Cancer treatment has increasingly shifted from a one-size-fits-all chemotherapy regimen toward therapies matched to the specific genetic mutations driving an individual patient’s tumor. That shift depends entirely on fast, accurate tumor profiling — and the FDA has now cleared an AI-powered tool designed to do exactly that. Tempus’s xT CDx, an AI-driven companion diagnostic, received FDA clearance this year, according to a roundup tracked by health-AI research firm DeciBio, giving oncologists a validated tool for matching patients to targeted treatments based on their tumor’s molecular profile.
What a companion diagnostic actually does
A companion diagnostic is a test that doesn’t just detect disease but determines whether a specific drug is likely to work for a specific patient, by checking whether their tumor carries the genetic mutation that drug targets. Tempus’s xT CDx is described as “tumor-only,” meaning it profiles the tumor’s genetic material directly rather than requiring a separate comparison sample of the patient’s normal tissue — a design choice that can simplify and speed up testing, particularly useful when a patient needs a treatment decision quickly.
Why AI is central to the process
Modern tumor profiling generates enormous volumes of genomic data per patient — far more than a human pathologist could manually interpret mutation by mutation in a clinically useful timeframe. AI models trained on large genomic and clinical outcome datasets help identify which of the many genetic variants detected in a tumor sample are actually clinically significant, flagging the mutations most likely to respond to a specific class of drug and filtering out the enormous number of genetic variations that have no bearing on treatment.
Part of a broader wave of AI diagnostic clearances
Tempus’s clearance landed alongside another significant approval in the same period: ArteraAI Breast, described as the first AI digital-pathology risk test specifically for breast cancer. Together, these clearances reflect a broader pattern in 2026 of the FDA moving AI-driven diagnostics from research tools into clinically reimbursable, prescribable products — a shift that has also drawn investment attention, including Roche’s reported agreement to acquire digital-pathology company PathAI for up to $1.05 billion.
The oncologist’s perspective versus the skeptic’s
For treating oncologists, a faster, reliable tumor-only profiling tool means patients — especially those with aggressive cancers where time matters — can start matched targeted therapy sooner rather than waiting weeks for traditional paired-sample genomic testing. But some pathologists and genomic medicine researchers have cautioned that tumor-only testing, without a matched normal-tissue comparison, can occasionally misclassify inherited genetic variants as tumor-specific mutations, a technical limitation that matters most in edge cases and underscores why these tools are cleared as “companion” diagnostics used alongside clinical judgment, not standalone decision-makers.
The regulatory backdrop
The FDA’s broader guidance on AI in drug and biological product development, issued in draft form in January 2025, remains the agency’s core reference point, and the FDA has said it has reviewed more than 500 regulatory submissions containing AI components since 2016 — a number that signals how deeply embedded AI already is in the diagnostics pipeline, even if individual clearances like xT CDx still draw attention case by case.
What’s next
With xT CDx cleared, the practical rollout question is adoption: how quickly oncology practices and hospital labs integrate the test into routine workflow, and whether insurers reimburse it consistently given its relatively novel tumor-only design. The DIA’s 2026 Global Annual Meeting in Philadelphia included dedicated sessions on regulatory oversight of AI tools in radiologic and pathologic assessment, signaling that regulators themselves are still working out validation standards for this category — meaning more clearances, and more scrutiny of how they’re validated, are likely before the field settles into a stable standard of care. For patients, the practical change may arrive quietly: a treatment-matching decision that once took weeks of waiting on a separate genomic lab could increasingly happen within the same visit cycle as diagnosis, provided hospital labs and insurers move at the same pace as the FDA’s own clearance process. Community oncology practices, which often lack the in-house genomics expertise of major academic cancer centers, are likely to be both the biggest beneficiaries of a streamlined tumor-only test and the slowest to adopt it, given the added cost and staff training any new diagnostic workflow requires.
Photo: https://kaboompics.com/ / PEXELS via Pexels
