Introduction
It is well recognized that the human immunodeficiency virus (HIV) invades the brain early in the course of infection, and is thought to cause neurocognitive impairment in up to 52% of patients (An, Groves, Gray, & Scaravilli, Reference An, Groves, Gray and Scaravilli1999; Heaton et al., Reference Heaton, Clifford, Franklin, Woods, Ake and Vaida2010). The typical neuropsychological disorder associated with untreated HIV infection is best described as a “subcortical” or “fronto-subcortical” dementia, as it involves cognitive (e.g., memory impairment, with relative sparing of recognition) (Grant, Reference Grant2008), neuropsychiatric (e.g., motivation, emotional control, social behavior) (Castellon, Hinkin, Wood, & Yarema, Reference Castellon, Hinkin, Wood and Yarema1998; Paul et al., Reference Paul, Flanigan, Tashima, Cohen, Lawrence, Alt and Hinkin2005), and motor symptomatology (e.g., motor slowing) (Hardy & Hinhn, Reference Hardy and Hinhn2002). Postmortem studies show that HIV has a predilection for the striatum as well as the white matter tracts connecting for the striatum with the cortex (Langford et al., Reference Langford, Letendre, Marcotte, Ellis, McCutchan and Grant2002; Wiley et al., Reference Wiley, Soontornniyomkij, Radhakrishnan, Masliah, Mellors, Hermann and Achim1998).
Past studies have consistently shown impaired function associated with frontostriatal circuits involved in visual attention and working memory (Chang et al., Reference Chang, Speck, Miller, Braun, Jovicich, Koch and Ernst2001; Ernst, Chang, Jovicich, Ames, & Arnold, Reference Ernst, Chang, Jovicich, Ames and Arnold2002), episodic memory (Castelo, Sherman, Courtney, Melrose, & Stern, Reference Castelo, Sherman, Courtney, Melrose and Stern2006), delay discounting (Meade, Lowen, MacLean, Key, & Lukas, Reference Meade, Lowen, MacLean, Key and Lukas2011), and a monetary incentive delay task (Plessis et al., Reference Plessis, du, Vink, Joska, Koutsilieri, Bagadia, Stein and Emsley2015). Furthermore, frontostriatal function loss is already observed during rest (Ortega, Brier, & Ances, Reference Ortega, Brier and Ances2015; Thomas, Brier, Snyder, Vaida, & Ances, Reference Thomas, Brier, Snyder, Vaida and Ances2013) and during simple motor paradigms (Ances et al., Reference Ances, Vaida, Cherner, Yeh, Liang, Gardner and Buxton2011). However, to date the impact of inhibitory control over voluntary movement has not been investigated. Such control entails concerted activation of both frontal regions and the striatum (Zandbelt & Vink, Reference Zandbelt and Vink2010). As HIV is known to involve the motor system clinically (Hardy & Hinhn, Reference Hardy and Hinhn2002), it is important to investigate how HIV impacts the various subsystems of this network involved in the control over the motor system.
Response inhibition represents a major component of frontostriatal functioning (Aron, Reference Aron2011). Although cognitive neuropsychological measures have demonstrated potential abnormalities of response inhibition in terms of Stroop task performance (Hinkin, Castellon, Hardy, Granholm, & Siegle, Reference Hinkin, Castellon, Hardy, Granholm and Siegle1999), the potential impact of HIV on the function of frontostriatal brain systems related to response inhibition has yet to be studied. Inhibition involves several sub processes, among which are (1) motor execution (i.e., GO responses), (2) outright stopping as an immediate reaction to a STOP signal (i.e., reactive inhibition) (Mink, Reference Mink1996; Vink et al., Reference Vink, Kahn, Raemaekers, van den Heuvel, Boersma and Ramsey2005), and (3) higher order cortical functions involved in proactive anticipation of stopping (i.e., proactive inhibition) (Aron, Reference Aron2011; Vink et al., Reference Vink, Kahn, Raemaekers, van den Heuvel, Boersma and Ramsey2005; Zandbelt & Vink, Reference Zandbelt and Vink2010).
