Predictive Biomarkers and Therapeutic Targets in Cancer

A predictive biomarker is a molecular feature of a tumor that indicates the likely benefit or lack of benefit from a particular therapy, distinguishing it from a prognostic biomarker, which reflects outcome regardless of treatment. Many predictive biomarkers correspond to therapeutic targets - driver alterations whose protein products can be directly inhibited - and identifying them is a central purpose of tumor molecular profiling.

Definition

A predictive biomarker is a molecular or cellular characteristic measured in tumor material that is associated with the likelihood of response or resistance to a specific therapy; a therapeutic target is the molecular alteration or pathway whose activity a treatment is designed to inhibit, and an actionable alteration is one linked to such a therapy by adequate evidence.

Scope

This entry covers the definition of predictive biomarkers, how they differ from prognostic markers, the concept of an actionable or targetable alteration, the assay types used to measure such biomarkers, and how evidence for clinical significance is graded. It is a conceptual and methodological reference and does not recommend therapies or testing for any individual.

Core questions

• What distinguishes a predictive biomarker from a prognostic biomarker?
• What makes a driver alteration a tractable therapeutic target?
• Which assay formats are used to measure predictive biomarkers, and what governs their validity?
• How is the strength of evidence linking a biomarker to a therapy graded and reported?

Key concepts

• Predictive versus prognostic biomarker
• Actionable alteration
• Therapeutic target
• Companion biomarker testing
• Resistance mechanisms
• Oncogene dependence
• Evidence tiers for clinical significance
• Assay analytic validity

Mechanisms

Predictive biomarkers work because a tumor that depends on a particular altered protein or pathway is selectively vulnerable to its inhibition - the principle of oncogene dependence. When an activating alteration both drives the tumor and can be measured, it serves simultaneously as a target and as a predictor of response, as with activating EGFR mutations and EGFR inhibitors in lung cancer. Biomarkers are measured by methods matched to the alteration type, including sequencing for mutations and fusions, in situ hybridization for amplifications and rearrangements, and immunohistochemistry for protein expression such as programmed death-ligand 1. The clinical significance of each finding is then graded against curated evidence in standardized tiers, and resistance can emerge when tumors acquire further alterations that bypass or restore the targeted pathway.

Clinical relevance

Predictive biomarkers are the link between molecular profiling and precision oncology, defining the subgroups in which targeted and immune therapies are studied. This entry explains the concepts, assays, and evidence frameworks involved; it characterizes how predictive evidence is generated and classified and is not a basis for selecting therapy or testing for any individual.

Epidemiology

The prevalence of targetable alterations varies widely by cancer type, with some defined predominantly by one or a few actionable drivers and others by a long tail of rare alterations. Genome-scale analyses show that oncogenic alterations across cancers converge on a limited set of signaling pathways, which frames how often a profiled tumor is expected to harbor a potentially targetable feature.

History

The predictive-biomarker paradigm took shape as specific tumor alterations were tied to response to matched therapies, beginning with examples such as activating EGFR mutations and EGFR inhibitors in lung cancer in the mid-2000s. As more targeted and immune therapies emerged, biomarker testing - including protein assays such as programmed death-ligand 1 immunohistochemistry - became integral to oncologic pathology, and professional standards were developed to grade and report the clinical significance of detected alterations.

Debates

How should predictive biomarkers be measured and thresholds set?

Different assays for the same biomarker can give discordant results, and cut-points for protein-based markers such as programmed death-ligand 1 vary across assays and contexts, making harmonization and threshold selection an ongoing methodological challenge.

Seminal works

• lynch-2004
• buttner-2017

Frequently asked questions

What is the difference between a predictive and a prognostic biomarker?

A predictive biomarker indicates whether a tumor is likely to benefit from a specific therapy, whereas a prognostic biomarker reflects the expected course of disease regardless of treatment; a single marker can sometimes carry both kinds of information.

What makes an alteration 'actionable'?

An alteration is considered actionable when adequate evidence links it to a therapy that targets the affected protein or pathway, so that detecting it can inform the rationale for that treatment; the strength of this link is graded in standardized evidence tiers.