Spatial Attention in HIV-1 Infection: A Preliminary Report Eileen M. Martin, Ph.D. DonnaJ. Sorensen, M.A. Lynn C. Robertson, Ph.D. Howard E. Edeistein, M.D. Valerie A. Chirurgi, M.D.
In a preliminary study, 30 nondemented human immunodeficiency virus (HIV-1) seropositive subjects without acquired immunodeficiency syndrome and 14 seronegative controls performed a reaction time measure of spatial attention. Compared with controls, sero positive asymptomatic subjects showed normal facilitation of reaction time at short cue-target intervals when attention was precued, but symptomatic subjects were impaired. However, asymptomatic subjects showed no evidence of normal inhibition of attention at the cued location at longer cue-target intervals, suggesting possible subtler spatial attentional deficits in this group. Cognitive slowing in HIV-1 infection may have an attentional component, with possible involvement of both automatic and controlled processes. (The Journal Neurosciences
of Neuropsychiatry 1992; 4:288-293)
and
Clinical
C
ognitive dementia
munodeficiency
slowing complex
is
a prominent feature of the associated with acquired im-
syndrome
(AIDS)1
and
can
be
dem-
onstrated reliably in patients with advanced human immunodeficiency virus (FIIV-1) infection using standard clinical neuropsychological tests of psychomotor functions, memory, and attention.2 Patients with constitutional symptoms but without AIDS-defining disorders show qualitatively similar but milder deficits on these clinical measures.3 The cognitive status of HIV antibodyseropositive asymptomatic subjects, as assessed by standard neuropsychological tests, is more controversial. Some investigators have reported evidence of impairment in asymptomatic subjects,4’5 but others have failed to demonstrate group differences between asymptomatic subjects and control subjects on these measures.6’7 These inconsistencies have led some investigators to conclude that the incidence of cognitive dysfunction among asymptomatic individUals is low and that positive findings can be attributed to experimental confounds such as substance abuse, psychiatric disorder, or employment of excessively liberal criteria for judgment of impairment.8 However, there is strong evidence that HIV-1 invades the central nervous system early in the course of systemic infection,9 and some asymptomatic patients complain of mild cognitive difficulties. A careful review of the literaReceived
September 13, 1991; revised December 12, 1991; accepted 1991. From the Departments of Neurology and Psychiatry and the Department of Medicine, University of California-Davis, and the Research Service, Veterans Affairs Medical Center, Martinez, California. Address reprint requests to Dr. Martin, Neurology Service (127), VA Medical Center, 150 Muir Rd., Martinez, CA 94553. Copyright C) 1992 American Psychiatric Press, Inc. December13,
288
VOLUME
4
#{149} NUMBER
3
#{149} SUMMER
1992
MARTIN
ture suggests that early central nervous system infection could be manifested by subclinical cognitive changes that are not detected reliably by the standard clinical instruments employed in most studies. Investigators have recently employed reaction time (RT) tasks developed in cognitive psychology as a means of evaluating subclinical cognitive slowing in asymptomatic subjects.10’6 These measures are well suited for studying patients with varying severity of infection because they are less subject to ceiling effects than clinical neuropsychological tests. In addition, specific component cognitive functions can be measured with RT tasks, and hypotheses about specific mechanisms of HIV-related cognitive dysfunction can be tested. For instance, tasks measuring primarily motor slowing (e.g., simple RT, which reflects the time to detect a stimulus and prepare and execute a motor response) or cognitive slowing (e.g., choice RT, which reflects these factors plus time to make some type of discrimination) can be administered, and different components of performance can be isolated. The potential contribution of RT tasks in the study of HIV-related neurobehavioral disorders is acknowledged in recently issued National Institute of Mental Health guidelines for assessment of AIDS-related cognitive changes.’7 Participants recommended inclusion of cognitive psychology tasks in the neurobehavioral battery because of their “greater complexity and potentially their increased sensitivity to subtle cognitive changes” (p. 964). Further, a recent longitudinal investigation demonstrated the prognostic value of RT measures by showing that performance of RT tasks was significantly correlated with rate of CD4 lymphocyte decline)6
Despite the potential advantages of cognitive neuropsychological measures, the literature on RT performance in HIV-positive subjects has been inconsistent. Some studies have found early deficits on RT tasks,’#{176}” but others have not.’2’4 It is difficult to evaluate comprehensively the available literature on RI performance in HIV-positive subjects. Experimental methods differ widely across studies and are not always described in sufficient detail for independent replication. The tasks employed have clearly differed in response requirements (e.g., single keypress go/no go versus choice of two or more keys), presence or absence of an alerting stimulus preceding the “go” stimulus, length of intertrial or interstimulus interval, and degree of task difficulty (e.g., simple two-choice discrimination versus complex discrimination). These parameters systematically influence absolute RI values; thus, obtained results are not directly comparable across studies. In addition, because of these differing task requirements, it cannot be assumed that the same underlying cognitive processes have been evaluated.
