Clinical and biomarker results from phase I/II study of PI3K inhibitor alpelisib plus nab-paclitaxel in HER2-negative metastatic breast cancer

Abstract

Purpose:

PIK3CA mutations are common in breast cancer and promote tumor progression and treatment resistance. We conducted a phase I/II trial of alpelisib (α-specific PI3K inhibitor) plus nab-paclitaxel in patients with HER2-negative metastatic breast cancer (MBC).

Experimental Design:

Eligible patients had HER2-negative MBC with any number of prior chemotherapies. Phase I was 3+3 dose-escalation design with three dose levels of alpelisib (250mg-300mg-350mg) daily, plus nab-paclitaxel 100mg/m² D1-8-15 every 28 days. Phase II was according to Simon’s two-stage design. PIK3CA mutations in tumor/circulating-tumor DNA (ctDNA) were assessed. Primary endpoints were recommended phase 2 dose (RP2D) and objective response rate (ORR). Additional endpoints included safety, pharmacokinetics, PFS, association of PIK3CA mutation with outcomes.

Results:

43 patients enrolled (phase I=13, phase II=30). 84% had visceral disease, 84% had prior taxane. No dose-limiting toxicities occurred in phase I. RP2D was alpelisib 350mg daily +nab-paclitaxel 100mg/m² D1-8-15. Hyperglycemia (G3:26%,G4:0%), neutropenia (G3:23%,G4:7%), diarrhea (G3:5%,G4:0%), and rash (G3:7%,G4:0%) were most common adverse events. Among 42 evaluable patients, ORR was 59% (CR=7%, PR=52%), 21% of whom had response lasting >12 months; median PFS was 8.7 months. 40% demonstrated tumor and/or ctDNA PIK3CA mutation; patients with tumor/ctDNA mutation demonstrated better PFS compared to those without mutation (11.9 vs 7.5 months, HR=0.44, p=0.027). Patients with normal metabolic status had longer PFS compared to prediabetic/diabetic patients (12.0 vs 7.5 months, p=0.014). No pharmacokinetic interactions were detected.

Conclusions:

Alpelisib plus nab-paclitaxel combination was well-tolerated and shows encouraging efficacy, especially in patients with PIK3CA-mutated tumor/ctDNA. Impact of metabolic status on response to this combination merits further investigation.

Introduction

Mutations and deregulations in the phosphatidylinositol-3-kinase (PI3K) pathway are common in breast cancer (1). PI3K pathway activation frequently occurs as a result of mutation of the gene encoding the PI3K catalytic subunit p110α (PIK3CA). Activation of the PI3K pathway promotes tumor growth and progression, as well as resistance to anticancer therapies like taxanes (2, 3).

Activating mutations in PIK3CA are noted in approximately 40% of patients with hormone receptor-positive, human epidermal growth factor receptor 2 (HER2)-negative breast cancer and in 8–10% of patients with triple-negative breast cancer (TNBC) (1, 4).

Alpelisib (BYL-719) is an oral α-specific PI3K inhibitor (PI3Ki) that selectively inhibits PI3Kα isoform (both wild-type and mutated p110α) and is significantly less active against the other class I isoforms β, δ, and γ (5). Targeting the alpha isoform of PI3K is expected to reduce the potential for treatment-related toxicity and improve the therapeutic window compared to inhibitors with less isoform specificity. Alpelisib has shown single-agent activity in patients with PIK3CA-altered advanced solid tumors (6). Recently, the phase III SOLAR-1 trial demonstrated that in patients with PIK3CA-mutated, hormone receptor-positive, HER2-negative advanced breast cancer who had received previous endocrine therapy, the addition of alpelisib to fulvestrant significantly prolonged progression-free survival (PFS) compared to fulvestrant monotherapy (7). These findings led to approval of alpelisib to be used in combination with fulvestrant by both the United States Federal Drug Agency and European Medicines Agency.

The primary objective of this phase I/II trial was to determine the safety and efficacy of alpelisib in combination with nab-paclitaxel in patients with HER2-negative metastatic breast cancer.

Patients and Methods

Study design and Participants

This was a single-arm phase I-II trial (NCT2379247). Eligible patients were females aged ≥18 years with locally advanced or metastatic HER2-negative (per 2013 ASCO/CAP guidelines (8)) breast cancer who had received at least one line of chemotherapy in either the advanced or neo/adjuvant setting, had measurable disease (per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1), Eastern Cooperative Oncology Group (ECOG) performance status ≤2, and adequate organ and marrow function. Participants were excluded for uncontrolled diabetes mellitus (fasting plasma glucose level >140 mg/dL [7.8 mmol/L] or glycosylated hemoglobin [HbA1C] >8%) or prior treatment with PI3K inhibitor. Previous endocrine therapy was allowed (no limit on number of prior endocrine therapies), and previous taxanes (except nab-paclitaxel) were allowed if >6 months since exposure. Participants with treated brain metastases were eligible if free from central nervous system symptoms and at least 3 months had passed since brain metastasis treatment.

