Onabotulinumtoxin type A (onabotA) has shown efficacy in chronic migraine (CM). Its precise mechanism of action, however, is unknown.
Onabotulinumtoxin type A (onabotA) has shown efficacy in chronic migraine (CM). Its precise mechanism of action, however, is unknown.
To analyze a potential relationship between calcitonin gene-related peptide (CGRP) and vasoactive intestinal peptide (VIP) levels and response to onabotA in CM.
Adult patients with CM were recruited. Matched healthy subjects with no headache history served as controls. CGRP and VIP levels were determined in samples obtained from the right antecubital vein by ELISA outside of a migraine attack and having taken no symptomatic medication prior to treatment with onabotA. OnabotA was administered according to the PREEMPT protocol every 12 weeks for at least two treatment cycles. A patient was considered as a moderate responder when both: (1) moderate-severe headache episodes were reduced by between 33 and 66%; (2) subjective benefit in a visual scale of 0-100 was recorded by the patient of between 33-66%. Patients were considered as excellent responders when both items improved >66%. Those without improvement of at least one-third in the two items were considered as nonresponders.
We assessed plasma samples from 81 patients with CM and 33 healthy controls. CGRP and VIP levels were significantly increased in CM population vs controls. CGRP and, to a lesser degree, VIP levels were significantly increased in responders vs nonresponders. For CGRP, a threshold of 72 pg/mL positively correlated with 95% of nonresponders. The probability of being a responder to onabotA was 28 times higher in patients with a CGRP level above the threshold of 72 pg/mL. Even though the sensitivity for the calculated threshold for VIP was poor, the probability that CM patients with low CGRP levels will respond to onabotA was significantly higher in those patients with high VIP levels.
Interictal CGRP and, to a lesser degree, VIP levels measured in peripheral blood are of great help in predicting response to onabotA.
Migraine is considered a neurovascular disorder. Either a cortical spreading depression phenomenon or changes in the modulating nociceptive inputs from the raphe and locus coeruleus nuclei from the brainstem are thought to activate the trigemino-vascular system (TVS), which releases vasoactive neuropeptides from the presynaptic nerve terminals, mainly calcitonin gene-related peptide (CGRP) and others, such as vasoactive intestinal peptide (VIP) around leptomeningeal and pericranial vessels.[3, 4] The local release of these neuropeptides induces vasodilation and neurogenic inflammation, which gives rise to the typical pulsating migraine pain.3-5 There are 2 types of migraine in terms of frequency: episodic migraine (EM) (fewer than 15 headaches per month) and chronic migraine (CM) (15 or more headache days per month). The diagnosis of CM is a clinical one. The International Headache Society defines CM as 15 or more headache days per month lasting >4 hours, with at least 8 or more days per month fulfilling migraine criteria. Although the source of pain persistence in CM is not known, it has been suggested that repeated episodes of TVS activation can sensitize central pain pathways and lead to migraine chronification.[7, 8] Supporting this concept, our group has recently reported that patients with active CM show increased interictal, peripheral levels of CGRP and, to a lesser degree, VIP.[9, 10] The levels of these neuropeptides could therefore be reliable biomarkers helping in a more objective diagnosis of CM.
The efficacy of pericranial injections of onabotulinumtoxin type A (onabotA) for the treatment of CM is well established. However, both its mechanism of action in this condition and potential predictors for response to onabotA in CM are not fully described. The aim of this study has been to analyze a potential relationship between interictal CGRP and VIP levels and response to onabotA treatment in a series of CM patients.
Adults attending our headache clinic who had been diagnosed by us as having CM according to current International Headache Society criteria and treated with onabotA were included in this study. All patients fulfilling criteria for analgesic overuse had been detoxified at least once for a minimum of 2 months. Exclusion criteria were pregnant or breast feeding women, excessive use of alcohol, and serious, active somatic or psychiatric diseases. Patients showing the comorbidities usually seen in CM, such as anxiety, depression, or fibromyalgia or with common vascular risk factors were not excluded. A detailed medical and headache history was available for all patients, who had attended our headache clinic a minimum of once per trimester during at least 12 months prior to entry in this study. All patients underwent a general physical and neurological examination. All participants had a normal magnetic resonance imaging study. Diagnosis of CM was confirmed by the use of monthly headache calendars in all patients. For the control group, we recruited matched healthy patients (medical students, residents, nurses, or physicians from our hospital) with a subjective absence of headache.
