A Phase Ib Dose-Escalation Study of the Oral Pan-PI3K Inhibitor Buparlisib (BKM120) in Combination with the Oral MEK1/2 Inhibitor Trametinib (GSK1120212) in Patients with Selected Advanced Solid Tumors

This article is featured in Highlights of This Issue, p. 663

The MAPK and the PI3K/AKT/mTOR (PI3K) pathways are frequently dysregulated in cancer. Molecular alterations in these pathways are implicated in tumorigenesis and resistance to anticancer therapies (1). KRAS mutations are found most frequently in pancreatic cancer (70%–90%; ref. 2), non–small cell lung cancer (NSCLC; 20%–30%; ref. 3), and colorectal cancer (30%–40%; ref. 4), whereas BRAF mutations occur in approximately 8% of human tumors, most frequently in melanoma (40%–60%; ref. 5), thyroid (39%), colorectal cancer (12%), and ovarian cancers (12%; ref. 6).

MAPK and PI3K pathways converge at multiple points, and preclinical data suggest that dual blockade may be synergistic (1, 7, 8), thus providing a rationale for evaluating safety, tolerability, and efficacy of PI3K and MEK inhibitor combinations in patients with tumors bearing genetic aberrations of these pathways.

Buparlisib (BKM120) is a potent and highly specific oral pan-class I PI3K inhibitor that does not inhibit mTOR and Vps34 kinases (9). MTD and recommended phase II dose (RP2D) of single-agent oral buparlisib are 100 mg/d (10). Key toxicities include hyperglycemia, neuropsychiatric disorders, liver toxicity, skin rash and hypersensitivity, and gastrointestinal toxicity. Trametinib (GSK1120212; Mekinist) is a reversible, highly selective allosteric inhibitor of MEK1/MEK2 activation and kinase activity (11, 12). MTD and RP2D of trametinib are 3 and 2 mg/d, respectively (13, 14). Key toxicities include rash, gastrointestinal effects, cardiac and vascular disorders, ocular toxicities, and interstitial lung disease. Trametinib has been approved as monotherapy for unresectable or metastatic BRAF V600E/K mutation-positive melanoma in the United States and Canada.

The aim of this phase Ib study was to determine MTD and/or RP2D for buparlisib combined with trametinib when administered orally to adult patients with selected advanced solid tumors, and then to evaluate safety and preliminary antitumor activity of MTD and/or RP2D in patients with advanced NSCLC, ovarian cancer, or pancreatic cancer with RAS or BRAF mutations in the expansion part of the study (clinicaltrials.gov registry identifier NCT01155453).

This phase Ib study was conducted at 7 hospitals across 5 countries in the United States, Canada, and Europe. Accrual period was May 2010 to January 2013. Dose-escalation part enrolled adult patients with advanced solid tumors harboring RAS or BRAF mutations such as colorectal cancer, melanoma, NSCLC, triple-negative breast cancer (TNBC), or pancreatic cancer. When MTD was defined, dose-expansion part was conducted in adult patients with measurable and evaluable (by RECIST version 1.0) advanced NSCLC, ovarian, or pancreatic cancer with RAS or BRAF mutations. Patients had baseline World Health Organization (WHO) performance status (PS) 0 to 2, and adequate organ function. Patients with anxiety assessed as grade ≥3 [National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 4] or ocular/retinal comorbidities associated with increased risk of central serous retinopathy or retinal vein occlusion were excluded. Prior treatment, including PI3K or MEK inhibitor therapy, was permitted if it had ended at least 4 weeks (or >1 cycle) before initiating study medication. Treatment was discontinued in cases of unacceptable toxicity, disease progression, at discretion of the investigator or by patient withdrawal.

This study was performed in accordance with the Declaration of Helsinki and the principles of Good Clinical Practice. The protocol was approved by an Institutional Review Board at each hospital, and all patients provided written informed consent before any study procedures.

