Assessment of cognitive function in systemic lupus erythematosus, rheumatoid arthritis, and multiple sclerosis by computerized neuropsychological tests

Authors


Abstract

Objective

Computerized neuropsychological testing may facilitate screening for cognitive impairment in systemic lupus erythematosus (SLE). This study was undertaken to compare patients with SLE, patients with rheumatoid arthritis (RA), and patients with multiple sclerosis (MS) with healthy controls using the Automated Neuropsychological Assessment Metrics (ANAM).

Methods

Patients with SLE (n = 68), RA (n = 33), and MS (n = 20) were compared with healthy controls (n = 29). Efficiency of cognitive performance on 8 ANAM subtests was examined using throughput (TP), inverse efficiency (IE), and adjusted IE scores. The latter is more sensitive to higher cognitive functions because it adjusts for the impact of simple reaction time on performance. The results were analyzed using O'Brien's generalized least squares test.

Results

Control subjects were the most efficient in cognitive performance. MS patients were least efficient overall (as assessed by TP and IE scores) and were less efficient than both SLE patients (P = 0.01) and RA patients (P < 0.01), who did not differ. Adjusted IE scores were similar between SLE patients, RA patients, and controls, reflecting the impact of simple reaction time on cognitive performance. Thus, 50% of SLE patients, 61% of RA patients, and 75% of MS patients had impaired performance on ≥1 ANAM subtest. Only 9% of RA patients and 11% of SLE patients had impaired performance on ≥4 subtests, whereas this was true for 20% of MS patients.

Conclusion

ANAM is sensitive to cognitive impairment. While such computerized testing may be a valuable screening tool, our results emphasize the lack of specificity of slowed performance as a reliable indicator of impairment of higher cognitive function in SLE patients.

Nervous system disease is common in patients with systemic lupus erythematosus (SLE) and encompasses a wide range of manifestations, ∼30–40% of which are attributable to SLE (1). Symptoms related to cognitive dysfunction are frequent, and formal neuropsychological assessment techniques have consistently found a higher frequency of cognitive impairment in SLE patients compared with healthy and disease controls (2–6). Variability in the frequency of such deficits is due to several factors, including bias in the selection of patients for study and operational decisions regarding the definition of cognitive impairment.

The American College of Rheumatology (ACR) has proposed a battery of neuropsychological tests for the assessment of cognitive function in SLE (7). Although comprehensive, there are several factors which limit the widespread use of these tests. For example, they are time consuming, require specialized training to administer, and are subject to large practice effects. Although not a replacement for more detailed neuropsychological assessment, computerized neuropsychological testing may facilitate the rapid and efficient screening by nonexperts of SLE patients for cognitive impairment.

The Automated Neuropsychological Assessment Metrics (ANAM) is one such computerized test battery, comprising subtests which are modified versions of standard neuropsychological tests (8, 9). ANAM was designed to examine information processing efficiency in tasks ranging from simple reaction time to attention/concentration, learning and memory, and executive functions (10). Most cognitive functions consist of multiple independent processes with diffuse neuroanatomical substrates, such that it can be difficult to determine exactly how a specific test score is achieved (11). However, reaction time paradigms that measure response latencies in addition to performance accuracy may be able to overcome some of these limitations and better dissociate overall slowing of performance from changes in higher-order cognitive functions.

Simple and choice reaction time tasks can provide reliable, valid, and sensitive measures that have psychometric properties comparable with conventional neuropsychological tests (12, 13). Simple reaction time tasks can be used to provide a “baseline” measure that accounts for sensorimotor processing speed, while choice reaction time paradigms can be used to examine efficiency within higher-order cognitive functions that may be more indicative of specific central nervous system (CNS) damage. ANAM has been shown to be sensitive to subtle cognitive impairments in traumatic brain injury (14, 15), early dementia (16), and multiple sclerosis (MS) (17). Potential advantages of the ANAM for use in patients with SLE include its reportedly limited dependence on proficiency in English or in reading ability as well as its strong association with performance on the neuropsychological test battery recommended by the ACR (18).

