Does Medicare cover pharmacogenomic (PGx) testing?

Sometimes, and narrowly. There is no broad national coverage for PGx. Most coverage runs through Medicare Administrative Contractors under MolDX Local Coverage Determinations, which generally cover a PGx test only when it informs a specific drug the patient is taking or actively considering and the gene-drug interaction is clinically actionable. There is also a National Coverage Determination for pharmacogenomic testing for warfarin response (NCD 90.1), but it covers CYP2C9 and VKORC1 testing only under coverage with evidence development. Pre-emptive panels ordered without a drug in play are frequently denied.

Why are pre-emptive PGx panels often denied?

Because the MolDX policies generally require that the drug under consideration be selected on clinical grounds first, with the PGx result then informing safe selection or dosing. Testing a broad panel before any medication is chosen, in case a result is useful someday, typically does not meet the reasonable-and-necessary standard. The contractor billing instructions also expect the drug name to be documented and submitted, which a no-drug pre-emptive order cannot satisfy.

What makes a gene-drug pair 'actionable' for coverage?

Under MolDX policy language, a use is generally considered actionable when the genotype may lead to selecting or avoiding a therapy or changing a dose, and that action is supported by the FDA drug label, an FDA warning or safety concern, or a CPIC Level A or B gene-drug interaction. If the pair does not meet one of those thresholds, coverage is unlikely.

Are oncology PGx tests like DPYD and UGT1A1 covered by Medicare?

Often, when ordered correctly. Following a series of contractor local coverage decisions, clinically indicated PGx testing for validated genes including DPYD and UGT1A1 is a covered service for Medicare beneficiaries across most of the United States, tied to the relevant therapy such as fluoropyrimidines for DPYD or irinotecan for UGT1A1. Coverage still depends on the patient being a candidate for that drug, correct documentation, and the applicable contractor's current policy.

Pharmacogenomic Testing Coverage: When Medicare Pays for PGx and When It Denies

A patient is about to start capecitabine. Their oncologist orders a DPYD test, the lab runs it, the result comes back, and Medicare pays. A few rooms away, a different clinic runs a 25-gene pharmacogenomic panel on a new patient who is on no relevant medication yet, "to have it on file." That claim gets denied. Same category of test, same CLIA-certified lab, opposite financial outcome. If you run a lab or a practice that bills PGx, understanding why those two stories end differently is the whole game.

Disclaimer: This is educational content, not medical, legal, or billing advice, and it is not a coverage determination for any specific patient or claim. Coverage rules change, Medicare Administrative Contractors apply their own local policies, and the only authoritative sources are the current NCD, LCD, and billing-article language. Always verify against CMS's current policy and consult qualified clinical, billing, and compliance professionals before making coverage or treatment decisions. Use this information at your own risk. Nothing here guarantees coverage for any test.

The shape of the problem

Pharmacogenomics sits in an awkward spot. Clinically, the field keeps maturing: the FDA maintains a Table of Pharmacogenetic Associations, and the Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes graded, genotype-based prescribing guidance. Commercially, though, the reimbursement landscape has not kept pace with the enthusiasm. There is no broad national mandate that says "Medicare covers PGx." Instead, coverage is assembled gene by gene, drug by drug, and contractor by contractor, and most of it turns on one question: was a specific drug actually in play when the test was ordered?

That single question separates the two outcomes in the opening. Targeted, single gene-drug PGx tied to a medication the patient is taking or actively considering is the version that tends to get paid. Pre-emptive, broad-panel PGx run before any drug decision is the version that tends to get denied. The rest of this article explains the rules that produce that split, and where otherwise-reasonable claims still fall apart.

The thin national layer: warfarin and not much else

When people assume there must be a national Medicare rule for PGx, they are usually thinking of one narrow determination. There is a National Coverage Determination for pharmacogenomic testing for warfarin response, commonly cited as NCD 90.1. It is worth understanding precisely because it shows how conservative the national posture has been.

