Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Sep 4.
Published in final edited form as: J Neurovirol. 2010 Oct 20;16(6):435–444. doi: 10.3109/13550284.2010.520817

Cerebral metabolite abnormalities in human immunodeficiency virus are associated with cortical and subcortical volumes

Ronald A Cohen 1, Jaroslaw Harezlak 2, Assawin Gongvatana 1, Steven Buchthal 3, Giovanni Schifitto 4, Uraina Clark 1, Robert Paul 5, Michael Taylor 6, Paul Thompson 7, David Tate 8, Jeffery Alger 7, Mark Brown 9, Jianhui Zhong 4, Thomas Campbell 9, Elyse Singer 7, Eric Daar 7, Deborah McMahon 10, Yuen Tso 11, Constantin T Yiannoutsos 2, Bradford Navia 12; the HIV Neuroimaging Consortium
PMCID: PMC4560459  NIHMSID: NIHMS352718  PMID: 20961212

Abstract

Cerebral metabolite disturbances occur among human immunodeficiency virus (HIV)-infected people, and are thought to reflect neuropathology, including proinflammatory processes, and neuronal loss. HIV-associated cortical atrophy continues to occur, though its basis is not well understood, and the relationship of cerebral metabolic disturbance to structural brain abnormalities in HIV has not been well delineated. We hypothesized that metabolite disturbances would be associated with reduced cortical and subcortical volumes. Cerebral volumes were measured in 67 HIV-infected people, including 10 people with mild dementia (acquired immunodeficiency syndrome [AIDS] dimentia complex [ADC] stage >1) via automated magnetic resonance imaging (MRI) segmentation. Magnetic resonance spectroscopy (MRS) was used to measure levels of cerebral metabolites N-acetylaspartate (NAA), myo-inositol (MI), choline-containing compounds (Cho), glutamate/glutamine (Glx), and creatine (Cr) from three brain regions (frontal gray matter, frontal white matter, basal ganglia). Analyses were conducted to examine the associations between MRS and cerebral volumetric measures using both absolute and relative metabolite concentrations. NAA in the mid-frontal gray matter was most consistently associated with cortical (global, frontal, and parietal), ventricular, and caudate volumes based on analysis of absolute metabolite levels, whereas temporal lobe volume was associated with basal ganglia NAA and Glx, and Cho concentrations in the frontal cortex and basal ganglia. Hippocampal volume was associated with frontal white matter NAA, whereas thalamic volume was associated with both frontal white matter NAA and basal ganglia Glx. Analyses of relative metabolite concentrations (referenced to Cr) yielded weaker effects, although more metabolites were retained as significant predictors in the models than the analysis of absolute concentrations. These findings demonstrate that reduced cortical and subcortical volumes, which have been previously found to be linked to HIV status and history, are also strongly associated with the degree of cerebral metabolite disturbance observed via MRS. Reduced cortical and hippocampal volumes were most strongly associated with decreased NAA, though reduced Glx also tended to be associated with reduced cortical and subcortical volumes (caudate and thalamus) as well, suggesting both neuronal and glial disturbances. Interestingly, metabolite-volumetric relationships were not limited to the cortical region from which MRS was measured, possibly reflecting shared pathophysiological processes. The relationships between Cho and volumetric measures suggest a complicated relationship possibly related to the effects of inflammatory processes on brain volume. The findings demonstrate the relationship between MRI-derived measures of cerebral metabolite disturbances and structural brain integrity, which has implication in understanding HIV-associated neuropathological mechanisms.

Keywords: brain, cerebral metabolites, cortical volume, HIV, magnetic resonance spectroscopy, morphometry, subcortical volume

Introduction

Human immunodeficiency virus (HIV) infection continues to cause cognitive impairments and brain dysfunction despite the fact that highly active antiretroviral therapy (HAART) has dramatically reduced mortality and the prevalence of HIV-associated dementia (Sacktor et al, 2002). Clinical and neuropathological studies suggest that HIV will continue to exert detrimental brain effects even when viral load is well controlled (Sacktor et al, 2002; Brew, 2004; McArthur, 2004; Valcour and Sacktor, 2002).

Structural brain abnormalities have been reported since the early years of the HIV epidemic, with evidence of reduced cortical volumes, particularly among people with advanced disease in the pre-HAART era (Aylward et al, 1995; Heindel et al, 1994; Jernigan et al, 1993). Following the advent of HAART, cortical brain changes received less attention, as it appeared that antiretroviral therapies were ameliorating many of the brain disturbances associated with HIV. Yet recent studies suggest that structural brain abnormalities continue to occur, particularly of the basal ganglia and cerebral white matter. For example, reduced white matter integrity has been found among a cohort of largely HAART-treated HIV-infected individuals compared to healthy controls (Gongvatana et al, 2009). Relatively few cortical volumetric studies exist in the post-HAART era, though emerging evidence suggests that cortical atrophy continues to occur (Thompson et al, 2005). We recently observed significant relationships between clinical variables and both cortical and subcortical volumes among people infected with HIV, with disease history (nadir CD4, duration of infection) being most strongly associated with cortical volumes (Cohen et al, 2010). This raises the concern that brain pathology incurred during the clinical history may increase vulnerability to neurodegeneration, with cortical changes analogous to premature aging (Brew et al, 2009; Valcour et al, 2004).

