Psilocybin and pain

Agent report for Kya, 14 September 2026, written after a friend asked in em-research for research on psilocybin as a painkiller. Working frame: probable clival chordoma in a 34-year-old woman, 51 kg, histology pending, severe trigeminal-distribution neuropathic pain scoring 7 to 9 daily on 50 to 90 mg of oral morphine a day, four possible seizure episodes since 5 September, a mass partly encasing the basilar artery and abutting both cavernous carotids, and a tumour growing forwards towards the sphenoid sinus. Every PMID verified through NCBI E-utilities on 14 September 2026. Trial registry checked the same day. No survival figures, by rule. Clinical reasoning is labelled.

Evidence levels. L1 randomised trial, meta-analysis or guideline. L2 prospective non-randomised trial or large cohort. L3 retrospective series. L4 case reports or very small series. L5 laboratory, animal or cell-line work, and expert reasoning.

a friend's categories. This sits between frontier and anecdotal. It has a small number of real randomised trials, all exploratory, none in cancer pain, and a much larger volume of enthusiastic secondary literature that reads as if the trials were positive when mostly they were not.

Tier A. The question and the short answer

Is psilocybin a painkiller?

On the current evidence: no, not in the sense that word normally carries, and not for the kind of pain the patient has. There is no randomised trial of psilocybin in cancer pain, none in neuropathic pain, and none in any pain at all where the primary endpoint was pain intensity and the result was positive.

What there is, in descending order of quality:

1. Two small randomised trials in primary headache disorders (migraine and cluster headache) from one Yale group, both explicitly labelled exploratory, one nominally positive on a secondary-sized sample, one negative on its primary outcome, and a 2026 follow-up in which an active placebo performed as well as the drug.
2. Two well-conducted randomised trials in cancer with large and durable effects, on depression and anxiety, not on pain.
3. Experimental pain work in healthy volunteers using LSD rather than psilocybin, one positive at 20 µg, one null at 15 µg.
4. A large historical literature from the 1960s on LSD in terminally ill patients that is uncontrolled, unindexed and unusable as evidence.

The 2026 critical review in Drugs is the fairest summary available and its own conclusion is that the field "is limited by small sample sizes, functional unblinding, and a lack of large, well-controlled randomized trials" (Boehnke 2026, PMID 42243569, L1 review of L2 to L5 material).

The finding that matters most for the patient is not about efficacy at all. It is that the mechanism psilocybin is proposed to work through is the same mechanism the patient has already told us she does not want. That argument is in Tier D and it is the reason this thread does not end in a recommendation.

Tier B. What the analgesia claim actually rests on

The historical layer. Kast's work in the mid 1960s gave LSD to patients dying of cancer and reported analgesia outlasting dihydromorphinone and pethidine by days. It is cited constantly. It is also uncontrolled, unblinded, unrandomised, published in journals that are not indexed in PubMed, and conducted in an era with no adverse-event standards. The modern authors who revisit it say so themselves: Ramaekers 2021 opens by noting that "clinical studies in the 1960s and 1970s have demonstrated profound analgesic effects of full doses of LSD in terminally ill patients, but this line of research evaporated after LSD was scheduled worldwide" (PMID 32842825, L4 at best). Treat it as the origin of the hypothesis, not as support for it.

The modern experimental layer, which is LSD and not psilocybin.

• Ramaekers 2021 (J Psychopharmacol, PMID 32842825, L1, n = 24 healthy volunteers, within-subject, placebo controlled). Single doses of 5, 10 and 20 µg LSD. 20 µg significantly increased cold pressor tolerance and reduced rated painfulness and unpleasantness, at 1.5 and 5 hours. Mean blood pressure rose, within the normal range. This is the single cleanest positive result anywhere in the field.
• Cavarra 2025 (Br J Pain, PMID 40917239, L1, randomised, placebo controlled, repeated dosing). 15 µg LSD, four administrations. No analgesic effect on any measure in the whole sample. A post-hoc subgroup without ceiling effects showed an increase in tolerance after the first dose only. The authors' conclusion is that 15 µg is probably below threshold.

So the best-designed direct tests of psychedelic analgesia are two cold-pressor studies in healthy people using a different drug, with one positive at one dose level and one null at a lower dose. Cold pressor tolerance in a healthy 25-year-old is a long way from a 3.8 cm mass on the clivus (clinical reasoning, L5).