Under NCD 90.1, testing of CYP2C9 and/or VKORC1 alleles to predict warfarin responsiveness is covered only through coverage with evidence development (CED) — meaning, in practice, for beneficiaries enrolled in a qualifying clinical study — and is limited to once per lifetime per beneficiary for patients who are candidates for warfarin anticoagulation and have not been previously tested. CMS's stated rationale was blunt: the available evidence did not demonstrate that this testing improves health outcomes for Medicare beneficiaries outside the CED context. (CMS NCD 90.1)

Read that again from a revenue perspective. The one squarely national PGx coverage rule is a coverage-with-evidence-development pathway with a once-in-a-lifetime cap. That is not a generous front door. It is a signal that the national program treats PGx as something to be paid for sparingly and only when tightly justified. Almost everything else you bill for PGx lives below the national layer, in contractor territory.

Where the real coverage lives: MolDX and the LCDs

Most PGx coverage decisions are made not nationally but by Medicare Administrative Contractors, and a large share of molecular testing runs through the MolDX program. MolDX maintains a family of Local Coverage Determinations on pharmacogenomics testing — you will see them under IDs such as L38294, L38335, L38337, L38394, and L38435 across the participating jurisdictions — and the participating contractors apply consistent criteria. (MolDX: Pharmacogenomics Testing L38294)

The core logic of these policies is what you need to internalize, and it is more specific than "is this test useful." The MolDX policies frame coverage around an actionable gene-drug interaction. In the policy's own framing, a use is generally actionable when the genotype result may lead to selecting a therapy, avoiding a therapy, or changing a dose — and that action must be grounded in one of a short list of authorities: the FDA drug label, an FDA warning or safety concern, or a CPIC Level A or B gene-drug interaction. If the gene-drug pair does not clear one of those bars, you should not expect coverage. (MolDX: Pharmacogenomics Testing L38435)

There is a second, equally important rule embedded in the same policies, and it is the one that catches pre-emptive testing. The medication has to be chosen first, on clinical grounds, and the PGx result then informs how to use it safely. In the policy's language, the selection of the medication must derive from clinical necessity rather than from the PGx test; once a therapeutic agent is selected, the test may be considered reasonable and necessary when the result is needed for the physician's decision about safely administering or dosing that drug. (MolDX: Pharmacogenomics Testing L38435)

Notice what that sequence does. It puts the drug before the test. A panel ordered "so we have results on file" inverts the order — it puts the test first and hopes a drug decision will later make it relevant — and that inversion is exactly what the policy is designed to exclude.

Single gene versus panel: a more nuanced line than "panels are bad"

It is tempting to summarize all of this as "single gene-drug tests get paid, panels get denied." That is close, but it oversells the panel ban and undersells the actual rule.

The MolDX policies do not categorically forbid multigene or panel testing. A panel can be payable when more than one gene on it is reasonable and necessary for the safe use of a medication, or when multiple drugs are genuinely under consideration for the patient. The disqualifier is not the number of genes; it is the absence of a drug decision driving the order. A two-drug regimen that implicates two actionable genes can support a multigene claim. A broad panel justified by "the patient might need one of these drugs someday" cannot. (MolDX: Pharmacogenomics Testing L38294)

And this is the heart of why pre-emptive, population-style PGx screening — the model some vendors market as "test once, use for life" — keeps colliding with Medicare. The clinical case for pre-emptive testing can be real. The coverage case, under current MolDX policy, generally is not, because the policies treat broad pre-emptive panels as not reasonable and necessary when there is no specific medication driving the test. The contractor billing and coding articles reinforce this operationally: they expect the performing lab to retain records of the drugs in use or under consideration and to submit the relevant drug information on the claim, with limits such as one test per drug in question per date of service. A pre-emptive panel with no drug attached simply cannot satisfy those instructions. (Billing and Coding: MolDX: Pharmacogenomics Testing A58395)

The oncology examples worth knowing

Two PGx examples come up constantly in the oncology and specialty-lab world, and they illustrate the "drug-tied, actionable" pattern cleanly.

