Inflammation and intervention: how does HIV cause AIDS and how does it cause disease despite ART
David Margolis, MD University of North Carolina for natap.org
the consequences of HIV replication clearly persist long after ART has suppressed viraemia … Interventions in the best of worlds could actually become therapies to be used to improve or normalise immune function and health outcomes, but least interventional clinical experiments might enlighten as to the drivers of HIV pathogenesis despite ART … Hydroxychloroquine (chloroquine) dampens the inflammatory response … in ART-suppressed patients without immunological response to ART … An array of immunological changes were seen … the potential toxicity of hydroxychloroquine would need to be weighed and tested
HIV infection was first thought to lead to AIDS by the uncontrolled replication of the highly cytopathic virus, killing CD4 cells and laying waste to the immune system. Some studies, however, suggest that the direct killing of lymphocytes by HIV is insufficient to account for immunodeficiency. A state of chronic, generalised immune activation has been suggested, in combination with viral replication, to be required to ignite progression to AIDS.
A recent paper in Cell by Warner Greenes group has revived the claim that HIV replication is central to CD4 depletion and immune collapse via the triggering of pyroptosis, a form of programmed cell death that engulfs surrounding, uninfected cells in the doom triggered by the viral death of the few.  However, this view was not represented at CROI.
The current majority view, reinforced by many related presentations at CROI, emphasised the role in HIV immunopathogenesis played by the mucosal immune dysfunction and its partner in guilt-by-association, bacterial translocation. It has been long observed that antiretroviral therapy (ART) suppression of viraemia usually does not fully resolve immune activation, although for most patients the proportion of cells bearing markers of activation are far below that of untreated patients. Two general types of studies at CROI focused on the issue of immune activation in HIV infection: 1) those outlining phenomena that were associated with immune inflammation, and that perhaps drive the process, and 2) those that used therapeutic interventions in an attempt to perturb the system and perhaps ameliorate pathology.
Observations on inflammation
Various soluble markers such as LPS and bacterial rDNA have been reported to be the result of microbial translocation, and associated with immune activation. Abdurahman and colleagues introduced bacterial flagellin-specific antibodies and an assay of soluble flagellin protein as new surrogate markers for immune activation. 
Flagellin is a potent inducer of the toll-like receptor, a human receptor designed to sense foreign proteins and respond by the induction of inflammation. The group showed higher levels of flagellin and flagellin antibodies in untreated patients than in treated patients, both groups being higher than uninfected controls. As with other such markers, flagellin markers were reduced but not completely by the use of ART. The investigators also noted that levels were higher in patients from Ethiopia and Vietnam than from Sweden, but given the diversity of these populations and the small sample size, it is difficult to make much of this difference. It remains to be seen if this biomarker is cheaper, more reproducible, or simpler to use in research setting than others that have been described.
Kamat and colleagues discussed a small but interesting study of the application of serological markers used to diagnose inflammatory bowel disease (IBD) to HIV-infected patients.  Antibodies to ASCA, pANCA, anti-OmpC, and anti-CBir1 were measured by ELISA in plasma from AIDS patients (n = 26) with low CD4 counts (<300 cells/mm3) and high plasma LPS (>80 pg/mL), and results correlated with clinical data. A weakness of this study was the fact that only 6 of the 26 patients were virologically suppressed, and in this report suppression was considered to be < 400 copies HIV RNA/ml. Further this was an advanced cohort with a mean CD4 count of 80 cells/µl. However antibodies linked to IBD was detected in 46%, and of these 75% had a Crohns-like pattern. Analysis showed a positive correlation between the flagellin antibody anti-CBir1 and IL-6 levels (r = 0.447, p = 0.048). The authors suggested that, like in IBD, these biomarkers might serve to monitor HIV-related gut inflammation. Should specific therapies to downmodulate inflammation be developed, such monitoring strategies might be useful, either in trials or in practice.
