Clival chordoma research
Home › Papers shared in the room

Shared papers and appraisals

Started 5 September 2026. Her partner posts research links, papers, screenshots or photos in the room; Kya saves a copy here named by date and time, reads it, and replies in the room with an honest appraisal: what it shows, how strong the evidence is, and whether it applies to the patient. Each item gets a short entry below.

Date and time Item (file) What it is Kya's appraisal (one line) Room message
2026-09-05 08:40 2026-09-05_0840_kyoshima-2003-repeat-operations-clival-chordoma.jpg (screenshot) Kyoshima K et al. Repeat operations in the management of clival chordomas: palliative surgery. J Clin Neurosci 2003;10:571-578. PMID 12948462. Three patients, 6 to 12 operations each over 8 to 17 years, cases 1978 to 1999 (her partner checked the dates), pre-endoscopic and pre-proton era. Narrow: repeat decompression relieves diplopia and visual symptoms in aggressive recurrent cases. Not a guide to a new diagnosis or to the patient's outlook; n=3. Abstract contains survival times, not repeated in the room per the prognosis rule. 1545700211580739628 (her partner), appraisal posted 08:48

2026-09-07 00:38, Cameron and Pauling 1976 (PNAS 73:3685), screenshot

Her partner shared the title page while watching a Dr John Campbell video on vitamin D and cancer. Appraisal: 100 "terminal" patients given 10 g/day ascorbate (IV then oral) vs 1,000 historical controls, survival about 4x longer (PMID 1068480). Non-randomised, historical controls, clinician-selected entry point; not reproduced. Two randomised placebo-controlled Mayo trials of 10 g/day oral (Creagan 1979 NEJM, n=150, PMID 384241; Moertel 1985 NEJM, n=100, PMID 3880867) found no benefit. Pauling's oral-vs-IV objection later gained biological support (pharmacological ascorbate reaches millimolar levels only IV), which is the basis of the modern IV trials; those still show no randomised survival benefit (see em-research/15, section 6).

2026-09-07 22:50, Wang XJ ... Zhang DD, "Nrf2 enhances resistance of cancer cells to chemotherapeutic drugs, the dark side of Nrf2", Carcinogenesis 2008 (PMID 18413364)

Her partner asked who wrote it. Abstract saved alongside. Appraisal: cell-line study from Donna Zhang's Arizona laboratory showing Nrf2 activation confers chemoresistance and Nrf2 knockdown restores sensitivity; heavily cited; the basis for caution about Nrf2 activators (sulforaphane and similar) during chemotherapy or radiotherapy. Not chordoma-specific. Follow-up (7 Sep 23:30): her partner asked about conflicts of interest. The 2008 paper was NIH-funded academic work; no declared interest known. The same laboratory (Zhang DD) later published brusatol as an Nrf2 inhibitor that enhances chemotherapy (Ren D et al, PNAS 2011, PMID 21205897, verified on PubMed), i.e. a subsequent research and possible commercial interest in the "Nrf2 protects tumours" narrative. Patent status not verified. Support for the narrative, verified 23:35: Jeong Y et al, Cancer Discov 2017, PMID 27663899 (KEAP1/NRF2 and TP53 mutations, lung squamous, radiation resistance; Diehn group, independent); DeNicola GM et al, Nature 2011, PMID 21734707 (Tuveson group, independent). Rojo de la Vega M et al, Cancer Cell 2018, PMID 29731393, "NRF2 and the Hallmarks of Cancer", is from the Zhang DD laboratory and is NOT independent; Kya wrongly presented it as such in the room and corrected it at 23:36. The PMID first quoted for Jeong (28073737) was wrong.

2026-09-08 00:04, PMID 38549144, Alvarez MR et al, J Neuroinflammation 2024, oral NaHCO3 immunomodulation via the splenic nerve

Her partner cited the PMID. Abstract saved alongside. Appraisal: rat study (female rats, 4 days oral sodium bicarbonate vs water); splenic shift to anti-inflammatory markers (M2 macrophages, Tregs) abolished by splenic denervation; no cancer, no disease model, no humans. The anti-inflammatory shift (Tregs, M2) is the direction associated with weaker anti-tumour immunity, so it does not argue for bicarbonate in cancer. Separate "alkalise the tumour" idea: mouse data (Robey 2009 and similar), one small human tolerability attempt, no patient outcome data. For the patient: not recommended; sodium load at gram doses with low BP and NSAID use.