As is the case for any diagnostic or screening procedure, the selection of controls is critical to the interpretation of test results. ANAM has previously been used in the evaluation of SLE patients in a limited number of studies (6, 18–22), but the inclusion of patients with other chronic diseases has been lacking. In the present study, we used ANAM to evaluate the cognitive performance of SLE patients as compared with patients with rheumatoid arthritis (RA), patients with MS, and healthy controls. We deliberately included patients with a chronic disease lacking CNS involvement (RA) and one which specifically involves the CNS (MS) to compare and contrast the frequency and characteristics of impaired performance on cognitive tasks as detected by ANAM in these conditions.

PATIENTS AND METHODS

Patients.

All study subjects provided informed consent for procedures approved by the Capital District Health Authority Research Ethics Board. Sixty-eight patients with SLE, 33 patients with RA, 20 patients with MS, and 29 healthy controls participated in the study. SLE and RA patients were recruited from the Dalhousie Lupus Clinic and general rheumatology clinics, respectively, in the Division of Rheumatology. All of these patients fulfilled the ACR classification criteria for SLE or RA (23, 24). Global SLE disease activity was measured by the SLE Disease Activity Index (SLEDAI) (25) and cumulative organ damage by the Systemic Lupus International Collaborating Clinics (SLICC)/ACR Damage Index (26). In RA patients, disease activity and impact were assessed according to the number of tender and swollen joints, erythrocyte sedimentation rate, C-reactive protein level, and Health Assessment Questionnaire score. MS patients were recruited from the Dalhousie MS Research Unit. All MS patients had stable, clinically definite, relapsing-onset MS according to the Poser criteria (27) and at worst required unilateral assistance to walk 100 meters. Neurologic disability was rated by the attending neurologists using the Expanded Disability Status Scale (EDSS) (28). This scale allows for grading of impairment due to MS on a scale of 0–10, where 0 indicates no impairment and 10 indicates death due to MS. All MS patients had an EDSS score of ≤6.0. Patients were excluded if they had comorbid neurologic or psychiatric disorders or a history of substance abuse or learning disability. Healthy control participants who met the same exclusion criteria were recruited through local advertisements. All participants had normal or corrected-to-normal vision and reported no vision problems at the time of the study.

The following data were collected on all study participants: age, sex, ethnicity, education, and medication use. In SLE and RA patients, neuropsychiatric events were characterized using the ACR case definitions and were diagnosed by clinical evaluation supported with appropriate investigations as per the ACR glossary (7). Attribution of neuropsychiatric events to SLE and non-SLE causes was determined as previously described (29).

ANAM testing.

The ANAM test battery (10) includes a variety of tasks designed to assess neurocognitive efficiency via measures of response time and accuracy. Most ANAM tasks resemble commonly used neuropsychological tests but have been modified to require a relatively simple subject–computer interface in which the required responses are either a yes/no or a same/different discrimination as indicated by pressing 1 of 2 mouse buttons. Each ANAM subtest is preceded by practice trials that include visual feedback regarding response accuracy, but test trials do not include feedback. Two simple reaction time tasks (20 trials each), in which participants are asked to respond as quickly as possible to a cue (an asterisk) in the center of the screen, are administered at the beginning and at the end of the ANAM.

