Brief summary of study findings and comparison with available literature
Our findings add to pre-existing evidence for the potential therapeutic value and safety of the preventive and acute use of nVNS as an adjunctive to prophylactic and abortive drugs in EM patients. We observed comparable responsiveness for head pain severity and frequency [
9‐
19]. A remarkable reduction in severe attacks was observed in our cohort, similiar responsiveness was observed for pharmacological interventions. Ferrari et al. performed a meta-analysis and reported up to 59% response rates after 2 h (improvement moderate/severe to mild/no pain), 30% pain free state after 2 h and 20% sustained pain free state (no headache recurrence or use of rescue medication 2–24 h after baseline) for acute pharmacological migraine interventions [
8]. Most recently, Martelletti et al. assessed additional secondary outcome parameter of the PRESTO study and found a higher rate of pain-freedom and pain-relief attacks as well as a higher rate of severity reduction in migraine patients compared to sham stimulation [
17]. This meaningful improvement was accompanied by a considerable rescue medication decrease [
18]. However, the antinociceptive head pain potential of the vagus nerve was confirmed in a further t-VNS study targeting the auricular branch at low frequencies (1 Hz) with 30% responder rate (responder defined as ≥ 50% reduction in headache days) [
19]. Additionally, our feasibility study demonstrated an improved sleep quality with adjunctive nVNS, which is in line with previously published data [
10,
11].
No association was found between those with excellent response compared to less favorable outcome and specific saliva OXY and IL-1β measures. Of note, our study enrolled EM and CM participants, as it has been well documented, that both subtypes encompass different characteristics and beyond doubt have been linked to different pathomechanisms [
5‐
7]. Overweight (BMI 25–30 kg/m
2) was present in 4 out of 12 patients, thus it cannot be excluded, that this fact may have a considerable impact on the increased inflammatory levels. In contrast to earlier hypothesis restricting the function of white adipose tissue (WAT) as a metabolic storage organ, current revised concepts consider WAT as an inflammatory endocrine active organ with the capability to promote or suppress peripheral and central inflammation via crosstalks between adipocytes (e.g. synthesis of leptin/adipokines) and the innate and adaptive immune system. Obesity as a low-grade chronic inflammation has been associated with tissue hypoxia/necrosis with consecutively upregulation of the pro-inflammatory response via cellular (M1/2 macrophage—Th1/Th2 cells phenotype transformation) and molecular (IL-1β, IL-6, TNF-α) pathways [
51]. Hence, future inflammatory migraine research should consider targeting peptides of the adipokine superfamily [
51].
Inter-ictal oxytocin and saliva levels were significantly higher in migraine patients compared to healthy controls at baseline and subtly increased after nVNS. So far, most of the reported studies determined cytokine, not oxytocin, in serum and compared inter-ictal versus ictal cytokine signaling. Significantly increased ictal IL-1β, IL-6 and TNF-α serum concentrations were measured in migraine patients with/without aura compared to post-ictal (after 1 week treatment) and healthy subjects not clearly indicative for a predictive value. Similar results were published addressed to migraine and peri-ictal cytokine analysis [
38‐
40]. In our trial, oxytocin and IL-1β screening was performed post-ictally and differed in the choice of the investigated biofluid (saliva). Indeed, it would have been of interest to collect ictal values and additional markers relevant for migraine such as CGRP, but was not performed according to our study protocol. As expected we found higher post-ictal concentrations in migraine patients compared to healthy controls. In line with the findings of Perini et al., it is likely, that ictally assessed saliva concentrations would have displayed higher values compared to our post-ictal results [
39]. Nevertheless, elevated saliva levels of inflammatory markers may serve as head pain susceptibility screening tool in migraine patients. After nVNS therapy, both mediators further increased along with an improved head pain state by preventive and abortive means. On the one hand, the increased oxytocin levels may be driven by the observed marked pain relief in our cohort. In several experimental studies, oxytocin has been demonstrated to be released in response to activated sensory neurons of the body (pinch, touch). In particular, electrical stimulation of somatic sensory neurons and afferent fibers of the vagal nerves was shown to effectively increase oxytocin plasma levels immediately after stimulation, which may support the observed increased oxytocin levels in our VNS treated study population [
52]. In addition, it is important to note, that 50% of our cohort suffered from migraine-associated aura. Cortical spreading depression as the electrophysiological correlate of aura has been suspected to evoke astrocytes (microglia) induced synthetization and immune response via IL-1β among other cytokines [
53]. Other human studies found a correlation between oxytocin concentrations and head pain intensity in migraine, contrary, we observed a trend towards association of migraine frequency (headache days/month) and oxytocin saliva levels [
53]. On the other hand and to our surprise, concentrations of pro-inflammatory IL-1β was higher after nVNS compared to baseline. Possible explanations may be the fact that although clinically improved none of the nVNS treated subjects could be classified as head pain free suggesting an ongoing inflammatory process. Importantly, preliminary data indicate that intra-nasally administered oxytocin (32 IU) reduces pain in two patients with chronic migraine headache. This effect was reduced in patients who had taken nonsteroidal anti-inflammatory drugs suggesting that the anti-nociceptive effect of oxytocin is cytokine-dependent [
33].