2026-09-09 01:27, PMID 40729189, Antonini et al, Reports (MDPI) 2023, glioblastoma case: cannabis, melatonin, oxygen-ozone

Her partner gave the title (the "GBM survivor" paper). Free full text from PMC (PMC12225241) saved alongside. Appraisal: single case report. The abstract says the patient refused conventional therapy; the case text says she had partial resection (Aug 2016), full radiotherapy 59.4 Gy in 33 fractions plus concurrent temozolomide (Oct to Nov 2016), then stopped temozolomide after one cycle. Tumour was MGMT-methylated, the subgroup that responds best to radiotherapy and temozolomide. The 2 cm residual had gone by Feb 2017, three months after radiotherapy finished, which is when radiotherapy response shows. The unconventional regimen (rectal ozone, melatonin escalated to 2 g/day, Bedrocan and Bedrolite up to 1 to 5 g/day each) started at the same time, so nothing separates its effect from the standard treatment. Long-term MGMT-methylated survivors exist without any of it. Senior author group has a cannabinoid research programme. No relevance to chordoma (different biology; no chordoma data). Melatonin dose used is a thousand times the patient's 2 mg.

2026-09-10 14:55, two papers from a friend (em-research)

11 September 2026, 20:48 — her partner asked about molecular hydrogen (H2)

Not a paper, a question, answered in em-research. Appraisal: plausible mechanism (Ohsawa 2007, Nature Medicine: selective scavenging of hydroxyl radical and peroxynitrite, leaving signalling radicals alone), very thin human evidence, nothing whatever in chordoma.

The only trial that bears on the patient's situation is Kang 2011 (Med Gas Res, PMID 22146004): 49 patients having radiotherapy for liver tumours randomised to hydrogen-rich water or placebo; quality of life better, oxidative stress markers down, and critically no difference in tumour response. L2, frontier. Everything else is reviews citing reviews, cell work, mice, and small uncontrolled advanced lung cancer series claiming survival benefit that will not bear weight. The 2026 Frontiers review adds no new patients.

The substantive issue raised with her partner: proton therapy kills chordoma partly through the free radicals H2 is designed to scavenge, so an antioxidant during radiotherapy could in principle protect the tumour too. Same argument as high-dose vitamin C in the Dalgleish note. Kang's null result on tumour response is the only direct evidence either way and points the right way, but n=49. NCT04175301 (pilot, hydrogen water during brain radiotherapy for high grade glioma) is the closest analogue and is small.

Safety genuinely benign: no meaningful toxicity, no interaction with morphine, gabapentin or diclofenac. Practical catches: dissolved H2 escapes water within minutes so most bottled product is inert by the time it is drunk; generators are expensive; inhalation kit is flammable.

Position given: not worth arguing against if she wants it, unlike graviola in the plants note it will not hurt her, but not ahead of food money, and ask the proton centre before starting it in the fortnight around any radiotherapy. Added to the questions for the proton centre.

Correction from her partner, 21:15. My "dissolved H2 escapes bottled water" objection does not apply to the product he is looking at: these are tablets dropped into water and drunk immediately, so the gas is generated in the glass. He is right. Two consequences worth holding: (1) those tablets are elemental magnesium plus an acid, and the hydrogen comes from Mg + 2H2O to Mg(OH)2 + H2, which is essentially the method used in Kang 2011 (a metallic magnesium stick in drinking water), so the tablets are closer to the tested intervention than any bottled product; (2) each tablet is therefore also a magnesium dose, often around 80 mg elemental, which must be counted against any separate magnesium supplement and against the renal caution (diclofenac, 51 kg, creatinine never seen).

