I watched the recent International Headache Society webinar, Updates in Cluster Headache, run by the IHS Cluster Headache Special Interest Group. I watched it as a patient who has lived inside this disease, not as a clinician looking in from outside. And I came away genuinely energised: the science is moving quickly, and the people driving it are serious, careful and clearly devoted to a disease the world mostly overlooks. What I want to add, from the patient's chair, is a hopeful nudge about where the next breakthroughs might come from, because I think cluster headache will be understood fastest when we connect the threads rather than studying them in silos.
That word, silos, is the whole idea, so let me show you what I mean before I spend ten thousand words on it. We have a group studying the hypothalamus and the body clock. We have a group studying inflammation and biomarkers. We have a newer thread on bioenergetics, the cell's energy economy. And we have a young, wide-open thread on the gut, diet and the microbiome. Each is doing careful, excellent work. The opportunity I keep seeing, from where I sit in the patient's chair, is how much they could gain by talking to each other. They are not four diseases. They are four windows onto one.
First, the credit that is due
Let me start where the webinar earned it, because the people doing this work are serious, and everything that follows is meant as encouragement and a patient's wish list, never as criticism.
Rolf Fronczek's pathogenesis update was, for me, the most alive part of the hour. He speaks at the speed of someone holding a dozen moving parts in his head at once, and that is exactly what cluster headache demands. His domain is the hypothalamus and the orexin system, and I will admit it is the part of the picture I knew least about going in. Orexin, also called hypocretin, is a pair of neuropeptides made by a small cluster of neurons in the lateral hypothalamus. They are key regulators of arousal: they help stabilise wakefulness, gate the boundaries between sleep stages, and tie into reward, autonomic tone and pain modulation. When the orexin signal collapses, you get narcolepsy. The reason this matters for us is that cluster headache is, above almost any other pain disorder, a timekeeping disease. Attacks come at the same hour of the night, in bouts that arrive in the same season, with a precision that feels less like illness and more like an alarm clock wired into the skull.
And there is human evidence connecting cluster headache to orexin biology. The Danish group, with Nunu Lund among the authors, showed that cerebrospinal hypocretin-1 is significantly reduced in both episodic and chronic cluster patients compared with controls (Barloese, Jennum, Lund and colleagues, 2015). Holland and Goadsby had already built the theoretical case that the hypothalamic orexinergic system may sit at the junction of the circadian timing and the trigeminal pain processing that together define the disease. So when Fronczek talks about pathogenesis, he is not gesturing at a vague "brain thing." He is pointing at specific rhythm-keeping tissue that appears dysregulated in cluster headache, and that may help govern the sleep and circadian timing around which our attacks cluster. (That attacks are specifically triggered by a particular sleep-stage transition is a long-standing hypothesis, not a settled fact; Naber and colleagues, 2019, review how loose the REM-sleep link actually turns out to be.) Hold on to that detail about sleep and energy demand. It comes back.
Nunu Lund's preventive-treatment update came from someone whose biomarker work I have followed closely and admire. Her group at the Danish Headache Center has produced some of the most consequential cluster headache findings of the last two years, and I will lean on them heavily below, because they happen to demolish one of the oldest and most damaging assumptions in this field. This is in no way a complaint about the calibre of the researchers. The science being done is genuinely good. My only hope is that the frame around it keeps widening.
An area I would love to see revisited: Vitamin D
When the question of the Vitamin D3 regimen came up in the Q&A, the answer moved very quickly: there had been an attempt to study it, the study did not work out, and then, almost in the same breath, came the caution that too much vitamin D can be toxic. I want to pick that exchange up gently and constructively, because I think there is a genuine opportunity hiding inside it.
