Objective and clinically-feasible analysis of diffusion MRI data can help predict dystonia after neonatal brain injury

Abstract Background Dystonia in cerebral palsy (CP) is debilitating but underdiagnosed. This precludes targeted treatment which is most effective if instituted early. Deep grey matter injury is associated with dystonic CP but can be challenging to quantify. Objective and clinically-feasible identification of injury that typically precedes dystonia could help determine which children are at highest risk for developing dystonia and thus facilitate early dystonia detection. Methods We examined brain MRIs in 4-5-day old neonates following therapeutic hypothermia for hypoxic-ischemic encephalopathy at a tertiary care center. Apparent diffusion coefficient (ADC) values in the striatum and thalamus were determined using a web-based viewer integrated with the electronic medical record (IBM iConnect Access). The notes of neonatal neurologists, pediatric movement disorders specialists, and pediatric cerebral palsy specialists (physicians who are likely the most familiar with motor phenotyping following neonatal brain injury) were screened for all subjects through 5 years of age for documentation of motor phenotype. Results In the 50 included subjects, striatal and thalamic ADC values significantly predicted dystonia with receiver operator characteristic areas under the curve of 0.862 (p=0.0004) and 0.838 (p=0.001), respectively. Striatal ADC values less than 1.014x10-3 mm2/s provided 100% specificity and 70% sensitivity for dystonia. Thalamic ADC values less than 0.973x10-3 mm2/s provided 100% specificity and 80% sensitivity for dystonia. Conclusions Lower striatal and thalamic ADC values predicted the development of dystonia in 4-5 day old neonates that underwent therapeutic hypothermia for HIE. Objective and clinically-feasible neonatal brain MRI assessment could help establish appropriate vigilance for dystonia in CP.