Oncology, PUMA Biotechnology, Merck, H3Biomedicine, survival; PFS: Progression-free survival; PgR: Progesterone receptor; Symphogen, OrigiMed, Immunomedics, Petra Pharma, G1 Therapeutics, PHQ: Patient Health Questionnaire; PI3K: Phosphatidylinositol-3-kinase; Athenex, and Daiichi Sankyo. C.A. receives or has received research grant PR: Partial response; RPPA: Reverse phase protein arrays; SD: Stable disease; support from Pfizer, Lilly, RADIUS, Bayer, Takeda, PUMA Biotechnology, and SNV: Single-nucleotide variants; TCPA: The Cancer Proteome Atlas Symphogen. C.A. holds stock options in Provista and Y-TRAP and serves in the Scientific Advisory Board of the Komen Foundation. Acknowledgements N.U.L. receives or has received research support paid directly to the The authors would like to thank the patients who participated in this clinical Institution from Genentech, Pfizer, Merck, and Seattle Genetics. N.U.L. has trial and their families, the referring physicians, and the clinical research served as a consultant or participated in advisory boards for Puma, Seattle coordinators who contributed to the successful completion of this study, in Genetics, Daiichi Sankyo, Denali Therapeutics, and California Institute for particular Jen Savoie (DFCI) and the SOLTI team. The authors would also like Regenerative Medicine. N.U.L. has received travel expense reimbursement to thank Kaitlyn Bifolck for editorial support. from Puma and Seattle Genetics. All other authors declare no relevant conflicts of interest. Authors’ contributions Conception and design: Nancy U. Lin, Jordi Rodon, Cristina Saura, Eric P. Author details 1 Winer, Carlos L. Arteaga, Ian E. Krop, Yisheng Li, David Solit, Gordon Mills, Department of Medical Oncology, Susan F. Smith Center for Women’s Lewis C. Cantley. Provision of study materials or patients: Nancy U. Lin, Jordi Cancers, Dana-Farber Cancer Institute, Harvard Medical School, 450 Brookline Rodon, Cristina Saura, Eric P. Winer, Ian E. Krop, Eva Ciruelos Gil, Begoña Avenue, Boston, MA 02215, USA. 2Department of Medical Oncology, Vall Bermejo, Joaquin Gavilá. Collection and assembly of data: Jennifer Savoie, d’Hebron University Hospital, Universitat Autònoma de Barcelona, Barcelona, Nancy U. Lin, Jordi Rodon, Cristina Saura, Ana C. Garrido-Castro, Romualdo Spain. 3Vall d’Hebron Institute of Oncology, VHIO, Barcelona, Spain. 4SOLTI Barroso-Sousa, Gordon Mills, David Solit, Hao Guo, Zhan Xu, Lewis C. Cantley. Breast Cancer Research Group, Barcelona, Spain. 5Present Address: Hospital Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 12 of 13 Sírio-Libanês, Brasilia, Brazil. 6Division of Biostatistics, Department of Data testing in breast cancer: American Society of Clinical Oncology/College of Sciences, Dana-Farber Cancer Institute, Boston, MA, USA. 7Hospital 12 de American Pathologists clinical practice guideline update. Arch Pathol Lab Octubre, Madrid, Spain. 8Clinic University Hospital, INCLIVA Biomedical Med. 2014;138(2):241–56. Research Institute, CIBERONC-ISCIII, Valencia, Spain. 9Fundación Instituto 14. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression Valenciano De Oncología, Valencia, Spain. 10Experimental Therapeutics severity measure. J Gen Intern Med. 2001;16(9):606–13. Group, Vall d’Hebron Institute of Oncology, Barcelona, Spain. 