The regimen itself is not simply internet folklore. It was developed by a chronic cluster patient, Pete Batcheller, a retired naval aviator, and it has been refined by the patient community since around 2011. It is high-dose vitamin D3, typically in the order of 10,000 IU per day, taken alongside a specific set of cofactors, magnesium, vitamin K2, vitamin A, zinc, boron, omega-3 and calcium, with the aim of raising serum 25-hydroxyvitamin D into a patient-regimen target range and holding it there. I have written the full history of how it emerged in From Cockpit to Clinic. A survey of 110 sufferers using the regimen was presented at the American Academy of Neurology's annual meeting (Neurology, 2014), with most respondents reporting meaningful reductions in attack frequency. That is a patient-reported conference abstract, not a randomised controlled trial, and I will be the first to say so. It is community testimony rather than controlled data, but it has been visible to anyone willing to look for well over a decade.
Now to the trial. There was a serious attempt to study it: a randomised, double-blind trial of high-dose vitamin D plus a multivitamin for cluster headache prevention, run out of UTHealth Houston with Mark Burish as principal investigator (ClinicalTrials.gov, NCT04570475). It is on the public record that the trial was terminated, and the documented reason given is stark and simple: "Low number met criteria to randomize." It enrolled 27 participants, began in September 2021, in the thick of the pandemic, and closed in 2024.
A trial that opens in late 2021, in the depths of the pandemic, and struggles to enrol is one of the most understandable stories in modern medicine, and I have nothing but sympathy for the team that tried. The way I read that result is not as a closed door but as an unfinished question: more than a decade of patient-reported benefit has so far produced one randomised trial, and that trial never quite got the chance to give us a clean answer. To me, that is an exciting opening rather than a disappointment, because the question is still wide open and very much worth doing well.
Then there is the toxicity question, which I think deserves a careful and generous reading, because the safety picture is genuinely debated rather than settled in either direction. The mainstream adult tolerable upper intake level, set by the US National Institutes of Health Office of Dietary Supplements, is 4,000 IU per day. So 10,000 IU per day is, plainly, an above-guideline dose, and I am not going to pretend otherwise. What I will say is that there is a serious, more permissive strand of the risk-assessment literature: Hathcock and colleagues (2007) argued, from the trial record in healthy adults, that 10,000 IU per day is a defensible upper limit, and Reinhold Vieth's analyses report that convincing hypercalcaemia from vitamin D involves 25-hydroxyvitamin D well above 200 nmol/L (about 80 ng/mL), and that intoxication-related hypercalcaemia is consistently accompanied by levels above roughly 220 nmol/L (about 88 ng/mL), with frank toxicity at much higher concentrations. That ~80-88 ng/mL floor overlaps the top of the regimen's own 80-100 ng/mL target — which is exactly why serum 25(OH)D and blood-calcium monitoring are essential. Those are real, citable positions, but they sit at the permissive end of an open debate, not at a consensus or a clinical safety guarantee. So the honest framing is not "10,000 IU is safe, full stop." It is that 10,000 IU per day is a debated, above-guideline dose, which the regimen pairs with blood-level and calcium monitoring, and that a blanket "toxicity" warning, without engaging any of that, dose, blood level, or monitoring, leaves the most useful part of the conversation still to be had.
Put the numbers next to standard paediatrics and the picture gets a good deal more interesting.
"Stoss" therapy, a single high-dose oral bolus of vitamin D, is used in selected deficient paediatric patients under specialist monitoring. Starship Children's Hospital's own guideline says intermittent high-dose therapy is known as STOSS therapy, recommends it for children with symptomatic deficiency, and lists 600,000 IU as the high-dose treatment for children over five with 25(OH)D below 50 nmol/L when serum calcium and phosphate are normal, with repeat 25(OH)D, calcium, phosphate and ALP checks three months later. A retrospective New Zealand study (Martin and colleagues, 2019) gave single oral boluses of 100,000 to 800,000 IU of cholecalciferol to 23 children aged 3 to 16 with inflammatory bowel disease; vitamin D levels rose as intended and the approach was generally well tolerated, though one child developed transient hypercalcaemia that resolved. That is paediatrics, not adult cluster headache. It is specialist, monitored medicine, not something done casually. But it does show that supervised medicine already works with single doses far larger than a daily 10,000 IU, when there is reason to.