Following current recommendations in our country (Spain), CM patients were considered for onabotA treatment if they have failed, due to either poor efficacy and/or tolerability, treatment with at least 2 prophylactic medications with demonstrated efficacy in migraine, and belonging to different pharmacological groups. In spite of taking oral preventatives, all patients treated with onabotA in this study continued to fulfill CM criteria, but oral preventatives were continued in order to look for a synergistic effect with onabotA. Without exception, we followed the PREEMPT protocol, that is 155-195 onabot U in 31-39 injection sites. All patients who were treated with onabotA received onabotA at least twice over 2 consecutive periods 12 weeks apart. A patient was considered as a moderate responder to onabotA when both: (1) according to the diary, moderate-severe headache episodes longer than 4 hours (or shorter if treated with symptomatic medications) were reduced by between 33 and 66%; (2) a subjective benefit according to a visual scale of 0-100 was also recorded by the patient of between 33 and 66%. Patients were considered as excellent responders when both subjective and objective items improved by more than 66%. Patients without improvement of at least one third in the 2 items were considered as nonresponders. Both patients and neurologists were blinded to the results of laboratory determinations when evaluating the efficacy of onabotA. The protocol was approved by our ethics committee and all patients signed an informed consent.
CGRP and VIP levels were determined in blood samples obtained before treatment with onabotA in our clinic. Patients rested in a supine position and blood samples were obtained from the right antecubital vein between 9:30 am and noon under fasting conditions in our clinic. The blood was collected, allowed to clot and serum was immediately separated after centrifugation for 10 minutes at 2000 x g. Aliquots were rapidly stored at −80 °C until assayed. All samples were obtained in the absence of acute moderate-severe pain and having taken no symptomatic medication in the previous 24 hours.
Serum CGRP and VIP levels were determined using commercial ELISA kits (USCN Life Science Inc, Hubei, China) strictly following manufacturer's instructions. Absorption levels were measured with a spectrophotometer from Bio-Rad (Hercules, CA, USA). The detection limit of the assay was <4.3 pg/mL for CGRP and <2.34 pg/mL for VIP.
CGRP and VIP levels are described by mean ± standard deviation, and quartile (P25, P50, and P75) values are also reported. Categorical variables are described by relative and absolute frequencies. The non-parametric k-sample Kruskal–Wallis test was used for the comparison of CGRP and VIP levels among the response group. The receiver operating characteristic (ROC) curve and the area under the curve (AUC) were used to measure the diagnostic quality of the CGRP levels. In addition, the Youden criterion (sensitivity + specificity) was used to compute the optimal threshold. Univariate binary logistic regression was also performed. Univariate odds ratio (OR) and 95% confidence interval (CI) are provided. Only the CGRP level was included in a multivariate binary logistic regression with a forward stepwise based on the likelihood ratio.
A total of 81 patients fulfilling CM criteria were included in this study. As a control group, 33 healthy women with no headache history (39.4 ± 13.2 years; 21-61 years) were recruited. The mean age of the CM patients was 46.2 ± 11.0 (range 23-65); only four (4.9%) were males. By history, the average time for which patients had suffered from CM was 10.2 ± 7.6 years. Main comorbidities and treatments taken by the patients when they were enrolled in this study are illustrated in the Table. Regarding the efficacy of onabotA treatment, 61 patients (75.3%) responded and the remaining 20 patients (24.7%) did not notice any significant response. Among those 61 responders, 41 (50.6%) and 20 (24.7%) showed moderate and excellent response, respectively. A number of demographic factors, clinical features, and comorbidities including age, duration of CM, a history of aura, analgesic overuse, depression, fibromyalgia, arterial hypertension or obesity, and treatment with triptans vs no triptan use, using preventatives in monotherapy vs polytherapy or taking topiramate vs not using topiramate were not significantly, or even numerically, different in patients who responded to onabotA vs those who did not respond.
