How chordoma cells run and what fuels the aggressive forms

For her partner, 10 September 2026. Answer to: how do dedifferentiated and poorly differentiated chordomas behave metabolically, what does "sarcoma-like" mean, does a chordoma break out of its slow growth, and what feeds that. Three tiers: the basics, the current papers, and the ideas outside the box with their evidence and their gaps. Every figure is sourced in research thread 21 (73 papers, PubMed IDs checked). Relative comparisons only; no survival figures. Where a line is my reasoning rather than a paper's finding it says so.

Tier A. The basics

What a chordoma cell runs on. A conventional chordoma cell is a notochord cell that never switched off its embryonic programme. It divides slowly: in skull base series the share of cells preparing to divide (the Ki-67 index) is mostly under 6 percent, and tumours above that grow measurably faster on repeat MRI (Holton 2000). Untreated skull base chordomas grew a mean 8 percent of their volume a month, with wide scatter (Wang 2020). On a PET scan they take up glucose modestly, with patchy uptake (Olson 2021), and in one small study most residual skull base chordomas lit up on a tracer for low oxygen but not on the glucose tracer (Mammar 2012). So a conventional chordoma is a low-flame tumour: slow, short of oxygen, fed by ordinary blood supply.

Dedifferentiated and poorly differentiated are two different things. A dedifferentiated chordoma is two tumours in one: ordinary chordoma sitting next to a high-grade sarcoma of spindle or irregular cells that divide fast, outgrow their blood supply and die in patches, and have lost brachyury (Meis 1987; Hung 2020). It is found in under 5 percent of chordomas. A poorly differentiated chordoma is one uniform tumour that keeps brachyury but has lost the gene INI1 (SMARCB1); it is mainly a tumour of children and young adults (Mobley 2010; Shih 2018).

What "sarcoma-like" means in practice. Faster division, dead patches inside the tumour, and spread through the bloodstream to lungs, bone and liver rather than only creeping locally. In the largest dedifferentiated series, 36 percent had spread at presentation against 3 percent for conventional chordoma, and after treatment 57 percent against 5 percent (Nachwalter 2019). Treatment shifts towards sarcoma chemotherapy alongside surgery and radiotherapy; in a pooled analysis of 130 cases, surgery was the one factor that held up (Liu 2021).

Does a chordoma "break out"? Sometimes, and it is documented three ways: sarcoma present from the start, sarcoma appearing at a recurrence, or sarcoma appearing after radiotherapy (Hanna 2008; Ikeda 1997). Across 130 cases, the from-the-start and later-appearing forms did not differ in age, site, size or spread (Vuong 2023). It is a change in the cells' genes: one patient's conventional tumour and its dedifferentiated recurrence went from 211 to 430 mutations, gaining TP53 and BRCA1 faults and losing the chromosome region that carries brachyury (Baluszek 2025). Nothing in the diet does that.

Fuel versus signals. Cancer cells grow because faulty signals inside them (here brachyury, the PI3K/AKT/mTOR growth pathway, loss of the brake gene CDKN2A, and in transformed tumours TP53) tell them to divide regardless of need; the extra sugar uptake follows the signal, which is why dividing cells burn glucose wastefully even with oxygen to spare (the Warburg effect; Vander Heiden 2009; Hanahan 2022). Blood glucose is held in a narrow band by the liver and pancreas whatever she eats; eating less does not starve the tumour, it starves her muscles, because the liver makes glucose from muscle and fat when food is short. Weight loss in cancer works the other way round: the tumour drives breakdown of the body's muscle and fat, and adequate nutrition is the mainstay of managing it (Baracos 2018). The European cancer nutrition guidelines advise against energy-restricting diets in anyone malnourished or at risk and find no adequate evidence for ketogenic diets in cancer (Arends 2017; Muscaritoli 2021). At 51 kg and down from 58 to 60, the priority is protein and calories, not restriction; that is my reasoning from those guidelines.

