Persistent post-traumatic headache vs. migraine: an MRI study demonstrating differences in brain structure

The majority of individuals with post-traumatic headache (PTH) have headache characteristics that are consistent with a migraine phenotype [1‐4]. The only clinical feature that then differentiates PTH from migraine is the head injury itself. The International Classification of Headache Disorders (ICHD) 3 beta criteria stipulate that a PTH must begin within 7 days of the head injury or within 7 days of being able to detect headaches following a head injury [5]. If the PTH persists for at least 3 months, it is classified as “persistent PTH” (PPTH). The ICHD criteria do not include any headache characteristics that differentiate PTH from other headache types such as migraine or tension-type headache. For the patient without a history of migraine who has head trauma and develops headache immediately following the trauma, the diagnosis of PTH is rather straightforward. However, even in such a situation, it is not clear if the head trauma caused a unique headache type (i.e. PTH) with a unique underlying pathophysiology or if the trauma unmasked an underlying propensity toward the development of migraine. Identification of differences in the pathophysiology of migraine and PTH would support the notion that migraine and PTH are truly distinct headache types.

The objective of this study was to compare measures of brain regional volume, cortical thickness, surface area, and brain curvature in a cohort of patients with PPTH attributed to mild traumatic brain injury (mTBI) to a cohort of patients with migraine. Differences in brain structure would serve as evidence that PPTH and migraine might be associated with different underlying mechanisms and thus should indeed be considered distinct headache types.

Eighty-four subjects, consisting of 28 subjects with PPTH, 28 with migraine, and 28 healthy controls were included (Table 1). Average age of subjects was 35.9 +/− 9 years and 56% were female. Comparing the patients with PPTH to those with migraine, there were no differences in age (PPTH: 35.1 +/− 9.6 years vs. Migraine: 37.5 +/− 8.5 years, p = .33) or headache frequency (16.6 +/− 7.8 days per month vs. 16.4 +/− 8.1 days per month, p = .93). There were differences in sex (PPTH: females 32% vs. Migraine: 68%, p = .02), number of years with headache (8.5 +/− 7.8 years vs. 17.1 +/− 9.3 years, p = <.001), depression scores (18.6 +/− 9 vs. 6.1 +/− 5.3, p = <.001), state anxiety scores (39.3 +/− 14.4 vs. 32 +/− 7.9, p = .02) and trait anxiety scores (47.2 +/− 12.6 vs. 36.4 +/− 9.8, p = <.001). The healthy control group had an average age of 35.2 +/− 9.1 years, 68% were female, average depression score was 1.5 +/− 1.9, average state anxiety score was 24.9 +/− 6, and average trait anxiety score was 27.4 +/− 5.5. Compared to the PPTH cohort, the healthy control cohort had similar age (p = .97), fewer females (p = .02), and lower depression (p < .001), state anxiety (p < .001), and trait anxiety scores (p < .001). Compared to the migraine cohort, the healthy control cohort had similar age (p = .33), the same number of females, and lower depression (p < .001), state anxiety (p < .001), and trait anxiety scores (p < .001).

Amongst the 28 individuals with PPTH, 21 (75%) had a phenotype that was consistent with a diagnosis of migraine (had the symptoms not been initiated by TBI), 4 with probable migraine, 2 with tension-type headache, and 1 was unclassifiable. Twenty-seven reported that their headaches were of moderate or severe intensity, 25 reported sensitivity to sound, 24 sensitivity to light, 22 nausea, 21 throbbing quality of headache, 19 worsening of headache with physical activity, and 11 vomiting. Five subjects had one TBI in their lifetime, 13 had 2 TBIs, two had 3 TBIs, one had 4 TBIs, two had 5 TBIs, and five had 6 or more TBIs. Of the TBIs leading to PPTH, 14 were due to explosions/blasts, 7 were due to sports injuries, 4 were due to motor vehicle accidents, and 3 were due to falls.

