PURPOSE: To investigate with statistical analysis the relationship bet
ween brain injury measured with magnetic resonance (MR) imaging and th
at measured with proton (hydrogen-1) MR spectroscopy. MATERIALS AND ME
THODS: Forty-two patients (34 female, eight male; mean age +/- SD, 38.
7 years +/- 13.1; age range, 6-60 years) with systemic lupus erythemat
osus (SLE) were examined with H-1 MR spectroscopy to measure N-acetyla
spartate (NAA), creatine (Cr), and choline (Cho) levels in normal-appe
aring white matter and with MR imaging to detect anatomic abnormalitie
s. RESULTS: Results of linear regression analysis revealed an associat
ion between the NAA/Cr ratio and anatomic abnormalities (P = .03). How
ever, only small focal lesions were independently related to NAA/Cr ra
tio changes (P = .04). Results of a similar analysis showed associatio
ns between the Cho/Cr ratio and anatomic abnormalities (P = .002). An
elevated Cho/Cr ratio and cerebral infarction were independently assoc
iated (P = .02), as were a decreased Cho/Cr ratio and severe cortical
atrophy P = .02). CONCLUSION: Cerebrovascular abnormalities underlie d
iffuse cerebral injury in SLE, with small vessel injury (ie, small foc
al lesions) primarily associated with a decreased NAA/Cr ratio and med
ium vessel injury (ie, infarct) primarily associated with an increased
Cho/Cr ratio. Statistical integration of H-1 MR spectroscopic and MR
imaging findings over large data sets provides insights into the relev
ance of individual MR imaging-visible brain abnormalities in SLE. This
statistical approach may be applicable to other systemic diseases com
plicated by brain injury.