HIV and the brain: longitudinal results from CHARTER and other studies
Nathan Geffen, CSSR
The purpose of the CNS HIV Anti-Retroviral Therapy Effects Research (CHARTER) study is to research how central and peripheral nervous system HIV-related complications are affected by ART. 
The study has been running since 2002 with several sites in the United States comprising about 1,600 patients. The CHARTER group presented numerous studies at CROI 2012.  Those studies that appear to have the most direct impact on diagnosing, predicting or preventing neurocognitive decline are discussed here.
Asymptomatic neurocognitive impairment (ANI) is associated with worse functional outcomes
Blackstone and colleagues assessed 578 participants with a comprehensive neuropsychological battery, self-report questionnaires of cognitive complaints and everyday functioning, as well as performance-based tasks measuring employment capacity and medication management. Of these 375 patients were classified as normal and 175 met criteria for ANI. A further 40 met criteria for symptomatic HIV associated neurocognitive disorder (HAND) of whom 14 had HIV-associated dementia.
The symptomatic HAND group reported significantly more symptoms of depression and had lower current CD4 counts. After controlling for current CD4 and depression symptoms, the ANI participants had worse employment capacity than the normal participants, but were comparable to the HAND group (p <0.001 for both comparisons). There were no between-group differences on the test of medication management. The researchers concluded their study suggests ANI is a less benign condition than is widely perceived. They indicate that their findings are consistent with research showing that mild neuropsychological impairment is associated with worse functional outcomes. They further suggest that performance-based tests of everyday functioning should be incorporated into the diagnosis of HAND. 
Longitudinal study examining neurocognitive impairment
Heaton and colleagues followed neurocognitive function in HIV-positive patients measured by comprehensive laboratory, neuromedical, and neurobehavioral assessments every 6 months, over 18 to 42 months. 
The study reported that 99 (22.7%) participants experienced neurocognitive decline, 266 (61%) remained stable, and 72 (16.5%) improved over 18 to 42 months. However, the only predictors in the multivariate Co regression analysis were having a confounding co-morbidity (RR 2.4; 95%CI 1.4, 4.0; p=0.0015), being off ART (RR 1.6; 95%CI 1.1, 2.5; p=0.025) and low CD4 count (RR 1.1: 95%CI 1.02, 1.21; p=0.017.
In another study by Heaton on what appears to be the same cohort, patients with ANI (n=84) and mild neurocognitive disorders (MND) (n=57) were more likely to experience statistically meaningful decline than neurocognitively normal patients (n=246) (23%, 30% vs 13%; p=0.004). Also ANI patients were less likely to improve than the neurocognitively normal group (7% vs 21%; p=0.008). These results were used to validate the ANI and MND as clinically important factors associated with reduced cognitive function over time and is consistent with Blackwell’s findings. 
Biomarkers for predicting neurocognitive performance
CHARTER researchers have been looking for biomarkers that predict cognitive decline. Cerebrospinal fluid (CSF) was collected from nearly 350 study participants, of whom 85% were on ART. Follow-up samples were collected within approximately a year for 70% of subjects. There were no associations at baseline between biomarkers and neurocognitive impairment. However interesting associations were found after follow-up. A lower baseline ratio of sphingomyelin to ceramide predicted a decline in neurocognitive performance (p=0.047). Sphingomyelin is a lipid that mainly occurs in nerve tissue. Ceramide is a lipid that occurs in large concentrations in cells and is one of the components of sphingomyelin. The authors note that the predictive potential of this ratio appeared to be driven by increases of certain species of ceramide over time. Lower levels of some multiple cholesterol esters were also associated with neurocognitive decline (p ranged from 0.046 to 0.007, depending on the species).
On the other hand, high levels of two triglycerides at baseline predicted neurocognitive improvement (p=0.005 and p=0.006 for the two species). At follow-up these were lower, suggesting they normalised over time. 
Another study by CHARTER researchers tested a panel of biomarkers to predict cognitive impairment. Just under 100 people with HIV were categorised into four groups: stably normal, stably impaired, reliably worsening and reliably improving. All underwent neurocognitive testing, phlebotomy, and lumbar puncture at two time points separated by a median of just over 6 months (IQR 5.6 to 70). The researchers measured CCL2, CXCL10, CX3CL1, CXCL12, IL-6, TNF-alpha, soluble TNF receptors (sTNFR, p75) and sCD14. 74% of patients were on ART at the first time point (median current CD4 of 394 cells/mm3 and median nadir of 110 cells/mm3), of whom 54% had undetectable viral loads in plasma and CSF. A combination of sCD14, CCL2, CXCL10, sTNFR, TNF-alpha predicted neurocognitive status in 92% of patients. Allowing a higher misclassification rate, 20%, meant that TNF-alpha could be removed from the panel.
