Effects of HIV-1 infection and aging on neurobehavioral functioning

preliminary findings

Abstract

Objective: 

The effects of aging on the presentation of HIV-associated neurocognitive disorders are largely unknown. In a cross-sectional observational study, we compared the neuropsychological profiles of 67 HIV-positive patients aged at least 50 years with those of 52 participants aged 35 years or less.

Methods: 

Participants received neuropsychological, psychiatric and neuromedical evaluations. Raw neuropsychological test scores were converted to demographically corrected T-scores; all were corrected for the effects of normal aging. Clinical ratings of impairment were assigned to the neuropsychological results.

Results: 

The two groups did not differ statistically with respect to demographic variables, percentage with AIDS, or CD4 cell counts. The ‘younger’ group had higher viral burdens in plasma and cerebrospinal fluid (CSF), and fewer were receiving antiretroviral treatment. The proportion of neuropsychologically impaired subjects in the ‘older’ group was slightly greater than in the younger group, and the older group tended to have higher rates of impairment across most ability domains. When group differences in CSF viral load were modeled statistically, both viral burden and age were significant predictors of neuropsychological impairment, together with a significant interaction between viral burden and age. Older individuals with detectable virus in CSF had twice the prevalence of neuropsychological impairment of those with undetectable levels. Among younger individuals, this proportion was not affected by viral load. Lifetime major depression, substance use disorder, and current depression symptoms were not associated with neuropsychological impairment.

Conclusion: 

Although further studies with larger and older samples are needed, this investigation suggests that older adults may be at greater risk of HIV-related neurocognitive dysfunction.

Introduction

With the advent of more efficacious antiretroviral treatment, the number of people infected with HIV who are living to middle and older age continues to increase. It is estimated that more than 90 500 adults over the age of 50 years are currently living with AIDS in the United States [¹]. In addition, the most recent (1998) survey of publicly funded test sites in the USA showed that 6.6% of new infections reported were in individuals aged 50 years or older [²].

Older age is associated with a number of factors that may complicate the course of HIV disease, including its neurocognitive manifestations. These include a greater incidence of cardiovascular, cerebrovascular, and metabolic morbidity (e.g. hypercholesterolemia, diabetes), the side-effects of medications used to treat such illnesses, and a less resilient immune system. Older age at seroconversion has been associated with a significantly increased risk of progression to AIDS [^3–9], including the development of Kaposi's sarcoma [¹⁰], perhaps as a result of age-related declines in thymic activity [¹¹] and immunocompetence [^12,13]. Older age has also been associated with poorer immune reconstitution in patients receiving highly active antiretroviral therapy (HAART) [^14,15]. Finally, older HIV-infected individuals are thought to be at risk of depression syndromes, which arguably might be associated with impaired performance on neuropsychological testing.

The effects of aging on the presentation of HIV-related neurocognitive disorders are largely unknown. Both normal aging and HIV disease are associated with declines in certain neurocognitive abilities. If aging can be seen as a condition of decreasing cognitive reserve [^16–18], then it is reasonable to speculate that aging in the context of HIV/AIDS may confer increased vulnerability to brain dysfunction.

An early epidemiological study in the pre-HAART era [¹⁹] showed that clinician-diagnosed HIV encephalopathy was three times as prevalent in elderly (≥ 75 years old) individuals compared with adolescents and adults under the age of 35 years, although measures of cognition and data on the differential diagnosis of other dementing illnesses were not available. Work from the Multicenter AIDS Cohort Study [¹⁶] failed to find an interaction between the effects of age and HIV status in a small group of individuals over the age of 55 years. Unpublished work from groups at the University of California at Los Angeles and the University of Miami has suggested that older age may play a role in the development of HIV-associated cognitive and functional deficits, particularly in patients with AIDS [^20,21].

The present study compared the neuropsychological profiles of older and younger HIV-infected adults in order to examine whether aging affects the likelihood or presentation of HIV-associated neurocognitive disorder. In addition, we explored the possibility of age-related differences in the relationship of viral burden and HAART with neurocognitive deficits.

