Predictive and prognostic biomarker strategy
A signature that separates groups in a discovery cohort is not yet a biomarker. The distance between those two things is where most programmes lose two years.
01The distinction
Predictive and prognostic are not the same thing
A prognostic biomarker tells you how a patient’s disease is likely to behave regardless of treatment. A predictive biomarker tells you how they are likely to respond to a specific intervention. They require different cohorts, different statistical designs, and they support entirely different commercial claims.
Conflating them is common and expensive. A programme that runs a single-arm discovery study and then claims treatment selection utility has not generated the evidence it thinks it has, and will be told so at the worst possible moment.
Candidate triage
Most discovery-stage signature lists contain more candidates than a programme can afford to develop. Triage weighs effect size against measurement feasibility, biological plausibility, existing intellectual property, and whether the analyte can be measured in a specimen type that will realistically be available in the clinic.
02The path to an assay
From signature to something measurable
- Analytical feasibility. Can this be measured reproducibly, in an available specimen, with a platform a clinical laboratory would adopt?
- Specimen and cohort strategy. What samples, from which patients, with what clinical annotation, and how many — before enrolment starts, not after.
- Assay format selection. IHC, multiplex immunoassay, qPCR or sequencing-based, chosen against the intended use rather than the technology to hand.
- Regulatory pathway alignment. Mapping to FDA IVD and companion diagnostic routes early, so the evidence generated is the evidence that will be asked for.
- Validation planning. Independent cohorts, pre-specified cut-points, and analytical validation distinguished from clinical validation.
03Track record
Signatures taken toward the clinic
Vasodilator protein expression in pulmonary arterial hypertension, with blood-pressure-regulating signature genes identified for potential in-vitro diagnostic development.
Genomic and proteomic biomarker discovery in PAH patient samples, including the role of O-glycosylation and GALNT-mediated modification in disease development.
Preventive biomarker discovery for breast cancer from microarray datasets across cell lines and surgical specimens, alongside identification of the molecular target of an antiprogestin tested as a chemoprevention agent.
Genome-wide analysis identifying IL-18 and FUCA2 as associated with diastolic function in sickle cell disease.
—Also
Related capabilities
Get started
Have a signature and no clear route to an assay?
Bring the discovery data and the intended clinical use. The gap between them is usually mappable in a single conversation.