Michael Eller; Peter J Goadsby


Expert Rev Neurother. 2013;13(3):263-273. 

In This Article

MRI in the Nonacute Headache Population

The following study is illustrative of problems in methodology in this area. In total, 530 patients underwent either MRI (n = 304) or CT (n = 226), and these were only a small proportion of the 3655 consecutive new patients referred to a tertiary headache center.[25] Demographics are only available for the total cohort: 69% were female, and the mean age was 42 years; 50% of the patients were diagnosed with migraine, 34% with TTH. A further 6.1% consisted of other primary headache diagnoses, while the remaining 10% of the patients represented a variety of secondary headache conditions, such as relating to trauma, drug effect or withdrawal or disorder of homeostasis. The choice to image was dependent upon which neurologist was seen, if any 'red' and 'yellow' flags were apparent,[9] and if any 'suspicious' findings were present. This variability in patient selection significantly compromises any interpretation of the study. Of the patients that underwent an MRI, 46% showed an insignificant abnormality – this was not defined, while 3% demonstrated a significant abnormality, mostly brain tumors. This contrasted with the CT findings, where 28% had an insignificant abnormality and 1% were found to have a significant lesion. Two patients in both the migraine and TTH groups had significant findings. The results are in keeping for those in the general community.

Another study demonstrated the use of 1.5 T MRI or CT brain scans in 1876 consecutive patients referred to the neurology service for headache.[26] The choice of modality was determined by a variety of factors, such as expediency and patient preference. Women represented around two-thirds of the study population, and the mean age was 38 years. Almost 50% had migraine, approximately 35% TTH, and nearly 11% an indeterminate phenotype. The remaining 4% included patients with less common phenotypes, such as cluster headache. CT scanning was conducted in 1432 patients, MRI in 580, and both in 136 patients. The rate of 'significant' lesions in those patients with a normal neurological examination was 0.9% (17 patients), and included three pituitary adenomas, an arteriovenous malformation, an ischemic stroke, two arachnoid cysts and a number of other tumors. Of these lesions, eight required surgical intervention. Of note, the yield of imaging in the group with indeterminate headache – those patients that were unable to be categorized by ICHD criteria – was 3.7%, higher than in TTH (0.8%) and migraine (0.4%). Despite the referral bias inherent in the study, and lack of predetermined imaging modality selection, this study represents a reasonable approximation to everyday practice.

A Japanese study retrospectively reviewed the 1-T MRI brain scans of 306 patients with episodic and chronic headache of more than 1 month, who also had a normal neurological exam.[27] Two-thirds of these scans utilized gadolinium enhancement. This group represents patients scanned for the indication of headache from an undefined and likely sizable pool. It is not clear if International Headache Society (IHS) criteria were used to classify patients, and which (if any) factors prompted the decision to image. They describe 'minor' abnormalities in 44%, without specifying the definition of this term. Only two patients had a 'significant' abnormality – one a subdural hemorrhage, another a pituitary adenoma. Their attached meta-analysis of MRI studies in migraine collected typically small studies using disparate patient-groups, using variable diagnostic criteria; in short, similar problems have plagued most imaging studies in the field. A 0.5% chance of finding a significant lesion among the 771 patients was found.

While an innocuous headache history coupled with a normal neurological examination does not preclude a significant intracranial lesion, intracranial pathology in common headache patients occurs at a rate roughly comparable with the general population. If it is poor practice to screen the general population, this corollary also applies to patients with a common headache phenotype, particularly in the absence of additional worrying symptoms and signs.