5.1 Introduction

While the initial reports from two patients indicating that their migraine headaches (MHs) had disappeared following cosmetic surgery were intriguing, one could not help being skeptical about the connection between cosmetic forehead procedures and migraine symptom disappearance. This onset rules out the possibility that the positive outcomes are related to a placebo effect, as claimed by the opponents of surgery, since neither the patients nor the surgeon was anticipating such a result. ^{1 – 3} It would have been totally imprudent to ignore such a report by two independent patients. It was, therefore, sensible to design a study to investigate this connection. Thus, a retrospective study was conducted to determine whether there was an association between the removal of the corrugator supercilii muscle and the elimination or significant improvement of MH. ⁴ Questionnaires were sent to 314 consecutive patients who had undergone corrugator supercilii muscle resection through endoscopic, transpalpebral, or open forehead rejuvenation procedures. The patients were queried as to whether they had a history of MH before the surgery and, if so, whether the headaches significantly improved or disappeared after surgery. If the answer was affirmative, then the patients were further questioned about the duration of the improvement or cessation of the headaches and the relationship to the timing of the surgery. After an initial evaluation of the completed questionnaires, a telephone interview was conducted to confirm the initial answers and to obtain further information necessary to ensure that the patients had a proper diagnosis of MH based on the International Headache Society criteria.

Of the 314 patients, 265 (84.4%) either responded to the questionnaire or were interviewed, or both. Of this group, 16 patients were excluded because of the provision of insufficient information to meet the International Headache Society criteria, the presence of organic problems, and other exclusions mandated by study design. Thirty-nine (15.7%) of the remaining 249 patients had MH that fulfilled the society criteria. Thirty-one of the 39 (79.5%) with preoperative migraine noted elimination or significant (>50%) improvement in MH immediately after surgery (p < 0.0001), and the benefits lasted over a mean follow-up period of 47 months. When the respondents with a positive history of MH were further divided, 16 patients (p < 0.0001; McNemar) noticed improvement over a mean follow-up period of 47 months, and 15 (p < 0.0001; McNemar) experienced total elimination of their MH over a mean follow-up period of 46.5 months. When divided by MH type, 29 patients (74%) had nonaura MH. Of these patients, the headaches disappeared in 11 patients, improved in 13 patients, and did not change in 5 patients (p < 0.0001). Ten patients experienced aura-type headaches, which disappeared or improved in 7 of the patients and did not change in 3 of the patients (p < 0.0001). This study suggested for the first time that there is indeed a strong correlation between forehead rejuvenation and potentially removal of the corrugator supercilii muscle and the elimination or significant improvement of MH. However, being a retrospective study, the results were open to criticism. This study offered additional compelling evidence against the placebo effect being the reason for the improvement or elimination of MH after surgery since these patients did not undergo surgery for treatment of MH nor were they aware of such a connection between their surgeries and MH.

Intrigued by the findings from this study, we designed a prospective pilot study investigating the role of corrugator resection and removal of the zygomaticotemporal branch of the trigeminal nerve in MH on a small group of patients (n = 29 [24 women, and 5 men], with an average age of 44.9 years [range: 24–63 years]). ⁵ The patients completed a comprehensive MH questionnaire. The research team neurologist evaluated the patients with moderate to severe MH to confirm the diagnosis using the criteria set forth by the International Headache Society. The team’s plastic surgeon injected 12.5 units of botulinum toxin A (BT-A) into each corrugator supercilii muscle group. The patients were asked to maintain an accurate log of their MH and to complete a monthly questionnaire documenting pertinent information related to their headaches. Patients who observed complete elimination of the MH following injection of BT-A underwent resection of the corrugator supercilii muscle group. Those who experienced only significant improvement and who had temple headaches underwent transection of the zygomaticotemporal branch of the trigeminal nerve with repositioning of the temple soft tissues, in addition to removal of the corrugator supercilii muscles. Once again, patients kept a detailed postoperative record of their headaches.

