A biomarker test now decides who enters a precision medicine trial, yet results often go unused and assays are not ready when screening opens.
For most of the history of clinical medicine, a laboratory result was an endpoint. It confirmed or ruled out a condition, supported a diagnosis or set a care pathway, and the lab's work was complete.
For anyone developing a precision medicine, the result is now the starting point. It says whether a patient carries the biomarker a therapy targets, whether they qualify for a trial and, in a basket or platform design, which arm they belong in.
Today, the test is the ticket to the trial, yet access to testing is uneven. Results often arrive in forms that neither a researcher nor a matching system can act on, and in many places there is no connection at all between the lab that produced the result and the team looking for patients.
Sponsors and contract research organizations experience these as enrollment problems. They are diagnostic problems first, and much of the solution can start in the lab.
Consider a pathology report that arrives by fax. Optical character recognition (OCR) can lift the text so a matching system can read it, but OCR is imperfect, and one misread field is an eligible patient missed.
Formatted PDFs fail the same way: easy for a physician to read, hard for a system to use for anything beyond the encounter they were written for. The root of the problem is that laboratories report results for the treating clinician, not for a matching algorithm.
The fix is to report the elements a matching system needs (the analyte, the value, the method and the specimen) as structured, coded data. Labs also must ensure that those data can flow into the systems used to find candidates rather than staying inside a document. A result alone still does not tell the whole story.
Joined with diagnosis, treatment history and prior response, it makes a patient's fit for a trial visible in a way no single value can. Large language models are good at reading across structured and unstructured records for those signals, and they are getting better.
But a matching tool is only as good as the result it is handed, and the judgment about whether a match is real stays with clinicians.
This is the part of the problem that is hardest to see from the sponsor's side. A trial team thinks of the eligibility test as something ordered at screening.
From the lab side, an assay a trial can rely on is a development project with its own timeline. Labs must select or build the method, validate it for the specimen types the trial collects, set it at the cutoffs the protocol requires and report it in a form the trial can use.
A lab that first hears about a marker when screening opens either has that assay or it does not. In my experience, the requests that reach a lab like ours are rarely for markers we cannot measure.
They are for versions of a test we have not yet validated: a different clinical cutoff, a different specimen type, a different reporting format. Each is weeks to months of work, and the clock does not start until someone tells us it is coming.
The cost of that gap does not end at approval, and it does not disappear once the test exists. Advanced lung cancer is the most mature precision-oncology setting there is.
Diaceutics' 2026 analysis found that testing improved there between 2019 and 2023. Even so, about two-thirds of eligible patients still did not get the most appropriate targeted therapy, because the loss moved downstream.
Among the patients who did get a result, the share lost at the treatment decision rose from 29% to 43%, and 86% of oncologists form a treatment plan before the full biomarker results are back.1 A result that arrives after the decision is, for that patient, a test that does not exist.
The same is true of a screening window. And lung cancer is the mature end of the curve. Most therapeutic areas now entering precision medicine are years behind it, with the assays they need still to be built.
The remedy is early engagement. When the lab is part of protocol design rather than a vendor chosen at site activation, the lab builds the assay alongside the program.
It is ready when screening opens, and it already exists on the day the therapy is approved, when the patients who need it are no longer trial participants but everyone.
Three questions separate a lab that supports enrollment from one that quietly slows it. Can you deliver results as structured, coded data, not only as a report?
If the honest answer is a PDF, plan for manual abstraction and the errors that come with it. Do your data policies, contracts and consents allow results to be used for research and trial matching?
Many labs have never had to answer this. The enterprising ones have reviewed their agreements and authorizations and can say what is permitted, for which patients and under what conditions.
When do you need to hear from us to have the assay validated before screening opens? A lab that answers "as early as possible, and here is what we need from you" is a partner.
A lab that answers "just send the order" has not thought about it. Keeping researchers, clinicians and laboratory staff in one conversation matters for the same reason.
Researchers know the population and the data the study needs. Clinicians know how decisions are actually made.
The lab knows what diagnostic information can and cannot support. A study designed with all three in the room is easier to enroll and easier to interpret.
That relationship also gives the lab time to prepare for what is coming. ~60% new oncology drugs approved in the United States in recent years require or recommend a biomarker test before use.2,3
Each of those therapies needs a test to identify eligible patients, screen for adverse effects and monitor response over time. The labs that see that demand early are the ones with the test ready when the therapy arrives.
Better-connected diagnostics also bear a persistent problem in clinical development: who gets into the trial. Access to advanced testing is uneven.
Patients treated outside major academic centers, and in some regions of the country, are less likely to be tested. When eligibility depends on a test, an access gap becomes an enrollment gap. Remote-care models that pair telehealth with testing and data capabilities can close some of that distance without asking every patient to travel to a research center.
Expanding access also means knowing who is missing. Diagnostic data linked to patient records show where the tested population diverges from the population a therapy is meant to treat.
No organization holds a complete dataset. But connecting diagnostic, clinical and research information lets researchers see the gaps and design studies that better represent the patients the drug is for.
Digital diagnostics can make trial matching proactive rather than retrospective: flagging a candidate when the result is generated, sometimes at the first clinical encounter, instead of searching records for signals that have been sitting there for months. That takes more than tools.
It takes results delivered in usable form, workflows that put the match in front of someone who can act on it, and a lab that treats supporting research as part of patient care rather than a separate line of work with different paperwork. As precision medicine advances, the line between supporting care and supporting research keeps fading, and laboratories belong on both sides of it.
Tests are expected to define a growing share of the next generation of therapies. The test should be ready before the trial has to ask for it.
About the Author
Chris Garcia, MD, is the chief digital innovation officer of Mayo Clinic Laboratories, a global leader in advanced laboratory testing, pathology services and diagnostics. In his current role as chief digital innovation officer, Dr. Garcia leads the development of Mayo Clinic Laboratories’ digital business products and services to drive growth and innovation. With his unique experience in data science and systems engineering, combined with his knowledge of the healthcare industry and clinical practice, Dr. Garcia leads the organization’s digital growth while supporting existing and emerging diagnostic business lines. He also serves as the medical director for BioPharma Diagnostics. Before joining Mayo Clinic in 2022, he held several leadership roles at Labcorp, including strategic director of Digital Pathology and Computational Pathology and director of Insight Analytics. His experience includes reference laboratory operations, artificial intelligence (AI) implementation in healthcare, and value-based care initiatives. His research interests are focused on AI/machine learning validation in medicine, digital transformation, and system design for healthcare diagnostics. Dr. Garcia earned his medical degree from the University of Illinois College of Medicine. He completed a residency in anatomic and clinical pathology at the University of Utah, and a fellowship in pathology informatics at Massachusetts General Hospital. Additionally, he holds a master’s degree in engineering management from the Massachusetts Institute of Technology.
References
