Few advances have shaped modern oncology more profoundly than the ability to match the right patient to the right therapy. Biomarkers play a central role in this process, helping guide treatment selection and supporting the continued expansion of precision medicine.¹ As researchers seek to identify the next generation of clinically relevant biomarkers, artificial intelligence (AI) is emerging as a promising tool for analyzing large and complex datasets that may contain previously unrecognized biological signals.² While much of the excitement surrounding AI focuses on its ability to identify new biomarkers, discovery alone is not enough. A biomarker’s value inherently depends on whether it can change clinical decision-making and be translated into routine care. This article explores those implications through three interconnected lenses: discovery, decision-making and implementation. 

Discovery

The path from biomarker discovery to routine clinical use is often lengthy, requiring researchers to identify promising biological signals, evaluate their relationship to clinical outcomes and validate their utility in patient care.¹ ² AI has attracted growing interest because it may help accelerate some of these early stages by identifying biomarkers associated with diagnosis, prognosis, treatment response, disease monitoring and other clinically relevant outcomes for further investigation.² However, accelerated discovery does not eliminate the need for rigorous clinical validation before new biomarkers are incorporated into practice.³ As the variety and volume of potential biomarker candidates grows, managed care stakeholders will need efficient approaches for evaluating emerging biomarker evidence and determining which candidates warrant consideration for future clinical adoption.   

Decision-making 

Biomarkers help inform some of the most important decisions in oncology. Consider a patient newly diagnosed with cancer. Biomarker testing may provide information about prognosis, identify the therapies most likely to benefit the patient and determine eligibility for clinical trials. These insights help guide initial treatment planning and may continue to inform care throughout the patient’s journey, often extending the impact of biomarkers in clinical decision-making well beyond diagnosis and early treatment selection.² As the disease evolves, biomarker information may help guide subsequent lines of therapy and support ongoing disease monitoring.² Ultimately, the growing role of biomarkers across multiple phases of care is expanding the scope of biomarker-informed decision-making. 

Although biomarker testing is increasingly integrated into oncology care, determining which biomarkers should be evaluated and when testing should occur can be complex.⁴ Clinicians are faced with a variety of questions: 

• Will a biomarker result influence a treatment decision today? 
• Could a biomarker result be needed to inform a future treatment decision? 
• Is additional testing likely to change management? 

As AI-driven biomarker discovery expands, providers may face growing difficulty keeping pace with emerging biomarkers, evolving testing recommendations and new treatment pathways. Managed care organizations can support decision-making through evidence-based pathways and timely guidance while using evidence related to clinical validity, clinical utility and patient outcomes to inform adoption of emerging biomarkers.  

Implementation 

Once a clinician determines that biomarker information is needed, practical questions quickly follow: 

• Which test should be ordered? 
• Is information on a single biomarker sufficient, or would broader profiling provide additional value? 
• Could other findings influence future treatment decisions? 

Ultimately, the amount of information needed to guide patient care determines how these questions are answered. In some situations, a focused, single-gene test may be sufficient to answer a specific clinical question.⁵ In others, broader approaches, such as next-generation sequencing (NGS), may evaluate multiple genes simultaneously and identify a wider range of actionable alterations.⁴ ⁵ While NGS may provide a more comprehensive assessment, it can also introduce additional considerations related to cost, interpretation of results and downstream clinical decision-making.⁵ For managed care organizations, clearly defining when focused testing or broader genomic profiling is appropriate can help align testing coverage with evidence-based treatment pathways. Testing decisions may also require coordination across medical and pharmacy policies, particularly when biomarker results directly determine access to targeted therapies.  

For some therapies, treatment eligibility depends on the results of a companion diagnostic test. Companion diagnostics create a direct link between biomarker testing and treatment selection by determining whether a patient is eligible for a specific therapy.⁶ Because treatment selection depends on the diagnostic result, testing and treatment policies are inherently interconnected. As AI expands the pool of potential biomarkers and biomarker-guided therapies, maintaining alignment between testing and treatment pathways may become increasingly important to ensure patients can access appropriate care in a timely manner.⁶ 

Key takeaways 

• Discovery: AI may accelerate biomarker discovery by helping researchers identify and prioritize potentially clinically relevant signals, but rigorous validation remains essential before adoption. 
• Decision-making: As the number of clinically actionable biomarkers grows, treatment decisions may become increasingly individualized, creating new opportunities for precision oncology while increasing the complexity of biomarker-informed care. 
• Implementation: The value of biomarker-guided care ultimately depends on effective implementation, including appropriate testing strategies, alignment between diagnostics and therapies, and evidence-based coverage policies. 

References:  

1. D'Avó Luís, A. B., & Seo, M. K. (2021). Has the development of cancer biomarkers to guide treatment improved health outcomes? The European Journal of Health Economics, 22(5), 789–810. https://doi.org/10.1007/s10198-021-01290-4  
2. Javaid, H., Petrescu, C. C., Schmunk, L. J., Monahan, J. M., O'Reilly, P., Garg, M., McGirr, L., Khasawneh, M. T., Al Lail, M., Ganta, D., Stubbs, T. M., Sun, B. B., Vitsios, D., & Martin-Herranz, D. E. (2025). The impact of artificial intelligence on biomarker discovery. Emerging Topics in Life Sciences, 8(2), 89–105. https://doi.org/10.1042/ETLS20243003  
3. Prelaj, A., Miskovic, V., Zanitti, M., Trovo, F., Genova, C., Viscardi, G., Rebuzzi, S. E., Mazzeo, L., Provenzano, L., Kosta, S., Favali, M., Spagnoletti, A., Castelo-Branco, L., Dolezal, J., Pearson, A. T., Lo Russo, G., Proto, C., Ganzinelli, M., Giani, C., ... & Pedrocchi, A. L. G. (2024). Artificial intelligence for predictive biomarker discovery in immuno-oncology: A systematic review. Annals of Oncology, 35(1), 29–65. https://doi.org/10.1016/j.annonc.2023.10.125   
4. DaCosta Byfield, S., Bapat, B., Becker, L., Reyes, C., Chatzitheofilou, I., Schroeder, B. E., Hostin, D., & Fox, J. (2025). Biomarker testing approaches, treatment selection, and cost of care among adults with advanced cancer. JAMA Network Open, 8(7), e2519963. https://doi.org/10.1001/jamanetworkopen.2025.19963  
5. Ewalt, M. D., West, H., & Aisner, D. L. (2019). Next generation sequencing—Testing multiple genetic markers at once. JAMA Oncology, 5(7), 1076. https://doi.org/10.1001/jamaoncol.2019.0453  
6. U.S. Food and Drug Administration. (2023, June 20). Companion diagnostics. https://www.fda.gov/medical-devices/in-vitro-diagnostics/companion-diagnostics