Comparative evaluation of MRS and SPECT in prognostication of patients with mild to moderate head injury

Magnetic resonance spectroscopy (MRS) and single-photon emission computed tomography (SPECT) have only been individually studied in patients with head injury. This study aimed to comparatively assess both in patients with mild to moderate head injury. Patients with a Glasgow Coma Scale (GCS) score of 9-14 who underwent MRS and/or SPECT were evaluated in relation to various clinical factors and neurological outcome at 3 months. There were 56 SPECT (Tc99m-ethylcysteinate dimer [ECD]) studies and 41 single voxel proton MRS performed in 53 patients, with 41 patients having both. Of the 41 who underwent MRS, 13 had a lower N-acetyl-aspartate/creatine (NAA/Cr) ratio, 14 had a higher choline (Cho)/Cr ratio, 19 were normal, and nine had bilateral MRS abnormalities. Of the 56 who underwent SPECT, 22 and 19 had severe and moderate hypoperfusion, respectively. Among those in Traumatic Coma Data Bank CT scan category 1 and 2, 50% had MRS abnormalities, whereas 64% had SPECT hypoperfusion, suggesting greater incremental validity of SPECT over MRS. In univariate analyses, GCS, moderate/severe hypoperfusion and bilateral SPECT changes were found to have significant association with unfavorable outcome (odds ratio 13.2, 15.9, and 4.4, and p values <0.01, 0.01, and 0.05, respectively). Patients with lower NAA/Cr ratio in MRS had more unfavorable outcomes, however this was not significant. In multivariate analysis employing binary logistic regression, GCS and severe hypoperfusion on SPECT were noted to have significant association with unfavorable outcome, independent of age, CT scan category, and MRS abnormalities (p values = 0.02 and 0.04, respectively). To conclude, ECD-SPECT seems to have greater sensitivity, incremental validity and prognostic value than single voxel proton MRS in select patients with head injury, with only severe hypoperfusion in SPECT significantly associated with unfavorable outcome independent of other confounding factors.