Buparlisib capsules and trametinib tablets were coadministered orally once daily in a 28-day cycle. First cohort received 30 mg/d of buparlisib + 0.5 mg/d trametinib. Doses were escalated according to an adaptive Bayesian logistic regression model (BLRM) with overdose control principle until MTD and/or RP2D was reached (Fig. 1A and B; refs. 15, 16). Dose escalation could not exceed MTD of either agent as monotherapy, and only one drug was escalated at a time. Cohorts of 3 to 6 evaluable patients were enrolled per dose combination. Multiple dose combinations could be studied simultaneously. Primary endpoint was incidence rate of dose-limiting toxicities (DLT) in cycle 1. DLT was defined as an adverse event (AE) or abnormal laboratory value assessed as at least possibly related to study medication, which also met the prespecified DLT criteria. AEs were classified according to NCI CTCAE version 4. Dose modifications were permitted for treatment-related toxicity. All patients underwent ophthalmologic examinations at baseline, each cycle completion, end of study, and as clinically warranted. Secondary objectives were safety, pharmacokinetics (PK), efficacy, and predictive/pharmacodynamic (PD) biomarkers. Tumor response was classified according to RECIST version 1.0.

Blood samples for serial PK evaluation for both buparlisib and trametinib were collected on days 1, 15, and 28 of cycle 1 in dose-escalation part; and on day 28 of cycle 1 in expansion part.

Archival and fresh tumor tissue samples, whenever feasible, were collected at baseline and at cycle 1 day 28 (dose-escalation part) or at baseline and day 15 (dose-expansion part) to investigate PD effects on molecular signaling, a secondary objective, and to evaluate potential biomarkers predictive of efficacy, an exploratory objective. Assessments included biomarker status/levels pre- versus posttreatment for PD and antitumor effect, S6 (p-S6) and ERK (p-ERK) phosphorylation, measured by immunohistochemistry (IHC). In addition, exploratory molecular alteration analysis on markers relevant to PI3K and MAPK pathways, such as PTEN protein expression and PIK3CA alteration, was completed on tumor tissue via IHC and next-generation sequencing, respectively.

The BLRM was fitted on the cycle 1 DLT data accumulated throughout the dose escalation to model the dose-toxicity relationship of buparlisib and trametinib when given in combination. At each decision time point, the adaptive BLRM identified combinations that meet the Escalation with Overdose Control criteria so that only combinations associated with a risk of excessive toxicity less than 25% could be considered for the next dose cohort. Dose recommendations were based on posterior summaries for each dose combination, including the mean, median, SD, 95% credibility interval, and the interval probabilities for under dosing (DLT rate >0.16), targeted (DLT rate between 0.16 and 0.35), and excessive toxicity (DLT rate > 0.33). A clinical synthesis of available toxicity information (including AEs that were not DLTs and laboratory tests), PK, PD, and efficacy information, as well as recommendations from the BLRM, were used to determine dose combination for the next cohort. Investigators, Novartis and GSK trial personnel participated in decision making. Selected doses could not violate overdose criteria. AEs, overall response rates (ORR), disease control rate (DCR), progression-free survival (PFS), duration of stable disease (SD), and overall survival (OS) were summarized.

As of July 18, 2013, 113 patients were enrolled and treated with buparlisib and trametinib; all included in safety and efficacy analyses. Baseline characteristics are summarized in Table 1. Nine combination dose levels were explored during dose escalation, varying from 30 to 90 mg for buparlisib and 0.5 to 2.0 mg for trametinib in a total of 66 patients (Fig. 1A and B). MTD was defined as 70 mg buparlisib and 1.5 mg trametinib. The BLRM indicated that estimated DLT rate at MTD and RP2D were 23% and 18%, respectively, and the posterior risk of DLT rate being in the excessive toxicity category was less than 0.1% for both MTD and RP2D, and thus met the EWOC criteria. Twelve (19%) of 62 evaluable patients during dose escalation experienced DLTs, all of which were reversible (Fig. 1B).

There were 5 of 15 (33%) evaluable patients with DLTs in the MTD cohort and 1 of 10 (10%) in the RP2D cohort. Most frequent DLTs among all evaluable patients were stomatitis (8/103, 8%), diarrhea, dysphagia, and creatine kinase (CK) increase (2/103, 2% each).

