A phase Ia/Ib clinical trial of metronomic chemotherapy based on a mathematical model of oral vinorelbine in metastatic non-small cell lung cancer and malignant pleural mesothelioma: rationale and study protocol

Vinorelbine is a semi-synthetic vinca-alkaloid that acts by inhibiting tubulin polymerization during mitosis. It is as active on the mitotic microtubules as other vinca-alkaloids, but less active against axonal microtubules, thus explaining its lower neurotoxicity [7].

Vinorelbine can be administered either intravenously or orally. The bioavailability of oral vinorelbine varies from 33 to 43 % [8, 9]. The tolerance is well documented and is comparable to the intravenous form [7].

The therapeutic efficacy has been reported clinically and might legitimize the use of vinorelbine in patients with lung cancer. However, this drug is also associated with a significant risk of haematological and non-haematological adverse events in these fragile and pre-treated patients [10].

Many features of vinorelbine make it an ideal candidate for the development of metronomic strategies. Vinca alkaloids, including vinorelbine, have an anti-angiogenic activity, demonstrated in vitro [11, 12] and in vivo [13‐28] after metronomic dosing. Due to its ease of administration, oral vinorelbine has paved the way for innovative treatment strategies through metronomic regimens.

In the first-line setting for advanced NSCLC, three phase III trials have investigated the combination of vinorelbine with platinum derivatives [13‐15]. These studies confirmed the superiority of a combination of platinum derivatives and vinorelbine over single-agent chemotherapy. Subsequently, several meta-analyses [16‐18] have demonstrated that the use of vinorelbine-based regimens may be less effective in controlling disease than other combinations with platinum derivatives in first-line therapy. Similarly, the use of single-agent vinorelbine as first-line therapy for advanced disease in fit elderly patients has been supplanted by the carboplatine/weekly paclitaxel doublet [19].

In the second-line setting, vinorelbine was one of the comparators (along with ifosfamide) of a phase III trial of second line docetaxel [20] in which docetaxel (75 mg/m2) led to better overall survival than the comparators, thus establishing it as the standard choice for second-line treatment. A retrospective study that evaluated 39 patients with stage IIIb or IV treated with vinorelbine after a median of two prior lines of treatment, showed a partial response rate of 7.7 % and stable disease in 25.6 % of patients [21].

However, several clinical trials have evaluated vinorelbine in metronomic regimens in different tumour types, including NSCLC [22‐24]. In a first metronomic clinical trial, conducted in 62 patients with advanced cancers, vinorelbine has been tested at doses ranging from 20 to 70 mg thrice a week [22]. At the end of this phase IA, the preferred effective dose was 50 mg thrice a week. Afterward, a phase IB trial, led by the same team in 101 patients with breast cancer, prostate cancer or NSCLC, compared the following dosages: 30 mg thrice a week, 40 mg thrice a week and 50 mg thrice a week. Despite the differences in dose, progression-free survival and tolerance did not differ in the 3 groups, with a total of 10 % of grade 3–4 neutropenia [23]. A metronomic chemotherapy Phase II trial of oral vinorelbine 50 mg thrice a week was conducted in 46 patients with treatment failure in the treatment of NSCLC. The response rate was 10.9 %. More side effects were recorded in this study, with mostly grade 3–4 neutropenia in 23.9 % of patients and febrile neutropenia in 10.9 % of them [24]. Recently, a prospective study recruited 43 chemotherapy naive elderly patients with stage IIIB-IV NSCLC, treated with oral vinorelbine 50 mg thrice a week. Overall response rate (ORR) was 18.6 % and tolerance was at an acceptable level [25].

Vinorelbine has demonstrated activity in MPM. The European Society for Medical Oncology (ESMO) guidelines consider that vinorelbine might be a reasonable choice in second-line therapy [26]. A non-comparative phase II study in which vinorelbine was administered at a dose of 30 mg/m² for 6 weeks in 63 patients previously treated with chemotherapy found a 16 % response rate and controlled disease in 68 % of patients. A major haematologic toxicity was noted, with a grade 3–4 neutropenia in 55 % of patients [27]. These second-line response rates were confirmed by a recent retrospective study on 59 patients included after a first-line treatment, including pemetrexed. The response rate was 15.2 %, with a 49.1 % disease control rate and a 6.2-month overall survival. In this study, vinorelbine, administered at a dose of 25 mg/m² D1-D8 every three weeks, had a better tolerance with grade 3–4 neutropenia in 8.4 % of patients [28].

