4.5 Surgical history — what was tried, and why most of it was abandoned

4.5.1 Timeline

historical + peer-reviewed Approximate chronology, assembled from the sources below:

Period	Development
Early 1900s	Sphenopalatine ganglion blockade used to terminate cluster attacks; “clinically difficult to perform”, never widely adopted (PMC4119587)
Mid–late 20th century	Destructive trigeminal procedures explored: gangliorhizolysis, rhizotomy, nerve section, glycerol/alcohol injection
1988	Mathew NT, Hurt W. “Percutaneous radiofrequency trigeminal gangliorhizolysis in intractable cluster headache.” Headache 1988;28:328-331 (cited in Leone, IHS)
1993	Kirkpatrick PJ, O’Brien MD, MacCabe JJ. “Trigeminal nerve section for chronic migrainous neuralgia.” Br J Neurosurg 1993;7:483-490 (cited in Leone, IHS)
1998	Lovely TJ, Kotsiakis X, Jannetta PJ — microvascular decompression series, Headache 1998;38:590-594 (PMID 11398301)
1998	Ford et al. — first report of gamma knife radiosurgery of the trigeminal nerve for chronic CH (cited in JNNP gamma knife trial)
2000–2001	Leone and colleagues, Milan: first hypothalamic DBS (Karger 2023) — the pivot from destructive to neuromodulatory surgery
2002	Peres MF et al. — greater occipital nerve blockade series, Cephalalgia 2002;22:520-522 (cited in Leone, IHS)
2003	Jarrar RG, Black DF, Dodick DW, Davis DH — trigeminal nerve section outcomes, Neurology 2003 (PMID 12707445)
2005	Donnet et al. — prospective gamma knife trial, negative and toxic (JNNP 2005;76(2):218)
2007	First implanted ONS series in CCH (PMC8665918)
2012–2013	SPG microstimulator: European approval Feb 2012; Pathway CH-1 published 2013
2019	Pathway CH-2 published; ATI collapses
2021–2026	ICON, L-ICON, Fogh-Andersen; SPG device commercially unavailable; destructive surgery essentially historical

4.5.2 Trigeminal nerve section / rhizotomy

peer-reviewed Jarrar RG, Black DF, Dodick DW, Davis DH. “Outcome of trigeminal nerve section in the treatment of chronic cluster headache.” Neurology 2003 Apr 22;60(8):1360-1362, DOI 10.1212/01.WNL.0000055902.23139.16, PMID 12707445. 17 patients with intractable chronic CH. The abstract’s conclusion is favourable: trigeminal nerve section “is an effective treatment with acceptable morbidity for a carefully selected group of patients.”

peer-reviewed But the outcome data tell a harsher story. A later review reports on the same series: 76% of the 17 patients experienced “long-term full or near complete pain relief” — “However, adverse effects were dramatic, with one death, cerebrospinal-fluid leaks, [and] a case of meningitis.” In the Jarrar series, two patients needed surgery for corneal anaesthesia to prevent blindness. The review also notes: “Other groups have reported negative outcomes with CH continuing after trigeminal nerve root section” (Curr Pain Headache Rep 2014;18(8):438 / PMC4119587).

One death and two sight-threatening corneal complications in a 17-patient series. The primary paper’s phrase “acceptable morbidity” and the review’s phrase “adverse effects were dramatic” describe the same data. That conflict is worth seeing clearly, and I have not smoothed it: the surgical literature of this era routinely used framing that a modern patient would not accept.

peer-reviewed Leone’s IHS history chapter lists the complications of destructive trigeminal procedures generally: diplopia, hyperacusis, jaw deviation, corneal anaesthesia, and anaesthesia dolorosa. It notes that complete trigeminal analgesia is necessary to obtain good results — i.e. you must deliberately deaden the whole face to have a chance of success — and that “the risk of a contralateral recurrence after surgery is rather high” in patients whose attacks alternate sides. Long-term ophthalmic follow-up is “highly recommended to avoid corneal ulcers” (Leone M, IHS chapter on non-pharmacological treatment).

Why abandoned: three converging reasons, per Leone: “few destructive procedures are associated with long-lasting benefit”, “side effects can be severely debilitating”, and candidates must be restricted to “chronically intractable patients with unilateral headaches and no history of side shift” (Leone, IHS). Add anaesthesia dolorosa — an intractable deafferentation pain in a numb face that is arguably worse than what you started with — and the arrival of non-destructive neuromodulation from 2001, and the procedure had no remaining rationale.

4.5.3 Percutaneous radiofrequency trigeminal gangliorhizolysis

peer-reviewed Mathew NT, Hurt W, Headache 1988;28:328-331. Leone’s review lists this among procedures “sometimes reported to be successful”, subject to the same requirement for complete trigeminal analgesia, the same complication list (diplopia, hyperacusis, jaw deviation, corneal anaesthesia, anaesthesia dolorosa), and the same high contralateral recurrence risk in side-shifting patients (Leone, IHS). No numerical success or complication rates are given in the sources I could retrieve — a gap, honestly noted.

peer-reviewed Radiofrequency directed at the SPG rather than the trigeminal ganglion has fared somewhat better and is still occasionally used: a prospective analysis of radiofrequency ablation or pulsed radiofrequency in 37 CCH patients, mean follow-up 68 months, found 30% not improved and 5 patients with total relief of headache and autonomic symptoms, with “a low complication rate… in comparison with SPG stimulation” (PMC8665918). In 3 patients pulsed radiofrequency failed and thermocoagulation of the SPG then succeeded, with follow-up ending at 11 months. The 2021 review grades SPG radiofrequency as: no RCTs, two small positive case series.

