HIV-infection in the brain, neurocognitive dysfunction and CNS drug penetration

Svilen Konov, HIV i-Base

Even though general interest in the effect of HIV on the brain and neurocognitive function has fallen off since the introduction of HAART as a result of reduced number of cases, it has renewed recently with reports that can help us to have a deeper insight into the mechanisms HIV uses to affect the central nervous system (CNS).

HIV enters the brain within days of primary infection and is found in cerebrospinal fluid throughout the course of disease. A study presented at CROI 2007 in Los Angeles, looked into the HIV-related dementia and its association with HIV DNA within monocytes. [1]

It was previously reported that high levels of HIV-1 DNA in peripheral blood mononuclear cells (PBMCs) is strongly associated with a diagnosis of HIV-1-associated dementia both in ART treated and non-treated subjects. The researchers in this study hypothesised that high HIV-1 DNA would be found specifically within CD14+/CD16+ subsets of PBMC. This hypothesis was based on suggestive laboratory observations. They followed two cohorts and HIV-1 DNA was assessed with real time PCR. The results showed that HIV-1 DNA was significantly higher in CD14+/CD16+ cells in people with HIV-associated dementia (HAD), regardless of current use of HAART , compared to those without HAD.

Another basic science study looked into oxidative stress in the lenticular nuclei (one of the structures, together with the caudate nuclei that makes up the basal ganglia) and the possible use of magnetic resonance spectroscopy in HIV neurocognitively impaired people. [2]

N-acetyl aspartate, choline and creatine are affected in patients with HIV-related dementia. These metabolic markers, however, are often normalised in certain CNS regions (i.e. the lenticular nuclei) after initiation of HAART. As a result, different surrogate biomarkers are needed. The investigators in this study hypothesised that lactate, a marker of inflammation and anaerobic glycolisis, and lipid, an indicator of cell membrane turnover due to oxidative stress, could serve as surrogate biomarkers. The team performed MRS (Magnetic Resonance Spectroscopy) of the lenticular nuclei of HIV-negative controls and 45 HIV-positive subjects on HAART with different degrees of HIV neurocognitive impairment. Analysis of variance among the various groups was performed.

The results showed that the lactate/creatine was significantly higher in HIV-positive people with moderate to severe neurocognitive impairment compared to HIV-negative controls. In contrast, both N-acetyl aspartate/creatine and choline/creatine ratios were similar in all groups. The researchers concluded that HIV induces inflammation and oxidative stress in HIV-positive patients despite HAART; that lipid and lactate are sensitive inflammatory biomarkers; and that the results may prompt use of additional neuroprotective medications in patients with neurological impairment.

A study by Spudich S from the University of California, San Francisco, US and colleague attempted to determine predictors of cerebrospinal viral load during primary HIV-1 infection in a cross sectional study. [3]

The group studied 46 subjects during the first six months of HIV infection (confirmed by seroconversion or less-sensitive ELISA testing) with CSF and blood analysis. The possible predictors investigated were blood CD4 and CD8 counts, plasma HIV-1 RNA levels, CSF white blood cell count and CSF protein with an outcome of interest the level of HIV RNA in the CSF. The number of days of infection was also included as a variable in a multivariate regression analysis.

The results showed that CSF HIV RNA levels are independently correlated to with plasma HIV RNA (increasing by 0.85 log for each 1 log increase in plasma, p<0.001). Elevations in CSF white blood cell count and protein were moderately but also statistically significantly correlated with CSF viral load (p<0.011 and p<0.025 respectively). Analysing the time since infection, the researchers concluded that early CSF HIV infection is driven primarily by plasma viral load.

CNS drug penetration and the impact of treatment on viral replication in the CSF is one of the primary interests for influencing neurocognitive disease in HIV infection.

Rosa and colleagues looked at CNS penetration of lopinavir/r monotherapy in 11 subjects in a sub-study of the IMANI-2 study. This is an ongoing, prospective, open-label investigation of LPV/r single agent in 40 treatment-naive patients. The 11 subjects had at least 24 weeks of LPV/r monotherapy before enrolment and two consecutive viral load results <50 copies/mL at least 4 weeks apart and at the most recent visit. A neurologist performed a lumbar puncture to all the participants. CSF viral load was measured with real time PCR. 10/11 were evaluable and one was excluded as a result of a traumatic tap: five males (two African American, one Caucasian and two Hispanic) and five females (three African American and two Caucasian) with a mean age of 42.

