Introduction

There is an unmet need to improve treatment options for patients with relapsed human epidermal growth factor receptor 2 (HER2)-negative metastatic breast cancer (BC) and non-small cell lung cancer (NSCLC) [1, 2]. Current standard treatment options for HER2-negative locally recurrent/metastatic BC include the single-agent taxanes paclitaxel and docetaxel [1]. When our study was initiated, docetaxel was a standard option for NSCLC following progression with first-line chemotherapy [3, 4].

The phosphatidylinositol 3-kinase (PI3K) signaling pathway is commonly altered in cancer [5, 6] and may be activated by gain-of-function mutations and/or amplification of the phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) gene [7,8,9]. PIK3CA encodes the α-isoform of the catalytic subunit of PI3K (PI3Kα) [8] and is mutated in ~40% of hormone receptor (HR)-positive BCs [10,11,12]. Increased PI3K signaling occurs frequently in lung cancer [13], including mutations in PIK3CA (2–3%) [14,15,16] and loss/low expression of the PI3K pathway suppressor, phosphatase, and tensin homolog (PTEN) (39–48%) [17].

Taselisib (GDC-0032), a potent and selective inhibitor of class I PI3Kα-, δ-, and γ-isoforms, has greater activity against tumor cells harboring PIK3CA mutations versus wild-type PIK3CA [18,19,20,21]. In PIK3CA-mutant BC models, taselisib plus docetaxel or paclitaxel enhanced activity versus taxanes alone [22]. A phase I dose-escalation study of single-agent taselisib suggested activity in PIK3CA-mutant BC and NSCLC [23] and a tolerable safety profile, with expected PI3K inhibitor class adverse events (AEs) [23,24,25,26]. Based on these results, the recommended single-agent dose of taselisib was 9 mg (capsule) [23]. The tablet formulation of taselisib has higher bioavailability than the capsule and 6 mg provides equivalent exposure to a 9 mg capsule [27].

Taselisib has clinical activity in patients with PIK3CA-mutant, estrogen receptor (ER)-positive, HER2-negative BC [28, 29]. In the neoadjuvant LORELEI study, addition of taselisib to letrozole significantly improved the overall response rate in ER-positive, HER2-negative early BC (vs. placebo plus letrozole; intention-to-treat and PIK3CA-mutant populations) [28]. In SANDPIPER, addition of taselisib to fulvestrant significantly increased progression-free survival (PFS) in patients with ER-positive, HER2-negative advanced BC, and PIK3CA mutations (vs. placebo plus fulvestrant) [29]. Tolerability was considered challenging, with frequent gastrointestinal toxicities and hyperglycemia, and a higher proportion of discontinuation due to AEs in the taselisib arm (vs. the placebo arm) [29].

This open-label, phase Ib study evaluated the safety and pharmacology of taselisib plus docetaxel or paclitaxel in locally advanced/metastatic BC or NSCLC, including PIK3CA-mutated cancers.

Methods

Patients

Eligible patients were ≥ 18 years and had histologically/cytologically documented breast adenocarcinoma with locally recurrent/metastatic disease (paclitaxel- or docetaxel-containing arms) or histologically documented advanced (stage IV)/recurrent NSCLC (docetaxel-containing arms). Patients with HR-positive BC had disease progression after ≥ 1 prior endocrine therapy in the adjuvant or metastatic setting. Patients with NSCLC had ≥ 1 prior anticancer regimen in an advanced setting and had docetaxel considered as appropriate per local guidelines. Patients had evaluable/measurable disease [Response Evaluation Criteria in Solid Tumours (RECIST) v.1.1], life expectancy ≥ 12 weeks, Eastern Cooperative Oncology Group performance status (ECOG PS) 0/1 at screening, and adequate hematologic and end-organ function (Supplementary methods).

This study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. Approval for the protocol and any material provided to the patient was obtained from independent ethics committees at participating institutions (Supplementary methods). All patients provided written informed consent.