Nagy and colleagues showed that antigen-presenting cells, capable of phagocytising foreign antigens and microbes, differed in their behavior in uninfected, HIV-infected, and HIV-infected but ART-suppressed patients.  Antigen-presenting cells from untreated HIV-positive patients on average produced more of the pro-inflammatory cytokines TNF, IL-6, and IL-12 when exposed to several microbes, or when stimulated through Toll-like receptors, the innate receptors that detect the presence of such pathogens. In most cases, the dysregulation or hyper-reponsiveness of antigen-presenting cells was ameliorated by suppressive ART. Assays of global cellular gene expression in APCs, illustrated that this dysregulation was correlated with global alterations in cellular gene expression. This observation added to the catalog of cellular immune dysfunction that has been observed in viraemic patients, partially ameliorated by effective ART. It remains to be seen if the duration of ART or the timing of it initiation after infection allows fuller normalisation of this immune response.
In thematic agreement with this observation, Funderburg and the ACTG 5248 investigators examined immune parameters in patients initiating a raltegravir/Truvada regimen.  The 37 patients included in this immunological substudy had moderately advanced disease, with a median CD4 count of 259 cells/mm3. In patients in whom viraemia was successfully suppressed, over the 72-week study markers of T cell activation such as the presence of the activation markers CD38 or HLA-DR, the marker of cell turnover Ki67, the levels of TNF receptor and the soluble receptor for lipopolysaccharide (sCD14) all fell precipitously in the first few days of therapy. However, in general, most markers were still measurably elevated after 8 weeks of therapy, and in most patients did not return completely after 2 years of therapy (week 72) to levels typically seen in uninfected patients. Similarly, bacterial lipopolysaccharide declined but did not normalise. The authors speculated that the association observed between continuing microbial translocation (higher LPS levels) and increased turnover (Ki67+) of central memory CD4 cells might mean that microbial translocation was the cause of CD4 cell depletion.
Similarly, Shive and colleagues from Case Western Reserve University studied a cohort of 61 patients who failed to enjoy immune reconstitution despite durable, successful ART (CD4 <350/µL after at least 2 years of HIV RNA below the limit of detection), and compared these immune failure (IF) patients to 21 HIV-negative controls, and 20 HIV-positive treated patients with immune reconstitution (immune success, IS). 
A panoply of soluble markers and mediators of inflammation were measured: the pro-inflammatory cytokine IL-6, the D-dimer coagulation marker, and markers of microbial translocation — the soluble CD14 LPS receptor and LPS itself. Overall, the mean levels of these markers were statistically different in ways that would be largely expected, given the prior work of this group and others. IL-6 and sCD14 levels were higher in IF than IS, and both higher than in uninfected patients. D-dimer levels were similar in IF and IS, but higher than healthy controls. LPS tended to be higher in IF and IS than in controls, but not significantly. In IF patients, but not IS patients, statistical correlations were seen between duration of untreated HIV infection and IL-6 and to D-dimer levels. What is unsatisfying is that there is considerable overlap in the individual data points, despite the statistically significant difference that can be demonstrated in mean or median values. If these parameters are critical drivers of immunopathogenesis, it is dissonant that these parameters are well within the normal range in a large number of patients with profound immune abnormalities.
A study of 9 patients from UCSF, who initiated therapy with advanced disease (mean nadir CD4 count 87 cells/mm3) and had maintained ART suppression for a median of 40 months, visually quantitated the density of collagen seen on staining of rectal biopsies.  Hunt and colleagues speculated that fibrosis in the lymphoid tissue of the GALT might impair functional T cell responses. In particular, it was hypothesised that the maintenance of CD8 cell responses might be impaired by fibrosis. As is generally the case in many such studies, a hyperactivated CD4 cell population was observed. Why abnormalities of lymphoid architecture would specifically lead to a dichotomised phenotype of CD38+ CD4+ cells but not CD38+ CD8+ cells was not addressed. Why the maintenance of an HIV-specific CD8+ T cell response would be expected in patients in whom viraemia had been suppressed for a median of almost 3.5 years was also unclear.