Comparison of the PPTH cohort to the migraine cohort revealed differences in area of the right lateral orbitofrontal area (p = .008), curvature of the right lateral orbital frontal area (p = .019), volume of the right lateral orbital frontal area (p = .025), and thickness of the left caudal middle frontal region (p = .025), left precuneus (p = .041), right supramarginal gyrus (p = .041), and left superior frontal region (p = .048). (Table 2, Fig. 1) For each of these regions, thickness, area and/or volume measurements in those with PPTH were less than those within the migraine cohort. Lateral orbital frontal region curvature was greater in the PPTH cohort compared to the migraine cohort, suggesting underlying structural damage. The significance of each variable within the ANCOVA is reported in Table 3. Amongst those regions that differed in structure when comparing PPTH to migraine, there were no significant correlations between the numbers of lifetime TBIs or years with PTH with brain structure.

To better interpret the differences in brain structure identified between PPTH and migraine, the seven regional structural measures that differed between the PPTH and migraine cohorts were compared between the PPTH group to a group of healthy controls and between the migraine group and the healthy controls. Amongst these regions, there were differences between the PPTH group and healthy controls for thickness of the right supramarginal gyrus (p = .001), area of the right lateral orbitofrontal region (p = .024), and thickness of the left superior frontal region (p = .035). For all of these comparisons, PPTH was associated with less area and cortical thickness compared to the healthy controls. Comparisons between the migraine group and the healthy control group demonstrated no significant differences in the seven structural measurements that differed when comparing the PPTH group to the migraine group.

The phenotypic similarities between migraine and PPTH justify the need for studies that investigate mechanistic similarities and differences between migraine and PPTH. The main finding of this study is that there are differences in brain structure between patients who have PPTH and those with migraine, perhaps suggesting that these two headache types are associated with distinct underlying pathophysiology despite their substantial similarities in symptoms.

Brain areas that differed between PPTH and migraine were located in lateral orbitofrontal, superior and middle frontal, precuneus and supramarginal gyrus regions. The observation that only three of the seven brain regional measurements differed when comparing PPTH to healthy controls and none of the seven brain regional measurements differed when comparing migraine to healthy controls suggests specificity of some brain structural differences to the comparison of PPTH vs. migraine. We are not aware of prior studies that have compared brain structure or function in a group of individuals with PPTH to a group with migraine and thus are unable to compare our findings to others. However, there are a few studies that have compared brain structure and metabolism in individuals with PTH compared to healthy controls. For example, a longitudinal voxel-based morphometry study by Obermann and colleagues compared 32 patients with PTH attributed to whiplash injury to healthy controls [15]. During the first 14 days post-TBI, there were no differences in brain structure. However, at 3 months, PTH was associated with decreased gray matter density in the anterior cingulate cortex and dorsolateral prefrontal cortex. As headaches subsided by 1 year, the gray matter density normalized in these brain areas. Amongst those who developed PPTH, at 1 year there was in increase in gray matter in the midbrain, thalamus and cerebellum. Assuming that the patients who developed PPTH and those who did not had similar brain injuries, this study supports the notion that brain structural changes in patients with PPTH are related to the continued headaches and are not due solely to the inciting brain injury. A magnetic resonance spectroscopy study of 17 individuals with PTH attributed to mTBI (9 with acute PTH and 8 with PPTH) found that compared to healthy controls those with PTH had metabolic abnormalities in the anterior frontal lobes, anterior and posterior medial frontal lobes, and medial parietal lobes [16]. Although there was little power in this study to detect differences between patients with acute PTH compared to those with PPTH, those with PPTH had non-significant but lower N-acetylaspartate values (a marker for neuronal health). In our study, there were no significant correlations between the structural measurements of regions that differed when comparing PPTH to migraine with the number of years that individuals had PPTH. However, our patient population differed than those in prior publications in that our patients had a longer duration of PPTH at the time of imaging (mean of 8.5 years). Studies of mTBI have also demonstrated abnormalities of brain structure in several of the regions that we identified to have different structure in PPTH compared to migraine [17‐19]. Most of these previously published studies have not reported whether the participants had PTH and they did not attempt to disentangle the effects of the brain injury itself on brain structure vs. the effects of post-TBI symptoms on brain structure.

