| |
Neurocognitive Impact of Substance Use in HIV Infection
|
| |
| |
Download the PDF here
From Jules: I don't really buy these results, I think it's like that if the study stratified by looking at patients with a history of heroin or meth use they would surely have found lingering neurocognitive deficits particularly at meth users, and longevity of use would matter as well.
JAIDS Journal of Acquired Immune Deficiency Syndromes: POST ACCEPTANCE, 30 June 2011
"On the basis of comprehensive neurocognitive testing HIV+ participants with either DSM-IV based SU disorders or self-reported histories of Non-syndromic, illicit substance use performed no differently, on average, than a carefully matched group of HIV+ participants without any indication of significant lifetime substance use. Similarly, participants with SU histories demonstrated neither a higher rate of HAND diagnoses nor more functional difficulties in everyday life.
This finding was unexpected, given the multiple studies that have linked abuse of specific substances, particularly alcohol (47) and methamphetamine (48) as well as polydrug use (49) to NC impairment; and the indications both from preclinical and clinical research suggesting that alcohol and other drugs may augment the neurotoxicity of HIV. The divergent findings may reflect the differences between the design of our study and most other studies which have examined HIV effects in the context of substance use. We compared NC function in HIV-infected groups with three levels of SU rather than comparing HIV-positive to HIV-negative persons with similar drug histories (30, 52). Two prior studies with similar design to ours reported more cognitive impairment among HIV-infected stimulant users relative to HIV-infected non- drug users (25, 53). However, participants in both studies had acute or recent use (within 24 months) and one study applied a less traditional cognitive NC task as the outcome measure (53). Thus, the discrepant findings between these two studies and ours may reflect differences in abstinence periods and NC measures.........Increased lifetime use of heroin was weakly associated with worse recall and working memory (both p values <.01). On the other hand, increased use of cocaine and cannabis was weakly associated with improved verbal fluency (p<.01)......Few studies have examined cognitive recovery from SU in the context HIV (59), but the process varies greatly in degree and course among people without HIV and for specific substances of abuse.......Our negative findings might also reflect type and severity of substance abuse in CHARTER as compared to some other studies. As one example, heavy methamphetamine exposure has been linked to greater likelihood of NC impairment, and neuroimaging indicators of brain dysfunction in HIV+ persons (25, 68, 69). However, in our study only about 12 % of subjects met lifetime methamphetamine dependence criteria......This study raises important questions. What are the best methods for summarizing complex polysubstance use in neuroAIDS research? What aspects of substance use are most relevant to functional impact: type(s), age of onset, duration, recency, etc.? How does the order and overlap of exposure to substances and HIV infection influence neuropathogenesis? What biomarkers are relevant in the study of HIV and SU? Finally, does illicit SU lead to more rapid neurocognitive decline or to a unique pattern or course of NC functioning over time?
In conclusion, while methodological challenges remain in cognitive neuroAIDS research with substance users, our study provides assurance that the impact of remote SU disorders and Non-syndromic use may be subtle, at least in the absence of acute or recent use/abuse. While this finding does not suggest neurocognitive safety of past substance use, it does indicate that substance use history, even when complex, does not necessarily merit exclusion of patients in short-term analyses of cognition in neuroAIDS research."
Byrd, Desiree A PhD; Fellows, Robert P MA; Morgello, Susan MD; Franklin, Donald BS; Heaton, Robert K PhD; Deutsch, Reena PhD; Atkinson, J Hampton MD; Clifford, David B MD; Collier, Ann C MD; Marra, Christina M MD; Gelman, Benjamin MD, PhD; McCutchan, J Allen MD, MSc; Duarte, Nichole A M.A.; Simpson, David M MD; McArthur, Justin MBBS, MPH; Grant, Igor MD; for the CHARTER group
Abstract
BACKGROUND: To determine how serious a confound substance use (SU) might be in studies on HIV-associated neurocognitive disorder (HAND) we examined the relationship of SU history to neurocognitive impairment (NCI) in participants enrolled in the CNS HIV Antiretroviral Therapy Effects Research (CHARTER) study.
METHODS: After excluding cases with behavioral evidence of acute intoxication and histories of factors that independently could account for NCI (e.g., stroke), baseline demographic, medical, SU, and neurocognitive data were analyzed from 399 participants. Potential SU risk for NCI was determined by the following criteria: lifetime SU DSM-IV diagnosis, self-report of marked lifetime SU, or positive urine toxicology (UTOX). Participants were divided into three groups: no SU (N = 134), Non-syndromic SU (N = 131), syndromic SU (N = 134) and matched on literacy level, nadir CD4, and depressive symptoms.