Interventions for inflammation
HIV pathogenesis appears to be extremely complex and multifactorial. Of course, it really is ultimately all about the virus, stupid, as without HIV replication there is no pathogenesis. However, the consequences of HIV replication clearly persist long after ART has suppressed viraemia. So the challenge that the field faces is to determine which the myriad effects of the virus on the immune system are the central and significant drivers of pathology, and are there any interventions that can improve health or outcomes beyond those benefits conferred by ART. A radical might even say what interventions can confer benefit in place of ART, but that seems a bit heretical in the so-called post-SMART era.
As described above, CROI presented many observations of immune-related phenomenology and associations of these with immune pathology. But to go beyond associative work, the relationship of phenomena to pathology must ultimately be tested with interventions. Interventions in the best of worlds could actually become therapies to be used to improve or normalise immune function and health outcomes, but least interventional clinical experiments might enlighten as to the drivers of HIV pathogenesis despite ART. Happily, several such studies were shared at CROI.
Several studies have demonstrated that low-level viraemia is unaffected by the intensification of suppressive ART with raltegravir. Lichtenstein and colleagues measured the immunological effects of raltegravir intensification.  This was an uncontrolled, single-arm study of 30 patients, with results from 26 being reported. At baseline, patients had CD4 counts <350 cells/mm3 or declining CD4 counts (average CD4 count 273 cells/mm3). ART suppression was in place for > 1 year, but an average of 7.2 years. This was therefore a somewhat unique cohort of patients with discordant immune responses despite very long-term suppression. After 1 year of raltegravir intensification, there was no change in absolute CD4 number, but modest increases in CD4% and CD4/CD8 ratio. So it seemed that therefore CD8 count declined. Also observed was a substantial reduction in activated CD4 cell percent, but it was not clear if this was a 30% decline from a high baseline or from a small number of residually activated cells.
In a larger, controlled crossover study of RAL intensification Ghandi led the ACTG 5244 team to examine the effect of RAL intensification.  This work had previously shown no change in low-level viraemia after addition of RAL, but here focused on other potential effects of intensification. In this study they could measure no change in the detection of 2-LTR circles, dead-end products of HIV reverse transcription that can be increased by RAL in vitro. Some patients had detected 2-LTR circles at baseline (2-LTRpos) and others not (2-LTRneg). Like the Lichtenstein group, Ghandi et al. examined CD4 counts and percentages, as well as CD4 and CD8 activation, again measured using the surface markers CD38 and HLA-DR. Neither the 2-LTR-positive nor 2-LTR-negative group had a significant decline in CD4% or CD8 cells that were CD38+HLA-DR+. This group had earlier reported a slight trend towards increase in CD4 cell counts over the 12 week period of raltegravir intensification  differing from the modest increase in CD4% but no change in CD4 cell count seen in the Lichtenstein study. Obviously, some of the differences in results might be related to the patient groups: stable patients in the Ghandi study and patients with persistently lower CD4 counts in the Lichtenstein study. But neither study really revealed profound immunological benefits of raltegravir intensification, suggesting that as a clinical maneuver intensification may not be a practical answer to poor immune reconstitution, and leaving doubt that ART intensification could further normalise immune function in a clinically meaningful way.
Massanella also reported on immunological findings in follow-up of their prior raltegravir intensification study.  In this study, the investigators had originally seen an increase in 2-LTR circles after RAL addition, but this was measured at two weeks after intensification, and did not persist. Ghandi et al. saw no increase in circles, but did not have samples to examine the two week timepoint, and could only measure later. In the Massanella study, patients in whom an increase in 2-LTR circles was seen also had a greater level of CD8 T cell activation (of HLA-DR+CD38+, HLA-DR+CD45RO+, and CD45RO+CD38+) prior to raltegravir, and a reduction after 24 weeks of raltegravir. Levels of CD8 activation returned to their prior elevated levels after raltegravir discontinuation. However, this deactivation/reactivation effect was not seen in most of the patients (22 of 34) in the study who did not have an increase of 2-LTR circles. Also unlike the Lichtenstein study but like the Ghandi study, no changes were observed in activation markers in CD4 T cells.
