Much of the benefit of antiretroviral therapy (ART) on HIV infection in the central nervous system (CNS) could be due to its indirect effect upon immune activation rather than the capacity of particular drugs to suppress HIV replication in the brain, according to presentations at the recent HIV neurology meetings in Venice.
Reports at the meetings suggest that ART may have profound effects upon HIV levels in the cerebrospinal fluid (CSF) even when it doesn’t get into the brain, according to important new data from an SIV monkey model — as well as the shank of clinical experience since ART became available. In fact, according to recent data in humans ART can even lower viral load in the CSF when the virus in the rest of the body has become resistant to it, suggesting that the existing model of HIV infection of the CNS may need some revision.
“Overt AIDS dementia complex in the developed world has become quite rare… and is very unusual in patients on antiretroviral therapy, particularly those with plasma viral suppression,” said Dr Richard Price of the University of California, San Francisco (UCSF). “Though there are exceptions that are very important, treatment is usually very effective in suppressing [cerebrospinal fluid] (CSF) HIV. As a rule, even in the face of drug resistance, the effectiveness in treating CSF infection has been remarkable. And if we look at acute decay kinetics, the virus can decay just as quickly in the CSF as in the plasma and that also was a surprise.”
But this shouldn’t be happening, according to the current models of the origins of HIV in the CSF —either it should be virus in transit or overflowing from the plasma, or it should come from an autonomous/compartmentalised infection of macrophages and microglia (both longer lived cells) within the protected environment of the CNS.
The conventional wisdom has thus been that ART with good systemic effectiveness but poor CNS penetration may be able to prevent or treat mild neurological problems but that patients with compartmentalised infection would usually need ART with good CNS penetration.
However, an increasing body of evidence shows that this doesn’t always matter, leading Dr Price to propose an alternative model: that most of the viral load that can be measured in the CSF is “amplified” by systemic immune activation — and so ART that reduces systemic activation of the immune system may usually, though perhaps not always, manage HIV in the CSF.
HIV in the CNS and compartmentalised virus
HIV enters the CNS very soon after primary infection, probably inside infected monocytes/macrophages, and once there, the virus can infect cells such as microglia. However, the significance of the establishment of a CNS reservoir at this time has never been fully understood, although some preliminary data suggest that developing unique species of virus “compartmentalised” in the CNS could be a sign of poor prognosis even at this stage.
What is clear from the literature, according to Dr Serena Spudich, also of UCSF, who gave a presentation on the early stages of HIV involvement in the CNS at the conference, is that roughly 10% of the people with primary infection develop a variety of neurological disorders such as encephalitis, aseptic meningitis, cerebellar ataxia (muscle coordination problems), spinal cord problems (myelopathy and Cauda Equina Syndrome), optic neuritis, Guillaine-Barre syndrome, facial palsy and polyneuropathies.
“But the timing of onset and nature of these clinical syndromes suggests immune-mediated disease,” she said. Also, these conditions tend to resolve spontaneously without requiring any treatment — at roughly the same time as the individual achieves a lower stable plasma viral load “setpoint”). In fact, the literature has long described a close association between systemic and neurological disease. For instance, patients with any seroconversion illness (likely as a result of high plasma viral load) tend to have more rapid cognitive impairment, while the presence of neurological symptoms is associated with more rapid systemic progression.
Investigations of compartmentalised virus — and its response to treatment
Greta Schnell of the University of North Carolina presented data from SIV-infected macaques and from a larger cohort of humans with varying degrees of neurological impairment looking at the origins of the virus population in the brain.
In the macaque model (with three animals), two patterns were observed: concordance between CSF and plasma viral population in two monkeys without signs of CNS inflammation, and discordance (compartmentalisation) in the third macaque. Compartmentalisation in this animal was associated with the presence of high levels of perivascular macrophages in the brain, and elevated levels of CD8 T-cells and levels of monocyte chemoattractant protein (MCP-1), which both indicate CNS inflammation whether in monkeys or humans.
In the cohort of 67 individuals analysis revealed highly discordant viral populations in the CSF, and plasma that correlated with neurological disease state. This would suggest that as patients progress to dementia, the virus in the CSF appears to switch from a plasma or mixed plasma/CNS source to a primarily CNS source.
Schnell and her colleagues then hypothesised that the virus in the CSF must thus come from the long-lived cells that HIV infects within the brain. But the results of a subsequent study in patients starting ART did not appear to support this.
