Occipital Injections for Trigemino-Autonomic Cephalalgias: Evidence and Uncertainties
- First Online:
- Cite this article as:
- Leroux, E. & Ducros, A. Curr Pain Headache Rep (2013) 17: 325. doi:10.1007/s11916-013-0325-z
Cluster headache is a debilitating disorder. Oral prophylactic treatments may act with a significant delay, cause side effects, or fail to control the attacks. Injections targeting the occipital nerve have raised interest for the management of CH. Their efficacy is thought to result from the anatomical convergence of trigeminal and cervical afferents in the trigeminal nucleus caudalis. Efficacy and safety of occipital injections are now documented by 2 randomized controlled trials and several case series, though the optimal technique and substance to be injected are still subject to discussion due to varied approaches in the published studies. The evidence supports the use of injected steroids, with or without the addition of an anesthetic. Side effects of local pain are common, but unlikely to be severe. Systemic effects related to steroid absorption are reported but infrequent. Occipital injections provide a rapid benefit on the frequency of attacks and can be used as an adjunct to an oral prophylactic for a quicker improvement. Whether or not this approach can be used without any oral prophylaxis is still to be determined. The technique is easy to master, has a low cost, and should be learned by physicians involved in CH management.
KeywordsCluster headacheGreater occipital nerveBlocksSuboccipital injectionsTrigemino-autonomic cephalalgias
The description of an anatomical convergence of the occipital nerves and the trigeminal nerve in the trigeminal nucleus caudalis at the brainstem level has changed our way of understanding headache and neck pain. Easily accessible, the greater occipital nerve (GON) has become a therapeutic target for a vast array of headaches. Trigemino-autonomic cephalalgias (TACs) are severe and highly impairing primary headaches of central origin and are not related to GON lesions. Nevertheless, the evidence supporting the use of GON injections for TACs, particularly cluster headache (CH), is growing. Many questions remain unanswered as to the substance or combination that is most effective, the number of injections needed, the optimal technique and the safety of repeated injections. This article reviews the available evidence on GON injections for the management of cluster headache and other TACs.
Greater Occipital Nerve Anatomy and Functional Convergence
The greater occipital nerve (GON) is a branch from the posterior ramus of the C2 cervical root, with an occasional contribution from C3 root by the cervical plexus of Cruveilhier [1–3]. This sensory nerve has no motor function. It emerges between the posterior arch of the atlas and the lamina of the axis. The nerve runs along the suboccipital triangle, over the obliquus capitis inferior muscle. It ascends between the obliquus capitis superior and the semispinalis capitis muscle, which it pierces before crossing the aponeurosis of the trapezius muscle [4, 5]. Near the occipital bone the nerve may divide into branches to supply the skin of the posterior aspect of the head up to the vertex. At the occipital nuchal line, the GON can lie in proximity to the occipital artery [5, 6], which can be palpated in some cases. Natsis et al. described that the nerve and the artery pierce the trapezius at separate points and sometimes well apart, as suggested by other authors .
Clinical relationship between the GON and the frontal area was suggested in 1940 by Hadden who described occipital neuralgia: «the first symptom is a suboccipital pain (…) then it spreads anteriorly in the temporal area and in the supra-orbital region» . In 1945 Skillern wrote: «a needle piercing the greater occipital nerve produced pain radiating to the ipsilateral supraorbital notch» . The stimulation of the greater occipital nerve induces increased central excitability of dural afferent input in rats . Bartsch also demonstrated increased responses in trigeminocervical nociceptive neurons to cervical input after stimulation of the dura mater . In healthy volunteers and in CH patients, Busch showed that a GON blockade did increase the R2 latencies of the blink reflex [12, 13]. These experiments suggest a 2-way functional continuum through the trigeminal nucleus caudalis (TNC) between the upper cervical segments and the trigeminal branches involved in cranial and facial nociception. This convergence may play a role in CH pathophysiology  and has supported the development of occipital nerve stimulation for cluster headache and other TACs, a technique that has shown success even in refractory cases [15–17].
Use of GON Block for Multiple Conditions
GON injections have been used to treat a vast array of secondary and primary headaches, from post-surgical pain from peripheral afferents to centrally-driven conditions such as cluster headache [18••, 19, 20•]. A response to a GON injection should not be considered diagnostic [21•]. The majority of these studies have been uncontrolled, and therefore the contribution of the placebo effect has not often been quantified. Advantages of GON injections are a low cost and potential accessibility even in developing countries where some drugs are not available or too expensive for patients.