JOURNAL
OF
NEUROPSYCHIATRY
et al.
Performance variability has also been a common problem for RI studies. Experiments have not always included sufficient numbers of experimental trials for adequate reliability,’8 possibly because RI tasks have often been administered as part of long neurobehavioral testing protocols. Finally, the theoretical basis for task selection has not always been specified, despite ample experimental
literature
on
RI
performance
in other
pa-
tient populations with similar dementias)2’ lask selection in many available studies appears guided by the implicit assumption that cognitive slowing should be evident on all RI tasks, regardless of a specific task’s cognitive
requirements.
Different
experimental
tasks
may measure different underlying cognitive functions, which may not necessarily be affected by HIV-1. In our laboratory, we are conducting a series of experiments evaluating cognition in HIV infection, using RI measures selected specifically to evaluate component cognitive functions underlying cognitive slowing. In an initial study,’5 we found that, compared with controls, both symptomatic and asymptomatic HIV-seropositive subjects showed an excessively long decision time, as measured by the discrepancy between simple and choice RI. In contrast, H1V-positive subjects performed normally on a control measure of rate of verbal encoding. These findings support the hypothesis that subtle cognitive slowing is present in early HIV-1 infection and does not represent a nonspecific effect across all speeded tasks. The present study investigated one possible mechanism of cognitive slowing. We evaluated the ability of HIV-seropositive
and
control
subjects
to move
attention
in the visual field without overt eye movements, using a task that has been employed to study spatial attention in patients with focal cortical lesions, schizophrenia, and subcortical dementing disorders.24 Developed by Posner, this “covert orienting” task measures time to detect a target following introduction of a visual cue. “Valid” cues summon attention to the location of the forthcoming target; “invalid” cues summon attention away from the target location, and thus subjects require additional time to disengage and move attention from an invalid cue to the target location. The “validity effect,” faster RI on valid than invalid trials (e.g., facilitation of speed of visual orienting when attention is precued), is most prominent with about 150 msec between cue and target. In addition, normal subjects typically show a reversal (faster RI to invalid than valid cues) at longer cue-target intervals. Ihis effect, the “inhibition of return,” indicates inhibition of movement of attention to previously attended locations, possibly representing facilitation of visual exploration of novel targets.26 For the purposes of this preliminary study, we were primarily concerned with the validity effect as the vari-
289
SPATIAL
ATTENTION
IN HIV-1
INFECTION
able of major interest because this effect has been studied more extensively in cognitive neuropsychology than the inhibition of return and because a more specific hypothesis about this effect could be formulated. We predicted that the validity effect would be diminished or delayed in H1V-seropositive subjects if cognitive slowing affected this type of attentional process.