A separate line of research has focused on characterizing cerebral metabolite disturbances among patients with HIV, in an effort to link the viral and immunological effects of HIV with neuropathological processes in the brain. Proton (1H) magnetic resonance spectroscopy (MRS) provides measures of specific metabolites that reflect different pathological processes. Numerous MRS studies of HIV have demonstrated abnormalities of the cerebral metabolites N-acetylaspartate (NAA), myo-inositol (MI), choline-containing compounds (Cho), and glutamate/glutamine (Glx) (Chang et al, 2004; Schifitto et al, 2009a; Schweinsburg et al, 2005; Paul et al, 2007; Lentz et al, 2009). Decreased NAA is known to reflect neuronal injury or loss, whereas increased Cho and MI have been linked to proinflammatory processes, and a breakdown of cell membranes (Ross and Bluml, 2001). Glx abnormalities have also been observed among patients with HIV, though the neuropathological implications are unclear. Furthermore, there is mounting evidence that cerebral metabolite abnormalities detected on MRS are associated with impaired neurocognitive function and that they are greatest among people with HIV-associated dementia (Chang et al, 2004; Paul et al, 2007, 2008; Patel et al, 2003; Yiannoutsos et al, 2004). Yet, the relationship of these cerebral metabolites to structural brain abnormalities in HIV has not been well delineated, and a full understanding of the evolution of cortical and subcortical injury in the context of chronic infection is still lacking.

To address this question, we examined the relationship between cerebral metabolite concentrations measured by MRS and the volumes of specific cortical and subcortical regions. Structural magnetic resonance imaging (MRI) and MRS data were obtained from a patient pool of HIV-infected patients followed in a large multicenter study designed to assess the effects of HIV infection on brain structure and function. The extent to which HIV clinical factors together with cerebral metabolite abnormalities contributed to association with brain volumes was also assessed. We hypothesized that reduced cortical and subcortical volumes would be significantly associated with the extent of metabolite disturbance.

Results

Table 1 summarizes the demographic and clinical characteristics of the sample.

Table 1.

Demographic and clinical characteristics of participants in the AIDS dementia complex (ADC) groups (continuous variables reported as mean ± SD, and categorical variables as percentages)

ADC stage
0 and 0.5 ≥1 Total
N 57 10 67
Gender (% male) 85 80 82
Age (years) 48.0 ± 8.2 48.7 ± 5.9 48.1 ± 7.8
Ethnicity (% Caucasian) 74 70 73
Education level (% some college or more) 60 80 63
% IV drug use 19 40 22
Duration of infection (years) 11.5 ± 5.8 16.0 ± 5.4 12.2 ± 5.9
CD4 count (log10 cells/ml) 2.55 ± 0.22 2.38 ± 0.43 2.52 ± 0.26
Nadir CD4 (log10 cells/ml) 1.48 ± 0.51 1.33 ± 0.38 1.46 ± 0.50
% undetectable Plasma RNA viral load 82 80 82
% known on ART at baseline (uninterrupted) 84 60 81
Open in a new tab

Note. The two ADC groups did not significantly differ on any demographic or clinical variables.

Two sets of analyses were performed with the MRS metabolites as predictors in the regression models. Absolute MRS measures were used in the first set of analyses, whereas the second set of analyses used metabolite/creatine (Cr) ratios. Table 2 shows the relationships between regional brain volumes and cerebral metabolite measures.

Table 2.

Relationships between regional brain volumes and cerebral metabolites

Absolute metabolite measures Metabolite/Cr ratios
β P Adjusted R2 P β P Adjusted R2 P
Gray matter volume .2396 .0003 Gray matter volume .1580 .0147
NAA in FGM .0158 <.0001 Cho in FGM .3078 .0081
Glx in FWM −.0445 .0131
NAA in BG .0430 .0311
Cho in BG −.2730 .0359
Ventricular volume .4335 <.0001 Ventricular volume .2184 .0035
NAA in FGM −.0240 <.0001 Cho in FGM −.4392 .0065
NAA in BG −.0108 .0265 NAA in BG −.0751 .0083
Glx in FWM .0069 .0720 Glx in FWM .0651 .0094
NAA in FWM −.0728 .0268
Cho in BG .3575 .0466
Frontal lobes volume .3020 <.0001 Frontal lobes volume .2202 .0034
NAA in FGM .0062 <.0001 Cho in FGM .1150 .0022
Glx in FWM −.0152 .0085
NAA in BG .0167 .0108
Cho in BG −.0851 .0403
NAA in FWM .0152 .0447
Temporal lobes volume .5404 <.0001 Temporal lobes volume .3942 <.0001
NAA in BG .0033 <.0001 Cho in FGM .0798 <.0001
Cho in FGM .0080 .0012 NAA in BG .0121 .0001
Cho in BG −.0100 .0031 Cho in BG −.0780 .0001
Glx in BG .0010 .0168
Parietal lobes volume .3255 <.0001 Parietal lobes volume .2755 .0001
NAA in FGM .0023 .0001 Glx in FWM −.0083 .0016
Glx in FWM −.0009 .0290 NAA in FGM .0111 0.0060
Thalamus volume .3089 <.0001 Thalamus volume .2662 .0003
NAA in FWM .4802 .0010 NAA in FWM 1.9652 .0124
Glx in BG .2670 .0021 Cho in FGM 7.1133 .0434
Glx in FGM .9514 .0742
Caudate volume .2687 .0002
NAA in FGM .4323 .0083
Glx in FGM .2175 .0690
Hippocampus volume .1694 .0036
NAA in FWM .3821 .0025
Open in a new tab

Note. Both absolute metabolite measures and metabolite/creatine ratios are included. Volumetric measures have been adjusted by total intracranial volume. Adjusted R2 values are for the whole model fit.

Absolute metabolite concentration analyses

Global brain volumes

Increased total gray matter volume was associated with increased NAA in the frontal gray matter (P < .0001). Increased ventricle volume was related to decreased NAA in the frontal gray matter (P < .0001), decreased NAA in the basal ganglia (P = .0265), and increased Glx in the frontal white matter (P = .072). There were no associations found between the metabolite levels and total white matter volume.

Cortical regional volumes

Increased frontal lobe volume was associated with increased NAA in the frontal gray matter (P < .0001). Increased temporal lobe volume was associated with increased NAA (P < .0001), decreased Cho (P = .0031), and increased Glx (P = .0168) in the basal ganglia; and increased Cho in the frontal gray matter (P = .0012). Increased parietal lobe volume was related to increased NAA in the frontal gray matter (P = .0001) and decreased Glx in the frontal white matter (P = .029).

