Abstract
BACKGROUND: The increasing clinical use of combining structural MRI (sMRI) with General Movements Assessment (GMA) or Hammersmith Infant Neurological Exam (HINE) before five months corrected age (CA) for early diagnosis of cerebral palsy (CP) lacks sufficient prognostic data for children with CP, especially those with Gross Motor Function Classification System (GMFCS) I. OBJECTIVE: Evaluate the predictive value of sMRI, GMA, and HINE individually and in combination for early CP diagnosis and assess accuracy across varying GMFCS levels in a regional cohort of preterm infants. METHODS: We performed sMRI between 39-44 weeks postmenstrual age and GMA and HINE between 12-18 weeks CA in 395 preterm infants born at ≤32 weeks' gestation across five NICUs in Greater Cincinnati. Brain abnormalities on sMRI included white matter injuries, cortical and deep gray matter lesions, or extensive cerebellar hemorrhage. Absent fidgety movements constituted abnormal GMA; abnormal HINEs were scores <56. The primary outcome was CP diagnosis at 22-26 months CA, classified by the GMFCS. We calculated sensitivity, specificity, positive predictive value, negative predictive value, and likelihood ratios for individual tests and combinations. RESULTS: Of 338 (86%) infants with complete follow-up, 48 (14.2%) showed sMRI abnormalities, 15 (4.6%) had abnormal GMA, and 69 (20.9%) had abnormal HINE. Thirty-nine children (11.5%) developed CP at age 2, of which 28 had GMFCS level I and 11 had GMFCS >II. The combination of sMRI and GMA achieved 100% specificity but only 22% sensitivity while the combination of abnormal sMRI and HINE demonstrated sensitivity of 32% and specificity of 98% for prediction of CP. Individual or combined tests showed far higher sensitivity (78-100%) for predicting CP in children with GMFCS levels II-V. CONCLUSIONS: The combination of sMRI with GMA or HINE demonstrated high specificity but low sensitivity for early CP diagnosis in a regional cohort of preterm infants. This approach appears effective for early detection of CP levels II-V but not for level I cases, the most prevalent type, underscoring the need for continued developmental follow-up for all very preterm infants and need for more sensitive diagnostic tools for early detection of CP. KEY POINTS: Questions: What is the individual and combined prognostic accuracy of sMRI, GMA, and HINE for early diagnosis of CP in preterm infants?Findings: In our prospective, regional study of preterm infants born at ≤32 weeks' gestation, we found that combining brain abnormalities on sMRI with abnormal GMA achieved 100% specificity but 22% sensitivity for diagnosing CP. Individual or combined tests showed far higher sensitivity (78-100%) for predicting CP in children with GMFCS levels II-V. Both individual and combined tests were poor predictors of GMFCS level I CP, the most common type.Meaning: While sMRI combined with GMA or HINE is effective for diagnosing CP with GMFCS levels II-V, this approach falls short for children with GMFCS level I.