Participants were enrolled at the University of Kansas Medical Center and at the Vanderbilt-Ingram Cancer Center. The study was approved by each institution’s human subjects research review board and was conducted in accordance with an assurance filed with and approved by the U.S. Department of Health and Human Services and in accordance with the U.S. Common Rule and the International Ethical Guidelines for Biomedical Research Involving Human Subjects. All patients provided written informed consent.

Procedures and Treatment

Nab-paclitaxel (100mg/m²) was administered intravenously on days 1, 8, and 15 of each 28-day cycle. Phase I of the study was a 3+3 dose escalation trial with three dose levels of alpelisib (250mg, 300mg, and 350mg administered orally once daily). Decision to escalate alpelisib dose was determined by safety evaluation (as described below). The dose of alpelisib identified as the recommended phase II dose (RP2D) in phase I was then administered in phase II. All patients were instructed to take second- or third-generation H₁-anti-histaminic prophylaxis for rash.

Toxicity was assessed using Common Toxicity Criteria for Adverse Events (CTCAE) version 4.03. Response (investigator-assessed) was evaluated according to RECIST version 1.1 every 8 weeks. During phase I, cycle 1 blood samples were collected pre-dose and 1.5, 2, 2.5, 3, 3.5, 4, 6, and 8 hours post-alpelisib dose (oral alpelisib dosing occurred one hour prior to the start of 30-minute nab-paclitaxel infusion) for pharmacokinetics (Supplementary Material).

Archived formalin-fixed paraffin-embedded (FFPE) tumor tissue (primary or metastatic site) was collected from all subjects. DNA and RNA isolated from FFPE tissue were subjected to next-generation sequencing (NGS) for assessment of PIK3CA mutation (see Supplementary Material for details) and RNA gene expression profiling, respectively. RNA signature scores generated by clara^T analysis and raw gene expression data (Xcel array, Almac Diagnostic Services), as well as immune cell subpopulation gene expression data, were compared between PIK3CA-mutated and non-mutated groups (Supplementary Material). Pre-treatment blood samples were collected from all subjects. Circulating tumor DNA (ctDNA) isolated from plasma samples was subjected to NGS for PIK3CA mutation assessment (Supplementary Material). Results of tissue or ctDNA analyses were not provided back to the treating physicians. Pre-treatment fasting plasma glucose and HbA1C values were used to determine baseline metabolic status for each patient, per the 2019 Classification and Diagnosis Guidelines of the American Diabetes Association (9). Per these guidelines, normal metabolic status is defined as hemoglobin A1C (HbA1C) <5.7% and fasting plasma glucose (FPG) <100 mg/dL. Prediabetic status is defined as HbA1C 5.7% to <6.5% and/or FPG 100 to <126 mg/dL. Diabetic status is defined as HbA1C ≥6.5% and/or FPG ≥126 mg/dL.

Endpoints

Primary objectives were to determine the RP2D of alpelisib + nab-paclitaxel for phase I and the overall response rate (ORR) in subjects treated at the RP2D for phase II. Secondary objectives included safety, pharmacokinetic evaluation, clinical benefit rate (CBR), progression-free survival (PFS), and overall survival (OS). Exploratory objectives included association of tumor/ctDNA PIK3CA alterations with response. ORR includes complete response (CR) plus partial response (PR). CBR includes CR, PR, plus stable disease (SD) ≥16 weeks.

Phase I was a 3+3 dose escalation design (three dose levels of alpelisib: 250mg, 300mg, and 350mg orally once daily, continuous dosing) with dose-limiting toxicities (DLT) assessed during the first treatment cycle. If two or more of the six patients experienced a dose-limiting toxicity, dosing escalation would cease and maximum tolerated dose (MTD) would be reached. RP2D was the next lower dose at which <1/6 subjects experienced a DLT. Dose of alpelisib was not increased beyond 350mg even if MTD not reached. (See Supplementary Material for dose escalation details).

Phase II was designed according to Simon’s two-stage minimax design to detect an improvement in ORR from 20% to 40%, with alpha 0.05 and a power of 80%. In the first stage, 18 participants were treated at the RP2D, with at least 5 responses necessary in order to proceed to the next stage. In the second stage, an additional 15 subjects were enrolled. If 11 or more subjects had an objective response among all 33 eligible subjects, the regimen would be considered promising.