"Sugar feeds cancer." True at the level of the cell, false as advice: dietary sugar does not raise blood glucose in a way a tumour can exploit, and no trial shows that cutting sugar slows an established tumour. The one place glucose and insulin demonstrably matter is with a particular class of drug (PI3K inhibitors), which raise insulin and are blunted by it; that is a signal, not a fuel, and not a drug she is on (Hopkins 2018).

Tier B. Current thinking

• Growth rate and prognosis. A Ki-67 above about 5 to 6 percent marks faster growth and earlier progression in skull base series (Holton 2000; Zhang 2019). No study reports the Ki-67 range of dedifferentiated versus conventional chordoma; the sarcoma part is by definition dividing fast.
• Glucose and PET. Conventional chordoma uses glucose at a low to moderate rate, less than a pituitary adenoma (Olson 2021; Tosaka 2017). Chordoma cells carry the standard glycolysis machinery (Wang 2020, cells). No study compares PET uptake in dedifferentiated against conventional chordoma; in one series of 18 sacral chordomas after carbon-ion therapy, a rising PET signal picked out the one tumour that later proved dedifferentiated (Evangelisti 2019). In the closest analogy, cartilage tumours, a high PET reading separated dedifferentiated from ordinary chondrosarcoma with 89 percent sensitivity (Gundavda 2022).
• Low oxygen. The hypoxia response protein HIF-1alpha is present in most chordomas and, in one spinal series, high levels predicted earlier recurrence or spread (Li 2012; He 2024). Low oxygen makes tumours resist radiotherapy and invade; it is one of the reasons proton dose matters.
• Brachyury's programme. Brachyury sits at the top of a network of about 160 genes controlling cell division and the tumour's matrix; switching it off arrests growth in the laboratory (Nelson 2012). Extra copies of the brachyury gene track with higher activity of the PI3K/AKT growth pathway and with earlier progression (Otani 2018). A direct brachyury-to-glucose link was looked for and not found; the connection runs through PI3K/AKT/mTOR, the pathway that couples growth signals to nutrient uptake.
• The growth pathway. In 50 chordomas the activated forms of AKT and its downstream partners were present in over 90 percent (Presneau 2009); chordoma cell lines have hyperactive AKT and mTOR and rapamycin slows them (Han 2009); 16 percent of sequenced chordomas carry a mutation in this pathway (Tarpey 2017).
• IDH. Chordoma does not carry IDH mutations; the main lookalike, chondrosarcoma, often does, so the mutation helps tell them apart (Arai 2012). A 2026 laboratory paper found chordoma depends on the normal IDH1 enzyme for its antioxidant supply and that blocking it shrank two xenografts (Pun 2026, cells and mice).
• Fats. A 2026 preprint, not yet peer reviewed, links recurrence and radioresistance to fat synthesis in the cells (Wei 2026). A proteomic study of 102 chordomas found three subtypes, one running on oxidative metabolism (Yin 2024).
• INI1 loss. INI1 and the enzyme EZH2 are opposing regulators of how DNA is packaged; losing INI1 leaves EZH2 unopposed and locks cells in a growth state (Wilson 2010). In poorly differentiated chordoma this loss drives progression, with CDKN2A also lost in the cell lines (Walhart 2023).
• What transformation acquires. TP53 mutation is the signature: in 9 of 23 dedifferentiated or poorly differentiated tumours against 5 of 445 conventional (Baluszek 2025). Loss of the brake gene CDKN2A is the commonest change acquired as any chordoma progresses, with its protein absent in about half of samples (Cottone 2020), and its deletion together with a 1p36 loss predicts early progression in a prospective clival series (Zenonos 2019).

Tier C. Outside the box, ranked by human evidence

None is standard care before histology, and each targets a marker that the biopsy will or will not show.