Learning and recall are examined using code substitution subtests in which participants are first asked to determine whether a series of number/symbol pairings are consistent with a standard set provided at the top of the screen (code substitution; 76 trials), and later to discriminate correct pairings from incorrect ones without the answer key (code substitution delayed; 36 trials). Working memory is assessed using both the mathematical processing subtest (20 trials), which requires participants to solve a series of mathematical operations and to determine whether the answer is greater than or less than 5, and a version of the Sternberg memory scanning paradigm (30 trials) that requires participants to memorize a fixed set of 6 upper case letters and then determine whether letters presented later are part of this set. Sustained attention is measured using a continuous performance subtest (81 trials), in which individuals are presented with a single digit every 950–1,200 msec and are asked to indicate whether each digit is the same or different from the one that directly preceded it. Visual–spatial processing is tested using the matching grids subtest (20 trials), in which participants are presented with two 4 × 4–block grid designs and are asked to indicate whether they are the same or different. Finally, the match to sample subtest (20 trials) is used to assess short-term memory, attention, and visual-spatial discrimination. It requires participants to memorize a 4 × 4–block grid design and then determine which of 2 designs, presented after a delay of 5,000–5,100 msec, is the same as that studied.

Data analysis.

Each ANAM subtest generates measures of mean reaction time, accuracy, and “throughput” (TP). TP uses an individual's reaction time and accuracy on a specific subtest to calculate their average number of correct responses per minute and is thought to offer a more “stable” index of performance on a timed test where speed–accuracy tradeoff may occur (30). However, a limitation of TP is that on tasks that assess higher cognitive functions (e.g., working memory or executive functions), it does not allow for differentiation between sensorimotor efficiency and mental processing efficiency. For instance, on an ANAM arithmetic task, the number of average correct responses per minute is a result of the efficiency of perceptual and motor processes (i.e., the efficiency of perceptual processing of the target and executing a motor program to press the response button) as well as the efficiency of the mental processes involved in mathematical computation and decision making. Because TP is calculated using mean reaction time per unit of performance time (i.e., mean reaction time per minute), it does not account for the contribution of sensorimotor efficiency to the final score. In clinical populations, this presents an obstacle to interpreting differences in performance efficiency both between and within groups of subjects. Sensorimotor slowing can be a relatively nonspecific index of neurologic or neuromuscular impairment, which can be affected to a similar degree by a variety of medical conditions. As such, the utility of TP in effectively differentiating clinically meaningful groups of subjects who differ in their cognitive efficiency can be questioned.

In order to overcome this limitation, while still providing a measure of overall performance that simultaneously takes account of speed and accuracy, we used the method first recommended by Townsend and Ashby (31) that has subsequently been referred to as “inverse efficiency” (IE) (32). IE is computed by dividing the mean speed of responding by the proportion of correct responses within each subtest. Like TP, IE accounts for any potential speed–accuracy tradeoff. However, since IE calculations are based on overall speed and accuracy (not just over a limited period of time), it is possible to “adjust” IE scores in order to minimize the influence of sensorimotor speed on the final outcome. In this study, the 2 simple reaction time subtests (SRT_1 and SRT_2) that require perception and motor response to a simple stimulus (an asterisk) were considered to represent an index of “pure” sensorimotor speed. For each participant, the average response time on both of these subtests was used to calculate a “mean simple reaction time” score that represented overall sensorimotor speed (SRT_m = [SRT_1 + SRT_2]/2). This SRT_m score was then subtracted from mean reaction time scores on individual ANAM subtests for each subject, and the remaining response time was presumed to reflect the speed of mental processing required for each subtest. Adjusted IE scores were thus calculated using these “remainder reaction time” scores, in order to provide a measure of the efficiency of cognitive processing on each subtest. Unlike TP, adjusted IE scores calculated in this manner should differentiate the effects of neurologic changes affecting cognitive processing from the effects of neurologic or other medical conditions on basic sensory and motor efficiency. Thus, differences in adjusted IE scores between clinical and control groups are more likely to reflect the specific cognitive effects of the clinical conditions.

