|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to characterize brain volumetric differences in HIV seropositive and seronegative men and to determine effects of age, cardiovascular risk, and HIV infection on structural integrity.
Magnetic resonance imaging was used to acquire high-resolution neuroanatomic data in 160 men aged 50 years and over, including 84 HIV seropositive and 76 seronegative controls. Voxel-based morphometry was used to derive volumetric measurements at the level of the individual voxel. Data from a detailed neuropsychological test battery were recombined into four summary scores representing psychomotor speed, visual memory, verbal memory, and verbal fluency.
Both age and HIV status had a significant effect on both gray matter (GM) and white matter (WM) volume. The age-related GM atrophy was primarily in the superior temporal and inferior frontal regions; the HIV-related GM loss included the posterior and inferior temporal lobes, the parietal lobes, and the cerebellum. Among all subjects, the performance on neuropsychological tests, as indexed by a summary variable, was related to the volume of both the GM and WM. Contrary to our predictions, the CVD variables were not linked to brain volume in statistically adjusted models.
In the post-HAART era, having HIV infection is still linked to atrophy in both GM and WM. Secondly, advancing age, even in this relatively young cohort, is also linked to changes in GM and WM volume. Thirdly, CNS structural integrity is associated with overall cognitive functions, regardless of the HIV infection status of the study volunteers.
The proportion of persons living with HIV/AIDS aged 50 years and older rose to 24% of all cases in 2005, up from 17.1% in 2001 (http://www.cdc.gov/hiv/topics/over50/resources/factsheets/over50.htm). While the total number of infected persons rose 20% in that time period, the increase in individuals over 50 was 58%. Factors normally associated with age-related neuropsychiatric syndromes may play an increasingly important role in understanding central nervous system (CNS) dysfunction occurring in older HIV + patients. Indeed, age-associated medical comorbidities are significant risk modifiers for HIV-associated neurocognitive disorder (HAND) [1–7]. There may be a differential effect of the presence of the APOE*4 allele in older HIV subjects , and diabetes is a critical comorbidity . The use of highly active anti-retroviral therapy (HAART) may be associated with abnormal amyloid deposition in the brain , and amyloid-beta and tau found in the cerebrospinal fluid may be related to HIV-associated dementia .
The Multicenter AIDS Cohort Study (MACS) has compared 345 asymptomatic HIV-infected men and 237 controls on neuropsychological test performance on Symbol-Digit Modalities and Trailmaking Tests over a 5-year period . There was no evidence of differences between these groups in either performance, or in year-to-year decline during the observation period. These findings indicate that psychomotor speed, a very sensitive measure of HIV-associated cognitive loss, is preserved over many years among HIV-infected individuals with controlled viremia.
We have also analyzed effects of cardiovascular factors on cognitive function in a cross-sectional study of 428 infected men and 207 uninfected controls aged 40 years and older . Carotid intima-media thickness and estimated glomerular filtration rate were significantly associated with slower psychomotor speed. HIV serostatus, however, was not associated with psychomotor speed in a covariate-adjusted model. There was no association between test performance and detectable HIV RNA, CD4+ cell counts, or AIDS in the HIV+ individuals. Abnormal coronary artery calcification, however, was a risk factor for poorer performance on memory tasks. Among the HIV-infected individuals only, the presence of detectable HIV RNA in plasma significantly increased risk of poorer memory performance.
The associations between brain structure and cognition in HIV infection have been reviewed by Paul  and by Thurnher . In addition to the well-documented findings of subcortical brain abnormalities, there is increasing evidence that the neocortex is injured in HIV infection [16, 17]. AIDS patients exhibit severe, selective gray matter thinning (10–15%) in a broad anatomic area that includes primary sensory and motor cortices of both hemispheres  and prefrontal and parietal cortical thinning were correlated with neuropsychological impairment. Loss of CNS structural integrity was also reflected in a 24% increase in ventricular volume in infected subjects, and these measurements also correlated with neuropsychological impairment. The volumes of the frontal horns provided good between-group discrimination . A recent study  demonstrated a loss of gray matter secondary to HIV disease in the anterior cingulate and temporal cortices, with a loss of midbrain white matter, as well.
This research report presents the findings of a voxel-based morphometry (VBM) analysis of the brain structural images of HIV-infected and seronegative participants of the MACS . We hypothesized that cardiovascular variables would be associated with increased neurological risk, as measured by effects on measures of both brain structure and cognition. We predicted that HIV-related effects in this HAART era sample would be attenuated relative to pre-HAART observations once cardiovascular risk variables were adjusted for in multivariate analyses.