What does this all mean? Massanella concluded that the reduction of immune activation in CD8 T cells after 48 weeks of intensification, and the subsequent reversion to pre-intensification levels after RAL discontinuation, suggested that CD8 T cell activation may reflect incomplete suppression of viral replication during apparently suppressive HAART. To this author, this conclusion seems difficult to reconcile, unless one accepts that plasma viraemia unaffected by raltegravir intensification is not proportional in some way to viral expression, new cell infection, new viral integration, and new viral production in another (tissue?) compartment.
This is because in patients with measurable 2-LTR circles and stable low-level viraemia (measurable but less than 50 copies/ml), Massanella showed that 2-LTR circles transiently increased after raltegravir presumably as cells that are newly infected form circles rather than viral integrants. But then concurrent with this, CD8 cell activation declines, presumably as fewer cells experience new viral integrants, and production of virions or viral proteins declines. However, plasma viral load does not change during RAL intensification, and ART resistance does not develop prior to raltegravir intensification, despite these putative ongoing rounds of reverse transcription, integration, and virion production. In my view, the results of these three studies are not concordant, and there is not one model that accounts for all the observations. It is possible some of the observed effects are marginal ones seen in small studies of uniquely selected patient populations.
Two other studies used the CCR5 entry inhibitor maraviroc as a probe in a manner similar to the RAL intensification studies. During other maraviroc studies, a rise in CD4 cells has been reported. The mechanisms that lead to this rise, and its clinical relevance, have been unclear. Wilkins and colleagues reported follow-up at 48 weeks from an ACTG maraviroc intensification study in patients with CD4 count <250 cells/mm3 after ART.  The initial 24-week intensification study had reported declines in CD4+ and CD8+ activation after maraviroc intensification; these changes were seen to partially reverse 24 weeks after maraviroc discontinuation. Absolute CD4 counts did not increase in this trial, as had been reported in some earlier MVC studies. That some measures of immune activation improved after maraviroc intensification, and then partially reversed after maraviroc discontinuation, support the idea that maraviroc was the direct cause of these effects.
An opposite result was found by Hunt and colleagues, who performed a randomised, placebo-controlled trial of 24 weeks of maraviroc intensification in patients with CD4 counts <350 cells/mm3 and plasma HIV RNA levels <48 copies/mL on ART for =1 year.  Most had been on suppressive ART for about 30 months. The primary outcome measurement examined was an immunological one: the change in percentage of activated (CD38+HLA-DR+) CD8+ T cells in peripheral blood. Maraviroc or placebo was added for 24 weeks, with virologic and immunologic evaluations at entry, week 4, week 24, and twelve weeks after maraviroc was stopped at week 36. Serial rectal tissue sampling was also performed in some patients.
HIV RNA by a single-copy assay declined for week 0 to week 4 in both arms and in both groups CD4 cell counts rose 37 cells/mm3, highlighting the danger of imputing biological effects to small changes in assays near the limit of detection, and the benefit of a placebo control. The overall primary results of the study were even more perplexing.
Starkly opposite of what was expected and hypothesised, through week 24 the percentage activated CD8+ T cells declined in the placebo arm but increased in the maraviroc arm. Again, surprisingly in the rectal tissue, biopsies showed a nearly two-fold increase in rectal CD4+ or CD8+ T cell activation in the maraviroc arm, and small decrease in the placebo arm. To complete the confounding observations, plasma lipopolysaccharide, a marker of bacterial translocation declined through week 24 in the maraviroc arm despite increasing markers of activation, while sCD14 (another marker associated with bacterial translocation) increased.
Again, what are we to make of these discordant results? As maraviroc binds human CCR5 cell receptors, these changes may reflect an immunomodulatory effect of maraviroc, and might not be related to any antiviral effect of the drug in these suppressed patients. But given the different findings in studies, one can only conclude that these discrepancies are related to subtly differing clinical populations, or that some of the changes must be ascribed to chance or biological variation. These results remain to be further confirmed, as does the demonstration that this modest immunomodulatory effect results in improved immune function, or even a clinically relevant outcome.