One cluster-headache vitamin D paper found a mean 25(OH)D level of 14.0 ng/mL in 28 patients, with 92.8% below 20 ng/mL. If a trial is designed around the patient-regimen target of roughly 80-100 ng/mL, it has to ask whether people ever reach that window. This simple model uses the adult loading-dose formula from van Groningen and colleagues; it is a trial-design illustration, not pharmacokinetics or medical advice.
With these defaults, the first 12 days of 10,000 IU/day are still nowhere near the selected patient-regimen target. The loading model gets there quickly enough to test the biological question.
So the most useful response to an adult regimen, I think, is not a blanket toxicity warning but a design conversation: dose, baseline level, target range, loading strategy, cofactors and monitoring. A patient starting around 14 ng/mL is not close to the 80-100 ng/mL range proposed by this patient regimen. If they are given a daily dose without a loading strategy, the early weeks of a trial may be spent slowly climbing while the attacks continue, which would test endurance more than biology. A future trial built around a prespecified blood-level target, with safety stopping rules, would, I think, be a far fairer and genuinely exciting test of the hypothesis.
The kidney-stone fear needs the same discipline. It is not nothing, and patients with renal disease, hypercalciuria, sarcoidosis, primary hyperparathyroidism or abnormal calcium handling need proper medical oversight. But vitamin D alone is not the kidney-stone bogeyman it is often made to be. In the New Zealand ViDA trial, monthly 100,000 IU vitamin D3 for a median 3.3 years did not increase kidney-stone events or hypercalcaemia. A 2016 meta-analysis found more hypercalcaemia and hypercalciuria signals, which is exactly why calcium monitoring matters, but it still did not find an increased kidney-stone risk. Put that beside standard cluster-headache preventives, verapamil with ECG monitoring and heart-block risk, lithium with renal and thyroid monitoring, corticosteroids with their systemic burden, topiramate with cognitive and stone risks, and the safety conversation looks very different. Not risk-free. Not casual. But hardly uniquely frightening.
The honest summary is hopeful and practical: the patient-reported signal is real and long-standing, the dose is above mainstream daily guidance, loading is already normal medicine in selected monitored contexts, and the questions worth designing around are baseline, target, loading, cofactors and monitoring. If a future study intends to test the regimen's proposed biological range, it should get participants into that prespecified range, hold them there, monitor them, and then test the hypothesis properly. That, to me, is a study waiting to happen, and a genuinely exciting one.
"More than a decade of patient-reported benefit deserves one clean, well-designed trial. That feels less like a grievance and far more like an opportunity."
On the Vitamin D3 regimenWhat stays abnormal between the attacks
Here is the finding I took most from the science, and it is the hinge on which much of what follows turns. A common working assumption has been that cluster headache behaves like a switch: the bout is the disease, and between bouts the patient is essentially back to baseline. That picture now looks too simple, and the people complicating it were on the webinar. Let me be careful about scope first, because this is exactly the kind of claim that gets oversold: none of what follows overturns the clinical definition of remission in the diagnostic criteria. What it says is narrower and still important. Several things we can measure do not return to normal when the attacks stop.
Two recent studies from the Danish Headache Center, both drawing blood interictally, between attacks, make the point. Søborg, Lund and colleagues (2025) measured PACAP-38, the neuropeptide now treated as a prime therapeutic target, and found it elevated in every cluster state compared with healthy controls, including episodic patients sampled during remission, where it ran about 34% higher than controls. Their interpretation is worth quoting: a "latently increased" level that "permanently lower[s] the threshold for activation of the trigeminovascular pathway." Then comes the inflammation. Lund and colleagues (2025), in Annals of Neurology, screened 45 cytokines across hundreds of samples and found that Oncostatin M was significantly elevated in all three cluster states, again including episodic remission, and that it was not explained by time since last attack, triptan use or verapamil. Their own words: the immune system "is altered in all 3 states of cluster headache compared with controls," and the alteration shows "specificity to the disease itself and not a secondary phenomenon." Worth noting that both are blood measures, plasma and serum, not direct readings from the brain.
