What a chordoma is made of

Written 7 September 2026 for her partner and the patient. Studies are cited by PubMed ID (PMID). The full thread with all 63 references is em-research/16. Written in terms of the dedifferentiated branch as the working frame, with the caveat that no one has yet seen the patient's tissue.

The cells

A chordoma is made of cells left over from the notochord, the rod that organised the spine when the patient was an embryo. Those cells were meant to disappear. A few stayed behind in the clivus (PMIDs 15166667, 15803192). At some point one of them kept a developmental switch called brachyury turned on, and lost a brake or two. Brachyury is both the marker pathologists stain for and the engine of the tumour (PMIDs 16538613, 22847733); some families carry an extra copy of its gene (PMID 19801981), and up to a quarter of tumours have gained a copy on their own (PMID 29026114).

That is nearly the whole genetic story. Chordoma has very few mutations compared with most cancers: a growth-pathway change in about one in six, and losses of the CDKN2A brake and a few chromatin genes (PMIDs 29026114, 33536423). Few mutations means slow division. Untreated, a conventional chordoma grows by roughly 8 percent of its volume a month (PMID 31926359). It is why the tumour can sit for years, and why ordinary chemotherapy, which works on fast-dividing cells, does little to the conventional type.

Under the microscope the classic cell is large and bubbly, the physaliphorous cell, sitting in lobules in a jelly-like matrix. The 2020 World Health Organization classification recognises three types: conventional (including a cartilage-like chondroid variant), dedifferentiated, and poorly differentiated (PMID 33438273).

Why it can hide

Slow is not the same as safe. A slow tumour in the clivus can reach several centimetres before it touches a nerve. On a brain MRI done without thin slices through the skull base and without contrast, a small one can look like normal bone marrow, or like a harmless notochord remnant that radiologists are taught to leave alone (PMIDs 25168701, 27206421, 23464492). So the November 2025 scan may have shown something that was reasonably read as nothing, or may have shown nothing at all. Which it was is a question for a skull base neuroradiologist with the actual images, which is why the national MDT email asks for exactly that.

What dedifferentiation is

"Dedifferentiation" means part of the tumour has stopped being chordoma. Under the microscope there are two tumours side by side with a sharp border: ordinary chordoma, and a high-grade sarcoma that has lost the notochord markers, brachyury included, and usually carries a new mutation in the TP53 guardian gene (PMIDs 32427623, 41420192). The sarcoma part has a rising mutation count and often loses the chromosome region that carries the brachyury gene, which is why the marker disappears.

It happens in fewer than 5 in 100 chordomas (PMID 27052403), under 1 in 100 in one large centre's series (PMID 32427623). It can be present from the start, appear when a tumour comes back, or follow radiotherapy; in the largest pooled analysis of 130 cases, 74 were present from the start and 39 arose later, and the two groups did no differently (PMID 36804481). When it is present, spread to lungs and bones becomes common rather than rare (one series: 57 percent of dedifferentiated against 5 percent of conventional, PMID 31338785), everything moves faster, and outcomes are worse. In that pooled analysis, complete surgery and radiotherapy were what predicted doing better; chemotherapy receipt did not separate the groups, though it is offered on sarcoma principles.

The third type

Poorly differentiated chordoma is a separate, mostly childhood type: brachyury is kept but a protein called INI1 (gene SMARCB1) is lost. It is a single fast population of epithelioid or rhabdoid cells rather than two tumours side by side, half the cases sit in the clivus, and the median age is about 11, with cases into the twenties (PMIDs 21057957, 29483606, 27067307). It has a targeted drug, tazemetostat, which has produced a documented exceptional response (PMID 30642912) and is in a children's trial (PMID 37228094). At 34 it would be unusual for the patient, but it is the type the rapid-growth note flagged and the INI1 stain answers it in a day.

[Correction 11 Oct 2026: tazemetostat was withdrawn from all markets in March 2026 (Ipsen, after SYMPHONY-1 showed second blood cancers in 5.7% vs 0); trials and compassionate access stopped. PMID 42599162; research 12 A3. Read anything above or below about it as history, not an option.]

Is it a mixture of cells with different metabolism? Yes

• Roughly half the cells in a chordoma are tumour cells; the rest are fibroblasts and immune cells. The tumour cells themselves come in several states, six in one single-cell study, including a stem-like, radiation-resistant group that tends to sit in one part of the tumour (PMIDs 35084549, 36127333).
• Conventional chordoma uses little glucose on PET and is patchy across the mass (PMID 33521875); residual tumour can be starved of oxygen yet PET-cold (PMID 22391104).
• Nobody has measured a dedifferentiated area on PET in chordoma specifically. In the closest related bone tumour, dedifferentiated chondrosarcoma, the high-grade part lights up far more than the low-grade part (PMID 27476566), and there is no reason to expect chordoma to differ. This is analogy, and I say so.
• Practical consequence: a single needle can miss the worst part. Two biopsies reported as conventional proved dedifferentiated at resection (PMID 27052403). The biopsy report should say where the sample came from and whether it matched the imaging, and the pathologist should stain for brachyury and INI1 as routine.

The three types side by side

What this means for the patient this week

• The tissue answers all of it. Brachyury, INI1 and TP53 staining, plus the whole genome sequencing already requested, tell the team which of the three she has. Until then the working frame is the aggressive branch, because it changes nothing about this week's actions and may save time.
• Ask that the biopsy is planned against the imaging so it samples the most active-looking part, and that the report states where the sample came from.
• If the team is willing, a PET scan before biopsy is a fair question: it can show a hot spot worth targeting and doubles as staging.
• The November 2025 images should be re-read by a skull base neuroradiologist next to the current ones. Whether a lesion was visible is not a judgement anyone can make from a report.