|Migraine without aura||47 (58%)|
|Migraine with aura||34 (42%)|
|Analgesic overuse||35 (43%)|
|Arterial hypertension||11 (14%)|
|Smoking habit||10 (12%)|
|Non-steroidal anti-inflammatory drugs||54 (67%)|
|Angiotensin converting enzyme inhibitors||20 (25%)|
|Other antidepressants||13 (16%)|
|Valproic acid||7 (9%)|
CGRP levels were significantly increased in CM patients (64.9 ± 31.0 pg/mL; range 11.4-157.7) as compared with healthy controls (33.3 ± 15.7 pg/mL; range 15.5-70.8; P < 10−10). CGRP levels in nonresponders (48.3 ± 21.2 pg/mL; range 11.4-110.8; P25 = 37.51, P50 = 45.03, P75 = 61.62) were significantly lower than those in responders (70.4 ± 31.9 pg/mL; range 12.8-157.7; P < .005), but still significantly higher (P < .001) than those of healthy controls. CGRP levels in moderate responders (66.1 ± 28.9 pg/mL; range 12.8-158.4; P25 = 42.88, P50 = 67.03, P75 = 85.48) were numerically lower than those of patients with excellent response (79.2 ± 36.6, range 22.0-157.7; P25 = 48.27, P50 = 83.14, P75 = 95.28, P = NS) (Fig. 1).
VIP levels were significantly increased in CM patients (173.7 ± 150.7 pg/mL; range 20.6-866.6) as compared with healthy controls (88.5 ± 62.3 pg/mL; range 15.5 ± 256.1; P < .001). VIP levels in nonresponders (115.5 ± 76.2 pg/mL; range 29.1-236.4; P25 = 53.23, P50 = 80.25, P75 = 197.31) were significantly lower than those in responders (189.7 ± 162.3 pg/mL; range 20.6-866.6; P < .05), but did not differ from those of controls. VIP levels in moderate responders (160.5 ± 120.9 pg/mL, range 20.6–534.0; P25 = 81.52, P50 = 126.69, P75 = 213.99) were numerically lower than those of patients with excellent response (245.3 ± 213.6 pg/mL; range 54.0-866.6; P25 = 78.88, P50 = 202.08, P75 = 309.28, P = NS) (Fig. 2). There was no significant correlation between either CGRP or VIP levels, response vs no response to onabotA, and any of the analyzed demographic factors, clinical features, and comorbidities (see above).
To evaluate the CGRP and VIP concentrations as potential predictors of response to onabotA in CM, the ROC curve and the AUC were measured. For CGRP, the optimal cut point (Youden index) was achieved at a concentration of 72 pg/mL with and AUC of 0.714 (95% CI: 0.594-0833). This threshold would classify correctly 49.2% of responders (sensitivity) and 95.0% of nonresponders (specificity) (Figs. 3 and 4). The probability of being a responder to onabotA was 28 (OR: 18.39) times higher in CM patients with a CGRP level above the threshold of 72 pg/mL. When the CGRP level was considered as continuous variable, the OR was 1.032 (95% CI 1.008-1.056), ie, for each unit (pg/mL) of CGRP level, the probability that a patient responds to the treatment is increased a 3.2%. For VIP, the Youden index was achieved at a concentration of 66 pg/mL (AUC 0.659; 95% bootstrap CI: 0.505-0.814). Contrary to CGRP, VIP threshold is sensitive (86.2%) but its specificity was very poor (50%). Consequently, this threshold would correctly classify 86% of CM patients as responders, which is not different from the global rate of response to onabotA in this series (78%). Among those CM patients with CGRP levels below 72 pg/mL, 28% had low VIP levels and just 33.3% responded as compared with 77.4% responders in the remaining 72% who had high VIP levels. Therefore, the probability of being a responder in CM patients with CGRP levels below the threshold was significantly higher in those patients with high VIP levels vs those with low VIP levels (OR: 6.857; 95% CI: 1.583-29.707; P = .012). Among CM patients with CGRP levels above the threshold, there was only one nonresponder who also had high VIP levels.