For the patient: the question is her level, which has never been measured, not the dose. She is on 2,000 IU/day, a sensible holding dose. Specific caution given to her partner: vitamin D raises calcium, she had ten seconds of bilateral lip and tongue tingling on 11 Sep which is the picture of low calcium or magnesium, and no calcium result has ever been seen, so raising the dose blind is pushing on a dial nobody has read. Vitamin D level added to the Wednesday bloods ask alongside calcium, magnesium, kidney function, cortisol and sodium. Cancer evidence unchanged: VITAL (n=25,871, 2,000 IU/day) no reduction in cancer incidence; D-Health (n=21,315) no reduction in cancer mortality; nothing in chordoma.

That selectivity is precisely why radiotherapy is the question: proton therapy kills chordoma partly via hydroxyl radicals, which is the species H2 scavenges, so on mechanism it could shield the tumour. Countervailing evidence is Kang 2011 (PMID 22146004), n=49 on radiotherapy for liver tumours: better quality of life, reduced oxidative damage markers, and no difference in tumour response. Reassuring in direction, but 49 patients, one tumour type, not chordoma.

Position restated and unchanged from 11 Sep: acceptable now, ask the proton centre before using it around radiotherapy; question to be written for the proton centre list.

Three conditions given, the first time critical. (1) Measure serum magnesium BEFORE starting: the tablets are elemental magnesium plus an acid, roughly 80 mg elemental magnesium per tablet, so starting them destroys the ability to establish whether she was deficient, and the bilateral perioral tingling of 11 Sep remains unexplained. Bloods first. (2) Count that 80 mg against any separate magnesium supplement, with the renal caution: magnesium is renally cleared, she is on diclofenac at 51 kg, and her creatinine has never been seen. (3) Keep two hours clear of gabapentin, which magnesium reduces the absorption of.

13 September 2026, 09:17 — her partner asked about sequencing hydrogen and hyperbaric oxygen around proton therapy

Question: "would it be reasonable to start use hydrogen up until 2 weeks before proton beam and then use HBOT after to help with healing".

Answered in the channel and written up as em-docs/em-hydrogen-hbot.md (+PDF) with full working in em-research/22-hydrogen-hbot-around-radiotherapy.md. There was no HBOT material anywhere in the repo before today.

Three findings. (1) Stopping hydrogen before radiotherapy is right, and the decisive evidence is not about hydrogen: Bairati 2005 (PMID 15812073, L1, n=540) gave vitamin E through head and neck radiotherapy and found fewer acute side effects but HR 2.88 for second primaries and HR 1.86 for recurrence-or-second-primary during the supplementation period, reversing after it stopped. The only hydrogen trial that bears on it (Kang 2011, PMID 22146004, n=49) found no difference in tumour response, which points the reassuring way but is small. (2) The plan as written had a gap: a chordoma proton course is seven to eight weeks, so "stop two weeks before" leaves hydrogen running through the treatment itself. Corrected to off for the whole course. Two weeks conceded as a sensible margin, since the gas clears in hours but the Nrf2 induction takes days to a week. (3) HBOT is the weaker half, not the stronger: Cochrane 2023 (PMID 37585677, 18 trials, 1,071 patients) found benefit for head and neck bone and soft tissue and for cystitis and proctitis, new evidence that it may NOT help osteoradionecrosis, and explicitly no important effect on neurological tissue, which is the tissue the patient would most want protected; and HOPON (PMID 30851351, L1) tested prophylactic HBOT and found 6.4 vs 5.7 percent, nothing.

The genuinely the patient-specific finding, which a generic answer would miss: she may be a high-risk HBOT patient. Compression requires every air space in the head to equalise, and she has a mass growing towards the sphenoid sinus, a right nostril blocked since 7 Sep, an open CSF leak question, and the prospect of a surgically repaired skull base defect. Unrepaired skull base defect, CSF leak and pneumocephalus are treated as contraindications; tension pneumocephalus after a single session is documented (PMID 19371560). Plus four possible seizure episodes against a background CNS oxygen toxicity seizure rate of roughly 1 in 2,000 to 10,000 sessions. Framed as a joint decision for the skull base surgeon and a hyperbaric physician, not a private booking.