For patients with normal performance at the first time point, a combination of sCD14, IL-6, CXCL12, CCL2 and sTNFR correctly classified the cognitive status of 94% at the second time point. Allowing for a 20% error rate, sCD14, CXCL12 and IL-6, correctly classified 82%, including all subjects in the stably normal group. For subjects with impaired performance at the first time point, CCL2, TNF-alpha, sCD14 and CX3CL1 classified 96% correctly. CCL2 and TNF-alpha correctly classified 81%, including all people in the stably impaired group.
The two most frequently identified biomarkers were sCD14 and CCL2. These are indicators of monocyte or macrophage activation. All cases of neurocognitive stability were correctly classified. 
Another biomarker study compared neurocognitive status in 34 HIV-positive patients virally suppressed on ART to 34 age-matched HIV-negative controls. Each patient had two visits. Differences between the two cohorts are not reported but one interesting finding was that of 13 subjects who were impaired at the first visit, 10 remained impaired at the second visit, and all but one of the 21 neuropsychologically normal subjects remained normal. Subjects who remained impaired showed little change in their baseline adjusted sCD163 level, while those who remained normal showed a drop in baseline adjusted sCD163 (least squares means: -1.1 versus -280; p=0.056). 
Role of central obesity and diabetes
Another CHARTER sub-study presented by Allen McCutchan and colleagues looked at the relationship between diabetes, obesity and cognitive decline in 130 HIV positive patients. 
Neurocognitive impairments was diagnosed in 40% of participants. Age and longer duration of infection predicted impairment. So did waist circumference but this was only measured in 55 participants. There was no assoiciation with BMI, HOMA score (a predictor of insulin resistance) and leptin levels. Self-reported diabetes was associated with impairment in patients in this sub-study. This contrasts with an analysis of the whole CHARTER cohort which found an association in patients older than 55 only but not patients younger than 55.
Cytomegalovirus (CMV) levels and cognitive decline
Letendre and colleagues studied 138 HIV-positive people to determine associations between CMV levels, neurocognitive characteristics, disease and demographics. CMV antibody concentrations were measured by enzyme-linked immunosorbent assay.
Higher CMV antibody levels were associated with older age (r = 0.23; p = 0.006), lower nadir CD4 cell counts (r = -0.34; p < 0.0001), ART use (p = 0.004), and worse global deficit score (r = 0.17; p = 0.04). For patients not taking ART higher CMV antibody levels were also associated with higher HIV RNA levels in CSF (r = 0.29; p = 0.05) but not in plasma. Multivariate analysis showed that worse global deficit score was associated with higher CMV antibody levels, more co-morbid conditions, and an interaction between CMV antibody level and plasma HIV RNA (p = 0.02).
Analysis of the interaction identified that higher CMV antibody levels were only associated with worse global deficit scores among subjects who had undetectable HIV RNA in plasma.
The authors conclude that higher CMV antibody levels were associated with worse neurocognitive functioning. They suggest their findings have implications for earlier initiation of ART, for the aging of the HIV population, and for the effect of CMV on HIV in the central nervous system. They also say that their findings add to existing data that suggest that CMV prophylaxis may be beneficial. 
Another finding relevant to older age and cognitive functioning in people with HIV comes from a study of over 205 CHARTER patients. These patients provided 162 CSF and 230 plasma samples. Tenofovir CSF (n=44) concentrations increased more steeply with age than plasma (n=44). Efavirenz concentrations increased in CSF (n=66) in patients older than 55 with a less steep and steadier increase with age for plasma (n=77) concentrations. Plasma (n=109) atazanavir concentrations slightly declined with age while CSF (n=58) concentrations remained stable. Higher ARV concentrations were also associated with worse neurocognitive functioning, which the authors note may indicate drug neurotoxicity. They concluded that more data in older HIV-positive people was needed to validate their findings. 
European AIDS Clinical Society (EACS) guidelines for diagnosing HAND not sensitive enough
Ignacio Perez-Valero and colleagues compared the recently released EACS guidelines and the HIV Dementia Scale (HDS) for diagnosing symptomatic and non-symptomatic HAND. 