Trigemino-autonomic cephalalgias (TACs) include cluster headache, paroxysmal hemicranias, and SUNCT (short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing) and SUNA (short-lasting unilateral neuralgiform attacks with cranial autonomic symptoms) [22, 23]. All these conditions are characterized by unilateral attacks of periorbital, frontal or temporal pain with autonomic signs. Parasympathetic hyperactivity includes conjunctival injection, lacrimation, nasal congestion or rhinorrhea, facial sweating, and eyelid edema. Horner’s sign (ptosis and miosis) suggests sympathetic hypoactivity . These conditions are distinguished by the frequency and duration of the attacks, with some degree of overlap and the coexistence of different phenotypes in some patients [25, 26]. Cluster headache is the most frequent form of TACs. The attacks last from 15 min to 3 h and can recur up to 8 times per day. Restlessness is a typical behavior during the attack. The episodic subtype (ECH) involves an alternance of active periods and remissions while the chronic cases have a baseline frequency of attacks with exacerbations . Chronic cases (CCH) often need continuous prophylaxis, while episodic cases need temporary prophylaxis during the bout. The management of cluster headache has been reviewed elsewhere [28–30].
Paroxysmal hemicrania is characterized by attacks of 2 to 30 min, with a frequency of 5 per day more than half of the time. Episodic and chronic forms are described. Paroxysmal hemicrania should respond to indomethacin by definition, but indomethacin resistant cases are described, and CH may also respond to indomethacin [31, 32]. Though not included in the TAC section of the ICHD-2 classification, hemicrania continua shares many clinical features with paroxysmal hemicranias [33–35], including absolute responsiveness to indomethacin in typical cases [36, 37]. Indomethacin has significant side effects and can be contraindicated in presence of gastric sensitivity, hypertension, or renal insufficiency, underlining the need for other options.
The attacks of SUNCT last from 5 to 240 s and may happen 3 to 200 times per day. SUNCT is rare and many cases have been related to underlying lesions. The management of SUNCT is a challenge, and therapeutic trials are absent due to a very low prevalence . For all TACs, a cure has not been found, and well-tolerated symptomatic therapeutic options are needed.
Open Studies of GON Injections for Cluster Headache
Studies of GON injections for cluster headache
N of patients
Substance injected (equivalent MP)
MP depo 120 mg
ECH >20 HAFD: 9/12
LD 2 % 2–5 mL
CCH >20 HAFD : 5/8
MP depo 160 mg
Good response: 3/16
LD 2 % 2–5 mL
Partial response: 6/16
No response: 7/16
MP depo 120 mg
Good response: 20/20
Part of larger study
LD 1 % 3–4 mL
Mean of 32 HAFD
TC 40 mg (MP 40 mg)
14 HAFD: 4/14
LD 1 % 3 mL
2–4 HAFD: 5/14
No response: 5/14
BM 18 mg and XL 2 % 0.5 mL (MP 96 mg) vs XL 2 % 0,5 mL
Remission in 8/13 active vs 0/10 placebo
HAFD >7 days in 11/13 active vs 0/10 placebo
Additional prophylaxis in 1/13 active vs 9/10 placebo
LD 2 %
Cessation of the bout
BM 2 mg (MP 13 mg)
LD 2 % 3 mL
Complete response: 10/19 (mean 17 HAFD)
ECH or CCH not specified
MP 80 mg
Partial response: 3/19 (mean 52 HAFD)
PL 2 % 5 mL
3 day benefit: 1/15
No benefit >24 h: 14/15
CVZ 3.75 mg (MP 50 mg)
From >2 to ≤2 attacks at day 15 20/21 active vs 12/22 placebo
3 injections (MP 150 mg)
Pain free from day 10 to day 30 in 11/21 active vs 4/22 placebo
48–72 h apart
Attacks over 15 days: 10 vs 30
Remission 7 days earlier
LD 2 % 2 cc
Complete response: 45 %
BM 21 mg
Partial response: 35 %
(MP 112 mg)
No response: 20 %
Randomized-Controlled Trials of GON Injections in Cluster Headache
Two studies are available. Ambrosini studied 16 ECH and 7 CCH patients, comparing 18 mg of betamethasone (mixture of short and long acting) and 0.5 cc of xylocaine 2 % with 0.5 cc of xylocaine 2 % only . ECH patients had a baseline median frequency of 1 attack per day, and CCH patients had a median of 2 attacks per day. 85 % of patients became attack free for 1 week after a single injection, 61 % staying attack-free for 1 month or more. No patient in the placebo group achieved this objective. Ten patients needed an additional prophylactic 1 week after the injection, 1 in the active group and 9 in the placebo group.
