A randomized, placebo-controlled phase III trial of masitinib plus gemcitabine in the treatment of advanced pancreatic cancer

introduction

Pancreatic cancer continues to be a disease with high unmet medical need, requiring new active agents. For over a decade, single-agent gemcitabine has been the standard first-line treatment for unresectable, locally advanced or metastatic pancreatic ductal adenocarcinoma (PDAC). Median overall survival (OS) is between 6 and 7 months and 1-year survival rates range between 17% and 25% [1, 2]. Numerous gemcitabine-based combination regimens evaluated in randomized trials have either failed to demonstrate significant improvement in OS or have shown statistically significant but rather modest survival benefits compared with gemcitabine alone; e.g. nab-paclitaxel plus gemcitabine recently reported a significant median OS gain of +1.8 months when compared with single-agent gemcitabine [1–3].

The potential therapeutic benefit of masitinib in combination with gemcitabine for the treatment of advanced PDAC has been previously reported in preclinical studies, wherein masitinib was shown to enhance the antiproliferative activity of gemcitabine in gemcitabine–refractory pancreatic cancer cell lines, and also in a clinical phase II trial [4, 5]. Exploratory analysis from the clinical study revealed two distinct patient subgroups with respect to masitinib treatment susceptibility, as evidenced by a plateau in the OS Kaplan–Meier curve between 9 and 17 months (see section A of the Supplementary Material, available at Annals of Oncology online). This observation could not be explained by patient–disease status leading to a hypothesis that there may be at least one subgroup of PDAC patients with particularly poor survival and susceptibility to masitinib plus gemcitabine treatment, the said subgroup being identifiable via a gene expression profile and/or another biological or clinical marker. Hence, future trials of masitinib in this indication would need to perform prospectively declared secondary subgroup analyses.

This observation is consistent with evidence that heterogeneity in tumor biology and microenvironment may be an important determinant of survival difference amongst groups of PDAC patients (i.e. aggressive versus relatively slow disease progression, as seen in routine clinical practice), which in turn leads to variability in terms of treatment susceptibility and potential failure of targeted drugs in the overall population [1, 6, 7]. It has been reported that such heterogeneity in PDAC patients may be associated with increased mast cell infiltration into the tumor or tumor microenvironment, both of which are prognostic factors for poor survival in PDAC [8, 9]. Masitinib is a potent oral tyrosine–kinase inhibitor (TKI) that targets a limited number of receptor tyrosine kinases including c-Kit, Lyn and Fyn, making it a highly selective inhibitor of mast cell function and activity [10].

methods

results

A total of 353 patients from 73 active centers (predominantly located in France, United States and the Czech Republic) were randomly assigned to receive masitinib plus gemcitabine or placebo plus gemcitabine. The safety population comprised all randomized patients who received at least one dose of either masitinib or placebo (N = 349). A CONSORT flow diagram for the study population as well as subgroups of interest and description of patient baseline characteristics are provided in sections D and E of the Supplementary Material, available at Annals of Oncology online. Baseline characteristics were generally well-balanced. The average number of post-study treatments was similar between treatment-arms at 1.1 ± 1.3 for the masitinib plus gemcitabine treatment-arm, and 1.0 ± 1.0 for the placebo plus gemcitabine arm, with the majority of patients receiving either single-agent gemcitabine (25% and 11%, respectively) or no additional treatment-line (27% and 31%) upon study discontinuation. Median exposure to masitinib or placebo in the safety population was 1.6 and 3.7 months, respectively, while median exposure to gemcitabine in the masitinib or placebo treatment-arms was 1.4 and 3.3 months, respectively; P = 0.001. At the data cut-off date, corresponding to a median follow-up of 26 months, one patient was ongoing treatment in the masitinib plus gemcitabine treatment-arm.

A summary of safety data is presented in Table 1. Overall toxicity increased for masitinib combined with gemcitabine when compared with single-agent gemcitabine. A higher frequency of serious and severe (grade 3 and 4) AEs, discontinuations, temporary interruptions and dose reductions was reported in the masitinib plus gemcitabine treatment-arm, although the occurrence of AE related deaths was lower in this treatment-arm than in the placebo plus gemcitabine arm. Hematological AEs contributed strongly to the discrepancy between treatment-arms, with the higher frequency reported for masitinib-treated patients due predominantly to an increase in neutropenia. No deaths were reported due to neutropenia in the masitinib plus gemcitabine treatment-arm, moreover, the occurrence of febrile neutropenia was similar between treatment-arms (1.7% for masitinib plus gemcitabine versus 0.6% for placebo plus gemcitabine), as were infections (30.6% versus 37.5%, respectively). Non-hematological AEs were typical of previously reported toxicity for masitinib, including vomiting, nausea and rash, but these generally resolved without sequelae and were not associated with any deaths.