Before treatment, subjects had median CD4 count of 264 cells/mm3 and a viral load of 4.28 log10. At the time of analysis, the median CD4 count was 493 cells/mm3 and all participants had a viral load <50 copies/mL. All subjects achieved suppression of CSF viral load <50 copies/mL. It would have been useful to know the baseline CSF viral load in these patients. Longer follow-up in larger numbers of patients is needed to determine the long-term durability of lopinavir/r monotherapy.

Two more studies looked into suppression of viral replication in the CSF.

Andrea Antinori and colleagues from the National Institute of Infectious Diseases in Rome looked at whether CSF penetration of combination ART was related to plasma viral load. The study included 2,785 treatment naive patients. [5]

They all had a plasma viral load of <80 copies/mL. A CSF penetration score (where 0 is low, 0.5-intermediate and 1 is high) was used. The score was based on the CHARTER definition, which in its turn is based on the pharmacodynamic and pharmacokinetic drug properties. The scoring of the regimen was achieved by adding the scores of the individual components of the combination. A Poisson regression was used to estimate the number of rebounds (virologic failures) per person years.

Baseline demographics included: 29% female, median age 36 years, 32% IDUs, 39% heterosexual, 37% HCV antibody positive, median CD4 count 260 cells/mm3 and HIV viral load 4.8 logs.

Although the results showed that higher values of current Cerebrospinal Penetration Scores (CPS) were associated with declining incidence rates of viral failures per 100 person years of follow up: 9.7 (5.7 to 15.6) if CPS 0.5; 8.1 (6.8 to 9.5) if CPS 1 to 1.5; 4.9 (4.1 to 5.7) if CPS>/=2, after adjusting for ART drugs currently used, however, the researchers found that the association between the increased penetration scores of ART in CNF and the lower incidence rate of viral failure was no longer statistically significant. The adjusted RR was <1 only in people starting combination ART with a low CD4 count.

A second study (ACTG 736) looked into whether the suppression of HIV replication in plasma requires suppression of HIV replication in the CSF. [6]

This was a multi-site longitudinal natural history study conducted at 15 adult ACTG sites (AIDS Clinical Trial Group) in the US. Eligibility criteria included: advanced HIV disease and either beginning a new HAART regimen or changing an existing regimen because of virologic failure. Participants’ neurolocical and neurocognitive state was checked by venipuncture and lumbar puncture, as well as by neuropsychological tests at baseline and after 24 and 52 weeks on the new combination.

Drugs considered to have good CNS penetration were zidovudine, abacavir, stavudine, nevirapine, efavirenz, lamivudine and indinavir. For the combination to be classed as having good CNS penetration, it had to include at least two of these ARVs. The association between the variables at each time point was done via generalised estimating equations with an autoregressive correlation structure.

101 subjects enrolled and 77 completed the study. The odds of undetectable CSF HIV RNA were 3.10 fold (95% CI 1.03 to 9.30) higher in those who took a combination with a good CNF penetration and 3.49 fold (95% CI 1.20 to 10.18) higher in treatment naive patients.

In addition, for every 1-log decrease in concurrent plasma HIV RNA, the odds of undetectable CSF HIV RNA increased 2.94-fold (95% CI 1.96 to 4.35). For every 1-log decrease in concurrent CSF HIV RNA, the odds of undetectable plasma HIV RNA increased 25-fold (95% CI 12.5 to 100). Even though the CIs of the study have a broad margin, the results still support the hypothesis that control of CSF virus is important for suppression of the plasma viremia. This may have clinical relevance in treatment experienced patients and any patients with neurological symptoms. It raise the question of whether these drugs may have a role for compartmental activity, even when resistance has accumulated in plasma.

A German study investigated the relationship between neuropsychological symptoms in patients where CSF viral load remained higher than plasma viral load. [7]

Apart from the fact that the investigators use terminology from the 1980s (i.e. HIV carriers, drug abusers, etc.), the study highlights an issue that is not well researched, both in terms of epidemiology, as well as its effect in clinical terms. The study recruited 142 patients consecutively, 33 (22.9%) of who had higher CNF than plasma viral load. Pure motor test, motor cognitive tests, motor cognitive visual test (grooved pegboard) and pure cognitive test were performed. Interestingly, patients with higher CSF than viral plasma load, even without overt clinical symptoms, showed marked pathological neuropsychological test profiles in contrast to individuals with higher plasma viral load who reveal only mild to moderate motor, but not cognitive deficits.