Subcortical structures and the hippocampus

Increased caudate volume was associated with increased NAA (P = .0083) and a trend towards increased Glx (P = .069) in the frontal gray matter. Increased thalamic volume was associated with increased NAA in the frontal white matter (P = .001), and increased Glx in the basal ganglia (P = .0021). Increased hippocampal volume was associated with increased NAA in the frontal white matter (P = .0025).

Relative metabolite concentration analyses

Global brain volumes

Increased total gray matter volume was associated with increased Cho in the frontal gray matter (P = .0081), decreased Glx in the frontal white matter (P = .0131), and increased NAA (P = .0311) and decreased Cho (P = .0359) in the basal ganglia. Increased ventricular volume was related to decreased NAA (P = .0083) and increased Cho (P = .0466) in the basal ganglia, decreased NAA (P = .0268) and increased Glx (P = .0094) in the frontal white matter, and decreased Cho in the frontal gray matter (P = .0065). Again there were no associations found between the metabolite levels and total white matter volume.

Cortical regional volumes

Increased frontal lobe volume was associated with increased NAA (P = .0108) and decreased Cho (P = .0403) in the basal ganglia, increased NAA (P = .0447) and decreased Glx (P = .0085) in the frontal white matter, and increased Cho in the frontal gray matter (P = .0022).

Increased temporal lobe volume was associated with increased NAA (P = .0001) and decreased Cho (P = .0001) in the basal ganglia, and increased Cho in the frontal gray matter (P < .0001). Increased parietal lobe volume was associated with increased NAA in the frontal gray matter (P = .006) and decreased Glx in the frontal white matter (P = .0016).

Subcortical structures

Only the thalamus was found be significantly associated with relative concentrations of the MRS metabolites. Increased thalamic volume was associated with increased NAA (P = .0124) in the frontal white matter, and increased Cho (P = .0434) and Glx (P = .0742) in the frontal gray matter.

Contribution of clinical measures

The addition of the clinical variables to the models did not change the MRS indices retained as the best predictors of the brain volumes. Furthermore, these additional variables were retained as significant predictors in only a few of the analyses. None of the clinical variables were retained as significant predictors for analyses conducted with relative metabolite concentrations. When absolute metabolite concentrations were analyzed, clinical variables were retained as predictors for total cortical gray matter, frontal gray matter, and caudate nucleus. Longer duration of HIV infection was associated with decreased total cortical gray matter volume (P = .0434), though the addition of this variable did not substantially increase the overall strength of the model (adjusted R2 = .2431, P = .0013). Participants with undetectable plasma HIV viral load exhibited increased frontal lobe volume (P = .0385), and the addition of this variable resulted in a 4.2% increase in the strength of the model (adjusted R2 = .3442, P < .0001). Undetectable HIV viral load was also associated with increased caudate volume (P = .0029), resulting in a 5.4% increase in model strength (adjusted R2 = .3234, P < .0001). The clinical indices were not retained as significant predictors of any other brain volumetric measure. Neither acquired immunodeficiency syndrome (AIDS) dementia complex (ADC) stage nor the interaction of ADC stage with any MRS metabolite measure was retained as significant predictors of any of the brain volumes.

Discussion

Initial optimism that HIV-associated structural brain abnormalities would be relatively rare following the widespread use of HAART have been tempered by recent evidence. Data from our group recently demonstrated reduced brain volumes in HIV-infected people relative to age-matched seronegative controls (Tate et al, 2010). Another recent study in the same cohort showed that cortical and subcortical volumes were significantly associated with HIV disease history and current clinical status (Cohen et al, 2010). In the current study, we extended these findings by demonstrating the link between cortical and subcortical brain volumes derived from structural MRI and cerebral metabolite measures derived from MR spectroscopy reflecting neuronal damage, cell membrane breakdown, inflammatory pathophysiology, and neurotoxicity. It is noteworthy that these findings were evident in the context of well-controlled HIV disease, as only about a third of the cohort exhibited clinically significant cognitive problems (i.e., ADC stage ≥0.5).

The most robust findings in this study involve the relationship between NAA and absolute metabolites concentrations. Cortical NAA was positively associated with both total cortical gray matter volume, and the volumes of specific cortical regions including the frontal and parietal cortex. Furthermore, decreased NAA levels in both the mid-frontal gray matter and basal ganglia were associated with increased ventricular volume. These findings are consistent with the idea that decreased NAA is an indicator of neuronal damage or loss.

Glx concentrations also emerged as significant predictors of many cortical and subcortical volumes. Though the direction of the relationships differed across regions, Glx tended to be negatively associated with cortical volume and positively associated with ventricular volume. These findings are consistent with previously reported associations between increased Glx and brain damage caused by chronic and acute hepatic encephalopathy, hypoxia, near drowning, and ornithine transcarbamylase deficiency (Binesh et al, 2006; Gropman and Batshaw, 2004; Kreis et al, 1996; Long et al, 2009; Sijens et al, 2008). Findings for other brain disorders have been mixed with regards to the direction of Glx effects. For example, reduced Glx concentrations have been observed in Alzheimer’s disease (Antuono et al, 2001; Jones and Waldman, 2004). In HIV, human and animal models have shown decreased levels of cerebral Glx (Lentz et al, 2009; Ratai et al, 2009; Chang et al, 1997), including patients with progressive multifocal leukoencephalopathy. Another study showed decreased frontal Glx concentrations, along with increased white matter integrity on diffusion tensor imaging among patients treated with lithium for HIV-associated cognitive impairment (Schifitto et al, 2009b). Accordingly, although alteration of cerebral Glx level appears to be common in brain disorders, whether increased or decreased levels are observed seems to depend on the nature of the underlying disturbance. Additional studies are needed to better understand Glx effects in HIV, particularly in the context of antiretroviral therapy and comorbidities such as hepatitis C, liver disease, and substance abuse.