As already reported by our group using in part subjects included here, this study first confirms that interictal CGRP and VIP levels measured in peripheral blood are increased in a large series of CM patients vs healthy subjects with no headache history. In fact, both CGRP and VIP levels in CM were twice those of controls, which should be interpreted as distant signs of activation of the sensory and parasympathetic arms of the TVS, respectively. The levels of these two neuropeptides, and especially of CGRP due to its lower variability, measured in peripheral blood and outside migraine attacks have been proposed as the first biomarkers helpful for a more objective diagnosis of CM in the context of a patient with daily or almost daily headaches and a history of migraine, which could be of value for a better selection of treatment for CM patients.[9, 10]
The impact of CM in terms of quality of life and economic burden is very relevant.13-15 Treatment of CM is not easy. Even though in clinical practice we use oral preventatives with efficacy in EM, objective evidence of efficacy in CM is available only for topiramate16-18 and, to a lesser degree, for valproic acid. It was not until this decade that the efficacy of pericranial injections of 155-195 U of onabotA was shown in two large controlled phase III trials. This efficacy has also been reported in several open studies20-23 and in this series in which three quarters of our patients showed an objective and subjective response to onabotA injections. The exact mechanism of action of pericranial injections of onabotA leading to migraine prevention is still unclear, and reliable potential predictors of response have not yet been identified. In the pooled analysis of the 2 phase III trials with onabotA in CM, there was no positive correlation between 85 possible clinical predictors and response to onabotA. The main finding of the present work is that interictal CGRP, and to a lesser degree, VIP levels are potentially of great help on predicting response to onabotA. In fact, both CGRP and VIP levels were significantly higher in CM patients responding to onabotA as compared with nonresponders. Due to relatively low numbers of patients, there was no significant difference in patients exhibiting an excellent response as compared with those showing just a moderate response, but both CGRP and VIP were numerically higher in the group who reported an excellent response, again suggesting a relationship between the levels of these two neuropeptides and the rate of response to onabotA. Due to its intrinsic numeric dispersion, the specificity of VIP data is poor. By contrast, CGRP levels are both rather sensitive, very specific, and show a high potency to predict response to onabotA in CM. This is exemplified by two data: first, the optimal CGRP threshold given by the ROC analysis, 72 pg/mL, allows us a correct prediction of response to onabotA in 95% of cases; and second, a CGRP level above this threshold multiplies the probability of response by 28. Taken together, these results indicate that increased CGRP levels, very probably reflecting a continuous activation of the sensory arm of the TVS, are a good biomarker for CM diagnosis and specifically its response to treatment with onabotA injections.
There were, however, CM patients with CGRP levels in the range of controls, 31 patients with CGRP below the threshold who responded (8 of them showed an excellent response), and there was 1 patient without response to onabotA who had increased CGRP levels. How can these results be interpreted? They suggest that, together with CGRP, there are probably other factors in the pathophysiology of CM,[4, 24, 25] such as VIP, pituitary adenylate cyclase-activating polypeptide (PACAP), or peptide histidine methionine (PHM), which are stored and released by the parasympathetic arm of the TVS. There are several arguments strongly supporting an involvement of the parasympathetic arm of the TVS in migraine pathophysiology, at least in some patients. Cranial autonomic parasympathetic symptoms, such as lacrimation, rhinorrhea, and eyelid edema, do appear, depending on criteria and study design, in 27% to 73% of migraine patients.27-29 Meningeal blood vessels receive dense parasympathetic innervation.[3, 4, 26] Activation and sensitization of nociceptors around extra- and intracranial vessels is a primary source of pain in migraine. It has been proposed that parasympathetic outflow to cephalic vasculature may trigger activation and sensitization of perivascular sensory afferents and thereby contribute to migraine pain chronification.[7, 25, 30, 31] Our finding of increased peripheral VIP levels in CM patients outside of migraine attacks could reasonably be interpreted as a distant sign of “permanent” activation of the parasympathetic arm of the TVS, at least in up to three quarters of patients who express parasympathetic symptoms during migraine attacks.27-29 Supporting a role for VIP at least in some CM patients and their response to onabotA, 7 out of the 8 patients with excellent response to onabotA and CGRP levels below the threshold showed increased VIP levels. Intriguingly, there were 4 patients with both low CGRP and VIP levels who showed clear response to onabotA. Release of other pain producing peptides, such as PHM or PACAP, not measured here could be the first explanation for these results. In fact, PACAP itself is able to induce migraine-like attacks, and very recently, it has been shown that PACAP levels, both ictal and interictally, are elevated in migraine patients. There is also another interesting explanation, of relevance for clinical practice, for these results. In the absence of an objective diagnostic marker, CM diagnosis is based on a clinical picture alone. There could be a group of patients with a phenotype mimicking that of CM who are actually suffering from other headaches, either primary or secondary. Even after being assessed by an experienced headache neurologist and a magnetic resonance imaging has been performed with normal results, other diagnoses, such as tension-type headache in a previous migraineur or psychogenic headache expressing as CM, are still possibilities, which would explain in part the relevant response to placebo in trials with onabotA. This could be an interesting point to be tested in future placebo-controlled clinical trials in CM and is a further example of the necessity of introducing objective markers, such as CGRP levels, in CM research to try to avoid other diagnostic mimics.