Only one human study assessed possible effects of cervical nVNS on the peripheral components of the neuro-immune reflex in healthy humans. Lerman and colleagues measured healthy individuals randomized to verum and sham cervical nVNS treatment. Chemokine levels assessed at baseline, and at 90 min and at 24 h after treatment showed a decrease in pro-inflammatory IL-1β, IL-8, and TNF-α levels and an increase in anti-inflammatory IL-10 levels, indicating that nVNS may inhibit pro-inflammatory cytokine release [
41]. Other human studies conceptualized to determine peripheral inflammatory profiles of subjects treated with surgically implanted invasive VNS (iVNS) were limited to depression and focal seizure with limited interpretations due to the uncontrolled study design [
36,
37]. However, comparable cytokine/chemokine data under VNS “off”-stimulation remain an open question.
The impact of oxytocin on trigemino-nociceptive signaling
Recently, a preclinical study determined oxytocin receptor expression and co-localization with calcitonin gene-related peptide (CGRP) in the trigeminal ganglion. Application of painful, facial electrocutaneous stimulation and adjunctive capsaicin-driven inflammation increased oxytocin expression in CGRP-containing trigeminal ganglion neurons, indicating the important role of oxytocin in migraine pathophysiology [
34].
The relationship between migraine and oxytocin was under investigation in a sophisticated experimental setting including measurement of electrophysiological (TNC firing response) and gene expression (C-fos) parameters after intranasal oxytocin administration compared to placebo [
35]. In the first, increased TNC firing rates after electro-cutaneous stimulation of the face were recorded and of note, attenuated by oxytocin, while in the second an increased C-fos expression in TNC neurons was observed after intra-peritoneal injection of nitroglycerin, which was revised in the oxytocin pre-treated group. In a next translational step, intranasal oxytocin was assessed in sham-controlled trials in EM and CM patients [
35]. Although no significant differences were observed after 2 h treatment for both migraine subtypes, the CM subgroup demonstrated a trend in favor of the verum treatment. Additional human trials indicate a stronger effect of oxytocin on frequency, rather than on severity. Interestingly, NSAID use was suspected to interfere with oxytocin effects by inhibiting cytokine synthesis [
35]. NSAID as an adjunctive rescue medication was present in 5 out of 12 patients in our study.
Migraine as a multi-network brain disorder displays altered sensory (nociception), cognitive, affective and circadian-dependent autonomic features (sleep, metabolism, thermoregulation), of which each component is able to drive head pain onset/attacks [
54‐
56]. Apart from its ictal functions, inter-ictal endogenous oxytocin has been linked to central sensitization (hyperalgesia, allodynia) and neurogenic inflammation in migraine pathophysiology. Such elevated inter-ictal oxytocin concentrations may reflect modulation of extracephalic pain perception and affective distress symptoms. Indeed, it would have been of great interest to quantify these inter-ical appearing clinical features relevant for migraine [
56]. Taken the complex and dynamic nature of migraine into account, the authors speculate, that molecular profiling of oxytocin may have a diagnostic and therapeutic potential for migraine-associated symptoms outside the ictal phase [
54,
55].
Pro-inflammatory IL-1 β associated effects in trigemino-nociceptive traffic
Zhang et al. recorded TG neuron response of meningeal nociceptors after local application of IL-1β and IL-6 on dural nociceptors and found that IL-1β, but not IL-6 was able to promote increased activity of TG neurons accompanied with increased mechano-sensitivity of intracranial nociceptors (meningeal afferent signaling) measured von Frey filaments [
42].
IL-1β has been suspected to induce upregulation of cyclo-oxygenase 2 mRNA (COX-2) expression in glial cells and neurons of the trigeminal ganglion (TG). These COX-2 dependent pathways lead to prostaglandine release from glial and neuronal TG cells, which in turn stimulates solely neurons of the TG to immediately (1 h after stimulation) produce CGRP, contrary to IL-1β, which demonstrated a delayed CGRP release pattern (24 h after stimulation) suggesting a glia-neuron interaction in the TG [
57]. Methylprednisolone reversed the IL-1β effects, but demonstrated no impact on prostaglandine induced CGRP release [
58]. Leptin, a metabolic marker produced by WAT cells, has been shown to interact with the COX-2 dependent pathways via crosstalks with IL-1β in glial cells and neurons of the hypothalamic-pituitary axis [
59].
