Dalgleish, vitamin D and adjuncts

Written 7 Sep 2026 for her partner's question. Sources are PubMed (searched by author and topic on 7 Sep 2026) plus the public pages named in the text. Where a claim is his public opinion rather than a published study, it says so. Web search was rate-limited; a few public pages could not be read and are marked as such.

1. Who he is

Angus Dalgleish is Professor of Oncology at St George's, University of London, and a consultant medical oncologist (melanoma and immunotherapy). His scientific standing rests on real work: he was first author on the 1984 Nature paper showing CD4 is the receptor for HIV (PMID 6096719), he is a Fellow of the Academy of Medical Sciences, and he has a 40-year record in cancer immunotherapy, mainly mycobacterial vaccines (SRL172 and then IMM-101), with over 300 PubMed entries.

Things to weigh alongside that:

• Commercial interests. He founded Onyvax (cancer vaccines, dissolved 2013), sits on the scientific advisory board of Immodulon (makers of IMM-101), is a named inventor on patents for low-dose naltrexone (LDN) in cancer, and has received grants from LDN Pharma, Jay Pharma (a cannabinoid company) and Celgene (his own conflict statements, PMIDs 35761946, 29500634, 31788420). He held stock options in Immunor, whose COVID vaccine candidate he co-authored (PMID 34192262). None of this makes his science wrong; it means his enthusiasm for LDN, cannabinoids and IMM-101 is not disinterested.
• Politics and COVID. He stood for UKIP in 2015 and campaigned for Leave. During the pandemic he promoted the lab-leak theory, co-wrote a paper arguing that BCG or M. obuense could protect against COVID (PMID 33013871), and later said publicly that COVID vaccines were "largely ineffective" and, via a colleague at a 2025 political conference, that they were "highly likely" a factor in cancers in the royal family. There are no peer-reviewed data behind the vaccine-cancer claim.
• Style. In his public writing he treats observational associations as settled and says randomised trials are "unnecessary" (see section 3). That is the opposite of how evidence normally works, and it matters when weighing his advice.

2. His relevant peer-reviewed papers

Searched and not found: any paper by him on vitamin D in human patients, on vitamin D and innate immunity, on vitamin D levels and immunotherapy response, or on vitamin C (zero PubMed results for Dalgleish plus ascorbate).

3. Vitamin D: what he has published vs what he says

Published. Three cell-line papers (2000 to 2012) on vitamin D analogues in pancreatic and breast cancer cells, one of which found the analogue blocked the cancer-cell death caused by another drug. Nothing on innate immunity. Nothing in patients.

Said publicly (not peer-reviewed). In a signed article for Positive Health Online (June 2019) he writes that 85% of patients in his melanoma clinic had 25(OH)D below the normal range, that patients with low vitamin D "respond poorly to immunotherapy" and to chemotherapy, that vitamin D is "vital for an effective immune response" (dendritic cell antigen presentation, killer cells), that randomised trials of supplementation are "unnecessary", and that every cancer doctor should "measure the vitamin D3 level and correct it before doing anything else". In the Daily Mail (28 May 2019, Lois Rogers) he is quoted: "We have more than 2,000 cancer patients where we have shown this and we have published several papers." PubMed has no such paper under his name. The Newton-Bishop melanoma finding he cites is real and independent: lower 25(OH)D at diagnosis was linked to thicker melanomas and worse survival in 872 patients (PMID 19770375), but that is an association, not proof that correcting it changes the outcome. The innate immunity biology is also real and independent of him: vitamin D is needed for macrophages to make the antimicrobial peptide cathelicidin when Toll-like receptors fire (Science 2006, PMID 16497887). A target of 100 nmol/L and a 10,000 IU dose are attributed to him on some forums; I could not read those pages and cannot verify he said it.

The strongest independent evidence.