Statistical Analysis

Overall frequencies and percentages were summarized for baseline characteristics. ORR and CBR were estimated with 95% confidence intervals (CI). PFS was defined as the time in months from the date of enrollment to the date of progression or death, whichever was earlier. OS was defined as the time in months from the date of enrollment to death as a result of any cause. Survival curves were assessed by the Kaplan-Meier method and groups compared by log-rank test. Cox regression modelling was used for multivariable analysis. All analyses were conducted using SPSS Statistics version 26 (IBM Corporation, Armonk, NY). P-values <0.05 (2-sided) were considered significant, without correction for multiple comparisons.

Results

Between April 2015 and May 2017, 43 patients were enrolled (N=13 phase I, N=30 phase II). Baseline characteristics of the study population are summarized in Table 1. Median age was 55 years, and 30% had triple-negative disease (defined as estrogen receptor and progesterone receptor immunohistochemistry <10% and HER2-negative). 84% of patients had visceral disease. The median lines of prior chemotherapy in the metastatic setting was 1, and 30% of patients had received two or more lines of chemotherapy for metastatic disease.

Table 1.

Baseline patient demographics and clinical characteristics

Characteristic – N (%)	All patients (N=43)
Age, years – median (range)	55 (34–72)
Subtypea
ER and/or PgR-positive	30 (70%)
Triple-negative	13 (30%)
Measurable disease	43 (100%)
Visceral disease	36 (84%)
Prior lines of chemotherapy for metastatic disease
0	10 (23%)
1	20 (47%)
≥2	13 (30%)
Prior taxane
Neo/adjuvant	26 (61%)
Metastatic	7 (16%)
Neo/adjuvant and metastatic	3 (7%)
None	7 (16%)
Prior CDK4/6 inhibitor	12 (28%)

^aER and/or PgR positivity defined as ≥10% by immunohistochemistry.

Abbreviations: CDK4/6, cyclin-dependent kinase 4/6; ER, estrogen receptor; PgR, progesterone receptor

Phase I results:

No dose-limiting toxicity occurred in the three patients treated at the 1^st alpelisib dose level of 250mg, three patients treated at the 2^nd dose level of 300mg, and seven patients treated at the 3^rd dose level of 350mg (1/7 patients treated at the 3^rd level stopped treatment within 7 days due to progressive brain metastasis, thus not evaluable for DLT). MTD of alpelisib was not reached, and RP2D of alpelisib was defined as 350mg orally daily in combination with nab-paclitaxel 100mg/m² on D1,8,15 every 28 days.

Toxicity

Diarrhea, hyperglycemia, fatigue, hematologic toxicity and peripheral neuropathy were the most common toxicities (Table 2). Diarrhea was noted in 81% of patients but was mainly grade 1 and 2 (5% grade 3, 0% grade 4). Hyperglycemia was noted in 70% of patients (26% grade 3, 0% grade 4). 28% of patients required metformin for management of hyperglycemia. Rash was noted in 44% of patients and was mainly grade 1 and 2 (7% grade 3, 0% grade 4). All patients received H₁ anti-histaminic prophylaxis for rash. 12% (5/43) of patients discontinued treatment due to toxicity (2 for pneumonitis; 1 each for infection, thrombocytopenia, kidney dysfunction). No patients discontinued treatment due to hyperglycemia, rash, or diarrhea. Alpelisib dose reduction was required for 11/43 (26%) patients (N=1 at the 250mg dose and N=10 at the 350mg dose), and nab-paclitaxel dose reduction was required for 12/43 (28%) patients.

Table 2.