The authors of the present review studied 28 ECH and 15 CCH patients, comparing cortivazol injections with placebo, without anesthetics [46••, 47]. Three injections were planned 48 to 72 h apart. Ten of 42 patients did not receive the third injection, including 6 patients in the active group who were in remission after the second injection. All ECH patients were also started on verapamil from the start of the trial. Chronic cases did pursue their usual prophylaxis. All patients included in this study had more than 2 attacks per day, with a mean of 3.9 attacks per day. The primary objective was to decrease the daily frequency of attacks to 2 or less per day following the last injection, and was reached by 95 % of the active group (21 of 22 patients) and 55 % of the placebo group (12 of 22 patients). Patients in the active group had a mean of 10 attacks over the 15 days following inclusion, compared with 30 days in the placebo group. Remission was reached by the ECH patients a mean of 7 days earlier in the active group. There were lower mean doses of verapamil (420 mg vs 564 mg) and fewer patients needing additional prophylactics (5 % vs 27 %) in the active group. At 30 and 90 days, the groups were not significantly different for remission status. This suggested that cortivazol injections were a useful adjunct treatment to rapidly break a period or exacerbation of CH, sparing the use of acute treatments, high dose verapamil, and additional prophylactics.
GON Injections for SUNCT, Paroxysmal Hemicrania, and Hemicrania Continua
Porta-Etessam reported a case of SUNCT in an 82-year-old female responding for 48 h to a GON block with bupivacaine . Due to the rarity of SUNCT, no large series or RCTs are available. Antonaci reported 7 patients with hemicrania continua (HC) and 6 with chronic paroxysmal hemicrania treated with GON blocks with lidocaine, but also lesser occipital (LON), and supra-orbital nerve (SON) blocks . Injections were performed after cessation of indomethacin. No patient responded to the GON or LON injection. Supra-orbital nerve injections had some positive effect for HC. Guerrero reported 9 patients from a 22 patient’s cohort who received GON and/or trochlear nerve injections for hemicrania continua because indomethacin had to be decreased . Injections were chosen based on the pain response to palpation of GON, SON, and trochlear nerves. Five patients received a GON block with 1 cc of 0.5 % bupivacaine and 1 cc of 2 % mepivacaine. Two had a complete benefit and 3 a partial benefit for more than 2 to 3 months. Four patients also received a trochlear injection of 4 mg triamcinolone, 3 with complete benefit and 1 with partial benefit. Afridi also reported 7 patients with hemicrania continua who received a total pf 10 GON injections with lignocaine 2 % and 80 mg of methylprednisolone. One patient had a complete response and 5 a partial response .
What is the Right Technique?
Many anatomical landmarks have been used for GON or suboccipital injections [21•]. Landmarks based on measurements vary and are not fitted to each patient’s anatomy. Many clinicians use the point of pain upon pressure (PUP) along the occipital ridge to localize the nerve but this technique has not been proven to be reliable. We find that in cluster patients, even during an active phase, the GON is not always sensitive to pressure, though allodynia is increasingly reported in CH [51, 52]. The palpation of the occipital artery, suggested by some authors, seems difficult to achieve and not always reliable. In the CH studies, the insertion of the needle was in general 2 cm below the occipital nuchal ridge, midway between inion and mastoid [18••]. We used a more medial approach, targeting the medial third between inion and mastoid [46••], because this point better reflects anatomical studies [5, 53, 54]. Echography-guided and CT-guided blocks have been reported to target the GON more precisely than blind injections [55–57], but these techniques have not been studied yet in TACs. However, a very precise injection might not be essential for therapeutic success as the injected liquid will diffuse in the subcutaneous tissue. We suggest targeting the medial third between inion and mastoid injection point, 2 cm under the ridge. Most importantly, the occipital bone has to be felt to avoid going too deep near the foramen magnum, and the plunger of the syringe has to be retracted prior injection to rule out a vascular puncture.