11Department 15. Spitzer RL, Kroenke K, Williams JB, Lowe B. A brief measure for assessing of Medical Oncology, Translational Genomics and Targeted Therapeutics in generalized anxiety disorder: the GAD-7. Arch Intern Med. 2006;166(10):1092–7. Solid Tumors, IDIBAPS, Hospital Clínic of Barcelona, Barcelona, Spain. 16. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. 12 Department of Biostatistics, Division of Basic Sciences, The University of New response evaluation criteria in solid tumours: revised RECIST guideline Texas MD Anderson Cancer Center, Houston, TX, USA. 13School of (version 1.1). Eur J Cancer. 2009;45(2):228–47. Communication, Northern Arizona University, Flagstaff, AZ, USA. 14Harold C. 17. Cheng DT, Mitchell TN, Zehir A, Shah RH, Benayed R, Syed A, et al. Memorial Simmons Comprehensive Cancer Center, UT Southwestern Medical Center, Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK- Dallas, TX, USA. 15Kravis Center for Molecular Oncology, Memorial Sloan IMPACT): a hybridization capture-based next-generation sequencing clinical assay Kettering Cancer Center, New York, NY, USA. 16Knight Cancer Institute, for solid tumor molecular oncology. J Mol Diagn. 2015;17(3):251–64. Oregon Health & Science University, Portland, OR, USA. 17Present Address: 18. Meric-Bernstam F, Akcakanat A, Chen H, Sahin A, Tarco E, Carkaci S, et al. Division of Basic Science Research, Department of Systems Biology, The Influence of biospecimen variables on proteomic biomarkers in breast University of Texas MD Anderson Cancer Center, Houston, TX, USA. 18Sandra cancer. Clin Cancer Res. 2014;20(14):3870–83. and Edward Meyer Cancer Center, Weill Cornell Medical College, New York, 19. Meric-Bernstam F, Akcakanat A, Chen H, Do KA, Sangai T, Adkins F, et al. NY, USA. 19Present Address: Department of Investigational Cancer PIK3CA/PTEN mutations and Akt activation as markers of sensitivity to Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, allosteric mTOR inhibitors. Clin Cancer Res. 2012;18(6):1777–89. TX, USA. 20. Sanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, et al. Oncogenic signaling pathways in the Cancer Genome Atlas. Cell. 2018; Received: 5 May 2020 Accepted: 11 October 2020 173(2):321–37 e10. 21. Elkabets M, Vora S, Juric D, Morse N, Mino-Kenudson M, Muranen T, et al. mTORC1 inhibition is required for sensitivity to PI3K p110alpha inhibitors in References PIK3CA-mutant breast cancer. Sci Transl Med. 2013;5(196):196ra99. 1. Lin NU, Vanderplas A, Hughes ME, Theriault RL, Edge SB, Wong YN, et al. 22. Xing Y, Lin NU, Maurer MA, Chen H, Mahvash A, Sahin A, et al. Phase II trial Clinicopathologic features, patterns of recurrence, and survival among of AKT inhibitor MK-2206 in patients with advanced breast cancer who women with triple-negative breast cancer in the National Comprehensive have tumors with PIK3CA or AKT mutations, and/or PTEN loss/PTEN Cancer Network. Cancer. 2012;118(22):5463–72. mutation. Breast Cancer Res. 2019;21(1):78. 2. Bauer KR, Brown M, Cress RD, Parise CA, Caggiano V. Descriptive analysis of 23. de Gooijer MC, Zhang P, Buil LCM, Citirikkaya CH, Thota N, Beijnen JH, et al. estrogen receptor (ER)-negative, progesterone receptor (PR)-negative, and Buparlisib is a brain penetrable pan-PI3K inhibitor. Sci Rep. 2018;8(1):10784. HER2-negative invasive breast cancer, the so-called triple-negative 24. Fruman DA, Rommel C. PI3K and cancer: lessons, challenges and phenotype: a population-based study from the California cancer registry. opportunities. Nat Rev Drug Discov. 