The half-life of the virus in the CSF (rate in the reduction in CSF viral load) was essentially the same in patients with compartmentalised and non-compartmentalised virus — all decayed relatively quickly, with some variations. If most of this virus in the CSF was being produced by HIV species in macrophages or microglia, viral loads should decline much more slowly.
So she concluded that the compartmentalised virus that is being measured in the CSF must come from a short-lived cell-type of unknown origin — which appears to contradict what most studies show are the HIV-infected cells in the brain tissue— primarily perivascular macrophages and microglia.
In another session, Dr Howard Fox of Scripps Institute, who presented data on a SIV monkey model of chronic infection, said that the variants within the CNS might not be evolving much at all — in fact, the virus in the brain might be archival (more similar to the species in circulation during primary infection). But it is also clear, that towards end-stage disease, infected monocytes and macrophages begin trafficking virus to the brain again — so altered variants could enter the brain at this time as well.
The role of activated monocytes and macrophages, and the benefit of non-CNS penetrant antiretrovirals
Rather than worry about compartmentalisation of virus per se, Dr Ken Williams of Harvard, an expert on macrophages, instead focused on how infected and activated monocytes and macrophages might be involved in the pathology of HIV-related neurological degeneration — and what sort of treatment might be necessary to stop it.
As already noted, activated monocytes and macrophages are involved in trafficking virus to the brain during primary infection, and once in the brain, these cells can do great damage, releasing inflammatory cytokines, TNF alpha, granzyme B and other inflammatory factors.
Dr Williams performed an experiment comparing the effects of combination ART in four monkeys versus four untreated monkeys who served as controls. But his choice of drugs was unusual: PMPA (the active compound in tenofovir), and racemic β-2′,3′-dideoxy-5-fluoro-3′-thiacytidine (which he called RCV but which is essential racemic FTC) — a combination that is roughly equivalent to Truvada.
He chose these drugs specifically because previous studies demonstrated that they do not penetrate the CNS, though they could treat virus systemically and in circulating monocytes. Furthermore, these antiretrovirals have been shown to reduce monocytes/macrophage activation.
Twenty-eight days after SIV infection and CD8 depletion —and after monocytes/macrophage activation and significant decreases in NAA — the monkeys were treated with the antiretrovirals. The effects were profound.
Even though treatment only reduced plasma viral load by a log or so and did not reduce it to undetectable levels, it was enough to stop monocyte/macrophage activation and infection in the periphery. This correlated to significant and rapid neurological stabilisation looking at NAA, with three of the animals completely recovered, while the third still showed some residual injury.
“Using this model, which I think is a reasonable model for HIV disease, it shows that you need active and continual monocytes/macrophage traffic into the brain [for ongoing disease], and lowering monocyte’s activation or infection seems to decrease disease,” said Dr Williams.
Dr William's model suggests that even modestly effective ART that dampens immune activation, could be enough to stop ongoing active neuronal injury in the brain — in other words, that the antiretrovirals used may not even need to get inside the CNS.
When is CNS-penetrating ART needed?
This would seem to go against the widespread belief that drugs that penetrate the CNS are needed to treat patients with HIV associated neurological disturbances.
Several previous studies have shown that ART with better neuropenetration scores could have greater effects upon viral load in the CSF. Indeed, Dr Scott Letendre of the University of California San Diego presented data from a number of patients from the CHARTER neurological study cohort, using an ultra-ultra sensitive viral load test that can detect HIV RNA copy numbers between 2.5-50 copies per ml. The findings showed that regimens with better neuropenetration were more likely to suppress viral load to truly undetectable levels in the CSF.
But it was not clear whether differences of such a small magnitude really have any clinically significant impact on neurocognitive performance. In fact, only three out of a dozen of the studies investigating the impact of neuropenetrating vs non-neuropenetrating ART that Dr Letendre highlighted have shown a significantly better impact on neurocognitive performance in recipients of neuropenetrative ART.
“There are conflicting data,” Dr Justin McArthur of John’s Hopkins University said on the following day of the meeting. “[Dr] Ned Sacktor working with our group showed that there was no effect on cognitive function or cognitive changes by the choice of antiretrovirals. You can choose an antiretroviral that had limited CNS penetration and it appeared that it still had some effect on neurocognitive impairment — about the same as a regimen that had putative good CNS penetration. But two more recent studies suggest that the choice of antiretrovirals does matter (Cysique L, 2004; Letendre S, 2006) and I think the field has now changed to reflect that view.”