Look at what these measured markers actually do across a cycle.
This is not an isolated pair of findings. It is a pattern, and the pattern is consistent:
- Hormones: Petersen, Lund, Jensen and colleagues (2024) identified compensated hypogonadism in men with cluster headache, and crucially found that androgen levels "did not differ between bouts and remissions." Not an artefact of the attack. A baseline state.
- Sleep: Lund and colleagues (2019) titled their paper almost as a rebuke to the switch model, showing disturbed sleep "is not the result of transient processes associated with the cluster period." The disturbance persists outside the bout.
- Symptoms: even clinically, the remission is not silent. Danish work on pre-cluster symptoms (2021) found a majority of patients reporting mild cluster symptoms and shadow attacks during remission, pointing to "an underlying pathophysiology extending beyond the cluster headache bouts."
Read those together and a more modest but still important conclusion follows: in episodic cluster headache, several measured biological markers remain abnormal during remission, not only during the bout. That does not redefine "remission" clinically, and it does not by itself prove a single, continuous year-round disease process. But it does complicate the comfortable idea that the patient is simply well between bouts, and it raises a question I find genuinely exciting: if some of the biology stays disturbed when the attacks stop, might there be real room to treat the year-round state, and not only the bout?
An engine running too hot: the energy hypothesis
If the disease never fully switches off, the obvious question is what, exactly, is the chronic abnormality. One thread that the webinar barely touched, and that I think deserves real attention, is bioenergetics, the cell's energy economy.
In January 2026, a Chinese-led group (Yu and colleagues, Journal of Pain Research) published a Mendelian randomisation analysis of 338 cerebrospinal-fluid metabolites against cluster headache risk. I want to describe it accurately, because precision is the whole point of this article: it is a genetic causal-inference study built on European datasets, not a measured patient cohort, and the authors themselves call the results preliminary. But the signal is striking. The single strongest hit was orotate, and 5-oxoproline (pyroglutamate) also came through. The authors grouped them explicitly as "energy and mitochondrial metabolites" and as markers of "disrupted glutathione metabolism," reinforcing "the hypothesis that mitochondrial dysfunction represents a shared pathogenic feature across primary headache syndromes."
Why those two molecules matter is the interesting part. Orotate sits on the pyrimidine-synthesis pathway whose committed enzyme, dihydroorotate dehydrogenase, is physically embedded in the inner mitochondrial membrane and feeds electrons directly into the respiratory chain. When mitochondrial respiration stalls, orotate metabolism shifts. 5-oxoproline is an intermediate of the cycle that builds and recycles glutathione, the cell's master antioxidant; its accumulation flags a depleted antioxidant tank. Put them together and you get a coherent, if still correlational, picture: an engine straining at the electron-transport chain, and an antioxidant supply running low. Exactly the state a chronically over-demanded, under-supplied tissue would produce.
This is not a lone idea. In migraine, the neuroenergetic hypothesis, a chronic brain-energy deficit lowering the threshold for attacks, is now a substantial literature (Del Moro and colleagues, 2022). And cluster headache layers onto it a strong relationship with oxygen, though I want to be careful to state which direction is solid and which is not. The abortive direction is rock solid: high-flow oxygen is a first-line acute treatment, proven in a definitive randomised trial (Cohen, Burns and Goadsby, 2009, JAMA) and endorsed in the European Academy of Neurology guideline (Mitsikostas and colleagues, 2023). The provoking direction is much softer. Lee Kudrow proposed in the 1980s that cluster patients autoregulate oxygen poorly, and reported oxygen-saturation falls preceding attacks (Kudrow, 1990), but that was an old, small study, and controlled hypoxia has not reliably triggered attacks in later provocation work, so I hold it loosely rather than asserting "hypoxia provokes." What survives is an asymmetry worth sitting with: oxygen reliably aborts attacks, but that does not by itself prove a primary oxygen-handling defect. It keeps oxygen-related mechanisms and energy metabolism on the list of questions worth studying carefully.
