Evaluation of sNfL as a Biomarker for Paclitaxel-Induced Peripheral Neurotoxicity Through an Integrated PKPD Model

Abstract Background Serum neurofilament light chain (sNfL), a biomarker of axonal damage, has shown promise in clinical studies for monitoring paclitaxel-induced peripheral neurotoxicity (PIPN). The latter involves pathological changes in PIPN sites such as the dorsal root ganglia, peripheral nerves, and brain. However, the mechanistic link between paclitaxel and NfL concentrations in these tissues remains poorly understood, necessitating preclinical investigation. Methods We developed a semi-mechanistic pharmacokinetic-pharmacodynamic model to characterize: (i) total and unbound paclitaxel concentrations in plasma, as well as in extracellular and intracellular compartments of PIPN sites; (ii) paclitaxel–tubulin complex formation; and (iii) NfL kinetics. The model was built using de novo-generated and previously reported data from rodents. Results Plasma pharmacokinetics of paclitaxel was captured using a two-compartment model, including Cremophor EL trapping and nonlinear tissue distribution. Paclitaxel pharmacokinetics in PIPN sites incorporated paclitaxel transport across the blood-to-PIPN sites barriers and paclitaxel–tubulin binding, described by capacity-limited kinetics with increased tubulin binding upon repeated plasma exposure. NfL kinetics in serum and cerebrospinal fluid were described using turnover models, with NfL leakage driven by paclitaxel–tubulin complex formation in PIPN sites. The model robustly predicted paclitaxel exposure across multiple doses and studies. While NfL predictions aligned with single-dose data, the model slightly underpredicted sNfL levels in an external validation dataset after repeated dosing of paclitaxel at 15 mg/kg, suggesting additional mechanisms may be involved. Conclusions Overall, the model successfully described the relationship between paclitaxel exposure and sNfL kinetics, offering a model-based framework for translational studies.