MTD was initially used during expansion part; however, after review of tolerability of first 10 patients enrolled, this dose was revised because of high incidence of AEs (mainly stomatitis and rash). On the basis of the Bayesian inference of the dose–DLT relationship using DLT information from patients enrolled in the dose-escalation part, as well as patients enrolled at the MTD level in dose-expansion part, the BLRM estimated that combination dose of 60 mg buparlisib with 1.5 mg trametinib satisfies the EWOC criteria, and this dose was established following review of available data as RP2D. All new patients were subsequently enrolled to the starting dose of RP2D of 60 mg buparlisib and 1.5 mg trametinib. In total, 47 patients were treated during dose expansion.

A total of 110 (97%) patients experienced AEs considered by investigators to be treatment related. Overall, 73 (65%) of 113 patients experienced treatment-related grade 3/4 AEs; most frequent (>5%) were CK increase (16/113, 14%), ALT (alanine aminotransferase) increase, and stomatitis (10/113, 9% each), maculo-papular rash and macular rash (9/113, 8% each), AST (aspartate aminotransferase) increase (7/113, 6%), thrombocytopenia, and dermatitis acneiform (6/113, 5% each; Table 2). Incidence of treatment-related grade 3/4 AEs at MTD and RP2D were 82% (28 AEs) and 58% (23 AEs), respectively.

Serious AEs (SAE) related to treatment were experienced by 21 of 113 (19%) patients; most frequent were stomatitis (5/113, 4%) and diarrhea (4/113, 4%). At MTD and RP2D, 9 (27%) and 5 (13%) of 40 patients, respectively, experienced treatment-related SAEs. No treatment-related deaths were observed.

Thirty-five (31%) of 113 patients permanently discontinued study drugs due to AEs, most common of which were ALT increase and rash macular (4/113, 4% each). At MTD and RP2D, 15 (44%) and 8 (20%) patients, respectively, discontinued study treatment due to AEs. AEs requiring dose interruption and/or dose reductions were experienced by 79 patients (70%) overall, of which most frequently reported (≥10%) were stomatitis (18/113, 16%), CK increase (15/113, 13%), and ALT increase (14/113, 12%). At MTD and RP2D, 29 (85% and 73%, respectively) patients, reported AEs requiring dose interruption and/or dose reductions.

At the MTD, 20 patients (58.8%) treated with buparlisib and 22 patients (64.7%) treated with trametinib had a relative dose intensity ≥75%. At the RP2D, 39 patients (97.5%) treated with buparlisib and 26 patients (65%) treated with trametinib had a relative dose intensity ≥75%.

Buparlisib was absorbed rapidly following oral administration (median T_max was 2.17 (1.0–12.3) hours at RP2D on cycle 1 day 15. After reaching peak plasma drug concentration (C_max), buparlisib concentrations decreased in a biexponential manner. At RP2D, geometric mean values of C_max and AUC0-24 on cycle 1 day 1 and cycle 1 day 15 were 340.87 ng/mL [coefficient of variation (CV)% 50.80] and 3,106.81 ng·h/mL (CV% 32.29) and 522 ng/mL (CV% 50.1), and 6,607 ng·h/mL (CV% 32.3), respectively. Geometric mean values of the accumulation ratio on cycle 1 day 15 and day 28 at RP2D dose level were found to be 2.12 (CV% 47.04) and 2.05 (CV% 48.88), respectively (Fig. 2).

The plasma concentration–time profile of trametinib also showed rapid absorption with a median T_max of 2.95 (1.5–12.3) hours at RP2D on cycle 1 day 15 (Fig. 2). Geometric means of C_max and AUC0-24 after a single dose at RP2D were 3.36 ng/mL (CV% 69.10) and 29.60 ng·h/mL (CV% 33.56). Geometric mean values of C_max and AUC0-24 on cycle 1 day 15 at RP2D and at MTD for C_max were 19 ng/mL (CV% 33.3), and 16 ng/mL (CV% 159.5), and for AUC0-24 were 325 ng·h/mL (CV% 32.2), and 391 ng·h/mL (CV% 50.1), respectively. Geometric mean values of the accumulation ratio on cycle 1 day 15 and day 28 at RP2D dose level were found to be 9.95 (CV% 28.55) and 10.37 (CV% 23.99), respectively. Steady state was achieved within 15 days for both drugs.