In a previous study in healthy volunteers, a modified version of a Stop Signal Anticipation Task (SSAT) was used to investigate cortical and subcortical functions involved in the inhibition of voluntary movement (see Figure 1). During this task, subjects were required to give timed GO responses with occasional STOP signals occurring at fixed probabilities. The probability of a STOP signal occurring was explicitly stated to allow participants to adjust their responses proactively. GO responses in the absence of STOP signals served as a baseline, and they mainly activated the primary motor cortex. The striatum was found not to be active during these baseline responses, as these involve simple button presses. The putamen, however, was found to be active bilaterally during successful STOP trials (i.e., reactive inhibition) (Zandbelt & Vink, Reference Zandbelt and Vink2010). Furthermore, it was found that proactive inhibition evoked activation in both frontal and striatal regions to facilitate STOP performance by slowing down GO responses. Proactive inhibition specifically engaged the right inferior frontal gyrus, bilateral parietal gyri, and the right striatum (Zandbelt & Vink, Reference Zandbelt and Vink2010).
Fig. 1 The Stop Signal Anticipation Task. There are two types of trials: (a) GO signal trials interspersed with (b) occasional STOP signal trials. (c) A changing color cue indicates the stop-signal probability, which varies from trial to trial.
We chose to investigate a cART naïve cohort for two reasons. First, it allowed us to investigate the impact of the illness on brain function without the confounding effects of medication. Indeed, cART has been shown to potentially confound the effects of HIV on the frontostriatal system, leaving its true impact uncertain (Chang, Yakupov, Nakama, Stokes, & Ernst, Reference Chang, Yakupov, Nakama, Stokes and Ernst2007). Second, unmedicated HIV-positive patients continue to form an important part of patients seen in clinical settings in sub-Saharan countries (Reda & Biadgilign, Reference Reda and Biadgilign2012). For example individuals in Sub-Saharan Africa are generally only eligible for cART with CD4 counts of 350 or 500 cells/mL. Also, many eligible individuals do not access care for various reasons and a substantial number default form care after 12 months (15–20%) (Fox & Rosen, Reference Fox and Rosen2010). Therefore, the effect of illness in the absence of cART remains an important question in these settings.
The aim of the present study was to investigate the effect of HIV infection on cortical and subcortical regions of the frontal-striatal system involved in the inhibition of voluntary movement. To this end, 22 cART naive HIV+ and 18 matched controls performed a stop signal anticipation task while being scanned with functional MRI.
We hypothesized that HIV infection would have the following effects: (1) motor execution as measured by baseline timed GO responses would be increased, as HIV infection often involves psychomotor slowing (Hardy & Hinhn, Reference Hardy and Hinhn2002; Navia, Jordan, & Price, Reference Navia, Jordan and Price1986); (2) reactive inhibition time as measured by the speed of inhibition, that is, STOP signal reaction time (SSRT) (Logan, Cowan, & Davis, Reference Logan, Cowan and Davis1984) would be increased, similar to findings in other subcortical dementia’s such as Parkinson’s disease (Gauggel, Rieger, & Feghoff, Reference Gauggel, Rieger and Feghoff2004); (3) subcortical dysfunction as evidenced by hypofunction in the putamen during reactive inhibition would be demonstrated (Zandbelt, van Buuren, Kahn, & Vink, Reference Zandbelt, van Buuren, Kahn and Vink2011); and (4) compromised higher cortical functioning would be reflected in the participants’ inability to proactively increase their response times in anticipation of STOP signals, as well as in associated cortical activation deficits in the inferior frontal gyrus during proactive inhibition (Zandbelt et al., Reference Zandbelt, van Buuren, Kahn and Vink2011).
Method
Participants
The study was approved by the Health Research Ethics Committee of Stellenbosch University and the Human Research Ethics Committee of the University of Cape Town, Cape Town, South Africa. Before enrolment, all participants provided written consent after receiving a full description of this study. Our sample was recruited from a medically stable clinic-attending population during routine HIV care and testing at Site C primary health care clinic, in Khayelitsha, Cape Town, South Africa. A total of 22 HIV+ participants were included in the study together with 18 gender, education, ethnicity, and age-matched healthy controls. The controls were HIV negative as confirmed by the enzyme-linked immunosorbent assay. All HIV+ patients included in the study were cART naïve.
Participants were screened using the Mini International Neuropsychiatric Interview (MINI) 6.0.0/MINI-PLUS 6.0.0. (Sheehan et al., Reference Sheehan, Lecrubier, Sheehan, Amorim, Janavs, Weiller and Dunbar1998). All HIV+ participants received a full medical examination by a clinician and were excluded if there was a current comorbid psychiatric/neurological or general medical condition that could confound the diagnosis of HIV associated neurocognitive disorders (HAND); any history of substance use/abuse as assessed by the Substance Abuse and Mental Illness Symptoms Screener (SAMISS) screening questionnaire (Pence et al., Reference Pence, Gaynes, Whetten, Eron, Ryder and Miller2005); a score for smoking greater than 1 as confirmed by the Kreek-McHugh-Schluger-Kellogg (KMSK) scale (Kellogg et al., Reference Kellogg, McHugh, Bell, Schluger, Schluger, LaForge and Kreek2003); were pregnant as confirmed by a urine pregnancy test; or were currently receiving treatment for tuberculosis. All participants were right handed as confirmed by the Edinburgh Handedness Inventory (Oldfield, Reference Oldfield1971).