The mechanistic layer. Castellanos 2020 (Reg Anesth Pain Med, PMID 32371500, L5) proposed the standard model: 5-HT2A agonism drives cortical neuroplasticity and reorganises the default mode network and the salience network, which in chronic pain are held in a maladaptive configuration. Boehnke 2026 (PMID 42243569) extends this to neuroplasticity, anti-inflammatory effects, large-scale network dynamics, and "higher-order psychological processes, such as pain acceptance and cognitive flexibility".

Read that list again. Three of the four proposed mechanisms are about how the brain relates to the pain signal. Only the anti-inflammatory strand is about the signal itself, and it is the weakest of the four (clinical reasoning, L5).

Tier C. The headache evidence, which is the only real clinical signal, and why it does not transfer

This is the part worth taking seriously, because it is the one place the observation came from patients first and the trials followed.

Sewell 2006 (Neurology, PMID 16801660, L3). Retrospective interviews with 53 cluster headache patients who had used psilocybin or LSD. Most reported attack termination, cluster period termination or remission extension. Uncontrolled, self-selected, recall-based. It launched the field.

Schindler 2021, migraine (Neurotherapeutics, PMID 33184743, L1, n = 10, double-blind, placebo-controlled crossover). Single oral psilocybin 0.143 mg/kg versus placebo. Weekly migraine days fell by 1.65 after psilocybin versus 0.15 after placebo, p = 0.003. Ten people. The striking finding, and the one that gets quoted, is that the reduction was not correlated with the intensity of the acute psychedelic effects — that is, the therapeutic effect and the trip came apart.

Schindler 2022, cluster headache (Headache, PMID 36416492, L1, n = 14 analysed, randomised, double-blind, placebo-controlled). Three doses of 0.143 mg/kg about five days apart. Attack frequency fell 3.2 per week on psilocybin versus 0.03 on placebo, p = 0.251 — negative on its primary outcome. Effect size moderate overall (d = 0.69), small in episodic disease (d = 0.35), large in chronic disease (d = 1.25 and sustained at d = 0.81 over eight weeks). The authors state plainly: "Efficacy outcomes were negative, owing in part to the small number of participants." A blinded extension (J Neurol Sci 2024, PMID 38581739, L2) reported a reduction in attack frequency.

Schindler 2026, migraine again (Headache, PMID 41459830, L1, n = 18, three parallel arms, active placebo). This is the honest one and it is the most important of the four. Single-dose psilocybin, two-dose psilocybin, or diphenhydramine 25 mg as an active placebo. Change in migraine days per week: placebo-placebo −0.7, placebo-psilocybin −2.0, psilocybin-psilocybin −1.7. Not significant, p = 0.102. Over eight weeks all three groups, including the active placebo group, improved by about 50 per cent. The authors' own plain-language summary says the placebo response "urges us to carefully consider how much of the improvement actually came from psilocybin". Blinding was still incomplete even with diphenhydramine.

Where that leaves the headache signal. One nominally positive trial in ten people, one negative trial in fourteen, and one trial in eighteen where an antihistamine did as well as the drug. That is a hypothesis, not a treatment.

And it does not transfer to the patient, for a specific reason. Cluster headache and migraine are primary trigeminovascular disorders. The patient's pain is compressive and infiltrative: a mass on the clivus involving the third and fourth nerves and extending into the cavernous sinus. Her retro-orbital pain was checked against a trigeminal autonomic cephalalgia pattern on the night of 13 September and failed on all three discriminators — the pain was persistent rather than episodic, there was no lacrimation and no nasal congestion (em-obs-log, 13 September). The one disorder where psychedelics have a signal is the disorder we already looked for and she does not have (clinical reasoning, L5).

Tier D. The cancer evidence, which is real, and the reason it cuts against the idea rather than for it

Grob 2011 (Arch Gen Psychiatry, PMID 20819978, L1, n = 12, double-blind, placebo-controlled, subjects as own control). Psilocybin 0.2 mg/kg in advanced cancer with reactive anxiety. Safe, with reductions in anxiety and depression at follow-up.

Griffiths 2016 (J Psychopharmacol, PMID 27909165, L1, n = 51, randomised double-blind crossover, high versus very low dose). "Substantial and sustained decreases in depression and anxiety."

Ross 2016 (J Psychopharmacol, PMID 27909164, L1, n = 29, randomised, active control niacin). "Rapid and sustained symptom reduction."

