The Administered Price Is the Real Forecast Variable

Every test in this market has a launch price with an expiry date

Under the Medicare programme, a new advanced diagnostic laboratory test is paid at the laboratory’s own list price for an initial period and then, once that period closes, at the median of actual claims. The initial window runs roughly nine months. What that means commercially is that the price a diagnostic company announces at launch is a temporary permission rather than a price, and the durable number is whatever the claims distribution produces afterwards.

The documented initial periods are a short list, and reading them in sequence is the most useful thing a reader can do with this market. Shield, code 0537U, ran from 1 April to 31 December 2025 with a launch price of USD 1,495.00. Signatera, code 0340U, ran from 1 July 2021 to 31 March 2022 at USD 3,500.00. Guardant360, code 0242U, ran from 1 April to 31 December 2021 at USD 5,000.00. FoundationOne Liquid CDx, code 0239U, ran over the same window at USD 3,500.00.

Four products, four launch prices spanning a factor of more than three, and one common feature: each was temporary. This is the variable that the liquid biopsy market report treats as the primary forecast input, ahead of test volume, because volume is what a payer permits and the rate is what a payer pays.

Thirty-four per cent in ten days, with no change to the product

The clearest demonstration is not a launch price at all. It is a revision inside an administrative file.

In the CY2025 clinical laboratory fee schedule, code 0340U was first posted at USD 2,919.60 and then revised to USD 3,920.00 in the file modified on 8 January 2025. That is a 34.26 percent move in ten days. Nothing about the product changed in those ten days: no new evidence, no label change, no coverage determination, no competitive entry.

Because a single administrative file is a thin source, the report corroborates it against a second, non-commercial one. The California Division of Workers’ Compensation issued an order adopting that fee schedule, and the order states the revision verbatim. Two independent documents, one of them not a vendor and not a commercial publisher, report the same number.

Now put that 34.26 percent next to the operating reality of the businesses involved. Non-GAAP gross margins in this market sit between 60 and 66 percent for the participants that disclose them. For a business at that margin, a 34 percent swing in the administered price is a larger variable than several quarters of volume growth, and it arrived through a file modification rather than a market event.

Why the median of claims is not the list price

The mechanism matters because it determines which direction the durable price moves. The median of actual claims is a distribution, and the distribution includes commercial payers paying above the Medicare rate, Medicaid paying below it, and claims that never appear because a prior authorisation was denied. A laboratory’s list price is a claim about what it would like to be paid. The median is a record of what it was paid.

For a test ordered across a mixed population, the median will usually sit below the list price, because the low end of the distribution is populated by exactly the payers a laboratory cannot refuse. That is the structural reason a launch price is not a forecast input, and why a report that models revenue as tests multiplied by published price will overstate every year after the first.

The same mechanism is visible in the broader category. The piece on the clinical laboratory fee schedule documents a legislated glide path of capped annual reductions from 2027, which is the same force applied to the whole schedule rather than to a single code.

What this does to a 16.20 percent forecast

The category is sized at USD 3,757.6 million in 2025 rising to USD 12,492.2 million in 2033, a multiple of 3.32 times and a compound rate of 16.20 percent. That rate is carried by volume: MRD volumes at the fastest participants grew 46 to 51 percent in 2025, and the report deliberately forecasts MRD at 20 percent because coverage language caps monitoring volume by design.

Price is the quieter constraint, and it is the one that has already moved. A category rate of 16.20 percent assumes realised price per test holds roughly flat, which is generous for a market where every participant that discloses both a test count and the revenue from those tests shows a decline. The sensitivity is direct: MRD at 15 percent rather than 20 percent takes the category rate to 14.21 percent, and screening at 20 percent rather than its base case gives 14.53 percent.

Both of those are volume assumptions. Neither addresses a 34 percent repricing, which is why the risk chapter treats the fee schedule as a category-level exposure and not a product-level one.

The one number I will not state

Shield’s initial period ended on 31 December 2025. What replaced it has not been verified, and rather than infer a rate from the direction of travel, the report leaves it open. That is not reticence for its own sake. A rate that will be cited for the next eight years of a forecast should come from a document, and when the document is not available the correct output is a stated gap.

This is the same discipline that governs the note on the colorectal screening determination, where a specificity figure measured against advanced neoplasia cannot be converted into one measured against colorectal cancer without the underlying data. In both cases the available evidence establishes a mechanism and does not establish a number, and a forecast that quietly fills the gap is worse than one that names it.

How to read a liquid biopsy forecast

Three questions, in order. Which price is the model using, and is it a list price or a realised one. Which coverage envelope is assumed, because coverage determines whether a test can be ordered at all. And what happens to the rate if the administered price moves by a third in either direction, because that has already happened once inside this decade and the mechanism that produced it has not been removed.

A forecast that survives those three questions is a forecast about a reimbursement system with a technology input. One that does not is a forecast about a technology, addressed to a payer who is not buying one.