METHODS
Procedure All subjects were administered the covert orienting task, which consisted of 400 RI trials (Figure 1) administered by an Al-compatible personal computer. A display of 3 boxes was present continuously on the computer monitor. On each trial, the subject fixated on the central box in the display. Each trial began with a 500-msec 880-Hz auditory warning stimulus, followed 100 msec after its offset by the appearance of a 300-msec cue (brightening of one of the two peripheral boxes located approximately 8 degrees on either side of the fixation box). The cue was
Subjects
Experimental subjects were I left-handed and 29 righthanded men, including 27 gay or bisexual subjects and 3 subjects with a history of transfusion or intravenous drug abuse. All experimental subjects were recruited through the Infectious Disease Clinic at the Martinez Veterans Affairs Medical Center, the Family Practice Clinic at Merrithew Memorial Hospital, or local physicians and were enrolled in a longitudinal study of cognition in HIV-l infection. Potential subjects with a history of neurologic disease, closed head injury with loss of consciousness, learning disability, major psychiatric disorder, current neuroleptic treatment, or current substance abuse were excluded from the study. All subjects had
HIV
volunteered
FIGURE
to participate
in
the
study,
were
without
overt clinical signs of dementia, and were capable of giving informed consent. Of the 30 subjects, 11 were clinically symptomatic (Centers for Disease Control stages IV-A or IV-C2),27 but without AIDS-defining disorders; 14 were completely clinically asymptomatic; and 5 were asymptomatic except for persistent generalized lymphadenopathy (CDC stages II and III, respectively). Subjects’ clinical staging was verified by their primary physicians on the basis of the physical examination and current laboratory values. All subjects were ambulatory and without serious clinical disease at the time of testing, which was conducted on an outpatient
1.
Demographic
data for all subjects Mean
± SD CD4 Lymphocyte
Subject
Age
Group positive,
positive,
(n
=
Count
symptomatic
(n=11) HIV
Education
38.7±8.1
14.1±2.4
324.4±209.6
36.2
15.0
559.0
asymptomatic
19)
Control (n=14)
± 6.6
34.5±11.8
1.
Procedure
task.
S
=
for the covert subject; SOA
± 2.5
15.5±2.1
=
± 270.8
Notobtained
orienting of visual attention stimulus-onset asynchrony.
E fixates
on central
box
EJ CUE:Right
Variable
LI
or left peripheral
SOA
box
(50,
brightens
150,
550,
1000
msec)
basis.
The control group comprised 14 healthy seronegative right-handed gay men (serostatus documented by enzyme-linked immunosorbent assay in all cases). The three groups (symptomatic, asymptomatic, and control) did not differ in mean age (F < 1) or years of education (F = 1.06, df = 2,41, P < 0.40; Table 1). Approximately 20% of the sample were African American or Hispanic. Seven experimental subjects and I control subject had a history of alcohol abuse, but no subject had abused alcohol or drugs for at least 3 years prior to study entry. Six asymptomatic and 9 symptomatic subjects were taking AZI. Three asymptomatic and 3 symptomatic subjects were taking anxiolytic or antidepressant medication. Written informed consent was obtained from all subjects.
290
TABLE
s TARGET
appears
in peripheral
box
/\ LI
VALID
J
EORJ
TRIAL:
Target
and
SAME
side
cue
INVALID on
Target
N /
S presses
VOLUME
response
4
TRIAL: on
OPPOSITE
side
of cue
#{149} SUMMER
1992
button
#{149} NUMBER
3
MARTIN
followed
consisted of an The target appeared 50, 150,550, or 1,000 msec after cue onset. (Using Posner’s terminology, the interval between the cue and the onset of the target stimulus is designated the “SOA,” or stimulus-onset asynchrony.) The subject was required to press a key using the dominant hand as quickly as possible following target onset, and time for the subject’s keypress was recorded. The trial was terminated with the subject’s response or after 3 seconds. Eighty percent of trials were “valid,” with cue and target appearing in the same position, and the remainder of the trials were “invalid,” with cue and target appearing in opposite positions. Stimulus order was randomly determined. Subjects were reminded periodically to keep their eyes on the fixation box. Eye movements were monitored by the experimenter, and trials with overt eye movements were discarded. Error trials (responding prior to target onset) were automatically readministered. All subjects also completed the Beck Depression Inventory28 and the “state” version of the State-Trait Anxiety Inventory, two self-report measures of psychological distress, in order to determine whether depression or anxiety significantly influenced RI performance. Data were excluded from analysis for I asymptomatic subject and I control subject who had two outlying RI scores greater than 3 standard deviations from the group mean. Log-transformed RI for each group was analyzed first in separate validity (valid vs. invalid) x SOA (50, 150, 550, 1,000 msec) repeated-measures analyses of variance (ANOVAs). Iwo-tailed tests were used for all analyses. asterisk
by a target stimulus, in one of the peripheral
which boxes.
et al.