Study design and treatment

This was an open-label, phase Ib, dose-escalation study (NCT01862081). Each arm had two stages: dose-escalation (stage 1) and dose-expansion (stage 2). Patients received taselisib with docetaxel (Arms A, C, D, and E) or taselisib with paclitaxel (Arms B and F), with taselisib at different dosing schedules (Fig. 1 and Supplementary methods). Patients in Arms A, C, D, and E received intravenous docetaxel (75 mg/m2) on Day 1 of each 21-day cycle and premedication with oral corticosteroids (per institutional guidelines and docetaxel treatment guidelines). Patients in Arms B and F received intravenous paclitaxel (80 mg/m2) on Days 1, 8, 15, and 22 of each 28-day cycle and could receive premedication with dexamethasone, diphenhydramine, and either ranitidine or famotidine (per institutional practice).

Fig. 1
figure 1

Participant flow diagrams. Study schema for patients treated with a taselisib plus docetaxel and b taselisib plus paclitaxel. AE adverse event, DLT dose-limiting toxicities, G-CSF granulocyte-colony stimulating factor, HER2 human epidermal growth factor receptor 2, MBC metastatic breast cancer, MTD maximum tolerated dose, NSCLC non-small cell lung carcinoma, qd once-daily, qw once-weekly, q3w every 3 weeks. Arms B and F enrolled 20 and 21 patients, respectively, in the expansion cohorts; however, the taselisib dosing information was not verifiable for two patients, and therefore these patients were not included in the safety-evaluable population

Full size image

Once the maximum tolerated dose (MTD), or maximum assessed dose in the absence of an MTD, of taselisib had been established in an arm from dose escalation, additional patients with each combination could be enrolled in stage 2 to confirm a potential recommended dose for future studies (Supplementary methods).

Outcomes

The primary endpoints were safety and tolerability, dose-limiting toxicities (DLTs), MTD, and identification of a recommended phase II dosing regimen. Secondary endpoints were pharmacokinetics (PK: taselisib and docetaxel; taselisib and paclitaxel; 6α-OH-paclitaxel), and preliminary assessment of the antitumor activity of taselisib plus docetaxel or paclitaxel. Exploratory endpoints are listed in the Supplementary methods.

Safety assessment

Safety was evaluated by monitoring all AEs, laboratory abnormalities, vital signs, and treatment exposure (Supplementary methods). AEs were defined and graded per National Cancer Institute-Common Terminology Criteria for Adverse Events (NCI-CTCAE) version 4.0. All patients with a verifiable record of taselisib dosing were included. Investigators assessed whether DLTs (Supplementary methods) that occurred within the DLT assessment window (Arms A, C, D, and E: Cycle 1, Days 1–21; Arms B and F: Cycle 1, Days 1–28) were possibly related to study treatment. The MTD was exceeded and a lower dose level evaluated if a DLT was observed in ≥ 33% of patients in a cohort (Supplementary methods). The Supplementary methods list AEs of special interest (AESIs).

Tumor assessments performed per RECIST v1.1 were recorded during the last week of the cycle and before the start of treatment in the next cycle (Supplementary methods). Objective responses were confirmed by repeat assessment ≥ 4 weeks after initial documentation. Clinical benefit rate (CBR) was defined as confirmed complete response, confirmed partial response, or stable disease lasting ≥ 6 months in all patients.

PK analysis

Blood samples were collected for PK characterization of taselisib, docetaxel, and paclitaxel. Individual plasma concentration versus time data and summary statistics were tabulated by treatment arm, stage, cycle, and dose level. The plasma concentration–time data were analyzed using non-compartmental methods employing Phoenix WinNonlin software (Certara USA, Inc., Princeton, NJ, USA) (Supplementary methods).

Biomarker assessments

PIK3CA mutation status was determined using local or central testing, with retrospective central confirmation using the Roche cobas® PIK3CA Mutation Test (Roche Diagnostics, Indianapolis, IN, USA) (Supplementary methods) [30]. Tumors were classified as PIK3CA-mutated if a positive result was obtained, PIK3CA-mutation not detected (MND) if no mutations were detected, or PIK3CA mutation status unknown if there was an assay failure/no tissue available for central confirmation.