HIV+ participants underwent detailed neuropsychological assessment within two weeks of neuroimaging and controls within 1 year for characterization purposes only. The test battery assessed the following cognitive domains: abstraction/ executive function, memory, learning, speed of information processing, verbal fluency, motor and sensory/perception (Grant, Reference Grant2008). From these scores, a Global Deficit composite score was derived (Carey et al., Reference Carey, Woods, Gonzalez, Conover, Marcotte, Grant and Heaton2004) using normative data from a larger parent study (Joska et al., Reference Joska, Westgarth-Taylor, Myer, Hoare, Thomas, Combrinck and Flisher2010).
The following laboratory measures were performed in the HIV+ participants within 2 weeks of neuroimaging: CD4 count, HIV viral load, Rapid Plasma Reagin for syphilis and thyroid stimulating hormone level. All participants received a urinary drug screen before being scanned. In view of the fact patient reports of duration of infection are highly unreliable in our present population we relied on pre-treatment CD4 count (as a proxy for nadir) and viral load as an estimation of disease progression as all participants were cART naive and being treated for the first time. While hepatitis C co-infection has been associated with increased risk and severity of cognitive impairment in HAND, participants were not screened as it is not endemic to the region (Amin et al., Reference Amin, Kaye, Skidmore, Pillay, Cooper and Dore2004). An experienced radiologist reviewed all of the scans for intracranial pathology that could potentially confound functional imaging measurement results.
Functional MRI
All scans were acquired on a 3 Tesla Siemens Allegra at the Combined Universities Brain Imaging Centre (CUBIC). During MRI image acquisition, 622 whole-brain two-dimensional-echo planar imaging images [repetition time (TR)=1600 ms; echo time (TE)=23 ms; flip-angle: 72.5 degrees; field of view (FOV): 256×256; 30 slices; 4 mm isotropic voxels) were acquired. Trial-by trial variability was accounted for by setting the total task length to 17 min. Excess scans were discarded.
For image registration, a T1 ME-MPRAGE weighted structural scan was acquired (TR=2530 ms; TE1=1.53 ms; TE2=3.21, ms; TE3=4.89 ms; TE4=6.57 ms; flip-angle: 7 degrees; FOV: 256 mm; 128 slices; 1 mm isotropic voxel size) (van der Kouwe, Benner, Salat, & Fischl, Reference van der Kouwe, Benner, Salat and Fischl2008).
Stop-Signal Anticipation Functional MRI Task
During the functional MRI (fMRI) experiment, participants performed the STOP signal anticipation task (Zandbelt & Vink, Reference Zandbelt and Vink2010). The experiment was performed using Presentation® software (Version 14.6, www.neurobs.com). The task is based on original work by Logan et al. (Reference Logan, Cowan and Davis1984) who proposed a horse-race model, suggesting that a response, either GO or STOP, is a result from a race between the GO process and the STOP process. The response is stopped when the STOP process finished before the GO process reaches execution threshold (Logan & Cowan, Reference Logan and Cowan1984). The task and experimental procedures were identical to those described before (Zandbelt & Vink, Reference Zandbelt and Vink2010). All participants received standardized training in task performance before scanning by a trained technician in their first language (isiXhosa).
During task performance, participants were presented with three background lines. In each trial, a bar moved at a constant speed from the bottom line to the top line, passing the middle line within 800 milliseconds (ms). On GO trials, participants were required to stop the bar with a button press using their right index finger, as close to the middle/colored line as possible. Should the bar reach the top line after 1000 ms, the GO trial was considered a failure. The inter-trial interval was kept at 1000 ms. On STOP signal trials the bar stopped on its own, indicating a STOP signal. During STOP trials the participant was required to withhold the button press (reactive response inhibition).
To measure proactive response inhibition, the probability of the stop signal was explicitly indicated to the participant by the color of the middle line. This allowed a participant to proactively anticipate a STOP signal in each STOP trial, by taking the stop-signal probability into account. There were five stop-signal probability levels: 0% (green), 17% (yellow), 20% (amber), 25% (orange), and 33% (red). In total, 414 go trials (0%, n=234; 17%, n=30; 20%, n=48; 25%, n=54; 33%, n=48) and 60 stop trials (17%, n=6; 20%, n=12; 25%, n=18; 33%, n=24) were presented in a single run in pseudorandom order.