The development of acute head pain has been associated with primary afferents activation of the TG driven by dural nociceptors, which are connected with the trigeminal nucleus caudalis (TNC). The TNC itself projects to the trigemino-cervical complex (TCC), and receives reciprocal input from the brainstem, the medulla oblongata, the hypothalamus-pituitary-axis, the thalamic nuclei (intralaminar nucleus of the thalamus) and cortical associated networks.
On the other hand, it has been well described, that the afferent properties of the vagus nerve project via the ncl. tractus solitarii to the locus coeruleus (LC), the dorsal raphe nucleus, the parabrachial plexus, the paraventricular nucleus of the hypothalamus and maybe directly to the TNC and the cervical spinal cord (trigemino-cervical complex, TCC). In view of the anatomic reciprocal connectivity of the vagus nerve, it may be reasonable, that cervical nVNS may impact trigeminovascular nociceptive signaling of pro- and anti-inflammatory markers in different biofluids (plasma, saliva, cerebrospinal fluid) [
20,
21,
26‐
31,
34,
35,
52,
57‐
59].
Inhaled (olfactory nerve) and/or ingestive (trigeminal nerve) chemical irritants have been suspected to promote pro-inflammatory mediators and to trigger neurogenic inflammation in a broad range of respiratory (asthma) and neurological disorders such as migraine. A dynamic and complex interplay between local (e.g. substance P, bradykinin) and distant efferent effects (adrenergic, cholinergic) characterizes in part the host response, in which mast cell derived immunomodulation plays a pivotal role. Hence, it cannot be excluded, that environmental factors may have an impact on inflammatory phenotyping and quantification of peripheral markers of the neuro-immune axis [
60,
61].
In particular, immunomodulatory mast cells (high affinity receptor FceR1) are capable to interact with the innate and adaptive immune response and poses an important role in the genesis of acute/chronic inflammation associated disorders (e.g. migraine). Immunglobuline E (Ig E), Toll-like receptors, IL-1β and IL-36 are known to activate mast cells and to drive a pro-inflammatory state, while IL-37 and IL-38 (member of the IL-1 cytokine family) act as an inhibitor of inflammation. Gallenga and colleagues reported that IL-38 is able to bind on the IL-36 receptor, which in turn blocks mast cell activation [
62]. It is important to note, that mast cell response occur in a time dependent manner with an immediate secretion and a delayed synthesis/release of inflammatory active peptides in order to establish a physiological host response. Among the mentioned IL-1 cytokine family, IL-1β increases IL 33 and TNF-α synthesis derived from mast cells. Furthermore, IL-33 interacts with monocytes and promotes mast cell differentiation, maturation and degranulation with subsequent secretion of pro-inflammatory cytokines/chemokines. Contrary, IL-37, an anti-inflammatory member of the IL-1 cytokine family, has the potential to counterbalance the IL-1β evoked innate and adaptive immune response. Therefore, the therapeutic anti-inflammatory value of IL-37 induced blocking of mast cells deserves further clarification in migraine research [
63,
64].
Limitations and future prospects for molecular inflammatory profiling in migraine
This study has several limitations including the uncontrolled design, the small-scale study cohort and a relatively short-term observation period. The main issue is related to the lack of a sham stimulation group (stimulation off) in order to discern what was due to patient’s expectation (treatment self-responsibility) in contrast to the real effects of nVNS. With respect to migraine as a complex brain disorder, such expectation associated with a novel device (like nVNS in our study) may represent a confounder. Different dosages of migraine drugs (preventive/abortive) administered in each patient may represent a confounder. Furthermore, cytokine levels may vary and depend on pre-analytic variables including sample processing, environmental factors, intra- and interindividual variability. The cytokine analysis performed in our study did not consider the dynamic nature of neuro-inflammation nor the circadian neurobiology, thus repetitive measurements are recommended in future trials. The role of the brain-immune-communication in primary headache disorders is of major interests. To this end, immune-phenotyping of migraine is in the beginning. Future research in this field should seek to investigate which immune pathways are overactivated in migraine, how immune cell hyperactivity is linked to disease susceptibility, and how environmental and genetic factors influence immune activation and disease manifestation. Immune-phenotyping should consider cell profiling in several flow cytometry panels and whole blood stimulation assays with a range of innate immune stimuli. Can we use molecular profiling to predict and individualize neurostimulation (nVNS) therapy? Can we target cytokines as diagnostic tools? Currently the answer is no as the precise mechanisms of the neuro-immune communication in migraine pathophysiology remains unclarified. Alternatively, advanced statistical methods capable to establish categorical-based dimensions may support the potential and the integration of biobank-based immune-phenotyping, thus help to define migraine specific characteristics and subsets (biotypes) of patients more or less likely to respond to neurostimulation therapy [
65,
66]. However, this study firstly approached to screen oxytocin and IL-1β in saliva and attempted to proof the feasibility of saliva analysis and undoubtedly, but was biased by the uncontrolled study design.