Patient QoL at baseline was similar between the treatment-arms (mean global health score of 53.5 ± 22.4 versus 53.9 ± 21.1 for the masitinib plus gemcitabine and placebo plus gemcitabine treatment-arms, respectively), as well as at the last patient visit (46.3 ± 23.7 versus 49.7 ± 21.7, respectively). The combination of masitinib plus gemcitabine did not, therefore, accelerate the decline in QoL with respect to single-agent gemcitabine.

The median OS for the overall population, the primary efficacy analysis, was similar for both treatment-arms; 7.7 months [95% CI (6.1; 10.6)] for masitinib plus gemcitabine and 7.0 months [95% CI (6.1; 10.6)] for placebo plus gemcitabine (all results reported hereafter relate to the multivariate analysis unless otherwise stated) (Table 2). The corresponding HR was 0.89 [95% CI (0.70; 1.13)]. Secondary analyses on surrogate survival endpoints of the overall population, e.g. progression-free survival or time-to-progression, were also similar between treatment-arms (data not shown).

Secondary analyses on OS did, however, show two subgroups of patients having particularly poor survival with single-agent gemcitabine, which was consistent with the study's hypothesis and prospectively declared subgroup analysis. These subgroups comprised patients with a genetic biomarker (overexpression of ACOX1 in blood), and patients with baseline pain intensity above a threshold of 20 mm as measured on a 100 mm visual analog scale (VAS). In both cases, the placebo plus gemcitabine patient cohorts divided into two distinct subgroups with survival reflecting aggressive or relatively slow disease progression (Figure 1), thus characterizing the defining variables of the prospectively declared secondary subgroup analysis (for further subgroup description see sections C and F of the Supplementary Material, available at Annals of Oncology online). Subsequent evaluation of the interaction between these variables and the combination of masitinib plus gemcitabine revealed a significant treatment benefit in both subgroups with respect to the placebo plus gemcitabine treatment-arm (Table 2).

Considering the patient cohort with pharmacogenomic data, 119 patients enrolled for the study had peripheral blood samples collected at baseline and were randomly assigned to the masitinib plus gemcitabine or placebo plus gemcitabine treatment-arms (n = 60 and n = 59 patients, respectively). The ACOX1 subgroup was determined following a pre-specified methodology as patients with overexpression of ACOX1 in blood defined as a delta cycle threshold (DCt) value of ≤3.05 (see section C of the Supplementary Material, available at Annals of Oncology online). In the overall pharmacogenomic population, a total of 40/119 patients (34%) were identified as being in the ACOX1 subgroup while 79/119 patients (66%) were assigned to its complement subgroup (i.e. absence of ACOX1 overexpression or non-ACOX1). In the ACOX1 subgroup, median exposure to masitinib or placebo was 1.8 and 2.4 months, respectively, while median exposure to gemcitabine in the masitinib or placebo treatment-arms was 2.1 and 1.9 months, respectively; P = 0.78. In the placebo plus gemcitabine treatment-arm, patients without ACOX1 overexpression (n = 39) had a significantly longer median OS compared with patients having ACOX1 overexpression (n = 20); 8.8 months [95% CI (5.6; 15.0)] versus 5.5 months [95% CI (3.4; 8.3)] (univariate model). The corresponding HR was 0.46 [95%CI (0.26; 0.82)], P = 0.007 (Figure 1A).

In the aforementioned ACOX1 subgroup, those patients treated with masitinib plus gemcitabine (n = 20) had a median OS of 11.7 months [95% CI (8.3; 19.9)] compared with a median OS of 5.6 months [95% CI (3.7; 12.9)] for the placebo plus gemcitabine treatment-arm (n = 20) (multivariate model), a statistically significant OS gain of + 6.1 months. The corresponding HR was 0.23 [95% CI (0.10; 0.51), P < 0.001] (Table 2). Overall survival rates at 6, 12, 18, and 24 months were respectively, 82%, 48%, 15%, and 11%, in masitinib plus gemcitabine treatment-arm versus 45%, 8%, 0.6%, and 0.3%, in the placebo plus gemcitabine treatment-arm. Safety in the ACOX1 subgroup was similar to the overall safety population (data not shown).

