Work

Yes, It Is Time to Reconsider How We Rate Cognitive Impairments in HIV Disease

Year: 2013

Type: article

Abstract: Meyers and colleagues have provided us with an excellent review, data analysis, and proposal regarding criteria for cognitive impairments in HIV disease. They identified the effect of different individual criteria for defining impairment, and the effects of different “false positive” rates on classification. This discussion is critical for our understanding of the incidence, prevalence, and temporal trends in HIV associated neurocognitive disorders (HAND), and is more generally relevant to any neurobehavioral syndrome without a clear antemortem biomarker to confirm the presence of a neuropathological process. The analysis of any neurobehavioral syndrome naturally must occur at least two levels: clinical and pathological. Although some diagnostic criteria (e.g., those for Alzheimer’s disease (AD)) make this distinction explicit, in more common parlance the distinction between HAND (or Alzheimer’s dementia) and the neuropathology of HIV disease (or AD) may be lost. This is particularly problematic in HIV disease as the neuropathological hallmarks of HIV encephalitis (the pathological correlate of HIV Dementia), the combination of white matter pallor, perivascular chronic inflammation and multi-nucleated giant cells, are now rare. The search for antemortem biomarkers has been generally unsuccessful [1], likely reflecting the multiple pathways through which CNS dysfunction can occur in HIV disease. Until such time as we are able to establish “ground truth” with regard to the underlying causes of HAND, we are left with a clinical decision. This is where the paper of Meyers and colleagues becomes so important. The presence and impact of “false positives” in neurocognitive disorders, particularly in the mild spectrum, receive scant attention from researchers. The likelihood of these errors is compounded by two common practices to increase sensitivity to ever-milder neurocognitive abnormalities: (1) extensive test batteries, resulting in uncorrected multiple comparisons, and (2) higher cutoff scores, increasing the overlap between critical portions of test score distributions in individuals with and without disease. The “cost” of increased sensitivity is necessarily a reduction of specificity, a problem rarely explored with regard to neurocognitive classification. Meyers and colleagues have given us the most comprehensive and systematic investigation of the cost to specificity as a function of number of cognitive tests, domains, cut-scores, sample size, inter-test correlations, and other factors. They conclude that the safest way to classify individuals is to use an average standard score across all of the measures in a cognitive domain, to lower the threshold for mild impairment to −1.5 Z, and to limit the number of domains evaluated. We offer an additional consideration: the classification of Asymptomatic Neurocognitive Impairment may well be clinically irrelevant. By definition, individuals without symptoms will not seek medical attention. The individuals with ANI might thus represent the lower end of the normal distribution of function. One of the difficulties of this discussion, however, is the apparent premise that there should be no degree of cognitive dysfunction among individuals without HIV infection. That is, all other things being equal, the goal appears to be to restrict “false positives” to as close to zero as is possible without sacrificing “true positives”, a feat made more difficult by the absence of a neuropathological standard. Unfortunately, as noted by Meyers and colleagues, there is a range of factors that can affect individuals’ test performance, and that may result in mildly (or less frequently, severely) impaired cognitive functions. The attribution of the cause of cognitive impairment was considered important even early in the epidemic [2, 3] and is now part of the diagnostic classification scheme [4]. These non-HIV-related causes can include histories of drug and alcohol use, traumatic brain injury, affective disorders, and the like. Thus, in addition to the problem of non-differential misclassification, we also have a problem with the attribution of risk. This problem becomes even more confounded in HIV disease as infected individuals enter middle and older age. In addition to the effects of HIV-infection on brain structure and function, now the normal consequences of aging - hypertension, diabetes, vascular disease, etc. - begin to take their toll. Even assuming a relatively low rate of true HIV-related neuropathology for individuals over the age of 65, for example 5%, when added to the pre-existing “false positive” rate (due to other pathologies, both age- and non-age-associated) this decreases even more the sensitivity to any HIV-related dysfunction, and may minimize the detectable efficacy of any intervention. When these factors are coupled with the change in the treated history of individuals with HIV disease, so-called cohort effects, the attribution of risk and interpretation of interventions becomes even more complicated. Given the complexity of problem, some might argue that it is simply too much to deal with, and that the only solution is to either abandon clinical trials related to HAND or to focus entirely on preclinical models. We strongly reject this sort of argument because we reject the premise that the problems are insurmountable. Meyers and colleagues have carefully explored one aspect of the problem – the classification of cognitive impairment - and they provide us with a specific target as they propose clear cutoff scores. This should lead to further investigation of the problem among larger cohorts that include neuropsychological evaluations, all of which should lead to greater clarity in interpreting the results of projects being completed in the post-HAART era. This is but one step in one part of the process, but it is an important step. We are all very grateful to our colleagues for pushing us forward.

Source: Neuroepidemiology

DOI: 10.1159/000355128

Authors: James T. Becker, Beth E. Snitz

Institution: University of Pittsburgh

Language: English

Topic: HIV Research and Treatment

Subfield: Virology

Field: Immunology and Microbiology

Domain: Life Sciences

SDG: Good health and well-being

Open Access status: bronze

Funders: National Institutes of Health, National Institute on Aging, National Institute of Mental Health, National Institute of Allergy and Infectious Diseases