The fact that increased Cho in the frontal gray matter was related to increase in overall cortical, frontal, and temporal gray matter volumes, and decreased ventricular volume was somewhat unexpected; as one might assume that the pathophysiology reflected by elevated Cho would also manifest itself through volume loss. Although the reason for this effect cannot be determined from this study, this finding suggests that inflammatory processes linked to Cho may actually contribute to an increase in brain volume, perhaps as a result of swelling and cellular responses to injury. Interestingly, the opposite effect was observed for Cho in the basal ganglia, as increased basal ganglia Cho level correlated with decreased cortical volumes and increased ventricular volume. The fact that the direction of effect for Cho depended on whether it was measured from the cortical gray matter versus basal ganglia suggests that the neuropathology (e.g., inflammation) associated with Cho may differ across different of brain tissues. The fact that different directions of Cho effects occurred within single statistical models for particular cortical areas suggests that the relationship between cerebral Cho levels and brain volume may be more complicated than for other metabolites.

MI failed to be retained as significant correlate of cortical or subcortical volumes in any of the statistical models. Cerebral MI levels have been linked to HIV status and history in past studies (Chang et al, 1999, 2002, 2003, 2004; Yiannoutsos et al, 2004; Sacktor et al, 2005; Tarasow et al, 2003). There is evidence that MI levels are associated with gliosis and inflammatory response (Yiannoutsos et al, 2004; Jones et al, 2004; Broom et al, 2007; Tkac et al, 2007). That it was not related to brain volume is noteworthy, though the meaning of this is not entirely clear at this point.

Interestingly, temporal lobe volume did not relate to metabolite concentrations in the same manner as other cortical regions. Whereas frontal and parietal volumes were most strongly associated with NAA levels in the mid-frontal gray matter, temporal lobe volume was more strongly associated with NAA, Cho, and Glx levels in the basal ganglia, and Cho level in frontal grey matter. The fact that hippocampal volume was associated with NAA level in the frontal white matter, and not grey matter, also indicates that temporal/hippocampal brain volumes may be differently affected than other regions.

It is noteworthy that the observed metabolite-volumetric relationships occurred in regions that were not confined to regions where MRS was sampled. For example, parietal lobe volume was associated with NAA level in the frontal gray matter; frontal lobe volume was not only associated with NAA level in the frontal lobe, but also in the basal ganglia; and temporal lobe volume was associated with basal ganglia NAA level. These results suggest that metabolite concentrations in a specific brain region can be sensitive to changes occurring more globally across the cortex, likely reflecting interrelated pathophysiological processes.

Comparison of the results derived from analyses of the absolute versus relative metabolite concentrations provides potentially useful insights into the findings from cerebral MRS studies in HIV. The magnitudes of the metabolite-volumetric relationships based on analysis of the absolute metabolite concentrations were generally greater. Furthermore, the analysis of absolute metabolite concentrations tended to point to the dominance of fewer metabolites being associated with brain volumes. In contrast, the analysis of relative metabolite concentrations yielded weaker overall magnitude of associations, but with a larger number of metabolites contributing to the association. For example, frontal lobe volume was strongly associated only with frontal gray matter NAA absolute concentrations with an overall model R2 of .30; but was related to NAA/Cr, Cho/Cr, and Glx/Cr measures in the frontal grey matter, frontal white matter, and basal ganglia with a significantly lower overall model R2 of .22. Frontal gray matter NAA typically did not emerge as a significant predictor in the relative concentration models. This was likely a direct result of the significant correlation between frontal gray matter NAA and Cr (R = .75, P < .05), thus diminishing the influence of the NAA/Cr ratio as a predictor. The implication of these findings is that relative concentrations may be useful for detecting smaller effects of Cho and Glx on brain volume in HIV, whereas absolute concentrations may provide a better estimate of metabolites account for the majority of volumetric effects.

The fact that few clinical variables were retained along with the metabolite levels as significant predictors of brain volume is also noteworthy. When the clinical variables are considered independent of the MRS measures, they are predictive of brain volumes (Cohen et al, 2010). Yet, these effects are relatively small compared to the association between cerebral metabolite concentrations and brain volumes, and the current findings indicate that when both types of data are considered together, the MRS measures are better at predicting both cortical and subcortical volumes. This is not surprising given the fact that the both cerebral metabolite and volumetric indices are both direct brain measures, whereas the clinical indices less directly reflect brain pathology. Furthermore, most patients in the study were neuroasymptomatic, HAART-treated, and had relatively well-controlled HIV disease. Among the few clinical variables that emerged, undetectable HIV viral load was significantly associated with increased frontal and caudate volumes, indicating the importance of virologic control. Duration of infection was related to decreased total cortical gray matter volume, pointing to the impact of disease history (Cohen et al, 2010). The lack of effects of ADC stage and its interaction with metabolite concentrations likely reflects the low rate of neurocognitive disorder in the cohort (85% with ADC stage <1). Importantly these findings also suggest that the association between cerebral metabolite concentrations and brain volume may exist independent of cognitive and functional status.

The current results demonstrated that clear relationships exist between cerebral metabolite levels and the cortical/subcortical brain volumes in HIV-infected individuals despite effective antiretroviral treatment. This reinforces previous findings that HIV-infected patients in the HAART era are still affected by brain disturbances. However, it remains unclear whether this reflects ongoing neuropathology associated with chronic HIV infection, or damage sustained at some point in the past, perhaps before viral load was adequately controlled with medication. Typically, gross morphometric changes are thought to occur as a late manifestation of neurodegenerative condition, such as Alzheimer’s disease. Yet, in the current study, relationships between the cerebral metabolites measured by MRS and brain volume were observed among patients who for the most part were asymptomatic and have good immune function and virologic control. In light of our previous findings that nadir CD4 and duration of infection are among the clinical measures most strongly associated with brain volumes in HIV-infected people (Cohen et al, 2010), it is possible that factors tied to patients history of infection are playing a significant role in affecting brain structure beyond current disease status.