Now layer the clock back on. Oliveira and colleagues (2024) showed that CLOCK-gene expression in cluster patients fluctuates abnormally across the year. The hypothalamus, Fronczek's hypothalamus, is one of the most energy-hungry, most circadian-driven structures in the brain, and our attacks cluster at night and around sleep, when its demands shift (the specific sleep-stage trigger, again, is a hypothesis rather than an established mechanism). A model in which a bioenergetic and oxidative strain interacts with a dysregulated molecular clock to time the bouts is, I think, the most fertile synthesis available right now. I hold it as a hypothesis, not a finding. But it is a hypothesis that connects Fronczek's hypothalamus, Lund's inflammation, Kudrow's hypoxia and the new metabolite data into one story instead of four.
The clue I would most love to see chased
Now to the part I find the most exciting clue we have. I want to state my actual claim precisely, because it is easy to misread. I am not saying "psychedelics are a nice alternative treatment." I am saying something narrower and, I think, far more important: the mechanism by which these compounds may abort or suppress attacks is the suspect we should be interrogating. If we can establish exactly how a drug switches an attack off, or suppresses a cycle, we may learn something fundamental about what an attack actually is. The treatment becomes the probe. And when you follow that probe, it leads straight back to the mitochondria that the metabolite study just flagged.
Two of the most striking interventions to emerge for cluster headache in living memory did not come from pharma pipelines; they came from patients. I want to be precise about the word "effective", because it would be wrong to imply these beat the front line: the strongest guideline-backed tools we have remain high-flow oxygen, subcutaneous sumatriptan and verapamil. But two patient-discovered interventions carry remarkably strong community signals, and that is its own kind of evidence about where to look. Vitamin D is one. Serotonergic psychedelics are the other. The psychedelic evidence is real, if early. Sewell, Halpern and Pope (2006) documented patients aborting attacks and terminating cycles with psilocybin and LSD. Schindler and colleagues (2015) found, in the Clusterbusters survey, that "infrequent and non-hallucinogenic doses were reported to be efficacious", the busting-dose phenomenon the community has described for years. The one randomised controlled trial of low-dose psilocybin (Schindler and colleagues, 2022) was, in honesty, negative on its primary endpoint in a tiny sample, though a blinded extension phase (2024) later showed a roughly 50% reduction in attack frequency. And here is the line that should stop everyone: in that trial, the change in attacks "[was] not correlated with the intensity of acute psychotropic effects." The therapeutic signal may be at least partly separable from the trip.
Which brings me to DMT, and I am going to be scrupulous about what is and is not established. There is, at present, no controlled clinical trial of DMT for cluster headache. A 2026 Neurology conference abstract reports an interim international survey of DMT use in cluster headache, but the reports that vaporised DMT can abort an attack within seconds, often at sub-hallucinogenic doses, remain survey and patient-community evidence, not trial data. They should be labelled as such. But communities that gave us vitamin D and the busting dose have earned the right to be taken seriously enough to test, and the mechanistic story behind DMT is where this gets genuinely exciting.
DMT is not only a serotonin-receptor drug. It is an endogenous agonist of the sigma-1 receptor (Fontanilla and colleagues, Science, 2009). The sigma-1 receptor is a chaperone protein that lives at the contact points between the endoplasmic reticulum and the mitochondria, the mitochondria-associated membranes, where it regulates calcium signalling into the mitochondria and supports cell survival under stress (Hayashi and Su, Cell, 2007). And here is the payoff that should make a cluster researcher sit up: Szabó and colleagues (2016) showed that DMT protects human neurons against severe hypoxia through the sigma-1 receptor, associated with reduced activity of HIF-1, the master regulator of the cell's response to low oxygen; Nardai and colleagues (2020) confirmed sigma-1-dependent protection against ischaemic brain injury in a living animal. Look at the machinery.
