Gulf War Illness (GWI), also known as Chronic Multisymptom Illness (CMI), affects 25–30% of the 697,000 veterans deployed to the 1990–1991 Persian Gulf War (Fukuda et al., 1998; Gray et al., 2002; RAC-GWVI, 2008; Li et al., 2011). Veterans with GWI suffer from a wide array of symptoms that include headaches, cognitive dysfunction, chronic pain, fatigue, and other complaints (Fukuda et al., 1998; Gray et al., 2002; Li et al., 2011). Gulf War veterans often meet the criteria for chronic fatigue syndrome (CFS) (Fukuda et al., 1994) and fibromyalgia (FM) strongly suggesting overlap in symptoms for GWI and CFS (Wolfe et al., 1990; Peres et al., 2002; Latremoliere and Woolf, 2009; Baraniuk and Zheng, 2010; Wolfe et al., 2010; Ravindran et al., 2011; de Tommaso et al., 2011).

Headache is one of the eight ancillary criteria in the CFS case definition (Fukuda et al., 1994; Baraniuk and Zheng, 2010; Ravindran et al., 2011). Complaints of headaches are also reported in Gulf War veterans and over 50% of FM patients (Gray et al., 2002; de Tommaso et al., 2011; Li et al., 2011). However, the type and severity of headaches are not evaluated as a part of the GWI, CFS, or FM case designation criteria (Wolfe et al., 1990; Fukuda et al., 1994, 1998; RAC-GWVI, 2008; Wolfe et al., 2010; Ravindran et al., 2011). We recently determined that migraines are the primary type type of headaches in CFS (Ravindran et al., 2011). Studies in FM have reported similar migraine predominance (de Tommaso et al., 2011). The exact relationships between headache subtypes, fatigue, pain, hyperalgesia, and other systemic complaints has been difficult to establish because of heterogeneous symptom presentation and/or lack of biomarkers that identify a distinct pathophysiological process in GWI.

One mechanism that has been proposed to drive the complaints of GWI and related illnesses is central sensitization (Latremoliere and Woolf, 2009; Baraniuk and Zheng, 2010). This process is defined as an amplification of responsiveness of central pain-signaling neurons that clinically manifest as hyperalgesia and/or allodynia (Latremoliere and Woolf, 2009). This normal adaptive neuronal process is essential for learning, but has been proposed to become dysregulated in each of these disorders and in even in migraineurs (Gebhart, 2004; DaSilva et al., 2007; Tietjen et al., 2009).

Neuroimaging studies report underlying dysfunction in brain energetics, the brainstem, and thalamocortical tracts in migraineurs (DaSilva et al., 2007). Alterations in these regions may lead to the loss of descending anti-nociceptive processes coupled with changes in cerebrovascular dynamics. Specifically, blood flow alterations in migraine headache are associated with dysfunction in the trigeminovascular system, which innervate cerebral blood vessels in the dura (Karatas et al., 2013). As a result of trigeminal activation, migraineurs demonstrate hyperalgesia, allodynia, and cognitive dysfunction during and between episodes (Gebhart, 2004; DaSilva et al., 2007; Tietjen et al., 2009; Baraniuk and Zheng, 2010; Karatas et al., 2013). These findings may lend credence to the theory of central sensitization.

Similar patterns of gray and white matter abnormalities and altered brain energetics in GWI, CFS, FM, and migraine suggest that common central mechanisms may contribute to the type of headaches and the cognitive impairments perceived as “brain fog” (Lutz et al., 2008; Barnden et al., 2011; Puri et al., 2012; Rayhan et al., 2013a,b). Understanding the pathophysiological mechanisms underlying migraine may identify the primary dysfunction that leads to the constellation of complaints in GWI, CFS, and other overlapping disorders (Figure 1).

