Individual Prediction-Based Dose Adaptation Of Capecitabine: In Silico Comparison With The Standard Method, Impact On Limiting Toxicity and On Antitimour Efficacy
Inès Paule (1), Michel Tod (1), Emilie Hénin (1), Benoit You (1,2), Gilles Freyer (1,2), Pascal Girard (1)
(1) EMR 3738 CTO, Faculté de Médecine Lyon-Sud, Université Claude Bernard - Lyon 1, France; (2) Service d’Oncologie Médicale, Hospices civils de Lyon, France
Objectives: Individual prediction-based dose adaptation (IPBDA) method using an ordinal data model of toxicity has been developed for anticancer drug capecitabine-induced hand-and-foot syndrome (HFS) and presented previously . A comparison with the standard dose adaptation by in silico clinical trials was performed and showed the superiority of the IPBDA in terms of toxicity, but did not inform about the impact on antitumor efficacy. When a model of colorectal tumours and capecitabine's effect on them has been published , we could extend the previous work by including the efficacy aspects into simulations and dose adaptation decisions. The objective was to find the most advantageous dose adaptation method in terms of grade>=2 HFS reduction without reducing antitumor efficacy by comparison with the standard approach for dose adaptation.
Methods: HFS and tumour models: HFS grade probabilities were described using a mixed-effects proportional odds Markov model . Tumours were described by a mixed-effects model relating the sum of the largest tumour diameters and drug doses .
Dose adaptation procedures: The standard method was to reduce the initial dose by 25% after second event with HFS grade>=2 and by 50% after the third one. The IPBDA consisted of: (1) estimating the individual random effects (EBEs of ETAs); (2) choosing the new dose so that the average risk of HFS grade>=2 over the next 3 weeks would be closest (but not higher) to the target risk. Dose reductions could be started after the first occurrence of HFS (even grade 1). Many variations were tested: different target risks, lower and upper limits for dose, conditions for allowing dose increase.
Comparison was made in terms of toxicity related criteria: % of patients having (reoccurring) events with grade>=2 HFS, average number of weeks with grade>=2 HFS, of events with grade >=2 HFS, their duration, % of patients who dropout due to HFS, as well as efficacy related criteria: % of patients having tumour response, % of patients who have progression of disease, relative change from baseline of tumour sizes.
Results: The best results in terms of HFS grade>=2 reduction and equivalent antitumour effect were found with IPBDA where the target risk was 4%, lower dose limit was 50% of the initial dose, dose increases were allowed up to 150% for patients without any HFS starting after 12 weeks or those having grade 1 for at least 6 consecutive weeks. The benefit was on average 10 days less of grade>=2 HFS, by reduced frequency and length of reoccurring events with grade>=2, 7% less of dropouts due to HFS.
Conclusions: IPBDA provided a small but clinically relevant improvement. Dose adaptation based on ordinal data model is limited by poor EBEs . However, in the case of HFS, the main hurdle to reduce toxicity was the insensitivity of response to dose changes.
 Paule, I. et al. Model-based dose adaptation of capecitabine for prevention of severe hand-and-foot syndrome: in silico comparison with the standard method. PAGE 17 (2008) Abstr 1260 [www.page-meeting.org/?abstract=1260]
 Claret, L. et al. Model-based prediction of phase III overall survival in colorectal cancer on the basis of phase II tumor dynamics. J Clin Oncol. 27, 25, 4103-4108 (2009)
 Henin, E. et al. A dynamic model of hand-and-foot syndrome in patients receiving capecitabine. Clin. Pharmacol. Ther. 85, 4, 418-425 (2009)
 Paule, I. et al. Estimation of Individual Parameters of a Mixed-Effects Dose-Toxicity Model for Ordinal Data. PAGE 19 (2010) Abstr 1785 [www.page-meeting.org/?abstract=1785]