• VITAL (PMID 30415629): 25,871 US adults, 2,000 IU/day vs placebo, 5.3 years. No reduction in cancer incidence (HR 0.96). A secondary analysis (PMID 33206192) found fewer metastatic or fatal cancers (HR 0.83, 95% CI 0.69 to 0.99), confined to people of normal weight.
• D-Health (PMID 35026158): 21,315 older Australians, 60,000 IU once a month, 5.7 years; achieved 115 nmol/L vs 77 in placebo. No reduction in deaths. Cancer deaths trended the wrong way (HR 1.15, 95% CI 0.96 to 1.39; 1.24 after excluding the first two years). The authors advise against this regimen in people who are already replete. A later paper reported no reduction in cancer incidence (PMID 40096917).
• Meta-analyses of randomised trials (PMIDs 30796437, 35676320, 24918818): no effect on getting cancer (RR 0.98), including in trials where blood levels exceeded 100 nmol/L (RR 0.95); a modest reduction in dying of cancer (RR 0.87, 95% CI 0.79 to 0.96), seen with daily dosing rather than large monthly boluses.

So the honest summary is: correcting deficiency is reasonable and cheap, daily dosing carries a small mortality signal, high-dose monthly bolus dosing in replete people did not help and may have harmed, and nothing here is specific to chordoma (PubMed: chordoma plus vitamin D returns zero papers).

The 100 nmol/L target. Thresholds: UK (SACN, NHS labs) call under 25 nmol/L deficient, 25 to 50 insufficient, over 50 sufficient. The US Institute of Medicine puts 50 nmol/L as adequate for 97.5% of people and 4,000 IU/day as the safe upper intake (PMID 21118827). The Endocrine Society in 2011 preferred 75 nmol/L and allowed up to 10,000 IU/day under supervision (PMID 21646368); its 2024 guideline reversed course and recommends against routine testing and against targeting levels above the IOM thresholds in healthy adults under 75 (PMID 38828931). Some advocates argue for 75 to 125 nmol/L (PMID 18844846). Toxicity: hypercalcaemia is the hazard; cases are reported from about 220 to 250 nmol/L upward and are common above 375 to 500 (PMIDs 18290725, 21646368). Dose to reach and hold 100: roughly 2.5 IU/kg per 1 ng/mL (2.5 nmol/L) rise (PMID 24037880); at 51 kg that is about 1,000 IU/day per 20 nmol/L, so going from 40 to 100 nmol/L needs about 3,000 IU/day, at the top of what the NHS allows without a doctor's supervision. Evidence that 100 beats 50: none. D-Health reached 115 nmol/L with no benefit and a cancer-death signal in the wrong direction; the meta-analysis subgroup above 100 nmol/L was null for incidence. Aiming for 100 is opinion; aiming for 50 to 75 with a blood test is evidence-based.

4. The compound that "reduces the chemotherapy dose"

What it is. Almost certainly the cannabinoid work with Wai Liu: Scott, Dalgleish and Liu, Int J Oncol 2017 (PMID 28560402), building on Anticancer Res 2013 (PMID 24123005). In HL60 leukaemia cells, pairs of cannabinoids (CBD plus THC, and others) were synergistic with cytarabine and vincristine, "allowing the dose of the cytotoxic agents to be dramatically reduced yet still remain efficacious", provided the cannabinoids were given after the chemotherapy. The 2022 paper adds LDN before CBD as a chemosensitiser in cell lines and mice (PMID 35179218). Everything here is dishes and mice. There is no human trial of cannabinoids lowering a chemotherapy dose in any cancer. An independent 2026 cell-line study did the same kind of experiment in osteosarcoma and chondrosarcoma cells with cisplatin plus THC and melatonin, 2.8 to 4.4-fold "dose reduction index" (PMID 42362763); again cells only, and chondrosarcoma is not chordoma.