Although GWI patients complain of headache symptoms that overlap with CFS, FM, and migraineurs, there is currently no study that specifically assesses this in GWI. We hypothesized that due to extensive symptom overlap, GWI subjects will have similar rates of headache complaints subtypes. As a first step toward testing this hypothesis, we focused on identifying clinical relationships between headache status, hyperalgesia (tenderness), pain, and fatigue ratings in GWI. In addition, we compared such complaints to CFS subjects and controls to further identify similarities and/or differences between groups. The cross-sectional design permitted estimation of the prevalences of headache, pain and fatigue in GWI. Comparisons of GWI to CFS subjects allowed validation and extension of previous findings. Migraine status, GWI, CFS, FM co-morbidity and other conditions were diagnosed based on history and physical examination rather than mail-in and/or online questionnaires. Detailed history and physical examination has proved to be highly reliable in previous studies (Ravindran et al., 2011). To supplement the research study, we review the current literature into migraines and potential mechanisms.

All of the GWI veterans recruited met CMI and CFS criteria (Fukuda et al., 1994, 1998). Overlap in case designation criteria is consistent with previous findings (Ravindran et al., 2011; Rayhan et al., 2013a,b). Scores for self-reported questionnaires, systemic and sinus pain thresholds, and migraine for GWI and CFS subjects were comparable. The high prevalence of migraine in CFS was verified and extended to include GWI (Peres et al., 2002; Ravindran et al., 2011). GWI had significantly lower SF-36 scores than controls indicating severe impairment of quality of life (Buchwald et al., 1996). This indicates a large unmet need to diagnose and treat migraine in GWI and CFS, and to expand the options for therapy of these conditions.

GWI subjects tended to be male due to the higher ratio of men in the military. In contrast, CFS and FM is predominantly diagnosed in the civilian female population (Reiffenberger and Amundson, 1996; Okifuji et al., 1999; Eisen et al., 2005; Stephen, 2005; RAC-GWVI, 2008; Li et al., 2011). The high prevalences of migraine in GWI and CFS were independent of gender. This finding may be the result of exclusion of menstrual and chronic childhood onset migraine from our study population, but is also consistent with the lack of association of migraine with estrogen, progesterone and other sex hormone receptor polymorphisms (Schürks et al., 2010).

The causes of GWI and CFS have been hotly contested with little consensus. However, recent work suggests that toxic exposures (sarin gas, pesticides, particulates from oil fires, etc.) and/or acetylcholinesterase inhibitor use in the Persian Gulf may be temporally associated with GWI onset (RAC-GWVI, 2008; Steele et al., 2012). Diverse infectious and stressor triggers have been proposed, but not proven, to underlie CFS and FM in the civilian population (Wolfe et al., 1990; Fukuda et al., 1994; Reiffenberger and Amundson, 1996; Peres et al., 2002; Ravindran et al., 2011).

The seemingly confusing symptomatology of these disorders requires a fresh, systems biology conceptual approach that integrates systems involved in assessing afferent inputs, spinal cord and brain processing, and the conscious perceptions that are conveyed as subjective complaints. Sensory afferent information from mucosal organs such as the nose (neurological non-allergic rhinitis of CFS (Baraniuk et al., 2005), esophagus (“nutcracker esophagus”), bowel (irritable bowel syndrome) (Lea et al., 2004), bladder (irritable bladder syndrome and interstitial cystitis) is poorly localized in the insular somatosensory cortex. Interoceptive visceral sensations may enter the consciousness of some subjects and be perceived as serious illnesses (Critchley and Harrison, 2013). Chronic irritation, inflammation, or centrally released mediators may lead to excessive glutamate release that activates neuronal AMPA receptors to promote the transmission of nociceptive messages (Rowbottom et al., 1998; Goadsby et al., 2002; Fukui et al., 2008) The resulting perceptual state may be an example of neural plasticity gone awry.

Dysregulation of neuropathic sensory relays to the dorsal pons may lead to excessive pain and tenderness (hyperalgesia, allodynia), fearful memories as in posttraumatic stress disorder, impaired executive decision-making (“brain fog”), and characteristic features of GWI and CFS. The synaptic molecular changes lead to neural hypersensitivity and hyperresponsiveness to extrinsic and central, intrinsic activation that is termed “central sensitization” (Latremoliere and Woolf, 2009). A range of gene polymorphisms such as ion channelopathies may modify the risk for illnesses development in the population. For example, certain variants of Panx1 ion channels may be dysfunctional and more vulnerable to stressors that promote widespread neurotransmitter release (Rowbottom et al., 1998; Goadsby et al., 2002; Fukui et al., 2008). In a rat model of migraine, activation of Panx1 channels on peripheral trigeminal neurons caused cerebrovascular dysregulation with changes suggestive of aura followed by headache (Karatas et al., 2013). Evidence supporting central sensitization has been found in each of these chronic nociceptive, interoceptive, and fatiguing illnesses (Malick and Burstein, 2000; Maizels and Burchette, 2004; Meeus and Nijs, 2007).

