In this study, we introduce a novel Bayesian meta-analysis framework to support the definition of a disease-agnostic Quality Target Product Profile (QTPP) for RNA-mAb constructs. The QTPP refers to a prospective summary of product characteristics critical for quality, safety, and efficacy. While traditionally QTPPs are defined qualitatively and product-specifically at late-stage in the product lifecycle, we propose a novel platform-level, probabilistic and computational approach that leverages prior knowledge and real-world data from both RNA vaccine and recombinant mAb fields.
Our dataset integrates pharmacokinetics (PK) and efficacy data from 13 RNA-mAb constructs, including four non-human primate studies and two clinical datasets, with matched comparator data from recombinant mAbs. First, we assess bioequivalence between RNA-mAb and recombinant mAb based on preclinical serum concentration data following intravenous administration. A fivefold reduction in RNA-mAb serum concentration is needed to match area under the curve (AUC) levels of recombinant mAb, suggesting advantageous PK characteristics. Significantly higher potency of RNA-mAb is also revealed across constructs. Between-study heterogeneity is also modelled to capture the effects of sequence design, chemical modifications, and LNP formulations.
Bayesian methods are further used to analyse missing data, propagate uncertainty in PK parameters (maximum serum concentration, clearance and half-life), and test for linearity between RNA dose and antibody serum concentration. Bayesian imputation techniques allow inference of incomplete time courses and support cross-study harmonization. Alternative administration routes are eventually compared under a unified posterior framework, highlighting the potential benefits of local administration for cancer applications. This enables an early readout of formulation and dose constraints for preclinical-to-clinical translation.
Importantly, we introduce a novel backward Bayesian inference method, where known human PK targets for more than 80 currently approved recombinant mAbs are used to estimate the required PK properties in preclinical models for RNA-mAb. This enables rational early-stage screening criteria during RNA-mAb discovery, potentially guiding construct selection, formulation, and escalation planning for different therapeutic applications.
Secondly, quality attributes at both the mAb expression level and RNA-LNP input level is assessed. Critically, RNA-encoded mAb glycosylation is modelled and compared with that of naturally occurring IgG and IgA, as well as recombinant mAbs, revealing distinct profiles and novel risks. The immunogenicity risks from high and repeated dosing, and the role of double stranded RNA (dsRNA) impurities are further analysed.
Overall, this is the first application of bayesian meta-analysis to systematically define QTPP for a novel therapeutic platform. This approach identifies data gaps, quantifies uncertainty, and supports decision-making across development. By enabling early-stage benchmarking and harmonisation across RNA-mAb constructs, the framework also provides a transparent, adaptable, and quantitative tool to guide future discovery and accelerate clinical translation.
Our findings also advance the comparison of RNA-mAbs and recombinant mAbs, and contribute to understanding whether RNA-based expression systems can eventually complement or replace recombinant antibody platforms, moving toward a more affordable and scalable alternative in the biopharmaceutical landscape.