The STOP signal delay (SSD), the interval between start of a trial and the STOP signal, was initially 550 ms and varied for each STOP signal according to a staircase procedure. That is, should a STOP trial be successful, the trial difficulty was increased by increasing the STOP signal delay by 25 ms. Should a STOP trial be unsuccessful, trial difficulty was decreased in the same manner. This technique assured an equal amount of successful and unsuccessful STOP trials. For more details on the SSAT, see Zandbelt and Vink (Reference Zandbelt and Vink2010).
Motor Execution: Baseline GO Reaction Time
Timed baseline GO responses were measured in the absence of the possibility of a STOP signal. To explore speeded responses in terms of simple reaction time we included the California Computerized Assessment Package (CALCAP) in our neuropsychological assessment battery (Miller, Satz, & Visscher, Reference Miller, Satz and Visscher1991).
Reactive Inhibition
Reactive inhibition was studied in terms of the SSRT (Logan et al., Reference Logan, Cowan and Davis1984; Zandbelt & Vink, Reference Zandbelt and Vink2010), which was calculated according to the integration method (Logan & Cowan, Reference Logan and Cowan1984) and pooled across all STOP signal probability levels.
Proactive Inhibition
In keeping with previous studies (Vink et al., Reference Vink, Kahn, Raemaekers, van den Heuvel, Boersma and Ramsey2005, Reference Vink, Zandbelt, Gladwin, Hillegers, Hoogendam, van den Wildenberg and Kahn2014; Zandbelt & Vink, Reference Zandbelt and Vink2010), proactive inhibition was measured as the effect of STOP signal probability on GO signal response time. Impaired proactive inhibition is evidenced by a reduced effect of STOP signal probability on GO signal response time (Vink et al., Reference Vink, Kahn, Raemaekers, van den Heuvel, Boersma and Ramsey2005, Reference Vink, Zandbelt, Gladwin, Hillegers, Hoogendam, van den Wildenberg and Kahn2014; Zandbelt & Vink, Reference Zandbelt and Vink2010). This indicates a weaker anticipation of a STOP signal. Statistical analysis of proactive inhibition consisted of a repeated measures analysis of variance on mean GO signal response times, with STOP signal probability and HIV serostatus as factors.
fMRI Data Analysis
Image data were modeled using SPM8 (www.fil.ion.ucl.ac.uk/spm/software/spm8/). Preprocessing and first-level statistical analyses were performed as previously described (Zandbelt & Vink, Reference Zandbelt and Vink2010). Preprocessing involved correction for slice timing differences by resampling all slices in time relative to the middle slice using Fourier interpolation. Re-alignment to the mean image was performed using fourth-degree B-spline interpolation to correct for head motion. Head motion parameters were analyzed to ensure that the maximum motion did not exceed a predefined threshold (scan-to-scan >3 mm) (Van Dijk, Sabuncu, & Buckner, Reference Van Dijk, Sabuncu and Buckner2012). Spatial normalization was performed using linear and non-linear deformations to the Montreal Neurological Institute template brain, and spatial smoothing using a 6-mm full-width at half-maximum Gaussian kernel to accommodate inter-individual differences in neuro-anatomy.
The fMRI data were modeled voxel-wise, using a general linear model, in which the following events were included as regressors: Timed GO signal trials with STOP signal probability >0%, successful STOP signal trials and failed STOP signal trials. For GO signal trials with a STOP probability above 0%, we also included a parametric regressor modelling the STOP signal probability level as well as variations in response time. GO baseline (0% STOP probability) as well as activity during rest was explicitly modelled.
The fMRI data were high-pass filtered (cut-off 128 Hz) to remove low-frequency drifts. For each participant, we computed four contrast images: (1) Baseline GO-activation, to measure motor response initiation, (2) activation during successful STOP signal trials versus failed STOP signal trials (to assess reactive inhibition), (3) activation during successful STOP signal trials versus GO signal trials in the 0% STOP signal probability context (also to assess reactive inhibition), and (4) the parametric effect of STOP signal probability on GO signal activation (to assess proactive inhibition). We computed two contrasts for reactive inhibition because there is no consensus currently, on which contrast is most appropriate for investigating reactive inhibition.