DHODH and orotate enter here
Hold that diagram against the energy hypothesis and you get a genuinely interesting line of inquiry, though I want to be careful not to oversell it, because every link in it is still provisional. The metabolite study (Yu and colleagues, 2026) is a Mendelian randomisation analysis: it uses genetic instruments to infer which metabolites may be causally related to cluster headache risk. It did not measure metabolites in patients, least of all during an attack, and it is not proof of mitochondrial dysfunction. The sigma-1 and hypoxia data on DMT (Fontanilla 2009; Szabó 2016; Nardai 2020) come from cells and animals, with no cluster headache anywhere in them. So this is not "two independent lines of evidence converging on a culprit." It is one preliminary genetic signal and one mechanistic story from the lab bench, each gesturing, separately, toward mitochondrial energy handling. That is a hypothesis worth chasing, not a verdict. But it is a good hypothesis, precisely because the question underneath it, what does the thing that aborts an attack actually do, is answerable with the right study.
I am presenting that as a hypothesis, clearly, not as fact, and the standard caution holds twice over: these compounds are largely illegal and, in DMT's case, untested in controlled cluster-headache trials. But this is the point I most want the researchers to hear. We do not need to wait for permission to find DMT interesting. We need to fund the mechanistic work, in cells and in animals, that asks how it might switch an attack off, if the patient reports are borne out, because the answer to that question may help explain what an attack is.
Part two: the thread I would love to see pulled
The gut, the diet, and an open frontier
If the disease is a whole-body, year-round, energy-and-inflammation problem, then there is one more system I think is brimming with untapped potential, and it is the one that sits upstream of inflammation, of serotonin, of vitamin D metabolism and of mitochondrial fuel all at once. The gut.
Start with the psychedelic thread, because it ties straight back to the clue we were just chasing. A 2025 viewpoint in ACS Chemical Neuroscience (Wang, Jun, Lin and Wang) lays out the bidirectional relationship between classic psychedelics and the gut microbiome, arguing that the interaction runs through "5-HT receptor signaling, neuroplasticity, and microbial metabolism," and hypothesising that microbes may influence psychedelic efficacy. In plain terms: the bugs in your gut may help set the serotonergic tone that a psychedelic then acts on, and may help explain why the same dose helps one patient and not another. I explored this terrain in Psychedelics and the Gut-Brain Axis, and the punchline is uncomfortable for a receptor-only model of how these drugs work: the response may be partly ecological, not just neurochemical.
Now the part that genuinely surprised me. In migraine, the gut-microbiome literature is not fringe and it is not thin. It is a serious, fast-growing, mechanistic body of work, though still one that calls for validation rather than declaring the case closed. A 2025 review (Zhang, Tran and Moskatel) notes that recent metagenome-wide association studies using Mendelian randomization "support the causal association between gut microbiota and migraine", which is to say they support a possible causal association; systematic reviews in the same area continue to stress that the findings need confirmation (Mugo and colleagues, 2025). A systematic review the same year (Gorenshtein and colleagues) found a consistent reduction in Faecalibacterium, an anti-inflammatory workhorse, and an excess of Veillonella in migraine patients. A 2026 multi-omics study in paediatric migraine (Tian and colleagues) found enrichment of an intestinal lipopolysaccharide-synthesis pathway with real diagnostic power and a depletion of kynurenic acid, a tryptophan-pathway metabolite. And a 2025 review (Sivri and Yıldıran) lists the shared mechanisms as "oxidative stress, neurogenic inflammation, mitochondrial dysfunction, and gut dysbiosis." Read that list again. Oxidative stress. Mitochondrial dysfunction. Neuroinflammation. These are the same words showing up in the cluster-headache energy and biomarker literature, arrived at from a completely different doorway.