Piconi and colleagues presented a creative study attempting to measure the effect of a direct immunodulatory drug in ART-suppressed patients without immunological response to ART. 
Hydroxychloroquine alters lysosomal pH and has been used to treat malaria. It immunological effects have led it its use in autoimmune diseases, and recently it has been shown to act by blocking the signaling of toll-like receptors on dendritic cells. In this way, chloroquine dampens the inflammatory response.
Twenty patients with nadir CD4 counts of about 50 cells/mm3 and CD4 counts on ART suppression persistently <200 cells/mm3 (median 184 cells/mm3) received 400 mg/day of hydroxychloroquine for six months. An array of immunological changes were seen, some of which reverted to or towards baseline two months after hydroxychloroquine was stopped. These included an increased CD4+ T cell percent, a reduction of activated CD4+T cells, memory and activated CD8+ cells and activated monocytes. Plasma LPS levels were rescued, and this reduction persisted after hydroxychloroquine was stopped. Although the potential toxicity of hydroxychloroquine would need to be weighed and tested, and the potential clinical benefits of these changes demonstrated more clearly, the use of a well-known and inexpensive drug probably deserves more study.
Inflammation and Intervention in the CNS
Echoing this work in the peripheral blood, there were some studies of inflammation and therapy in the CNS presented at CROI. Carsenti-Dellamonica and colleagues measured neurocognitive impairment by psychometric testing, and in 179 patients in whom a plasma sample to measure LPS was available, found that higher LPS levels were associated with neurocognitive impairment. Again, as discussed above, this association may not be causal. The other independent association in multivariate analysis found with neurocognitive impairment was cellular proviral DNA, a measure associated with long-term exposure to viraemia and advanced disease. 
In a similar way, Lyons and colleagues found that soluble CD14 was associated with impaired performance on neurocognitive testing. In 97 patients with nadir CD4 counts <300, they found plasma sCD14 levels were higher in subjects with neurocognitive test scores indicating global impairment. However, their cohort was challenging, 38% HCV-positive, 59% using illicit drugs, only 72 of the 97 on ART, and of 57/72 patients had HIV RNA >1000 copies/mL. While the authors suggested the potential utility of plasma sCD14 as a peripheral biomarker to monitor progression of neurocognitive dysfunction related to HIV, this observation may be complicated due to the populations coinfection, drug use, and incomplete ART suppression rates. 
Letendre studied the phosphorylated form of the CSF protein Tau, and found that Tau levels were higher in older patients on ART with greater immune recovery, and that this in turn was associated with poorer memory function. Again, the utility and specificity of this biomarker for predicting and following HIV-related neurological disease will have to be expanded and broadened in other patient cohorts. 
Dahl and colleagues performed an antiviral intensification study focusing on the CSF. HIV-1-infected patients on therapy for >2 years with plasma suppression <50 copies/mL for >1 year and CSF suppression <50 copies/mL at screening were enrolled in a randomised, open-label study of a 12-week course of raltegravir intensification. HIV RNA in the CSF was detected, but was generally lower than in plasma: HIV-1 was detected in the CSF in 8/36 samples (median 0.3 copies/mL) and in the plasma of 26/40 plasma samples (median 1.4 copies/mL) with a low inter-patient variability. As in the plasma, there was no effect of raltegravir intensification on CSF HIV RNA. 
Similarly, Price and colleagues added raltegravir to suppressive regimens should test if this resulted in a further decrease in brain viral replication and its stimulation of intrathecal immune activation. They found no evidence that RAL intensification reduced intrathecal immunoactivation in virally suppressed subjects. 