This study was approved by the ethical standards committee on human experimentation at each of the MACS sites. Written informed consent was obtained from all participants prior to their undergoing research procedures.
The MACS is a four-site study of the natural and treated history of HIV infection among men who have sex with men. Volunteers were enrolled in three waves: 1984/1985, 1987/1990, and in 2001/2003 (primarily from racial/ethnic minorities). The MACS has tracked cognitive functions among the study participants for the past 24 years using screening tools (Trail Making, Symbol-Digit Substitution Tests) and has followed a sub-cohort with more detailed testing for approximately 20 years.
The MACS cardiovascular disease (CVD) substudy included men, age ≥40 years, with no self-reported history of heart disease (heart attack, heart surgery, and other heart illness) or cerebrovascular disease, and weight <300 lb ; the baseline visit was completed between April 2004 and January 2006. In 2007, each site conducted magnetic resonance imaging (MRI) examinations of 40 MACS participants (total N=160), including high-resolution anatomic sequences.
The men who were approached for participation in the MRI study were all part of the CVD substudy and were matched on a 1:1 basis as a function of age, education and race (Caucasian vs. Other). At the Chicago and Baltimore sites it was possible to create 20 pairs of subjects (HIV±) within site. However, because of the distribution of cases in Los Angeles and Pittsburgh, these sites had to match cases across the centers. The resulting distribution of cases by serostatus thus differed by site. Demographic characteristics of the subjects are shown in Table 1.
Beginning in 2005 the MACS instituted a new schedule of assessing cognitive function within the entire study cohort. Between 2005 and 2007, all MACS participants completed a standard neuropsychological test battery; those subjects who scored within normal limits were scheduled to return bi-annually. Those subjects whose performance fell below normal limits are retested on a semi-annual basis. In addition to these tests, each of the volunteers who were enrolled in this study completed additional tests which more closely conform to those mentioned in the revised research criteria for HAND . All of the neuropsychological test data were sent to the Pittsburgh site where the raw scores from the tests were either transformed into demographically adjusted T scores , or to standard scores derived from published norms.
The MRI scanning sequences were taken from the protocol developed by the Alzheimer’s disease neuroimaging initiative (ADNI) for use with scanners with 3 Tesla field strengths (http://www.adni-info.org/images/stories//mritrainingmanualv1.pdf). Three of the sites housed a Siemens 3 T Trio scanner (maximum gradient slew rate: 200 mT/m/s; maximum gradient strength 40mT/m), with the Siemens phase-array head coil. One of the sites (Los Angeles) used a Siemens Allegra scanner. The sequences were (in order): localizer scan, transversal (axial) proton density, transversal (axial) T2-weighted, transversal (axial) FLAIR, coronal MP-RAGE (8–10 min), and transversal (axial) diffusion tensor imaging. The sagittal MP-RAGE sequence used for this analysis was: FOV=256 mm; slices= 160; TR=2,300 ms; TE=2.91 ms; TI=900 ms; flip angle=9 °; and thickness=1.2 mm. Only the data from the MP-RAGE will be described in this report.
In order to standardize the acquisition of the data, the subjects were required to remove any dentures, hair clips, combs, earrings, necklaces, etc. and to remove all upper body clothing with metallic trim, such as zippers, buttons, or embroideries that could cause artifact. In order to assist with the verification of scan orientation, a fiducial marker (vitamin E or fish oil capsule) was taped to the subjects’ right temple. A procedures manual, modeled after that of ADNI, was distributed to the technologists and investigators at each site. All of the de-identified MRI data were transferred from each site to a central location at the University of Pittsburgh. They were subsequently stored on local hard disks with the first author of this report. A copy of all of the MRI data was also sent to the central data repository in Baltimore.
Subclinical CVD was assessed using electron beam tomography or multidetector computed tomography to measure coronary artery calcium and ultrasound examination of the carotid artery to measure carotid IMT, plaque, and stiffness/distensibility. Laboratory measures included total cholesterol, low and high density lipoproteins, glucose, insulin, glycosylated hemoglobin, and standardized blood pressure and heart rate measures. GFR was estimated using a standard protocol. The CVD variables are shown in Table 1.