Summary statistics and t-tests were used to examine differences in the demographic characteristics of the groups. Linear regression was used to examine whether group differences in performance were affected by fatigue, as measured by the difference in simple reaction times at the beginning versus the end of the ANAM session for each subject. Overall group differences for all 4 groups as well as pairwise comparisons between groups across all subtests for both adjusted IE scores and for TP scores were analyzed using O'Brien's generalized least squares test (33, 34). All comparisons were adjusted for age and education. The frequencies of impaired performance of tasks based on adjusted IE scores across groups were compared. Possible predictors of impaired task performance within the 3 clinical groups were examined by ordinal regression. For simple reaction time, overall group differences were examined by linear regression, adjusting for age and education. The frequency of impairment based on simple reaction time and its relevant predictors were checked by logistic regression.

RESULTS

Patient characteristics.

The demographic characteristics of each group are shown in Table 1. SLE patients were older than healthy controls (t = −2.14, P = 0.04), and RA patients were older than MS patients (t = −2.99, P = 0.004) and controls (t = −3.73, P < 0.0001). The number of years of education was lower in RA patients compared with controls (t = −2.65, P = 0.01), and in SLE patients compared with controls (t = −2.77, P = 0.007). When these group differences in age and education were accounted for, all clinical groups had higher scores than controls on self-reported symptoms of depression (SLE t = −2.75, P = 0.007; RA t = −2.92, P = 0.005; MS t = −2.56, P = 0.02). Similarly, SLE and MS patients had higher scores than controls on self-reported anxiety (SLE t = −2.21, P = 0.03; MS t = −2.05, P = 0.05). None of these mean scores for the individual groups approached the Hospital Anxiety and Depression Scale cutoff scores of ≥11, recommended for the identification of the probable presence of depression and/or generalized anxiety disorder (35). SLE patients had mild disease activity and low cumulative organ damage, as reflected by SLEDAI and SLICC/ACR Damage Index scores. The disease-specific summary scores for RA and MS also indicated low levels of disease activity and disability.

Table 1. Demographic and clinical characteristics of the patients and controls*
 SLE patients (n = 68)RA patients (n = 33)MS patients (n = 20)Controls (n = 29)
  • *

    Except where indicated otherwise, values are the mean ± SD. SLE = systemic lupus erythematosus;– = not applicable; HADS-D = Hospital Anxiety and Depression Scale depression subscale; HADS-A = Hospital Anxiety and Depression Scale anxiety subscale; NPSLE = neuropsychiatric SLE; SLEDAI = SLE Disease Activity Index; SLICC/ACR = Systemic Lupus International Collaborating Clinics/American College of Rheumatology; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; HAQ = Health Assessment Questionnaire; EDSS = Expanded Disability Status Scale; ASA = acetylsalicylic acid; NSAIDs = nonsteroidal antiinflammatory drugs.

  • Biologic agents were tumor necrosis factor α inhibitors (etanercept, infliximab, or adalimumab) in the rheumatoid arthritis (RA) group and interferon-β therapies or glatiramer acetate in the multiple sclerosis (MS) group.

No. of women/men63/532/120/027/2
Age, years45.5 ± 13.449.8 ± 10.241.9 ± 7.440.2 ± 9.8
Ethnicity, %    
 White92.693.9100100
 Other7.46.100
Education, years14.9 ± 2.514.4 ± 3.215.1 ± 2.716.5 ± 2.0
Years since diagnosis11.9 ± 9.512.0 ± 11.06.9 ± 5.1
HADS-D score3.9 ± 4.24.6 ± 4.24.0 ± 2.82.1 ± 2.3
HADS-A score6.0 ± 4.25.6 ± 4.26.4 ± 3.94.2 ± 3.4
Cumulative neuropsychiatric events, % of patients66.245.510020.7
Cumulative NPSLE events, % of patients32.4
Neuropsychiatric events <4 weeks of assessment, % of patients45.630.303.4
NPSLE events <4 weeks of assessment, % of patients25.0
SLEDAI4.4 ± 4.2
SLICC/ACR Damage Index1.3 ± 1.9
Tender joint count1.8 ± 3.7
Swollen joint count1.9 ± 3.1
ESR14.5 ± 13.7
CRP3.9 ± 3.7
HAQ1.0 ± 1.1
EDSS3.1 ± 1.5
Current medications, % of patients    
 Prednisone (mean ± SD mg/day)16.2 (13.0 ± 18.9)6.1 (5.8 ± 2.7)0 (0)0 (0)
 Biologic agents033.3800
 ASA (low dose)25000
 NSAIDs16.242.400
 Coxibs1.59.100
 Antimalarials48.545.500
 Methotrexate14.763.600
 Azathioprine10.3000
 Mycophenolate4.4000
 Cyclophosphamide2.9000

