In Silico Development of a Retatrutide Injectable Formulation for a Multidose Cartridge

Abstract

This study developed an evidence-weighted and probabilistic framework for selecting a candidate liquid formulation of retatrutide at 10 mg/mL in a 3 mL multidose cartridge. Four formulation platforms were ranked by a prespecified multi-criteria decision analysis. Uncertainty in weights and scores was propagated in 100,000 Monte Carlo iterations. The selected composition contained retatrutide 10.0 mg/mL, tromethamine 1.211 mg/mL, D-mannitol 40.0 mg/mL, and phenol 5.0 mg/mL at pH 7.50 ± 0.10. A first-order Arrhenius scenario model then propagated uncertainty in the degradation rate, activation energy, degradant yield, high-molecular-weight species yield, and activity relationship. The Tris-mannitol-phenol platform achieved a median decision score of 4.12 out of 5 and ranked first in 99.997% of simulations. At 5 °C and month 12, the predicted median assay was 98.77% of T0 with a 95% uncertainty interval of 97.83-99.31%; total degradation products were 1.20% [0.83-1.86%]. At 25 °C and month 6, the corresponding estimates were 92.39% [89.30-94.61%] and 5.28% [3.62-7.61%]. The probability of maintaining at least 98% assay at 5 °C through 12 months was 95.3%. These are conditional computational outputs rather than observations. The analysis identifies a rational lead composition and quantifies which assumptions require experimental confirmation.

The authors' abstract, as published at the source. Preprints.org, 2026 · DOI ↗