Conceded to her partner where he was right: protons deposit more of their energy as direct, clustered DNA damage at the Bragg peak than photons do, so the radical-scavenging worry really is smaller with protons. Smaller, not absent, and never measured in a patient.

Also rebuilt em-syrris-appointment-sheet.pdf, which had been stale since several rounds of additions, and added questions 18 to 20 (antioxidant list during protons, whether a skull base repair contraindicates HBOT, who decides) plus a "things to tell him" line saying the family is reading around adjuncts and wants to do it with the team rather than around them.

13 September 2026, 09:27 — her partner quoted a result and asked about the p values

Quoted line, no source given: "A higher number of adjuvant chemotherapy cycles was also related to longer PFS and OS (p = 0.049 and p = 0.030, respectively)", then "These p values" and "Been a while since I did stat analysis".

Paper NOT identified. Searched for it; the closest candidate was Xu 2026, "Institutional experience and pooled survival analysis of chemotherapy based multimodal management in poorly differentiated chordoma" (Discover Oncology, PMID 42165926, 5 institutional plus 25 pooled = 30 patients, univariable only, no multiplicity correction), but its p values are 0.029, 0.041 and 0.009, so it is a different paper. Asked her partner for the title or link rather than guessing. That candidate is worth holding anyway: it is a 2026 pooled analysis of chemotherapy in poorly differentiated chordoma, anthracycline and ifosfamide, all INI1/SMARCB1 lost, which is one of the branches that applies to the patient.

Answered the statistics question in full, since that is what he actually asked. Four points: (1) 0.049 and 0.030 are the weakest results that still count as positive, one clearing by a thousandth; (2) p is the chance of seeing a result this extreme if nothing is there, not the probability the finding is true; (3) multiple comparisons, a dozen-plus variables tested against survival with no correction means one or two marginal hits is what noise looks like; (4) and the structural one that matters most, immortal time bias, for which "more cycles means longer survival" is the textbook example: you must be alive to receive eight cycles, so survival partly causes the cycle count. Plus confounding by indication, since cycles continue in patients tolerating and responding. Said plainly that I would disbelieve this finding at p = 0.001 unless cycles were handled as a time-dependent covariate or by landmark analysis, in a multivariable model.

Flagged the branch that decides relevance: chemotherapy does essentially nothing in conventional chordoma, but dedifferentiated and poorly differentiated chordoma are treated as sarcomas and the cycle question is live there.

Identified, 09:29. her partner: "This is from a paper Chris Syrris authored... trying to get through some of his work before Wednesday." The paper is Ashkan, Baig Mirza, Soumpasis, Syrris et al., "MGMT Promoter Methylation: Prognostication beyond Treatment Response", J Pers Med 2023 Jun 14;13(6):999, PMID 37373988, King's College Hospital; Mr Syrris is fourth author. It is a glioblastoma study, n=69, operated with 5-ALA, and "adjuvant chemotherapy" means temozolomide. Nothing in it transfers to the patient clinically.

Three things given back to her partner. (1) The immortal time bias objection is not hypothetical here, it is the settled answer in this literature: a meta-analysis of four randomised trials, 882 patients, six versus up to twelve TMZ cycles, found no OS benefit (23 vs 24 months, p = 0.73) and no significant PFS benefit (p = 0.27) (Attarian 2021, Front Oncol, PMID 34900732). Retrospective association says more cycles help; randomised evidence says extending cycles does nothing. That gap is the bias. (2) In fairness to the paper, the cycles sentence is a secondary unadjusted by-the-way. The load-bearing claim, MGMT methylation treated as a continuous variable, was handled properly: p = 0.008 and 0.006, holding at p = 0.034 and 0.042 after adjustment for extent of resection. Said so explicitly so her partner does not walk into Wednesday thinking his surgeon's paper is poor work.