Steroids and Anesthetics: Are They Necessary, Are They Sufficient?
Suboccipital injections of steroids alone cannot be called nerve “blocks”, since they cause no anesthesia. Local anesthetics inhibit the sodium channels of the pain afferents, the unmyelinated C fibers, and thinly myelinated Aδ fibers . Ashkenazi suggested that steroids offer no additional benefit to anesthetics alone for transformed migraine . In the CH studies, only Busch used anesthetics alone, with negative results . The study by Antonaci for paroxysmal hemicranias and hemicrania continua using only lidocaine was also negative . Anthony tried lidocaine alone on 5 CH patients during a period but between attacks, and this did not prevent attacks to recur 24 h later . Peres used a relatively low dose of steroid , and the results were not as good as other studies. Only the trial by Leroux et al. compared a steroid alone with placebo, with positive results. All the other studies published used a combination. Altogether, these data strongly supports that GON injections of steroids are efficient. Whether a combination with anesthetics is superior to steroids only is not known.
It has been hypothesized that suboccipital steroids may exert their effect through systemic absorption. Oral steroids are used as a transitional treatment for CH, but despite wide clinical experience of efficacy, the supporting literature is limited to a class III trial and small reports [29, 60, 61]. A usual course of oral steroids for CH would require at least 50 mg of prednisone or the equivalent daily for 1 week then a taper, adding-up to 500 mg or more. Recurrence of attacks upon weaning is a common problem. The dose equivalent calculated for the mixtures used by Leroux (for 3 injections) and Ambrosini, are 187.5 and 135 mg of prednisone, which is much lower than what would be needed orally. Although suboccipital steroid injections may provoke systemic effects showing that some absorption is present [62, 63], a complete systemic absorption is unlikely. Anthony showed in a study on cervicogenic headache that 94 % of 180 patients receiving suboccipital methylprednisolone improved for a mean of 23 days compared with 0 % of 50 patients injected intra-muscularly at a remote site . In his study on CH patients, he also reported that 5 patients receiving 120 mg of IM methylprednisolone for another bout were not improved. The local mechanism of action of injected steroids on neurotransmission is still under study. Intra-thecal methylprednisolone decreased spinal glial activation in a rat model of neuropathic pain . Local injection of steroids block C-fiber activation and diminishes allodynia [65, 66]. In humans, methylprednisolone added to lidocaine improved the outcomes of neuropathic pain . Overall, locally injected steroids probably have a direct effect on neurotransmission and fewer side effects than systemic steroids.
Are Repeated Blocks Advisable?
Most published series used a single injection, ipsilateral to the pain. Anthony repeated the injections for 7 ECH patients, either because of failure of the first injection or recurrence of headaches . Six of these patients obtained a benefit. In the same study, all chronic cases had 2 injections, and 2 patients even had 3. The delay between injections was not detailed. Ambrosini also repeated injections for 6 patients, 4 in the placebo group with no description of the effect. The 2 patients in the active group improved after this second injection for 6 and 26 days. The study by Leroux used 3 repeated injections, 48 to 72 h apart [46••]. This approach was based on the French team’s 10 years’ experience in managing CH patients with more than 2 attacks per day. No study has compared one with multiple injections. Repeating the injection a few days later in patients with more than 2 attacks per day is worth trying. Gantenbein suggests that bilateral injections can be more efficient than a unilateral one , but the majority of studies still support the use of unilateral injections.
What are the Possible Side Effects, in the Office and Long Term?