Agin-Liebes 2020 (J Psychopharmacol, PMID 31916890, L2, long-term follow-up of the NYU cohort at about 4.5 years). Benefits largely maintained.

These are good trials. They are the reason psilocybin is taken seriously at all. Every one of them measured psychological and existential distress. None of them measured pain as a primary outcome.

Now the argument that actually decides this for the patient.

On 13 September the patient said she wants to come off morphine, all of it, and gave her reason in her own words: "It doesn't actually work as a painkiller. It just makes me care less about the pain."

That is an accurate description of opioid action on the affective rather than the sensory-discriminative component of pain, and it was logged as clinical information rather than a complaint (em-obs-log, 13 September; appointment sheet section 7).

The proposed mechanism of psilocybin in pain is the same dimension. Pain acceptance, cognitive flexibility, salience network reconfiguration, a changed relationship to the signal. If psilocybin worked in the patient exactly as its advocates describe, it would do a more thorough version of the thing she has just told us she dislikes about the drug she is already on (clinical reasoning, L5).

The steelman, stated fairly, because it is not a trivial objection. There is a real difference between sedated indifference, every hour of every day, with constipation, cognitive fog and a rising dose and a durable shift in how the pain is held, after one or two supervised sessions, with no daily drug and no tolerance. The patient's complaint may be against the chronic pharmacological blunting rather than against relief of the affective component as such. If the Schindler mechanism holds — therapeutic effect uncoupled from the acute psychedelic experience — then in principle the affective benefit could be obtained without living inside it. That is the strongest version of the case and it is worth stating before dismissing it.

Why it still does not carry. The uncoupling finding comes from two trials totalling twenty-four people in a different disease. The durability claim in cancer is for mood, not pain. And the thing the patient most needs is the sensory component touched, which nothing here addresses. The honest position is that psilocybin is a plausible future treatment for the affective dimension of chronic pain and has nothing to offer the sensory dimension of a compressive skull base tumour (clinical reasoning, L5).

Tier E. Safety, and why the patient is a worse candidate than an average patient

The general safety picture is better than the drug's reputation.

Hinkle 2024 (JAMA Psychiatry, PMID 39230883, L1, 214 studies, analysable adverse-event data from 114 studies, 3,504 participants). No serious adverse events in healthy participants. Serious adverse events in approximately 4 per cent of participants with pre-existing neuropsychiatric disorders, including worsening depression, suicidal behaviour, psychosis and convulsive episodes. Common non-serious events: headache, anxiety, nausea, fatigue, dizziness. No deaths by suicide, persistent psychotic disorders or HPPD in contemporary trials. The authors flag that only 23.5 per cent of studies since 2005 described systematic adverse-event assessment.

Freitas 2025 (Aust N Z J Psychiatry, PMID 39670342, L1 systematic review, 24 trials). Most common events: elevated blood pressure, headaches, nausea, vomiting, fatigue, anxiety. No deaths attributed to psilocybin.

Now apply that list to this patient. Six specific problems, in order of seriousness.

1. Seizures. She has had four brief episodes since 5 September — metallic taste, strange smell, unreality — which are on the sheet as possible temporal lobe seizures, with an EEG requested and not yet done. Convulsive episodes appear in the JAMA serious-adverse-event list precisely in the group with pre-existing neurological or psychiatric conditions. Giving a powerful 5-HT2A agonist to someone with an unexplained possible seizure disorder and an unimaged medial temporal lobe is the wrong order of operations (clinical reasoning, L5).

2. Blood pressure. Elevated blood pressure is the single most consistently reported physiological effect, in both systematic reviews and in both LSD studies. The patient's mass partly encases the basilar artery and abuts both cavernous carotid segments. A deliberate transient hypertensive episode in a woman with tumour wrapped around her posterior circulation is a risk nobody has quantified because nobody has studied it (clinical reasoning, L5).

3. Vomiting. Nausea and vomiting are common. She has had hiccups and a taste of vomit, which are on the sheet as possible medullary involvement, and coughing on Oramorph which is the third pointer to the brainstem. Vomiting hard raises intracranial pressure. If there is any question of a CSF leak — and there is, it is question 5 on Wednesday's sheet — this is not a risk to take casually (clinical reasoning, L5).

4. Headache. Psilocybin causes delayed transient headache dose-dependently in healthy volunteers (Johnson 2012, Drug Alcohol Depend, PMID 22129843, L1). She is already at 7 to 9.