(3) The more useful find, which is directly relevant to the patient: Mr Syrris is a co-author on Baig Mirza, Boardman, Okasha, Syrris et al., "Fat in the Fossa and the Sphenoid Sinus: A Simple and Effective Solution to CSF Leaks in Transsphenoidal Surgery", J Neurol Surg B Skull Base 2023;84(2):143-156, PMID 36895808, 439 patients at King's. Their fat-in-fossa-and-sphenoid technique cut postoperative CSF leak requiring intervention to 4.4 percent against 20.3 percent for multilayer repair and 12.6 percent for no repair, with fewer reoperations, fewer lumbar drains and shorter stays. The patient has an open CSF leak question and may have endonasal surgery, so this has been added to the appointment sheet as question 5a.

His wider publication list (14 papers, all King's neurosurgery): glioma with fluorescence guidance and extent of resection, transsphenoidal and skull base, a proposed replacement for the Knosp classification in acromegaly, chronic subdural outcomes, lumbar drain infection, trigeminal neuralgia after MVD. An outcomes-and-numbers surgeon working in exactly the patient's corridor, which also means the quantitative questions on the sheet will land rather than irritate. Note: "Syrris P" in PubMed is Petros Syrris, a cardiac geneticist, and is a different person.

13 September 2026, 23:10 — her partner asked about Essiac tea

Answered no, without qualification, and written up as an addendum to em-docs/em-plant-compounds.md. This is the first item in that file where the objection goes beyond absence of evidence.

No human trial has ever shown Essiac does anything to a cancer; it was tested and failed at Memorial Sloan Kettering and the NCI between the 1950s and 1980s, at UBC in 2005 and at Michigan in 2006, and the Natural Standard systematic review found no high quality trial of any traditional use (PMID 19476742). The one substantial human study, TEA-BC, a retrospective cohort of 510 Ontario women with breast cancer, found no quality of life or mood benefit, with users scoring worse on two subscales, which the authors correctly attribute to confounding since users were younger with more advanced disease (PMID 17212569).

The decisive finding is Kulp 2006 (PMID 16541326): Essiac and Flor-Essence stimulated proliferation of all four human breast cancer cell lines tested, ER positive and ER negative, through both ER mediated and ER independent mechanisms. Cell culture, breast cancer not chordoma, but it is the only directional signal in the literature and it points the wrong way.

Three the patient-specific hazards given more weight than the efficacy question: oxalate from sheep sorrel and rhubarb root, in a 51 kg woman on a daily NSAID whose creatinine the family has never been shown; additive anthraquinone laxative effect from rhubarb root on top of two senna nightly, thirty six hours after fresh rectal bleeding; and slippery elm mucilage blunting absorption of her modified release morphine, modified release diclofenac and gabapentin.

Told him explicitly that he has been right to push on other items, vitamin D and the NAD+ objection especially, and that this is not one of them. Saying no clearly is worth more to him than hedging.

14 September 2026, 17:00. Psilocybin as a painkiller (a friend, em-research)

A friend asked for the research on psilocybin for pain. Written up as brief em-docs/em-psilocybin-pain.md and research thread 23; both on the site.

Appraisal. No randomised trial anywhere shows psilocybin lowers pain intensity in any patient group. The direct experimental work used LSD in healthy volunteers: positive at 20 micrograms (PMID 32842825), null at 15 micrograms (PMID 40917239). The clinical signal is confined to migraine and cluster headache from a single Yale group, forty-two patients across three randomised trials, of which one was nominally positive (PMID 33184743), one was negative on its primary outcome by the authors' own statement (PMID 36416492), and the newest, with an active placebo, found diphenhydramine 25 mg did as well as the drug (PMID 41459830). The four good cancer trials measured depression and anxiety, not pain.

The decisive point is not efficacy, safety or legality. Psilocybin's proposed mechanism in pain is acceptance, cognitive flexibility and a changed relationship to the signal, which is the affective dimension. That is precisely what the patient objected to about morphine on 13 September in her own words. The steelman, that a durable one-off shift differs from daily blunting, is stated in the brief and does not carry on twenty-four patients in a different disease.

