Antemortem Memory Impairment Screen Performance Is Correlated With Postmortem Alzheimer Pathology
Year: 2003
Type: article
Abstract: To the Editor: The Mini-Mental-State-Examination (MMSE) and the Memory Impairment Screen (MIS) are recommended as screening tests for dementia.1 Both tests have high reliability and validity in screening for dementia. Possible advantages of the MIS, a four-word free and cued recall test, over the MMSE include its brevity and ease of use.2 The biological validity of most screening tests, including the MIS, is not established. We examined the relationship between the MIS and Alzheimer's disease pathology in the prospective autopsy series of the Einstein Aging Study (EAS). EAS participants were randomly recruited from population lists of Medicare-eligible recipients (aged ≥70). Subjects received clinical and neuropsychological evaluations at enrollment and 12- to 18-month follow-up visits. The Blessed-Information-Memory-Concentration test (BIMC) was used to assess mental status. Dementia was diagnosed at consensus case conferences, attended by study neurologists, neuropsychologist, and social worker using Diagnostic and Statistical Manual of Mental Disorder, Revised, Fourth Edition criteria The MIS has been administered in the EAS since 1996.2 A score of less than 5 (range 0–8) is optimal for detecting dementia.2 Informed consent was obtained at entry, and subjects were asked to consider eventual participation in an autopsy program. Autopsies were performed within 24 hours of death, blinded to clinical status. Senile plaques and neurofibrillary tangles (NFT) were counted in multiple cortical, hippocampal, and subcortical regions with thioflavin-S fluorescent microscopy and immunostaining for tau antibodies. A Braak stage was assigned, ranging from I (minimal hippocampal pathology) to VI (maximal pathology involving association cortices).3 Alzheimer's disease was diagnosed using National Institute of Aging—Reagan Institute criteria.4 Neocortical amyloid deposits without neuritic pathology was termed pathological aging. MIS is recommended for screening for any dementia. Hence, we examined Spearman correlations of final MIS scores with Alzheimer markers in 21 cases who were administered the MIS, from 29 EAS autopsies done since 1996. Because the MIS has better validity in detecting Alzheimer's disease,2 15 cases representing a spectrum of severity of Alzheimer pathology, from nondemented subjects with no or minimal Alzheimer pathology (n = 6) to pathological aging (n = 4) and Alzheimer's disease (n = 5), were examined. Braak stage comparisons were made with the Kruskal-Wallis test. The mean age ± standard deviation at death was 87.1 ± 8.1. There were 15 women (71.4%). Median interval from MIS administration to death was 15 months (range 1–52). Seven of nine subjects meeting criteria for dementia at their last evaluation had abnormal MIS scores (sensitivity 78%, specificity 92%). Two other subjects were determined to have dementia after caregiver interviews were conducted after death. Pathological diagnoses in these subjects included Alzheimer's disease (n = 5), vascular dementia (n = 4), dementia with Lewy bodies (n = 1), and frontotemporal dementia (n = 1). Antemortem MIS scores had a sensitivity of 73%, specificity of 90%, and positive predictive value of 89% for predicting Alzheimer pathology. In comparison, we found that the MIS had a sensitivity of 87%, specificity of 96%, and positive predictive value of 85% to detect Alzheimer's disease (base rate 20%) using clinical diagnosis as the criterion standard.2 Eleven subjects had Braak stage I–II (median interval from MIS to death, 14.0 months), seven had stage III–IV (32.5 months), and three had stage V–VI (3.0 months). Higher Braak stages were associated with worse performance on BIMC (H = 4.99, P = .08) and MIS (H = 10.03, P = .007). Final mean MIS scores were 6.8 ± 2.3 for subjects in Braak stage 0–II, 5 ± 2.8 in stage III–IV, and 0 ± 0 in stage V–VI. Final mean BIMC scores were 6.12 ± 11, 12.4 ± 11.2, and 28 ± 5.6, respectively. Final MIS scores correlated with Braak stage (Table 1). MIS correlated better with NFT than senile plaque. The MIS and Braak staging correlations remained significant (Spearman r = −0.557, P = .03) after partial correlations were made to adjust for the influence of senile plaques. As predicted, the MIS was better correlated with pathological markers in the 15 cases representing a spectrum of Alzheimer pathology. Because the MIS is a memory screen, it is not surprising that it has better accuracy for clinically diagnosing Alzheimer's disease than other dementia subtypes.2 This study shows that antemortem MIS performance is also highly correlated with Alzheimer pathology. We previously reported that free and cued recall test performance declines at earlier Braak stages than BIMC.5 Early Braak stages may not have a behavioral correlate or have mild behavioral changes, which the MIS may detect.5 The recent introduction of MIS and low autopsy rate, similar to other community studies, explains our small sample. In contrast to clinical or institutional autopsy series, our sample included a significant proportion of dementia-free subjects. The long interval between MIS administration and death in some cases (Braak stages III–IV) should bias towards the null. The MIS and Braak stage correlation is improved (Spearman r = −0.835, P = .001) if the analysis is restricted to 11 cases with MIS done within a year of death. Establishing biological validity is necessary to validate screening tests. The present study corroborates previous results2 and supports the discriminative validity of MIS.
Language: English
Subfield: Psychiatry and Mental health
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2064894388