Background Despite a resurgence of research in functional neurological disorder (FND), knowledge of its pathophysiology is limited. To date, structural neuroimaging studies have typically been underpowered and have produced heterogenous findings. Mega-analyses combine individual raw participant data from multiple studies, substantially increasing statistical power and the ability to probe the complex pathophysiology of FND.

Methods In this FND Society Neuroimaging Committee initiative, international research groups with magnetic resonance imaging data of functional motor or seizure patients and healthy controls were invited to send T1-weighted images (pre-processed or unprocessed) and associated clinical variables. Key morphometric features such as cortical thickness, cortical surface area, and subcortical volumes were extracted using the Destrieux cortical atlas (Freesurfer 6.0.0) and harmonised to account for multi-centre variance with an empirical Bayesian approach (ComBat). Following ANOVAs corrected for multiple comparisons, cohorts were compared with MANCOVAs adjusted for age, sex, and total intracranial volume.

Results Fifteen groups contributed. The final dataset consisted of 564 age- and sex-matched controls and 493 patients, some of whom had both symptom subtypes (overall 314 motor symptoms, 193 seizures). Mean age 38.1 years for FND; 75.5% female. The FND cohort showed reduced cortical thickness in regions including the bilateral paracentral gyri and sulci, bilateral superior precentral sulci, and bilateral superior frontal sulci; varying fontal, occipital, and temporal regions showed reduced cortical surface area; volumes were reduced in regions including the bilateral putamen and hippocampi. In every case, the differences were small (η_p ² <0.016). Significant regions survived a manual leave-one-out analysis. The motor FND subtype contributed most to the differences. Binarised clinical variables were available for ~60% patients. Some of the reduced cortical thicknesses, particularly in the superior frontal and precentral regions, were associated with a history of depression. None of the differences were associated with history of anxiety, antidepressant use, or illness duration (years).

Discussion This international collaboration mega-analysis represents the largest structural imaging study in FND to date. Results suggest FND is associated with small reductions in cortical thickness, surface area, and volumes, particularly in frontal, motor, and subcortical regions. Future research should investigate whether findings are of aetiological significance or are secondary to the condition, and whether they are specific to FND or are also seen in related disorders. Further granular analysis, for example via receptor density mapping, could further elucidate the pathophysiology of this disabling disorder. Consensus regarding the minimum clinical variables to collect in future FND research is required.