P
In a preliminary study, 30 nondemented human immunodeficiency virus (HIV-1) seropositive subjects without acquired immunodeficiency syndrome and 14 seronegative controls performed a reaction time measure of spatial attention. Compared with controls, sero positive asymptomatic subjects showed normal facilitation of reaction time at short cue-target intervals when attention was precued, but symptomatic subjects were impaired. However, asymptomatic subjects showed no evidence of normal inhibition of attention at the cued location at longer cue-target intervals, suggesting possible subtler spatial attentional deficits in this group. Cognitive slowing in HIV-1 infection may have an attentional component, with possible involvement of both automatic and controlled processes. (The Journal Neurosciences
of Neuropsychiatry 1992; 4:288-293)
and
Clinical
C
ognitive dementia
munodeficiency
slowing complex
is
a prominent feature of the associated with acquired im-
syndrome
(AIDS)1
and
can
be
dem-
onstrated reliably in patients with advanced human immunodeficiency virus (FIIV-1) infection using standard clinical neuropsychological tests of psychomotor functions, memory, and attention.2 Patients with constitutional symptoms but without AIDS-defining disorders show qualitatively similar but milder deficits on these clinical measures.3 The cognitive status of HIV antibodyseropositive asymptomatic subjects, as assessed by standard neuropsychological tests, is more controversial. Some investigators have reported evidence of impairment in asymptomatic subjects,4’5 but others have failed to demonstrate group differences between asymptomatic subjects and control subjects on these measures.6’7 These inconsistencies have led some investigators to conclude that the incidence of cognitive dysfunction among asymptomatic individUals is low and that positive findings can be attributed to experimental confounds such as substance abuse, psychiatric disorder, or employment of excessively liberal criteria for judgment of impairment.8 However, there is strong evidence that HIV-1 invades the central nervous system early in the course of systemic infection,9 and some asymptomatic patients complain of mild cognitive difficulties. A careful review of the literaReceived
September 13, 1991; revised December 12, 1991; accepted 1991. From the Departments of Neurology and Psychiatry and the Department of Medicine, University of California-Davis, and the Research Service, Veterans Affairs Medical Center, Martinez, California. Address reprint requests to Dr. Martin, Neurology Service (127), VA Medical Center, 150 Muir Rd., Martinez, CA 94553. Copyright C) 1992 American Psychiatric Press, Inc. December13,
288
VOLUME
4
#{149} NUMBER
3
#{149} SUMMER
1992
MARTIN
ture suggests that early central nervous system infection could be manifested by subclinical cognitive changes that are not detected reliably by the standard clinical instruments employed in most studies. Investigators have recently employed reaction time (RT) tasks developed in cognitive psychology as a means of evaluating subclinical cognitive slowing in asymptomatic subjects.10’6 These measures are well suited for studying patients with varying severity of infection because they are less subject to ceiling effects than clinical neuropsychological tests. In addition, specific component cognitive functions can be measured with RT tasks, and hypotheses about specific mechanisms of HIV-related cognitive dysfunction can be tested. For instance, tasks measuring primarily motor slowing (e.g., simple RT, which reflects the time to detect a stimulus and prepare and execute a motor response) or cognitive slowing (e.g., choice RT, which reflects these factors plus time to make some type of discrimination) can be administered, and different components of performance can be isolated. The potential contribution of RT tasks in the study of HIV-related neurobehavioral disorders is acknowledged in recently issued National Institute of Mental Health guidelines for assessment of AIDS-related cognitive changes.’7 Participants recommended inclusion of cognitive psychology tasks in the neurobehavioral battery because of their “greater complexity and potentially their increased sensitivity to subtle cognitive changes” (p. 964). Further, a recent longitudinal investigation demonstrated the prognostic value of RT measures by showing that performance of RT tasks was significantly correlated with rate of CD4 lymphocyte decline)6
Despite the potential advantages of cognitive neuropsychological measures, the literature on RT performance in HIV-positive subjects has been inconsistent. Some studies have found early deficits on RT tasks,’#{176}” but others have not.’2’4 It is difficult to evaluate comprehensively the available literature on RI performance in HIV-positive subjects. Experimental methods differ widely across studies and are not always described in sufficient detail for independent replication. The tasks employed have clearly differed in response requirements (e.g., single keypress go/no go versus choice of two or more keys), presence or absence of an alerting stimulus preceding the “go” stimulus, length of intertrial or interstimulus interval, and degree of task difficulty (e.g., simple two-choice discrimination versus complex discrimination). These parameters systematically influence absolute RI values; thus, obtained results are not directly comparable across studies. In addition, because of these differing task requirements, it cannot be assumed that the same underlying cognitive processes have been evaluated.