Treatment-related adverse events

Event – N (%)	All grades	Grade 1	Grade 2	Grade 3	Grade 4
Diarrhea	35 (81%)	22 (51%)	11 (26%)	2 (5%)	0 (0%)
Hyperglycemia	30 (70%)	8 (19%)	11 (26%)	11 (26%)	0 (0%)
Nausea	28 (65%)	21 (49%)	7 (16%)	0 (0%)	0 (0%)
Fatigue	27 (63%)	11 (26%)	14 (33%)	2 (5%)	0 (0%)
Peripheral neuropathy	25 (58%)	15 (35%)	9 (21%)	1 (2%)	0 (0%)
Anorexia	22 (51%)	16 (37%)	6 (14%)	0 (0%)	0 (0%)
Electrolyte imbalance	22 (51%)	10 (23%)	7 (16%)	5 (12%)	0 (0%)
Anemia	21 (49%)	12 (28%)	4 (9%)	5 (12%)	0 (0%)
Mucositis	21 (49%)	14 (33%)	6 (14%)	1 (2%)	0 (0%)
Neutropenia	20 (47%)	3 (7%)	4 (9%)	10 (23%)	3 (7%)
Dysgeusia	19 (44%)	17 (40%)	2 (5%)	0 (0%)	0 (0%)
Rash	19 (44%)	14 (33%)	2 (5%)	3 (7%)	0 (0%)
Infection	13 (30%)	1 (2%)	10 (23%)	2 (5%)	0 (0%)
Edema	12 (28%)	8 (19%)	4 (9%)	0 (0%)	0 (0%)
Vomiting	11 (26%)	6 (14%)	5 (12%)	0 (0%)	0 (0%)
Myalgia	10 (23%)	7 (16%)	3 (7%)	0 (0%)	0 (0%)
Nail changes	10 (23%)	7 (16%)	3 (7%)	0 (0%)	0 (0%)
Liver enzyme increase	9 (21%)	8 (19%)	1 (2%)	0 (0%)	0 (0%)
Alopecia	8 (19%)	3 (7%)	5 (12%)	0 (0%)	0 (0%)
Arthralgia	8 (19%)	6 (14%)	1 (2%)	1 (2%)	0 (0%)
Dry skin/mouth	8 (19%)	7 (16%)	1 (2%)	0 (0%)	0 (0%)
Dyspepsia	8 (19%)	5 (12%)	3 (7%)	0 (0%)	0 (0%)
Dyspnea	8 (19%)	6 (14%)	1 (2%)	1 (2%)	0 (0%)
Cough	7 (16%)	4 (9%)	3 (7%)	0 (0%)	0 (0%)
Weight loss	7 (16%)	2 (5%)	5 (12%)	0 (0%)	0 (0%)
Musculoskeletala	7 (16%)	4 (9%)	2 (5%)	1 (2%)	0 (0%)
Creatinine imbalance	5 (12%)	2 (5%)	2 (5%)	1 (2%)	0 (0%)
Flatulence	4 (9%)	4 (9%)	0 (0%)	0 (0%)	0 (0%)
Watering eyes	4 (9%)	4 (9%)	0 (0%)	0 (0%)	0 (0%)
Blurred vision	3 (7%)	3 (7%)	0 (0%)	0 (0%)	0 (0%)
Decreased lymphocyte count	3 (7%)	1 (2%)	2 (5%)	0 (0%)	0 (0%)
Dizziness	3 (7%)	3 (7%)	0 (0%)	0 (0%)	0 (0%)
Dry eyes	3 (7%)	2 (5%)	1 (2%)	0 (0%)	0 (0%)
Fever	3 (7%)	3 (7%)	0 (0%)	0 (0%)	0 (0%)
Gait disturbance	3 (7%)	3 (7%)	0 (0%)	0 (0%)	0 (0%)
Gastrointestinal - otherb	3 (7%)	2 (5%)	1 (2%)	0 (0%)	0 (0%)
Headache	3 (7%)	3 (7%)	0 (0%)	0 (0%)	0 (0%)
Pneumonitis	3 (7%)	1 (2%)	2 (5%)	0 (0%)	0 (0%)
Abdominal pain	2 (5%)	2 (5%)	0 (0%)	0 (0%)	0 (0%)
Lymphedema	2 (5%)	0 (0%)	2 (5%)	0 (0%)	0 (0%)
Prolonged corrected QT interval	2 (5%)	0 (0%)	2 (5%)	0 (0%)	0 (0%)
Pulmonary - otherc	2 (5%)	0 (0%)	2 (5%)	0 (0%)	0 (0%)
Thrombocytopenia	2 (5%)	1 (2%)	1 (2%)	0 (0%)	0 (0%)
Otherd	20 (47%)	17 (40%)	2 (5%)	1 (2%)	0 (0%)

^aGrade 3 generalized muscle weakness (n=1); grade 2 muscle weakness lower limb (n=2); grade 1 events (n=1 each): muscle weakness lower limb, left arm pain, generalized muscle weakness, and toe pain.

^bGrade 2 colitis (n=1); grade 1 stomach tightness (n=1); and grade 1 unspecified gastrointestinal disorder (n=1).

^cGrade 2 pleural effusion (n=1) and grade 2 stridor (n=1).

^dGrade 3 thromboembolic event (n=1); grade 2 epistaxis (n=1); grade 2 hot flashes (n=1); grade 1 fall (n=2); grade 1 allergic rhinitis (n=2); and grade 1 events (n=1 each): anxiety, bone pain, epistaxis, flu-like symptoms, hoarseness, hot flashes, hypertension, hypoalbuminemia, postnasal drip, sore throat, urinary incontinence, urinary urgency, and visual disturbance.

