The primary goal of this investigation was to determine if a positive HIV serostatus was associated with poorer procedural learning (PL) in a sample of individuals with a history of substance dependence. Several important conclusions emerged. First, compared to the HIV− group, HIV+ individuals showed evidence of poorer overall performance on PL tasks, particularly on the two measures with complex motor demands (i.e., Rotary Pursuit and Star Mirror Tracing). Secondly, both HIV+ and HIV− participants evidenced significant improvements in performance across trial blocks of all three tasks, suggesting successful PL. Finally, there was no evidence to substantiate a specific deficit in procedural learning among HIV+ compared to HIV− participants.
Our findings are consistent with deficits in complex motor functions and processing speed (typically indexed by slowed performance) among HIV+ participants, which have been well-documented in the scientific literature (e.g., Hardy et al., 2002
; Heaton et al., 1995
; Martin, Sorensen, Edelstein, & Robertson, 1992
), and are often referred to as a "hallmark" neurocognitive feature resulting from known abnormalities in white matter and subcortical nuclei of HIV+ participants. Of the tasks that our participants were administered, the Rotary Pursuit Task (RPT) and Star Mirror Tracing Task (SMT) both place time demands on participant performance and depend on intact complex motor functions whereas the Weather Prediction Task (WPT) does not. Although in theory our subjects’ performance might be influenced in part by peripheral neuropathy, some evidence argues against this explanation. Peripheral neuropathies are reported in 10 – 15% of HIV+ patients, and are more common in advanced disease stages, typically affect the lower extremities first, and often do not produce objective motor deficits in the upper limbs (for reviews see Verma, 2001; Wulff, Wang, and Simpson, 2000). Given the low prevalence of immunological AIDS diagnoses in our ambulatory community sample of HIV+ participants, we do no think that peripheral neuropathies (if present) would be of sufficient severity to affect performance on the measures we employed. However, future studies may benefit from detailed evaluation of peripheral neuropathy, particularly among subjects with more advanced disease. Furthermore, inclusion of additional neuropsychological tests assessing specifically simple motor skills and processing speed would help to understand better the mechanisms for the complex motor deficits that we observed in the HIV+ sample.
It is important to note that multiple structures in addition to the caudate and putamen contribute to performance of both motor and non-motor tasks of PL, including prefrontal, occipital, parietal cortex, and cerebellum (e.g., Grafton et al., 1992
; Jenkins, Brooks, Nixon, Frackowiak, & Passingham, 1994
; Poldrack et al., 1999
; Poldrack et al., 2001
). PL tests of motor skills may rely more heavily on different structures (e.g., putamen and cerebellum) in these systems when compared to PL measure without motor demands (e.g., WPT). Furthermore, neural mechanisms that may prove compensatory for WPT may not necessarily generalize to SMT and RPT. Recent literature suggests that in some conditions individuals may recruit medial temporal lobe structures and partially rely on declarative memory to perform the WPT (e.g., Foerde 2006; Moody 2004). Further evidence indicates that individual elements of the circuitry supporting procedural learning might compensate for possible dysfunction in other structures in the network. For example, Fera and colleagues (2005)
compared performance of young and older healthy adults on the WPT and found no significant group differences in task performance. However, older adults demonstrated much greater activation of parietal cortex on fMRI.
HIV-associated neurocognitive deficits were detected in our sample of individuals with history of substance dependence on two of three PL tasks, and our finding of significant improvements across Trial Blocks on all tasks indicate that the measures employed in this investigation were sensitive indicators of PL in our sample. However, despite evidence of poorer performance on the RPT and SMT, our findings did not demonstrate specific PL deficits among HIV+ substance dependent individuals compared to HIV− controls. Contrary to our hypothesis, both HIV+ and HIV− individuals demonstrated significant improvements in performance across Trial Blocks and their rate of improvement across Trial Blocks did not differ significantly across all three PL tasks. However, the absence of a non-substance using HIV+ control group and a group of healthy controls precludes coming to definitive conclusions on whether performance among our HIV+ subjects was "within normal limits" or if there are possible interactions between substance use and HIV on PL performance. Recruitment of such groups is currently underway in our laboratory to address these questions in future studies.