Analysis of PIK3CA mutations in plasma circulating tumor DNA (ctDNA) was performed using the BEAMing Digital PCR platform (Sysmex Inostics GmbH, Hamburg, Germany) (Supplemental methods).

Tumors were classified as PIK3CA-mutated if a mutation was detected by central tissue or plasma ctDNA testing.

Statistical methods

Safety analysis was based on patients who received ≥ 1 taselisib dose (safety-evaluable population) and patient data were collected until study discontinuation. Efficacy analysis was based on the intention-to-treat population of all patients enrolled. Data were summarized (medians and standard deviations) and Kaplan–Meier methodology used for time-to-event analysis. Analyses were performed by PIK3CA status using tissue samples (docetaxel arms) or both ctDNA and tissue samples (paclitaxel arms); both sample types were analyzed in the paclitaxel arms to better understand the impact of PIK3CA mutation status. Data cutoff was the last patient last visit (June 9, 2017).

Results

Patient characteristics

Eighty patients were enrolled from 11 centers [USA (n = 7), Spain [2], Belgium [1], and Canada [1]] into either docetaxel-containing (n = 21) or paclitaxel-containing arms (n = 59) (Fig. 1). The taselisib safety-evaluable population comprised all patients who received docetaxel and 57/59 who received paclitaxel; taselisib dosing information was not verifiable for one patient each from the expansion cohorts of Arms B and F. At the time of data cutoff, all patients had discontinued taselisib and were no longer on -study (Fig. 1).

In the docetaxel-containing arms, 15 patients had BC and six had NSCLC (Table 1). In the paclitaxel-containing arms, 56 safety-evaluable patients had BC, and of these, 45 had HR-positive BC, while 11 had triple-negative BC. One patient with NSCLC was inadvertently enrolled into a paclitaxel-containing arm (expansion cohort F) and allowed to continue on study treatment.

Table 1 Baseline demographics and clinical characteristics in patients (safety evaluable)
Full size table

In the docetaxel-containing arms, study discontinuation was most commonly due to progressive disease (90.5%) and death (9.5%); disease progression (95.2%) was the most common reason for taselisib discontinuation. In the paclitaxel-containing arms, study discontinuation was most commonly due to disease progression (75.4%), AEs (12.3%), and withdrawal of consent (7.0%); disease progression (68.4%) and AEs (22.8%) were the most common reasons for taselisib discontinuation.

Safety

In the docetaxel-containing arms, 2/4 patients who received the 3 mg taselisib capsule plus docetaxel (Arm A) had DLTs (one experienced Grade 4 neutropenia; one had both Grade 3 stomatitis and Grade 4 decreased neutrophil count). The MTD was exceeded in Arm A with 3 mg taselisib capsules (equivalent to 2 mg tablets) once-daily (qd) for 21 consecutive days and Arm A was closed to enrollment. In Arm C, there were no DLTs, and the taselisib tablet was escalated from 2 mg to 4 mg; however, Grade 4 neutropenia was reported in both the 2 mg (3 cases) and 4 mg (1 case) cohorts. In Arm D, while no DLTs were reported, there were two cases of Grade 4 neutropenia in the 2 mg tablet cohort. Arm D was therefore closed, and Arm E opened with mandatory granulocyte-colony stimulating factor (G-CSF) prophylaxis, and taselisib escalated from the 2 mg to the 4 mg tablet; there were no DLTs and no cases of Grade 4 neutropenia reported.