Acute episodes of migraine and its consequences may provide a model for central sensitization (Figure 1). The triggering event may be an acute, extreme, and prolonged depolarization called cortical spreading depression (CSD) (Pietrobon, 2005; Smith et al., 2006; Eikermann-Haerter and Ayata, 2010). The molecular and electrical events initiating CSD involve glial and neuron cell body depolarization with an extreme efflux of K+ out of cells, and influx of Na+ and Ca2+ (Smith et al., 2006; Ayata, 2010; Charles, 2010).

Returning the intracellular K+ concentration to the high level required to repolarize these cells can take several minutes instead of milliseconds. The prolonged depolarization leads to oligemia that may cause transient hypoxia, depletion of oxygenated neuroglobin reserves, (Casado et al., 2005) and anaerobic glucose metabolism with lactic acid production. Release of autoregulatory vasodilators leads to profound reactive cortical vessel dilation and hyperemia. In animal models, the CSD spreads outwards in concentric fashion at the rate of 3–5 mm per minute as a wave of intense depolarization with oligemia followed by autoregulatory hyperemia. The change in blood flow can be detected by arterial spin labeling using 3 Tesla fMRI (T2 weighted perfusion changes) (Swartz and Kern, 2004). Peri-infarct depolarizations may be a similar phenomena in strokes, brain trauma, and other head injuries (Strong and Dardis, 2005; Fabricius et al., 2006; Strong et al., 2007). CSD may be analogous to the chaotic, stochastic cellular depolarization of atrial fibrillation, premature ventricular contractions (PVC's), esophageal spasm, or epilepsy.

Chronic CSD-like depolarization in migraine, GWI and CFS may promote central sensitization and progressive dysfunction of the anterior cingulate gyrus (ACC) and other neuroanatomical loci (Jensen et al., 2009; Obermann et al., 2009), “Neural plasticity” (Pietrobon, 2005) may reinforce conditioned memories and contribute to affective dysfunction, anxiety, fear and posttraumatic stress disorder (PTSD), (Woodward et al., 2006) fatigue, pain, hyperalgesia, allodynia and cognitive dysfunction (“brain fog”) (Meeus and Nijs, 2007). Neurovascular dysfunction may cause gray matter thinning and white matter abnormalities (prevalence = 16–40%; OR = 3.9, 95% CI = 2.26–6.72) that accentuate these disabilities (Burgmer et al., 2009; May, 2009; Schwedt and Dodick, 2009).

Alternative central sensitization hypotheses of migraine such as the “Unitary Hypothesis” suggest a primary role for cortical and limbic structures (Burstein and Jakubowski, 2005). Burstein and Jakubowski proposed that the well-known triggers of migraine stimulated multiple hypothalamic, limbic, and cortical nuclei that project axons to the superior salivatory nucleus (SSN). The SSN is a preganglionic parasympathetic nucleus that activates the postganglionic sphenopalatine ganglion. The release of acetylcholine, vasoactive intestinal peptide (VIP), and nitric oxide (NO) may initiate vasodilation of meningeal vessels. Wall stretching would then stimulate mechanoceptors on trigeminal afferent neurons that convey the signal of migraine neurovascular pain.

