Neuropsychological Correlates of Basal Ganglia and Medial Temporal Lobe NAA/Cho Reductions in Traumatic Brain Injury FREE

Proton magnetic resonance spectroscopy (¹H-MRS) can test neurochemical status in patients with traumatic brain injuries (TBIs). The technique allows in vivo detection of neurochemical alterations such component as N-acetylaspartate (NAA), creatinine/phosphocreatinine, and choline-containing compounds (Cho) in selected tissue volumes.¹ A decreased concentration of NAA reflects neuronal death, and an increased concentration of Cho is related to inflammation, demyelination, and membrane synthesis or repair. In TBI the main findings are reduced NAA concentrations and increased Cho concentrations.²

Proton magnetic resonance spectroscopy metabolic abnormalities are present in the acute or subacute period in patients with TBI,^{3 - 6} persist after 6 months' evolution^{3 ,5} and correlate with outcome on clinical scales.^{3 ,5 - 8} To our knowledge, only 2 research groups correlated an NAA concentration decrease with neuropsychological deficits.^{3 - 4 ,9}

In previous studies MRS voxels were placed in neocortical gray matter^{3 - 4 ,9} or white matter,^{3 - 6 ,9 - 10} but none focused on brain structures such as basal ganglia or hippocampus which are sensitive to the hypoxic-ischemic, metabolic, and molecular events in severe head injuries.^{10 - 11} The aim of this study was to relate regional neurochemical abnormalities in the basal ganglia and medial temporal lobe to memory and frontal lobe impairment in TBI.

Twenty patients (mean [SD] age, 25.6 [7.15] years; mean [SD] educational status, 11.4 [3.0] years) who had moderate (n = 4) and severe (n = 16) TBI were matched to 20 healthy control subjects according to mean (SD) age (25.4 [7.4] years), sex (16 males), and mean [SD] educational status (12 [3.0] years). Severity was determined using Glasgow Coma Scale (GCS) scores. Patients with moderate TBI had a GCS score between 13 and 8, patients with severe TBI had a GCS score below 8. Patients were recruited from the Neurotraumatology Unit, Vall d'Hebron University Hospital, Barcelona, Spain. Inclusion criteria were a GCS score of 13 or less, an absence of focal lesions in the regions of interest on computed tomography, and a normal educational history. The neuroradiologist (N.B.) evaluated patients' present magnetic resonance imaging (MRI) data. Three patients showed lesions in the prefrontal poles, 4 in the temporal poles, and 1 in both frontal and temporal poles. T2-weighted images showed basal ganglia hyperintensities in 5 patients and hippocampal hyperintensities in 8 patients. Exclusion criteria were abnormal premorbid IQ, aphasia, dysarthria, or motor impairment precluding neuropsychological evaluation. All subjects participated voluntarily; none had a history of TBI or neurological or psychiatric diseases. The local ethical committee approved the study. Signed informed consent was obtained from participants or their parents or guardians. Sample characteristics are listed in Table 1.

We used a 1.5-T magnetic resonance scanner (Signa 5.4; General Electric, Milwaukee, Wis) to obtain 2 ¹ H-MRS voxels (2 × 2 × 2 cm³), one in the left basal ganglia and the other in the left midtemporal region from a coronal section (Figure 1). Water-suppressed spectra were acquired using a double-spin, echo point-resolved spectroscopy sequence with a repetition time of 1500 milliseconds and echo times of 114 milliseconds (basal ganglia voxel) and 35 milliseconds (hippocampus voxel). We obtained NAA (at 2.0 ppm) and Cho (at 3.15 ppm) concentration peaks for both locations. Spectra were analyzed using the manufacturer-supplied spectroscopy software package of the MR system.

The battery neuropsychological tests evaluated frontal and medial temporal lobe functions usually impaired after TBI. To assess memory we used the Rey Auditory Verbal Learning Test, the memory subtests from the Rey Complex Figure 1 Test, and the Warrington Face Recognition Memory Test. Frontal lobe functions were evaluated using the Word Fluency Test (verbal fluency), Continuous Performance Test (attention and information processing speed), Backward Digit Span (working memory), Trail Making Tests (parts A and B) (visual scanning, motor speed, attention, and mental flexibility), Symbol Digit Modalities Test (visual scanning, tracking, and motor speed), and Grooved Pegboard Test (fine motor speed). Global adjustment to activities of daily living and general outcome were assessed using the extended Glasgow Outcome Scale. Neuropsychological tests were administered by a single neuropsychologist (M.A.) blind to clinical and spectroscopic data.

All statistical analyses were done using SPSS version 11.0 for Windows (SPSS Inc, Chicago, Ill). The Kolmogorov-Smirnov test showed that all continuous variables had normal distribution. The t test was, therefore, used for independent samples to compare group means in both neuropsychological and MRS variables. The Pearson product moment correlation test was used to correlate spectroscopic metabolites and neuropsychological tests. Since this was an exploratory study, we did not correct for multiple comparisons, but we applied a 2-tailed level of significance of P<.01.