In the docetaxel-containing arms, all patients experienced ≥ 1 AE (Table 2). The most common treatment-emergent AEs (≥ 40%) were fatigue (52.4%), diarrhea (47.6%), and alopecia, nausea, and neutropenia (42.9%, each) (Table S1). All patients had ≥ 1 taselisib-related AE, most commonly diarrhea and fatigue (47.6% each). Grade ≥ 3 AEs were common (90.5%), with neutropenia and decreased neutrophil count the most frequent (33.3% each) (Table S2). Nine patients (42.9%) experienced ≥ 1 serious AE (SAE), with the majority (23.8%) being Grade 3. SAEs reported in ≥ 2 patients were BC progression and pneumonia (Table S3). Two patients (9.5%) experienced taselisib-related SAEs; Grade 2 gastritis in one, and Grade 4 respiratory failure and Grade 5 sepsis in another. Three patients (14.3%) died (Table 2), one due to sepsis (considered to be related to taselisib) and the other two due to disease progression. One patient (4.8%) had Grade 4 respiratory failure that led to withdrawal of taselisib, while 28.6% had AEs that led to withdrawal of docetaxel (Table 2). AESIs reported were diarrhea (47.6%), hyperglycemia (28.6%), and rash-AE group term (AEGT) (28.6%); all were Grade 1–2.

Table 2 Safety overview (safety-evaluable; treatment-emergent)
Full size table

In the taselisib plus paclitaxel arms, four DLTs were reported: one patient in Arm B, Cohort 1 had Grade 3 maculopapular rash, one in Arm B, Cohort 2 had Grade 3 mucosal inflammation. In Arm B, the MTD was not reached and an expansion cohort was opened (4 mg tablet qd dose level). Two DLTs were reported in Arm F (Grade 3 rash, Grade 4 decreased neutrophil count). The MTD was exceeded at the 6 mg tablet 5 days on/2 days off schedule, and an Arm F expansion cohort was opened at the 4 mg tablet 5 days on/2 days off schedule.

In the taselisib plus paclitaxel arms, all patients experienced ≥ 1 AE (Table 2). The most common treatment-emergent AEs were diarrhea (61.4%), nausea (47.4%), fatigue (43.9%), and alopecia (42.1%) (Table S4). Fifty-five patients (96.5%) had ≥ 1 taselisib-related AE, with diarrhea (54.4%) and fatigue (38.6%) the most frequent. Grade ≥ 3 AEs were frequent (78.9%), with hyperglycemia and decreased neutrophil count (14.0% each) the most common (Table S5). Twenty-three patients (40.4%) had ≥ 1 SAE, which were mostly Grade 3 (29.8%). The most common SAEs were pneumonitis (7.0%) and BC progression (3.5%) (Table S6); two patients (3.5%) died in the paclitaxel-containing arms, both due to BC progression. Fourteen patients (24.6%) had AEs that led to withdrawal of taselisib, most commonly due to pneumonitis (7.0%), followed by diarrhea (3.5%). The rate of AEs leading to taselisib discontinuation was 36.8% in the Arm B and 15% in the Arm F expansion cohorts. Sixteen patients (28.1%) had AEs that led to withdrawal of paclitaxel, most commonly neurotoxicity, peripheral sensory neuropathy, and pneumonitis (5.3% each). The AESIs reported were diarrhea (61.4%), rash-AEGT (52.6%), hyperglycemia (28.1%), pneumonitis (14.0%), and colitis-AEGT (5.3%). In the Arm B expansion cohort, 6/19 (31.6%) patients had pneumonitis/interstitial lung disease, including 4/19 (21.1%) at Grade 3. In the Arm F expansion cohort, the pneumonitis rate was 5% (Grade 3 event).

Exposure

In the docetaxel-containing arms, median duration of taselisib exposure was 3.9 months (range 0.5–15.2) (Table S7a), median number of doses was 56 (range 8–460), and mean total cumulative dose for all patients was 208.3 mg (range 32–1004). Median duration of docetaxel exposure was 3.5 months (range 0.7–12.5). The mean total number of doses for all patients was 6.3 (range 2–19; median 6.0).

In the paclitaxel-containing arms, median duration of taselisib exposure was 3.2 months (range 0.2–39.1) (Table S7b). In all patients, the total mean number of doses was 128.7 (range 5–789, median 68) and the mean total cumulative dose was 415.1 mg (range 20–2552). The median duration of paclitaxel exposure was 3.0 months (range 0.0–27.4). The mean total number of doses for all patients was 13.9 (range 1–70; median 12).