The differential susceptibility regions of the brain to volume loss and their relationship to cerebral metabolite levels measured by MRS points to the potential power of these methods for studying and measuring the effects of host and viral factors as well as their interactions, and together provide a framework to further understand structural change in the HIV infected brain in the setting of chronic disease and HAART. It should be emphasized that these findings are based on cross-sectional analysis of baseline data. Validation of these results will depend on longitudinal studies integrating multimodal neuroimaging data, including MRS and morphometry, along with neurocognitive and clinical measures. The relationship between neuronal loss and the more acute pathologiscal processes associated with HIV on brain structure needs to be disentangled in future studies.

Methods

Clinical sample

Sixty-seven HIV-infected patients enrolled in a longitudinal study of HIV were assessed as part of their baseline examinations. The sample had a larger proportion of men (82%) than women. Patients came from HIV clinics at three sites (University of Rochester, University of Colorado, and Stanford University) and consisted of the first set of individuals for whom brain morphometry data were available. Table 1 summarizes the demographic and clinical characteristics of the sample.

This clinical cohort consisted of individuals who were middle-aged (mean = 48.1, SD = 7.8 years) and relatively well educated (63% with some college or more education). Most had been infected for over 10 years (66%) and had a CD4 nadir <50 cells/ml (59%). A majority (81%) of participants were HAART-treated at the time of the study. Accordingly, we observed more intact current CD4 levels (on the log10 scale, mean = 2.52, SD = 0.26), and only a small proportion (18%) of participants with detectable plasma viral load.

Neurocognitive status was assessed and using the AIDS dementia complex staging scale, as previously described (Brew et al, 1995). Patients were assessed on both clinical and neuropsychological tests and rated as no impairment, stage 0; subclinical impairment, stage 0.5; mild impairment, stage 1; moderate, stage 2; or severe, stage 3. Neurocognitive impairment was defined as performance of at least 1.0 standard deviation below normative values on two or more neuropsychological tests or at least 2.0 standard deviations below normative values on one or more tests within any neurocognitive domains. Subsequent to the inception of this study, Antinori and colleagues published a new classification in which ADC stages 0.5 and 1 would correspond to mild cognitive disorder or motor neuron disease (MND) and ADC stage 2 or greater would correspond to HIV-associated dementia or HAD (Antinori et al, 2007). Ten participants (17%) currently exhibited at least mild AIDS dementia complex (ADC stage ≥1), whereas the remaining 57 individuals were neuroasymptomatic (NA).

Structural MRI data acquisition and processing

High-resolution whole-brain structural images were acquired using a T1-weighted MPRAGE sequence with the following parameters: TE = 3.57 ms, TR = 2730 ms, flip angle = 7°, FOV = 256 × 256 mm, 1 × 1 × 1 mm resolution. Brain volumes were measured with the Individual Brain Atlases using Statistical Parametric Mapping IBASPM (Alemán-Gómez et al, 2006), toolbox for the Statistical Parametric Mapping 5 (SPM5; Wellcome Department of Imaging Science; www.fil.ion.ucl.ac.uk/spm/) software package running under Matlab. Image processing involved segmentation of individual brain volumes into gray matter, white matter, and cerebrospinal fluid (CSF) compartments, yielding estimated volumes for each compartment, along with the total intracranial volume. Segmented brain volumes were normalized via nonlinear registration to the MNI152 template, and gray matter voxels were labeled according to a predefined anatomical atlas (Tzourio-Mazoyer et al, 2002). Labeled brain volumes were then inverse-transformed into their native spaces, yielding estimated volumes of specific brain regions according to the atlas labels. To correct for head size variability, ratios of gray matter, white matter, and ventricular volumes, and lobular volumes to total intracranial volume were used as volumetric measures.

MRS data acquisition and processing

Single-voxel 1H spectra were acquired using a customized version of the PRESS sequence. Voxels 6 cc in volume were prescribed in three regions: midline frontal gray matter, right or left frontal white matter in the centrum semiovale, and right or left basal ganglia. The hemisphere for frontal white matter and basal ganglia regions was alternated from subject to subject. Field homogeneity and water suppression were adjusted using automated algorithms from GE and Siemens. Water suppressed spectra were collected with TE/TR = 35/3000 ms, bandwidth = 2500 Hz, 128 averages, NEX = 8.

In addition, the customized pulse program automatically collects single-scan fully relaxed water FIDs from each voxel at 7 different echo times (TE = 30, 45, 65,100, 200, 500, and 1500 ms; TR = 15 s) from which metabolite concentrations are calculated. To control for a possible instrument bias, we collected phantom MRS data concurrently with the subject evaluation using the identical protocol as described above. The metabolite levels obtained from the phantoms provide a summary of conditions present at the time of the subject evaluation thus explaining some of the variations across sites and time points.

The time domain spectral data were transferred to an FTP server at the central MRS processing site Hawaii. The metabolite ratios NAA/Cr, Cho/Cr, MI/Cr, and Glx (= Glu + Gln)/Cr were determined using the LC Model spectral analysis software (Provencher, 2001) and an unsuppressed water FID at TE = 30 ms for eddy-current correction. This technique yields interindividual variations of about 10% to 15%, and intrasubject variability of 3% to 8%, for the concentrations of the major metabolite peaks (Lee et al, 2003).

Statistical analysis

We tested whether the MRS metabolites (NAA, Cho, MI, Glx) measured in three brain regions (frontal gray matter, frontal white matter, and basal ganglia) were differentially associated with brain volumes using linear regression. Three sets of volumetric measures were used as dependent variables: (1) gray matter, white matter, and total ventricular volumes; (2) frontal, parietal, and temporal lobe cortical volumes; and (3) caudate, putamen, thalamus, and hippocampus volumes. The MRS metabolites serving as the independent measures were analyzed using two different methods: (1) absolute metabolite concentrations; (2) ratios of metabolite/Cr. The rationale using both of these approaches is that clinical research literatures in HIV now exist for MRS using both absolute and relative metabolite concentrations. Subsequently, HIV clinical variables (current CD4, nadir CD4, plasma HIV RNA, years since HIV diagnosis, antiretroviral treatment status), hepatitis C virus (HCV) status, ADC stage were entered into the regression analyses to examine the added value of these variables in predicting brain volumes.