CHARTER’s comprehensive neurobehavioral assessments that involve several hours of comprehensive testing were used as the gold standard.
|Sensitivity %||Specificity %|
|EACS screen for symptomatic HAND||57||95|
|EACS screen for HAND||15||91|
|HDS screen for symptomatic HAND||52||64|
|HDS screen for HAND||50||73|
While the authors stated that neither EACS nor HDS screens had sufficient sensitivity for detecting cases for referral, concluding that EACS sensitivity is especially poor if the full range of HAND is considered, they failed to consider that the EACS guidelines were established to provide increased awareness for a simple intervention, based not only on the limited time that most doctors have with patients, but also that for many doctors, assessing NCI is not currently a significant aspect of HIV management. These results do not mean that easy to use evaluations that within minutes can clarify, even roughly, the urgency for some referrals, or do not have a place in clinical care.
The effect of HIV on the brain remains an important aspect of care and these data help. While advanced HIV disease causes HAND and dementia, the biological mechanisms are poorly understood. Whether HIV or ARVs contribute to cognitive decline in asymptomatic patients, especially at higher CD4 counts and/or controlled viraemia, with or without ART is unclear.
The results from CHARTER may help predict diagnosis but evidence of sub-clinical changes, while worrying, do not suggest different management, other than perhaps more careful observation.
Higher rates of cognitive problems in HIV positive compared to negative people, even on stable ART, are subject to confounding and the difficulty of an appropriately matched control.
The biomarker studies from CHARTER are interesting, but given the large number of biomarkers that were considered, some of these may be chance associations and their findings still need to be validated.
The study reporting an association between central obesity, diabetes and NCI did not report on the possibility of reverse causality – that NCI may have contributed to poor diet, but this is a US study so the diet may have been regionally normal – or whether a common cause be responsible for NCI and diabetes.
While the authors suggested “avoiding ARV drugs that induce central obesity might protect patients from or reverse neurocognitive impairment” this is easy to say but more complicated to interpret with any degree of precision, given that central lipohypertrophy has been associated with all classes of ARVs and no single ARV has been shown to be clearly protective.
Three key questions remain unanswered: Does pre-AIDS HIV infection significantly affect cognitive functioning? What are the long-term effects of HIV on cognitive functioning? Can earlier ART improve cognitive outcomes in people with HIV?
Hopefully research from the START trial, which has a neurology substudy, will provide data at higher CD4 counts (>500 cells/mm3), together with any impact of earlier ART. The substudy will have 300 recruits in each arm. 
The higher prevalence of impairment in HIV positive people suggests that neurocognitive assessment should be addressed in guidelines and integrated into routine care.
Unless stated otherwise, all references are to the Programme and Abstracts of the 19th Conference on Retroviruses and Opportunistic Infections (CROI), 5-8 March 2012, Seattle.
- CHARTER. Welcome to CHARTER as a resource.
- CHARTER. 2012. CHARTER investigators present at CROI 2012!
- Blackstone K et al. 2012. How “asymptomatic” is HIV-associated asymptomatic neurocognitive impairment? 19th CROI, 2012, Seattle. Poster abstract 497.
- Heaton R et al. 2012. Prevalence and predictors of neurocognitive decline over 18 to 42 Months: a CHARTER longitudinal study. 19th CROI, 2012, Seattle. Poster abstract 474.
- Heaton R et al. 2012. Asymptomatic mild HIV-associated neurocognitive disorder increases risk for future symptomatic decline: a CHARTER longitudinal study. 19th CROI, 2012, Seattle.
- Haughey N et al. 2012. A lipidomic analysis of CSF from HIV-infected subjects identifies metabolites that predict changes in cognitive performance. 19th CROI, 2012, Seattle. Poster abstract 472.
- Letendre S et al. 2012. A concise panel of biomarkers diagnoses and predicts neurocognitive status in HIV-infected individuals. 19th CROI, 2012, Seattle. Oral abstract 82.
- Burdo T et al. 2012. Elevated sCD163 Is a marker of neurocognitive impairment in HIV-infected individuals on effective ART. 19th CROI, 2012, Seattle. Oral abstract 81.
- McCutchan JA et al. 2012. Role of central obesity, diabetes, and metabolic variables in HIV-associated neurocognitive disorder. 19th CROI, 2012, Seattle. Poster abstract 490,
- Letendre et al. 2012. Higher cytomegalovirus (CMV) antibody concentrations are associated with older age, lower nadir CD4+ cell counts, and worse global neurocognitive (NC) functioning in people with HIV disease. 19th CROI, 2012, Seattle. Poster abstract 466,
- Croteau D et al. 2012. Older age is associated with higher antiretroviral concentrations in cerebrospinal fluid in HIV-infected individuals. 19th CROI, 2012, Seattle. Poster abstract 592.
- Perez-Valero I et al. 2012. Validation of the EACS guidelines 2011 Algorithm for detecting HAND in the CHARTER cohort. Paper 19th CROI, 2012, Seattle. Poster abstract 508.
- UNC School of Medicine. Start neurology substudy.