Effective and objective evaluation of biological processes are essential components of evaluating patients, particularly in the context of drug development. Biomarkers are essential components in this regard.

However, in order to translate the results provided by biomarkers, regulators and clinical trialists alike require unambiguous interpretation and communication.

To address this, the FDA-NIH Joint Leadership Council established a framework and vocabulary for how biomarkers can be interpreted, including their use and their significance: the Biomarkers, EndpointS, and other Tools (BEST) glossary.

BEST is developed to accurately distinguish between different biomarker properties, in the context of biomedical research, clinical practice, drug development and regulatory considerations and use. The most frequently used classes of biomarkers in relation to drug development include diagnostic, prognostic, monitoring, response, and predictive biomarkers.


Figure 1. Diagnostic, prognostic, monitoring, and response biomarkers.

A diagnostic biomarker can accurately separate healthy conditions from those affected by disease.

A prognostic biomarker can identify patients that are at a higher likelihood of experiencing disease progression or adverse outcome.

A monitoring biomarker can be used to track development of disease over time or monitor response to treatment.

A response biomarker can be used to pharmacodynamically study the response to treatment.

Diagnostic biomarkers

A diagnostic biomarker can be used to detect or confirm pathological changes resulting in disease, or to identify subtypes of disease within a patient population (figure 1a). An accurate diagnosis based on a biomarker requires thorough testing and validation and comes down to clinical sensitivity (the fraction of patients testing positive who has the disease) and specificity (the fraction of patients testing negative who does not have the disease).

In addition to that, a diagnostic biomarker needs to have a high positive predictive value (PPV, the patients who test positive and have the disease) and negative predictive value (NPV, the patients who test negative and do not have the disease). In general, a diagnostic biomarker will have to be carefully evaluated in the intent-to-diagnose population, weighing in sensitivity, specificity, PPV and NPV, to ensure proper separation of populations and minimization of false positive- and negative results.

Prognostic biomarkers

A prognostic biomarker assesses the decreased or increased likelihood of a clinical event occurring (figure 1b). They are often used at a defined baseline (i.e., enrollment or date of starting treatment) in a clinical context to estimate the risk of disease progression or occurrence of severe adverse events. Importantly, a prognostic biomarker is associated with the likelihood of future clinical events occurring disregarding therapeutic intervention.

Prognostic biomarkers are often used as eligibility criteria in the context of drug development and clinical trials, as they may identify patients more susceptible to benefit from the therapy due to increased likelihood of disease progression and adverse events. In fact, a substantial number of clinician studies fail because of lack of disease progression, thus making it impossible to detect a therapeutic benefit. This provides a mean for increasing statistical power in clinical trials, which often include a time-to-event or event rate as outcome measures.

Monitoring biomarkers

A monitoring biomarker is measured repeatedly throughout a trial period or in assessment of disease in patients (figure 1c) and is usually used to monitor the change over time and the scale of change of a physiological parameter (i.e., disease activity). The importance of monitoring disease state becomes evident when disease activity leads to disease progression, which may augment the need for a different treatment strategy.

If patients are treated, changes in the monitoring biomarker may also mean that the biomarker can be used for monitoring the pharmacodynamic effects of the therapeutic intervention. Monitoring biomarkers can serve many purposes and provide most clinical utility, particularly in guiding clinical care, drug development and interpretability/credibility of interventional trials.

Response biomarkers

A response biomarker quantifies a biological response to therapeutic intervention (figure 1d). Response biomarkers can be divided in two overall groups; pharmacodynamic biomarkers and surrogate endpoints, albeit the classification is not mutually exclusive. A pharmacodynamic biomarker is used to describe a therapeutic effect in a patient as a response to therapeutic intervention. It can reflect the biological activity of the intervention, but does not necessarily lead to conclusions on efficacy, risk of outcome or mechanism of action of the intervention.

Pharmacodynamic biomarkers can be very useful for establishing proof-of-concept. Surrogate endpoints are response biomarkers which reflect clinical outcomes, i.e., a beneficial effect on the biomarker leads to a beneficial clinical impact. Surrogate endpoints are in large part further interpreted by the level of clinical validation and are considered as candidate surrogate endpoints, reasonably likely surrogate endpoints, and validated surrogate endpoints.

A reasonably likely surrogate endpoint entails strong mechanistic support for how an impact on the surrogate endpoint correlates with an endpoint intended to reflect clinical benefit, whereas a validated surrogate endpoint is confounded by clear mechanistic evidence of a direct relationship between the surrogate endpoint and specific clinical benefit.

Predictive biomarkers

A predictive biomarker is used to identify patients who are more likely to respond to- and benefit from a specific therapeutic intervention. The terminology predictive biomarkers is drug-biomarker specific and requires an actual IVD development approved by the FDA.  Predictive biomarkers are often used for trial enrichment, by selecting patients based on their eligibility or to stratify patients into biomarker positive- and negative groups.

In drug development, predictive biomarkers are often chosen by how they directly reflect treatment induced changes in the specific pathophysiological pathway from which they are derived. Predictive biomarkers are often tested in Phase II and validated in phase III together with the choice of treatment, to assess whether the biomarker is truly predictive, as the difference observed in risk of outcome correlated to the biomarker can be due to the prognostic ability of the biomarker and the clinical benefit could occur irrespective of received therapeutic intervention in the biomarker positive group.

Please don't hesitate to contact us if you have any questions or other inquiries.

Are you interested in learning more about Nordic Bioscience?
Enter your information in the form and a representative will contact you shortly.

By submitting this form you agree to our terms and conditions.