All linear regression models were adjusted for age, sex, and race. All statistical analyses were done using R-2.9.2 (R Core Development System: http://www.rproject.org). Final linear regression models were selected by minimizing Akaike information criterion (AIC) (Akaike, 1974; Burnham and Anderson, 2002), which balances the model fit and its complexity. Increasing the number of parameters in the models improves their fit to the data, but at a cost of increased complexity. AIC balances the goodness of fit and the number of included covariates by penalizing the number of parameters in the model. The best model is the one with the lowest AIC. This method is more robust than the traditional stepwise selection procedures and produces parsimonious models balancing the goodness of fit and model complexity. To ameliorate the selection of models that might contain non-significant variables, we used a bootstrap procedure on the results of the initial fit, and only chose the variables for the final models that were selected in more than 70% of the bootstrapped samples.

Acknowledgments

The authors would like to acknowledge and express their appreciation for the efforts on all the MRS consortium sites involved in this multicenter project, including the Colorado Clinical and Translational Research Institute (RR025780, Sokol, RJ), Principal Investigator [PI], and the staff, clinicians, and investigators from these sites that made this study possible. The authors also appreciate the support of the National Institutes of Health who provide the support and funding for this research, including R01 NS036524 “Proton MRS Studies of Cerebral Injury in HIV Infection” (Brad Navia, PI); R01 MH074368 “Age Effects on HIV-Associated Brain Dysfunction” (Ronald Cohen, PI).

The authors alone are responsible for the content and writing of the paper.

Footnotes

Declaration of interest: The authors report no conflicts of interest.