Another alternative is the central periaqueductal gray matter (PAG) hypothesis. The PAG has been considered a migraine generator. Positron emission tomography scans performed during spontaneous migraine attacks found activation of the PAG, dorsal pons near the locus ceruleus, and ACC (Weill et al., 1995). Activation of the ACC is not fully explained, but could indicate interference with executive decision-making and other cognitive functions (Rayhan et al., 2013a). Sumitriptan relieved the migraine symptoms, but the PAG remained activated. This suggested that the PAG may act as a “migraine generator” with the latent potential to trigger additional attacks. Other lines of evidence implicating the PAG in migraine include the correlation between the duration of illness with iron deposition (a marker of dysfunction) near the PAG, caudate and putamen (Aurora, 2009; Kruit et al., 2009). Dihydroergotamine binds to the PAG and dorsal raphe nucleus, (Goadsby and Gundlach, 1991) while naratriptan injection into the PAG has an antinociceptive effect (Bartsch et al., 2004).

Dysregulated brain stem release of NO and calcitonin gene related peptide (CGRP) may represent a molecular mechanism to initiate migraine and facilitate central sensitization show in Figure 1. These mediators may act synergistically (Panconesi et al., 2009) to initiate: (i) inhibition of anti-nociceptive functions; (ii) phantom, self-generated sensory phenomena that may include pain, visual (e.g., photosensitivity (Recober et al., 2010) auditory, and vestibular aberrations; (iii) retrograde activation of trigeminal ganglion afferents leading to CGRP release from branched nerve endings around dural and pial vessels (reverse of 3–5 on Figure 1); (iv) unregulated nociceptive responses from vascular stretch receptors (5 on Figure 1); and (v) facilitate the other ascending nociceptive and cognitive dysfunction (7–11 on Figure 1).

The “neurolimbic hypothesis” extends the role of the PAG as a “migraine generator” to include activation of a proposed neurolimbic pain network (Maizels et al., 2012). In this heuristic, migraine represents that end result of dysfunctional brain network activities, rather than the consequence of defects in individual ion channels, other proteins, or anatomical nuclei. Maizels et al. propose that brainstem pain-modulating circuits have bidirectional connections with the limbic system (amygdala, insula, ACC, orbito-frontal cortex, prefrontal cortex, hypothalamus) that interact tonically to modulate migraine expression 974).

This hypothesis is based on the existence of a major network of trigeminovascular - sensitive neurons that innervate the thalamus, with thalamic relays to widespread regions of the cortex (Noseda et al., 2011; Karatas et al., 2013). This pathway originates in the trigeminal nerve endings in pial and dural vessel walls. These primary trigeminal neurons innervate the cell bodies of second-order trigeminovascular neurons that are located in laminae I and V of the dorsal horn of the caudal medulla and upper cervical C1 and C2 spinal segments. The secondary ascending neurons have been traced electrophysiologically to the midbrain, where they bifurcate into an axon projecting to thalamic posterior (Po), ventral posteromedial (VPM), and parafascicular nuclei, and a second axonal limb that arborizes and terminates in anterior, lateral, perifornical and posterior hypothalamic nuclei, and lateral preoptic areas. The third-order trigeminovascular—responsive thalamic neurons convey signals to wide variety of cerebral regions. VPM and Po neurons have axons that bifurcate and issue dense terminal arborations in the trigeminal barrel-field region of the primary somatosensory cortex (S1BF) and secondary somatosensory cortex (S2). Additional Po neurons project to extra-trigeminal regions of S1, the primary auditory (AuD/Au1), retrosplenial (RSA), parietal association (PtA), primary and secondary visual (V1/V2) cortices, and primary and secondary motor cortices (M1/M2). Neurons from the laterodorsal (LD) and lateroposterior (LP) thalamic nucleus project to M1 and M2 regions, trigeminal and extratrigeminal S1 regions, secondary visual cortex mediomedial (V2MM), and the ectorhinal cortex (Aurora, 2009; Burstein et al., 2010; Noseda et al., 2011). This pathway connects the primary meningeal nociceptive signals to the functionally distinct and anatomically remote cortical areas that are involved in migraine—related dysregulation of affect, motor function, visual and auditory perception, spatial orientation, memory retrieval, olfaction, and allodynia (Burstein et al., 2010).