The CVD variables were reduced to categorical variables (present/absent, normal/abnormal) based on standard criteria or the distribution of values within the HIV− group . These included the presence of hypertension (resting BP >130/90, or self-report of HTN, or use of anti-hypertensive medications) and diabetes (self-report or use of anti-diabetic medications). The CES-D score was calculated for each subject, and those scoring above 16 were classified as “depressed” for the purpose of this analysis. Education was classified as high school or less, 13–15 years, and college or greater. We classified each participant with reference to current use, or any use within the past 5 years of illicit drugs.
The neuropsychological test data were reduced to z scores based on the distribution of the entire study sample. The data were then recombined into four summary scores representing psychomotor speed (Symbol-Digit Substitution, Trailmaking Part A), Visual Memory (Recall of Visual Reproductions), Verbal Memory (Logical Memory Recall), and Verbal Fluency (letter and category word generation). The signs were adjusted in each case so that positive scores represented better performance. A summary score was computed by taking the mean of the four composite variables (see Table 1).
The MRI data were first processed through a non-parametric non-uniform intensity normalization  to reduce between scan and between-site differences in the images. This was followed by a bias correction (within SPM2) in order to help improve spatial registration. We then used a recursive implementation of the brain extraction tool  from the FMRIB Software Library (http://www.fmrib.ox.ac.uk/fsl/) to strip off the skull and other extraneous tissue.
We created a normal template image using the seronegative control subjects, and estimated the prior probabilities of each tissue class (i.e., gray, white, and CSF) for use in the segmentation routines. We then used the VBM2 script (http://dbm.neuro.uni-jena.de/vbm/vbm2-for-spm2/) for normalization and segmentation of the data. The resulting maps of gray matter had been modulated to render the values in each of the 1×1×1 mm voxels as a volume, which were then smoothed using a 10×10×10 mm Gaussian filter to reduce the effects of registration error and render the data more amenable to parametric analysis. At this stage, 12 of the initial 160 scans (7.5%) did not pass our quality control checks, and these subjects were excluded from further analysis.
For all of the VBM analyses, total intra-cranial volume was entered as a covariate. In addition, because the subjects were scanned at different sites (and on two different model scanners), we also entered four binary dummy variables to adjust for possible between-site differences. For all analyses, the default threshold for reporting statistical significance was set at a False discovery rate of P<.05, with an extent threshold of 100 voxels . The mean whole brain image created from the seronegative control subjects was used to project all study findings.
There are three main findings from this analysis. Firstly, age has a significant effect on both GM and WM volume at the voxel level. Secondly, HIV disease had an independent effect on GM and WM volume at the voxel level. Thirdly, among all subjects, the performance on neuropsychological tests, as indexed by a summary variable, was related to the volume of both the GM and WM. Contrary to our predictions, the CVD variables were not linked to brain volume in statistically adjusted models.
Figure 1 shows the results of the VBM analysis of gray matter projected onto a single-subject template of the cortical surface. The effects of age are shown in the right-hand columns, and the effects of HIV disease in the left-hand columns. Both variables were entered into the same analysis simultaneously, with total intra-cranial volume and study site as covariates. There was no significant interaction between age and HIV group. The age-related GM atrophy is focused primarily in the superior temporal and inferior frontal regions, with additional tissue loss in the medial temporal and cingulate cortices (cf. ). By contrast, the HIV-related GM loss was more widely distributed. It includes the posterior and inferior temporal lobes (right > left), the parietal lobes, and the cerebellum (see Fig. 1).
White matter atrophy was significantly associated with advancing age (see Fig. 2), but much less so with HIV status. Older subjects, regardless of their serostatus had atrophy in the peri-ventricular white matter and frontal and temporal regions.
We then analyzed the associations between the cognitive summary score and GM and WM volumes. Instead of entering HIV status into the SPM model, we instead entered the Global Impairment Rating for each participant and the resulting analysis was similar to a multiple regression model, with total intra-cranial volume and study site as covariates. We found a significant link between GM and test performance in the right temporal lobes, predominantly. Consequently, we created a regional mask using PickAtlas (http://www.fmri.wfubmc.edu/cms/software#PickAtlas) and focused our analysis on the right temporal pole (i.e., we excluded the remainder of the brain from the search space). We found extensive regions of the anterior temporal lobe that were linked to performance on the neuropsychological tests, as indexed by the cognitive summary score (see Fig. E-1 in the Electronic supplementary materials). Analysis of WM also revealed that atrophy was associated with lower scores on the neuropsychological tests (see Fig. 3). The regions were in the same WM areas as those observed for the aging effect, but were less extensive.