5. Interactions. True serotonin toxicity requires an MAOI or a serotonin-releasing agent; psychedelics combined with non-MAOI serotonergic drugs are low risk (Malcolm 2022, Psychopharmacology, PMID 34251464, L5 review). Her current list is favourable here: morphine, gabapentin, diclofenac, omeprazole, paracetamol, melatonin, senna. Morphine is weakly serotonergic and the combination is not a recognised toxicity risk. This is the one safety domain where she is not unusual. Note for the future: if amitriptyline or duloxetine is added as the neuropathic adjuvant — both are on the list in research thread 20 — that changes.

6. Set and setting, in the real sense. The cancer trials selected patients who were psychologically stable, prepared over weeks, and supported through an eight-hour session by two trained monitors. The patient is in unrelieved pain, has small children, is awaiting a biopsy of a tumour nobody has named yet, and has had no preparation of any kind. The trials do not support the drug in isolation; they support a protocol (clinical reasoning, L5).

Tier F. Legality and access in the UK, which is the practical blocker

Psilocybin is a Class A drug under the Misuse of Drugs Act 1971 and, more restrictively, a Schedule 1 substance under the Misuse of Drugs Regulations 2001. Schedule 1 means no recognised medicinal use in UK law. It cannot be prescribed. Possession is an offence. A Home Office licence is required even to hold it for research, which is the reason UK psilocybin research is slow and expensive.

There is no compassionate access, named patient, or specials route for psilocybin in the UK. The only lawful route to it is enrolment in a licensed clinical trial.

Trial registry checked on 14 September 2026 (ClinicalTrials.gov API, interventional studies with psilocybin, recruiting or not yet recruiting):

• United Kingdom: three studies, none in pain. Self-harm, major depressive disorder, and a healthy volunteer study.
• Cancer pain, anywhere: two, both United States. NCT06001749 "Psilocybin in Cancer Pain Study" (recruiting) and NCT06827054 "Psilocybin with psychotherapy for improving chronic pain in cancer patients requiring opioids" (not yet recruiting).
• Chronic neuropathic pain: two, both Canada. NCT06731335 and NCT06518720.
• Headache: one, United States. NCT06464367, mechanistic studies in headache disorders.

So: there is no trial she could join in this country, and the overseas trials are in different diseases and would not take a patient with an untreated skull base tumour, no histology, and possible seizures.

The route people actually take is illegal purchase and unsupervised self-administration. That is where the real harm in this area lives — unknown dose, no monitoring, no blood pressure cuff, and nobody in the room who knows what to do. It should be named rather than talked around (clinical reasoning, L5).

Tier G. What this displaces, which is the real cost

The reason to be firm here is not that psilocybin is dangerous in the abstract. It is that there are four untried conventional options for the patient's pain sitting in the repo right now, all of them evidenced, all of them legal, and none of them yet obtained:

1. Gabapentin is at 900 mg a day against a licensed maximum of 3,600, with no titration schedule ever written. This is described in the appointment sheet as the single biggest untapped option in the house, and it is still true on 14 September.
2. Dexamethasone, requested and not yet answered, for nerve pain from a compressive mass.
3. A sphenopalatine ganglion block, which is the specialist option with the best fit to her anatomy (research thread 20).
4. Opioid rotation to oxycodone, converted from the true daily total rather than the background dose alone.

Each of these has better evidence in neuropathic cancer pain than everything in this thread combined, and each is obtainable this week from a prescriber. The opportunity cost of pursuing psilocybin is not zero; it is the attention of the people who could be chasing those four (clinical reasoning, L5).

Tier H. The answer in three lines

• As an analgesic: no. No randomised trial anywhere shows psilocybin reduces pain intensity in any patient population, and the two best direct tests used LSD in healthy volunteers with one positive and one null result.
• As a treatment for the distress around a serious illness: yes, genuinely, on four good randomised trials, and that is a real and separate thing worth knowing about for later.
• For the patient specifically, now: no, and the strongest reason is not safety or legality but that its proposed mechanism is the affective dimension, which is the exact thing she said she does not want done to her.