The potential effect of fatigue on test performance was examined by linear regression, comparing group differences in the change between the score on the simple reaction time subtest administered at the start of ANAM testing (SRT_1) and the score on the simple reaction time subtest administered at the end of the session (SRT_2). Individual subjects' mean reaction times on SRT_1 were subtracted from those on SRT_2, and larger differences were taken as an indication of a decline in overall speed of sensorimotor processing that could be attributed to fatigue. However, the absence of any differences between groups on simple reaction time difference scores (P = 0.82) suggested that any effects of fatigue on ANAM performance were equal across groups.

Group differences in performance on ANAM testing.

The mean simple reaction time group differences were examined using ordinary linear regression. After controlling for the effect of age (P < 0.01), controls were found to have lower mean simple reaction time than patients with MS (P < 0.001), SLE (P < 0.05), or RA (P < 0.05), whereas the only differences between the patient groups was a lower mean simple reaction time for SLE patients than for MS patients (P < 0.05) (Figure 1). Average group performance on each of the ANAM subtests is illustrated in Figure 2 (for TP scores) and Figure 3 (for adjusted IE scores). Overall group differences were found for both TP (F[3,140] = 7.49, P < 0.01) and adjusted IE (F[3,138] = 4.16, P = 0.01). MS patients demonstrated the lowest TP scores and the highest adjusted IE scores (i.e., least efficient performance among the groups), whereas controls had the highest TP and lowest adjusted IE scores.

Figure 1.

Simple reaction time (SRT) scores in the 4 test groups (healthy controls, patients with rheumatoid arthritis [RA], patients with systemic lupus erythematosus, and patients with multiple sclerosis [MS]). Mean simple reaction time scores were calculated by combining participants' performance on the 2 simple reaction time subtests of the Automated Neuropsychological Assessment Metrics ([SRT_1 + SRT_2]/2). Bars show the mean and SEM.

Figure 2.

Subtest throughput (TP) scores in the 4 test groups (healthy controls, patients with rheumatoid arthritis [RA], patients with systemic lupus erythematosus, and patients with multiple sclerosis [MS]). TP represents the average number of correct responses per minute. Higher scores represent better performance. The individual subtest data points for each group are connected for illustration purposes only. Bars show the mean ± SEM. CPT = continuous performance; ST6 = Sternberg memory scanning paradigm; CDS = code substitution; CDD = code substitution delayed; MTG = matching grids; MSP = match to sample; MTH = mathematical processing.

Figure 3.

Subtest inverse efficiency (IE) scores in the 4 test groups (healthy controls, patients with rheumatoid arthritis [RA], patients with systemic lupus erythematosus, and patients with multiple sclerosis [MS]) adjusted for simple reaction time. Adjusted IE scores were calculated by first subtracting individual average simple reaction time scores from individual average reaction time scores on the other Automated Neuropsychological Assessment Metrics subtests. The resulting scores were divided by the proportion of correct responses on each subtest to account for the speed–accuracy tradeoff and to reflect the efficiency of performance. Higher scores represent poorer/less efficient performance. The individual subtest data points for each group are connected for illustration purposes only. Bars show the mean ± SEM. See Figure 2 for other definitions.