And we have a clue in our own disease that should have set this thread on fire years ago: diet. In 2018, Di Lorenzo and colleagues ran a small, open-label, single-arm trial of a modified Atkins ketogenic diet in eighteen patients with drug-resistant chronic cluster headache, the hardest group to treat that exists. The mean number of attacks per month fell from around 109 to around 31. Fifteen of the eighteen responded; eleven reached a full resolution of headache. A ketogenic diet plausibly touches several of the threads above, and I want to phrase this as plausibility rather than proof: it shifts the brain toward ketones as an alternative fuel, which is why it is of interest wherever mitochondrial energy metabolism is in question, and it is known to remodel the gut microbiome (in animal models of other brain disorders, some of the benefit can even be transferred by faecal transplant). None of those mechanisms has been shown to be what helped these particular cluster patients; they are candidate explanations, not demonstrated ones, and the trial itself was small and uncontrolled. But a provocative result in the most refractory patients we have, sitting on top of so many converging threads, is exactly the kind of finding that deserves a wave of follow-up. Eight years on, a properly powered trial, ideally with a microbiome arm built in, feels like one of the most promising studies still waiting to be run.
I think there is a simple, structural reason it has not happened yet, and I say it with sympathy rather than blame, because it is about the system and not the people in it. A diet has no patent. A bacterial community has no obvious molecule to license. The machinery of modern therapeutic research, the funding, the trials, the careers, is naturally built to find a druggable target and own it. When the most promising lead is "eat differently" or "change your gut ecology," there is no product at the end of the rainbow, and the funding engine is slow to start. That is nobody's fault; it is an incentive structure. And naming it honestly is the first step to working around it, through public funding, philanthropy and patient-led research that does not need a molecule to license.
And here is an encouraging sign. Commercial actors are clearly alert to this. There is a granted United States patent (US 9,987,224 B2, listed by Google Patents as assigned to Seed Health) describing a method of "addressing the avoidance of migraines, cluster headaches and dizziness by adjusting an individual's microbiome", delivered, of all things, through adhesive oral strips that seed beneficial bacteria. I want to be careful: a patent is a claim, examined for novelty, not a proof of efficacy, and I am not endorsing the device or saying it is in active development for cluster headache. But the existence of that patent tells you something useful. It is at least a sign of commercial and intellectual-property interest in the microbiome-headache space, and that makes the absence of direct cluster-headache trials more conspicuous.
A conversation worth reopening
There is one more silo, and it is the one we have been most hesitant to reopen. In the 1970s and 1980s, clinicians like John Graham and Lee Kudrow wrote candidly about the cluster patient as a whole person, the personality, the psychology, the life history. Graham's "leonine" portrait and his description of "a source of fierce but repressed anger" behind a mask of competence, Kudrow's personality work (1974), these were bold, human observations. Somewhere along the way we decided they were unscientific and quietly dropped them. Today's literature is genetics, peptides, receptors and stimulators. Cleaner. Safer. And, I think, incomplete.
I have written about this at length in The Fire That Turns Inward and The Lion in the Mirror. The psychodynamic tradition, Allan Abbass and colleagues' 2008 Cephalalgia paper chief among it, argues that emotional factors and somatisation can be clinically relevant in some chronic headache patients, and that direct assessment and management of those factors may matter. That paper called for proper trials in headache populations. For cluster headache specifically, that call has never been answered.
And here is an honest gap I will name rather than paper over. There is now a real literature linking adverse childhood experiences to primary headache disorders, including migraine, and to chronic pain; one migraine cohort reported childhood trauma in around 40% of participants, with attachment disruption mediating chronicity. For cluster headache, I could not find a dedicated ACE or trauma study. I do not get to call that settled science, so I will not. What I can offer is the thing science is supposed to start from: an observation worth a hypothesis. On a support call of seven cluster patients, I mentioned growing up without a father and a hard childhood. Four others, unprompted, described strikingly similar early lives. That is anecdote, not evidence. But it is exactly the kind of anecdote that, in a braver era, sent a Graham or a Kudrow looking. And it is not as disconnected from the rest of this article as it might first appear: early-life stress can shape the gut microbiome, inflammatory set-points, and stress physiology. The psychology is not a separate magisterium from the biology. It may be one of its upstream taps.
The body does not exist in isolation. It is not a collection of neuropeptide blocks waiting to be antagonised one at a time. Psychology clearly shapes distress, disability, coping and comorbidity, and it may shape disease course in ways cluster headache has not yet tested. Studying that openly would only add to our rigour, never threaten it. It is one of the most human frontiers this field has left to explore.