Radiotherapy, which is her likely treatment. The same group showed CBD plus THC before irradiation increased radiosensitivity in glioma cells and shrank tumours in an orthotopic mouse model (PMID 25398831), and that heat shock protein inhibitors improve CBD's effect (PMID 26504004). Independent groups reproduced CBD plus radiation synergy in glioblastoma cells (PMIDs 29088769, 30783513). The counter-case: in mouse models of medulloblastoma and ependymoma, in-vitro synergy did not translate into any survival benefit in vivo (PMID 33477420). Human data are thin: two uncontrolled Austrian case series (9 and 15 patients) giving CBD 400 to 600 mg/day alongside standard radiochemotherapy for glioblastoma reported longer survival than expected, with no control arm (PMIDs 31570484, 35403130); and one randomised phase 1b in 21 patients with recurrent glioblastoma (nabiximols spray plus temozolomide vs placebo) was safe with an exploratory survival signal it was not powered to test (PMID 33623076). None of this involves protons, chordoma, or the skull base.

Conditional on her histology.

• If conventional (or chondroid) chordoma, the commonest outcome: chemotherapy is not standard because conventional chordoma is insensitive to cytotoxic drugs (PMIDs 28750274, 38691867). A compound that lowers the dose of a drug she will not be given has no bearing on her treatment. The only transferable idea is the radiosensitisation one, and that has never been tested in a human chordoma or with proton beam.
• If dedifferentiated chordoma (a sarcoma-like component; 7 of 86 in one sequenced cohort enriched for advanced disease, PMID 40802536; usually quoted under 5%): sarcoma-type chemotherapy (doxorubicin, ifosfamide) is offered. Two complete remissions to ifosfamide or a six-drug regimen were reported in 1993 (PMID 8334623); a pooled analysis of 130 dedifferentiated cases found chemotherapy receipt did not separate outcomes and that complete resection and radiotherapy were what predicted survival (PMID 36804481). Here the cannabinoid question would at least be about a drug she might receive, but the answer is still: cell-line data only, leukaemia lines, different drugs, nothing in a human sarcoma.
• If poorly differentiated (SMARCB1/INI1-deficient) chordoma: a distinct, aggressive, mainly paediatric entity (PMID 27067307); in the largest series, 19 patients, median age 11, range 1 to 29, 53% at the clivus, treated with surgery, radiotherapy and chemotherapy together, with worse survival than other subtypes (PMID 29483606; also 40 French children, PMID 29361006). Adult cases at 34 are rare but the 5 Sep rapid-growth note already flagged the subtype. Chemotherapy is used, and the EZH2 inhibitor tazemetostat has produced responses in INI1-deficient tumours including poorly differentiated chordoma (PMIDs 29141919, 30642912); MD Anderson's protocol routes INI1-deficient chordoma to targeted-therapy trials (PMID 38691867). Cannabinoid data in this setting: none.

  [Correction 7 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.]

5. IMM-101

Heat-killed Mycobacterium obuense, injected intradermally, developed by Immodulon with Dalgleish as adviser and trialist. Evidence: phase 1 in melanoma, n=18, safe (PMID 21930686). IMAGE-1, randomised open-label phase 2 in 110 patients with advanced pancreatic cancer: gemcitabine plus IMM-101 vs gemcitabine, median survival 6.7 vs 5.6 months, not significant; significant in the pre-defined metastatic subgroup, 7.0 vs 4.4 months (PMID 27599039). Four melanoma patients did well with IMM-101 plus checkpoint inhibitors (PMID 30107850). Six prostate patients on IMM-101 plus bromelain and/or LDN, uncontrolled (PMID 35761946). Radiotherapy combination data exist: an Erasmus single-arm trial (LAPC-2) gave stereotactic radiotherapy plus IMM-101 to 38 patients with locally advanced pancreatic cancer; safe, no IMM-101-related serious events, median survival 19.0 months, "comparable" to their earlier trial of radiotherapy alone (PMID 36813171), with an immunology substudy showing T and NK cell activation (PMID 36358718) and stable quality of life (PMID 39888694). So it is safe with photon radiotherapy and there is no evidence it adds benefit. No phase 3 result, no licence anywhere, no trial in sarcoma, chordoma or any bone tumour (PubMed search returned none). It is available only in trials or on a named-patient basis. One report of M. obuense bloodstream infection in a pneumonia patient shows the live organism is not wholly harmless, though the product is heat-killed (PMID 31002048). Relevance to the patient: none direct.