We assessed group activation differences in predefined a priori regions of interest (ROIs), based on activation maps acquired in a previous experiment (Zandbelt & Vink, Reference Zandbelt and Vink2010), in which an independent sample of healthy volunteers performed the same task. These ROIs were defined using a cluster-level threshold (cluster-defining threshold p<.001, cluster probability of p<.05, family-wise error corrected for multiple comparisons). They included the primary motor cortex as activated during baseline GO responses; the right striatum during reactive inhibition and the right inferior frontal gyrus during proactive inhibition. We chose the right inferior frontal gyrus as our primary region of interest during proactive inhibition, as the inferior frontal regions have been shown to be affected by HIV in tasks of visual attention and working memory (Plessis et al., Reference Plessis, du, Vink, Joska, Koutsilieri, Stein and Emsley2014). An exploratory voxel wise whole-brain analysis was also performed for each of the above contrasts, testing for group differences using independent sample t-tests (family-wise error corrected for multiple comparisons).
Results
Demographics
Forty participants were initially recruited into the study. We excluded five patients for the following reasons: Tested positive for benzodiazepines (n=1); Did not receive a full physical examination (n=1); Problems with response box (n=2); Poor scan quality leading to failure in pre-processing (n=1). One healthy control was excluded to excessive movement (6.7 mm between scan movement) with a notable drop in signal to noise as assessed with in house quality assessment software (Geissler et al., Reference Geissler, Gartus, Foki, Tahamtan, Beisteiner and Barth2007; Stöcker et al., Reference Stöcker, Schneider, Klein, Habel, Kellermann, Zilles and Shah2005). Therefore, 17 HIV+ participants and 18 healthy controls were included in the final analysis (see Table 1).
Table 1 Demographic characteristics of the diagnostics groups.
Note. Age, viral load, and CD4 data represent mean±SD. Education and GDS data represent median and interquartile range between 25th and 75th centiles. AIDS defining CD4 count of 350 cells/μl used.
F, female; M, male; GDS: Global Deficit Score (22); HC: Hhealthy Ccontrols, HIV: HIV+ participants. AIDS defining CD4 count of 350 cells/μl used.
The groups did not differ with regard to age, gender, ethnicity, or education level. All HIV participants were ambulant and healthy enough to participate in the experiment. No significant pathology was found on the structural MRI scans. As expected, most of the present sample is female as the HIV+ population of sub-Saharan Africa consists mostly of women infected by heterosexual contact (UNAIDS, 2012).
Behavioral Results
Motor execution
Response times for baseline GO trials (a STOP signal probability of 0%) were close to the target response time of 800 ms (794±7 ms), for controls, whereas HIV+ participants were significantly slower (M=827±8; t(33)=−3.067; p=.004; r=0.471). Reaction time assessment according to CALCAP confirmed that HIV participants showed significant signs of motor slowing (Control: M=295±16 ms; HIV: M=367±19 ms; t(32)=−2.818; p=.008; r=0.446). Despite this slow baseline response speed, there was no difference in baseline GO accuracy between groups (Control: M=96±0.6; HIV: M=96±0.7; t(33)=0.673; p=.506; r=0.116), indicating that all subjects were able to perform the basic task regardless of baseline response speed.
Reactive inhibition
The speed of reactive inhibition (SSRT) did not differ between the groups (Control: M=270±3 ms; HIV: M=267±6 ms; t(33)=0.429; p=.670; r=0.075). As STOP errors were manipulated according to subject performance as previously described (Pence et al., Reference Pence, Gaynes, Whetten, Eron, Ryder and Miller2005; Zandbelt & Vink, Reference Zandbelt and Vink2010), there was no difference in STOP related errors (t(33)=−1.191; p=.242; r=0.203).
Proactive inhibition
There was a significant main effect of STOP probability on reaction time regardless of disease status, showing response slowing as the STOP probability increased (F(2.1,70.4)=16.808; p<.001) in trials where a STOP signal could occur (see Figure 2). Although there was no significant group by STOP probability interaction (F(2.1,70.4)=1.516; p=.226), there was a trend toward a group effect (F(1,33)=3.235; p=.081). This shows that both groups were able to slow down their responses proactively during GO trials in which a STOP signal could occur, with the HIV+ group being on average slower than controls.
Fig. 2 Effect of stop-signal probability on go-signal response time across groups with error bars indicating the standard error of the mean.