These clinical trials showed the limits of empirical approaches to determine appropriate metronomic strategies in oncology. These first results, obtained in different populations and with empirical treatment regimens, suggest the development of new approaches, such as bio-mathematical modelling, for the selection of metronomic schedules. Complex situations inherent to the administration of anticancer agents can be managed using pharmacokinetic-pharmacodynamic models (PK-PD). Previous studies showed that numbers of critical questions have to be addressed when switching to a metronomic dosing regimen. Which dose should be used? What is the best dosing schedule? Among the most effective protocols, which one will be the least toxic? Because countless schedules are possible, testing them empirically, either clinically or using non-clinical models, seems to be an unachievable goal. Mathematical modelling can be an effective tool to address the above questions [29]. The modelling approach can transpose into mathematical language the action of the drug on the body, both for efficacy and toxicity. Once the model is built, a search for an optimal solution (i.e., protocol achieving greater efficacy while respecting pre-defined constraints on toxicity) can be performed in silico by the iterative simulation of thousands of different schedules. Studies have provided further support for this method [30].

The aim of the clinical trial presented below is to effectively use oral vinorelbine in a metronomic schedule, developed and validated by mathematical modelling, in patients with NSCLC or MPM after failure of standard treatment.

This study is a prospective non-randomized phase Ia/Ib trial of the single agent oral metronomic vinorelbine. The study will take place in the Phase I Oncology Unit (Centre d’Essais Précoces en Cancérologie de Marseille APHM CLIP²), 264 rue Saint-Pierre, 13005 Marseille, France.

Several model-based strategies to address the issue of keeping drug-related toxicities under control while improving anti-tumour efficacy [31‐33] have been developed. Recently, our group has developed a PK-PD mathematical model specifically dedicated to the management of oral metronomic chemotherapy [29, 34]. The model is divided into 3 parts:

With adjusted parameters, the PK-PD model described adequately the clinical data published by Briasoulis et al. (30, 40 and 50 mg, thrice a week, D1, D3, D5) both in terms of efficacy and toxicity. In a second step, the model was used to simulate, for a known average PK profile, alternate continuous metronomic schedules, achieving higher efficacy while being tolerated. A dosing regimen weekly D1, D2 and D4 with 60, 30 and 60 mg dose respectively was selected. Simulation of this protocol, named VTP (Vinorelbine Theoretical Protocol), resulted in more favourable efficacy profile and significant reduction of variability in response while maintaining a total dose per week of 150 mg (Fig. 1).

To be eligible for inclusion, patients will have to provide signed informed consent forms before undergoing any study-related procedures and have the willingness and ability to comply with scheduled visits, laboratory tests and other study procedures. All patients must be aged 18 or older, have histologically or cytologically proven metastatic NSCLC or MPM, and a progression of disease after standard treatments. All patients must have at least one measurable target by RECIST 1.1. and an Eastern Cooperative Oncology Group (ECOG) performance status between 0 and 2. An adequate haematological (neutrophil count ≥ 1,500/mm³, platelet count ≥ 100,000/mm³, haemoglobin ≥ 9.0 g/dL), hepatic (total bilirubin ≤ 1.5 x ULN, aspartate transaminase (AST) and alanine transaminase (ALT) ≤ 2.5 x upper limit of normal (ULN), or AST and ALT ≤ 5 x ULN (if liver function abnormalities are due to cancer) and renal (creatinine clearance based on the Cockcroft-Gault formula ≥ 45 ml/min) function is required. Time between the end of the previous treatment and inclusion in the clinical study will have to be 4 weeks for experimental treatment as part of a trial, or 3 weeks for chemotherapy, or if this time is less than 4 weeks, 5 times the half-life of previous treatment for targeted therapy. Patients will be ineligible to participate if they have a peripheral neuropathy grade > 1, an uncontrolled cardiac disease requiring treatment (heart failure, angina pectoris, arrhythmia), a recent myocardial infarction (≤6 months) or a history of cancer other than basal cell carcinoma, treated cervical intraepithelial neoplasia or any other cancer treated without recurrence for at least 2 years. Patients with an active infection or any other serious medical or psychiatric condition that could affect participation in the trial (according to the opinion of the investigator) will also be excluded.