PK analysis

Based upon the relatively sparse sampling schedule for all study drugs, taselisib displayed moderate absorption and demonstrated approximately proportional and linear increases in maximum observed plasma concentration (Cmax), area under the curve from time 0 to the last measurable concentration (AUC0–last), and area under the curve during 24 h (AUC0–24) after a single dose and at steady state (following 15 days of daily dosing), with increasing dose levels from 2 mg to 6 mg capsule or tablet equivalent. On average, steady-state qd exposures (AUC0–24) were consistent with predicted efficacious exposures at dose levels of 2–6 mg. Data from the expansion cohorts for Arms B and F (4 mg tablet dose) indicated that the Cycle 1, Day 15 pre-dose (Cmax) levels of taselisib were 33.1 ng/mL (68.8 ng/mL) and 15.4 ng/mL (31.5 ng/mL), respectively. Docetaxel [31] and paclitaxel [32] demonstrated PK characteristics similar to previously reported values at all taselisib dose levels.

Clinical activity

Clinical activity was assessed in the intention-to-treat population who received ≥ 1 dose of study drug (including the two patients excluded from the taselisib safety-evaluable population), with response rates reported in those with measurable disease at baseline.

Across all doses of taselisib plus docetaxel, 7/20 patients (35.0%) with baseline measurable disease had partial responses and none had a complete response (Table 3). Objective response rate was 35.0% (7/20) and CBR was 45.0% (9/20; Table 3). In the seven patients who responded to taselisib plus docetaxel, median duration of response (DoR) was 5.5 months [95% confidence interval (CI) 3.1–5.5] (Table 3). DoR was 2.6 months and 3.1 months in the two patients with PIK3CA MND, while in the four with detectable PIK3CA mutations, DoR ranged from 3.5 to 12.5 months. In the 21 patients who received any dose of taselisib plus docetaxel, median PFS was 4.1 months (95% CI 2.7–6.8) (Table 3). PFS ranged from 1.2 (censored) to 8.3 months in 11 patients with PIK3CA MND and from 6.2 to 15.1 months in five with PIK3CA mutations.

Table 3 Clinical activity
Full size table

Across all doses of taselisib plus paclitaxel, 1/54 patients (1.9%) with baseline measurable disease had a complete response and 10/54 (18.5%) had partial responses (Table 3; Fig. S1a). Objective response rate was 20.4% (11/54 patients) and CBR was 27.8% (15/54) (Table 3). In the 11 patients who responded to taselisib and paclitaxel, median DoR was 7.3 months (95% CI 4.4–12.7) (Table 3). Among seven patients with PIK3CA MND, DoR ranged from 2.1 to 36.6 months, while in four with detectable PIK3CA mutations, DoR ranged from 1.9 (censored) to 13.3 months (Table 3). In the 59 patients who received any dose of taselisib plus paclitaxel, median PFS was 4.1 months (95% CI 3.0–7.1) (Table 3). PFS by PIK3CA mutation status is shown in Table 3 and Fig. S1b. Table S8 shows efficacy in the paclitaxel-containing arm by PIK3CA mutation status by tumor tissue central analysis.

Among all 18 responders treated with taselisib plus taxanes, eight had received prior taxane therapy.

Discussion

This was an open-label, multicenter, phase Ib dose-escalation study of oral taselisib in combination with either docetaxel (Arms A, C, D, and E) or paclitaxel (Arms B and F) in patients with HER2-negative, locally recurrent/metastatic BC or NSCLC. The observed safety profiles of taselisib, paclitaxel, and docetaxel were generally consistent with the known safety profiles of PI3K-AKT-mTOR inhibitors or taxanes; however, pneumonitis was higher than expected (14%) in the taselisib plus paclitaxel arms.

DLTs were reported in the 3 mg taselisib capsule (qd for 21 consecutive days) plus docetaxel arm. DLTs were not observed in the other investigated dosing schedules and the MTD was not reached for other schedules. However, Grade 4 neutropenia was commonly observed with this combination, requiring G-CSF support throughout dose escalation and preventing the opening of the expansion cohorts.