Methods

Subjects

The subjects were 119 HIV-positive volunteers participating in studies of the neurobehavioral consequences of HIV/AIDS at the University of California, San Diego (the HIV Neurobehavioral Research Center, the California NeuroAIDS Tissue Network, and the program project on NeuroAIDS Effects of Methamphetamine). Individuals were excluded if they had any non-HIV-related conditions of sufficient severity to affect their neuropsychological functioning (e.g. developmental disabilities, psychosis, neurological or metabolical disorders, non-remote substance dependence or current substance abuse). Participants were selected for the ‘older’ group (n = 67) on the basis of having been at least 50 years old at the time of their visit; those in the ‘younger’ group (n = 52) were required to be no older than 35 years. From a pool of 377 younger individuals, the current sample was chosen to match the older group as closely as possible with respect to years of education, proportion of non-white participants, sex, and stage of HIV disease (see Table 1 and Table 2). The maximum age of participants in the older group was 67 years. Their education ranged from 8 to 19 years. The minimum age of participants in the younger group was 20 years, with their education ranging from 10 to 18 years.

Table 1: Background characteristics by age group (no statistically significant differences besides age).

Table 2: Baseline medical characteristics by age group.

Procedure

Neurobehavioral assessment

The methods of neurobehavioral assessment have been described in detail elsewhere [²²]. Briefly, all study participants completed neurocognitive evaluations using a comprehensive neuropsychological battery designed to assess functioning in the areas of attention/working memory, speeded information processing, learning, recall, verbal fluency, abstraction/problem solving, and motor ability. The Appendix lists the tests used to assess each domain of cognitive functioning. Participants tested before June 1999 received a battery that was approximately 4 h long, whereas those tested after that date received a battery that took approximately 2 h to complete.

Raw test scores were converted to T-scores (standard scores with a mean of 50 and standard deviation of 10) using demographically corrected norms to account for the effects of age, education, sex, and ethnicity, as appropriate and available for each measure. A neuropsychologist routinely rated cognitive impairment according to a subject's performance in each neuropsychological domain, as well as a rating of global functioning, without being aware of his/her HIV status. These clinical ratings are assigned on a nine-point scale in which: 1, above-average functioning; 2, average; 3, below average; 4, borderline; 5, definite mild impairment; 6, mild–moderate impairment; 7, moderate impairment; 8, moderate–severe impairment; and 9, severe impairment [²³]. A rating of global impairment required at least two ability domains with ratings in the impaired range.

Psychiatric evaluation

As part of their neurobehavioral assessment, participants were evaluated for major depression and substance use disorders using either the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) [²⁴] or the Psychiatric Research Interview for Substance and Mental Disorders [²⁵], depending upon the parent research protocol. Both instruments were administered by trained clinicians to establish diagnoses using criteria set forth by DSM-IV [²⁶]. In addition, depression symptoms occurring during the week before testing were assessed using the Beck Depression Inventory (version I) [²⁷].

Neuromedical evaluation

Participants underwent a general physical examination as well as a detailed neurological examination administered by a physician or trained research nurse. Blood samples were obtained to verify HIV status, CD4 cell count (and other cell populations), and plasma levels of HIV-RNA burden. The HIV-RNA level in the cerebrospinal fluid (CSF) was ascertained for 78 cases. RNA levels were obtained using the Amplicor HIV-1 Monitor Test (Roche Molecular Systems, Inc., Branchburg, NJ, USA).

Statistical analyses

Group differences in continuous outcome variables were analysed using two-tailed t-tests. For discrete variables we used χ² tests. In addition, analyses of variance and logistic regression were used to test the main effects and interactions between the age group and variables of interest on neuropsychological functioning.

Results

Background variables

As mentioned in the Subjects section, the groups were comparable with respect to demographic variables (see Table 1), the proportion of subjects with AIDS, and Centers for Disease Control (CDC) classifications based on 1993 guidelines [²⁸]. However, the younger group had higher levels of both plasma [t₍₁₀₅₎ = −3.5, P = 0.0007] and CSF HIV RNA [t₍₇₆₎ = −3.1, P = 0.003]. The majority of participants in both groups were receiving HAART, but the proportion of untreated subjects tended to be greater in the younger group (overall χ² = 6.4, P = 0.10). The duration of HIV infection was estimated on the basis of self-reported dates of either the last antibody negative test or the first positive test, or the likely date of exposure when the former were not available. The older group had an average estimated duration of infection of 12 years (sd = 5.9), compared with 8 years for the younger group (sd = 5.6) [t₍₁₀₀₎ = 3.6, P < 0.0005].