Twenty-four of 29 patients (82.8%; p < 0.001) reported a positive response to the injection of BT-A, 16 (55.2%; p < 0.001) observed complete elimination, 8 (27.6%; p < 0.04) experienced significant improvement (at least 50% reduction in intensity or severity), and 5 (17.2%) did not notice a significant change in their MH. Twenty-two of the 24 patients who had a favorable response to the injection of BT-A underwent surgery, and 21 (95.5%; p < 0.001) observed a postoperative improvement. Ten patients (45.5%; p < 0.01) reported elimination of MH and 11 patients (50.0%; p < 0.004) noted a considerable improvement. For the entire surgical group, the average intensity of the MH reduced from 8.9 to 4.1 on an analogue scale of 1 to 10, and the frequency of MH changed from an average of 5.2 per month to an average of 0.8 per month. For the group who only experienced an improvement, the intensity fell from 9.0 to 7.5 and the frequency was reduced from 5.6 to 1.0 per month. Only one patient (4.5%) did not notice any change. The follow-up ranged from 222 to 494 days, the average being 347 days.

This study confirmed the potential value of surgical treatment of MH since as many as 21 of 22 patients benefited significantly from the surgery. It was also evident that injection of BT-A is an extremely reliable predictor of surgical outcome. This study further confirmed our findings from the retrospective study. However, being a pilot study, it needed further validation since it did not include a control group. The study also raised the question as to why all of the patients did not respond to surgery if the surgery was indeed the answer. Furthermore, now that the connection was seemingly strong, it was time to start exploring the rationale behind the efficacy of the surgical treatment of MH.

It was about this time that research indicated the role of BT-A in treatment of MH. ⁶ Based on the findings from the use of BT-A and preliminary surgical findings, we suspected that the elimination of the muscle function was the reason for these surprising positive findings. Additionally, after carefully listening to the patients who had undergone surgery, it became clear that most of these patients who had observed incomplete or partial elimination of MH had pain in the sites other than those that were the target of the surgery such as the occipital region and the retrobulbar area. Anatomical study of these sites seemed to be the next logical step.

Our investigation related to the anatomy of the greater occipital nerve (GON) revealed some enthralling findings. ⁷ The objective of this study was to determine the GON course, conceivable anatomic variations, and potential sites of entrapment or compression. Twenty fresh cadaver heads were dissected to trace the course of the GON from where the occipital nerve penetrates the semispinalis muscle. It was discovered that the nerve travels through the thickness of the muscle on every specimen. Standardized measurements were used on 14 specimens to determine the location of the emergence of the nerve using the midline and occipital protuberance as landmarks. The point of emergence of the GON from the semispinalis muscle was determined to be centered approximately 3 cm below the occipital protuberance and 1.5 cm lateral to the midline.

As we were completing this anatomical study, a patient who had undergone septorhinoplasty reported that his MH had disappeared. Searching for a potential connection in the literature, we came across an article from Switzerland demonstrating that indeed this surgery had been performed for MH and improved or eliminated MH on many patients. ⁸ In this study, 299 patients with various types of headaches, including migraine, cluster, and so-called idiopathic headaches, were operated on between 1973 and 1991. Septal correction, resection of the middle concha, ethmoidectomy, and sphenoidectomy on the corresponding headache side or occasionally on both sides were carried out. Most patients (235; 78.5%) were totally asymptomatic postoperatively; 34 (11.5%) had a sensation of pressure in the head on rare occasions but no further migraines, and 30 (11%) continued to experience headaches that occurred only rarely and were mild and of short duration.

Up to this point, the prevailing theory behind MH pathogenesis included four basic concepts: (1) neuronal hyperexcitability during the interictal phase; (2) cortical spreading depression (CSD) as the basis of aura; (3) trigeminal nerve activation at a peripheral and central origin that accounts for the headache; and (4) the provocative concept that progressive central sensitization is possibly related to periaqueductal gray matter damage. Of these four concepts, our findings supported peripheral activation of the trigeminal nerve. MH is postulated to be caused by dilatation of large vessels innervated by the trigeminal nerve. Vasodilatation is the consequence of release of calcitonin gene-related peptide, substance P, and neurokinin A found in the cell bodies of trigeminal neurons. What prompts the release of these peptides remains unclear. We proposed that it may be the mechanical stimulation of the potentially hyperexcited, peripheral sensory nerves that instigates the cascade of events. In fact, in 3 out of 4 trigger sites that we studied, the sensory nerves traverse the muscles. As to the fourth site, contact between the turbinates and the deviated septum may cause MH in some patients.

