Potential Time Course of Human Immunodeficiency Virus Type 1–Associated Minor Motor Deficits: Title and subTitle BreakElectrophysiologic and Positron Emission Tomography Findings FREE

HUMAN immunodeficiency virus type 1 (HIV-1)–associated cognitive/motor complex (HIV-1–CMC) is the predominant clinical manifestation of HIV-1 in the brain. One defining feature is an "acquired abnormality in motor function or performance verified by clinical examination, neuropsychological tests or both."¹ Although the complex was already described in the first clinical review,² clinical³ and neuropathological⁴ studies have not defined it conclusively, and there has been an ongoing controversy about nomenclature and clinical and neuropathological definitions.^{1 ,5} The presence of possible subclinical manifestations was initially not generally accepted. The debate also has focused on finding appropriate variables to quantify "HIV-1–associated minor cognitive/motor disorders."¹

There is unequivocal evidence that the basal ganglia play a pivotal role in HIV-1 infection of the brain.^{6 - 7} Imaging studies using positron emission tomography (PET) to measure regional cerebral glucose metabolism (rCMRGlc) have revealed hypermetabolism in the basal ganglia^{8 - 11} and thalamus^{8 - 9} or have described changes in the temporal⁸ or parietal¹¹ lobes in early HIV-1–associated brain disease. One of these studies mentioned that relative hypermetabolism of striatal regions is the most characteristic finding in patients with subclinical dysfunction.¹⁰ As HIV-1–associated brain infection progresses, cortical and subcortical gray matter hypometabolism⁹ or globally reduced cortical uptake of glucose¹² is observed. However, the functional relevance of these findings is still poorly understood because attempts to correlate metabolic abnormalities and functional variables conclusively have failed so far. In one study,⁹ PET results were highly correlated with measures of fine motor control, whereas in another report by the same group,¹⁰ only a poor correlation was found.

Our investigation correlates PET imaging data of the prefrontal-striatal circuitry with motor performance. Motor performance was quantified using a motor test battery allowing assessment of basal ganglia motor function, a sensitive indicator of HIV-1–associated minor motor disorders.^{13 - 16}

Nineteen HIV-1–positive men (15 homosexual, 3 bisexual, and 1 heterosexual) were recruited from the Department of Neurology, Heinrich-Heine-University, Düsseldorf, Germany. Ages ranged from 25 to 62 years (mean, 42 ± 11 years). Patients covered the entire spectrum of motor function, ranging from normal psychomotor speed to highly pathologic psychomotor slowing. Mean time span since establishment of seropositivity was 4.8 ± 4.3 years. Data on plasma viral load were available for 11 patients (below detection threshold, 5 patients; <1000 copies/mL, 1 patient; 1000-10,000 copies/mL, 1 patient; 10,000-30,000 copies/mL, 4 patients). Nine patients had acquired immunodeficiency syndrome (AIDS).¹⁷ Three patients had not yet received any antiretroviral treatment (no ART); 7 used only reverse transcriptase inhibitors (ie, received ART); and 9 used highly active antiretroviral therapy (HAART). None of the patients fulfilled the clinical criteria for HIV-1–CMC,¹ ie, the AIDS dementia complex. None had evidence of HIV-1–associated myelopathy. None of the patients had any history of cerebral lymphoma or opportunistic cerebral infection. One patient had signs of HIV-1–associated polyneuropathy. All patients underwent scoring with regard to the 14 motor examination items of the Unified Parkinson's Disease Rating Scale (UPDRS). Mehrfach-Wortwahl-Test, Version B (MWT-B)¹⁸ and Raven's progressive matrices¹⁹ were performed. To assess the psychopathological status, the Arbeitsgemeinschaft für Methodik und Documentation in der Psychiatrie (AMDP) system²⁰ and the Hamilton Depression Rating Scale²¹ were used. Patient data were compared with those of a group of 15 healthy HIV-1–negative volunteers (10 men, 5 women; mean age, 38 ± 11 years) with no histories of any neurologic, psychiatric, or seizure disorders. Results of general physical and neurologic examinations were unremarkable. Written informed consent was obtained in accordance with guidelines of the Declaration of Human Rights, Helsinki, 1975. The study was approved by the university ethics committee. All individuals were informed that the purpose of this study was to investigate the rCMRGlc pattern during resting wakefulness. None had undergone previous PET scanning.

Axial, coronal, and sagittal T1- (repetition time [TR], 2000 milliseconds; echo time [TE], 20 milliseconds) and axial and coronal T2-weighted (TR, 2000 milliseconds; TE, 80 milliseconds) images and proton density–weighted magnetic resonance (MR) images were acquired using a 1.0-T scanner (Siemens Magnetom; Siemens, Erlangen, Germany).

