Alicja Puszkiel (1,2), Stéphane Goutagny (1,3), Guilhem Bousquet (4,5), Jeanick Stocco (6), Philippe Decq (3), Lucie Chevillard (1), Xavier Declèves (1,2)
(1) Inserm UMR-S1144, Faculty of Pharmacy, Université Paris Cité, Paris, France (2) Laboratory of Pharmacology and Toxicology, Cochin University Hospital, Assistance Publique Hôpitaux de Paris, Paris, France (3) Department of Neurosurgery, Beaujon University Hospital, Assistance Publique Hôpitaux de Paris, Paris, France (4) Oncology Department, Avicenne University Hospital, Assistance Publique Hôpitaux de Paris, Paris, France (5) Inserm UMR-S942, Université Paris Cité, Paris, France (6) Department of Pharmacy, Beaujon Hospital, Assistance Publique Hôpitaux de Paris, Paris, France
Objectives: Trastuzumab, an anti-HER2 monoclonal antibody (mAb), is used as standard treatment in HER2-positive breast cancer patients at the recommended dose of 6 mg/kg administered intravenously (IV) every three weeks (q3w). About 5% of breast cancer patients develop leptomeningeal carcinomatosis (LC) consisting of metastatic invasion of subarachnoid space in the central nervous system (CNS) [1]. The distribution of mAbs trough blood-brain barrier into CNS is limited because of their large molecular weight [2]. Therefore, intrathecal (IT) administration of mAbs in patients with LC was proposed in order to achieve therapeutic concentrations in the cerebrospinal fluid (CSF). As this practice is still rare, IT doses and dosing regimens remain currently empirical. The objectives of this study were to:
- describe pharmacokinetics (PK) of trastuzumab in human plasma and ventricular CSF after simultaneous IT and IV administration using a minimal physiologically-based pharmacokinetic model (PBPK),
- perform simulations of alternative dosing regimens of trastuzumab to achieve therapeutic concentrations in CSF.
Methods: A previously published minimal PBPK model for mAbs disposition consisting of four compartments (tight and leaky tissues, plasma and lymph) was used [3]. An additional compartment for ventricular CSF was added in the model assuming that trastuzumab after IT administration in the left ventricle is cleared to plasma through CSF flow rate multiplied by lymphatic reflection coefficient. The distribution of mAb from plasma to CSF was described by CSF flow rate multiplied by blood-brain barrier reflection coefficient. The system-specific parameters were fixed to physiological or previously published values from Cao et al. and Chang et al. [3, 4]. Plasma trastuzumab clearance (CLp) was estimated based on rich PK data from a patient initially treated with IV doses of 6 mg/kg every three weeks (q3w) and IT doses of 50 mg q3w, then 50 mg q1w followed by 50 mg every three days (q3d) for a total treatment time of 261 days [5]. A total of 61 CSF and 39 plasma concentrations were collected. Secondly, the model was fitted to data from a second patient who received IT doses of 150 mg q1w and IV doses of 6 mg/kg q2w, for which 21 trastuzumab CSF and 24 plasma concentrations were available. The comparison between observed and model-predicted concentrations was evaluated using median absolute prediction error (MPE) [6].
The final model was used to simulate alternative IT dosing regimens. For the steady-state CSF concentration, the target to be reached was above 60 mg/L, which is mean plasma steady-state trough concentration [Css,min] observed in patients with HER2-positive cancer treated with trastuzumab at recommended IV dose [7]. Trastuzumab CSF concentrations after repeated IT doses of 100 and 150 mg q1w and q3d and continuous perfusion at 0.4, 0.8 and 1.0 mg/h rates were compared in terms of achievement of this target. Modeling and simulations were performed in NONMEM (version 7.5, Icon plc).
Results: The minimal PBPK model showed adequate description of trastuzumab PK in plasma and CSF with MPE of 29% [interquartile range, IQR: 11 – 43] for plasma and of 39% [IQR: 22 – 67] for CSF data in the patient with rich PK sampling. The estimated trastuzumab CLp was 0.324 L/day (relative standard error, RSE = 13%) that is consistent with previously reported values [8]. The PK data of the second patient showed MPE of 23% [IQR: 6.3 – 34] and 76% [IQR: 24 – 86] for plasma and CSF, respectively. Trastuzumab showed fast clearance from CSF to plasma with Cmin,ss of 0.56 and 0.85 mg/L for 100 and 150 mg q1w, respectively. Repeated dosing of 100 and 150 mg q3d resulted in Cmin,ss of 10.3 and 15.4 mg/L, respectively. Trastuzumab CSF steady-state concentrations [Css] of 25.4, 51.7 and 64.6 mg/L were achieved after continuous perfusion at 0.4, 0.8 and 1.0 mg/h, respectively.
Conclusions: We successfully developed a minimal PBPK model to describe trastuzumab PK in human plasma and CSF. The differences in trastuzumab PK between the two patients suggest complexity of mAbs PK in the CNS possibly influenced by disease progression in patients with LC (inflammation, target-mediated drug disposition…). The simulations showed that a continuous intraventricular perfusion at 1.0 mg/h rate could allow to achieve similar Css to that observed in plasma after repeated IV dosing and for which clinical efficacy was observed.
References:
[1] Gauthier et al., Ann Oncol. 2010 Nov;21(11):2183-2187.
[2] Ovacik & Lin, Clin Transl Sci. 2018 Nov;11(6):540-552.
[3] Cao et al., J Pharmacokinet Pharmacodyn. 2013 Oct;40(5):597-607.
[4] Chang et al., J Pharmacokinet Pharmacodyn. 2019 Aug;46(4):319-338.
[5] Bousquet et al., J Clin Oncol. 2016 Jun 1;34(16):e151-5.JCO 2016
[6] Sheiner & Beal, J Pharmacokinet Biopharm. 1981 Aug;9(4):503-12
[7] Baselga et al., Ann Oncol. 2001;12 Suppl 1:S49-55.
[8] Bruno et al., Cancer Chemother Pharmacol. 2005 Oct;56(4):361-9.
Reference: PAGE 30 (2022) Abstr 10082 [www.page-meeting.org/?abstract=10082]
Poster: Drug/Disease Modelling - Absorption & PBPK