JOURNAL
OF
NEUROPSYCHIATRY
et al.
Performance variability has also been a common problem for RI studies. Experiments have not always included sufficient numbers of experimental trials for adequate reliability,’8 possibly because RI tasks have often been administered as part of long neurobehavioral testing protocols. Finally, the theoretical basis for task selection has not always been specified, despite ample experimental
literature
on
RI
performance
in other
pa-
tient populations with similar dementias)2’ lask selection in many available studies appears guided by the implicit assumption that cognitive slowing should be evident on all RI tasks, regardless of a specific task’s cognitive
requirements.
Different
experimental
tasks
may measure different underlying cognitive functions, which may not necessarily be affected by HIV-1. In our laboratory, we are conducting a series of experiments evaluating cognition in HIV infection, using RI measures selected specifically to evaluate component cognitive functions underlying cognitive slowing. In an initial study,’5 we found that, compared with controls, both symptomatic and asymptomatic HIV-seropositive subjects showed an excessively long decision time, as measured by the discrepancy between simple and choice RI. In contrast, H1V-positive subjects performed normally on a control measure of rate of verbal encoding. These findings support the hypothesis that subtle cognitive slowing is present in early HIV-1 infection and does not represent a nonspecific effect across all speeded tasks. The present study investigated one possible mechanism of cognitive slowing. We evaluated the ability of HIV-seropositive
and
control
subjects
to move
attention
in the visual field without overt eye movements, using a task that has been employed to study spatial attention in patients with focal cortical lesions, schizophrenia, and subcortical dementing disorders.24 Developed by Posner, this “covert orienting” task measures time to detect a target following introduction of a visual cue. “Valid” cues summon attention to the location of the forthcoming target; “invalid” cues summon attention away from the target location, and thus subjects require additional time to disengage and move attention from an invalid cue to the target location. The “validity effect,” faster RI on valid than invalid trials (e.g., facilitation of speed of visual orienting when attention is precued), is most prominent with about 150 msec between cue and target. In addition, normal subjects typically show a reversal (faster RI to invalid than valid cues) at longer cue-target intervals. Ihis effect, the “inhibition of return,” indicates inhibition of movement of attention to previously attended locations, possibly representing facilitation of visual exploration of novel targets.26 For the purposes of this preliminary study, we were primarily concerned with the validity effect as the vari-
289
SPATIAL
ATTENTION
IN HIV-1
INFECTION
able of major interest because this effect has been studied more extensively in cognitive neuropsychology than the inhibition of return and because a more specific hypothesis about this effect could be formulated. We predicted that the validity effect would be diminished or delayed in H1V-seropositive subjects if cognitive slowing affected this type of attentional process.