Company status (verified 7 Sep 2026, 01:05 UK)

Immodulon Therapeutics Limited (company 06223291) entered administration on 17 December 2024 (Companies House filing AM01; administrators Quantuma Advisory; progress reports January and July 2026; still in administration). ClinicalTrials.gov (API query 7 Sep 2026): the company's own trials are all completed or terminated, the last being IMM-101 with checkpoint inhibitors in melanoma (NCT03711188, completed December 2021, no results posted); the expanded-access programme for pancreatic cancer (NCT04137822) is listed as no longer available; an investigator-led phase 2 of gemcitabine, pembrolizumab and IMM-101 in pancreatic cancer (NCT06498518) was withdrawn in May 2025 before recruiting; a Canadian phase 3 of IMM-101 for COVID-19 prevention (NCT04442048) completed December 2025. Interpretation: a missed primary endpoint in 2016, no funded phase 3, and insolvency in 2024; no safety signal in the record. The product currently has no manufacturer and no active supply route.

6. Vitamin C

Dalgleish has no publications on it (PubMed, his name plus ascorbate: zero).

Oral. Absorption saturates: blood levels plateau at about 70 to 80 micromol/L on 200 to 400 mg/day and doses above 1 g add little (PMID 15068981). Two double-blind randomised trials of 10 g/day oral vitamin C in advanced cancer (150 and 100 patients) showed no benefit in symptoms or survival (PMIDs 384241, 3880867). Up to 1 g/day is safe; more causes loose stools and adds to oxalate load. There is no cancer-outcome evidence for oral vitamin C at any dose.

High-dose intravenous ("pharmacological ascorbate"). The rationale comes from Chen, Levine and colleagues: at millimolar concentrations, reachable only by infusion, ascorbate acts as a pro-drug that generates hydrogen peroxide in the tissue fluid and kills cancer cells selectively in the dish (PNAS 2005, PMID 16157892), and slows aggressive xenografts in mice (PNAS 2008, PMID 18678913). Human studies, in order of strength: phase 1 with gemcitabine and erlotinib in 14 pancreatic patients, no added toxicity (Monti, PLoS One 2012, PMID 22272248); phase 1 with gemcitabine, n=9 completing, well tolerated (PMID 23381814); phase 1/2a in ovarian cancer, n=27, less chemotherapy toxicity, survival not the endpoint (Ma, Sci Transl Med 2014, PMID 24500406); mechanism plus phase 1/2a in pancreatic cancer (Polireddy, Sci Rep 2017, PMID 29215048); and, most relevant to her because it involves brain tumours and radiotherapy, the Iowa work: ascorbate with radiotherapy plus temozolomide in glioblastoma, 11 patients, safe, median survival 18 months against a historical 14 to 16, alongside a lung cancer arm (Schoenfeld, Cancer Cell 2017, PMID 28366679; the full phase 1 report PMID 31427282). The Iowa single-arm phase 2 in glioblastoma has now reported: median survival 19.6 months vs 14.6 in historical controls, with an MRI iron marker predicting who benefits (PMID 37773633; imaging companion papers PMIDs 33260088, 38285244). Historical-control comparisons in glioblastoma are unreliable; this needs a randomised trial, which has not been done. Randomised evidence overall: one small positive trial (metastatic pancreatic cancer, n=34, survival 16 vs 8.3 months, PMID 39369582) against two larger negative ones, a phase 3 in 442 colorectal patients with no progression-free survival benefit (PMID 35929990) and a placebo-controlled phase 2 in prostate cancer stopped for futility with survival numerically worse on vitamin C (PMID 39076107). No randomised trial has shown that intravenous vitamin C prolongs survival. A systematic review found the best-supported effect is on quality of life and chemotherapy tolerance, not on the cancer (PMID 24867961).