Of 113 evaluable patients in the study, 7 [6.2%; 90% confidence interval (CI), 2.9–11.3; 6 ovarian and 1 NSCLC patients harboring KRAS mutations] had a best response of partial or complete responses (PR or CR).

Of the 21 patients with ovarian cancer, 4 were enrolled in the dose-escalation phase and 17 in the dose-expansion phase (Table 1A). The majority were well-differentiated serous cancer (15 patients, 71%) and were KRAS mutated (19 patients, 91%); Table 1B summarizes additional characteristics. These patients were heavily pretreated with median of three prior lines of therapy (range, 1–14). ORR was 29% (1 confirmed CR, 5 confirmed PR) and 50% at RP2D (1 confirmed CR and 3 confirmed PR; Table 3). An additional 2 (10%) of 21 patients experienced a best target lesion reduction of at least 30% without subsequent confirmation of response (one progressed, and the second underwent resection of nontarget lesion; Fig. 3A). DCR was 76% and median duration of SD was 11 months. Median PFS in all patients with ovarian cancer was 7 months (95% CI, 4.2–12.9). At time of analysis data cutoff date, median OS was not reached as a majority (18 patients, 86%) of patients with ovarian were alive. Three patients were continuing on study treatment as of July 18, 2013; one patient with ovarian cancer and CR subsequently progressed in December 2013 at cycle 15, 1 patient with NSCLC with SD progressed in August 2013 after 8 cycles and the third patient with ovarian cancer and SD was still ongoing on cycle 19 as of June 2014.

Of the 17 patients with NSCLC, 9 (53%) achieved a best overall response (BOR) of SD and 1 (6%) patient with KRAS mutation attained confirmed PR (Fig. 3A). One additional patient with NSCLC (6%) experienced a best percentage reduction of 54% in sum of longest diameters (SLD) without subsequent confirmation of the response criterion (Fig. 3A). Median PFS was 4 months (95% CI, 1.8–5.3) and median OS was 5 months (95% CI, 3.9–NA). For the patients with pancreatic cancer, 12 (50%) achieved a BOR of SD. Median PFS was 2 months (95% CI, 1.8–3.4) and median OS was almost 5 months (95% CI, 3.8–5.8).

Multiple linear regression models were performed to estimate the association with the best percentage change in SLD. There was no statistically significant exposure–response relationship, potentially due to the limited number of patients treated. Treatment duration exposure data of patients who started at MTD or RP2D dose illustrated that most responders had a durable response despite reductions from the starting dose (Supplementary Figs. S2 and S3).

For the PD assessment, a reduction in expression of p-S6 and p-ERK was observed relative to baseline. Overall, mean percentage change in the cytoplasmic levels of p-S6-240, p-S6-235, and p-ERK from baseline at cycle 1 day 28 was −25% (n = 8), −14% (n = 8), and −34% (n = 4), respectively. Highest inhibition of pS6-235 was observed in 2 patients with ovarian cancer (56.2% and 48.1% decrease from baseline, respectively), both were partial responders. There was no evidence of strong corresponding p-S6-240 or p-ERK inhibition and one of these 2 partial responders showed even increase of p-ERK levels when on study treatment. Overall, it was not possible to determine any association between efficacy and PD markers, possibly due to limited available data (Supplementary Fig. S1A–S1C).

Exploratory analyses examined any relationship between molecular status at baseline and efficacy (Fig. 3A and B). Although not statistically significant, patients with ovarian cancer with a KRAS G12V alteration appeared to be more responsive to study treatment compared with patients with other alterations (Fig. 3B). These analyses were hypothesis generating, and outcome should be interpreted cautiously because of small sample size with further study needed to confirm. No clear associations between markers relevant to PI3K and MAPK pathways and clinical activity were found (Fig. 3A).