2014;13(2):140–56. Cancer. 2007;109(9):1721–8. 25. Baselga J, Im SA, Iwata H, Cortes J, De Laurentiis M, Jiang Z, et al. Buparlisib 3. Lin NU, Claus E, Sohl J, Razzak AR, Arnaout A, Winer EP. Sites of distant plus fulvestrant versus placebo plus fulvestrant in postmenopausal, recurrence and clinical outcomes in patients with metastatic triple-negative hormone receptor-positive, HER2-negative, advanced breast cancer (BELLE- breast cancer: high incidence of central nervous system metastases. Cancer. 2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet 2008;113(10):2638–45. Oncol. 2017;18(7):904–16. 4. Dent R, Hanna WM, Trudeau M, Rawlinson E, Sun P, Narod SA. Pattern of 26. Di Leo A, Johnston S, Lee KS, Ciruelos E, Lonning PE, Janni W, et al. metastatic spread in triple-negative breast cancer. Breast Cancer Res Treat. Buparlisib plus fulvestrant in postmenopausal women with hormone- 2009;115(2):423–8. receptor-positive, HER2-negative, advanced breast cancer progressing on or 5. Kassam F, Enright K, Dent R, Dranitsaris G, Myers J, Flynn C, et al. Survival after mTOR inhibition (BELLE-3): a randomised, double-blind, placebo- outcomes for patients with metastatic triple-negative breast cancer: implications controlled, phase 3 trial. Lancet Oncol. 2018;19(1):87–100. for clinical practice and trial design. Clin Breast Cancer. 2009;9(1):29–33. 27. Juric D, Janku F, Rodon J, Burris HA, Mayer IA, Schuler M, et al. Alpelisib plus 6. Thorpe LM, Yuzugullu H, Zhao JJ. PI3K in cancer: divergent roles of isoforms, fulvestrant in PIK3CA-altered and PIK3CA-wild-type estrogen receptor-positive modes of activation and therapeutic targeting. Nat Rev Cancer. 2015;15(1):7–24. advanced breast cancer: a phase 1b clinical trial. JAMA Oncol. 2019;5(2):e184475. 7. Cancer Genome Atlas Network. Comprehensive molecular portraits of 28. Baselga J, Dent SF, Cortés J, Im Y-H, Diéras V, Harbeck N, et al. Phase III human breast tumours. Nature. 2012;490(7418):61–70. study of taselisib (GDC-0032) + fulvestrant (FULV) v FULV in patients (pts) 8. Marty B, Maire V, Gravier E, Rigaill G, Vincent-Salomon A, Kappler M, et al. with estrogen receptor (ER)-positive, PIK3CA-mutant (MUT), locally advanced Frequent PTEN genomic alterations and activated phosphatidylinositol 3- or metastatic breast cancer (MBC): Primary analysis from SANDPIPER. J Clin kinase pathway in basal-like breast cancer cells. Breast Cancer Res. 2008; Oncol. 2018;36(18_suppl):LBA1006-LBA. 10(6):R101. 29. McRee AJ, Marcom PK, Moore DT, Zamboni WC, Kornblum ZA, Hu Z, et al. A 9. Gewinner C, Wang ZC, Richardson A, Teruya-Feldstein J, Etemadmoghadam phase I trial of the PI3K inhibitor buparlisib combined with capecitabine in D, Bowtell D, et al. Evidence that inositol polyphosphate 4-phosphatase patients with metastatic breast cancer. Clin Breast Cancer. 2018;18(4):289–97. type II is a tumor suppressor that inhibits PI3K signaling. Cancer Cell. 2009; 30. Martin M, Chan A, Dirix L, O'Shaughnessy J, Hegg R, Manikhas A, et al. A 16(2):115–25. randomized adaptive phase II/III study of buparlisib, a pan-class I PI3K 10. Rodon J, Brana I, Siu LL, De Jonge MJ, Homji N, Mills D, et al. Phase I dose- inhibitor, combined with paclitaxel for the treatment of HER2- advanced escalation and -expansion study of buparlisib (BKM120), an oral pan-Class I breast cancer (BELLE-4). Ann Oncol. 