Ana-Marija Grisic (1), Yulia Vugmeyster (2), Brigitte Brockhaus (1), Mary Ruisi* (2), Haiqing Dai (2), Akash Khandelwal (1)
(1) Merck KGaA, Darmstadt, Germany, (2) EMD Serono Research & Development Institute, Inc., Billerica, MA, USA; an affiliate of Merck KGaA, Darmstadt, Germany, *Affiliation at the time the study was conducted.
Objectives: Avelumab is a fully human IgG1 monoclonal antibody that mediates its anticancer activity by binding to programmed death-ligand 1 (PD-L1) expressed on tumor cells. By blocking the interaction of PD-L1 with PD-1 on T cells, immune responses are restored and antibody-dependent cell-mediated cytotoxicity is induced. The pharmacokinetics (PK) of avelumab has been characterized previously in adult patients.1 The objective of this study was to characterize avelumab exposure in pediatric patients.
Methods: The data analyzed originated from a multicenter, open-label, international, phase 1 clinical trial (ClinicalTrials.gov NCT03451825) in pediatric patients (n=21) with refractory or relapsed malignant solid tumors or lymphoma. Initial patients (n=6) received avelumab at 10 mg/kg every 2 weeks (q2w; initially approved dose in the adult population). After the initial dose of 10 mg/kg was confirmed safe, the dose was escalated to 20 mg/kg q2w (n=15) to achieve an exposure similar to that of adults. PK samples comprised Ctrough at weeks 1, 3, 5, 9, 13, 15, 25, 37, and 49 and every 12 weeks thereafter, and concentration at the end of infusion at weeks 1, 5, 13, and 37, with additional samples in cycle 1 at 48-96 h in patients <10 kg and at 3 h and 48-96 h in patients ≥10 kg. Data were analyzed using a nonlinear mixed-effects modeling approach. A 2-compartment model with time-varying clearance (CL) pre-established in the adult population was used. Different approaches for characterizing exposure were investigated, including predictions of pediatric exposure from the adult PK model, estimation of PK parameters from pediatric PK data only and pooled pediatric and adult data, and estimation using the frequentist prior approach. NONMEM (v7.3.0), PsN (v4.4.8), Pirana, R, and RStudio were used for model development and pre- and post-processing.
Results: In total, 151 PK observations (46 for 10 mg/kg q2w and 105 for 20 mg/kg q2w) were available for analysis. Patients had a median baseline age of 12 years (range, 3.1-17.1) and median baseline body weight of 37.3 kg (range, 13.4-78.65). Data from the pediatric trial alone were not informative enough to support the estimation of the parameters of the model with the structure pre-established in adults. Due to much higher informativeness of adult data, in the case of pooling of adult and pediatric data, parameter estimates were shrinking toward the estimates from the adult model. The frequentist prior approach ($PRIOR in NONMEM), whereby estimation of the parameters related to time change in CL, peripheral volume of distribution (V2), and intercompartmental exchange rate (Q) were informed by prior estimates from the adult model, demonstrated potential in overcoming the limitations of other approaches. For fixed-effect parameters, RSE was ≤25%; for random-effect parameters, it was ≤45%. Visual predictive check demonstrated good predictive performance of the model. Baseline CL and central volume of distribution (V1) for the typical pediatric patient with weight of 37 kg (pediatric median) were estimated to be 0.0185 L/h (RSE 7%) and 1.98 L (RSE 4%), respectively. The parameters informed by prior information were close to the adult values, with V2 and Q of 0.9 L and 0.0325 L/h, respectively, and slightly lower maximal decrease in CL (2%). The effect of body weight was higher in pediatric patients, with an estimated power exponent of 0.89 and 0.63 for the effect on CL and V1, respectively, compared with 0.545 and 0.475 found in adults; these estimates are similar to those previously reported for another drug from the same class2.
Conclusions: In this study, the frequentist prior approach was found suitable for overcoming limitations of sparse pediatric data analysis arising from the phase 1 trial. In this pediatric population the effect of body weight on CL of avelumab was found to be stronger than values reported for adults. The underlying reasons for this difference (e.g. due to differences in tumor types, disease burden, adults vs. pediatric) remain to be investigated.
References:
[1] Wilkins JJ, et al. CPT Pharmacometrics Syst Pharmacol. 2019;8:415427.
[2] Shemesh CS, et al. J Immunother Cancer. 2019;7:314.
Reference: PAGE 29 (2021) Abstr 9600 [www.page-meeting.org/?abstract=9600]
Poster: Drug/Disease Modelling - Paediatrics