At first glance, our findings may appear to contradict those presented in the seminal paper by A. Martin and colleagues (1993)
, that is widely cited as providing evidence for procedural learning deficits associated with HIV infection. However, findings from these two studies are considerably less discrepant with careful comparison of their investigation with ours. A. Martin et al. (1993)
found that a small subset of approximately 24% of the HIV+ group in their study and 7% of HIV− controls (ns= 7 and 1, respectively) could be characterized as “poor learners” on the basis of minimal improvement in performance from the first to last Rotary Pursuit trial. The level of the neurotoxin quinolinic acid in CSF among five HIV+ "poor learners" was significantly higher than in 18 HIV+ "good learners" and quinolinic acid levels were correlated with improvements in Rotary Pursuit performance. However, they also found that HIV+ subjects, as a group
, did not show a deficit in rate of improvement across Trial Blocks on the Rotary Pursuit and the majority of subjects performed in the normal range. Thus, analyses of task performance in both the A. Martin study subjects and our groups showed no significant HIV serostatus by Trial Block interactions, which we deem necessary to establish a specific procedural learning deficit from HIV. In fact, despite significant differences in sample characteristics and our use of individuals with history of substance dependence, the results of both investigations are remarkably similar. Further, our conclusions are bolstered by our larger study sample and use of three separate PL tasks. Therefore, although both investigations found evidence for complex motor problems among HIV+ participants as a whole, neither investigation substantiates specific PL deficits, per se, among HIV+ persons. To our knowledge, no study has demonstrated this latter type of impairment among HIV+ participants. One can speculate that that the paucity of published investigations on HIV-associated PL deficits may stem, in part, from a "file drawer" effect, as other studies with negative outcomes may have gone unpublished (Rosenthal, 1979
Several findings from more detailed studies of the neural systems underlying procedural learning help to clarify our results. Most initial investigations suggesting that PL is dependent on integrity of striatum were conducted with clinical populations with known basal ganglia disease (i.e., Parkinson and Huntington' disease). However, the pattern of neuropathology associated with these diseases typically extends well beyond striatum and are much more severe than in HIV. The caudate is often noted as a critical structure for PL, but widespread circuitry beyond striatum is often shown to be active in neuroimaging studies of PL (e.g., Grafton et al., 1992
; Jenkins, Brooks, Nixon, Frackowiak, & Passingham, 1994
; Poldrack et al., 1999
; Poldrack et al., 2001
). Thus, some researchers have postulated that the caudate is an important, but not essential, structure for procedural learning (e.g., Schmidtke, Manner, Kaufmann, & Schmolck, 2002
). As we not above, several studies have suggested that individuals may rely on brain systems outside striatum to perform some PL tasks. It is possible that neuropathology in striatum must reach a threshold level of severity or must extend substantially to additional brain regions for frank deficits in procedural learning to emerge. Brain functioning in our HIV+ sample was compromised enough to manifest with deficits in complex motor skills, but it may be that injury to neural systems that support procedural learning was insufficient to produce detectable deficits.
We acknowledge that HIV has been shown to disrupt basal ganglia preferentially, but HIV-associated neuropathology is diffuse and affects additional brain systems, including white matter, prefrontal cortex, and hippocampus (e.g. Jernigan et al., 1993
; Pomara, Crandall, Choi, Johnson, & Lim, 2001
; Reyes, Mohar, Mallory, Miller, & Masliah, 1994
). Furthermore, correlations between basal ganglia neuropathology and neurocognitive performance are not always detected, suggesting that there is not an invariable relationship between striatal integrity and neuropsychological performance in the HIV literature (cf. Paul et al., 2002
). For example, Moore and colleagues (2006)
found significant correlations between a global index of neuropsychological performance obtained shortly before death of HIV+ participants and postmortem measures of neurodegeneration in hippocampus and midfrontal cortex, but not putamen. Striatal pathology in HIV may not be of sufficient severity, in most cases, to produce deficits in procedural learning as have been reported with disorders that severely damage basal ganglia and associated circuits (e.g., Parkinson's and Huntington's Disease).
With one exception, we found no consistent evidence to suggest that markers of immune status related to performance in our sample. WPT performance was significantly worse among HIV+ participants with detectable viral load in plasma, compared to those with undetectable viral load (p = 0.03). However, performance on the WPT did not differ significantly as a consequence of CD4 count or ARV status. Furthermore, performance on the SMT and RPT, which differed based on HIV serostatus, did not differ significantly when HIV+ groups were stratified by CD4 count, plasma viral load, or ARV status. Although the observed differences in WPT performance might have resulted from Type-I error, one might speculate that HIV effects may first manifest on tests involving complex motor skills (e.g., SMT and RPT) and then progress to more "cognitive" tasks (e.g., WPT). Therefore, participants with more advanced disease would be more likely to demonstrate differences in performance on the WPT, in addition to the RPT and SMT. However, only 18% of HIV+ participants in the current study met criteria for an immunological AIDS diagnosis, 44% had undetectable HIV viral load in plasma, and 86% were on ARVs, with 50% on HAART. Future studies may benefit from including HIV+ participants with more advanced disease and from examination of additional medical markers of HIV disease severity (e.g., biomarkers of neuroinflammation, viral load in CSF, or neuroimaging data).
Our sample consisted predominantly of African-American men with a high school education, most of who were in their 4th
decade of life and were recruited from the Chicago greater metropolitan area; thus, our results may not generalize to participants of differing demographics. Nonetheless, our findings are very similar to those of A. Martin et al. (1993)
, who tested a sample of men recruited from the U.S. military that was generally younger, more highly educated, and of higher estimated IQ than our sample. Further studies that include more diverse samples will be better poised to systematically examine possible interactions with demographic variables, including possible gender effects.
In summary, the results from the current investigation showed evidence of poorer performance on procedural learning tasks that were consistent with general difficulties in complex motor skills among HIV+ individuals with substance dependence compared to matched HIV− controls. However, there was no evidence for specific procedural learning deficits among the HIV+ group, as both groups showed significant and comparable improvements in performance across Trial Blocks on motor and non-motor tasks of procedural learning. Although our results suggest that both groups achieved asymptote within the trials we administered, future investigations may examine if HIV+ individuals may achieve levels of performance comparable to controls on motor skills tasks if they receive additional training. Overall, our findings contribute new knowledge to an initially promising but understudied research area within the literature on HIV and neurocognition. Furthermore, the current study is unique in its use of multiple measures of procedural learning (with and without motor demands), a fairly large sample size, and its focus on examining the effects of HIV among individuals with history of substance dependence. The ability to implicitly acquire skills gradually and incrementally may impact an individual's ability to learn and implement new tasks at work, efficiently complete routine tasks at home, and carry out skills taught in clinical treatment programs. Understanding procedural learning in HIV+ substance dependent individuals will help to more accurately characterize the impact of HIV and substance use on neurobehavioral performance.