Sources

• Boehnke KF, Pouyan N, Aday JS. Classic psychedelics for chronic pain: a critical review of the literature and practical advice for clinicians. Drugs 2026. PMID 42243569.
• Castellanos JP and colleagues. Chronic pain and psychedelics: a review and proposed mechanism of action. Reg Anesth Pain Med 2020. PMID 32371500.
• Ramaekers JG and colleagues. A low dose of lysergic acid diethylamide decreases pain perception in healthy volunteers. J Psychopharmacol 2021. PMID 32842825.
• Cavarra M and colleagues. A randomised placebo-controlled study of the effects of LSD microdosing (15 µg) on pain perception in healthy volunteers. Br J Pain 2025. PMID 40917239.
• Sewell RA, Halpern JH, Pope HG. Response of cluster headache to psilocybin and LSD. Neurology 2006. PMID 16801660.
• Schindler EAD and colleagues. Exploratory controlled study of the migraine-suppressing effects of psilocybin. Neurotherapeutics 2021. PMID 33184743.
• Schindler EAD and colleagues. Exploratory investigation of a patient-informed low-dose psilocybin pulse regimen in the suppression of cluster headache. Headache 2022. PMID 36416492.
• Schindler EAD and colleagues. Psilocybin pulse regimen reduces cluster headache attack frequency in the blinded extension phase of a randomized controlled trial. J Neurol Sci 2024. PMID 38581739.
• Schindler EAD and colleagues. Comparing single- and repeat-dose psilocybin with active placebo for migraine prevention. Headache 2026. PMID 41459830.
• Grob CS and colleagues. Pilot study of psilocybin treatment for anxiety in patients with advanced-stage cancer. Arch Gen Psychiatry 2011. PMID 20819978.
• Griffiths RR and colleagues. Psilocybin produces substantial and sustained decreases in depression and anxiety in patients with life-threatening cancer. J Psychopharmacol 2016. PMID 27909165.
• Ross S and colleagues. Rapid and sustained symptom reduction following psilocybin treatment for anxiety and depression in patients with life-threatening cancer. J Psychopharmacol 2016. PMID 27909164.
• Agin-Liebes GI and colleagues. Long-term follow-up of psilocybin-assisted psychotherapy for psychiatric and existential distress in patients with life-threatening cancer. J Psychopharmacol 2020. PMID 31916890.
• Hinkle JT and colleagues. Adverse events in studies of classic psychedelics: a systematic review and meta-analysis. JAMA Psychiatry 2024. PMID 39230883.
• Freitas RR and colleagues. The safety of psilocybin-assisted psychotherapy: a systematic review. Aust N Z J Psychiatry 2025. PMID 39670342.
• Johnson MW and colleagues. Psilocybin dose-dependently causes delayed, transient headaches in healthy volunteers. Drug Alcohol Depend 2012. PMID 22129843.
• Malcolm B, Thomas K. Serotonin toxicity of serotonergic psychedelics. Psychopharmacology 2022. PMID 34251464.
• Trial registry: ClinicalTrials.gov API v2, queried 14 September 2026. NCT06001749, NCT06827054, NCT06731335, NCT06518720, NCT06464367, NCT06798636, NCT06793397, NCT06815653.
• Misuse of Drugs Act 1971 (Class A); Misuse of Drugs Regulations 2001 (Schedule 1).

Tier I. Addendum, 14 September 2026 17:30, after a friend pushed back

A friend's reply in em-research: "Okay. But just saying a trial came up... There is evidence of it helping with this type of pain it sounds like. The patient wants off the morphine."

I listed the trials in Tier F and then talked past them. That was a gap and this addendum closes it. Two concessions first, because both are real.

Concession 1. The reframe is correct and my brief answered the wrong question. "Is psilocybin an analgesic" and "could psilocybin help the patient come off morphine" are different questions. Two of the five relevant trials are about opioid reduction rather than pain reduction, and I treated the whole field as if it were about analgesia. A friend is right that the second question is the one that matters in this house.

Concession 2. My central argument was stated too strongly. I argued that the affective mechanism is the problem, because the patient objected to exactly that. But NCT06827054's own rationale is the "total pain" construct, the Cicely Saunders idea that pain has psychological, spiritual and social components. That is mainstream palliative care, not fringe, and treating the affective dimension is a legitimate clinical aim rather than a sleight of hand. The argument stands as applied to the patient's stated preference; it does not stand as a general objection, and I wrote it as if it did.

Now the trials, read individually rather than counted. All five queried in full on 14 September 2026.