Resting state fMRI functional connectivity studies of interictal migraine subjects provide additional support for abnormal brainstem—limbic system interactions (Mainero et al., 2011). Functional connectivity is the property of brain regions to harmoniously signal to each other, and indicates their capability to act in cooperative fashion. Migraineurs had increased functional connectivity between the PAG and thalamus, posterior parietal cortex, anterior insula, somatosensory cortex that are involved in somatosensory, interoceptive, and nociceptive processing. Those subjects with a high frequency of migraines had increased connectivity between the PAG, anterior insula, nucleus cuneiformis, and hypothalamus, but reduced connectivity of PAG with the prefrontal cortex, ACC, amygdala, and medial thalamus. Decreased connectivity between the PAG and anterior and posterior cingulate cortex, hippocampus, putamen, and posterior insula has been confirmed in a separate group who had frequent migraines (Maleki et al., 2011a). Women had more significantly reduced functional connectivity of the PAG to the posterior cingulate cortex and amygdala than men (Maleki et al., 2011b). Functional connectivity alterations were more pronounced in adults than children (Lebel et al., 2011). When allodynia was present, migraineurs had decreased connectivity between the PAG, prefrontal cortex, ACC, and anterior insula. Therefore, magnetic imaging studies in GWI and CFS should take into account potential changes in baseline cerebrovascular hemodynamic responses (Rayhan et al., 2013b), gray matter interconnectivity, and white matter integrity (Rayhan et al., 2013a). Disruption of these intricate neural pathways may represent a new paradigm for understanding complex neurological diseases.

These objective findings have important clinical implications for the subjective presentation of migraine, GWI and CFS subjects. A common presentation is:

“… a middle-aged woman with chronic migraine and medication overuse, as well as fibromyalgia. In addition, there is anxiety and depression, fatigue and insomnia, and the familiar exhaustive list of psychotropics and antiepileptic drugs tried and failed” (Maizels and Burchette, 2004).

This type of patient may seem exceedingly frustrating, since many physicians find the panoply of symptoms daunting and their armamentarium bare. The physician's quandary can be deepened as they investigate the history further. Regardless of whether they consider themselves “lumpers” or “splitters,” the presence of one functional syndrome increases the probability of others (Bradley, 2008). The situation is all the more exasperating since current medical dogma does not provide a satisfactory explanation for the underlying pathophysiology. Maizels et al. (2012) provide a logical approach based on their neurolimbic model, knowledge of the dysfunctional of serotonergic neural pathways, and central nature of the nociceptive, interoceptive, somatosensory, and fatigue complaints (Table 5).

This cross-sectional investigation had several limitations. Although presence and absence of aura was identified, during subgroup analysis of migraines subjects were not grouped based upon the presence or absence of aura due to the small sample size. It was not designed to correlate migraine phenotypes in GWI with triggers such as fasting, loss of sleep, odors, hormonal changes, and stress (Fukui et al., 2008). Longitudinal changes in migraine severity, systemic hyperalgesia, fatigue, and functional status await larger epidemiological and treatment studies. Analysis of magnetic imaging correlates are pending.

GWI and CFS subjects have a high prevalence of migraine headaches, comparable patterns of systemic and cognitive complaints, and share overlapping phenomenological case designation criteria. CFS and GWI had lower systemic pain thresholds indicating systemic hyperalgesia and central sensitization. The loss of descending inhibitory antinociceptive aminergic signals may contribute to spinal sensitization, but cannot explain the many other cerebral, brainstem, and autonomic abnormalities. It is tempting to speculate that the parallel findings of GWI, CFS, and migraine indicate a shared underlying pathophysiological mechanism of central nervous system neural pathways that may account for chronic pain, fatigue, and cognitive dysfunction (“brain fog”). Interdisciplinary studies that address these diverse components of illness are required to identify relevant dysfunctional pathways and molecular mechanisms that may lead to reductionist approaches for evaluation of future drug treatments.

James N. Baraniuk and Murugan K. Ravindran designed the protocol. Rakib U. Rayhan, Murugan K. Ravindran, and James N. Baraniuk administered and conducted the study. Rakib U. Rayhan and James N. Baraniuk interpreted and completed the statistical analysis. Rakib U. Rayhan and James N. Baraniuk wrote, edited, and gave final approval for the manuscript.