Presented at AACR Annual Meeting 2026
Jeannette Fuchs1, Diana Rocha Gomes2, Christophe Mas1, Eric Durandau3, Antoine Attinger2, Anna Pokorska-Bocci1
Debiopharm International S.A., Lausanne, Switzerland │ 1Personalised Medicine │ 2Translational Pharmacology │ 3Data&Bioinformatics
Summary
Assessment of target expression for mono- and bi-specific antibodies (bs)Abs and antibody-drug conjugates (bs)ADCs can be challenging due to variable correlation between bulk RNASeq, proteomics and immunohistochemistry (IHC) data. While IHC is sufficient to address single target expression levels at single cell level, multiplexed proteomics is required to assess spatial distribution and co-expression of multiple targets for bsAb and bsADC use.
To address these challenges, we performed an in-depth characterization of HER3 and an undisclosed target (Protein X) across multiple omics platforms in patient-derived xenografts (PDXs) for a bsADC targeting HER3 and Protein X.
HER3 mRNA indeed correlates only moderately with Mass Spectrometry-derived (MS) HER3 protein intensities (Figure A). Similarly, HER3 protein intensities do not properly reflect IHC-derived HER3 protein levels (Figure B) besides not capturing the details of HER3 heterogeneity as IHC does (Figure C). Notably, none of these analytes sufficiently predicts response of in vivo PDXs (S=sensitive, R=resistant) to the HER3/Protein X bsADC.
Finally, the proportion of HER3- and/or Protein X-positive cells was assessed by multiplexed proteomics to investigate the co-expression pattern of both targets and their effect on response to HER3/Protein X bsADC treatment in vivo (Figure D, E). While co-expression by itself was non-predictive, the 3D multifactorial combination of HER3 and Protein X expression levels, together with the proportion of coexpressing cells, show a tendency towards HER3/Protein X bsADC resistance being associated with higher H-scores of both targets (Figure F). However, another PDX model with similar features (#2-S) showed sensitivity to the same treatment.
A larger cohort of PDX models with a larger target expression range needs to be analysed to build more reliable associations between target expression and in vivo response.