The MTD was not reached for 3 mg and 6 mg taselisib capsules (qd for 28 consecutive days) plus paclitaxel. The 4 mg taselisib tablet (qd for 28 consecutive days in a 28-day cycle) plus paclitaxel was not recommended for further investigation based on discontinuations due to AEs and the higher number of patients with pneumonitis in Arm B. The 31.6% rate of pneumonitis/interstitial lung disease (21% at Grade 3) observed in the Arm B expansion cohort with qd taselisib was numerically higher than observed in other studies with single-agent paclitaxel or taselisib [23, 33].

The 4 mg taselisib tablet (5 days on/2 days off) plus paclitaxel (Arm F) was better tolerated than the qd schedule (Arm B), with a lower rate of discontinuation due to AEs and pneumonitis. The observed decrease in trough concentration (and Cmax) on Cycle 1, Day 15 for the 5 days on/2 days off dosing suggested that lowered exposure over the 2-day dosing holiday provides a sufficient break to improve drug safety. Plasma concentrations of taselisib, combined with docetaxel or paclitaxel, were generally consistent with studies where single-agent taselisib was administered. Paclitaxel and docetaxel exposures when given with taselisib were consistent with single-agent therapies.

Addition of PI3K inhibitors to taxanes is challenging. In BELLE-4, there was no improvement in PFS with the pan-PI3K inhibitor buparlisib plus paclitaxel versus placebo plus paclitaxel in patients with HER2-negative advanced BC (overall or PI3K-activated population), with a higher rate of discontinuations due to AEs in the buparlisib arm [34]. In PEGGY, pictilisib plus paclitaxel (vs. placebo plus paclitaxel) did not improve PFS in the intention-to-treat or PIK3CA-mutated subgroup, in patients with HR-positive, HER2-negative locally recurrent/metastatic BC, and there were a significant number of dose modifications due to AEs in the pictilisib arm [35].

In our study, and despite its relative selectivity for mutant PI3Kα, combining taselisib with paclitaxel or docetaxel was challenging. Alpelisib (BYL719), another PI3Kα inhibitor, plus nab–paclitaxel, showed encouraging efficacy and manageable toxicity in HER2-negative metastatic BC, particularly in PIK3CA-mutant tumors [36]. Highly selective PI3Kα inhibitors, such as GDC-0077, which showed tumor regression in PIK3CA-mutant breast tumor xenografts, may circumvent the narrow therapeutic index of PI3K inhibitors seen in recent clinical trials [37].

Inhibition of downstream effectors of PI3K may improve safety and/or efficacy. Ipatasertib, an AKT inhibitor, plus docetaxel or paclitaxel, was well tolerated, with improved PFS and overall survival [38, 39]. PFS was longer in patients with triple-negative BC who received ipatasertib plus paclitaxel, versus placebo plus paclitaxel [40]. In addition, paclitaxel plus AZD5363, a highly selective small-molecule AKT inhibitor, significantly improved PFS and overall survival, versus paclitaxel plus placebo, in first-line triple-negative BC [41]; the most common Grade ≥ 3 AEs were diarrhea, infection, neutropenia, rash, and fatigue [41].

In this study, taselisib plus taxanes showed preliminary evidence of antitumor activity, including partial responses in advanced BC. Our study was not randomized, but the median PFS for taselisib plus paclitaxel was less than for the paclitaxel control arms from two prior randomized studies [34, 35]. Since this was a non-randomized study in a heterogenous patient population, efficacy results should be interpreted with caution. The small population studied and multiple potential confounding factors, including tumor type (triple-negative BC; HR-positive, HER2-negative BC; NSCLC), line of therapy, dose, and schedule, mean that no firm conclusions can be reached on whether taselisib adds to the antitumor activity of taxanes or whether tumor PIK3CA mutation status plays a role.

In this phase Ib study, taselisib plus taxanes had a challenging safety profile that may be partly mitigated by intermittent dosing. While there was some evidence of antitumor activity in locally recurrent or metastatic HER2-negative BC, the overall benefit–risk profile was not clearly advantageous, and no further development of taselisib plus taxanes is currently planned.