Sacktor and colleagues reported the results of a minocycline trial aiming to treat HIV-associated neurocognitive disorders (HAND). HIV-1-infected individuals with progressive neurocognitive decline were enrolled in a prospective, randomised, double-blind, placebo-controlled study of minocycline, across 16 sites. Participants were randomised to receive either minocycline 100 mg orally or matching placebo orally every 12 hours. 107 HIV-infected individuals, of whom 51% had mild cognitive impairment, and 38% had HIV dementia (mild-MSK Stage 1). Minocycline was safe and well tolerated in individuals with HIV-associated cognitive impairment, but cognitive improvement was not observed in this study. As with the periphery, there are not yet good tools to intervene in the mild inflammation that persists in many patients on ART, and not a clear understanding of the risks and benefits of potential interventions. 
Unless stated otherwise, all references are to the Programme and Abstract Book of the 17 Conference on Retroviruses and Opportunistic Infections (CROI), 27 February3 March 2011, Boston.
- Doitsh G et al. Abortive HIV infection mediates CD4 T cell depletion and inflammation in human lymphoid tissue. Cell. 2010 Nov 24;143(5):789-801.
- Abdurahman S et al. Elevated flagellin-specific antibodies during HIV-1 infection. Poster abstract 309.
- Kamat A et al. Serological markers for IBD in AIDS patients with evidence of microbial translocation. Poster abstract 315.
- Nagy L et al. HIV-1 Modulation of APC Response to Pathogenic and Non-pathogenic Bacteria Contributes to Immune Activation. Poster abstract 316.
- Funderburg N et al. Delayed reduction in CD4 T cell turnover following viral control correlates with markers of microbial translocation in treatment-na patients receiving RAL-based ART: preliminary results from ACTG A5248. Poster abstract 318.
- Shive C et al. Immunologic failure despite suppressive ART is related to increased inflammation and evidence of microbial translocation. Poster abstract 320.
- Hunt P et al. Gut-associated lymphoid tissue fibrosis is associated with CD4+ T cell activation and poor HIV-specific CD8+ T cell responses during suppressive ART. Poster abstract 319.
- Lichtenstein K et al. Reduction of chronic inflammation occurs in selected patients with RAL intensification. Poster abstract 276.
- Gandhi R et al. RAL intensification of patients on long-term suppressive ART does not increase 2-LTR HIV DNA circles in PBMC: results from ACTG A5244. Oral abstract 51.
- Gandhi R et al. The effect of raltegravir intensification on low-level residual viremia in HIV-infected patients on antiretroviral therapy: a randomized controlled trial. PLoS Med. 2010 Aug 10;7(8). pii: e1000321.
- Massanella M et al. Dynamics of CD8 T cell activation during RAL intensification and after Its discontinuation in HAART-suppressed subjects. Poster abstract 281.
- Wilkins T et al. Effect of adding and removing MVC on immune activation in HIV+ patients on suppressive ART: results from ACTG A5256. Poster abstract 574.
- Hunt P et al. Immunomodulatory effects of MVC intensification in HIV-infected individuals with incomplete CD4+ T cell recovery during suppressive ART. Oral abstract 153LB.
- Piconi S et al. Immunomodulatory effects of hydroxychloroquine in HIV-infected ART-treated immunological non-responders. Poster abstract 382.
- Carsenti-Dellamonica H et al. LPS may be a predictive factor even in mild forms of HIV-associated neurocognitive impairment: sub-analysis of the Neuradapt Study. Poster abstract 404.
- Lyons J et al. Plasma-soluble CD14 is a strong predictor of impaired neurocognitive test performance in attention and learning domains in HIV-infected subjects. Poster abstract 405.
- Letendre S et al. Higher levels of phosphorylated Tau in CSF are associated with HIV infection, older age, antiretroviral use, and worse prospective memory. Poster abstract 406.
- Dahl V et al. Low levels of HIV-1 detected in the CSF of suppressed subjects Enrolled in a RAL Intensification Study and in Elite Controllers. Poster abstract 423.
- Price R et al. A pilot study of the effect of RAL intensification on CNS immunoactivation in virally suppressed subjects. Poster abstract 424.
- Sacktor N et al. Minocycline treatment for HAND: results from a multicenter trial. Poster abstract 421.