On further analysis, both groups showed, as expected, a main effect of response slowing relative to baseline (F(1,33)=31.171; p<.001). There was no group by response speed interaction, indicating that both groups showed an equal amount of proactive slowing relative to their own baseline (F(1,33)=2.42; p=.129). Finally, HIV+ had the same accuracy on GO trials where a STOP signal could occur (Control: M=97±0.5%; HIV: M=95±0.8%; t(33)=1.745; p=.090; r=0.291). Both groups had a STOP accuracy close to the target of 50%, indicating that the stepwise difficulty adjustment during STOP trials were successful (Control: M=51±1.5%; HIV: M=54±1%; t(33)=−1.522; p=.137). It should be noted that, as the difficulty was adjusted according to individual performance, we expected no difference in STOP accuracy.
fMRI Results
Motor execution
Both groups showed equal activation in the motor cortex on baseline GO responses, indicating normal motor cortical function during the timed responses (t(33)=0.320; p=.751; r=0.056).
Reactive inhibition
When comparing successful STOPs versus unsuccessful STOPs, we found hypo-activation in the right putamen (t(33)=2.157; p=.038; r=0.352) in the HIV+ group during this task. Further exploratory ROI analysis revealed the same effect to be present in the left putamen (t(33)=2.136; p=.040; r=0.348) (see Figure 3). No further differences were found on an exploratory whole brain analysis (see Figure 4). An exploratory regression analysis with putamen activation as dependent variable and global deficit score, viral load, and age as predictors, revealed no further significant results in the HIV+ group. We did not find any significant correlations between cognitive domain scores and reactive inhibition activation (see Table 2).
Fig. 3 Putamen activation during reactive inhibition in both HIV+ participants and seronegative controls.
Fig. 4 Whole-brain activation during reactive inhibition in HIV+ participants and healthy control subjects (HC). One-sample t-tests of (a) successful STOP signal versus failed STOP signal activation and (b) successful STOP signal versus GO signal 0% activation, for HIV+ and healthy control subjects. Significant activation clusters (p<.05, family-wise error-corrected) are displayed on the normalized and skull-stripped group-average brain (neurological orientation). HC, control subject; L, left; R, right; HIV+, HIV-positive participant.
Table 2 Cognitive domains tested (Mean Z value corrected via normal control group) as well as Pearson’s correlations performed for the left as well as the right putamen activations during reactive inhibition.
SE=standard error.
Proactive inhibition
We found no difference in function associated with proactive inhibition between groups in the inferior frontal gyrus (t(33)=−0.036; p=.972; r=0.006). An exploratory whole-brain analysis was also performed, which also showed no group differences. Exploratory analysis revealed no relationship between frontal task activation and cognitive domain scores.
Discussion
In this study, we investigated frontostriatal functioning during an inhibition task that assessed motor execution, reactive inhibition and proactive inhibition, in 17 cART naïve HIV+ participants and 18 matched healthy controls. The HIV+ participants had significantly slower baseline GO reaction times indicating impaired baseline motor execution. Both groups demonstrated similar responses on behavioral measures of reactive inhibition as well as proactive inhibition. During fMRI, however, in HIV+ participants the putamen was found to be hypo-active during reactive inhibition. There were no significant fMRI differences between groups in the cortex during proactive inhibition. This could indicate a relative sparing of higher cortical function with a specific dysfunction of the more basic functions of the basal ganglia in a cART naïve population during reactive inhibition.
We investigated motor execution in terms of timed GO responses and found HIV+ participants to be significantly slower. HIV+ participants displayed a normal inhibition process in terms of speed of inhibition. Given that their timed GO responses were abnormally slow, these findings suggest that the SSRT and the GO process are dependent on different fronto-striatal pathways. This is consistent with animal studies reporting that SSRT is sensitive to cortical lesions and GO reaction time is disrupted by subcortical lesions (Eagle, Baunez, et al., Reference Eagle, Baunez, Hutcheson, Lehmann, Shah and Robbins2007).
The HIV+ group proactively reduced their response speed similar to controls. This indicates that, despite abnormalities found in their baseline GO responses, HIV+ participants could still anticipate STOP signals and slow down their responses to facilitate stopping. As proactive inhibition requires higher cortical regions to appropriately respond to complex environmental cues and successfully communicate this information to the striatum (Zandbelt et al., Reference Zandbelt, van Buuren, Kahn and Vink2011; Zandbelt & Vink, Reference Zandbelt and Vink2010), it suggests higher cortical functions are largely intact in the present sample.
Our finding of no differences between HIV+ participants and controls in the function of the motor cortex during timed GO responses differs from a previous study in a cART treated group. In the latter study, increased motor cortical activation was found in HIV infected participants using a motor-tapping paradigm (Ances, Roc, Korczykowski, Wolf, & Kolson, Reference Ances, Roc, Korczykowski, Wolf and Kolson2008). Our finding suggests that the cortex is relatively spared in cART naïve HIV+ individuals (Chang et al., Reference Chang, Yakupov, Nakama, Stokes and Ernst2007) as seen here during response execution.