More knowledge about the anatomy of the zygomaticotemporal branch of the trigeminal nerve seemed in order at this time. A study was conducted to determine the site of emergence of this nerve from the deep temporal fascia and to identify the number of its accessory branches and their locations. An initial pilot study, conducted on 20 patients, concluded that the main zygomaticotemporal branch emerges from the deep temporal fascia at a point on average 17 mm lateral and 6 mm cephalad to the lateral palpebral commissure, commonly referred to as the lateral canthus. These measurements, however, were obtained after dissection of the temporal area, rendering the findings less reliable. Another study included 20 consecutive patients, 19 women and 1 man, between the ages of 26 and 85 years, with an average age of 47.6 years. ⁹ Those who had a history of previous trauma or surgery in the temple area were excluded. Before the start of the endoscopic forehead procedure, the likely topographic site of the zygomaticotemporal branch was marked 17 mm lateral and 6 mm cephalad to the lateral orbital commissure on the basis of the information extrapolated from the pilot study. The surface mark was then transferred to the deeper layers using a 25-gauge needle stained with brilliant green. After endoscopic exposure of the marked site, the distance between the main branch of the trigeminal nerve or its accessory branches and the tattoo mark was measured in posterolateral and cephalocaudal directions. In addition, the number and locations of the accessory branches of the trigeminal nerve were recorded.

On the left side, the average distance of the emergence site of the main zygomaticotemporal branch of the trigeminal nerve from the palpebral fissure was 16.8 mm (range: 12–31 mm) in the posterolateral direction and an average of 6.4 mm (range: 4–11 mm) in the cephalad direction. On the right side, the average measurements for the main branch were 17.1 mm (range: 15–21 mm) in the posterolateral direction and 6.65 mm (range: 5–11 mm) in the cephalic direction. Three types of accessory branches were found in relation to the main branch: (1) accessory branch cephalad, (2) accessory branch lateral, and (3) accessory branches in the immediate vicinity of the main branch. This anatomical information has proven colossally helpful in injection of BT-A in the temporalis muscle to eliminate the trigger sites in the parietotemporal region and surgical management of MHs triggered from this zone.

Armed with the anatomical information and having an understanding that there are at least four common trigger sites, we designed a prospective randomized study. ¹⁰ The purpose of this study was to investigate the efficacy and safety of surgical deactivation of MH trigger sites. The study was designed by a team of a board-certified neurologist who was a headache specialist, a biostatistician, and a plastic surgeon. The number of patients in each group and the randomization process were decided by the experienced biostatistician. Of 125 patients diagnosed with MH and selected by the neurologist of the team using the International Headache Society criteria, 100 were randomly assigned to the treatment group and 25 served as controls, with the 4:1 allocation decided by the biostatistician considering our results from the retrospective and prospective pilot studies. Patients in the treatment group were injected with BT-A for identification of trigger sites (not for therapeutic reasons). Eighty-nine patients who noted improvement in their MH for 4 weeks following injection of BT-A underwent surgery. Eighty-two of the 89 patients (92%) in the treatment group who completed the study demonstrated at least 50% reduction in MH frequency, duration, or intensity compared with the baseline data; 31 (35%) reported elimination and 51 (57%) experienced improvement over a mean follow-up period of 396 days. In comparison, 3 of 19 control patients (15.8%) recorded reduction in MH during the 1-year follow-up (p < 0.001), and no patients observed elimination. All variables improved significantly for the treatment group when compared with the baseline data and the control group, including the Migraine-Specific Quality of Life questionnaire (MSQ), the Migraine Disability Assessment score (MIDAS), and the Short Form-36 Health Survey (SF-36). The mean annualized cost of migraine care for the treatment group ($925) reduced significantly compared with the baseline expense ($7,612) and the control group ($5,530; p <0.001). The mean monthly number of days lost from work for the treatment group (1.2 days) was reduced significantly compared with the baseline data (4.41 days) and the control group (4.4 days; p < 0.003).

