Category verbal fluency performance may be impaired in amnestic mild cognitive impairment

To study category verbal fluency (VF) for animals in patients with amnestic mild cognitive impairment (aMCI), mild Alzheimer disease (AD) and normal controls. Method Fifteen mild AD, 15 aMCI, and 15 normal control subjects were included. Diagnosis of AD was based on DSM-IV and NINCDS-ADRDA criteria, while aMCI was based on the criteria of the International Working Group on Mild Cognitive Impairment, using CDR 0.5 for aMCI and CDR 1 for mild AD. All subjects underwent testing of category VF for animals, lexical semantic function (Boston Naming-BNT, CAMCOG Similarities item), WAIS-R forward and backward digit span, Rey Auditory Verbal Learning (RAVLT), Mini-Mental Status Examination (MMSE), and other task relevant functions such as visual perception, attention, and mood state (with Cornell Scale for Depression in Dementia). Data analysis used ANOVA and a post-hoc Tukey test for intergroup comparisons, and Pearson’s coefficient for correlations of memory and FV tests with other task relevant functions (statistical significance level was p<0.05). Results aMCI patients had lower performance than controls on category VF for animals and on the backward digit span subtest of WAIS-R but higher scores compared with mild AD patients. Mild AD patients scored significantly worse than aMCI and controls across all tests. Conclusion aMCI patients may have poor performance in some non-memory tests, specifically category VF for animals in our study, where this could be attributable to the influence of working memory.


Resumo -
Mild cognitive impairment (MCI) is a clinical entity in patients with objective cognitive problems (most often episodic memory) but without impairment in daily life activities, 1 having a greater likelihood of transforming into dementia, most often Alzheimer disease (AD), than in the normal population. 2 MCI can be classifi ed according to the clinical presentation of symptoms into amnestic MCI (aMCI), multiple domain or single non-memory domain MCI. 1,2 Thus, by defi nition, aMCI presents with exclusive memory defi cit, sparing other cognitive domains such as language, visuospatial perception or executive functions. Nonetheless, aMCI individuals may present some nonmemory-related poor performance in specifi c neuropsychological tests, following a pattern similar to AD, 3 and continue to be classifi ed as amnestic rather than multiple domains MCI. This classifi cation is based on the clinical judgment that poor performance in one test is not enough to consider an entire cognitive domain as impaired.
Verbal fl uency (VF) for animal's names is a simple and widely used task that can reveal impairment in early phases of AD, 4 where a recent study points to impairment even in aMCI. 3 Category VF involves several cognitive aspects, such as semantic knowledge, executive function and working memory. Henry et al. suggested that verbal fl uency is "an excellent way of evaluating how subjects organize their thinking and ability to "organize output in terms of clusters of meaningfully related words". 5 Our aim was to compare verbal fl uency (category: animals) in healthy controls and patients diagnosed as aMCI and mild AD, hypothesizing that these two groups of patients have similar performance, because impairment of this function is common even in early stages of AD.

Methods
We studied 45 subjects, comprising 15 with aMCI and 15 with mild AD attended at the Unit for Neuropsychology and Neurolinguistics (UNICAMP Clinic Hospital), along with 15 controls. Routine laboratory examinations for dementia assessment (including B12 and folate dosage, sorology for syphilis, thyroid hormones) and brain computed tomography were carried out in all patients. The local ethics committee approved this research.
MCI in our clinic is a clinical diagnosis carried out by trained neurologists using a standardized mental status battery and was based on the following criteria of the International Working Group on Mild Cognitive Impairment: 1 (i) the person is neither normal nor demented; (ii) there is evidence of cognitive deterioration shown by either objectively measured decline over time and/or subjective report of decline by self and/or informant in conjunction with objective cognitive defi cits; and (iii) activities of daily living are preserved and complex instrumental functions are either intact or minimally impaired. We included only patients older than 50 years who had a CDR (Clinical Dementia Rating) 6 of 0.5. This classifi cation was performed by using a semi-structured interview.
For probable AD diagnosis, we used the criteria of the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and Alzheimer's Disease and Related Disorders Association (ADRDA) 7 , including only patients classifi ed as CDR 1. Exclusion criteria were history of other neurological or psychiatric diseases, head injury with loss of consciousness, use of sedative drugs within 24 hours of the neuropsychological assessment, drug or alcohol addiction and prior exposure to neurotoxic substances. The control group consisted of subjects with CDR 0 and no previous history of neurological or psychiatric disease, or memory complaints.
Neuropsychological evaluation comprised the following tests: 1) Verbal fl uency (VF) for animals' category (the score was the total number of different animal names given the by patient in one minute). 2) Mini Mental Status Examination (MMSE), 8 Brazilian version. 3) Episodic memory was evaluated using the Rey auditory verbal learning test (RAVLT). 9 4) Boston Naming Test (BNT-translated and culturally adapted version for Brazilian population by Dr. Cândida Camargo -Psychiatry Institute, Medicine School, University of São Paulo). 10 The BNT score was the sum of spontaneous correct responses plus correct responses following a semantic cue. 5) CAMCOG's subscale of similarities between pairs of nouns. 11 The patients were asked " In what way are they Data analysis by means of Systat software used ANOVA and a post-hoc Tukey tests for intergroup comparisons of demographic and cognitive scores, as well as Pearson coeffi cient for correlation between tests. Statistical signifi cance considered was p<0.05. Table 1. Demographics results of amnestic mild cognitive Impairment (AMCI), Alzheimer disease (AD), and normal control subjects.