Psychiatric variables

Both groups obtained mean Beck Depression Inventory scores in the mildly depressed range. DSM-IV psychiatric diagnoses were available for 47 older and 27 younger subjects, as this assessment was omitted during some years because of protocol changes. There were no significant differences with regard to the lifetime or 12-month prevalence of major depressive disorder, although there was a somewhat greater proportion of older participants who met criteria for a current episode of the disorder (see Table 1). Likewise, there were no significant differences in the lifetime prevalence of substance use disorders, although a somewhat greater proportion of older participants reported lifetime substance use disorders. Any substance: lifetime: 73% of older versus 50% of younger subjects, 12-month: 23 versus 11%; alcohol: lifetime: 56 versus 48%, 12-month: 12 versus 7%; cannabis: lifetime: 44 versus 19%, 12-month: 5 versus 4%; cocaine: lifetime: 44 versus 26%, 12-month: 9 versus 0%; amphetamines: lifetime: 40 versus 33%, 12-month: 7 versus 4%; sedatives: lifetime: 16 versus 19%, 12-month: 2 versus 0%; opioids: lifetime: 25 versus 19%, 12-month: 7 versus 4%; hallucinogens: lifetime: 12 versus 19%, 12-month: 2 versus 0%; other: lifetime: 0 versus 7%, 12-month: none.

Neuropsychological variables

The length of the neuropsychological battery (4 versus 2 h) did not influence the prevalence of neuropsychological impairment; therefore the results were combined. The majority of participants were assessed using the 2 h neuropsychological battery (73% of older and 65% of younger subjects). The proportion of participants found to be neuropsychologically impaired with each battery was not statistically different (64% with the 2 h versus 50% with the 4 h, χ² = 1.98, P = 0.16). Also, this modest difference corresponds to the fact that recruitment in recent years favored participants with more advanced disease. Among those who took the more recent 2 h battery, 84% had AIDS, compared with 47% for the 4 h battery. As would be expected, individuals with AIDS were more likely to be impaired (67%) than those without AIDS (41%), irrespective of age. However, there was little difference in the prevalence of neuropsychological impairment between the older and younger groups within batteries (9% greater for the older group with the 2 h and 11% greater for older group with the 4 h), and no difference that cannot be explained on the basis of disease status. Groups were therefore combined across test batteries, as both assessments were designed to measure the same abilities and have a roughly equivalent sensitivity to impairment.

Although there was a slight tendency for the older group to show higher overall rates of neuropsychological impairment, globally and on most ability domains (see Fig. 1), none of these differences between the groups were statistically significant.

Fig. 1.: Rates of global and domain-specific neuropsychological impairment by group. NP, Neuropsychological ░ Older; □ younger.

When participants with either a recent substance use disorder (typically abuse rather than dependence) or major depression diagnosis, or both, were removed from analyses, the rates of impairment in each group did not change substantially. The difference in prevalence of impairment in the older group remained approximately 10% higher than in the younger group. To illustrate, in a logistic regression, a lifetime or 12-month diagnosis of major depression did not predict the presence of neuropsychological impairment (lifetime: whole model test χ² = 1.93, P NS; group effect χ² = 0.09, P NS; major depression χ² = 0.81, P NS; interaction χ² = 1.35, P NS. 12-month: whole model test χ² = 1.31, P NS; group effect χ² = 0.05, P NS; major depression χ² = 0.02, P NS; interaction χ² = 1.16, P NS).