The common adverse effects related to injection of BT-A included discomfort at the injection site in 27 patients after 227 injections (12%), temple hollowing in 19 of 82 patients (23%), neck weakness in 15 of 55 patients (27%), and eyelid ptosis in 9 patients (10%). The common complications of surgical treatment were temporary forehead and occipital anesthesia on 89 patients (100%) on the operated sites, temporary dryness of the nose in 12 of 62 patients who underwent septum and turbinate surgery (19.4%), rhinorrhea in 11 (17.7%), intense scalp itching in 7 of 80 patients who underwent forehead surgery (8.8%), and minor hair loss in 5 patients (6.3%). Sinus infection was observed in 3 patients (3.4%). After this experience, we altered our postoperative antibiotic course and had patients continue taking antibiotics while their intranasal surgical sites were healing and from thereon, sinus infections have been eliminated. Recurrence of the septal deviation was noted in 12 patients (13.5%), but these were not symptomatic. Intense itching was reported by 7 patients (7.9%). All 10 parameters in validated tools demonstrated measurable and statistical significant improvement in the MSQ MIDAS, and SF 36.

This study demonstrated that surgical deactivation of migraine trigger sites can eliminate or significantly reduce migraine symptoms effectively and safely. Interestingly, when the results of the surgery within the first 3 postoperative months were compared to the subsequent three quarters, there was a steady, statistically significant pattern of improvement in each quarter. This finding is totally contrary to what one would expect from a placebo effect, where the initial higher success rate would decline over time. Additionally, when the surgical outcome was analyzed independently at each trigger site, there was significant increase in both improvement and elimination rates, which indicated that some of the failures or suboptimal results had to do with the fact that we failed to detect and deactivate all trigger sites in every patient. This has been confirmed with a more recent study. ¹¹

In a 5-year follow-up arm of the above study, ¹² patients were examined by the research team and were asked to complete the MH log, the SF-36, MSQ, and MIDAS questionnaires that they had completed previously, at a minimum of 60 months after surgery. The neurologist and the surgical team were able to follow 79 patients for 5 years. Ten patients who had observed significant improvement but not elimination underwent deactivation of additional (different) trigger sites during the 5-year follow-up period in an attempt to eliminate their headaches. This opens the study to criticism. However, the overall outcome with or without inclusion of these 10 patients was no statistically significant difference. It would have been improper to deny the opportunity for additional improvement in the quality of life for these patients who had agreed to participate in the study merely to prevent criticism.

Sixty-one of 69 (88%) patients continued to have a positive response to the surgery after 5 years (p < 0.0001). Twenty (29%) reported complete elimination of MH, 41 (59%) noticed a significant decrease in frequency, severity, and duration of MH, and 8 (11%) observed no significant change (p < 0.0001). All of these endpoints were assessed independently. When compared to the baseline values, all 10 measured variables including MIDAS, MSQ, and SF-36 improved significantly at 60 months (p < 0.0001), confirming that the achieved positive results were enduring.

To offer the ultimate evidence of the efficacy of surgical decompression of MH trigger sites, we planned a study to include a sham surgery arm with enormous inherent challenges. After obtaining institutional review board (IRB) approval from two separate boards, patients with a single trigger site or MH predominantly emanating from one site were recruited and randomized to receive placebo surgery or real surgery. ¹³ Two board-certified and fellowship-trained headache specialists selected the patients based on the International Headache Society criteria. The number of subjects assigned to the real surgery and sham surgery groups as well as their randomization technique into the groups was decided by a highly respected and experienced biostatistician. The patients underwent injection of BT-A to simulate the surgery effect by paralyzing the muscle (not for therapeutic reasons). The patients in the sham surgery group underwent the incision and the nerves and muscles were identified, but the muscles were not touched and the nerves were not decompressed.

Like any other study completed in our center, all data were collected and stored by the study nurse coordinator without any input from the surgical team and delivered directly to the biostatistician of the team. The data were analyzed and the results section of the article was written by the biostat-istician who also prepared the tables and graphs without any input from the surgical team.

Seventy-five of 76 patients enrolled in the study completed the 1-year follow-up. Forty-nine of those patients served in the treatment group and 26 were assigned to the sham surgery group. Forty-one of the 49 (83%) patients benefitted from the real surgery compared to 15 of 26 (56%) patients in the sham surgery group (p = 0.014). While 28 (57%) patients in the real surgery group observed complete elimination, only 1 patient in the sham surgery group reported elimination (p < 0.0001). ¹³