Pairwise comparisons between the 3 disease groups on overall ANAM performance, adjusted for age and education, demonstrated lower TP for MS patients than for either SLE patients (t = 3.03, P < 0.01) or RA patients (t = 4.64, P < 0.01) while the SLE and RA groups did not differ (t = 0.44, P > 0.05). The same pattern of results held for the adjusted IE scores, with MS patients demonstrating higher adjusted IE scores in comparison with both SLE patients (t = 2.54, P = 0.01) and RA patients (t = 3.03, P < 0.01), while the SLE and RA groups once again did not differ (t = 0.28; P > 0.05).

Finally, pairwise comparisons between the control group and each of the 3 disease groups were conducted, adjusting for age and education. On measures of TP, controls differed from both the MS group (t = 3.42, P < 0.01) and the SLE group (t = 2.53, P < 0.05) but not from the RA group (t = 1.10, P > 0.05). However, for measures of adjusted IE score, only the MS group differed from controls (t = 3.55, P < 0.01), while no differences in adjusted IE scores were found between controls and either the SLE patients (t = 1.52, P > 0.05) or the RA patients (t = 0.53; P > 0.05).

Group differences in frequency of impairment on ANAM tasks.

In order to explore potential differences between patient groups in their frequency of overall cognitive impairment on ANAM subtests, subjects' adjusted IE scores on each subtest were converted to Z scores using the distributions of the control group performance as the reference. Patients were considered impaired on an ANAM subtest if their Z score differed from that of controls by 1.5 or more (i.e., performance ≥1.5 SD worse than the mean performance of the control group). In a separate analysis, this same approach was also used to examine impairment in mean simple reaction time. Predictors of cognitive impairment within groups were examined using ordinal regression, with overall frequency of impaired performance of tasks used as the outcome variable within groups. Only age was shown to be positively associated with a higher frequency of impaired performance of tasks in all 3 disease groups (P < 0.01). None of the other variables, including disease duration, were related to cognitive impairment within groups. The same held true for the analyses of simple reaction time impairment, with age again being the only variable associated with a higher probability of impaired simple reaction time (P < 0.01).

MS patients had impaired performance on a greater number of subtests than SLE patients (P < 0.05) and a trend toward impaired performance on a greater number of subtests than RA patients (P = 0.07), even after adjustment for age. SLE and RA patients did not differ (P = 0.58). Although 61% of RA patients and 50% of SLE patients had impaired performance on ≥1 ANAM subtest, this was in contrast to a frequency of 75% for MS patients. Only 9% of RA patients and 11% of SLE patients had impaired performance on ≥4 ANAM tasks (i.e., more than half of the tasks) whereas 20% of MS patients were impaired to this extent. For simple reaction time, there were no differences in the proportion of impaired subjects between the patient groups (MS patients 45%, SLE patients 43%, and RA patients 51%).

DISCUSSION

Cognitive impairment is reported to be one of the most common manifestations of neuropsychiatric SLE (NPSLE) (2–6). Although sometimes profound in individual cases, the majority of patients have subtle and frequently subclinical cognitive deficits, which are evanescent rather than progressive over time (36–39). Recently, the use of computerized neuropsychological testing, such as ANAM, has revealed poorer performance in SLE patients compared with healthy controls (6, 18–22). Our study was designed to compare and contrast the cognitive abilities of SLE patients with those of patients with other chronic diseases as well as with those of healthy controls, as determined using ANAM. We have confirmed that SLE patients have slower performance on cognitive tasks than that of controls but comparable with that of RA patients and better than that of MS patients with mild neurologic disability. Furthermore, conventional outcomes of ANAM testing did not distinguish between the nonspecific reduction of performance efficiency and more specific abnormalities in higher cognitive functions, evaluated such as working memory and executive tasks.

