META-WWOX: AI/XAI-Enabled Biomarkers of Metabolic Disease Progression

We are developing a collaborative project to identify and validate multimodal biomarkers of metabolic disease progression and treatment response by integrating longitudinal clinical data, RWD and multi-omic profiles using AI/XAI. We seek partners with large longitudinal cohorts, biobanks/RWD and complementary epigenomic, metabolomic or proteomic expertise.

Project concept: Type 2 diabetes and metabolic syndrome are highly heterogeneous, with variable progression, complications and treatment response. We propose that clinically relevant trajectories can be predicted by combining longitudinal clinical data with molecular signatures reflecting regulatory and metabolic states.

A key mechanistic element is WWOX as a regulatory hub within transcriptomic, epigenomic and metabolic networks, rather than a standalone biomarker. We will test whether WWOX-associated regulatory states contribute to patterns linked to progression and treatment response.

Our existing foundation: Through collaboration with an Internal Medicine Department, we have access to clinical data and biological material from patients with diabetes, metabolic syndrome and related disorders, including laboratory parameters, treatment histories, outcomes and follow-up. We can also prospectively collect blood samples (serum, plasma, PBMCs) for molecular profiling, providing a real-world translational cohort for biomarker discovery. We also have experience in RWE studies and integration of heterogeneous clinical datasets with outcome analyses (e.g. PMID: 41238080).

Our research background combines transcriptomics, systems biology, predictive modelling and WWOX-related regulatory networks. Previous work demonstrated WWOX-associated effects on signalling and differentiation (PMID: 31543760) and TGFα-EGFR regulatory rewiring (PMID: 35290621). We have also developed network-based and transcriptomic biomarker approaches (PMID: 33273537; 37176106).

Multi-omic progression atlas for disease monitoring: In a characterised subcohort, we propose integrating:

• transcriptomics;

• epigenomics and promoter regulation;

• metabolomics and/or proteomics;

• blood-derived molecular profiles;

• clinical and laboratory data;

• treatment history and response;

• comorbidities, complications and imaging features where available.

WWOX will be evaluated through expression, transcript usage, promoter activity, epigenetic regulation and downstream network activity within an unbiased multi-omic framework.

AI/XAI and biomarker development: Molecular and longitudinal clinical data will be integrated using established machine-learning and AI/XAI approaches. Explainable models will identify key drivers of prediction and support development of clinically interpretable biomarker signatures. Main objectives include prediction of:

• progression to advanced metabolic disease;

• deterioration of metabolic control;

• response or non-response to treatment;

• patient subgroups with distinct trajectories.

Our expertise includes feature selection, longitudinal and survival modelling, subgroup discovery, internal validation, discrimination, calibration and biomarker optimisation (PMID: 31319941).

Our capabilities:

• biomarker discovery and statistical validation;

• transcriptomic and multi-omic analysis;

• systems biology, GSEA, WGCNA and network analysis;

• integration of molecular, clinical and RWD data;

• machine learning and AI/XAI modelling;

• longitudinal outcome modelling and patient stratification;

• biostatistics and clinical study methodology;

• processing of blood-derived patient samples;

• clinical utility assessment, HTA and health economics.

Partnership sought:

• large longitudinal cohorts, biobanks and RWD infrastructures with repeated measurements and long-term outcomes;

• epigenomics and promoter-regulation expertise;

• metabolomics and/or proteomics;

• independent validation cohorts;

• clinical expertise in diabetes and metabolic disorders.

Expected outcome: The project aims to deliver a clinically validated multimodal predictive biomarker integrating molecular and longitudinal clinical data to predict metabolic disease progression and treatment response. We aim to lead multi-omic biomarker discovery, systems-level integration, AI/XAI interpretation and statistical validation, while partners contribute large-scale longitudinal/RWD resources and complementary omics technologies.