RESULTS: While approximately 50% of the participants were diagnosed with HAND, a MANCOVA of neurocogntive summary scores, covarying for UTOX, revealed no significant effect of SU status. Correlational analyses indicated weak associations between lifetime heroin dosage and poor recall and working memory, as well as between cannabis and cocaine use and better verbal fluency.
CONCLUSIONS: These data indicate that HIV neurocognitive effects are seen at about the same frequency in those with and without historic substance abuse, in cases that are equated on other factors that might contribute to NCI. Therefore, studies on neuroAIDS and its treatment need not exclude such cases. However, the effects of acute SU and current SU disorders on HAND require further study.
Introduction
HIV infection and abuse of alcohol and illicit substances are independently associated with neurologic, neurocognitive, and neuropathologic complications (1-6). When concurrent, drugs of abuse can exacerbate the neuropathogenesis of HIV by suppressing immunity and increasing viral replication in the central nervous system (7- 9). Some research has confirmed synergistic deleterious effects of substance use (SU) on neurocognitive functioning and the progression of HIV-related neurocognitive disorders (10-15). Given these findings, some investigators have questioned the appropriateness of including substance abusers in neurocognitive HIV studies. Importantly, the number of HIV-infected persons with significant substance use histories has been increasing since the 1980's (16). To exclude these individuals from participation in neuroAIDS research would limit understanding of the neurological manifestations of HIV in SU patients and decrease generalizability of findings.
Despite concerns regarding the synergy of HIV infection and substance use, studies have produced conflicting results (17-26). Primary contributing factors to these conflicting results are methodologic and include heterogeneity of study designs, variable comparison groups (e.g., normative data sets, matched seronegatives), methods of SU history ascertainment (e.g., self-report versus objective measurement; syndromic diagnoses versus report of misuse), small sample sizes, and importantly, lack of control for other impactful co-factors (e.g., literacy level, depression, and other comorbid conditions with potential CNS effects; 27). Further, most studies of mono-drug abuse and HIV-related cognitive functioning fail to control for past or present use of other substances, even though published data confirm that substance users typically use multiple substances over their lifetimes and even simultaneously (28, 29). Even neurocognitive studies in which polysubstance users with and without HIV were well matched demographically have yielded mixed results (30, 31). Thus, the appropriateness of including substance abusing participants in cognitive research remains uncertain.
To date, no study of a large, multi-site, diverse HIV-infected cohort has examined the impact of substance use histories on neurocognitive functioning while simultaneously considering other relevant demographic and medical factors. To address this limitation, the present study selected cases from the CHARTER study that differed in substance use status, but were relatively free of other comorbidities (e.g., head injuries) that might confound interpretation of the contribution of substance use to NC function. The aim of the current study was to determine whether persons with histories of illicit substance use deserve special inclusion consideration in cognitive neuroAIDS research. We hypothesized that: 1) HIV-infected participants with histories of prior SU disorders and/or recent use of illicit substances would be more neurocognitively (NC) impaired than HIV-infected patients without such histories, and 2) cumulative lifetime dose of substances would be negatively correlated with NC functioning.
Results:
Demographic and medical characteristics
Demographics from the matched SU groups are presented in Table 1. As planned, the no SU, Syndromic SU and Non-syndromic SU groups were similar on most relevant demographic and medical variables, including duration of HIV infection, CART status and adherence levels (all p values >.01); the exceptions were proportion with hepatitis C infection (9.8%, 30.8%, and 16.4%, respectively, p<.001) and years of education (13.5, 12.5, and 13.3, respectively, p<.01). Of note, these variables had minimal effect on the overall demographically adjusted test scores (see Appendix 1) and thus were not covaried in analyses.
Because the matching process eliminated many subjects from the overall Syndromic SU group, the Syndromic SU subset (n = 134) that was selected for analysis was compared to the larger unselected Syndromic SU group (n = 808). The selected subset was similar to the larger group on age, education, WRAT-3 scores, nadir CD4, BDI scores, urine toxicology, Hepatitis C, and average number of SU diagnoses (all p values >.01). However, the selected Syndromic SU group had a smaller proportion of Hispanics (6.3% vs. 16.4%, respectively, p <.001). Analyses comparing NC test performance of these two groups revealed no statistically significant differences in global or domain T-scores (all p values >.01).
When the Non-syndromic SU group was compared to the Syndromic SU group on substance use characteristics, more persons in the Syndromic group had history of intravenous drug use (36 % vs 10%, p <.001) but equal proportions had positive urine toxicology for illicit substances at the time of testing (16% vs 23%, p > .01). The Syndromic SU group reported using significantly greater dosages of all substances, with the exception of heroin, over the lifetime than the Non-syndromic SU group (p values <.01; see Appendix 3).