METHODS
Procedure All subjects were administered the covert orienting task, which consisted of 400 RI trials (Figure 1) administered by an Al-compatible personal computer. A display of 3 boxes was present continuously on the computer monitor. On each trial, the subject fixated on the central box in the display. Each trial began with a 500-msec 880-Hz auditory warning stimulus, followed 100 msec after its offset by the appearance of a 300-msec cue (brightening of one of the two peripheral boxes located approximately 8 degrees on either side of the fixation box). The cue was
Subjects
Experimental subjects were I left-handed and 29 righthanded men, including 27 gay or bisexual subjects and 3 subjects with a history of transfusion or intravenous drug abuse. All experimental subjects were recruited through the Infectious Disease Clinic at the Martinez Veterans Affairs Medical Center, the Family Practice Clinic at Merrithew Memorial Hospital, or local physicians and were enrolled in a longitudinal study of cognition in HIV-l infection. Potential subjects with a history of neurologic disease, closed head injury with loss of consciousness, learning disability, major psychiatric disorder, current neuroleptic treatment, or current substance abuse were excluded from the study. All subjects had
HIV
volunteered
FIGURE
to participate
in
the
study,
were
without
overt clinical signs of dementia, and were capable of giving informed consent. Of the 30 subjects, 11 were clinically symptomatic (Centers for Disease Control stages IV-A or IV-C2),27 but without AIDS-defining disorders; 14 were completely clinically asymptomatic; and 5 were asymptomatic except for persistent generalized lymphadenopathy (CDC stages II and III, respectively). Subjects’ clinical staging was verified by their primary physicians on the basis of the physical examination and current laboratory values. All subjects were ambulatory and without serious clinical disease at the time of testing, which was conducted on an outpatient
1.
Demographic
data for all subjects Mean
± SD CD4 Lymphocyte
Subject
Age
Group positive,
positive,
(n
=
Count
symptomatic
(n=11) HIV
Education
38.7±8.1
14.1±2.4
324.4±209.6
36.2
15.0
559.0
asymptomatic
19)
Control (n=14)
± 6.6
34.5±11.8
1.
Procedure
task.
S
=
for the covert subject; SOA
± 2.5
15.5±2.1
=
± 270.8
Notobtained
orienting of visual attention stimulus-onset asynchrony.
E fixates
on central
box
EJ CUE:Right
Variable
LI
or left peripheral
SOA
box
(50,
brightens
150,
550,
1000
msec)
basis.
The control group comprised 14 healthy seronegative right-handed gay men (serostatus documented by enzyme-linked immunosorbent assay in all cases). The three groups (symptomatic, asymptomatic, and control) did not differ in mean age (F < 1) or years of education (F = 1.06, df = 2,41, P < 0.40; Table 1). Approximately 20% of the sample were African American or Hispanic. Seven experimental subjects and I control subject had a history of alcohol abuse, but no subject had abused alcohol or drugs for at least 3 years prior to study entry. Six asymptomatic and 9 symptomatic subjects were taking AZI. Three asymptomatic and 3 symptomatic subjects were taking anxiolytic or antidepressant medication. Written informed consent was obtained from all subjects.
290
TABLE
s TARGET
appears
in peripheral
box
/\ LI
VALID
J
EORJ
TRIAL:
Target
and
SAME
side
cue
INVALID on
Target
N /
S presses
VOLUME
response
4
TRIAL: on
OPPOSITE
side
of cue
#{149} SUMMER
1992
button
#{149} NUMBER
3
MARTIN
followed
consisted of an The target appeared 50, 150,550, or 1,000 msec after cue onset. (Using Posner’s terminology, the interval between the cue and the onset of the target stimulus is designated the “SOA,” or stimulus-onset asynchrony.) The subject was required to press a key using the dominant hand as quickly as possible following target onset, and time for the subject’s keypress was recorded. The trial was terminated with the subject’s response or after 3 seconds. Eighty percent of trials were “valid,” with cue and target appearing in the same position, and the remainder of the trials were “invalid,” with cue and target appearing in opposite positions. Stimulus order was randomly determined. Subjects were reminded periodically to keep their eyes on the fixation box. Eye movements were monitored by the experimenter, and trials with overt eye movements were discarded. Error trials (responding prior to target onset) were automatically readministered. All subjects also completed the Beck Depression Inventory28 and the “state” version of the State-Trait Anxiety Inventory, two self-report measures of psychological distress, in order to determine whether depression or anxiety significantly influenced RI performance. Data were excluded from analysis for I asymptomatic subject and I control subject who had two outlying RI scores greater than 3 standard deviations from the group mean. Log-transformed RI for each group was analyzed first in separate validity (valid vs. invalid) x SOA (50, 150, 550, 1,000 msec) repeated-measures analyses of variance (ANOVAs). Iwo-tailed tests were used for all analyses. asterisk
by a target stimulus, in one of the peripheral
which boxes.
et al.
P