Safety. Haemolysis and methaemoglobinaemia in G6PD deficiency, so a G6PD test is mandatory before any infusion (PMID 32561141); oxalate kidney injury in anyone with reduced kidney function; falsely high glucose-meter readings; and a laboratory signal for pro-thrombotic red-cell changes (PMID 26139164).

With radiotherapy: contested. At pharmacological IV doses ascorbate is a pro-oxidant; in animal models it sensitised pancreatic tumours to radiation while protecting normal bowel (PMID 30254147), and the glioblastoma studies above gave it during photon radiotherapy without excess toxicity. At ordinary oral doses it is an antioxidant, and the long-standing caution is that antioxidant supplements may blunt the free-radical damage radiotherapy relies on; most radiotherapy units advise against high-dose antioxidants during a course. Both positions rest on mechanism and small studies; neither has proton data, and nothing exists in chordoma.

7. Bottom line for the patient, plainly

What transfers. One thing: check her 25-hydroxyvitamin D and, if it is under 50 nmol/L, replete it with daily vitamin D3, 2,000 to 4,000 IU (50 to 100 micrograms) a day, and re-test in 8 to 12 weeks. Aim for 50 to 75 nmol/L. That is supported by the trials and by every guideline; it costs pennies; and after months of pain, poor intake and being indoors she is likely to be low. Ask the GP to add it to Monday's bloods. Do not exceed 4,000 IU/day without a doctor and a blood level. Do not chase 100 nmol/L; the one large trial that got there found no benefit and a possible harm signal.

What does not transfer. Cannabinoids to lower a chemotherapy dose: cell-line data in leukaemia, no human trial anywhere, and chemotherapy is not on her table unless histology shows a dedifferentiated or poorly differentiated tumour, in which case it is still cell-line data. Cannabinoids as radiosensitisers: mice and a handful of uncontrolled glioblastoma patients, no proton data, no chordoma data. IMM-101: safe, unproven, pancreatic and melanoma only, trials or named-patient only, nothing in bone tumours. LDN: cell lines and case series; blocks morphine, so it is incompatible with her current pain control (see research 13). Vitamin C: oral doses above about 500 mg add nothing measurable; high-dose IV is a trial intervention with two negative large randomised trials and one positive small one, none in her tumour.

Safety cautions specific to her.

• Vitamin D excess causes hypercalcaemia: nausea, vomiting, constipation, thirst, passing lots of urine, tiredness, confusion, drowsiness, weakness. Every one of those overlaps with morphine and gabapentin side effects and with the neurological signs the team is watching. At 2,000 to 4,000 IU/day this is not a realistic risk; at 10,000 IU/day for months without a blood test it is. Anyone giving her calcium or "vitamin D plus K2" high-dose combinations should stop and check a level first.
• Cannabinoids (THC especially, CBD at high dose too) sedate and slow thinking. She is on morphine by the clock and is starting gabapentin. Stacking a third sedative makes it harder to tell drug effect from tumour effect, worsens balance, and THC can affect eye movements in someone who already has ptosis and double vision. CBD also inhibits liver enzymes that clear several drugs; morphine and gabapentin are not much affected, but anything added later might be. If she wants CBD for symptoms, that is a conversation with the prescribing team, with the diary running, and not during the window when they are assessing her nerves.
• High-dose IV vitamin C: G6PD test first, kidney function first, never during the radiotherapy course without the radiation oncologist's agreement.

Word count about 2,950 plus the table, above the 900 to 1,300 asked for, because three sections were added mid-task (subtype-conditional chemotherapy, the 100 nmol/L target, and vitamin C with named trials).