2017;28(2):313–20. PI3K inhibitor, in patients with advanced solid tumors. Investig New Drugs. 31. Kim SB, Dent R, Im SA, Espie M, Blau S, Tan AR, et al. Ipatasertib plus 2014;32(4):670–81. paclitaxel versus placebo plus paclitaxel as first-line therapy for metastatic 11. Bendell JC, Rodon J, Burris HA, de Jonge M, Verweij J, Birle D, et al. Phase I, triple-negative breast cancer (LOTUS): a multicentre, randomised, double- dose-escalation study of BKM120, an oral pan-Class I PI3K inhibitor, in blind, placebo-controlled, phase 2 trial. Lancet Oncol. 2017;18(10):1360–72. patients with advanced solid tumors. J Clin Oncol. 2012;30(3):282–90. 32. Schmid P, Abraham J, Chan S, Wheatley D, Brunt AM, Nemsadze G, et al. 12. Hammond ME, Hayes DF, Dowsett M, Allred DC, Hagerty KL, Badve S, et al. Capivasertib plus paclitaxel versus placebo plus paclitaxel as first-line American Society of Clinical Oncology/College of American Pathologists therapy for metastatic triple-negative breast cancer: the PAKT trial. J Clin guideline recommendations for immunohistochemical testing of estrogen Oncol. 2020;38(5):423–33. and progesterone receptors in breast cancer (unabridged version). Arch 33. Sharma P, Abramson VG, O'Dea A, Pathak HB, Pessetto ZY, Wang YY, et al. Pathol Lab Med. 2010;134(7):e48–72. Clinical and biomarker results from phase I/II study of PI3K inhibitor BYL 719 13. Wolff AC, Hammond ME, Hicks DG, Dowsett M, McShane LM, Allison KH, (alpelisib) plus nab-paclitaxel in HER2-negative metastatic breast cancer. J et al. Recommendations for human epidermal growth factor receptor 2 Clin Oncol 2018;36(15_suppl):1018. Garrido-Castro et al. Breast Cancer Research (2020) 22:120 Page 13 of 13 34. Juric D, Castel P, Griffith M, Griffith OL, Won HH, Ellis H, et al. Convergent loss of PTEN leads to clinical resistance to a PI (3) Kalpha inhibitor. Nature. 2015;518(7538):240–4. 35. Schmid P, Adams S, Rugo HS, Schneeweiss A, Barrios CH, Iwata H, et al. Atezolizumab and nab-paclitaxel in advanced triple-negative breast cancer. N Engl J Med. 2018;379(22):2108–21. 36. Robson M, Goessl C, Domchek S. Olaparib for metastatic germline BRCA- mutated breast cancer. N Engl J Med. 2017;377(18):1792–3. 37. Litton JK, Rugo HS, Ettl J, Hurvitz SA, Goncalves A, Lee KH, et al. Talazoparib in patients with advanced breastcancer and a germline BRCA mutation. N Engl J Med. 2018;379(8):753–63. 38. Peng W, Chen JQ, Liu C, Malu S, Creasy C, Tetzlaff MT, et al. Loss of PTEN promotes resistance to T cell-mediated immunotherapy. Cancer Discov. 2016;6(2):202–16. 39. Juvekar A, Burga LN, Hu H, Lunsford EP, Ibrahim YH, Balmana J, et al. Combining a PI3K inhibitor with a PARP inhibitor provides an effective therapy for BRCA1-related breast cancer. Cancer Discov. 2012;2(11):1048–63. 40. Matulonis UA, Wulf GM, Barry WT, Birrer M, Westin SN, Farooq S, et al. Phase I dose escalation study of the PI3kinase pathway inhibitor BKM120 and the oral poly (ADP ribose) polymerase (PARP) inhibitor olaparib for the treatment of high-grade serous ovarian and breast cancer. Ann Oncol. 2017;28(3):512–8. 41. Barroso-Sousa R, Keenan TE, Pernas S, Exman P, Jain E, Garrido-Castro AC, et al. Tumor mutational burden and PTEN alterations as molecular correlates of response to PD-1/L1 blockade in metastatic triple-negative breast cancer. Clin Cancer Res. 2020;26(11):2565–72. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.