We performed PET scanning with a PET camera (PC 4096/7 WB; Scanditronix, Uppsala, Sweden) as described by Seitz et al.²² The in-plane image resolution (full-width half-maximum) in the reconstructed images was 7.1 mm and the axial resolution was 6.5 mm.²³ Before scanning, the subject's left hand was warmed to about 40°C for 45 minutes to ensure arterialization of blood in the dorsal hand veins (oxygen saturation of >90%). Subjects rested comfortably on a padded table, and the PET gantry was aligned with the orbitomeatal line. Room noise was minimized and room lights were dimmed. Subjects were asked not to move or to speak but to keep their eyes open during the examination to avoid falling asleep. Serum concentrations of fludeoxyglucose F 18 (FDG) were determined in samples of arterialized venous blood after intravenous bolus injection of 185 MBq. Twenty-minute scanning was centered around 45 minutes after injection, according to steady-state FDG kinetics.²⁴ The rate of rCMRGlc was quantified according to the method described by Phelps and colleagues.²⁵ The kinetic constants and the lumped constant 0.52 were taken from Reivich et al.²⁶ The rate of rCMRGlc was determined in visually placed, anatomically oriented, homologous regions of interest (ROIs) in both hemispheres. The ROIs were drawn on each axial image along the boundary of the gray matter by using anatomical criteria provided by Talairach and Tournoux.²⁷ Since regional hypometabolism might change the size of a subjectively drawn region, regions on both hemispheres were traced as mirror images so that they would not differ in size by more than 10%. In our control group, the mean metabolic asymmetry between homologous regions in the cerebral hemispheres was 2.2% ± 1.6%. In addition, coregistered T1-weighted MR images were available for anatomical verification of the ROIs. The ROIs were selected to cover the prefrontal-striatal-thalamic loop^{28 - 29} (including the frontomesial and cingulate cortex; mean metabolic asymmetry, 2.6% ± 1.9%), the dorsolateral prefrontal cortex (mean metabolic asymmetry, 2.0% ± 1.5%), the thalamus (mean metabolic asymmetry, 2.5% ± 1.97%), and the basal ganglia (mean metabolic asymmetry, 2.1% ± 1.3%). For comparisons, the primary visual cortex (mean metabolic asymmetry, 1.9% ± 1.4%) was chosen. Figure 1 illustrates ROI placement.

The ROI data were averaged across 2 axial slices. Each mean ROI value was normalized by dividing it by the value of the individual's whole-brain metabolism to control for global differences in rCMRGlc between subjects and to allow comparisons between groups. Whole-brain metabolism was determined by averaging the values of gray matter ROIs from all slices, taking into account the actual size of single ROI areas relative to whole-brain area (volume). Values were rated as pathologic if they fell outside the range of the mean plus 2 SDs. Only the data of an investigator (C.A.) who was blind to all clinical information were used for this analysis.

Electrophysiologic motor testing included the analysis of the following: (1) postural tremor of the outstretched hands (tremor peak frequency [TPF]); (2) most rapid alternating movements of index fingers (MRAM); and (3) most rapid voluntary isometric index finger extensions (MRC). The variables measured in the last context were simple reaction time (RT), ie, the time span between a short go signal and the onset of contraction, and contraction time (CT), ie, the time span between the onset and the peak force of a contraction (methodological details are described elsewhere¹³ ).

Normal values for 98 HIV-1–negative subjects have been published.³⁰ Individual values were rated pathologic if a frequency (TPF or MRAM) was below the mean − 2 SD or if a time span (RT or CT) was longer than the mean + 2 SD.

Statistical analysis was performed using a commercially available software package (Statview; Abacus Concepts, Inc, Berkeley, Calif).

Unless otherwise indicated, data are given as mean ± SD.

Mean premorbid verbal intelligence (MWT-B) score was 107 ± 16; current nonverbal intelligence (Raven) score, 114 ± 13. These values did not differ among the patients. None of the patients showed more than mild impairment in the following domains: disturbances of attention and memory, formal disorders of thought, disturbance of affect, and disturbances of drive and psychomotility. None showed major depression.

Electrophysiologic test results are summarized in Table 1. Five patients showed completely normal motor results in all 8 variables; 2 patients showed only MRAM slowing of 1 hand. All other patients showed pathologically prolonged CT (bilateral in 10; unilateral in 2), alone or in combination with slowing of MRAM or prolonged RT. Tremor peak frequency was normal in all patients. Electrophysiologic variables were significantly correlated between hemispheres (P values of Fisher's r-to-z transformation: TPF, P = .03; MRAM, P = .002; RT, P<.001; CT, P = .01). Electrophysiologic variables did not correlate with the current CD4 cell counts.