Results
The results of demographic data are shown in Table  1 and neuropsychological evaluation in Table 2. aMCI subjects were similar to controls in age (p=0.576) and education (p=0.483). aMCI subjects performed similar to controls in CAMCOG's item of similarities (p=0.789) and Boston Naming Test (p=0.582) but performed worse than controls in verbal fl uency (p<0.001), MMSE (p=0.034), backward digit span (p<0.05), delayed recall (p<0.001) of RAVLT, CAMCOG's item of similarities (p=0.789) and Boston Naming Test (p=0.582).
AD patients were older than aMCI (p=0.012) but not control subjects (p=0.121). The educational level of the AD group was lower than that of controls (though not statistically signifi cant). These patients scored lower than controls and aMCI subjects on all tests, except the forward digit span. The cognitive performance of mild AD was worse than aMCI, which in turn was poorer than controls.
The analysis of relationships between tests in the groups showed statistically signifi cant correlations only between VF and RAVLT delayed recall in the AD group (r=0.545; p<0.05) and between VF and BNT in the aMCI group (r= 0.540; p<0.05). In AD group, FV tended to correlate to BNT, but not reaching statistical signifi cance (p=0.066). Scores on the Cornell Scale for Depression did not correlate to any of the cognitive tests: F (2,42)=0.929; p=0.403.

Discussion
Our fi ndings showed that aMCI patients performed worse than controls but better than mild AD on the category VF task. This task involves not only speed and ease of word production, but also lexical-semantic fi eld selection, executive function and working memory, in keeping track of what words have already been said. Some authors have found poor performance on category VF in MCI patients, and have interpreted this fi nding as a degradation of semantic networks. [15][16][17] We suggest that working memory and attention, rather than semantic or executive function deficits, may have infl uenced VF in our patients, since aMCI subjects had signifi cantly lower scores on backward digit span test yet normal performance in semantic and executive tasks (neither anamnesis nor objective cognitive tests used in our diagnostic process showed executive dysfunction in any patients classifi ed as aMCI). In fact, Perry et al. 18 have shown that defi cits in attention are more prevalent than defi cits in semantic memory in early AD. Similarly, our results on lexical semantic tests such as BNT and CAMCOG's similarities, showed no difference between aMCI and controls. Thus, our fi ndings suggest that semantic knowledge is not impaired and cannot explain the poor performance of this group of patients in category VF. AD patients' low VF was correlated to their impaired RAVLT delayed recall. A plausible explanation for this fi nding could be that our VF task partly depends on active retrieval (lexical-semantic selection) of animals' names from long-term declarative memory, also the case in the RAVLT delayed recall task. On the other hand, it is diffi cult to explain why VF was correlated to BNT in the aMCI group, since this group performed as well on the BNT as did controls. Nevertheless, the fact that FV was correlated to BNT in the aMCI group and also tended to correlate in the AD group, suggests that both groups may have impairment of some linguistic competence involved in lexical-semantic selection, although this was not specifi cally tested in our study.
We have found that aMCI patients may have poor performance in some non-memory tests, specifi cally category VF for animals, and that this could be attributable to the infl uence of working memory. However, further studies using more comprehensive testing of VF, including a phonemic task, as well as more specifi c tests for executive function and lexical-semantic selection in a larger sample are needed for more robust conclusions to be drawn.