This phase 1b study demonstrated that oral pan-PI3K inhibitor buparlisib and oral MEK1/2 inhibitor trametinib can be safely combined with an RP2D of buparlisib 60 mg and trametinib 1.5 mg daily in patients with BRAF/RAS-mutant solid tumors. The relative dose intensity reflected better tolerability with RP2D compared with MTD of the combination. However, long-term treatment was very challenging for both dose combinations. Other PI3K and MEK inhibitor combination studies have shown similar toxicity profiles to this study, with fatigue, gastrointestinal, and cutaneous toxicity being most predominant (17–22), limiting ability to administer both agents at the single-agent MTDs in combination.

Exposure and maximum concentration of buparlisib appeared to be lower than in the first-in-man (FIM) study (CBKM120X2101; ref. 10) both after a single dose and at steady state, although the accumulation ratios of AUC0-24 on days 15 and 28 were similar to what was observed in monotherapy, suggesting that this decrease might be due to a change in bioavailability. Similar phenomenon was observed in other combination trials; however, the underlying mechanism is unknown.

Exposure of trametinib appeared to be increased in the combination compared with the 2 mg dose in the FIM study (13), whereas no major change in its maximum concentration was observed. Accumulation of trametinib is highly variable making it difficult to establish if this increased exposure is due to a potential interaction with buparlisib. When comparing known PK characteristics of the two drugs, no clear mechanism could be identified for a potential interaction.

Preliminary results in other PI3K and MEK inhibitor combination studies have shown early signs of antitumor activity in colorectal cancer, renal cell carcinoma, ovarian cancer, NSCLC, endometrial cancer, melanoma, and prostate cancer (17, 19–21). In a recent phase I trial testing SAR245409 (PI3K/mTOR inhibitor) combined with pimasertib in 46 patients, there were two PRs with the combination (KRAS-mutated colorectal cancer with neuroendocrine features and KRAS/PIK3CA–mutated low-grade ovarian cancer; ref. 19).

In our study, promising antitumor activity [ORR 29% (at RP2D ORR was 50%), DCR 76%, and median PFS 7 months] was observed in RAS/BRAF–mutant ovarian cancer. In a phase II study of single-agent MEK inhibitor selumetinib in 52 patients with low-grade serous ovarian carcinoma, CR or PR was observed in 8 patients (15%), SD was observed in 34 patients (65%), and median PFS was 11.0 months (23). There were 14 of 34 (41%) patients with KRAS mutations, of which 2 (14%) showed a tumor response; this was not significantly different from the 25% response rate in KRAS wild-type tumors (5/20 patients; P = 0.672); however, median PFS in patients with ovarian cancer with KRAS mutations was not reported (23). Two ongoing trials are investigating MEK inhibitors in low-grade serous ovarian cancer: A phase III study with single-agent MEK inhibitor binimetinib (MEK162, MILO study) and a phase II/III study with trametinib.

Alternative dosing schedules of this combination might improve long-term tolerability. Currently, a phase 1b, open-label study (CMEK162X2109) is evaluating an alternative schedule of the PI3Kα inhibitor BYL719 combined with MEK inhibitor binimetinib MEK162 (MEK162 given 14 days on, followed by 7 days off and continuous dosing of BYL719) in patients with RAS- or BRAF-mutant advanced solid malignancies, such as ovarian cancer.

It has been shown that targeting the PI3K/AKT/mTOR pathway seldom works when a single agent is used (24). The current generation of PI3K inhibitors is designed to produce more specific inhibition and potentially improve toxicity profiles, although none are currently approved for treatment of ovarian cancer. Activation of these pathways and role played in promoting ovarian cancers are not completely understood (25). Results of our study indicate that buparlisib and trametinib combination treatment can inhibit PI3K and MAPK pathway signaling, although it is unclear if the degree of suppression achieved with plasma concentrations delivered is sufficient for sustained pathway blockade.

