From Prognostic Signal to Drug Target
Aurora’s discovered CV-death proteins map to established and emerging druggable biology.
Risk Scores Are Cheap; Mechanism Is Valuable
Most prognostic models are black boxes: useful for stratification, silent on biology. Aurora’s penalised-Cox signatures are interpretable by construction, so each selected protein carries a direction of effect and a magnitude. When those proteins cluster in coherent, druggable pathways rather than scattering as noise, they become leads for target identification and for choosing which patients a mechanism is most likely to help.
The Discovered Proteins Map to Druggable Pathways
Top penalised-Cox coefficients, baseline + clinical + proteomics + PRS model. UK Biobank HF cohort.
The strongest prognostic proteins are not arbitrary. They sit on pathways with approved drugs or active clinical programmes, which means a prognostic hit can double as a mechanistic hypothesis and a ready-made patient-selection marker.
Pathway, Marker, and Therapeutic Relevance
| Pathway | Aurora markers | Therapeutic relevance |
|---|---|---|
| RAAS / renal perfusion | Renin, EDN1 | Target of ACE inhibitors, ARBs, MRAs, and endothelin antagonists. |
| Metabolic / systemic stress | ANGPTL4, GDF-15 | Active antibody and small-molecule programmes across cardiometabolic disease. |
| Fibrosis / remodelling | SPON1, HGF, SCGB3A1 | Anti-fibrotic and tissue-repair mechanisms of growing interest in HF. |
| Congestion / wall stress | NT-proBNP, ANGPT2 | Canonical readouts of cardiac load; pharmacodynamic and enrichment markers. |
Direction of effect is consistent with biology: RAAS activation, metabolic stress, and fibrosis markers all rise with worse prognosis, while protective markers (for example HPGDS) move the other way. That coherence is the signal that these are real mechanism, not overfit.
Why It Matters for Discovery and Development
A single Aurora run yields three assets at once: a validated prognostic model, a ranked set of mechanistic hypotheses, and the patient-selection markers to test them. Because the platform generalises to any indication by editing the cohort definition, the same approach can surface target and stratification hypotheses across a portfolio, turning population-scale data into a faster route from biology to a defensible development plan.
CEO, Hurdle
He/Him. Tom is CEO at Hurdle, a diagnostic-as-a-service company. Tom is a specialist in Epigenetics, Machine Learning, and Computational Biology.
LinkedIn →Working on something similar? We can talk through how Hurdle would approach it.
Talk to us →