4.5.4 Gamma knife radiosurgery

peer-reviewed After Ford et al.’s positive 1998 report, Donnet et al. ran a prospective open trial: 10 patients (9 men, 1 woman), mean age 49.8 (32–77), mean CCH duration 9 years (2–33), targeting the cisternal segment of the trigeminal nerve with a single 4 mm collimator at 80–85 Gy maximum dose; mean follow-up 13.2 months (J Neurol Neurosurg Psychiatry 2005;76(2):218).

Outcome	n
No improvement	2
No further attacks	3
Dramatic improvement (a few attacks/month or very few over 6 months)	3
Pain-free then recurrence (1 week; 2 weeks)	2

Complications: paraesthesia without hypoaesthesia 3; hypoaesthesia 1; deafferentation pain 1. The authors concluded that “the rate and severity of trigeminal nerve injury appeared significantly higher than in trigeminal neuralgia”, that the study “did not support the positive results of Ford et al.”, and that morbidity was “significant in relation to the low rate of pain cessation, making the procedure less attractive even for the more severely affected subgroup of patients” (JNNP 2005).

peer-reviewed Systematic review: Franzini A, Clerici E, Navarria P, Picozzi P. “Gamma Knife radiosurgery for the treatment of cluster headache: a systematic review.” Neurosurg Rev 2022;45(3):1923-1931, DOI 10.1007/s10143-021-01725-9, PMID 35112222. 5 studies, 52 patients described, 48 analysed (trigeminal nerve 34, SPG 1, both 13):

• Initial meaningful pain reduction: 60–100% across individual studies; aggregate 37/48 (77%).
• Meaningful pain reduction persisting at last follow-up: 20/48 (42%) — i.e. roughly half of initial responders lost the benefit.
• Trigeminal sensory disturbances: 28/48 (58%).
• Deafferentation pain: 3/48 (6%).
• Conclusion: gamma knife targeting the trigeminal nerve or SPG is associated with “a frequent risk of trigeminal disturbances” and “possible deafferentation pain”; long-term results are “controversial”.

Why largely abandoned: a 58% rate of permanent-risk trigeminal sensory disturbance and a 6% rate of deafferentation pain, in exchange for a benefit that persists in 42% — and the benefit is irreversible in neither direction (you cannot turn radiation off). Compared with a reversible, non-destructive implant, the risk-benefit maths collapsed.

4.5.5 Microvascular decompression (MVD)

peer-reviewed Lovely TJ, Kotsiakis X, Jannetta PJ. “The surgical management of chronic cluster headache.” Headache 1998;38(8):590-594, DOI 10.1046/j.1526-4610.1998.3808590.x, PMID 11398301. University of Pittsburgh — the Jannetta group, who invented MVD for trigeminal neuralgia.

• 28 patients (2 bilateral) underwent 39 operations: microvascular decompression of the trigeminal nerve, alone or combined with section and/or MVD of the nervus intermedius. Mean follow-up 5.3 years.
• Initial success (≥50% relief): 22/30 first-time procedures (73.3%).
• >90% relief: 15/30 (half).
• Long-term success (“excellent or good”): 46.6%.
• Repeat procedures: 7 of 8 failed at long-term follow-up (87.5% failure).
• “Morbidity and neurological deficit from the operations was minimal” (no numerical complication rate given).
• Conclusion: “Chronic cluster headache remains a debilitating and poorly controlled syndrome.”

Why abandoned for CH (but retained elsewhere): the durable success rate of ~47% after a posterior fossa craniotomy, with essentially no benefit from reoperation, is a poor trade. MVD remains a first-line procedure for trigeminal neuralgia and is still occasionally considered for SUNCT/SUNA — indeed the 2014 review concluded that “except for SUNCT/SUNA treated with microvascular decompression, neurostimulation techniques will be the choice in the future” (PMC4119587). For cluster headache specifically, it is essentially historical.

4.5.6 Greater occipital nerve blockade (still used, and standing apart from the rest)

peer-reviewed Peres MF, Stiles MA, Siow HC, et al. “Greater occipital nerve blockade for cluster headache.” Cephalalgia 2002;22:520-522. 14 patients: 4 good response, 5 moderate response, 5 no response (cited in Leone, IHS).

peer-reviewed Leone’s assessment: the injection method “was widely used but had not been subjected to systematic evaluation”, and without controlled studies it is hard to know whether the effect is “a pure corticosteroid effect on occipital muscles or… a specific effect on the greater occipital nerve” (Leone, IHS).

GON blockade is the one procedure on this list that was never abandoned. It is minimally invasive, reversible, cheap, and is explicitly listed by the EHF among the preventive treatments that count towards refractoriness criteria (EHF consensus 2014). It requires “appropriate clinical experience” to perform. If you have not had GON blockade, you have not exhausted the non-device options.

4.5.7 The EAN’s summary judgement on surgery

peer-reviewed “Surgical procedures are not indicated in most patients with cluster headache” (EAN 2023).

peer-reviewed Leone’s framing remains the standard: surgery is “a last-resort measure for treatment-resistant cluster patients and should be considered only after all pharmacologic options have been exhausted” (Leone, IHS).

4.5.8 The pattern worth naming

Read as a whole, the surgical history of cluster headache shows a single recurring failure mode: destroying the anatomical structure through which the pain is transmitted does not reliably stop cluster headache, because cluster headache is not generated there. Complete trigeminal analgesia — a fully numb face — was required for the best results, and even then attacks continued in some patients, recurred on the other side in others, and were sometimes replaced by anaesthesia dolorosa. The functional imaging that showed posterior hypothalamic activation in 1998, and Leone’s DBS work from 2000–2001, reframed CH as a central disorder with peripheral expression, and every therapy since has been modulatory and reversible rather than destructive. That shift — from cutting to stimulating — is the single most important development in the surgical history of this disease.