Neuromedical variables

Because the groups differed in their viral load, and CSF viral load has been associated with neuropsychological deficits [²⁹], the higher viral load in the younger group may have attenuated group differences in neuropsychological impairment. In order to test this possibility we employed a logistic regression with neuropsychological impairment as a dichotomous outcome and log₁₀ CSF RNA value and group membership as independent variables. When CSF viral load was included in the model, both age and viral load were found to be significant predictors of neuropsychological impairment (χ² = 3.84, P = 0.05 and χ² = 4.91 P = 0.03, respectively). This analysis also permitted us to investigate possible age differences in the relationship of viral burden to neuropsychological impairment. With the logistic regression described above, we found a significant interaction for the effects of age × CSF viral burden (χ² = 4.41, P = 0.04). A plot of the interaction shows that whereas CSF HIV-RNA levels were similar between younger neuropsychologically impaired and neuropsychologically normal subjects, among older subjects, the CSF viral burden was higher in those who were neuropsychologically impaired than in those who were neuropsychologically normal (see Fig. 2). This relationship was not duplicated for the plasma viral burden.

Fig. 2.: Interaction between age group and viral burden showing that levels of cerebrospinal fluid HIV RNA differ according to neuropsychological impairment in older subjects but not in younger subjects. CSF, Cerebrospinal fluid. —— Neuropsychologically normal; ▪ neuropsychologically impaired.

Table 3 summarizes the results of similar logistic regressions for all neuropsychological ability domains. In these analyses, age group was a significant predictor of impairment in abstraction, attention/working memory, and learning; in all cases older age was associated with more impairment. Interestingly, impairment of these same three ability domains was also predicted by both the CSF viral load and the interaction of age group with viral load. Also consistent with the pattern shown for global neuropsychological functioning in Fig. 2, all of these interaction effects reflected a greater association between CSF viral burden and neuropsychological impairment in the older group.

Table 3: Results of logistic regressions using age group membership and cerebrospinal fluid viral burden as predictors of impairment in seven neuropsychological ability domains.

To illustrate the interaction between age group and CSF viral burden, we divided each group according to whether the CSF viral load was in the detectable or undetectable range, and compared the proportions of neuropsychologically impaired individuals. Among the older group, 81% of those with detectable CSF HIV RNA were neuropsychologically impaired, compared with 42% of those with undetectable levels (χ² = 6.16, P = 0.01). Among the younger group, these proportions were 54 and 57%, respectively (χ² = 0.03, NS). That is, in the older group, detectable levels of virus in the CSF doubled the likelihood of neuropsychological impairment, whereas in the younger group the rates of impairment were similar regardless of the CSF viral load.

As the older group had, on average, been infected for a little over 4 years longer than the younger group, it is possible that the apparent lack of a relationship between CSF viral load and neuropsychological impairment in the younger group resulted from ‘transient’ CSF infection. The theory here is that earlier in the course of infection, virus detected in the CSF reflects trafficking across the blood–CSF barrier from plasma, whereas later in the infection CSF virus is likely to be ‘autonomous', or derived from independent HIV replication in the central nervous system (CNS) [^30,31]. Repetitive re-seeding of the CNS compartment by HIV over a longer period in older individuals may increase the size of the CNS reservoir, raising the probability of autonomous CNS infection and associated neurocognitive impairment. A crude surrogate marker for estimating the source of viral RNA in the CNS is pleocytosis, or elevated white blood cell numbers in the CSF. We therefore compared the two age groups for the presence or absence of pleocytosis using a cutpoint of 5 or more cells/mm³ of CSF, and found that 5% of the older and 12% of the younger participants showed pleocytosis at the time of their visit. This difference did not reach statistical significance.

We also examined the possibility of age group differences in the relationship between the viral burden, effects of HAART, and neuropsychological impairment. HAART appeared to be equally effective in reducing the viral burden in plasma and CSF in both age groups. Older subjects on HAART had mean plasma HIV-RNA levels of 2.82 (sd = 0.99) log₁₀ copies, compared with 4.08 (sd = 1.11) for those not on HAART. Among the younger subjects grouped according to HAART status, the mean plasma viral loads were 3.38 (sd = 1.31) and 5.10 (sd = 0.92) log₁₀, respectively [age group effect F_(1,103) = 11.93, P = 0.0008; treatment effect F_(1,103) = 42.42, P < 0.0001; interaction F_(1,103) = 1.0, NS]. With respect to virus in the CSF, older subjects on HAART had a mean of 1.98 (sd = 0.42) log₁₀ copies of HIV RNA versus 2.65 (sd = 0.61) for those not on HAART. In the younger subjects, these figures were 2.58 (sd = 1.08) and 2.89 (sd = 0.86) log₁₀ copies, respectively [age effect F_(1,76) = 4.59, P = 0.04; treatment effect F_(1,76) = 6.38, P = 0.01; interaction F_(1,76) = 0.90, P NS].