We tested the effects of time since infection, CD4+ cell counts, and viral load on GM and WM integrity among the seropositive men in the study. None of the effects was significant at an FDR of p<.05. The same was true when we tested the effects of drug use, diabetes, and hypertension on structural integrity among all study participants.
These data make several important points. Firstly, in the post-HAART era, having HIV infection is still linked to atrophy in GM, and to a lesser extent WM. Secondly, advancing age, even in this relatively young cohort, is also linked to changes in GM and WM volume. Thirdly, CNS structural integrity is associated with overall cognitive functions, regardless of the HIV infection status of the study volunteers.
Our findings are fully consistent with the accumulating data that the neocortex is affected in HIV disease, even among individuals with mild or no cognitive dysfunction . We have previously identified, in a separate non-MACS study, those brain regions that were selectively vulnerable to HIV disease by computing 3D maps of cortical gray matter thickness . Prefrontal and parietal cortical thinning in both brain hemispheres was linked with neuropsychological impairment. This loss of CNS structural integrity was also reflected in an expansion of the lateral ventricles, estimated as a 24% increase in size of the ventricles in the HIV-infected participants [19, 29].
Chiang and colleagues  used tensor-based morphometry to visualize the brain regional atrophy in that same group of HIV-infected individuals. Significant atrophy was found bilaterally in the primary and association sensorimotor areas, basal and medial frontal lobes (~15–20% deficit), corpus callosum, cingulum, putamen, globus pallidus, and thalamus (~10–15% deficit). Atrophy of these regions, particularly in the white matter, correlated with cognitive impairment and CD4+ cell counts. Leporé  developed a more sensitive method of analysis of these data, retaining the full deformation tensors and applying a manifold version of Hotelling’s test. With this approach, consistent but more extensive patterns (relative to the standard methods) of structural abnormalities were detected.
We  and others [17, 20] have shown that the extent of GM atrophy in HIV-infected individuals is related to the duration of infection. One interpretation of these findings is that it is due to the “legacy effect”—that is, the individuals with long-term HIV infection (like many of the men in this study) were also the ones with no exposure to treatment, use of only monotherapy, and use of HAART only after significant immune suppression and rampant viral replication. HIV viral proteins are neurotoxic, the GM and WM volumetric differences may reflect neuronal loss, reduced dendritic complexity, synaptic loss , and associated white matter degeneration. Because HAART (and other) medications often fail to significantly permeate the blood brain barrier , brain degeneration still may proceed even in patients treated with anti-retroviral medication. Thus, we may be seeing the residual effects of uncontrolled HIV infection.
The fact that we did not see extensive WM changes as a consequence of HIV infection is also consistent with the more recent neuropathological findings that white matter pallor and demyelination are now much less common than in the early days of the epidemic in the USA. However, prior to the widespread use of HAART, structural MRI studies consistently revealed the presence of white matter lucencies . White matter abnormalities could be diffuse or focal, and seen as signal increases on T2-weighted images , by T1 relaxation times, or using diffusion tensor imaging [37, 38].
One question that arises from this analysis that is particularly important for clinical neuroradiologists as well as researchers is why there was a correlation between volume of white matter and cognitive function. White matter lesions are more common in older individuals (e.g., Fig. 2) and correspond to loss of myelinated axons, gliosis, enlarged perivascular spaces, and small vessel disease [39–41]. White matter damage is also linked to the presence of hypertension, and may precede the subsequent development of stroke and dementia [42, 43]. There is a range of data documenting the relationship between white matter damage and cognition (e.g., [44–47]), and data from the Cardiovascular Health Study suggest that this association is at least partially mediated by damage to the GM (Raji et al., personal communication). Specifically, WM damage can be considered as a marker of small vessel disease, which has an impact on GM integrity and subsequently impairs cognition. In the present case, because markers of subclinical vascular disease were unrelated to GM volume, we would predict (although cannot prove) that the majority of the association between WM atrophy and cognition is due to the direct link, with less of the variance accounted for by the indirect pathways through GM damage secondary to small vessel disease. However, this will need to be tested directly, preferably with longitudinal imaging data.