Electrophysiologic test results were compared with the results of the UPDRS. Five of the 7 patients with normal CTs scored 0; the others, 5 and 6. Of the 12 patients with pathologically longer CTs, only 2 scored 0, and 10 scores were in the pathologic range (1-18; median, 5). The UPDRS scores and CT values of the right hand were weakly but significantly correlated (Spearman rank correlation, ρ = 0.47; P = .05). Individual results of CD4 cell counts, UPDRS, and CT values are provided in Table 2.

The MR imaging scans were evaluated by a neuroradiologist (F.W.) who was blind to all clinical data. All patients had normal findings on MR imaging scans.

Nondemented HIV-1–positive patients demonstrated a characteristic metabolic pattern of mean rCMRGlc alterations compared with controls. The sizes and metabolic rates in ROIs showed no significant differences between patients and controls. Normalized metabolic rates are given in Table 3. Patients generally had elevated rates of glucose metabolism in the basal ganglia of both sides and depressed metabolism in the frontomesial cortex bilaterally. After Bonferroni adjustment, however, the ratio of basal ganglia to frontomesial cortex metabolism was significantly higher for the patients than for the controls. No significant metabolic alterations were found in the thalamus, prefrontal cortex, or primary visual cortex.

To examine for a possible correlation between rCMRGlc alterations and motor performance, we analyzed each patient's data separately. Normalized metabolic ratios for the basal ganglia of both sides are presented in Table 2. These ratios did not correlate with immune function as represented by CD4 cell counts.

The present study was designed to test the hypothesis that PET scanning can expose metabolic abnormalities constituting the pathophysiological and functional correlate of HIV-1–associated minor motor deficits. The PET findings in our patients were clearly abnormal in the prefrontal-striatal circuitry. These data corroborate earlier findings of relative hypermetabolism in the basal ganglia,⁹ especially in the striatum. Hypermetabolism is one of the most characteristic findings in HIV-1–positive patients with subclinical neurologic dysfunction.¹⁰ Secondarily, late cortical and subcortical hypometabolism⁹ or globally reduced cortical uptake of glucose have also been described.¹² This combination of hypermetabolism and hypometabolism may well be the reason that in our patients, metabolic alterations in the basal ganglia failed to reach significance at a group statistical level. However, on an individual basis, 7 patients showed significantly pathologic unilateral or bilateral hypermetabolism, and in 1 patient, unilateral basal ganglia metabolism was significant reduced. In contrast to reports describing hypermetabolism in the thalamus^{8 - 9} or alterations in the temporal⁸ or parietal¹¹ lobes, we did not find such changes in our patients but did detect a second, significantly hypometabolic region, the frontomesial cortex (including the cingulate cortex). On an individual basis, 9 patients showed significantly pathologic unilateral or bilateral hypometabolism in the frontomesial cortex/cingulate cortex. These areas are known to be closely connected with the basal ganglia.

For an individual analysis of patient data (Table 2), we focused on disturbances in the metabolism of the basal ganglia and compared these findings with levels of motor function. Five patients (patients 1-5) showed normal metabolic rates in the basal ganglia on both sides and in the CTs of both hands. Four of these 5 patients scored 0 in the UPDRS motor items. We consider these patients to have been asymptomatic at the time of examination (phase 1 in Figure 2).

Three patients (patients 17-19) showed metabolic rates well below the mean in the basal ganglia bilaterally; in patient 19, the rate was 2 SDs below the mean. These patients showed highly pathologic motor performance (much longer CTs) and none of them scored 0 in the UPDRS motor items. We consider these patients to have had HIV-1–associated minor motor deficits at the time of examination. According to the literature, such cases most likely belong to a fairly advanced, hypometabolic, more symptomatic phase of the disease (phase 4 in Figure 2).

Analysis of the data of the remaining 11 patients revealed a much more heterogeneous picture and could provide insight into the metabolic and electrophysiologic dynamics of HIV-1–associated abnormalities of the central nervous system (CNS). Three patients (patients 6-8) had fairly early stages of relative basal ganglia hypermetabolism that were generally accompanied by normal or only slightly pathologic motor performance on one side. An additional 4 patients (patients 9-12) had marked hypermetabolism and their motor performance was much more clearly pathologic. In 4 patients (patients 13-16), the basal ganglia were less hypermetabolic than in patients 9 through 12, but not as severely hypometabolic as in patients 17 through 19. Motor performance in these 5 patients was generally more pathologic than in patients 6 through 12.