Dr Letendre espouses the view that CNS penetration is particularly important in patients with CD4 cell counts around or below 200 cells/mm3 and neurological symptoms — people who are more likely to have autonomous compartmentalised infection. However, there are, as of yet, no data from randomised clinical trials comparing ART with good penetration to ART without in patients with neurological disorders, although one is now starting.
However, the interpretation of data from such a trial could be complicated by a number of factors — such as the inability of ART to repair already established brain injury.
“Neuronal loss is presumably permanent, even when CNS inflammation is ‘burnt out,’ said Dr McArthur. “So, an individual who has had neurocognitive disturbance with neuronal loss - no matter what you do to that individual, that degree of neuronal loss will be pretty well fixed. The role of HAART in reversing neurocognitive deficits — while it’s good and while it can have an impact, [there] is usually only… a partial response.”
This may have confounded the results of one of the last large NIH-sponsored randomized controlled studies performed in patients with HIV dementia, the abacavir add-on trial, recently published in PLoS Medicine by Professor Bruce Brew of the University of New South Wales and colleagues. The idea behind the study, which got underway shortly after ART came into widespread use, was to add high dose abacavir (which has good CNS penetration) vs. placebo to the existing ART regimen in people with established HIV dementia.
“Over the course of this study, neurocognitive improvement was seen both in the placebo group and in the abacavir-treated group with no statistical significant differences, and there were declines in CSF HIV RNA in both groups though somewhat better in the abacavir group,” said Dr McArthur.
He noted at the time the study began that the investigators did not anticipate the potent and durable effect of ART. In addition, many of people with HAD in this study, “probably 80% or higher - had no measurable inflammatory markers in the spinal fluid,” he said, “suggesting in fact that this had ‘burn out.’”
This underscores the need to identify reliable markers that can be used in real time in patients to differentiate between active disease and inactive disease. Notably, Dr Brew gave a presentation at the Venice meeting on one of these, neurofilament protein (NFL), a marker of axonal brain injury when detected in the CSF. He stressed that the use of NFL could improve the interpretation of trial results by showing whether participants still had active vs. burnt out disease, rather than relying upon neurological tests that could just be measuring residual deficits.
Use of such a marker, once clinically validated, might also protect patients with burnt out disease from having their otherwise successful ART regimens changed, or from having additional antiretrovirals added unnecessarily to their regimen.
The success of sub-optimal therapy
But Dr Price stressed that while there may be some people with progressive neurological disease who need ART with good CNS penetration, what is much more striking are the vast numbers of people who seem to do well on ART that doesn’t reach its full potency in the CNS.
“Some of the drugs get in well, but most of the regimens that we prescribe to patients, are going to be weaker in terms of penetration [and] potency in the nervous system than systemically,” he said.
And yet, dementia is not becoming epidemic and the development of drug resistance in the CSF is not rampant.
For the most part, his observations have been similar to what Greta Schnell reported, that viral load drops just as fast in the CSF as in the plasma — which shouldn’t be happening if the regimens are indeed ‘less potent’ in the brain.
To better understand what he called “the disproportionate effect of treatment” in the CSF, Dr Price and colleagues performed a cross-sectional analysis of plasma vs. CSF viral loads, and immune activation markers in 123 neuroasymptomatic HIV-infected individuals who were grouped according to their treatment status: Off (off treatment) (n=57); failures, who continued taking ART regimens despite the fact that their plasma viral loads were above 500 copies/ml (n=35) (the reason for failure was resistance); and successes with plasma viral load below 500 copies/ml (n=47).
With a few exceptions, the successes had CSF viral loads that were the same as or lower than their plasma viral load. But even though the failures and those off treatment had similar plasma viral loads, the CSF viral load in those continuing to take failing ART was consistently much lower, and the difference in the plasma/CSF ratio was about two logs in the failures vs. 1 log in those off treatment.
“So there is something about failed therapy that has more of an impact in the CSF than in the blood,” said Dr Price.
Earlier longitudinal studies in his patients showed that viral load in the CSF fell below levels in the plasma; that the rates of CSF decay and plasma decay were similar in most that CSF approached or reached the limit of detection, and that pleocytosis resolved. This occurred in several patients followed on failing therapy as well. However, when they looked at patients who interrupted failed therapy, although CSF viral load had been low at baseline, it rebounded once the individuals went off treatment, the rebound was greater in the CSF than in the plasma, and CSF white blood cell counts also increased.