We then examined the effects of HAART on neuropsychological impairment by age group. Among the older group, subjects not receiving HAART were 19% more likely to be neuropsychologically impaired than those on HAART (76 versus 57 %), compared with a 2% difference among the younger participants (54 versus 52%). However, in a logistic regression, the interaction between age group and treatment status was not statistically significant in predicting neuropsychological impairment (χ² = 0.91, P = 0.34) (see Fig. 3).

Fig. 3.: Proportion of neuropsychological impairment according to antiretroviral treatment by age group. NP, Neuropsychological. ▒ Highly active antiretroviral therapy (HAART); □ non-HAART/no antiretroviral therapy.

Discussion

In this study we set out to explore the possibility that aging in the context of HIV infection may render individuals more vulnerable to neurocognitive dysfunction. We found some evidence that this may be the case, in that above a threshold of CSF viral load, older individuals were more likely to be neurocognitively impaired than their younger counterparts. This would be consistent with theories of cognitive reserve and with the general understanding that aging increases susceptibility to a variety of health insults. That neurocognitive impairment was not related to current depression symptoms or to lifetime histories of major depression or substance use disorders is consistent with earlier work [^32,33], and the mild severity of depression symptoms. We emphasize that this is a preliminary study of a complex topic, and it has some significant limitations. We used convenience samples of HIV-positive participants who were not recruited to investigate the possibility of an interaction between age and HIV on CNS and neurobehavioral functioning. As such, the samples are relatively small, with the associated limitations in power to detect anything but substantial group differences in the neurobehavioral outcome variables.

Although we used age-corrected norms to account for normal aging, the ideal design for detecting an interaction between aging and HIV effects should also include older and younger HIV-negative controls. In addition, as this is a cross-sectional study, we cannot rule out the possibility that some of the findings (or lack of findings) were caused by cohort differences unrelated to age per se. We did make efforts to minimize such differences by carefully selecting a younger HIV-positive comparison group that was similar to the older group with respect to other demographic variables (sex, educational level, ethnicity) and stage of HIV disease. The groups were also similar regarding a reported history of previous substance use disorders, and individuals with other potential confounds were excluded from both groups. Nevertheless, the groups were not comparable in other potentially important ways, such as the estimated duration of HIV infection, current antiretroviral treatment status, and viral loads. Although we did attempt to deal with these factors in multivariate statistical modeling of the results, the fact remains that ‘older age’ in this study may be confounded with other subject or disease variables that cannot be understood or adequately controlled for in a strictly cross-sectional study. An additional important limitation of this preliminary study is that our ‘older’ group was not very old. The fact that our group of over 50-year-olds had a mean age of 53.3 years shows that only a few (four) were aged 60 years or older. Clearly, future studies of this type should include much more substantial numbers of participants in their 60s and beyond. It is quite possible that ‘age effects’ will be more robust in samples containing larger proportions of people with HIV infection who are older than 60 years of age.

Ultimately, what is needed to answer the questions that we have begun to explore is a larger-scale longitudinal study of younger and older HIV-positive and HIV-negative adults. Larger sample sizes are required to understand possibly significant cohort differences and to provide the statistical power to detect potentially complex age effects that may be important but are not necessarily large. The longitudinal component is needed to improve the strictly correlational nature of a cross-sectional design, and provide more secure insights into cause–effect relationships.

With these caveats in mind, the current results suggest that older HIV-infected individuals may indeed have at least a somewhat greater risk of or vulnerability to CNS complications. Despite the fact that our younger group contained more untreated individuals and had higher average viral burdens in both plasma and CSF, slightly fewer of them evidenced neuropsychological impairment (54 versus 64%). When the CSF viral burden and age group were considered together in a multivariate prediction model, both were found to be significant predictors of neuropsychological impairment, with a provocative suggestion that older age may confer increased sensitivity to the presence of virus in the CNS with regard to neurocognitive dysfunction.