• NCT06731335, psilocybin for chronic neuropathic pain, Canada, phase 2, n = 30, recruiting. Its own registered summary states: "While theoretical mechanisms demonstrate promise, there is no clinical evidence. This vacuum of clinical evidence has been occupied by a 'psychedelic hype bubble' with media communications touting psychedelics as 'miracle cures'. The mismatch between evidence and perception creates an urgent need for RCT to fill this significant gap." That is the investigators, in the registry, describing the state of the field. It is the strongest single line in this thread and it is not mine.
• NCT06001749, psilocybin in cancer pain, Dana-Farber and MGH, phase 2, n = 15, recruiting. Requires advanced cancer "unlikely to be cured or controlled with treatment" and chronic opioid therapy at an oral morphine equivalent of 200 mg a day or more. The patient is on 20 to 50 mg a day and has an unbiopsied tumour that is potentially curable. She fails both criteria, and fails the second by a factor of four to ten.
• NCT05585229, psilocybin-assisted psychotherapy for tapering opioids, Canada, open-label pilot, n = 10. This is the closest in intent to what a friend is asking about. It requires a noncancer chronic pain condition, a stable opioid dose for 90 consecutive days, and at least one previous failed taper attempt. The patient fails all three.
• NCT06827054, psilocybin with psychotherapy for chronic pain in cancer patients requiring opioids, phase 2, n = 20, not yet recruiting. Requires a stable pain regimen for at least one month with no intention to adjust it during the study. The patient's regimen has changed on most days this week.
• NCT06518720 (TRANSCEND), neuropathic pain with comorbid treatment-resistant depression, n = 16. Requires a primary DSM-5 diagnosis of major depressive disorder meeting treatment-resistance criteria. Not the patient's situation.

So the honest reading is not "trials exist, therefore there is evidence." Four of the five are phase 2 feasibility or pilot studies with ten to thirty participants. A recruiting trial is the registration of a question. Equipoise is a condition of ethical approval: if the answer were known, the placebo or control arm could not be justified. And every one of these five, read against the patient's actual position, excludes her, most of them on more than one criterion.

The part of a friend's point that survives all of that, and it does survive. the patient wants off morphine. That is the goal. The question is what stands in the way of it.

It is not motivation. She halved her own MR from 20 mg to 10 mg on 13 September without being asked, against my advice, and she was right to. It is not psychological acceptance, and it is not craving or dependence in the addiction sense, which is what the psilocybin opioid literature actually addresses. What stands in the way is that she is at 7 to 9 and nothing else is holding the pain. The floor has to be built before the ceiling can come down. Psilocybin does not build the floor, and none of these five trials claims it does.

Unchanged conclusion, for a narrower reason than before. No for the patient now. Not because the idea is unserious, and not because the mechanism is illegitimate, but because every trial that exists excludes her, the one lawful route in the UK does not exist, and the thing blocking her taper is sensory pain that this does not touch.

What does get her off morphine, in order, all obtainable this week: gabapentin titrated from 900 towards 1,800 to 2,400 mg with a written schedule; dexamethasone if the team will authorise it pre-biopsy; a sphenopalatine ganglion block; and, if a swap is wanted rather than a reduction, oxycodone converted from the true daily total. Then taper the morphine 10 to 25 per cent at a time behind that floor. The agreed measure stands: under 60 mg a day is a genuine reduction, over 80 is a swap.

Tier J. Addendum, 14 September 2026 17:45. Expected effectiveness in the patient, if a trial she could join existed

A friend, in em-research: "I'm not asking about the feasibility of her joining current trials, but an analysis of the potential effectiveness if she were to join a new one that arises."

Fair, and it is the third time he has had to sharpen the question. I answered "is it an analgesic", then "can she enrol". He is asking for a forecast. Here it is, with the reasoning exposed so the reasoning can be argued with rather than the number.

J1. The organising insight: which kind of pain the hypothesis is actually about

Look at what the psychedelic pain field studies, without exception:

Fibromyalgia (two trials). Chronic low back pain (two). Chronic pelvic pain. Migraine. Cluster headache. Phantom limb. Chronic neuropathic pain after failed drug trials.

Every one of these is a condition where the peripheral generator is absent, minimal, or static, and the pathology is substantially in central processing. Fibromyalgia and chronic low back pain are the textbook nociplastic syndromes. Migraine and cluster are primary central disorders. Phantom limb is deafferentation pain in a limb that no longer exists. The Canadian neuropathic trial recruits people who have already failed two guideline drugs, which selects hard for centralised pain.