The study measured immune activation by evaluating levels of CD8 cells co-expressing the CD38 and HLA-DR receptors (markers of activation) in blood and plasma. It found that treatment (whether failed or successful) achieved a statistically significant reduction in CD8 activation in both compartments. What happens with CD4 cell activation was less clear because they are infected and killed in greater numbers in both those off or on less effective treatment.
“Despite the fact that plasma viral load isn’t that much lower in those on failed treatment, the immune activation is significantly lower,” said Dr Price. Across plasma HIV RNA levels, T-cell activation in both fluids was lower in the failures than in individuals off treatment, although the association between CSF viral loads and CD8 activation did not differ by treatment status.
“It's not known why immune activation is lower in failed therapy,” said Dr Price. “This is not an original observation, it's been seen before (though not in the CSF) but the reason for it remains obscure. It could be reduced fitness [of the virus].”
To explain the findings, Dr Price hypothesises that the level of immune activation drives the CSF viral load in relation to the plasma viral load — and that the effect of failed therapy can be explained by its effect on immune activation. This may correlate well with Dr William’s findings.
“It may very well be that macrophage activation occurs in parallel, and that CD8 activation is a surrogate for generalised activation,” he said.
This would lead to a more complicated model of CSF infection and the effects of treatment, in which immune activation is involved both in directing infected cells into the CSF and CNS, while the availability of activated CD4 cells in the CSF amplifies viral load whether from transitory source or autonomous infection in the CNS.
Treatment reduces plasma viral load (and thus the transitory component) but probably has less potent direct effects upon the protected autonomous and amplified components. However, it can have a potent indirect effect on these the autonomous and amplified viral components by interrupting immune activation, trafficking and amplification — even less potent therapy if it reduces immune activation. This model could explain the rapid viral load decay rate in both his patients and the cohort described by Greta Schnell.
“We think the amplified component is probably the predominant component being measured in the CSF,” he said. He added that “this model would be compatible with a degree of genetic compartmentalisation after primary infection, would explain the disproportionate effect of ART in the CSF through reduction of amplified infection,” could potentially enhance the evolution of more potentially neuropathic variants, and since amplified variants may derive from the brain, it could allow us to see brain derived virus.”
But Dr Price shied away from saying that systemically effective ART therapy is all that is needed to treat neurological disease in people with HIV.
“The caveat I would say about our cohort was that they were neurologically asymptomatic people — the role of penetration is very likely to be important in people with major CNS infection. I wouldn’t dismiss penetration, but I don’t think that it’s as important as we make it out to be for general systemic therapy prior to the development of dementia.”
One final possible ramification of this model is that treatment responses in the amplified component of virus in the CSF could obscure some clinically important autonomous infections in the brain and potentially mislead clinicians into thinking there was a response to therapy (yet another reason to develop reliable biomarkers for active neurological disease).
Whether the presence of low levels of ongoing HIV replication in the CNS, or even the persistence of chronically infected cells is associated with significant progressive disease was a matter of much debate on the final day of the meeting.
Williams K. HIV infection and the effects of antiretroviral drugs in macrophages. Second HIV Infection and the Central Nervous System: Developed and Resource-Limited Settings, Venice, Italy, 2007.
Spudich S. Central nervous system involvement in primary HIV infection. Second HIV Infection and the Central Nervous System: Developed and Resource-Limited Settings, Venice, Italy, 2007
Schnell G et al. Virus compartmentalization and trafficking through the central nervous system. Second HIV Infection and the Central Nervous System: Developed and Resource-Limited Settings, Venice, Italy, 2007
Letendre S. Neuropenetration and efficacy of antiretroviral therapy in the central nervous system. Second HIV Infection and the Central Nervous System: Developed and Resource-Limited Settings, Venice, Italy, 2007.
Price R.W. Immune activation and the disproportionate effect of antiretroviral treatment on cerebrospinal fluid HIV. Second HIV Infection and the Central Nervous System: Developed and Resource-Limited Settings, Venice, Italy, 2007.
McArthur J. The changing phenotype of HIV dementia in the HAART era: potential mechanisms. Evolving Mechanisms of HIV Neuropathogenesis in the HAART era: Domestic and Global Issues, Venice, Italy, 2007.