Biomarker

A biomarker is a defined characteristic measured as an indicator of normal or pathological biology, or response to intervention. FDA-NIH BEST Glossary (2016) categories: (1) Diagnostic — identifies or confirms a condition (HbA1c for diabetes); (2) Monitoring — tracks disease progression or treatment response over time (PSA post-prostatectomy); (3) Pharmacodynamic/Response — shows biological effect of a drug at the target (BCR-ABL inhibition by imatinib); (4) Predictive — identifies patients likely to respond to a treatment; (5) Prognostic — indicates likely clinical course regardless of treatment; (6) Safety — signals adverse effects (ALT for hepatotoxicity); (7) Susceptibility/Risk — indicates increased likelihood of disease. See pharmacogenomics guide for genetic biomarkers.

Predictive biomarker: identifies patients likely to respond to a specific treatment. Examples: KRAS WT → respond to anti-EGFR (cetuximab, panitumumab) in CRC; HER2 overexpression → trastuzumab benefit in breast/gastric cancer; BRCA1/2 mutation → PARP inhibitor (olaparib) benefit; PD-L1 TPS ≥50% → first-line pembrolizumab benefit in NSCLC. Prognostic biomarker: indicates likely outcome regardless of treatment. Examples: Ki-67 in breast cancer (high = worse prognosis regardless of chemotherapy), TP53 mutation in CLL (poor prognosis), Oncotype DX score (25-gene recurrence score — both prognostic and predictive in some contexts). Many biomarkers have both predictive and prognostic roles — distinction requires randomised trial data with a biomarker-negative control arm. See clinical trial guide.

A PD biomarker demonstrates on-target biological activity of a drug. Used in Phase I/II to identify the biologically active dose and confirm mechanism. Examples: BCR-ABL inhibition (imatinib): residual kinase activity in CML cells inversely correlates with disease response; PSA decline rate: PD biomarker of AR pathway inhibition (enzalutamide/abiraterone) in prostate cancer; Tumour mutational burden (TMB): PD/predictive biomarker for checkpoint inhibitor response; ctDNA clearance: PD biomarker for early response to targeted therapies; HbA1c change from baseline: PD biomarker of antidiabetic drug efficacy. Contrast with: therapeutic index (safety/efficacy ratio in a population), surrogate endpoint (PD biomarker used as a substitute for clinical endpoint in regulatory approval). See clinical trial guide for surrogate endpoint validation requirements.

A surrogate endpoint is a biomarker validated as a substitute for a clinical outcome in a regulatory trial. FDA accelerated approval pathway uses "reasonably likely to predict" clinical benefit — stronger than a plain biomarker, weaker than full surrogate validation. Validated surrogates: LDL-C (→ CV events per statin trials), HIV RNA viral load (→ AIDS/mortality), CD4 count (HIV), HbA1c (→ diabetic complications per DCCT/UKPDS). Accelerated approval examples: tumour ORR (objective response rate) as surrogate for OS in oncology — requires post-market confirmatory trials. Surrogate failure: flecainide reduced PVCs (VT suppression as surrogate) but increased mortality in CAST trial. Biomarker qualification: FDA Biomarker Qualification Program formalises the evidentiary standard. See clinical trial guide for NNT interpretation and surrogate endpoint limitations.

FDA Biomarker Qualification Programme (BQP): formalises a biomarker's acceptability for a defined context of use (COU) in all drug development programs — a public good, not sponsor-specific. Three qualification stages: (1) Initiation — FDA accepts the COU for review; (2) Consultation and advice — FDA reviews evidence package; (3) Full qualification — published qualification decision. Qualified biomarkers: total kidney volume (TKV) by MRI for ADPKD progression enrichment; alanine aminotransferase (ALT) for drug-induced liver injury (safety biomarker — already standard but formally qualified). FDA BQP ↗. CDISC and BEST glossary standardise biomarker terminology. Related: drug-receptor theory, clinical trial guide.