Six Per Cent of Price Is the Whole Growth Story

The line that is missing from every summary

Cancer diagnostics grew quickly in 2025 and was paid less for each test it delivered. Those two sentences are both true of the same companies in the same year, and only one of them appears in the announcement.

Guardant Health grew oncology revenue 26 percent on oncology volume growth of 34 percent. Those numbers cannot both be right unless price fell, and the company’s own disclosures let you calculate by how much: realised revenue per oncology test declined from roughly USD 2,635 to roughly USD 2,476, a fall of 6.04 percent. Multiply volume growth of 1.34 by a price factor of 0.9396 and the product is 1.259 — which is the 26 percent revenue growth the company reported. The reconciliation is exact, and it is the clearest read available on what this market’s growth actually consists of.

Once you have that identity, it becomes a lens. Every participant in tumour profiling that discloses both a test count and the revenue those tests produced shows the same direction, and the differences between them are differences of degree rather than of sign.

Three companies, one direction

Veracyte grew testing revenue 18 percent on volume growth of 19 percent, which is a realised decline of 1.09 percent per test. The product mix is narrower and more guideline-anchored than Guardant’s, and the price held almost flat — but it did not rise.

Myriad Genetics’ tumour profiling revenue fell 3 percent in the same period while Prolaris volume rose 12 percent in the fourth quarter. That is the same identity producing a worse outcome: volume growth was not sufficient to hold revenue, and the gap between the two lines is price.

Exact Sciences, Natera, Tempus and NeoGenomics all grew, and all did so in a reimbursement environment where the administered rate is set by a fee schedule rather than by the seller. That environment is the subject of the piece on the clinical laboratory fee schedule, and it explains why an industry with genuinely improving products cannot raise the price of them.

What a falling realised price does to a forecast

The identity matters because it constrains what a forecast can claim. If realised price per test declines at 6 percent a year and volume grows at 12 percent, revenue grows at about 5 percent, not 12. Any projection that extrapolates test volume into revenue without a price term is assuming away the mechanism that is currently operating in every disclosed set of numbers in the category.

The oncology molecular diagnostics and tumour profiling market report therefore builds from segment revenue rates rather than from test volumes, and it sets those rates by asking what the buyer is paying for rather than how many tests exist. Therapy selection is 10 percent. Minimal residual disease is 20 percent. Classifiers are 8 percent. The remainder is 12 percent.

That ordering is not a comment on clinical value. It is a comment on where price holds. A therapy-selection panel competes with other therapy-selection panels and with tissue testing, and its price is contested. A classifier that answers a question a guideline asks — should this patient receive adjuvant chemotherapy, yes or no — is paid at a rate that does not move much, because the payer is buying the answer rather than the assay.

The annuity is the difference, and it shows up in gross margin

The clearest evidence for the segment split is not a growth rate but a margin. Natera ran 800,800 oncology tests in 2025, up 51.6 percent, and gross margin rose from 60.3 percent to 64.7 percent as volume compounded. Guardant grew oncology volume 34 percent and took non-GAAP gross margin from 62 percent to 66 percent, while realised price per test fell 6 percent, because the cost per test fell faster.

Now compare the one-shot panel. Myriad’s tumour profiling business shrank on falling price. NeoGenomics turned USD 727 million of revenue into a 5.91 percent adjusted EBITDA margin across a broad menu. The pattern is consistent: a test ordered once at a physician’s discretion does not improve with scale, because scale does not reduce the interpretive labour and does not improve the price. A test ordered repeatedly by protocol improves on both, because the same specimen type and the same workflow are amortised across more cycles.

That is the whole argument for putting MRD at 20 percent and therapy selection at 10 percent, and it is arithmetic rather than enthusiasm. Serial monitoring builds an annuity. A discretionary panel sells into a competitive market at a price the payer administers.

The base is a construction, and the hole in it is larger than anywhere else in molecular diagnostics

Nine attributed pools reconcile to USD 5,414.0 million of named revenue: Guardant USD 683.6 million, Natera USD 760.8 million, Exact USD 717.0 million, Tempus USD 687.9 million, NeoGenomics USD 472.6 million, Veracyte USD 483.6 million, Myriad USD 121.7 million, Roche USD 1,066.8 million and QIAGEN USD 420.0 million. The remainder of the USD 6.50 billion base — USD 1,086 million, or 16.7 percent — is unlisted and regional laboratory testing.

That residual is far larger than the 6.8 percent gap in the PCR market, and the reason is structural. Most tumour profiling happens inside hospital laboratories that publish nothing, because the testing is a cost centre inside a health system rather than a product sold to a market. A report that sized this category from listed company filings alone would be sizing the addressable portion of the market, not the market.

Two rows carry most of the judgement. Natera does not disaggregate oncology revenue, so its pool is 800,800 disclosed tests at an estimated USD 950 realised — the largest single estimate in the table, and labelled as one. Roche has no oncology split at all, so molecular is attributed at 35 percent of CHF 2,527 million converted at 1.2062; a 50 percent attribution would add USD 457 million to the base.

Twelve point one per cent in the first year and 12.71 per cent compounded

The two rates are different and the difference is arithmetic rather than a discrepancy. The segment build produces 12.09 percent in the first forecast year. Compounded across the full period the rate is 12.71 percent, because MRD grows faster than the category and its weight rises from 21.85 percent of the base to 36.1 percent of it by 2033. A faster segment compounding inside a slower one pulls the compound rate above the initial rate every year it gains weight.

The sensitivities are set by the two segments that matter. MRD at 15 percent instead of 20 percent gives a category compound rate of 11.17 percent. Therapy selection at 7 percent instead of 10 gives 11.42 percent. Both are within the range of outcomes disclosed companies have already produced, which is the point: this is not a market where the uncertainty is whether it grows, it is a market where the uncertainty is which part of it does.

What would break it

Three things, and none of them is clinical. If realised price per test keeps falling at 6 percent while volume growth decelerates toward the low teens, revenue growth converges on the low single digits regardless of how good the assays become. If coverage for serial monitoring narrows — the liquid biopsy report documents a coverage policy that pays for MRD only where it precludes other surveillance, which caps volume by design — then MRD’s 20 percent is not available. And if the four participants currently losing money on volume growth stop buying share with price, the volume line falls before the price line recovers.

The category’s growth is real. It is being purchased, and the purchase price is visible in the one number that never makes the summary: what each test earned.