1. Regional hyperthermia with sarcoma chemotherapy. The strongest evidence of the list: in 329 patients with high-risk soft tissue sarcoma, heating the tumour region during chemotherapy improved local control and survival in a randomised trial (Issels 2018). Gap: limb and trunk sarcomas; nothing at the skull base; heating next to the brainstem and carotids is not established. Relevant only if a dedifferentiated part is confirmed and a sarcoma protocol chosen.
2. mTOR inhibitors, imatinib with everolimus. In 43 patients with progressing chordoma, 9 shrank by a lenient measure and 1 by the strict one, a third stopped for toxicity, and response tracked with how many tumour cells had the pathway active (Stacchiotti 2018). In a pooled analysis of 12 phase 2 trials, it was the only regimen that held progression off longer than the rest (Cote 2026). Gap: conventional, advanced disease; the effect is holding, not shrinking.
3. Tazemetostat, the EZH2 inhibitor. Licensed for INI1-negative epithelioid sarcoma (Gounder 2020); one metastatic poorly differentiated chordoma had a lasting response (Gounder 2019). Gap: applies only if the INI1 stain is negative, which is unusual at 34; the stain is routine and is the question to ask.

[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.] 4. Immune checkpoint drugs. Across a basket trial of rare sarcomas including 34 chordomas, about 6 percent responded (Blay 2023); one INI1-negative chordoma in a child responded to nivolumab (Williamson 2021). The target protein PD-L1 is scarce on chordoma cells (Chan 2025). Gap: low response unselected; the markers that would select are not settled. 5. CDK4/6 inhibitors for CDKN2A loss. Palbociclib in 28 patients with that loss held the disease in 39 percent with no shrinkage (Teleanu 2025). Gap: stabilisation only. 6. Brachyury vaccines. Safe and immunogenic, but the randomised trial of one vaccine with radiotherapy was stopped for futility (DeMaria 2021) and a second missed its target (Cote 2024). 7. EGFR inhibitors, afatinib. The only EGFR drug active across chordoma cell lines, and it also degrades brachyury (Magnaghi 2018). The European trial in patients is not published in PubMed; treat quoted results as unverified. 8. Metformin. No chordoma study. The definitive test elsewhere, 3,649 breast cancer patients randomised to metformin or placebo, was negative (Goodwin 2022). No reason to expect chordoma to differ. 9. Glucose-blocking drugs (2-deoxyglucose, dichloroacetate). Cell work only in sarcoma; in a phase 1 trial 2-DG caused high blood sugar in everyone and heart rhythm changes in a fifth, with one response in 34 (Raez 2013). No chordoma data, no human efficacy signal. 10. Ketogenic and fasting diets. The honest tally: a ketogenic diet alone did nothing measurable in recurrent glioblastoma (Rieger 2014); a calorie-restricted ketogenic diet with fasting added nothing to re-irradiation in a randomised trial and cost 2 kg (Voss 2020); a fasting-mimicking diet around chemotherapy improved scan responses in a breast cancer trial that excluded anyone underweight (de Groot 2020); a ketogenic diet during chemotherapy showed a benefit in one contested trial (Khodabakhshi 2021). Meta-analyses find these diets lower glucose and body weight but not the growth hormone IGF-1 or quality of life (Salido-Bueno 2024; Zhao 2022), and a review of fasting around chemotherapy found weight loss in every study and called the claims premature (Bozzetti 2022). Every positive trial excluded underweight patients, and conventional chordoma is not treated with chemotherapy in the first place. For the patient, at 51 kg and losing, the expected result is more muscle loss with no demonstrated tumour effect: not recommended on the current evidence. 11. IDH1 and fat-synthesis blockers. Two 2026 laboratory leads (Pun 2026; Wei 2026 preprint). No human data. 12. PET-guided biopsy. In breast cancer the biopsy missed the most active part of the tumour in 14 percent, more often when the tumour was over 3 cm (Koolen 2013); two spinal chordomas read as conventional on needle biopsy proved dedifferentiated at surgery (Kato 2016). The practical ask: sample any region that looks different on MRI or hottest on PET, and record where the needle went.

What this means for the week

1. The subtype is answered by two routine stains, brachyury and INI1, plus TP53, p16 and Ki-67: ask for all five on the biopsy report, and for the sample location.
2. Food does not feed the tumour in any controllable way; weight loss does harm. Protein and calories, dietitian input, no restriction diets.
3. Every drug in tier C targets a marker; none applies until the marker is known, and the ones with the best evidence are for advanced or confirmed-sarcoma disease.