The results of this analysis build on the growing body of evidence that among long-term HIV-infected individuals, control of viral replication and of immunological competence appears to protect against the CNS abnormalities and cognitive dysfunction that were so common prior to the advent of HAART. Although HIV disease still has an impact on brain structure, the effects appear to be less extensive than those associated with age. As the risk factors for HAND continue to change , and we develop a better understanding of the multiple factors that can affect brain structure and function, being able to place HAND in the context of normal age-related processes becomes increasingly important.
The authors are grateful to the volunteers and the staff of the Multicenter AIDS Cohort Study for the additional time and effort that they contributed towards the successful completion of this project. This study was supported in part by funds from the National Institute for Allergy and Infectious Diseases to the collaborating MACS sites: UO1-AI-35042, 5-MO1-RR-00052 (GCRC), UO1-AI-35043, UO1-AI-37984, UO1-AI-35039, UO1-AI-35040, UO1-AI-37613, and UO1-AI-35041.
Conflict of interest Dr. Miller is the author of the CalCAP reaction time program and has a financial interest in this software.
The Multicenter AIDS Cohort Study (MACS) includes the following: Baltimore: The Johns Hopkins University Bloomberg School of Public Health: Joseph B. Margolick (Principal Investigator), Haroutune Armenian, Barbara Crain, Adrian Dobs, Homayoon Farzadegan, Joel Gallant, John Hylton, Lisette Johnson, Shenghan Lai, Ned Sacktor, Ola Selnes, James Shepard, Chloe Thio. Chicago: Howard Brown Health Center, Feinberg School of Medicine, Northwestern University, and Cook County Bureau of Health Services: John P. Phair (Principal Investigator), Joan S. Chmiel (Co-Principal Investigator), Sheila Badri, Bruce Cohen, Craig Conover, Maurice O’Gorman, David Ostrow, Frank Palella, Daina Variakojis, Steven M. Wolinsky. Los Angeles: University of California, UCLA Schools of Public Health and Medicine: Roger Detels (Principal Investigator), Barbara R. Visscher (Co-Principal Investigator), Aaron Aaronow, Robert Bolan, Elizabeth Breen, Anthony Butch, Thomas Coates, Rita Effros, John Fahey, Beth Jamieson, Otoniel Martínez-Maza, Eric N. Miller, John Oishi, Paul Satz (deceased), Harry Vinters, Dorothy Wiley, Mallory Witt, Otto Yang, Stephen Young, Zuo Feng Zhang. Pittsburgh: University of Pittsburgh, Graduate School of Public Health: Charles R. Rinaldo (Principal Investigator), Lawrence A. Kingsley (Co-Principal Investigator), James T. Becker, Ross D. Cranston, Jeremy J. Martinson, John W. Mellors, Anthony J. Silvestre, Ronald D. Stall. Data Coordinating Center: The Johns Hopkins University Bloomberg School of Public Health: Lisa P. Jacobson (Principal Investigator), Alvaro Munoz (Co-Principal Investigator), Stephen R. Cole, Christopher Cox, Gypsyanber D’Souza, Stephen J. Gange, Janet Schollenberger, Eric C. Seaberg, Sol Su. NIH: National Institute of Allergy and Infectious Diseases: Robin E. Huebner; National Cancer Institute: Geraldina Dominguez; National Heart, Lung and Blood Institute: Cheryl McDonald; National Institute of Mental Health: Pim Brouwers.
James T. Becker, Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. Department of Psychology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. Neuropsychology Research Program, Suite 830, 3501 Forbes Avenue, Pittsburgh, PA 15213, USA.
Victoria Maruca, Spalding University, Louisville, KY, USA.
Lawrence A. Kingsley, Department of Infectious Diseases and Microbiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA.
Joanne M. Sanders, Department of Epidemiology, Bloomberg School of Public Health, The Johns Hopkins University, Baltimore, MD, USA.
Jeffery R. Alger, Department of Neurology, University of California Los Angeles, Los Angeles, CA, USA.
Peter B. Barker, Department of Radiology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Karl Goodkin, Neuropsychiatric Institute, University of California Los Angeles, Los Angeles, USA.
Eileen Martin, Department of Psychiatry, University of Illinois at Chicago, Chicago, IL, USA.
Eric N. Miller, Neuropsychiatric Institute, University of California Los Angeles, Los Angeles, CA, USA.
Ann Ragin, Department of Neurology, Feinberg School of Medicine, Northwestern University, Chicago, USA.
Ned Sacktor, Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Ola Selnes, Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD, USA.