We still do not have a complete understanding of the pathophysiology of CM or the real mechanism of action of onabotA in this entity. It is well established, however, that activation of the TVS has a crucial role and leads to afferent and efferent release of neuropeptides, especially CGRP. This facilitates a peripheral inflammatory response and vasodilatory response and causes activation of second-order neurons involved in pain transmission. In most vessels, the release of neuropeptides causes endothelium- and nitric oxide-independent vasodilation through a direct action on smooth muscle cells mediated both by cyclic adenosine monophosphate and by activation of adenosine triphosphate-dependent K + channels.[33, 34] Persistent release of CGRP and possibly other neuropeptides is thought to induce sensitization of central trigeminal neurons, and therefore migraine chronification, by triggering a signaling pathway mediated by brain-derived neurotrophic factor leading to increased expression of the P2X receptors. These peptidergic central neurons use L-glutamate as their primary neurotransmitter.[35, 36] CGRP, acting via a unique receptor complex, increases neurotransmitter release at these levels, which could lead to the central sensitization underlying chronic pain states such as CM.[7, 8] Our results, showing high CGRP and VIP levels in CM patients and a significant relationship between increased levels of these neuropeptides and response to onabotA, support, first, a crucial role of these neuropeptides in the pathophysiology of CM in humans, and second, that inhibition of local release of these neuropeptides is the likely mechanism of action of onabotA in CM, as previously had been hypothesized from experimental models.37-39 To fully demonstrate this possibility, it would be necessary to confirm a correlation between clinical response and a decrease in the levels of these neuropeptides, especially of CGRP, during onabotA treatment.
This study has several limitations. First, data shown here still include a limited number of CM patients and controls without headache. Second, our results in CM patients come from a selected clinical population. We do not know the true specificity of the increases in neuropeptides shown here, though at least for CGRP, it has been shown that levels in tension-type headache, cervicogenic headache, and in cluster headache (outside a cluster period) are below those seen in CM. One of our potential confounders could be the fact that, due to ethical reasons, CM patients treated here with onabotA were also taking oral preventatives. However, this could make our results even more relevant as these drugs should also reduce to some degree the activation of TVS and as a consequence decrease neuropeptide release. Finally, longitudinal studies with several determinations before and after onabotA are necessary to demonstrate the consistency of our data. With these limitations in mind, our data show that interictal CGRP, and to a lesser degree VIP, levels are reliable markers first for a CM diagnosis and second for predicting response to onabotA, which confirms a crucial role of these neuropeptides in the sensitization process required for migraine chronification and suggest that the mechanism of action of onabotA in CM includes a local inhibition of the release of CGRP and other pain producing neuropeptides.
This study was supported by the PI11/00889 FISSS grant (Fondos Feder, ISCIII, Ministry of Economy, Spain). P.M.C. is supported by grant MTM2011-23204 of the Spanish Ministry of Science and Innovation.