The brain regions that differed in structure when comparing individuals with PPTH to those with migraine in this study have all been previously demonstrated to participate in pain processing. Frontal regions play roles in the affective and cognitive evaluation of pain [20, 21]. Frontal regions have previously been found to have abnormal structure, function, and functional connectivity in individuals with different headache types including migraine, cluster headache, and medication overuse headache [22‐26]. The precuneus, a core region of the default mode network that is responsible for self-referential processing and interoception, participates in the determination of pain sensitivity and pain thresholds [27‐30]. Prior studies have implicated the precuneus and the default mode network in headache disorders including migraine, medication overuse headache, and cluster headache [22, 30‐33]. The supramarginal gyrus is likely involved in cognitive evaluation of pain including pain empathy [34, 35]. The supramarginal gyrus has previously been demonstrated to have atypical function and structure in groups of individuals with migraine and medication overuse headache [36, 37]. Although there are sufficient data to support a role for these brain regions in pain processing and headache, the explanation as to why the structure of these regions differs in individuals with PPTH compared to those with migraine is yet to be elucidated. It is plausible that these are brain regions that are simply more susceptible to the effects of mTBI and once damaged they contribute to the initiation and persistence of PTH.

The different measures used in this study (volume, area, cortical thickness, curvature) provide complementary information about brain structure. Regional volumes, surface area, cortical thickness and curvature are plastic measures of brain structure that commonly show alterations in the presence of aging, learning, and neurodegeneration related to disease. Brain curvature provides a measure of cortical folding, with increased curvature indicating areas of sharper cortical folds. Increased cortical curvature has been associated with white matter damage, aging, neurodegenderative disease, and mild TBI [38]. Although regional volumes have been most commonly used to measure brain structure and compare brain structure between subject cohorts, measurements of area and cortical thickness are likely to be more sensitive to small changes in brain structure. Regional brain area, cortical thickness, and volume have been previously identified to differ in cohorts of individuals with headache compared to healthy controls and to contribute to subclassification of headache types [39‐41].

Limitations: Similar to individuals with migraine and PPTH in the community and in clinical practice, our research participants had co-morbid medical conditions, and they were utilizing medications. For example, patients with PPTH had higher anxiety and depression scores than patients with migraine and healthy controls. Although we controlled for many potentially confounding variables in our ANCOVA, including depression and anxiety scores, it is not entirely possible to decouple the effects of these co-morbid illnesses on brain structure from the effects of migraine and PPTH on brain structure. However, study results would be of little value in understanding migraine and PPTH if we excluded patients with typical disease and only enrolled the very rare individual who has migraine or PPTH in isolation. Inclusion of typical patients with migraine and PPTH makes the results from this study generalizable to the typical individual with migraine or PPTH. Furthermore, the subject cohorts were not exactly matched for age and sex. Although we attempted to do so, the realities of our patient populations resulted in enrolling more men with PPTH (mostly armed forces veterans) and more women with chronic migraine. Sex was controlled for in our ANCOVA and was significant for only one of the brain measurements that differed between migraine and PPTH, but it still could have impacted our results to some extent. Perhaps the biggest limitation of this analysis is the inability to differentiate the effects of TBI from that of PPTH on brain structure. It is not clear if the differences in brain structure identified in this study are due to PPTH or if the differences would be similar in individuals who have persistent post-TBI symptoms without headache. We are currently trying to enroll those rare patients who do not have headache but do have other persistent symptoms following a mild TBI for comparison to the PPTH group.

In conclusion, although there was substantial overlap in symptoms between individuals with PPTH and those with migraine, the structure of several brain regions differed in individuals with PPTH compared to those with migraine. These structural differences suggest that the pathophysiology of PPTH might be different than that of migraine and support the classification of PPTH and migraine as distinct headache types. Additional studies are planned to confirm these imaging findings and to determine their specificity.