Efficacy

In the primary efficacy analysis among 33 patients treated at the RP2D, ORR was 52% (95% CI: 34%−70%) (CR n=2, PR n=15). Objective responses were noted at all dose levels of alpelisib and were maintained over time (Figure 1A–C). In an intention-to-treat analysis including all 42 evaluable patients treated in phase I and II portions of the trial (1/43 patients not evaluable for response), ORR was 59% (95% CI: 44%−75%), with an additional 21% demonstrating SD for ≥16 weeks, for CBR of 80% (95% CI: 69%−93%). Among patients with triple-negative and hormone receptor-positive (HR+) disease, ORR was 58% (7/12, 95% CI: 26%−91%) and 60% (18/30, 95% CI: 41%−79%), respectively. Table 3 provides additional ORR details by line of treatment, taxane exposure, and breast cancer subtype. 29% (12/42) of patients received prior CDK4/6 inhibitor therapy; of these, 58% (7/12) had an objective response (CR n=2, PR n=5).

Figure 1: Antitumor activity of alpelisib plus nab-paclitaxel based on RECIST v1.1.

(A) Best percentage change from baseline in the sum of longest diameters of target lesions. (B) Longitudinal change from baseline in the sum of longest diameters of target lesions. (C) Durability of response. Panels A and B include patients who received at least one cycle of alpelisib and had available at least one post-baseline tumor assessment (N=39). Panel C includes all patients evaluable for response (N=42). Abbreviations: TNBC, triple-negative breast cancer.

Table 3.

Objective response by prior treatment and subtype

	All evaluable (N=42)	Prior lines of chemotherapy in metastatic setting			Prior taxane exposure		Subtype
Response – N (%)		0 (N=9)	1 (N=20)	≥2 (N=13)	Yes (N=36)	No (N=6)	ER/PgR+b (N=30)	TNBC (N=12)
ORR	25 (59%)	6 (67%)	13 (65%)	6 (46%)	22 (61%)	3 (50%)	18 (60%)	7 (58%)
CR	3 (7%)	0 (0%)	2 (10%)	1 (8%)	3 (8%)	0 (0%)	1 (3%)	2 (17%)
PR	22 (52%)	6 (67%)	11 (55%)	5 (38%)	19 (53%)	3 (50%)	17 (57%)	5 (42%)
SD ≥16 weeks	9 (21%)	1 (11%)	4 (20%)	4 (31%)	7 (19%)	2 (33%)	9 (30%)	0 (0%)
CBRa	34 (80%)	7 (78%)	17 (85%)	10 (77%)	29 (81%)	5 (83%)	27 (90%)	7 (58%)

^aCR + PR + SD ≥16 weeks

^bER and/or PgR positivity defined as ≥10% by immunohistochemistry.

Abbreviations: CBR; clinical benefit rate; CR, complete response; ER, estrogen receptor; ORR, overall response rate; PgR, progesterone receptor; PR, partial response; SD, stable disease; TNBC, triple-negative breast cancer.

At the time of data cutoff (October 16, 2019), one patient remained on study, and the median duration of follow-up was 17.3 months (range 1.8–40.0 months). Median duration of response was 5.0 months (range 2.0–40.0 months), and 21% (7/34) of patients had response lasting >12 months. Median PFS and OS were 8.7 months (95% CI: 5.5–11.9 months) and 18.5 months (95% CI: 9.6–27.4 months), respectively (Figure 2A, Supplementary Figure 1A). Seven patients discontinued nab-paclitaxel and continued alpelisib monotherapy (after 24, 19, 8, 6, 6, 5, and 4 cycles of the combination therapy, respectively). Progressive disease was noted for patients on monotherapy after a median of 15 weeks (range 8–43 weeks).

Figure 2: Kaplan-Meier estimates of progression-free survival.

(A) All evaluable patients. (B) All evaluable patients, by PIK3CA mutation status. Abbreviations: CI, confidence interval; HR, hazard ratio; PFS, progression-free survival.

Efficacy by PIK3CA status

Tumor tissue PIK3CA mutation status was available for 34/42 (81%) patients (inadequate tissue=6, assay failure=2), and ctDNA PIK3CA status was available for all 42 patients. Overall, 40% (17/42) of patients had PIK3CA mutation detected in tumor tissue and/or ctDNA (N=9 detected in both tissue and ctDNA, N=4 detected in tissue only, N=1 detected in ctDNA only, N=3 detected in ctDNA but tissue not available). Clinical benefit rate was higher in the PIK3CA-mutated subgroup compared to those without tumor/ctDNA PIK3CA mutation (CBR=100% vs 68%, OR=1.47, p=0.013).

Median PFS was 11.9 months (95% CI: 6.2–17.6) in the PIK3CA-mutated subgroup compared to 7.5 months (95% CI: 2.3–12.6) in those without tumor/ctDNA PIK3CA mutation (HR=0.44, 95% CI: 0.21–0.93, p=0.027, Figure 2B). Median OS was numerically higher in the PIK3CA-mutated subgroup compared to those without tumor/ctDNA mutation (26.7 vs 14.9 months, HR=0.59, 95% CI: 0.27–1.29, p=0.19) (Supplementary Figure 1B).

