Perception of Fragmented Letters by Patients With Pathologically Confirmed Dementia With Lewy Bodies or Alzheimer Disease.

Abstract

Patients with dementia with Lewy bodies perform worse than those with Alzheimer disease (AD) on tests of visual perception, but the clinical utility of these tests remains unknown because studies often had clinically diagnosed groups that may inadvertently cross-contaminate Lewy body disease (LBD) with pure AD pathology, used experimental tests not easily adaptable for clinical use, and had no way to examine relationships between the severity of LBD pathology and degree of cognitive impairment. Therefore, we sought to determine whether performance on a widely used clinical test of visuoperceptual ability effectively differentiates between patients with autopsy-confirmed LBD or AD and correlates with the severity of LBD pathology. Patients with mild to moderate dementia (n = 42) and cognitively healthy controls (n = 22) performed a Fragmented Letters Test in which they identified letters of the alphabet that were randomly visually degraded by 70% and additional visuospatial and episodic memory tests. At autopsy, dementia cases were confirmed to have LBD (n = 19), all with concomitant AD, or only AD (n = 23). Severity of α-synuclein pathology in the hippocampus and neocortex was rated on an ordinal scale. Patients with LBD performed worse than those with AD (B = -2.80 ± 0.91, p = 0.009) and healthy controls (B = -3.34 ± 1.09, p = 0.01) on the Fragmented Letters Test after adjustment for age, sex, education, Mini-Mental State Examination score, and ability to name intact letters. Patients with AD did not differ from controls (B = -0.55 ± 1.08, p = 0.87). The test effectively distinguished between patients with LBD or AD with 73% sensitivity and 87% specificity, and the area under the curve in receiver operating characteristic analyses was 0.85 (95% CI 0.72-0.95), higher than for standard tests of visuospatial ability (Block Design; 0.72; CI 0.35-0.75) or memory (California Verbal Learning Test, trials 1-5; 0.55; CI 0.57-0.88). Fragmented Letters Test scores were negatively correlated with LBD pathology density ratings in hippocampus and neocortical regions (Spearman rs = -0.53 to -0.69). Fragmented Letters Test performance can effectively differentiate patients with LBD pathology from those with only AD pathology at a mild to moderate stage of dementia, even when LBD occurs with significant concomitant AD pathology, and may also be useful for gauging the severity of cortical α-synuclein pathology in those with LBD.