When the data of all the subgroups are compared, a hypothesis of disease kinetics can be set up (Figure 3). Initial increasing hypermetabolism turns toward hypometabolism as the disease progresses.

We correlated CT values of both hands with a metabolic index for the basal ganglia on an individual basis. The sum of the differences between the mean rate of basal ganglia metabolism (1.35 each side) and the individual metabolic rates was calculated: eg, for patient 6 (1.49 − 1.35) + (1.50 − 1.35) = 0.14 + 0.15 = 0.29, or basal ganglia index = 29. Second- and fourth-order polynomial regression yielded comparable results. A potential time course for the development of minor motor deficits is presented in Figure 2.

Our interpretation of individual data is compatible with the results of an earlier study⁹ describing a positive correlation between measures of fine motor control and metabolic abnormalities (which would appear relatively early in the course of the disease, eg, in phase 2 in Figure 2). A later study by the same group¹⁰ found no correlation at all (which would be the case at the time at which hypermetabolism turns toward hypometabolism or "pseudonormal" metabolism, ie, transition from phase 2 to phase 3 in Figure 2). Examination of motor performance is therefore helpful because it may differentiate between patients without any minor cognitive/motor deficits and those with pseudonormal brain metabolism in a later stage of disease. Basal ganglia dysfunction could represent the pathophysiological correlate of minor motor deficits. With regard to more advanced stages of the disease, pathologic motor performance is known to precede the clinically overt manifestation of dementia.^{15 ,31} Pathologically prolonged CTs can therefore serve as a marker of more advanced disease.¹⁶ What these motor tests basically examine is the capacity to perform and control single or alternating voluntary/involuntary rhythmic movements, ie, movements in which the basal ganglia play an important role.³² The tests are therefore a sensitive quantitative variable for recognizing a major feature of all basal ganglia disorders, ie, slowing of movement. This slowing cannot be attributed to peripheral nerve abnormality.³³ In normal motor function,³⁴ the maximal force of a contraction is independent of the amplitude of the contraction curves. Mildly abnormal CT curves in HIV-1 infection resemble those of patients in early phases of Huntington disease.^{13 ,35} More impaired HIV-1–positive patients show additional findings consistently seen in Parkinson or Wilson disease.³⁶ The PET scans of patients with manifest Huntington disease³⁷ or Wilson disease³⁸ show basal ganglia hypometabolism resembling that in our late-phase HIV-1–positive patients. One may speculate about the possible effects of ART and HAART on our patients. An improvement of psychomotor slowing has been demonstrated during use of ART¹⁴ and HAART.³⁹ This effect cannot account for our findings, because patients 7, 9, 11, and 13 to 19 had pathologic slowing despite treatment. They obviously did not respond to ART or HAART. It is therefore unlikely that patients 13 through 19 showed normalized metabolic function of the basal ganglia due to effective therapy, and yet continued to show pathologic motor slowing. Conversely, patients 1 through 5 showed consistent normal motor performance. However, CNS abnormalities may respond to ART or HAART at certain defined points during the course of the disease that are not yet known.

Finally, there is no close correlation between CNS abnormalities (as measured by psychomotor speed and PET findings) and immunological status (CD4 cell count). This could result in part from the fact that the immunological status of a patient may improve differently under therapy than does the CNS. Most patients actually had lower CD4 cell counts before the time of PET scanning (eg, patient 17 had a documented CD4 cell count of 0.069 × 10⁹/L [69 cells/µL] before therapy and 0.679 × 10⁹/L [679 cells/µL] at the time of PET scanning). Thus a pathologic process within the CNS conceivably may have begun before the immunological response to ART or HAART.

There is substantial theoretical and clinical evidence that CT in particular is a specific marker of basal ganglia function and can describe and quantify basal ganglia dysfunction reliably in HIV-1–associated minor motor disorders. Together with neuroimaging procedures, these tests may prove to be an appropriate tool to assess the time course of HIV-1–associated brain disease in larger prospective studies. Individual data could be essential for establishing specific neurologic therapeutic strategies.

Accepted for publication March 29, 2000.

The authors would like to thank Kristina Rascher, MD, for carefully reviewing the manuscript and for helpful linguistic advice.

Corresponding author and reprints: H. J. v. Giesen, MD, Department of Neurology, Heinrich-Heine-Universität Düsseldorf, Postfach 10 10 07, D-40001 Düsseldorf, Germany (e-mail: giesenhj@uni-duesseldorf.de).