References

  1. Akaike H. A new look at the statistical model identification. IEEE Trans Autom Control. 1974;19:716–723. [Google Scholar]
  2. Antinori A, Arendt G, Becker JT, Brew BJ, Byrd DA, Cherner M, Clifford DB, Cinque P, Epstein LG, Goodkin K, Gisslen M, Grant I, Heaton RK, Joseph J, Marder K, Marra CM, McArthur JC, Nunn M, Price RW, Pulliam L, Robertson KR, Sacktor N, Valcour V, Wojna VE. Updated research nosology for HIV-associated neurocognitive disorders. Neurology. 2007;69:1789–1799. doi: 10.1212/01.WNL.0000287431.88658.8b. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Aylward EH, Brettschneider PD, McArthur JC, Harris GJ, Schlaepfer TE, Henderer JD, Barta PE, Tien AY, Pearlson GD. Magnetic resonance imaging measurement of gray matter volume reductions in HIV dementia. Am J Psychiatry. 1995;152:987–994. doi: 10.1176/ajp.152.7.987. [DOI] [PubMed] [Google Scholar]
  4. Antuono PG, Jones JL, Wang Y, Li SJ. Decreased glutamate + glutamine in Alzheimer’s disease detected in vivo with (1)H-MRS at 0.5 T. Neurology. 2001;56:737–742. doi: 10.1212/wnl.56.6.737. [DOI] [PubMed] [Google Scholar]
  5. Alemán-Gómez Y, Melie-García L, Valdés-Hernandez P. IBASPM: toolbox for automatic parcellation of brain structures. Paper presented at Organization for Human Brain Mapping meeting; June 11–15, 2006; Florence, Italy. 2006. [Google Scholar]
  6. Brew BJ. Evidence for a change in AIDS dementia complex in the era of highly active antiretroviral therapy and the possibility of new forms of AIDS dementia complex. AIDS. 2004;18(Suppl 1):S75–S78. [PubMed] [Google Scholar]
  7. Burnham KP, Anderson DR. Model selection and multimodel inference: a practical-theoretic approach. 2nd ed. New York: Springer; 2002. [Google Scholar]
  8. Broom KA, Anthony DC, Lowe JP, Griffin JL, Scott H, Blamire AM, Styles P, Perry VH, Sibson NR. MRI and MRS alterations in the preclinical phase of murine prion disease: association with neuropathological and behavioural changes. Neurobiol Dis. 2007;26:07–717. doi: 10.1016/j.nbd.2007.04.001. [DOI] [PubMed] [Google Scholar]
  9. Brew BJ, Crowe SM, Landay A, Cysique LA, Guillemin G. Neurodegeneration and ageing in the HAART era. J Neuroimmune Pharmacol. 2009;4:163–174. doi: 10.1007/s11481-008-9143-1. [DOI] [PubMed] [Google Scholar]
  10. Binesh N, Huda A, Thomas MA, Wyckoff N, Bugbee M, Han S, Rasgon N, Davanzo P, Sayre J, Guze B, Martin P, Fawzy F. Hepatic encephalopathy: a neurochemical, neuroanatomical, and neuropsychological study. J Appl Clin Med Phys. 2006;7:86–96. doi: 10.1120/jacmp.v7i1.2151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Brew BJ, Rosenblum M, Cronin K, Price RW. AIDS dementia complex and HIV-1 brain infection: clinical-virological correlations. Ann Neurol. 1995;38:563–570. doi: 10.1002/ana.410380404. [DOI] [PubMed] [Google Scholar]
  12. Chang L, Ernst T, Leonido-Yee M, Walot I, Singer E. Cerebral metabolite abnormalities correlate with clinical severity of HIV-1 cognitive motor complex. Neurology. 1999;52:100–108. doi: 10.1212/wnl.52.1.100. [DOI] [PubMed] [Google Scholar]
  13. Chang L, Ernst T, Tornatore C, Aronow H, Melchor R, Walot I, Singer E, Cornford M. Metabolite abnormalities in progressive multifocal leukoencephalopathy by proton magnetic resonance spectroscopy. Neurology. 1997;48:836–845. doi: 10.1212/wnl.48.4.836. [DOI] [PubMed] [Google Scholar]
  14. Chang L, Ernst T, Witt MD, Ames N, Gaiefsky M. Relationships among brain metabolites, cognitive function, and viral loads in antiretroviral-naive HIV patients. NeuroImage. 2002;17:1638–1648. doi: 10.1006/nimg.2002.1254. [DOI] [PubMed] [Google Scholar]
  15. Chang L, Ernst T, Witt MD, Ames N, Walot I, Jovicich J, DeSilva M, Trivedi N, Speck O, Miller EN. Persistent brain abnormalities in antiretroviral-naive HIV patients 3 months after HAART. Antivir Ther. 2003;8:17–26. [PubMed] [Google Scholar]
  16. Cohen RA, Harezlak J, Schifitto G, et al. Effects of nadir CD4 count and duration of human immunodeficiency virus infection on brain volumes in the highly active antiretroviral therapy era. J NeuroVirol. 2010;16:25–32. doi: 10.3109/13550280903552420. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Chang L, Lee PL, Yiannoutsos CT, Ernst T, Marra CM, Richards T, Kolson D, Schifitto G, Jarvik JG, Miller EN, Lenkinski R, Gonzalez G, Navia BA. A multicenter in vivo proton-MRS study of HIV-associated dementia and its relationship to age. NeuroImage. 2004;23:1336–1347. doi: 10.1016/j.neuroimage.2004.07.067. [DOI] [PubMed] [Google Scholar]
  18. Gropman AL, Batshaw ML. Cognitive outcome in urea cycle disorders. Mol Genet Metab. 2004;81(Suppl 1):S58–S62. doi: 10.1016/j.ymgme.2003.11.016. [DOI] [PubMed] [Google Scholar]
  19. Gongvatana A, Schweinsburg BC, Taylor MJ, Theilmann RJ, Letendre SL, Alhassoon OM, Jacobus J, Woods SP, Jernigan TL, Ellis RJ, Frank LR, Grant I. White matter tract injury and cognitive impairment in human immunodeficiency virus-infected individuals. J NeuroVirol. 2009;15:187–195. doi: 10.1080/13550280902769756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Heindel WC, Jernigan TL, Archibald SL, Achim CL, Masliah E, Wiley CA. The relationship of quantitative brain magnetic resonance imaging measures to neuropathologic indexes of human immunodeficiency virus infection. Arch Neurol. 1994;51:1129–1135. doi: 10.1001/archneur.1994.00540230067015. [DOI] [PubMed] [Google Scholar]
  21. Jernigan TL, Archibald S, Hesselink JR, Atkinson JH, Velin RA, McCutchan JA, Chandler J, Grant I. Magnetic resonance imaging morphometric analysis of cerebral volume loss in human immunodeficiency virus infection. The HNRC Group. Arch Neurol. 1993;50:250–255. doi: 10.1001/archneur.1993.00540030016007. [DOI] [PubMed] [Google Scholar]
  22. Jones RS, Waldman AD. 1H-MRS evaluation of metabolism in Alzheimer’s disease and vascular dementia. Neurol Res. 2004;26:488–495. doi: 10.1179/016164104225017640. [DOI] [PubMed] [Google Scholar]
  23. Kreis R, Arcinue E, Ernst T, Shonk TK, Flores R, Ross BD. Hypoxic encephalopathy after near-drowning studied by quantitative 1H-magnetic resonance spectroscopy. J Clin Invest. 1996;97:1142–1154. doi: 10.1172/JCI118528. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lentz MR, Kim WK, Lee V, Bazner S, Halpern EF, Venna N, Williams K, Rosenberg ES, Gonzalez RG. Changes in MRS neuronal markers and T cell phenotypes observed during early HIV infection. Neurology. 2009;72:1465–1472. doi: 10.1212/WNL.0b013e3181a2e90a. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Long LL, Li XR, Huang ZK, Jiang YM, Fu SX, Zheng W. Relationship between changes in brain MRI and (1)H-MRS, severity of chronic liver damage, and recovery after liver transplantation. Exp Biol Med (Maywood) 2009;234:1075–1085. doi: 10.3181/0903-RM-118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lee PL, Yiannoutsos CT, Ernst T, Chang L, Marra CM, Jarvik JG, Richards TL, Kwok EW, Kolson DL, Simpson D, Tang CY, Schifitto G, Ketonen LM, Meyerhoff DJ, Lenkinski RE, Gonzalez RG, Navia BA. A multi-center 1H MRS study of the AIDS dementia complex: validation and preliminary analysis. J Magn Reson Imaging. 2003;17:625–633. doi: 10.1002/jmri.10295. [DOI] [PubMed] [Google Scholar]