What I am hoping for
So let me bring the silos together, because the whole hope of this piece is that they belong together.
We now have decent evidence that several measured aspects of cluster headache biology stay abnormal between bouts, not just during them: PACAP-38, Oncostatin M, hormones, sleep. We have an emerging, still-preliminary bioenergetic signature: the genetic metabolite data, the migraine neuroenergetic literature, Kudrow's older hypoxia work, the reliability of oxygen as an abortive, the dysregulated clock. We have two patient-discovered interventions, vitamin D and psychedelics, whose proposed mechanisms, anti-inflammatory and sigma-1/mitochondrial, point back toward that same biology. We have a gut-microbiome story that is robust in migraine and mechanistically overlapping at the level of inflammation and energy, but one I could not find a single dedicated human cluster-headache study of, plus a refractory-patient ketogenic result that has been left to gather dust. And we have an abandoned psychological tradition that the rest of pain medicine is now quietly rediscovering. None of these is settled. But they are not separate curiosities either. They look like windows onto one system, and that is reason enough to study them together.
What I am hoping the research community might take on is not radical. It is this:
- Treat the disease, not just the bout. If measurable biology stays abnormal in remission, design prevention and trial endpoints around the year-round state, including the persistent peripheral inflammatory and neuropeptide abnormalities now documented, rather than only attack counts during a bout.
- Use the treatments as probes. Fund the mechanistic studies of why psychedelics, oxygen and ketones work. The compound that aborts an attack is the best instrument we have for discovering what the attack is, and right now a promising mechanistic lead, sigma-1 signalling and mitochondrial energy, is sitting largely unexamined in this disease.
- Run the gut study. I could not find a dedicated peer-reviewed microbiome study in cluster headache. Do the obvious one. Then properly power the ketogenic trial that 2018 pointed toward, with a microbiome arm built in.
- Reopen the psychology question. Run the ISTDP and adverse-childhood-experience studies in cluster headache that have never been run. I suspect you will find something genuinely useful: data, and a richer picture of the whole patient.
- Test the modality, do not assume the monotherapy. Today's evidence-based treatment is acute oxygen and triptans, with preventives such as verapamil and, in selected cases, corticosteroids, lithium, topiramate or galcanezumab. Nutrition, the microbiome, high-dose supplementation and psychotherapy are not established cluster headache treatments, and I am not claiming they are. My argument is narrower: a whole-system disease deserves a research agenda that actually tests whole-system care, rather than ruling it out by never looking.
I am genuinely excited by this research, and I want to be completely clear about that, because everything in this piece comes from someone who cares deeply and feels real hope, not someone who has given up. The genetics, the imaging, the biomarkers, the new targets, all of it is moving quickly. So here is my hope, offered with real gratitude to the people who have spent their careers on a condition the world overlooks: let us keep widening the lens to the whole patient, let us keep building bridges between the silos, and let us keep listening to the community that has so often pointed toward the answers. The most important questions are also the most exciting ones, and I genuinely cannot wait to see where you take them.
A note on safety and scope. I am a patient and a researcher, not a doctor, and nothing here is medical advice. High-dose vitamin D should only be undertaken with a clinician and with monitoring of serum calcium and 25-hydroxyvitamin D, and it is not safe for everyone, including people with conditions such as sarcoidosis or primary hyperparathyroidism. A ketogenic diet is a significant medical intervention that should be supervised, particularly for anyone with metabolic, kidney or cardiac conditions. The psychedelic and DMT material is discussed as science and as patient experience, not as encouragement; these substances are largely illegal and, for cluster headache, remain unproven, with no controlled cluster-headache trial evidence for DMT. The microbiome device patent referenced is a legal claim, not evidence of efficacy. Where I have described something as a hypothesis, I mean exactly that. The peer-reviewed sources below are where the established claims come from; the patient-reported material is identified as such in the text.