Cognitive dysfunction, evaluated using formal clinical neuropsychological assessment techniques, has been reported in up to 80% of SLE patients (40), although most studies have found a prevalence between 17% and 66% (41, 42). Many individual patients have subclinical deficits. For example, a review of 14 cross-sectional studies of cognitive function in SLE revealed subclinical cognitive impairment in 11–54% of patients (41). A single pattern of SLE-associated cognitive dysfunction has not been found, but commonly identified abnormalities include overall cognitive slowing, decreased attention, impaired working memory, and executive dysfunction (e.g., difficulty with multitasking, organization, and planning). Since the majority of SLE patients with cognitive impairment have relatively mild deficits, the careful selection and assessment of cognitive performance in control groups is of critical importance for defining expected levels of function in healthy individuals and those with chronic diseases other than SLE. Although cognitive impairment may be viewed as a distinct subset of NPSLE, it can also serve as a surrogate of overall brain health in SLE patients, and it may be affected by a variety of factors, including other neuropsychiatric syndromes.

ANAM has been used in previous cross-sectional (6, 18–20, 22) and longitudinal (21) studies to evaluate cognitive performance in SLE. It has been validated by comparison with the ACR recommended battery of neuropsychological tests (18–20) and found to have a sensitivity for the detection of cognitive impairment of 76.2%, a specificity of 82.8%, and overall correct classification of 80% (18). The frequency of cognitive impairment reported for studies using ANAM was 69% in adult SLE patients with a mean disease duration of 8 years (6) and 59% in patients with childhood-onset SLE who had a mean age of 16.5 years at the time of assessment (19). In adult patients studied within 9 months of SLE diagnosis, cognitive impairment was present in 21–61% of cases, depending on the stringency of the definition of impairment (22).

All of those previous studies used either published ANAM norms, which were derived mostly from young males in the US military (43), or healthy controls recruited from a single academic center (6) to determine the presence of cognitive impairment. In our study, the frequency of cognitive impairment was 11–50% in SLE patients, depending on the stringency of the decision rules, when compared with locally recruited healthy controls. However, this frequency was comparable with that seen in patients with RA (9–61%) and lower than that in MS patients (20–75%). That the performance of SLE patients on ANAM testing was better than that of patients with stable demyelinating disease is perhaps to be expected. However, the comparable frequency of abnormalities in patients with RA, which does not primarily affect the CNS, raises concerns regarding the presumed etiology of deficits detected by ANAM testing.

Cognition is the sum of intellectual functions that result in thought. It includes reception of external stimuli, information processing, learning, storage, and expression. Disturbance of even one of these functions can result in disruption of normal thought production and can present as cognitive dysfunction, perhaps most easily seen on sensitive timed measures of performance. In all previous studies of SLE, the primary outcome of performance on ANAM subtests has been TP, a measure of the average number of correct responses per minute, which has been proposed as the best ANAM indicator of cognitive efficiency (10, 20). However, in the assessment of higher cognitive functions, TP does not distinguish between reduced sensorimotor efficiency and impaired mental processing. Thus, we generated adjusted IE scores as an indicator of efficiency of mental processing that, while still controlling for individual differences in speed–accuracy tradeoff, provided a measure that more clearly reflected the speed of information processing. The finding that TP scores in both MS and SLE patients differed significantly from those in the healthy controls but that adjusted IE scores were significantly different only for the MS patients suggests that the slowing in ANAM performance in SLE patients as a group is not necessarily attributable to a decline in their efficiency of mental processing.

In both SLE and non-SLE populations, there can be several potential causes of reduced performance of cognitive tasks other than primary CNS disease (44). Many of these have been addressed in the current study. For example, the analyses of differences in ANAM scores between groups were adjusted for age, the strongest predictor of cognitive decline, and years of education. The potential impact of fatigue, mood, concurrent neuropsychiatric disease, and disease duration was not significant. The patient groups also had comparable disease status, in that patients with SLE, RA, and MS all had quiescent disease activity with relatively limited disability at the time of ANAM testing. There was also considerable overlap in the medications used by patients with SLE and RA. Indeed, it is possible that subtle nonspecific neurotoxicity from long-term use of immunosuppressive agents (45, 46) may account for nonspecific cognitive slowing in some SLE and RA patients.