Metabolic status and efficacy

Among 42 patients, 38% had normal metabolic status, 52% were prediabetic, and 10% were diabetic. Patients with normal metabolic status had longer PFS compared to those who were prediabetic or diabetic; median PFS of 12.0 months (95% CI: 7.9–16.1) for those with normal metabolic status and 7.5 months (95% CI: 4.8–10.2) for prediabetic/diabetic patients (HR 0.36, 95% CI:0.16–0.84, p=0.014, Supplementary Figure 2A). The impact of metabolic status on PFS was more pronounced in the PIK3CA-mutated subgroup (N=17), where median PFS was 25.8 months (95% CI: 7.7–43.9) for patients with normal metabolic status (N=6) vs 5.8 months (95% CI: 5.5–6.1) for prediabetic/diabetic patients (N=11) (HR=0.10, 95% CI:0.01–0.78, p=0.008, Supplementary Figure 2B). In the patient subgroup without PIK3CA mutation (N=25), median PFS for patients with normal (N=10) vs prediabetic/diabetic (N=15) metabolic status was similar (6.4 months vs 7.5 months, p=0.29) (Supplementary Figure 2C).

On multivariable analysis (variables included: PIK3CA status, breast cancer subtype, baseline metabolic status, prior lines of chemotherapy 0 vs ≥1), presence of PIK3CA mutation and normal metabolic status were significantly associated with longer PFS (HR=0.43, 95% CI: 0.20–0.91, p=0.027 and HR=0.36, 95% CI: 0.15–0.82, p=0.016, respectively).

Gene expression signatures

Tumor samples adequate for RNA extraction were available for 30 patients, and RNA sequencing was successful for 19 patients (N=7 with PIK3CA mutation and N=12 without mutation). Analysis of published gene expression signatures revealed that angiogenesis (10) was upregulated, whereas MYC targets, tumor necrosis (11), and TP53 deficiency (12) pathways were downregulated in the PIK3CA-mutated group compared to the non-mutated group (Supplementary Figure 3A). On CIBERSORTx analysis, PIK3CA-mutated tumors appeared to exhibit lower imputed abundance of monocytes and naïve B-cells (Supplementary Figure 3B).

Pharmacokinetics

The study included limited pharmacokinetic analysis for 13 patients (phase I). Plasma samples were analyzed for alpelisib concentration for cycle 1 pre-dose and 1.5, 2, 2.5, 3, 3.5, 4, 6, and 8 hours post-dose. For alpelisib, obtained values for Tmax, Cmax, and AUC (0–8 hrs) and the dose proportionality observed for the latter two parameters were consistent with previous published reports (Supplementary Table 1A) (6). Paclitaxel was measured in both plasma and ultra-filtered plasma (to quantify total and free paclitaxel, respectively) at end of infusion (EOI), and at 0.5, 1, 1.5, 2, 4, and 6 hours post-EOI. Plasma concentration profiles, Cmax, and AUC values were consistent with previous reports of paclitaxel monotherapy at all alpelisib doses (Supplementary Tables 1B–C) (13). We thus conclude that no significant pharmacokinetic interactions were detected for the co-administration of alpelisib and nab-paclitaxel.

Discussion

In this phase I/II study, the combination of alpelisib and nab-paclitaxel demonstrated encouraging efficacy with a manageable safety profile in patients with HER2-negative metastatic breast cancer. Over two-thirds of the patients in our study had received prior chemotherapy for metastatic disease, and a third had received more than two lines of chemotherapy. We noted an ORR of 59% and CBR of 80% in this pre-treated patient population, with efficacy maintained in patients receiving treatment in the second line or beyond, where ORR was 61%. The majority (84%) of patients in this study had prior taxane exposure (for early or metastatic disease), and efficacy was noted in patients with prior taxane exposure. Although numbers are small to draw definitive conclusions, efficacy appeared similar in HR+ breast cancer (ORR 58%) and TNBC (ORR 60%). Importantly, 21% of patients had response lasting longer than one year, suggesting durability of response in some patients. In the 29% of patients who had received prior CDK4/6 inhibitor therapy, response rate was 67%.

