Francis W. Ojara (1,2), Andrea Henrich (1,2), Niklas Hartung (5), Markus Joerger (3), Max Roessler (4), Nicolas Frances (6), Wilhelm Huisinga (5), Charlotte Kloft (1)
(1) Dept. Clinical Pharmacy & Biochemistry, Institute of Pharmacy, Freie Universitaet Berlin, Germany, (2) Graduate Research Training Program PharMetrX, Germany, (3) Medical Oncology and Clinical Pharmacology, Dept. of Internal Medicine, Cantonal Hospital St. Gallen, Switzerland, (4) CESAR central office, Vienna, Austria, (5) Institute of Mathematics, Universitaet Potsdam, Germany, (6) Dept. of Translational Modelling and Simulation, F. Hoffman La Roche Ltd, Switzerland.
Objectives:
Peripheral neuropathy (PN), a dose-limiting adverse event of paclitaxel (PTX) affects > 20% of patients on PTX therapy and negatively impacts quality of life. Covariates e.g. PTX dose and time of plasma concentration above 0.05 μM have previously been identified as influential by comparing odds of PN for patients at different covariate levels [1,2,3]. However, the above analyses do not account for impact of time on treatment and censored observations (unknown incidence times) on the risk of PN. The aim of this analysis was to describe the time-course of occurrence of clinically important PN (CIPN, grades 2 or 3) and explore the influence of different treatment and patient characteristics on the risk of CIPN to support dose adaptation and hence reduce the occurrence of PTX-associated CIPN.
Methods:
Patients from the CEPAC-TDM study (n = 366), who received PTX in combination with carboplatin or cisplatin every 3 weeks for ≤ 6 cycles [4], were included. PN symptoms, severity, start and end dates were recorded and classified using the common toxicity criteria (version 4.0) [5]. Time-to-event (TTE) analysis was employed to describe the risk of incidence of CIPN (event) during treatment. Constant, Weibull, Gompertz and cycle-varying hazard models were examined. Impact of covariates, PTX dose, age, weight, sex, and smoking status were jointly evaluated in a full covariate model [6] with covariate selection based on prior clinical and mechanistic knowledge, and information content of the different covariate categories. Statistical and clinical significance of covariates were derived from the distribution of hazard ratios corresponding to different covariate levels. Statistical significance was attained if distribution of hazard ratios didnot include 1 (the null value) whereas the region of clinical significance was set to +/- 20% of the null value. The PTX dose-CIPN risk profile was further explored by simulating incidence of CIPN at two clinically relevant dose levels: 200 mg/m2 and 175 mg/m2, with a standard treatment schedule
(6 cycles, 21 days each). Dataset formatting was performed in R (3.4.3) and TTE analysis in NONMEM (7.3.0).
Results:
105 PN events were reported with generally higher incidences at cycle start and gradual decline across the cycle. The cycle-varying hazard model, describing a surge in risk of CIPN at cycle start and gradual decline across each cycle, best captured the data. PTX relative dose and age had both statistically and clinically significant impact on the risk of CIPN whereas weight, sex and smoking status were not statistically significantly associated with CIPN. A 19% increase in risk of CIPN with 200 mg/m2 over 175 mg/m2 was predicted: hazard ratio (95% CI), 1.19 (1.06, 1.33). The risk of CIPN increased with age. There were also trends towards increase in risk of CIPN with increased weight and smoking status. Since platinum drug type (cisplatin or carboplatin) and diabetic status (diabetic or non-diabetic) had unbalanced proportions (less than 20% for the less represented category) these categorical covariates were not included in the full covariate model.
Conclusions:
We described the occurrence of CIPN and determined the impact of various treatment and patient characteristics on the risk of CIPN using TTE analysis. With this methodology we additionally accounted for the information about time on treatment and censored observation on risk of PN. The risk of CIPN increased with increase in administered dose and age. The model enables evaluation of the impact of treatment and patient characteristics on the risk of CIPN.
References:
[1] Argyriou A et al. Crit. Rev. Oncol. Hematol., 66: 218, 2008
[2] Kraff S et al. Cancer Chemotherapy. Pharmacol. 75: 975, 2015.
[3] Ojara FW et al. PAGE 26 (2017): Abstr 7278
[www.page-meeting.org/default.asp?abstract=7278]
[4] Joerger M et al. Ann. Oncol. 27: 1895, 2016.
[5] National Institute of Health, National Cancer Institute, Common Terminology Criteria for Adversed Events (CTCAE), 2009
[6] Gastonguay MR. PAGE 20 (2011), Abstr 2229
[www.page-meeting.org/?abstract=2229]
Reference: PAGE 27 (2018) Abstr 8610 [www.page-meeting.org/?abstract=8610]
Poster: Drug/Disease Modelling - Oncology