Finally, to confirm that the inclusion of social drinkers in the no SU group did not inadvertently select high use drinkers, we contrasted the median reported number of lifetime drinks in this group (930; range = 3 to 32,433) to that of the Syndromic SU group (12,800; range = 3 to 145,061). This large difference (p < 0.001) supports the notion that undiagnosed Syndromic levels of drinking were unlikely in the no SU group.
General substance use characteristics
In the Syndromic SU group, the three most frequent diagnoses were for alcohol, cocaine and cannabis. The frequency of past diagnoses for this group is presented in Table 2. Nearly half (47%) of the group had 3 or more lifetime diagnoses (maximum number = 8). The most frequent combinations of multi-substance use disorders were for alcohol and cocaine (43.0%), cocaine and cannabis (29%), and alcohol, cocaine and cannabis (23.8%). All diagnoses were for past syndromic use with average time since any use of an illicit substance ranging from 2.5 to 10 years for each substance though 29% of this group reported use of illicit substances within the last year.
Neurocognition in Substance Use Groups
Group performance scores on the neurocognitive battery are presented in Table 3. Average global and all seven cognitive domain T-scores, controlling for positive urine toxicology by MANCOVA, were similar among the groups (F(16, 776) = 1.46; p>.01), even when the two substance use groups were combined and compared to the no-SU group (F(8, 388) = 2.33, p>.01). No significant differences were detected in the diagnosis of HIV-associated neurocognitive disorders between the SU groups (Table 4; p>.01).
Functional Status
Substance use status had no effect on the proportion of participants with functional dependence (χ 2 (4)=1.14, p>.01) or the mean number of functional complaints (F (2, 395)= .14, p>.01). Neither did it impact the total number of subjective cognitive symptoms on the PAOFI (See Table 4; F (2, 396) = .22, p>.01).
Quantity of use and NC functions
We next examined if lifetime "dosage" of alcohol and illicit substances was related to NC function. Bivariate correlational analyses of quantified lifetime SU exposure with NC domain T-scores were performed on participants from the overall sample of 1284 to increase power (Table 5). Increased lifetime use of heroin was weakly associated with worse recall and working memory (both p values <.01). On the other hand, increased use of cocaine and cannabis was weakly associated with improved verbal fluency (p<.01). No other associations reached statistical significance. A similar analysis of duration of substance use (the number of days of reported use) and Global T-score did not reach statistical significance (p>.01). Finally, we explored potential relationships between duration and quantity of lifetime tobacco use and NC. These correlations did not approach statistical significance (p>.01).
DISCUSSION
On the basis of comprehensive neurocognitive testing HIV+ participants with either DSM-IV based SU disorders or self-reported histories of Non-syndromic, illicit substance use performed no differently, on average, than a carefully matched group of HIV+ participants without any indication of significant lifetime substance use. Similarly, participants with SU histories demonstrated neither a higher rate of HAND diagnoses nor more functional difficulties in everyday life.
This finding was unexpected, given the multiple studies that have linked abuse of specific substances, particularly alcohol (47) and methamphetamine (48) as well as polydrug use (49) to NC impairment; and the indications both from preclinical and clinical research suggesting that alcohol and other drugs may augment the neurotoxicity of HIV (25, 50, 51) The divergent findings may reflect the differences between the design of our study and most other studies which have examined HIV effects in the context of substance use. We compared NC function in HIV-infected groups with three levels of SU rather than comparing HIV-positive to HIV-negative persons with similar drug histories (30, 52). Two prior studies with similar design to ours reported more cognitive impairment among HIV-infected stimulant users relative to HIV-infected non- drug users (25, 53). However, participants in both studies had acute or recent use (within 24 months) and one study applied a less traditional cognitive NC task as the outcome measure (53). Thus, the discrepant findings between these two studies and ours may reflect differences in abstinence periods and NC measures.
The similarity in NC performance between the SU groups in the current study could be due to several factors. First, most of our participants with lifetime substance use disorders were not currently syndromic and reported their use as occurring years prior to the time of assessment. Thus, the prolonged period of abstinence may have allowed for reversal of substance use effects, although our cross-sectional study design does not allow for a definitive test of this hypothesis. The literature on cognitive recovery following abstinence from drugs other than alcohol is limited and has produced of mixed findings. Investigations involving acute/short term withdrawal have reported deficits across multiple cognitive domains (54, 55). In contrast, normalization of NC functioning has been demonstrated to occur in the months and years after achieving stable abstinence in alcoholics (56, 57), and users of cannabis (58) and methamphetamine (59). Few studies have examined cognitive recovery from SU in the context HIV (59), but the process varies greatly in degree and course among people without HIV and for specific substances of abuse (60- 62). In fact, recency of alcohol and drug use is reported as a stronger predictor of NC functioning than chronicity of prior use (47, 63).