Although further development of older generation PI3K inhibitors was hampered by the toxicity profile of these agents, there were hints of efficacy in PIK3CA-mutated disease in previous studies (14, 15). SOLAR-1 results have now validated PIK3CA mutation as a required selection criterion for efficacy of alpelisib, which is an α-specific PI3K inhibitor. SOLAR-1 showed statistically significant and clinically meaningful prolongation of PFS with alpelisib plus fulvestrant compared to placebo plus fulvestrant in patients with PIK3CA-mutated HR+, HER2-negative advanced breast cancer (PFS 11 vs 5.7 months, HR 0.65, p<0.001) (7). In our study 40% of patients had PIK3CA-mutated disease. The PIK3CA-mutated subgroup of our study demonstrated higher CBR (100% vs 68%) and longer PFS (11.9 vs 7.5 months) than those without mutation. The longest duration of response (40 months, ongoing) was observed in a patient with TNBC with PIK3CA mutation. Clinical studies do not show differential efficacy of taxanes based on PIK3CA mutations in HER2-negative metastatic breast cancer (16). On the contrary, efficacy of alpelisib, at least in combination with endocrine therapy, is related to PIK3CA mutations. Thus, the observed higher activity of the alpelisib plus nab-paclitaxel combination in the PIK3CA-mutated subgroup of our study is likely being driven by alpelisib. It should, however, be acknowledged that given the single-arm nature of our trial, the relative contribution of each drug in the entire study population or in the PIK3CA-mutated subgroup cannot be determined.

The efficacy seen in our study compares very favorably with what has been reported for single-agent nab-paclitaxel in phase II/III trials, where ORR ranges from 37–64% (first-line setting) to 14–21% (second-line or greater) (17–19). Previous studies of nab-paclitaxel show median PFS ranging from 3.0–6.1 month (17–19). In this context, our observed PFS of almost 12 months in the PIK3CA-mutated group appears very promising. This combination of alpelisib and nab-paclitaxel should be investigated further in randomized studies.

In the phase I portion of our trial, the RP2D of alpelisib was determined without observing any dose-limiting toxicities at the three dose levels, and responses were noted at all three dose levels of alpelisib. The most common toxicities were gastrointestinal, hyperglycemia, and hematologic. Overall incidence of diarrhea was 81%, majority of which was grade 1 or 2, with only 5% reporting grade 3 diarrhea. Rash was noted in 63% of patients and was also mainly grade 1 and 2 in severity (7% grade 3, no grade 4). All patients in our study received second/third-generation H1 anti-histaminic prophylaxis for rash, which probably contributed to low rates of grade 3–4 rash. It is to be noted that nab-paclitaxel is also associated with some risk of diarrhea and rash, which likely contributed to higher overall rates of diarrhea and rash we observed compared to SOLAR-1 (20). The incidences of grade 3 diarrhea and rash in our study are similar to SOLAR-1 (20). Hyperglycemia is an on-target effect of alpelisib, and grade 3 hyperglycemia was noted in 26% of the patients in our study (no grade 4 hyperglycemia events). One-quarter of patients required metformin for management of hyperglycemia. No patient discontinued treatment due to hyperglycemia, rash, or diarrhea. Twenty-six percent of patients required alpelisib dose reductions for toxicity management. Hematologic toxicity, peripheral neuropathy, and musculoskeletal adverse events were expected toxicities related to nab-paclitaxel, although the rate of grade 3 or higher peripheral neuropathy (2%) was lower than what has previously been reported with this agent. No new toxicity signals associated with either alpelisib or nab-paclitaxel were observed.

It has been shown that hyperglycemia induced by PI3K inhibition leads to increase in insulin levels and this glucose-insulin feedback can reactivate PI3K signaling in mouse models even in the presence of PI3K inhibitor (21). Dietary/pharmacological measures (e.g. ketogenic diet, sodium-glucose cotransporter inhibitors) can diminish this insulin feedback, thus increasing treatment efficacy of PI3K inhibitor (21). Patients with normal metabolic status in our study experienced longer PFS compared to pre/diabetic patients. This effect was remarkably pronounced in the PIK3CA-mutated subgroup. It is plausible that baseline insulin resistance in pre/diabetic patients may have played a role in decreasing the efficacy of the alpelisib. A somewhat similar observation was made in the international SANDPIPER trial, where differences in taselisib efficacy were observed according to geographic distribution (lesser efficacy in patients enrolled in Latin America/Eastern Europe compared to other parts of the world) (15). Possible differences in degree of insulin resistance, diet, and perhaps hyperglycemia management in patients from different geographic regions may have contributed to these findings. These observations highlight the potential role of insulin resistance and glucose-insulin feedback on optimizing clinical efficacy of PI3K inhibitors. The impact of metabolic status/insulin resistance and diet on effectiveness of PI3K inhibitors should be evaluated further in ongoing studies.

