Figure 1. Flowchart.
Procedures.
A comprehensive clinical assessment was systematically performed before treatment start.
During the treatment period, clinical assessment of toxicities as well as blood count, liver, renal and thyroid functions were assessed every 2 weeks for the first 3 months and then monthly. Patients were instructed to monitor blood pressure at home (18,19). All adverse events were prospectively graded according to the National Cancer Institute – Common Toxicity Criteria for Adverse Events (NCI-CTCAE) version 5.0 (20). In case of grade 3 or 4 toxicity, pazopanib was suspended until improvement to grade 1-2. CT scan and/or MRI tumour evaluations were recommended every three months until progression.
The recommended starting dose of pazopanib was 800 mg/day. However, for patients with a higher risk of toxicities (ECOG PS ≥ 2, age ≥ 75 years old, albumin < 30 g/L, cardiovascular background), a lower starting dose (200 to 600 mg/day) could be prescribed at the discretion of physician. Subsequently, doses could be adjusted in 200 mg increments or decrements based on tolerance (21).
Pharmacokinetic assessments.
Blood samples were drawn at steady state every two weeks for the first three months and then monthly. Blood was collected into 5 ml lithium heparinized Vacutainer tubes at any time over the administration interval. Samples were centrifuged at 3,000 rpm for five minutes at 4°C and transferred to polypropylene tubes and kept at -20°C until assay. Plasma pazopanib concentration was measured using high-performance liquid chromatography coupled with UV detection. The intra- and inter-assay coefficients of analytical variability were less than 8% and 10%, respectively. The lower quantification limit was 1.2 mg/L. The PAZ Cmin was estimated using a Bayesian method. The population PK model of Yu et al. (11) was implemented in the Bayesian estimator.
Study endpoints.
Regarding efficacy, the primary endpoint was PFS at 3 months in STS patients. The main objective was to determine whether PAZ Cmin ≥ 27 mg/L at day 15 (D15) was associated with a longer PFS at 3 months. PFS was calculated as the time between the first day of pazopanib to tumour progression or death. Tumour progression was assessed using RECIST 1.1 criteria when measurable lesions were present or was established by the referent oncologist based on clinical findings, with retrospective confirmation by two oncologists (PBR and JA). The cut-off of 3 months for the PFS was chosen for two reasons. First, it corresponded to the first radiologic assessment, secondly it was in accordance with the primary endpoint of PALETTE trial (6).
The secondary endpoints were overall survival (OS) in STS patients and incidence of dose-limiting toxicity (DLT) during the first three months in both STS and BS patients. OS was calculated as the time between the first day of pazopanib to death (all causes included). DLT was defined as any clinical or biological grade 3 or 4 toxicity leading to treatment dose reduction, interruption (temporary stop) or permanent discontinuation.
Statistical analysis.
Statistical analyses were performed with the software R (version 4.0.3). Groups were compared with a Student’s t-test for quantitative variables, and Chi2-test for qualitative variables. Based on the results of our exploratory study and the PALETTE trial (6), we calculated that we would need to enrol 82 patients to show a 35% difference of 3-month PFS (70% vs 35% in patients with a Cmin ≥ 27 and < 27 mg/L respectively) with a two-sided 5% significance level and an 80% statistical power. Univariate and multivariable logistic regression models were used to test the association of bio-clinical variables with 3-month PFS. Variables associated with significant p-value in univariate analyses and potential confounders (initial daily dose, histological subtype) were included into multivariable models, except tumour grade (not applicable to some histological subtypes). Interaction tests revealed no significant subgroup differences. Survival curves were obtained with Kaplan-Meier estimates and compared with log-rank test. Univariate and multivariable Cox proportional hazards regression was used to identify variables associated with OS. The proportional hazards assumption was checked for each model using graphical methods based on Kaplan-Meier curves and the scaled Schoenfeld residuals. Univariate logistic regression models were used to test the association of bio-clinical variables with DLT.
All p-values were two-sided, and the level of significance was set at p < 0.05.
Our cohort study fulfils the STROBE criteria for the reporting of observational studies in epidemiology (22).
RESULTS
Patients and treatment.
One hundred and eighteen patients, mostly with STS (81%), were eligible for statistical analysis. Their baseline characteristics are outlined in Table 1. The median treatment duration was 3.9 months (range 0.4-51.0). At least one dose change was performed in 51 patients (43%), mainly within the first 3 months (36%).