The first is DPYD, relevant to fluoropyrimidine chemotherapy such as 5-fluorouracil and capecitabine. Certain DPYD variants reduce the enzyme activity that clears these drugs, raising the risk of severe toxicity, which is why testing is done before treatment and why a result can change the dose or the choice of agent. Following the wave of contractor local coverage decisions, clinically indicated DPYD testing is reimbursed by Medicare across most of the country, and clinical guidance such as the NCCN colon cancer guideline supports offering DPYD testing to patients planning fluoropyrimidine therapy. (Ho et al., Clinical Pharmacology & Therapeutics, 2025)

The second is UGT1A1, relevant to irinotecan. Reduced-function UGT1A1 variants impair clearance of irinotecan's active metabolite and are associated with increased risk of severe neutropenia and diarrhea. As one oncology-practice review put it, a series of local coverage decisions by Medicare contractors means clinically indicated PGx testing for validated genes including both UGT1A1 and DPYD is a covered service for Medicare beneficiaries across most of the United States. Both pairs are also recognized in the FDA's Table of Pharmacogenetic Associations as associations with therapeutic management implications. (JCO Oncology Practice, 2022; FDA Table of Pharmacogenetic Associations)

The pattern is the takeaway, not just the two genes. Each is tied to a specific drug class the patient is a candidate for. Each clears the actionability bar through the FDA label and CPIC guidance. Each is ordered before treatment, when the result can actually change the plan. That is what a payable PGx claim looks like.

Where otherwise-reasonable PGx claims still get denied

Even when a test is, in principle, coverable, the claim can still fall over. A few recurring failure modes are worth naming, because they are the ones that turn a clinically sound order into a write-off.

The first is the no-drug order — the pre-emptive panel discussed above. If the documentation does not establish a specific medication being taken or actively considered, the policy's reasonable-and-necessary gate is not met, full stop.

The second is the actionability gap. The gene-drug pair has to clear the FDA-label, FDA-safety, or CPIC Level A/B threshold. A pair that is biologically interesting but does not meet one of those authorities is unlikely to be covered, no matter how the order is written.

The third is missing drug documentation on the claim. The MolDX billing articles expect the relevant drug information to be retained and submitted, and they impose frequency limits. A claim that omits the drug, or that exceeds the per-date-of-service limits, invites denial even when the underlying test was appropriate. (Billing and Coding: MolDX: Pharmacogenomics Testing A58395)

The fourth is jurisdictional drift. Because so much of this is contractor-level, the same test can be handled differently across MACs, and policies are revised on their own schedules. A workflow validated against last year's LCD revision in one jurisdiction is not automatically safe in another, or even in the same one after an update.

The strategic reality: PGx coverage is a moving, gene-level rulebook

Step back and the structural point is hard to miss. PGx reimbursement is not governed by one rule you can learn once. It is governed by a national determination that covers essentially one drug under evidence development, plus a set of contractor LCDs and billing articles that grant coverage only for actionable gene-drug pairs tied to a real medication decision — and those pairs are defined by external authorities (the FDA label, FDA safety communications, CPIC's evolving Level A/B list) that keep changing as the science advances. CPIC has reviewed hundreds of gene-drug pairs and published guidance on a growing subset; the FDA table is periodically updated. Every one of those updates can move the coverage line for a specific test. (CPIC overview, Clinical Pharmacology & Therapeutics)

So getting paid for PGx is not really a clinical problem or even a billing problem in isolation. It is a tracking problem. Whether a given test is payable depends on the gene, the drug, the indication, the patient's actual medication context, and the current language of the specific contractor's policy — all of which drift over time. A lab that wins on PGx is one that knows, per gene and per drug, what the live coverage criteria are and what the order needs to look like to satisfy them.

That is the category Converus works in: keeping the per-gene, per-drug, per-payer coverage rules current so reimbursement teams can order and bill against today's policy rather than last year's. You do not have to use a tool to act on the insight, though. The insight is the point. In pharmacogenomics, coverage is narrow, it is specific to the gene-drug-indication triple, and it changes — so the operators who get paid are the ones who treat policy tracking as a standing function, not a once-a-year project.