Safety specifics for the patient: raised blood pressure against a partly encased basilar artery, convulsions in the 4 per cent serious-adverse-event rate among people with pre-existing neurological conditions (PMID 39230883) against four possible seizure episodes and no EEG, vomiting against a possible CSF leak. Serotonin interaction risk with her current list is low.

Legal position: Schedule 1, unprescribable, no compassionate route. Registry checked 14 September: three UK psilocybin trials recruiting, none in pain.

Verdict: no, and the week is better spent on the four untried conventional options (gabapentin titration, dexamethasone, sphenopalatine ganglion block, opioid rotation).

15 September 2026, 10:12. Dal Bello 2022, ketogenic diet in gliomas (her partner, em-research)

File: 2026-09-15_1012_PMID 36145228_dal-bello-2022-ketogenic-diet-gliomas.jpg. Dal Bello S et al. Ketogenic Diet in the Treatment of Gliomas and Glioblastomas. Nutrients 2022;14(18):3851. PMID 36145228, PMC9504425. Full text read.

Type: narrative review, not systematic; no search strategy, no risk-of-bias assessment; 91 references. Level: summarises L5 animal work and L3/L4 small uncontrolled human series.

What it fairly shows: in glioma mouse models the picture is more positive than thread 24 gave it credit for. Several models showed slower growth or longer survival, mostly at 4:1 to 6:1 ratios or combined with calorie restriction, radiotherapy, bevacizumab or the glutamine blocker DON. 3:1 diets showed no effect; medulloblastoma models showed no benefit at any ratio; rat gliomas upregulated MCT1 and oxidised ketones (De Feyter 2016), which the authors flag as contradicting metabolic rigidity. Human data: ERGO 2014 (20 patients, all progressed), and series of 6, 8, 9, 11 and 12 patients; tolerable, some quality of life and seizure benefit, no controlled survival evidence.

Limits: does not cite ERGO2 (Voss 2020, the randomised trial, negative) or Sperry 2020, both published before it. Much of the animal benefit is entangled with calorie restriction, which is exactly what the patient at 51 kg cannot do. Glioma lives in glucose-dependent brain; chordoma is a bone tumour, and chondrosarcoma, her main lookalike, burns ketones (Vargas-Lopez 2024). The authors' own conclusion names metabolic plasticity as the unaddressed problem.

Verdict: a fair concession owed on glioma mouse data; does not change the position for the patient. Keep the fat for calories, do not pursue ketosis as treatment, keep calorie restriction off the table. Threads 24 and 25.

2026-09-24 21:05 Newell 2021, n-3 LCPUFA during cancer treatment (PMID 33801979, Cancers 13:1206)

Her partner pasted the MDPI page header. It's a scoping review of 57 clinical trials to Aug 2020, all small and mixed, with no chordoma or bone sarcoma. It reports weight maintained or gained, better quality of life, and signals for survival and fewer serious adverse events. The authors themselves call for RCTs on dose and type. Verdict: the weight and cachexia benefit is the real one and matches ESPEN 2021 (PMID 33946039). There's no tumour effect for the patient. Advice given: start after the 29 Sep biopsy, not before. OPERA (PMID 23128104, cardiac surgery, perioperative fish oil) found no extra bleeding, but nothing goes in new before the procedure without the team knowing. Oily fish is already in the food plan. Follow-up, same evening. Her partner asked about the claim that "DHA concentrates in tumour cells and triggers apoptosis". Real, but preclinical. The cell death is ferroptosis in acidic tumour cells (Dierge 2021, Cell Metab, PMID 34118189, cells plus mice). The best human data is the Bougnoux 2009 single-arm phase II (PMID 19920822, n=25, metastatic breast cancer, 1.8 g DHA a day with anthracycline chemotherapy; patients with higher plasma DHA did better), as a sensitiser to chemo, not alone. PubMed "chordoma docosahexaenoic" gives 0 hits. Told her partner it's reasonable after the biopsy for weight, with no effect on treatment decisions, and to tell the radiotherapy planners.

Research thread RE · source file em-research/mark-shared/README.md · last edited 14 October 2026