Our negative findings might also reflect type and severity of substance abuse in CHARTER as compared to some other studies. As one example, heavy methamphetamine exposure has been linked to greater likelihood of NC impairment, and neuroimaging indicators of brain dysfunction in HIV+ persons (25, 68, 69). However, in our study only about 12 % of subjects met lifetime methamphetamine dependence criteria. The selection of study populations may also account for why our findings differ from earlier literature, as prior investigations reporting differences between SU groups may not have adjusted for group differences on NC-relevant co-factors such as literacy level or other sources of neurological injury (e.g., traumatic brain injury, stroke). Thus, some prior studies might have found greater NC deficits in HIV+ drug abusers partly because of difficulty in adjusting for comorbidities that often track with drug abuse.
Another potential explanation for apparent lack of drug effect on neurocognitive performance in the present study could be lack of sensitivity of our neurocognitive battery to substance associated brain injury. We consider this to be unlikely, given that a similar battery of tests identified both separate and additive neurocognitive effects of methamphetamine and HIV in a prior study (25), and also detected improvement in methamphetamine users who became stably abstinent (59). Nevertheless, we cannot rule out the possibility that a different test battery might yield different results.
Lifetime dosages of most substances did not predict NC performance in this study. This observation is consistent with lack of success in relating lifetime drinking estimates to neurocognition in many studies of alcoholism (e.g., see review by Rourke & Grant, 2009 (47) and similar findings in Cherner et al. (69) who also were not able to relate lifetime estimates of methamphetamine use to NC impairment. The exceptions in the current study were associations between heroin use and worse recall and working memory scores and cocaine and cannabis use and improved verbal fluency. While statistically significant, the correlation coefficients reflecting the strengths of these associations were weak and therefore may reflect true, mild substance specific effects.
This study has several significant and unique strengths. First, multiple methods were used to characterize illicit substance use, allowing for a rigorous approach to classifying such SU and its absence. Studies utilizing non-drug using comparison groups ("controls") often identify these through the absence of syndromic SU diagnoses. This method may misclassify persons who use substances heavily, but underreport their use or its functional impact. For example, among persons without a current diagnosis in our sample, 23% had positive urine toxicology for illicit substances, highlighting the value of toxicology screening. The multiple detection methods employed here provides increased confidence in identification of our "no substance" control group. The importance of avoiding misclassification of "controls" is underscored by reports that HIV- infected persons with positive urine toxicology yet denial of recent substance use were impaired in attention and learning/memory relative to HIV-infected persons who accurately reported recent drug use (70). This difference suggests that substance use non-disclosure and acute drug effects may be associated with cognitive dysfunction. A final strength is that our groups were screened for exclusion of severe neurologic confounds and matched to minimize the influence of factors that might influence NC test performance, namely, literacy level, depression, nadir CD4 count, and CART adherence level.
Given the complex histories of polysubstance use in our participants, we lacked statistical power to investigate individual effects of specific substances. The average number of syndromic SU diagnoses in our sample was 3 with a maximum of 8. Instead, we grouped users across classes of substances. This method is relevant to our primary aim: to determine if neuroAIDS cognitive research is confounded by historic substance use (broadly defined) as all assessed substances have demonstrated negative effects, at least acutely, on neurocognition (71-72).
This study raises important questions. What are the best methods for summarizing complex polysubstance use in neuroAIDS research? What aspects of substance use are most relevant to functional impact: type(s), age of onset, duration, recency, etc.? How does the order and overlap of exposure to substances and HIV infection influence neuropathogenesis? What biomarkers are relevant in the study of HIV and SU? Finally, does illicit SU lead to more rapid neurocognitive decline or to a unique pattern or course of NC functioning over time?
In conclusion, while methodological challenges remain in cognitive neuroAIDS research with substance users, our study provides assurance that the impact of remote SU disorders and Non-syndromic use may be subtle, at least in the absence of acute or recent use/abuse. While this finding does not suggest neurocognitive safety of past substance use, it does indicate that substance use history, even when complex, does not necessarily merit exclusion of patients in short-term analyses of cognition in neuroAIDS research.
|
|
| |
| |
|
|
|
