Multi-modal feature learning to prioritize ADCs with favorable half-life in mice
Clinical Snapshot
PICO Framework
| P — Population | C57BL/6 mice (experimental model for antibody-drug conjugates and monoclonal antibodies) |
| I — Intervention | Multi-modal machine learning framework using XGBoost models with antibody features, conjugation parameters, and linker-payload properties |
| C — Comparator | Different feature representation approaches (sequence descriptors vs AlphaFold structural properties vs protein language model embeddings) |
| O — Outcomes | Half-life prediction accuracy measured by Pearson correlation coefficient and Spearman rank correlation |
Bottom Line
This study presents a promising machine learning framework for predicting ADC half-life in mice using multi-modal features including antibody structure, conjugation parameters, and linker-payload properties. The models achieved strong predictive performance (r=0.76-0.79) and could potentially reduce the resource-intensive empirical testing currently required for ADC optimization. However, the work is limited to mouse models with uncertain translation to human pharmacokinetics, and the dataset size is relatively modest. While the approach shows promise for early-stage screening in pharmaceutical development, validation in human studies and expansion to other pharmacokinetic parameters would strengthen clinical applicability. The framework represents a valuable tool for ADC developers but should complement rather than replace traditional PK studies.
Key Findings
P Value: Not reported in abstract
Effect Size: Pearson correlation coefficient r = 0.76 (cross-validation), r = 0.79 (temporal validation)
Primary Outcome: Half-life prediction accuracy in C57BL/6 mice
Nnt Or Sensitivity: Spearman rank correlation ρ = 0.81 for ESM-2 protein language model
Confidence Interval: Not reported in abstract
Clinical Application
Computationally feasible for routine screening once models are implemented Relevant for Australian biotech companies developing ADCs, could inform TGA submission strategies by optimizing PK properties early in development Pharmaceutical companies developing ADCs, particularly for early-stage screening
Abstract
Favorable pharmacokinetic (PK) properties of antibody-drug conjugates (ADCs) enable higher systemic exposure and are associated with improved clinical efficacy, yet ADC PK optimization remains largely empirical and resource-intensive. We developed an in silico framework to predict half-lives of monoclonal antibodies (mAbs) and ADCs in C57BL/6 mice using multi-modal feature learning. To address structural complexity of ADCs, each compound was deconvoluted into three components: (1) antibody features represented using sequence descriptors, AlphaFold-based structural properties, or protein language model (PLM) embeddings; (2) conjugation-related features, including drug-to-antibody ratio and conjugation sites; and (3) physicochemical properties of linker-payloads. XGBoost models were trained using experimental half-life data from 118 mAbs and ADCs. Structure-based and PLM approaches achieved strong cross-validation (Pearson's r = 0.76), outperforming sequence-only models. In temporal validation of 32 mAbs and ADCs, the structure-based models demonstrated robust generalization (r = 0.79). PLM embeddings achieved comparable rank ordering (Spearman's ρ up to 0.81 for ESM-2), but lower quantitative accuracy relative to structure-based models. Overall, our study presents a scalable machine learning framework for early PK screening of mAbs and ADCs prior to synthesis and in vivo testing.
References
- 1.Zhu, X.-W., Jani, K., Gu, W., Guterman, M., Polyakov, V. R., Gerber, H.-P., & Rubas, W. (2026). Multi-modal feature learning to prioritize ADCs with favorable half-life in mice. mAbs. https://doi.org/10.1080/19420862.2026.2657646
Related Research
CPT: pharmacometrics & systems pharmacology
Structured Schemas for Provenance-Rich, LLM-Assisted QSP Model Calibration
2 Aug 2026
Journal of pharmacokinetics and pharmacodynamics
Diffusion models for virtual populations and pharmacometric simulations
1 Aug 2026
Journal of chemical information and modeling
HyperDC: A Non-Uniform Hypergraph Framework for Dual- and Higher-Order Drug Combination Recommendation Across Diverse Complex Diseases
29 July 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service