Side effects of greater occipital nerve injections
Number of patients
Volume and substance used
Side effects (number of patients)
MP depo 120 mg
LD 2 % 3–5 mL
MP depo 160 mg
Syncope (1) Significant pain at the injection site (5) moderate for 2, severe or prolonged for 3
LD 2 % 3–5 mL
MP depo 120 mg
Common: occipital fullness
LD 1 % 3–4 mL
Dizziness for 1–2 days (8)
TC 40 mg
LD 1 % 3 mL
BM+XL 2 % 0.5 mL
Transient pain at the site of injection (2)
XL 2 % 0.5 mL
LD 2 % 3 mL
No report specific to CH patients
MP 80 mg
Alopecia (1), syncope (1), dizziness (3)
Prilocaine 2 % 5 mL
No major events reported
Facial hypoesthesia for 4 days (1)
CVZ 3.75 mg (equals MP 50 mg) 3 injections
Neck pain (62 % vs 45 %)
Non-cluster headache (33 % vs 18 %)
Attack schedule changed (3), nausea (2), muscle pains and itching (2)
Vertex paresthesias (1), hand paresthesias (1), Ipsilateral blocked nose (1), facial edema (1), eye dysesthesia (2), eye redness (1)
LD 2 % 2 mL
Facial edema(4), sleep disturbances (4), acne (3), local neck pain (3), oral candidiasis (2), eye dysesthesia (1), heartburn (2), bradycardia (1)
BM 21 mg
Is this Technique Reliable, Are Results Reproducible?
Patients and physicians always hope that a given treatment will work every time. In the Ambrosini study, some patients not responding to a first injection did respond with another. In the Gantenbein study, the response to the second injection (more than 1 month after the first) was reported and compared with the response to the first injection for 24 patients. For patients with an initial complete response, 60 % had complete response again, 30 % a partial response, and 10 % had no response. For patients with initial partial response, none had a complete response, 71 had a partial response again, and 29 had no response. When there was no benefit to the first injection, the second injection generated a complete response for 57 % and no response for 43 %. This suggests that the reproducibility of the benefit from this approach is modest. Even a failure the first time does not preclude the success of another trial. Part of the variation may be attributed to the variability if the technique.
Methodological Issues for Future Trials
Future studies on GON or suboccipital injections for CH should be placebo controlled. Daily frequency of attacks should be recorded instead of subjective terms such as «partial response». The number of headache-free days is an interesting and clinically relevant endpoint but less sensitive than the precise number of attacks or acute treatment use. The severity of attacks is not always accurately evaluated when patients treat early with rapid-acting drugs such as sumatriptan. Evolution of attacks over at least 30 days should be recorded for recurrence. Cluster headache is difficult to study because spontaneous remissions may give false impressions of therapeutic efficacy, especially when the treatment is started after many weeks of attacks. The spontaneous remission bias should be avoided by including only patients who are early in their bout or known to have bouts of a long duration. We did learn from our trial that the HIT-6 and MIDAS are poorly applicable to cluster headache patients because first, attacks can be controlled rapidly and allow a return to function, and second, «lying down» is not a behavior observed with CH attacks. A tool evaluating the quality of life in CH is still lacking. Side effects should be carefully recorded.
Further studies could investigate the efficacy of GON injections without oral prophylaxis to break a bout in ECH, compare injections of steroids alone with injections of steroids with anesthetics, and compare suboccipital with remote intra-muscular injection of steroids.
GON blocks and suboccipital steroid injections targeting the GON have shown efficacy for transitional treatment of CH. They can be used instead of oral steroids, and can be added to other oral prophylaxis such as verapamil and lithium. They may be safer than courses of oral steroids. The exact injection procedure has not been validated, but an approximate anatomical landmark is easy to learn and use. Evidence does not support the use of anesthetics alone. Steroids alone or in combination with anesthetics may be used until further studies show the superiority of one approach over another. The total dose of steroids should be at least 80 mg and no more than 160 mg of methylprednisolone. Repeated GON injections can be used for patients with more than 2 attacks per day. Side effects are not uncommon, though rarely severe, and should be discussed with patients.
Conflict of Interest
Dr. E. Leroux serves as a board member for Pfizer, Merck, Allergan, and Tribute Pharmaceuticals. Dr. Leroux serves as a consultant for Merck and Allergan. Dr. Leroux has received a grant from Allergan. Dr. Leroux has received unrestricted educational grants from Merck, Pfizer, Allergan, and Teva. Dr. Leroux has received payment for lectures from Pfizer, Merck, Johnson and Johnson, Purdue, and Teva.
Dr. Anne Ducros serves as Associated Editor of Cephalalgia. Dr. Ducros has received an academic research grant from CIRC. Dr. Ducros has received payments for lectures from Almirall, Merck Sharpe and Dome, and Pfizer.