Patients will benefit all useful premedications and post-medications.

During the continuation of protocol-defined treatment, laboratory tests (cell blood count) and clinical examination will be conducted at D1 of each week. Tumour response will be assessed at baseline and every 6 weeks from the 1^st day of treatment by performing a cerebral, thoracic, abdominal and pelvic CT scan. All assessments will be performed by investigators using RECIST version 1.1 (1.1 modified for MPM). The treatment will be administered until radiological disease progression, unacceptable toxicity, or withdrawal of consent.

After the end of the treatment period, patients will be seen every month or contacted by phone to collect data on potential adverse events and survival. The planned duration of the trial is estimated to be 36 months (duration of inclusions: 12 months; Follow-up: up to 24 months after the last patient inclusion).

The primary endpoint of the phase Ia is the tolerance, assessed by CTC v4.0, with an objective of less than 10 % of patients with haematological toxicity (neutropenia) grade 3–4.

The primary endpoint of the phase Ib is the objective response according to RECIST 1.1 (NSCLC) or modified RECIST 1.1 (MPM).

Secondary endpoints include Overall Response Rate (ORR), Duration of Response (DR), Duration of Disease Control (DDC), Disease Control Rate (DCR), Progression-Free Survival (PFS), Overall Survival (OS), Tolerance (assessed according to CTC V4), Quality of life assessed by LCSS (Lung Cancer Symptom Scale) and EORTC QLQ C30 and C13, the PK-PD relationship of oral vinorelbine under VTP and OVTP schedules.

Individual pharmacokinetic analysis will be based on two sets of four blood samples to determine the individual pharmacokinetic profiles of vinorelbine. Sampling times will be calculated by a D-optimality method based on literature data on the pharmacokinetics of vinorelbine [38].

During this clinical trial, an ancillary study on biomarker analyses will be conducted: circulating endothelial cells (CEC), fibroblast growth factor 2 (FGF2), vascular endothelial growth factor (VEGF), interleukin-8 (IL-8), and thrombospondin-1 (TSP1) and Tie2/Tek. The results will be analysed to determine any correlation with the data obtained by RECIST in terms of tumour response and/or efficacy.

The expected number of patients is 32, with the following distribution:

Twenty patients will receive the OVTP schedule.

The number of patients in the study is based on the Fleming one-step method, with a type I error (α) equal to 5 % and a type II (β) equal to 10 %.

The statistical hypotheses are determined by:

The following will be tested: H0: p = p0 ≤ 25 % versus H1: p ≥ p1 = 50 %. According to these hypotheses, the calculation shows that 20 patients should be included in the study.

All patients receiving at least one dose of study drug will be included in the evaluation of safety. All patients who received at least one dose of study drug will be included in the intention-to-treat population and included in the assessment of effectiveness. Patients with valid pharmacokinetic data will be included in the pharmacokinetic analyses.

All efficacy parameters will be analysed descriptively. For the ORR and DCR, estimated 90 % and 95 % confidence intervals will be provided. For the PFS, the duration of the response and overall survival, the Kaplan-Meier method, the median durations and 95 % confidence intervals will be calculated.

It is expected that a consultation meeting will be organized between the members of the project and the pharmaceutical company Pierre Fabre, after the inclusion of 12 patients in Phase I (4 weeks after the D1 of the 12^th included patient). This meeting will aim to analyse the adequacy of the model prediction and clinical outcomes (review of toxicity and clinical efficacy) in perspective with other results held by Pierre Fabre on trials underway elsewhere.

The transition to the second stage will take place after a favourable decision of the monitoring committee.