Considering the prospectively declared subgroup based on a baseline clinical characteristic, i.e. pain intensity tested once at baseline, 312 patients from the mITT population had VAS data available. The ‘pain’ subgroup, 137/312 patients (44%), included all patients reporting a VAS score of >20 mm, this threshold being consistent with established precedent and defined prior to unblinding (see section F of the Supplementary Material, available at Annals of Oncology online). Comparison was made against patients reporting negligible baseline pain intensity, defined by a VAS < 5 and not requiring opioid analgesics to manage disease-related pain, referred to hereafter as the ‘no pain’ subgroup (n = 68/312 patients, 22%). All remaining patients, i.e. those with a baseline VAS ≥ 5 but <20 or VAS < 5 but taking analgesics opioids (n = 107/312, 34%) formed a third subgroup. In the ‘pain’ subgroup, median exposure to masitinib or placebo was 1.5 and 2.5 months, respectively, while median exposure to gemcitabine in the masitinib or placebo treatment-arms was 1.4 and 2.3 months, respectively; P = 0.17. In the placebo plus gemcitabine treatment-arm, patients from the ‘no pain’ subgroup (n = 34) had a significantly longer median OS than patients in the ‘pain’ subgroup (n = 73), 16.9 months [95% CI (13.2; 22.2)] versus 5.6 months [95% CI (4.4; 8.0)] (univariate model). The corresponding HR was 0.30 [95% CI (0.18; 0.48), P < 0.001] (Figure 1B).

In the aforementioned ‘pain’ subgroup those patients treated with masitinib plus gemcitabine (n = 64) had a median OS of 8.0 months [95% CI (5.8; 11.5)] compared with a median OS of 5.4 months [95% CI (3.7; 8.3)] for the placebo plus gemcitabine treatment-arm (n = 73) (multivariate model), a statistically significant OS gain of + 2.6 months. The corresponding HR was 0.62 [95% CI (0.43; 0.89), P = 0.012] (Table 2). Overall survival rates at 6, 12, and 18 months, were respectively, 58%, 32%, and 18%, in the masitinib plus gemcitabine treatment-arm versus 44%, 18%, and 8%, in the placebo plus gemcitabine treatment-arm. Safety in the pain subgroup was similar to the overall safety population (data not shown). One also notes that the frequency of patients reporting back pain as an AE during treatment was significantly lower (P = 0.004) in the masitinib plus gemcitabine treatment-arm than in the placebo plus gemcitabine treatment-arm of the safety population (Table 1).

Internal validation of masitinib's effect in patients from the ‘pain’ subgroup is provided through analysis of survival data in patients consuming high doses of opioid analgesics at baseline (>1 mg/kg/day), referred to hereafter as the ‘high opioid’ subgroup (n = 34). Briefly, it is a fair assumption that such patients were experiencing moderate to severe cancer-related pain to justify initiation of such pain management measures and are, therefore, comparable to patients from the ‘pain’ subgroup. Patients in the exploratory ‘high opioid’ subgroup and treated with masitinib plus gemcitabine (n = 20) had a median OS of 8.5 months [95% CI (6.0; NA)], whereas patients treated with placebo plus gemcitabine (n = 14) had a median OS of 6.0 months [95% CI (3.5; NA)]. This corresponds to a survival benefit of 2.5 months and HR of 0.43 [0.17; 1.06]; P = 0.23.

discussion

Although no discernible difference between treatment-arms was observed for the primary endpoint in the overall population, this study did identify subgroups with remarkably poor survival while under single-agent gemcitabine. Patients with overexpression of ACOX1 or baseline pain (VAS > 20 mm) had a worse prognosis (median OS of 5.6 and 5.4 months, respectively) with respect to the overall population (median OS of 7.0 months) and historical median OS data for gemcitabine-treated patients (typically 6.5 months) [1]. Such data illustrate that the markers of ACOX1 expression in blood and baseline pain intensity may have prognostic value, with patients from these subgroups experiencing aggressive disease progression while receiving single-agent gemcitabine. It is estimated that together, these subgroups encompass 63% of the entire PDAC population (i.e. 34% of patients who were identified as belonging to the ‘ACOX1’ subgroup and 29% of patients who were identified as belonging to the ‘pain’ subgroup with no overexpression of ACOX1, amounting to 63% of patients in one or the other subgroup).