As predicted by findings of both clinical (Navia et al., Reference Navia, Jordan and Price1986) and post mortem studies (Langford et al., Reference Langford, Letendre, Marcotte, Ellis, McCutchan and Grant2002; Wiley et al., Reference Wiley, Soontornniyomkij, Radhakrishnan, Masliah, Mellors, Hermann and Achim1998), we found subcortical regions to be primarily affected as evidenced by putamen hypofunction during reactive inhibition. Although task based fMRI studies have reported subcortical involvement with HIV infection (Ances et al., Reference Ances, Vaida, Cherner, Yeh, Liang, Gardner and Buxton2011; Melrose, Tinaz, Castelo, Courtney, & Stern, Reference Melrose, Tinaz, Castelo, Courtney and Stern2008) studies differ with regards to the directionality of this involvement. For example, increased activation in the striatum has been reported during a fMRI motor tapping paradigm (Ances et al., Reference Ances, Vaida, Cherner, Yeh, Liang, Gardner and Buxton2011) and decreased activity during a semantic event-sequencing task (Melrose et al., Reference Melrose, Tinaz, Castelo, Courtney and Stern2008). A possible explanation for these differences from the present study could be that behavioral tasks differed between studies. Additionally, a positron emission tomography (PET) study, reported baseline hypometabolism in the basal ganglia in HIV+ participants with moderate motor-slowing, whereas basal ganglia hypermetabolism was found in HIV+ participants with normal motor performance (Giesen et al., Reference Giesen, von, Antke, Hefter, Wenserski, Seitz and Arendt2000). As fMRI studies in HIV often include elderly subjects (Plessis et al., Reference Plessis, du, Vink, Joska, Koutsilieri, Stein and Emsley2014), differences in age across studies could also be a confounding factor, as HIV+ effects on the brain is thought to co-vary with increased age (Ances, Ortega, Vaida, Heaps, & Paul, Reference Ances, Ortega, Vaida, Heaps and Paul2012; Chang, Holt, Yakupov, Jiang, & Ernst, Reference Chang, Holt, Yakupov, Jiang and Ernst2013; Holt, Kraft-Terry, & Chang, Reference Holt, Kraft-Terry and Chang2012). Another possible explanation for the directional difference of our data is the effect of cART (Chang et al., Reference Chang, Yakupov, Nakama, Stokes and Ernst2007), as previous fMRI studies included samples mostly on treatment (Plessis et al., Reference Plessis, du, Vink, Joska, Koutsilieri, Stein and Emsley2014). Furthermore, two arterial spin labelling studies have reported reduced baseline regional cerebral blood flow in the striatum of HIV+ participants, which further supports our finding of hypo-activity of these subcortical structures at baseline (Ances et al., Reference Ances, Roc, Wang, Korczykowski, Okawa, Stern and Detre2006, Reference Ances, Sisti, Vaida, Liang, Leontiev, Perthen and Ellis2009). Our findings therefore confirm the impact of HIV on the function of the striatum and extend these results by demonstrating striatal hypofunction in a cART naïve sample during reactive inhibition.
We found no difference in cortical activity during proactive inhibition. This is seemingly in contrast to functional studies performed in the past that do indeed find relative cortical hyperactivity in HIV (Chang et al., Reference Chang, Tomasi, Yakupov, Lozar, Arnold, Caparelli and Ernst2004; Melrose et al., Reference Melrose, Tinaz, Castelo, Courtney and Stern2008; Plessis et al., Reference Plessis, du, Vink, Joska, Koutsilieri, Stein and Emsley2014). These studies largely included participants on cART treatment, however. As cART treatment has been shown to be associated with increased cortical activity (Chang et al., Reference Chang, Yakupov, Nakama, Stokes and Ernst2007), this could serve as a possible explanation for this difference.
We did not find any significant correlations between cognitive test scores and fMRI activation. This is not surprising given that fMRI has been shown to be more sensitive to cerebral pathology than cognitive testing (Ernst et al., Reference Ernst, Chang, Jovicich, Ames and Arnold2002). Also, the study may be underpowered due to the small sample size. Further investigation is, therefore, warranted.