All patient groups had significantly poorer performance as compared with the healthy controls on timed tests involving sensorimotor and cognitive processing of information. However, for RA patients, the fact that they differed from controls on simple reaction time, but not on TP or adjusted IE score, suggests that in most, their abnormal ANAM results reflected deficiencies in sensorimotor speed. Nevertheless, regardless of the definition of impairment used, at least some proportion of the RA group demonstrated abnormal ANAM performance based on their adjusted IE scores and therefore demonstrated poor cognitive processing efficiency, possibly due to their long-term use of immunosuppressive agents (45, 46).

Interpretation of the findings in MS patients is more straightforward. Not only were they slower on simple reaction time relative to the controls, but they also differed from all other patient groups on adjusted IE scores, which is arguably the most accurate measure of cognitive processing efficiency derived from the ANAM. Moreover, they were the only patient group that differed from controls in adjusted IE scores. Thus, in patients with relatively mild and stable CNS disease, ANAM is clearly sensitive to slowed cognitive processing efficiency.

The performance of SLE patients was closer to that of RA patients than MS patients. Overall, SLE patients did not differ from RA patients, making it difficult to claim that ANAM is particularly good at identifying specific CNS disease in SLE patients. While there was an overall difference in TP between the SLE and healthy control groups, as has been reported previously for ANAM (6, 43), this was not true for adjusted IE scores, nor were SLE patients more often impaired than RA patients when adjusted IE cutoffs were used. As with RA, at least a portion of the SLE patients studied seemed to demonstrate a combination of sensorimotor processing deficiencies, together with impaired cognitive processing efficiency. In some SLE patients, this may again reflect nonspecific neurotoxicity of immunosuppressive agents (45, 46), while for others, it may well be the consequence of direct autoimmune inflammatory effects of SLE on the nervous system.

There are a number of limitations to the current study. First, neither formal neuropsychological assessment nor neuroimaging was used to look for clinical behavioral and neuroanatomical correlates of the abnormalities found on ANAM testing. Second, there were demographic differences, especially in age, between some groups that required adjustment in the analyses for these differences. Finally, the patients and controls did not undergo repeat testing with ANAM to determine whether the detected differences remained stable over time. The strengths of our study are the use of locally recruited healthy controls rather than published norms to determine expected performance on ANAM testing, the inclusion of 2 distinct disease control groups in order to examine the relative severity and etiology of cognitive impairment in the SLE patients, and the generation of adjusted IE scores to more specifically evaluate higher cognitive functions and the efficiency of mental processing.

What, therefore, is the role of ANAM in the assessment of cognitive dysfunction in SLE patients? It is clear from our findings and those of previous studies that ANAM is sensitive to the detection of both CNS and non-CNS causes of reduced cognitive efficiency. Furthermore, it may detect both the effects of a specific neuroanatomical injury, such as demyelination, or a generalized nonspecific reduction in overall slowing of cognitive function, which is likely present in a variety of chronic inflammatory diseases. ANAM cannot be used to determine impairment of specific domains of cognitive abilities and was not designed as a substitute for formal neuropsychological assessment. Future studies are required to determine its role in screening for cognitive impairment in SLE patients. Its value in the evaluation of changes in the cognitive performance of SLE patients over time may lie in its sensitivity, which was evident in the present study and previous studies, but this will require longitudinal assessment, ideally with concurrent neuropsychological and neuroimaging investigations.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Hanly had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Hanly, Farewell, Fisk.

Acquisition of data. Hanly, Omisade, Fisk.

Analysis and interpretation of data. Hanly, Omisade, Su, Farewell, Fisk.

Ancillary