As expected, NAA/Cho ratios were significantly lower in both regions in patients with TBI than in controls. Patients with TBIs differed significantly from controls in all neuropsychological tests (Table 2).

The NAA/Cho ratio in basal ganglia correlated with measures of fine motor speed and attention: Backward Digit Span (r = 0.63, P = .003), Trail Making Test A (r = 0.58, P = .007) (Table 3). In controls, the relationships between neurochemical levels and neuropsychological performance were not significant. For the whole sample of participants, in the hippocampus we found that the NAA/Cho ratio achieved significant correlations on the Rey Auditory Verbal Learning Test (r = 0.40, P = .009) and on facial memory recognition (the Warrington Face Recognition Memory Test) (r = 0.50, P = .001). In basal ganglia the NAA/Cho ratio correlated with Backward Digit Span (r = 0.64, P = .00) and Trail Making Test A (r = 0.56, P = .00).

We attempted to relate NAA/Cho concentrations to clinical data, in particular to the presence of hypoxia (minimum cerebral perfusion pressure), a minimum brain tissue oxygen concentration (PtiO₂), and maximum intracranial pressure. We only found a trend toward correlation between the number of local hypoxic episodes (brain tissue oxygen concentration <15 mm Hg) and the NAA/Cho ratio in the hippocampus (r_s = −0.66, P = .02).

Relationships between ¹H-MRS measurements of metabolic alteration and neuropsychological impairment have been previously reported.^{3 - 4 ,9} Our results agree with these findings. However, whereas those studies assessed a decrease in the NAA concentration to measure diffuse brain damage, we were looking for focal relationships. As basal ganglia have strong connections with neocortex frontal lobes¹² and are involved in cognitive functions, we expected to find a relation between frontal alterations and the decrease in the basal ganglia NAA concentration. Similarly, as the medial temporal regions contain the hippocampus, we expected a relation between a decreased NAA concentration and memory dysfunctions.

Our main finding was the relationship between frontal lobe neuropsychological impairment at 6 months and a decrease in the basal ganglia NAA concentration in patients with TBIs. We found significant correlations between the NAA/Cho ratio with tests that measure frontal lobe functions such as visual scanning, fine motor speed, attention, and working memory. Hippocampal NAA/Cho ratio correlations were less consistent; we only found a tendency for visual recognition.

For the whole sample (patients and controls) we observed significant correlations in both NAA/Cho voxels. The metabolites in the hippocampal region correlated with memory tests and in the basal ganglia region with frontal lobe tests. However, these correlations also reflect the between-group differences in both neuropsychological and spectroscopic findings.

Our voxels focused on gray subcortical structures, but we cannot guarantee that the NAA level decrease was not in fact due to the presence of white matter changes. However, the different correlation patterns observed in the voxels and their relationship with the functional properties of the target structures suggest specificity. So voxel location is a factor in detecting long-term neurological deficits in TBI. We found that decreases in the basal ganglia NAA level reflect neuropsychological sequelae.

Basal ganglia and hippocampus are sensitive to hypoxia and ischemia, and the action of excitotoxins on glutamate receptors influences the development of acute central nervous system injury.^{11 ,13} We were unable to relate the NAA level decrease with clinical data reflecting hypoxia or ischemic damage. Further studies with larger samples are needed to investigate these factors. To conclude, our results suggest that abnormal metabolite concentrations in basal ganglia are related to frontal lobe deficits.

Corresponding author: C. Junqué, PhD, Departament de Psiquiatria i Psicobiologia Clínica, Universitat de Barcelona, Institut d'Investigacions Biomediques August Pi i Sunyer, C/Casanova 143, 08036 Barcelona, Spain (e-mail: cjunque@ub.edu).

Accepted for publication October 17, 2003.

Author contributions: Study concept and design (Mrs Ariza and Dr Junqué); acquisition of data (Mrs Ariza and Drs Bargalló and Olondo); analysis and interpretation of data (Mrs Ariza and Drs Junqué, Mataro, Poca, Bargalló, Olondo, and Sahuquillo); drafting of the manuscript (Mrs Ariza and Drs Junqué and Bargalló, ); critical revision of the manuscript for important intellectual content (Drs Mataró, Poca, Bargalló, Olondo, and Sahuquillo); statistical expertise (Mrs Ariza); obtained funding (Drs Junqué, Poca, and Sahuquillo); administrative, technical, and material support (Mrs Ariza and Drs Junqué, Mataró, Bargalló, Olondo, and Sahuquillo); study supervision (Dr Junqué).

This study was supported by the grants PM 98-0192 from Ministerio de Ciencia y Tecnologia, Madrid, Spain, and 2001SGR 00139 from the Generalitat de Catalunya, Catalunya, Spain, and by a research grant from the Ministerio de Ciencia y Tecnologia, Barcelona, Spain (Mrs Ariza).