  27. McArthur JC. HIV dementia: an evolving disease. J Neuroimmunol. 2004;157:3–10. doi: 10.1016/j.jneuroim.2004.08.042. [DOI] [PubMed] [Google Scholar]
  28. Provencher SW. Automatic quantitation of localized in vivo 1H spectra with LC Model. NMR Biomed. 2001;14:260–264. doi: 10.1002/nbm.698. [DOI] [PubMed] [Google Scholar]
  29. Paul RH, Ernst T, Brickman AM, Yiannoutsos CT, Tate DF, Cohen RA, Navia BA. Relative sensitivity of magnetic resonance spectroscopy and quantitative magnetic resonance imaging to cognitive function among nondemented individuals infected with HIV. J Int Neuropsychol Soc. 2008;14:725–733. doi: 10.1017/S1355617708080910. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Patel SH, Inglese M, Glosser G, Kolson DL, Grossman RI, Gonen O. Whole-brain N-acetylaspartate level and cognitive performance in HIV infection. AJNR Am J Neuroradiol. 2003;24:1587–1591. [PMC free article] [PubMed] [Google Scholar]
  31. Paul RH, Yiannoutsos CT, Miller EN, Chang L, Marra CM, Schifitto G, Ernst T, Singer E, Richards T, Jarvik GJ, Price R, Meyerhoff DJ, Kolson D, Ellis RJ, Gonzalez G, Lenkinski RE, Cohen RA, Navia BA. Proton MRS and neuropsychological correlates in AIDS dementia complex: evidence of subcortical specificity. J Neuropsychiatry Clin Neurosci. 2007;19:283–292. doi: 10.1176/jnp.2007.19.3.283. [DOI] [PubMed] [Google Scholar]
  32. Ross B, Bluml S. Magnetic resonance spectroscopy of the human brain. Anat Rec. 2001;265:54–84. doi: 10.1002/ar.1058. [DOI] [PubMed] [Google Scholar]
  33. Ratai EM, Pilkenton SJ, Greco JB, Lentz MR, Bombardier JP, Turk KW, He J, Joo CG, Lee V, Westmoreland S, Halpern E, Lackner AA, Gonzalez RG. In vivo proton magnetic resonance spectroscopy reveals region specific metabolic responses to SIV infection in the macaque brain. BMC Neurosci. 2009;10:63. doi: 10.1186/1471-2202-10-63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Sijens PE, Alkefaji H, Lunsing RJ, van Spronsen FJ, Meiners LC, Oudkerk M, Verkade HJ. Quantitative multivoxel 1H MR spectroscopy of the brain in children with acute liver failure. Eur Radiol. 2008;18:2601–2609. doi: 10.1007/s00330-008-1049-z. [DOI] [PubMed] [Google Scholar]
  35. Sacktor N, McDermott MP, Marder K, Schifitto G, Selnes OA, McArthur JC, Stern Y, Albert S, Palumbo D, Kieburtz K, De Marcaida JA, Cohen B, Epstein L. HIV-associated cognitive impairment before and after the advent of combination therapy. J NeuroVirol. 2002;8:136–142. doi: 10.1080/13550280290049615. [DOI] [PubMed] [Google Scholar]
  36. Sacktor N, Skolasky RL, Ernst T, Mao X, Selnes O, Pomper MG, Chang L, Zhong K, Shungu DC, Marder K, Shibata D, Schifitto G, Bobo L, Barker PB. A multicenter study of two magnetic resonance spectroscopy techniques in individuals with HIV dementia. J Magn Reson Imaging. 2005;21:325–333. doi: 10.1002/jmri.20272. [DOI] [PubMed] [Google Scholar]
  37. Schweinsburg BC, Taylor MJ, Alhassoon OM, Gonzalez R, Brown GG, Ellis RJ, Letendre S, Videen JS, McCutchan JA, Patterson TL, Grant I. Brain mitochondrial injury in human immunodeficiency virus-seropositive (HIV+) individuals taking nucleoside reverse transcriptase inhibitors. J NeuroVirol. 2005;11:356–364. doi: 10.1080/13550280591002342. [DOI] [PubMed] [Google Scholar]
  38. Schifitto G, Yiannoutsos CT, Ernst T, Navia BA, Nath A, Sacktor N, Anderson C, Marra CM, Clifford DB. Selegiline and oxidative stress in HIV-associated cognitive impairment. Neurology. 2009a;73:1975–1981. doi: 10.1212/WNL.0b013e3181c51a48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schifitto G, Zhong J, Gill D, Peterson DR, Gaugh MD, Zhu T, Tivarus M, Cruttenden K, Maggirwar SB, Gendelman HE, Dewhurst S, Gelbard HA. Lithium therapy for human immunodeficiency virus type 1-associated neurocognitive impairment. J NeuroVirol. 2009b;15:176–186. doi: 10.1080/13550280902758973. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Thompson PM, Dutton RA, Hayashi KM, Toga AW, Lopez OL, Aizenstein HJ, Becker JT. Thinning of the cerebral cortex visualized in HIV/AIDS reflects CD4+ T lymphocyte decline. Proc Natl Acad Sci U S A. 2005;102:15647–15652. doi: 10.1073/pnas.0502548102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Tkac I, Dubinsky JM, Keene CD, Gruetter R, Low WC. Neurochemical changes in Huntington R6/2 mouse striatum detected by in vivo 1H NMR spectroscopy. J Neurochem. 2007;100:1397–1406. doi: 10.1111/j.1471-4159.2006.04323.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, Mazoyer B, Joliot M. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. NeuroImage. 2002;15:273–289. doi: 10.1006/nimg.2001.0978. [DOI] [PubMed] [Google Scholar]
  43. Tate D, Harezlak J, Price J, Dewey J, Yiannoutsos C, Campbell T, Daar E, McMahon D, Schifitto G, Zhong J, Taylor M, Singer E, Alger J, Brown M, Cohen R, Navia B. Human Brain Mapping. Barcelona, Spain: 2010. [June 6–10, 2010]. Progressive brain atrophy in chronic HIV infection: longitudinal analysis of the HIVNC Cohort Study. [Google Scholar]
  44. Tarasow E, Wiercinska-Drapalo A, Kubas B, Dzienis W, Orzechowska-Bobkiewicz A, Prokopowicz D, Walecki J. Cerebral MR spectroscopy in neurologically asymptomatic HIV-infected patients. Acta Radiol. 2003;44:206–212. doi: 10.1080/j.1600-0455.2003.00028.x. [DOI] [PubMed] [Google Scholar]
  45. Valcour V, Sacktor N. HIV-associated dementia and aging. J Ment Health Aging. 2002;8:295–306. [Google Scholar]
  46. Valcour V, Shikuma C, Waters M, Sacktor N. Cognitive impairment in older HIV-1 seropositive individuals: prevalence and potential mechanisms. AIDS. 2004;18(Suppl 1):S79–S86. doi: 10.1097/00002030-200401001-00012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Yiannoutsos CT, Ernst T, Chang L, Lee PL, Richards T, Marra CM, Meyerhoff DJ, Jarvik JG, Kolson D, Schifitto G, Ellis RJ, Swindells S, Simpson DM, Miller EN, Gonzalez RG, Navia BA. Regional patterns of brain metabolites in AIDS dementia complex. NeuroImage. 2004;23:928–935. doi: 10.1016/j.neuroimage.2004.07.033. [DOI] [PubMed] [Google Scholar]

RESOURCES