Two more studies that may have clinical relevance were presented during the last day’s poster abstract sessions. The first one was an attempt to clarify when is the right time to start ART in order to avoid the HIV-related neurocognitive impairment, having as a reference the patient’s nadir CD4 cell count [8].

The study had a cross-sectional design and included 64 subjects. It excluded (for some reason) patients who “reported drug abuse”, or who had psychiatric or neurological disorders. The percentages of subjects with neurocognitive disorders were compared through different CD4 count cut-offs, used in different settings as a signal for initiation of ARV therapy, namely: 200, 250, 300 and 350 cells/mL. The results are shown in Table 1 below.

Table 1: CD4 cut-offs and percentage of participants with neurocognitive disorders

Nadir cd4 count Percentages
=200 vs >200 73.1 vs 52.6
=250 vs >250 66.7 vs 53.3
=300 vs >300 63.9 vs 56.5
=350 vs >350 57.1 vs 62.5

Although the results were not statistically significant, there was a trend for a higher percentage of patients with neurocognitive impairment at the lower strata of nadir cut offs. The only two statistically significant results were observed in over 100 paired areas (both in respect to the nadir 200 cut off): attention/working memory (span backward, p=0.032) and executive functions (TMT-b, p=0.020). Even though the researchers conclude that their results confirm that nadir CD4 count is a potential risk factor for HIV-related neurocognitive impairment, additional larger and better powered studies are necessary to ascertain the findings, so that they can become a part of the recommendations for the time of initiating therapy in specific sub-populations.

The other study with potential clinical implications looked into other than ART medications, approved by the regulatory bodies for different indications and that could possibly benefit -HIV-positive patients with neurocognitive problems [9]. In vitro data support that serotonin reuptake inhibitors and HMG-CoA reductase inhibitors (statins) affect the replication of HIV. The study analysed the results of 658 patients. The data were collected via comprehensive neuropsychological tests and HIV RNA measurements in plasma and CSF. 467 (71%) of the participants used ART, 195 (30%) used SRI and 63 (10%) used statins. SRI users were less likely to have HIV RNA levels > 50 copies/mL in CSF (29% vs. 37% in non-SRI users, OR 0.69, p=0.05).

This association was particularly clear for three of the seven SRIs used by the participants, namely citalopram, setraline and trazodone and was limited to those participants who were not on concomitant ART. The SRI users also performed better on neuropsychological tests (median global deficit score 0.37 vs. 0.47, p=0.04). Statin users were also less likely to have HIV RNA>50 copies/mL in CSF (16% vs. 37%, p<0.0001). They, however, showed the strongest association in those who used ART (2% vs. 18%, p<0.0001). Statin use was not associated with better neuropsychological performance.


All references are to the Programme and Abstracts of the 14th Conference on Retroviruses and Opportunistic Infections, 25-28 February 2007, Los Angeles.

  1. Shiramizu B, Shikuma C, et al. HIV-1 associated dementia associated with HIV DNA within monocytes in subjects treated or not treated with ARV medications. Oral abstract 114.
  2. Ances B, Roc A, et al. MRS can detect HIV-induced inflammation and oxidative stress in the lenticular nuclei. Oral abstract 116.
  3. Spudich S, Gisslen M, et al. Predictors of cerebrospinal fluid viral burden during primary HIV-1 infection. Oral abstract 115.
  4. Rosa Y, Letendre S, et al. Single-agent therapy with lopinavir/ritonavir controls HIV-1 viral replication in the central nervous system. Poster abstract 381.
  5. Antinori A, Cozzi-Lepri A, et al. Could cerebrospinal fluid ART penetration scores predict viral rebound in ART-naive individuals who achieve plasma viral load suppression on their first therapy? Poster abstract 382a.
  6. Marra C, Sinha S, et al. Supression of HIV replication in plasma requires suppression of HIV replication in cerebrospinal fluid. Poster abstract 382b.
  7. Arendt G, Nolting T, et al. Distinct neuropsychological deficit pattern in HIV-1+ patients with higher CSF than plasma viral load. Poster abstract 385.
  8. Munoz-Moreno J, Rodriguez C, et al. Recommended earlier initiation of ART based on nadir CD4 cell count as a risk factor for HIV-related neurocognitive impairment. Poster abstract 383.
  9. Letendre S, Marquie-Beck J, et al. Clinical evidence of antiviral activity of serotonin reuptake inhibitors and HMG-CoA reductase inhibitors in the central nervous system. Poster abstract 384.

Links to other websites are current at date of posting but not maintained.