HIV Articles  
Benefit or Toxicity from Neurologically Targeted Antiretroviral Therapy? EDITORIAL COMMENTARY
  Clinical Infectious Diseases March 15 2010;50:930-932
Bruce J. Brew
St. Vincent's Centre for Applied Medical Research Centre, Department of Neurology and HIV Medicine, St. Vincent's Hospital, Darlinghurst, New South Wales, Australia
The treatment, and indeed prevention, of HIV-associated neurocognitive disorders [1] is becoming an increasing concern. These disorders have not disappeared with the widespread use of highly active antiretroviral therapy (HAART). Indeed, although the incidence of HIV-associated dementia (a more severe form of HIV-associated neurocognitive disorders) has halved, the prevalence has approximately doubled [2]. Milder forms of HIV-associated neurocognitive disorders are more common than HIV-associated dementia and have not changed with the use of HAART, despite undetectable HIV RNA level in plasma and cerebrospinal fluid, thereby excluding poor adherence as a cause [3, 4]. Indeed, in the AIDS Clinical Trials Group Longitudinal Linked Randomized Trials study [5], 458 (39%) of 1160 patients were found to be impaired at baseline, 217 (22%) of 991 patients remained impaired during the entire study, and 128 (21%) of 615 patients became impaired during the study despite optimal medical management.
There is skepticism within the HIV-treating community as to whether these milder deficits amount to anything of clinical significance, with some in the community predicting an increased rate of mortality [6], job loss [7], a decrease in the ability to drive [8], a potential increased susceptibility to neurodegenerative diseases [9], and poor adherence to treatment [10]. This last prediction (ie, poor adherence treatment) is perhaps the most important because it underpins at least some of the later consequences through the promotion of resistance to antiretroviral drugs (ARVs) and viral escape [11].
Why should HIV-associated neurocognitive disorders remain and continue to occur despite HAART? Indeed, there is a paradoxical benefit: the milder forms of cognitive impairment continue in the face of powerful ARVs, whereas more severe impairment has responded to such potent drugs. There are several possibilities. First, there may be a "legacy" effect, namely, that the impairment reflects residual damage from the pre-HAART era, when therapies had limited efficacy. This is unlikely to be a major explanation. Mild impairment can fluctuate with periods of normal function for about 12% of patients, even for those who are virally suppressed [4, 12]. More importantly, HIV-associated neurocognitive disorders may occur subsequent to starting HAART [3-5, 13]. Although an HIV-associated neurocognitive disorder might be a manifestation of an immune restoration disorder for some patients, this is unlikely to be the case for the majority of patients, because its frequency exceeds that associated with HIV-associated dementia related to immune restoration disorder and because its occurrence is temporally remote from the commencement of HAART [14]. Second, persistent impairment may result from another condition. However, such confounders have been excluded in several studies [4, 13]. Third, some ARVs may be toxic to the central nervous system (CNS). Despite abundant evidence that nucleoside reverse-transcriptase inhibitors can disturb mitochondrial function [15] and that protease inhibitors can disturb proteosomal function [16], leading to neuropathy and lipodystrophy, there is little evidence of CNS toxicity from ARVs. By use of magnetic resonance spectroscopy (MRS), Schweinsburg et al [17] did show that patients had lowered N-acetylaspartate levels when treated with stavudine and didanosine, indicating the possibility of neuronal damage.
Similarly, Marra et al [18] observed some deterioration in the neuropsychological function of patients being tested while taking ARVs with better penetration into the CNS (ie, "neuro-HAART"). Both studies, however, had small numbers of patients. Fourth, for some patients, there may be ongoing low-grade brain infection, despite plasma and cerebrospinal fluid viral loads being below detection. This is likely the most important reason, because it is known that ARVs vary in their ability to cross the blood brain barrier [19]. Indeed, of the current ARVs, only nevirapine, indinavir, lopinavir, amprenavir, abacavir, zidovudine, stavudine, emtricitabine, darunavir, raltegravir, and maraviroc have cerebrospinal fluid levels greater than the half maximal inhibitory concentration [20-23].
Thus, there is theoretical concern at least that some patients with mild HIV-associated neurocognitive disorders, those with subclinical HIV-associated neurocognitive disorders, and those at risk for HIV-associated neurocognitive disorders may not be receiving an optimal HAART regimen (ie, neuro-HAART). It is in this context that the study by Winston et al [24] in this issue of journal is particularly important, in terms of study design and therapy. Winston et al [24] demonstrate that a relatively complex substudy can be feasibly added to a larger parent study. Previously, such neurological substudies have been hampered by concerns regarding the complexity of the assessment tools. Winston et al [24] have clearly shown that studies using tools such as abbreviated neuropsychological tests and magnetic resonance imaging (MRI) brain scans are feasible, even with multicenter international studies. With regard to therapy, their finding that a neuro-HAART regimen results in a better outcome for neuropsychological functioning but an inferior outcome for assessing cerebral metabolites with MRS is important but open to several interpretations.
One possible interpretation, which was suggested by Winston et al [24], is that neuro-HAART is a "mixed blessing." It is beneficial neuropsychologically but toxic at a cerebral metabolism level. However, there are other possibilities apart from those 2 outcomes being correct; for example, both outcomes may be inconclusive, or only one may be correct. The results may be inconclusive because of the relatively small number of patients, and we must remember that only some (not all) patients develop HIV-associated neurocognitive disorders. Additionally, there is the issue of multiple comparisons without statistical correction confounding a clear therapeutic signal.
Alternatively, it is possible that neuro-HAART was beneficial and that the MRI changes were misleading. The patients in each of the arms may have been mismatched neuropsychologically. To enter the study, patients had to be free of HIV-associated dementia, but it is possible that some had milder forms of neuropsychological impairment that went unrecognized. As such, there may have been a mismatch between the arms of the study: patients with subclinical HIV-associated neurocognitive disorders may have been present in greater numbers in the neuro-HAART arm but may have had less neuronal reserve (as shown by the N-acetylaspartate-to-creatine ratio on an MRI brain scan) to show improvement. Moreover, the significance of the greater inflammatory changes (as shown by the myo-inositol-to-creatine ratio on an MRI brain scan) in the neuro-HAART may reflect the so-called good inflammation (namely, for repair).
Conversely, it is possible that neuro-HAART was toxic and that the neuropsychological tests were misleading. The neuropsychological benefit may have occurred as a result of the practice effect. However, the CogState computerized cognitive test battery has been specifically developed to minimize such an effect.
How are the findings of Winston et al [24] to be placed in clinical context? Most studies of neuro-HAART point to its benefit [25, 26]. The concern over the possibility of toxicity is reasonable, but the evidence is sparse. Nonetheless, the only way to resolve this complex problem will be to conduct a randomized, controlled clinical trial on this subject.
  iconpaperstack view older Articles   Back to Top