Clinical and Genetic Predictors of Paclitaxel Neurotoxicity Based on Patient-Versus Clinician-Reported Incidence and Severity of Neurotoxicity in the ICON7 Trial

S. B. Park; J. B. Kwok; R. Asher; C. K. Lee; P. Beale; F. Selle; M. Friedlander


Ann Oncol. 2017;28(11):2733-2740. 

In This Article


Patients and Study Design

The study was undertaken using clinical data and peripheral blood DNA samples from the ICON7 trial, a multicentre, phase III trial undertaken in 1528 patients with ovarian cancer.[12] Details of the trial methods, results and quality-of-life outcomes for the entire cohort have been reported previously.[12–14] ICON7 translational research samples were provided following peer review and approval by the translational research subgroup and trial steering committee. All patients provided written informed consent. The protocol was approved by Independent Ethics Committees and in accordance with the Declaration of Helsinki. Only ICON7 patients providing consent for translational research sample collection and matched DNA and clinical data were available for this analysis (N = 454 patients). In ICON7, patients with stage 1 to 4 epithelial ovarian cancer were randomised to receive six cycles of three weekly carboplatin (area under the curve 5 or 6) and paclitaxel (175 mg/m2) with or without bevacizumab (7.5 mg/kg).[12] Patients with pre-existing Grade ≥2 neuropathy were excluded from the trial. Grade ≥2 neuropathy required paclitaxel delay until resolution to grade 1 (National Cancer Institute Common Toxicity Criteria Version grading below grade 2-Moderate symptoms; limiting instrumental ADL and grade 3-Severe symptoms; limiting self care ADL), followed by dose reduction. Grade 3 neuropathy required paclitaxel cessation.

Outcome Measures

The National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) version 3 Neuropathy Sensory subscale was used by clinicians to report neuropathy severity. The European Organisation for Research and Treatment of Cancer quality of life questionnaire–ovarian cancer module (EORTCQLQ-OV28) provided paper-based data on patient-reported neuropathy,[15] utilising two questions: 'Have you had tingling hands or feet?' and 'Have you had numbness in your fingers or toes?' Patients scored symptoms within the past week on a four-point scale corresponding to 'not at all', 'a little', 'quite a bit' and 'very much'. High scores represent greater symptom burden. Outcome measures were collected at each treatment cycle followed by 6 weekly assessments for 36 weeks. Follow-up assessment was undertaken at 15 and 18 months post-treatment commencement.

Polymorphism Selection and Assessment

DNA samples were obtained from the ICON7 biobank (University of Leeds, UK). Analysis was restricted to patients of Caucasian ancestry due to differences in allele frequencies, as per prior studies.[16] Six microtubule-associated polymorphisms were selected for analysis based on effects on microtubule dynamics: microtubule associated protein tau (MAPT; rs242557 and rs1052553[17]), glycogen synthase kinase-3 beta (GSK3B; rs6438552 and rs3755557[16]), CEP72 rs924607[11] and β-tubulin (TUBB2; rs909961[18]). Polymorphisms were genotyped using TaqMan Probe Genotyping Assays (Thermo Fisher Scientific, Waltham). Genotypes were scored to reflect their effects on gene transcription—alleles associated with lower expression were assigned values of '0' and opposing alleles assigned '1'. Additive effects of polymorphisms within the same gene were examined by a sum score. Additional details provided in supplementary methods, available at Annals of Oncology online.

Statistical Analysis

Patient demographics were summarised as percentages if categorical and medians/means if continuous variables. Clinician-reported neuropathy (CRN) was considered to be significant if NCI-CTCAE grade ≥2. Significant patient-reported neuropathy (PRN) was based on a response of 'quite a bit' or 'very much' on either neuropathy-related question of the EORTC QLQ-OV28 scale. CRN duration was based on dates of adverse events recorded during study visits. Neuropathy was considered persistent if there was no documented resolution by 18-month follow-up. The κ statistic was used to determine agreement between PRN and CRN. For polymorphism analysis, deviation from Hardy–Weinberg equilibrium was tested with chi-square tests (Supplementary Table S1, available at Annals of Oncology online). Wilcoxon signed rank tests were utilised to examine differences between groups with and without bevacizumab. Proportional hazards regression models modelled the cumulative dose to first incidence of neuropathy (dose-to-event analysis). Patients who did not experience significant neuropathy were censored at their final cumulative dose. Kaplan–Meier curves were constructed to demonstrate probability of neuropathy in relation to cumulative dose. The univariate association of baseline demographics with CIPN was investigated using proportional hazards models. All associations with P < 0.1 were used as candidate variables for a multivariable model with stepwise selection procedure. P values <0.05 were considered statistically significant. No correction for multiple testing was undertaken, in light of the exploratory analyses, which should be considered when interpreting the results. Analyses were conducted using SPSS and Stata version 13.