The slow GO processes and relatively normal STOP processes found in HIV infection in the present study could be explained on the basis of dopaminergic deregulation: Supportive evidence is forthcoming from animal studies suggesting that dopamine deregulation could result in an abnormally slow motor responses, leaving the speed of inhibition spared relative to controls (Bari, Eagle, Mar, Robinson, & Robbins, Reference Bari, Eagle, Mar, Robinson and Robbins2009; Eagle, Tufft, Goodchild, & Robbins, Reference Eagle, Tufft, Goodchild and Robbins2007). Studies performed in rats revealed that neither the administration of the mixed D1/D2 receptor antagonist cis-flupenthixol (Eagle, Tufft, et al., Reference Eagle, Tufft, Goodchild and Robbins2007) nor dopamine-associated transport blockade by GBR-12909 influenced SSRT in a modified version of the SSAT (Bari et al., Reference Bari, Eagle, Mar, Robinson and Robbins2009). However, in the same studies GO reaction time was found to speed up in response to dopamine re-uptake blockade as well as methylphenidate administration (Bari et al., Reference Bari, Eagle, Mar, Robinson and Robbins2009; Eagle, Tufft, et al., Reference Eagle, Tufft, Goodchild and Robbins2007). Our finding of a slow GO process in the presence of a normal speed of inhibition indicates that the exact biochemical nature of striatal dysfunction in HIV infection requires further elucidation. As the striatum is modulated by dopaminergic activity (Frank, Reference Frank2005), the present finding of striatal hypofunction could relate to dopamine deregulation. This could be further investigated by PET using dopamine ligands in conjunction with fMRI functional measurements. Prospective pre- and post-cART prospective studies would help to clarify the striatal and cortical effects of cART (Chang et al., Reference Chang, Yakupov, Nakama, Stokes and Ernst2007).
A strength of our study is the exclusion of important confounds that could influence fronto-striatal function such as cART (Chang et al., Reference Chang, Yakupov, Nakama, Stokes and Ernst2007), comorbid depression and substance abuse. The effects of age should also be minimal (Ances et al., Reference Ances, Vaida, Yeh, Liang, Buxton, Letendre and McCutchan2010; Holt et al., Reference Holt, Kraft-Terry and Chang2012), due to the relatively young age of our cohort. We could not demonstrate a link between slow GO processes and brain function. Although we found normal motor cortical activation in HIV+ participants, we could not rule out differences in other parts of the motor system, as our task involved only simple timed motor responses. We also could not demonstrate a behavioral difference in terms of reactive STOP accuracy, which related to putamen activation. This was due to an inherent limitation of our task design: As putamen activation relies on balanced correct versus incorrect STOP factors, task difficulty was adjusted to achieve equal successful and unsuccessful STOP for both groups. We did however demonstrate that the putamen is hypoactive in HIV infection when controlling for task performance in this way. Our study is further limited by a small sample size due to the carefully selected nature of our sample.
Taken together, our findings support the hypothesis that HIV infection primarily affects the basic functions of the putamen during reactive inhibition, with relative sparing of cortical function during proactive inhibition.
Acknowledgments
The authors thank Dr. Hetta Gouse from the University of Cape Town, as well as Prof. Rob Paul from the University of Missouri-St. Louis, for their assistance in determining the neuropsychological assessment battery as well as giving input into the neuropsychological assessments. The authors also thank Dr. Karen Cloete and Dr. Sanja Kilian from Stellenbosch University for their support in preparation of the manuscript, and Mr. Teboho Linda from the University of Cape Town for assisting in the recruiting of participants. Financial Disclosures: S. Du Plessis has received support from a National Research Fund International Research Training Grant (IRTG 1522), as well as support from the Medical Research Council of South Africa, Biological Psychiatry Special Interest Group of the South African Society of Psychiatrists as well as the HIV research Trust (HIVRT14-049). M. Vink, PhD and A. Bagadia have reported no biomedical financial interests or potential conflicts of interest. Joska, PhD has received support from the Medical Research Council of South Africa. E. Koutsilieri, PhD is supported by an EDCTP Strategic Primary Grant (SP. 2011.41304.065.) as well as a “Verein für Parkinson-Forschung” grant. D. Stein, PhD has received research grants and/or consultancy honoraria from Abbott, Astrazeneca, Biocodex, Eli-Lilly, GlaxoSmithKline, Jazz Pharmaceuticals, Johnson & Johnson, Lundbeck, Novartis, Orion, Pfizer, Pharmacia, Roche, Servier, Solvay, Sumitomo, Sun, Takeda, Tikvah, and Wyeth. R. Emsley, PhD has participated in speakers/advisory boards and received honoraria from AstraZeneca, Bristol-Myers Squibb, Janssen, Lilly, Lundbeck, Servier, and Otsuka. He has also received research funding from Janssen, Lundbeck and AstraZeneca.