Our findings demonstrate clinical activity of a PI3K and MEK inhibitor combination in KRAS-mutant ovarian cancer, for which there is a lack of effective treatment options. We observed a higher ORR in patients with KRAS-mutant ovarian cancer compared with a single-agent MEK inhibitor in the same population. A hypothesis might be that PI3K and MEK inhibitor combination therapy is more active in certain KRAS-mutant genotypes such as G12V. Although not statistically significant, patients in this study with KRAS G12V–mutated ovarian cancers seemed to benefit more from the combination treatment. Emerging evidence in colorectal cancer and NSCLC (26–28) indicates that not all KRAS-mutant genotypes are alike in terms of their biology, and may respond differently to the same treatment. However, additional data are required to evaluate this hypothesis. Various questions arise from our data, including whether KRAS mutation is a predictive biomarker in low-grade serous ovarian cancer, and whether other histologic subtypes of ovarian cancers with RAS mutations will benefit from MEK inhibitor therapy or PI3K and MEK inhibitor combinations. To further investigate sensitivity of low-grade ovarian cancer and the role of KRAS mutations and signaling pathways in PI3K and MEK inhibitor combination therapy, additional studies are warranted.

The combination of oral pan-PI3K inhibitor buparlisib and MEK1/2 inhibitor trametinib has an MTD of buparlisib 70 mg and trametinib 1.5 mg daily and an RP2D of buparlisib 60 mg and trametinib 1.5 mg daily in patients with mutant BRAF/RAS solid tumors. Predominant toxicities of the combination were gastrointestinal and dermatologic, requiring frequent dose interruption and/or dose modification. This oral combination demonstrated promising antitumor efficacy for patients with ovarian cancer with KRAS-mutant tumors.

P.L. Bedard reports receiving commercial research grants from Astra Zeneca, Bristol-Myers Squibb, GlaxoSmithKline, Novartis, and Roche/Genentech; and is a consultant/advisory board member for Pfizer and Roche. J. Tabernero is a consultant/advisory board member for GlaxoSmithKline and Novartis. F. Janku reports receiving other commercial research support from Novartis. J. Vansteenkiste is a consultant/advisory board member for Novartis. C.D. Britten reports receiving a commercial research grant from Novartis. N.T. Le has ownership interest (including patents) in GlaxoSmithKline. K. Carter and D. Csonka are employees of and have ownership interest (including patents) in Novartis Pharma AG. D. Demanse and M. Peters are employees of Novartis Pharma AG. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Tabernero, N. Le, M. Peters, H. Nauwelaerts, C. Sessa

Development of methodology: J. Tabernero, D. Demanse, M. Peters, H. Nauwelaerts, C. Sessa

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.L. Bedard, J. Tabernero, F. Janku, Z.A. Wainberg, L. Paz-Ares, J. Vansteenkiste, E. Van Cutsem, J. Pérez-García, A. Stathis, C.D. Britten, K. Carter, D. Demanse, A. Zubel, H. Nauwelaerts, C. Sessa

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.L. Bedard, J. Tabernero, F. Janku, Z.A. Wainberg, L. Paz-Ares, J. Vansteenkiste, E. Van Cutsem, J. Pérez-García, A. Stathis, C.D. Britten, N. Le, K. Carter, D. Demanse, D. Csonka, M. Peters, A. Zubel, H. Nauwelaerts, C. Sessa

Writing, review, and/or revision of the manuscript: P.L. Bedard, J. Tabernero, F. Janku, Z.A. Wainberg, L. Paz-Ares, J. Vansteenkiste, E. Van Cutsem, J. Pérez-García, A. Stathis, C.D. Britten, N. Le, K. Carter, D. Demanse, D. Csonka, M. Peters, A. Zubel, H. Nauwelaerts, C. Sessa

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Nauwelaerts, C. Sessa

Study supervision: Z.A. Wainberg, L. Paz-Ares, K. Carter, D. Demanse, M. Peters, A. Zubel, H. Nauwelaerts, C. Sessa

The authors thank the patients, their families, care givers, and the staff and doctors at the participating centres. GlaxoSmithKline provided trametinib to Novartis for this study. Although the study was sponsored by Novartis, GlaxoSmithKline provided valuable input to the design and execution of the combination development plan. Editorial support was provided by Dr. Simone Bentley of Spectrum Regulatory Solutions and funded by Novartis.

Novartis funded study design, collection, analysis and interpretation of data, editorial support and was responsible for the decision to submit the article for publication.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.