That is not a coincidence of recruitment convenience. It is the hypothesis. 5-HT2A agonism is proposed to work by reorganising the default mode and salience networks and restoring cortical plasticity, and that is a treatment aimed at a brain that has learned pain, not at a nerve that is being crushed (clinical reasoning, L5).

the patient is the opposite case. She has a 3.8 cm mass on the clivus, growing, in contact with the third, fourth and sixth nerves, the trigeminal distribution and the cavernous sinus, with new signs appearing weekly. Her peripheral generator is large, active and getting larger. The central sensitisation she undoubtedly also has is layered on top of a live structural drive, not standing in place of one.

This is the single most important point in the whole thread, and it is mechanistic rather than evidential. It would still hold if every trial listed above had been positive.

J2. Component by component

Take her pain apart and forecast each piece separately.

Sensory-discriminative, from compression and infiltration. Expected effect: near zero. Nothing in the proposed mechanism acts on a peripheral generator. No trial has tested a psychedelic against pain from an active growing tumour.

Central sensitisation and wind-up layered on top. Expected effect: small but not zero, and unmeasurable in her. She has had 12 to 14 months of poorly controlled pain, which is enough to establish central changes. This is the component the hypothesis genuinely targets. The problem is that any central gain gets overwritten by the peripheral drive it sits on.

Affective, the suffering component. Expected effect: real, possibly large. This is the one place the evidence is good, on four randomised trials. Two caveats. Those trials recruited patients with clinically significant anxiety or depression, and the patient's distress has never been characterised, so the headroom may be smaller. And it is the dimension she has explicitly asked not to have treated.

Opioid taper facilitation. Expected effect: near zero, for a structural reason. The psilocybin opioid literature, and the stronger alcohol and tobacco literature behind it, works by restoring motivation and reducing craving. The patient has no deficit in either. She halved her own MR on 13 September unasked and against my advice. A treatment that corrects a deficit gives nothing when the deficit is absent. Her taper is blocked by pain, not by wanting.

Opioid-induced hyperalgesia. This is the strongest case for it and I want to state it properly. Her daily oral morphine went from 30 mg on 6 September to 90 mg on 10 and 11 September with no fall in score. That is the textbook signature of OIH. OIH is a state of maladaptive central pain amplification, driven partly by loss of descending inhibition, and it is mechanistically much closer to the nociplastic phenotype the psychedelic hypothesis targets than her tumour pain is. If any component of her pain were going to respond, this is the one. There is no clinical evidence for it, only the mechanistic adjacency. But it is a real argument and it is the best one available on this side.

J3. A forecast, labelled as what it is

These are subjective clinical-judgement probabilities, not derived from data. They are offered because a refusal to estimate is less useful than a number with its workings shown. Argue with the reasoning above, not the digits.

• Sustained reduction of 30 per cent or more in her pain score (the usual threshold for a meaningful analgesic effect), attributable to psilocybin, in her current state: roughly 10 to 15 per cent.
• Meaningful reduction in distress and fear, measured properly: roughly 50 to 60 per cent. This is the good bet and it is a different outcome.
• Enabling a morphine reduction that would not otherwise have happened: roughly 15 per cent, and the number is confounded because she is already reducing without it.
• A serious adverse event given seizure history, unimaged temporal lobes, and tumour in contact with the posterior circulation: higher than the 4 per cent seen in the neuropsychiatric trial populations, and not quantifiable.

The asymmetry is the finding. The outcome most likely to move is the one she has asked us not to chase, and the outcome she wants is the least likely to move.

J4. What would change these numbers, which is the useful half

The estimates above are conditional on her state now. Four things would raise them materially.

1. Successful treatment of the tumour. If she has surgery and protons and is left with a static deafferentation injury to the trigeminal system with no growing lesion behind it, her phenotype converts from structural to nociplastic. That is precisely the population the hypothesis is about. Persistent neuropathic facial pain after skull base treatment is common and is currently treated badly. My 10 to 15 per cent becomes something I would want to look at again properly.
2. A negative EEG and a benign explanation for the four episodes, which removes the largest single safety objection.
3. Histology and a controlled disease course, which removes the "overwritten by progression" problem. Any central adaptation has a half-life; a growing tumour has a shorter one.
4. A trial with an active placebo and a pain-intensity primary endpoint reporting positive. Nothing like that exists yet. The 2026 migraine trial, which is the only one that used a real active control, is the reason to be sceptical rather than hopeful.

J5. The conclusion, which has moved

The idea is not wrong. It is early, and it is aimed at the wrong phase of her illness.