Both parameters of ACOX1 and baseline pain (VAS > 20 mm) also suggested predictive value with the masitinib plus gemcitabine treatment-arm showing a statistically significant median OS gain of + 6.1 months [HR = 0.23 (0.10; 0.51)] in the ACOX1 subgroup and + 2.6 months [HR = 0.62 (0.43; 0.89)] in the pain subgroup when compared with single-agent gemcitabine. Although there was increased toxicity with the addition of masitinib to gemcitabine, safety remained within acceptable limits with application of appropriate risk management measures and there was no overall detrimental effect on QoL. Therefore, the combination of masitinib and gemcitabine for the treatment of advanced PDAC appears to exhibit a positive benefit–risk ratio for these subpopulations. Of note, the pharmacogenomic examination of RNA expression in peripheral blood samples also identified a set of ten genes with high discriminatory power, albeit ambiguous biological plausibility, with ACOX1, representing the single most important gene to explain OS (see section C of the Supplementary Material, available at Annals of Oncology online).

There is an emerging consensus that under certain circumstances it is possible for a subgroup to be considered of clinical significance (see section G of the Supplementary Material, available at Annals of Oncology online). The present study has met these criteria. For example, internal consistency supporting the clinical plausibility of each subgroup is provided from independent patient samples (see sections C and F of the Supplementary Material, available at Annals of Oncology online). Considering biological plausibility, it is thought that the presence of baseline pain (VAS > 20 mm) or an overexpression of ACOX1 effectively identifies those patients with a pro-tumoral T-helper cell type-2 (Th2) immune response, a condition caused in part by increased mast cell activity in the tumor microenvironment or by transcriptional or physiological alterations favoring M2-polarization of tumor-associated macrophages (TAM) (see section H of the Supplementary Material, available at Annals of Oncology online). For instance, mast cells have been implicated with the development of neuropathic pain in PDAC patients and skewing macrophage polarization towards a pro-tumoral M2-type [11, 12]. Furthermore, recent preclinical data from KrasG12D driven mouse models of PDAC with pain or spontaneous chronic pancreatitis show that pancreatic tumor lesions of masitinib-treated mice have decreased mast cell count and reduced intra-tumoral vascularization and innervation when compared with control mice (Dubreuil P, 2014; personal communication). Other nascent research suggests masitinib may induce the recruitment of macrophages with a potential for antitumoral activity within the tumor (Hermine O, 2014; personal communication). Thus, mechanisms of action associated with masitinib apparently converge towards favoring a preferential accumulation of antitumoral M1-macrophages in the tumor microenvironment with concomitant reduction of oxidative stress effects. Presentation of these supportive data fall beyond the scope of the current clinical paper with additional translational research needed to fully elucidate such mechanisms; as such, these preclinical data will be reported in full elsewhere.

In conclusion, the survival benefit observed for PDAC patients with overexpression of ACOX1 in blood or reporting baseline pain of VAS > 20 mm when treated with masitinib plus gemcitabine, coupled with manageable toxicity suggests a positive benefit–risk ratio. This has led to the initiation of a confirmatory study that may support the use of masitinib plus gemcitabine as a new treatment option for these two subgroups of PDAC patients.

funding

This study was financially supported by AB Science, Paris, France (no grant number) and by Acobiom, Montpellier, France (no grant number). The Sponsors were involved in the study design; data processing (monitoring, collection, analysis, and interpretation); manuscript preparation and submission. The corresponding author had full access to all the data and had final responsibility for the decision to submit for publication. All authors approved the final manuscript.

disclosure

Masitinib is under clinical development by the study sponsor, AB Science (Paris, France). AM, CM, and YA are an employees and shareholders of the study sponsor AB Science. OH and PD are consultants and shareholders of AB Science. LM served as a paid consultant to AB Science. DP and DR are employees and shareholders of the study sponsor Acobiom. All other authors declare no conflicts of interest.

Supplementary Material