Upon gene expression analysis, PIK3CA mutation was associated with enhanced angiogenesis and MYC downregulation. These findings are not surprising, since hyperactivity of PI3K is known to increase tumor angiogenesis and TCGA pan-cancer analysis has shown MYC and PIK3CA mutations to be mutually exclusive (22–24). PIK3CA-mutated cancers may be less likely to have oncogenic dysregulation of the MYC axis; however, evasion of PI3K-targeted therapy can occur via MYC/eIF4E amplification, thus this pathway remains potentially important for acquired resistance to PI3K-directed therapy (24). Using CIBERSORTx analysis we found lower imputed frequency of monocytes and naïve B-lymphocytes in PIK3CA-mutated cancers. A small sample size and lack of direct immune cell quantification preclude any definitive conclusions, but several previous pre-clinical reports have suggested that PI3K hyperactivity blunts tumors immunogenicity (25, 26). If confirmed in other clinical studies, these findings may inform future trial design of PI3Ki combinations (immunotherapy, MYC-targeting agents, VEGF inhibitors, etc.).

In addition to PI3K inhibition, AKT inhibition is also currently being explored as a way to target the PI3K pathway in breast cancer. Two phase II trials have demonstrated that addition of oral AKT inhibitors to first-line paclitaxel therapy improves PFS in patients with PIK3CA/AKT1/PTEN-altered metastatic TNBC (4, 27). A phase II trial in patients with hormone receptor-positive metastatic breast cancer has also demonstrated improvement in PFS with addition of an AKT inhibitor (capivasertib) to fulvestrant (28). The recently reported IPATunity130 (cohort A) trial, however, failed to show improvement in PFS with addition of oral AKT inhibitor ipatasertib to first-line paclitaxel in PIK3CA/AKT1/PTEN-altered metastatic TNBC (29). The clinical efficacy of AKT inhibitors in breast cancer, especially the ability of these agents to target altered PIK3CA and PTEN, which are upstream of AKT in the PI3K/AKT pathway, is being investigated in ongoing studies. Overall survival results from IPATunity130 are awaited, and another ongoing phase III trial (NCT03997123) is assessing capivasertib in combination with first-line paclitaxel chemotherapy in patients with metastatic TNBC.

Although the results are intriguing, our study has limitations. As a single-arm trial, the relative contribution of alpelisib versus nab-paclitaxel towards efficacy cannot be determined. The small number of patients in subgroup analyses similarly limits our ability to draw conclusive links between PIK3CA mutation or metabolic status and benefit from alpelisib. Our study does, however, raise important hypotheses to be explored in future work. This combination warrants further investigation in randomized trials. Indeed, an ongoing phase III clinical trial (EPIK-B3, NCT04251533) is assessing alpelisib plus nab-paclitaxel as first or second-line treatment for patients with advanced TNBC and PIK3CA mutation or PTEN loss. A phase II trial (NCT04216472) evaluating alpelisib plus nab-paclitaxel as neoadjuvant treatment for anthracycline-refractory early-stage TNBC with PIK3CA or PTEN alterations is also currently enrolling patients. Results of these ongoing trials will address the role of alpelisib in TNBC. Other ongoing studies are also evaluating alpelisib in combination with HER2-targeted therapy, anti-androgen therapy, and anti-estrogen therapy in biomarker-selected patient populations (NCT03207529, NCT04208178, NCT03056755).

Supplementary Material

2

Supplementary Figure 1: Kaplan-Meier estimates of overall survival. (A) All evaluable patients. (B) All evaluable patients, by PIK3CA mutation status. Abbreviations: CI, confidence interval; HR, hazard ratio; OS, overall survival.

3

Supplementary Figure 2: Kaplan-Meier estimates of progression-free survival by baseline metabolic status. (A) All evaluable patients. (B) Patients with PIK3CA mutation. (C) Patients without PIK3CA mutation.

4

Supplementary Figure 3: Gene expression signatures by PIK3CA mutation status. (A) Published gene expression signatures.¹ Hu et al., Oncogene (2005); 24: 1212–1219. ²Hallmark signature within clara^T platform. ³Bredholt et al., Oncotarget (2015); 6: 39676–39690. ⁴Knijnenburg et al., Cell Reports (2018); 23: 239–254. (B) CIBERSORTx gene expression analysis of immune cell subpopulations.

Statement of translational relevance.

PIK3CA mutations are common in breast cancer and promote tumor progression and treatment resistance. This phase I/II trial evaluated the α-specific PI3K inhibitor alpelisib in combination with nab-paclitaxel in patients with HER2-negative metastatic breast cancer. The alpelisib plus nab-paclitaxel combination demonstrated a manageable safety profile with encouraging efficacy in both hormone receptor-positive and triple-negative breast cancer. Twenty-one percent of patients had response lasting >12 months, suggesting durability of response. Efficacy of the combination was higher in patients with PIK3CA mutation and those with normal (vs prediabetic/diabetic) metabolic status. Alpelisib in combination with nab-paclitaxel is under investigation in an ongoing biomarker-selected phase III clinical trial. The impact of metabolic status on response to alpelisib-based therapy merits further investigation.