Its window is not now, with an untreated growing mass and pain at 7 to 9. Its window is twelve to eighteen months from now, if she is left with persistent neuropathic facial pain after definitive treatment, at which point she would be a genuinely good candidate rather than a poor one, and the trials that exist today would be several generations further on.

That is worth keeping on file rather than closing. Added to the long-term pain thread as a deferred option.

Tier K. Addendum, 14 September 2026 18:15. Dose, and why foraged material cannot be dosed

A friend, in em-research: "What's the dose you anticipate being required for that kind of benefit? Assuming liberty caps as we are in the UK."

K1. The trial doses, and which outcome each buys

Doses are weight-adjusted where the trial did so. The patient is 51 kg, which is well below the 70 to 80 kg on which most fixed-dose protocols were built.

The important structural point: the dose depends entirely on which outcome is being sought.

• The distress benefit, which is the outcome I forecast at 50 to 60 per cent in Tier J, comes from the high-dose trials. For the patient that is roughly 15 to 22 mg, a full psychedelic dose producing several hours of profound altered state requiring two trained monitors.
• The headache signal, which is the only pain signal in the field, came from 7.3 mg at her weight, a much lower dose, and the finding there was that benefit did not correlate with the intensity of the experience.

So there is no single answer. The dose for the outcome she wants least is roughly three times the dose for the outcome with the only pain evidence behind it.

K2. Liberty caps, and the arithmetic that answers the question by dissolving it

Psilocybe semilanceata is the relevant UK species, fruiting on grassland from roughly late August to November.

The definitive measurement is Christiansen 1981 (Planta Med, PMID 17401967, L5 analytical), which assayed Norwegian P. semilanceata across habitats and found psilocybin in dried material varying from 0.17 to 1.96 per cent of dry weight. That is an 11.5-fold range within one species in one country. The same paper found the highest percentage in the smallest mushrooms but the highest absolute milligrams in the largest, so size is not a usable proxy in either direction.

Run the trial doses against that range. Dried weight required:

At a nominal 25 mg dried weight per cap, the 25 mg target spans roughly 50 to 590 caps depending on which field the mushrooms came from and which year.

That is the answer to the question, and it is not a number. The same handful of mushrooms is either a sub-threshold dose or three times a full clinical dose, and there is no way to tell which from inspection. Every clinical trial in this thread used synthetic psilocybin of known purity, assayed and weighed, precisely because this variability makes the natural material unusable as a medicine.

Three further problems with foraged material, all real:

• Composition differs. Wild P. semilanceata carries baeocystin and norbaeocystin alongside psilocybin, in ratios the 1994 comparative work found to be distinctive to this species (Gartz 1994, J Ethnopharmacol, PMID 7967658, L5). Their pharmacological contribution is not characterised. Synthetic psilocybin contains none of them.
• Content degrades. Recent analytical work found significant variability in psilocybin-to-psilocin ratios attributable to storage conditions (Rodríguez 2025, ACS Omega, PMID 41322617, L5). Material picked last autumn is not the material it was.
• Misidentification is the genuine lethal risk in UK foraging, not the psilocybin. Galerina marginata contains amatoxins, the same class as death cap, causes fulminant hepatic failure, and overlaps in season and grassland-edge habitat. This is the mechanism by which people in this country actually die from mushroom picking.

K3. Where I stop, and why

I have set out the pharmacology because it is published, and because the variability finding is itself the most important safety fact in it, but I am not going to convert it into a dose for the patient.

The reason is specific rather than general. The three objections already on the table are seizures against four uncharacterised episodes and no EEG, a blood pressure rise against tumour in contact with the basilar artery and both cavernous carotids, and vomiting against a possible CSF leak and possible medullary involvement. Every one of those risks scales with dose. An unmeasurable dose makes all three unbounded. That is the argument, and it does not depend on the law or on disapproval.

K4. What I can actually do about the thing he wants

A friend: "Would love to see a trial like this pop up."

Set up on 14 September 2026: a monthly automated check of ClinicalTrials.gov and the ISRCTN registry for interventional psilocybin studies in pain, cancer pain or neuropathic pain with a UK site, reporting into em-research when something new appears. Script memory/context/em-research/trial-watch.py, cron entry recorded in CRON-JOBS.md.

The conditions under which a hit would actually matter are in Tier J4: after definitive treatment, with histology known, the seizure question resolved, and persistent facial neuropathic pain from a static injury. That is the patient this hypothesis is about, and the patient may become her.