Félicien Le Louedec (1), Maud Maillard (1), Manon Launay (2), Bernard Royer (3), Marie-Christine Etienne-Grimaldi (4), Camille Tron (5), Céline Narjoz (6), Hugo Alcaran (7), Fabienne Thomas (1), on behalf of the Groupe de Pharmacologie Clinique Oncologique (GPCO)
(1) Laboratoire de Pharmacologie, Institut Claudius Regaud, IUCT-Oncopole et Centre de Recherches en Cancérologie de Toulouse, Inserm UMR1037, Université Paul Sabatier, Toulouse, France, (2) Laboratoire de Pharmacologie et Toxicologie, CHU de Saint-Etienne, Saint-Etienne, France, (3), Laboratoire de Pharmacologie Clinique et Toxicologie, CHU de Besançon, Besançon, France (4) Laboratoire d’Oncopharmacologie, UPRC EA 7497, Centre Antoine Lacassagne, Nice, France, (5) Laboratoire de pharmacologie CHU de Rennes, Univ Rennes, CHU Rennes, Inserm, EHESP, Irset (Institut de recherche en santé, environnement et travail) - UMR_S 1085, F-35000 Rennes, France, (6) Assistance Publique des Hôpitaux de Paris, Hôpital européen Georges-Pompidou, Service de biochimie, Paris, (7) Service de Biochimie et Biologie Moléculaire, CHRU de Tours, Tours, France
Objectives:
Dihydropyrimidine dehydrogenase (DPD) deficiency testing is mandatory in France [1] and recommended by the European Medicines Agency before initiating a fluoropyrimidine (FP)-based treatment in cancer patients [2]. This consists of measuring the uracil plasma concentration (U), the physiological substrate of DPD. For example, a partial deficiency is defined as U > 16 ng/mL and justifies decreasing FP dose. However, strict pre-analytical conditions must be respected to guarantee test interpretation since U increases if blood is not immediately centrifuged and stored at -20°C until being analyzed. To provide robust guidelines for this pre-analytical process, several laboratories have conducted their own stability studies, but they lack rigorous methodology: not respecting identical times across studies, using small data size. The objective of the current study is to provide recommendations for this pre-analytical process using a non-linear mixed-effects modeling approach.
Methods:
U was determined on residual aliquots of patients sampled for routine DPD deficiency testing in multiple centers. Stability before or after centrifugation (whole blood vs plasma) was explored, either at a cold (4°C) or at room temperature (RT), and at various times (minimum twice, from 30 min to 48 h post-sample). Data of U vs length of time until -20°C storage was analyzed longitudinally with a population approach where each sample was considered as an individual. Profiles at RT were fitted using a Gompertz model to capture the saturation of U increase. Profiles at 4°C were fitted using an exponential increase model. The difference between whole blood and plasma was implemented using different fixed effects parameters. Inter-individual variability was set as log-normal, and residual error as additive on log-transformed U. Parameter precision was obtained with the covariance step. Modeling was performed using the First-Order Conditional Estimation with Interaction algorithm in NONMEM 7.4. The maximum acceptable time limit for processing the samples was then assessed through simulations. Four cohorts (combining RT/4°C and blood/plasma) of 500 U vs time profiles were simulated, and the time period when U did not increase by more than 20% in 95% of the profiles was measured. This procedure was repeated 500 times by varying the population parameters to account for their uncertainty and to derive 90% confidence intervals (90%CI). Uncertainty distribution was implemented as multivariate Normal on fixed effects parameters and as inverse Wishart on random effects parameters. Simulations were performed in R 4.0, mainly with the mrgsolve (1.0.0) [3] and simpar (0.1.0) packages [4].
Results:
241 profiles (626 U values) from seven hospitals were available for analysis. The Gompertz model adequately fitted the saturation in U increase observed at RT, although the carrying capacity parameter (i.e., Umax) was fixed at 200 ng/mL. A residual error of 13.5% was obtained, in accordance with the assay error. At RT, simulations showed that U was stable for 0.93 h (90%CI 0.50-1.48 h) in whole blood, and for 0.95 h (0.43-2.96 h) in plasma. These time periods were higher if whole blood and plasma samples were stored at +4°C: respectively, 10.9 h (5.20-21.5 h) and 8.21h (3.49-19.8 h).
Conclusions:
A population approach successfully quantified the impact of pre-analytical steps in measuring U. It led to recommending that pre-analytical steps be carried out within a maximum of one hour at RT to adequately measure U and to accurately detect DPD deficiency. Storage at 4°C increases sample stability, but precise recommendations for time periods cannot be made due to the large 90%CI, explained by the data sparsity at later measurements. Non-linear mixed-effects modeling made it possible to analyze data collected from various centers using different collection times to assess stability. This methodology thus added statistical power and robustness to the conclusions. It also illustrates an original use of a population approach outside the domains of PK/PD and disease modeling. Finally, the R code used to prepare the simulation study dealing with uncertainty is available as a proper R package at https://github.com/FelicienLL/uncrtnty.
References:
[1] Méthodes de recherche d’un déficit en dihydropyrimidine deshydrogénase visant à prévenir certaines toxicités sévères associées aux traitements incluant une fluoropyrimidine (5-fluorouracile ou capécitabine). Haute Autorité de Santé. https://www.has-sante.fr/jcms/c_2891090/fr/methodes-de-recherche-d-un-deficit-en-dihydropyrimidine-deshydrogenase-visant-a-prevenir-certaines-toxicites-severes-associees-aux-traitements-incluant-une-fluoropyrimidine-5-fluorouracile-ou-capecitabine. Accessed March 28, 2022.
[2] EMA. EMA recommendations on DPD testing prior to treatment with fluorouracil, capecitabine, tegafur and flucytosine. European Medicines Agency. https://www.ema.europa.eu/en/news/ema-recommendations-dpd-testing-prior-treatment-fluorouracil-capecitabine-tegafur-flucytosine. Published April 30, 2020. Accessed March 28, 2022.
[3] Baron KT, Gillespie B, Margossian C, Pastoor D, Denney B, Singh D, et al. Mrgsolve: Simulate from ODE-Based Models.; 2020. https://CRAN.R-project.org/package=mrgsolve. Accessed March 28, 2022.
[4] Simpar. Metrum Research Group; 2021. https://github.com/metrumresearchgroup/simpar. Accessed March 29, 2022.
Reference: PAGE 30 (2022) Abstr 9962 [www.page-meeting.org/?abstract=9962]
Poster: Drug/Disease Modelling - Oncology