A model-based extrapolation enabled labelling of emicizumab in haemophilia A paediatric patients <1 year old despite lack of clinical data
Sylvie Retout, Hans-Peter Grimm, Claire Petry, Christophe Schmitt, Nicolas Frey
Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center, Basel, Switzerland
Objectives: Emicizumab is a bispecific humanised monoclonal antibody (mAb) that binds activated factor (F) IXa and FX to activate FX, mimicking the function of missing or defective FVIIIa in patients with haemophilia A (PwHA) [1]. European Medicine Agency (EMA) among others approved a maintenance dose of 1.5 mg/kg/week from birth to adulthood in PwHA with inhibitors against FVIII. No data in PwHA
Methods: A popPK model was available for emicizumab, developed on a dataset of 191 PwHA, including 17 PwHA aged 2 to < 6 y, and only 4 PwHA aged 1 to < 2 y. The model included a body weight (BW) effect on the apparent clearance (CL/F) and volume, with estimated exponents of 0.891 (RSE=4.0%) and 1.02 (RSE=3.5%) respectively; CL/F was also impacted by albumin concentrations, increasing with decreasing albumin levels. For the PK extrapolation to PwHA < 1 y old, a known albumin variation with age [2] was implemented in the popPK model and the increase in BW with age was accounted for by using an actual covariate database from 693 infants tau,SS) at the dose of 1.5 mg/kg/week in PwHA 1 y old could be used to extrapolate CL/F to PwHA < 1 y old; (2) or assuming that CL/F follows an age-based maturation function combined with the classical fixed allometric exponent of 0.75 for BW effect as described in [4]. Simulated AUCtau,SS were then translated into bleeding event risk using an existing exposure–efficacy model [5].
PBPK simulations using SimCyp Version 15 [6] were also carried out to investigate whether a more mechanistic description of age-related differences could further improve the confidence in PK projections, especially in PwHA
Finally, the emicizumab popPK model was compared with published popPK models of other mAbs in children and infants.
Results: The popPK simulations with the maturation function predicted the lowest reduction of AUCtau,SS (27%) compared to PwHA >1 y old. At those levels, the efficacy of emicizumab is expected to be maintained, with exposures still at the plateau of effect.
Predicted CL/F with PBPK approach were 15%-20% higher than the ones predicted by the emicizumab popPK model for patients aged 3 months to 1 y and twofold higher for neonates. However, those predictions remained highly uncertain due to the lack of validation of the PBPK approach for mAbs in paediatrics, the absence of data for the ontogeny of key mechanisms (e.g., FcRn), and also the fact that the PBPK model over-predicted CL/F (up to +40%) in age ranges where patient data were available. Those PBPK predictions were however provided to the EMA, highlighting that the methodology was not robust enough yet to confidently extrapolate PK in infants.
Lastly, of the very few published popPK models of mAbs [8], only the palivizumab model [4] included an explicit age-based maturation function, whereas the others included only allometric scaling approach using BW. The palivizumab model was developed on a dataset of 1684 patients from birth to 2 y of age, and its use of a maturation function was therefore considered as the most robust extrapolation approach. The emicizumab CL/F was found very comparable to the current knowledge regarding CL/F of mAbs in paediatrics, especially to palivizumab’s, even for PwHA
Conclusions: By leveraging emicizumab models (i.e: popPK, PBPK and exposure-response), together with literature data, the proposed dosing of emicizumab for young infants was deemed appropriate although no data in PwHA < 1 y old were available. That full model-based extrapolation, together with a high unmet medical need and assumptions for disease and PK-pharmacodynamics similarities compared to PwHA >1 y old, was considered acceptable and led to the approval of emicizumab in PwHA with FVIII inhibitors in all age groups in the European Union countries.
References:
[1] Shima M et al. N Engl J Med. 2016;374:2044-53.
[2] Johnson TN et al. Clin Pharmacokinet. 2006;45:931-56.
[3] Jorga K et al. Clin Pharmacol Ther. 2016;100:761–769.
[4] Robbie GJ et al. Antimicrob Agents Chemother. 2012;56:4927–4936.
[5] Jonsson F et al. PAGE 27 (2018) Abstr 8451 [www.page-meeting.org/?abstract=8451]
[6] Simcyp Version 15.1, Simcyp Limited. Sim-Healthy Volunteer population Library. July, 2016.
[7] Li L et al. AAPS J. 2014;16(5):1097-109.
[8] Barrett JS et al. Clin Pharmacol Ther. 2012 ;92: 40-49.
[9] Fuhrmann S et al PAGE 26 (2017) Abstr 7143 [www.page-meeting.org/?abstract=7143]
