Clinical validity of a borderline group in CSF biomarker classification in Alzheimer’s disease

Abstract Background Efforts to translate the NIA‐AA AT(N) dichotomic research framework of Alzheimer’s disease (AD) into clinical practice should evaluate how to deal with borderline cases. We characterise a peri‐threshold group of individuals regarding Aß classification, using the Aß 42 /Aß 40 ratio in cerebrospinal fluid (CSF). Method We conducted a retrospective study of consecutive patients referred to a university hospital memory clinic for diagnostic evaluation of cognitive impairment. We identified individuals (n = 31) with a Aß 42 /Aß 40 ratio within a prior defined borderline zone. Two additional groups, matched for clinical severity, age and sex, were used for comparisons: a low Aß 42 (normal Aß 42 /Aß 40 ratio >0,5; pathological Aß 42, n = 23), and a typical AD reference group (A+T+(N+)), n = 50). We compared demographic, clinical, neuropsychological and neuroimaging characteristics between the groups. Results Included patients had a median age of 72 years and presented either with mild cognitive impairment (CDR 0.5) or mild dementia (CDR 1). Groups did not differ in demographic and clinical characteristics (education, disease duration, prevalence of vascular risk factors). In comparison to the typical AD reference group, patients in the borderline group were overall similar, except for better mean performance on the MMSE (26 vs. 21, p<0.05), verbal delayed recall (mean z‐score ‐1.9±1.6 vs. ‐3.1±1.9, p<0.05) and verbal recognition (mean z‐score ‐1.65±1.1 vs. ‐2.5±1.8, p< 0.05). They also showed a less severe degree of parietal atrophy (Koedam rating score, p<0.05). Compared with the individuals in the low Aß 42 group, the borderline group only had a better performance in semantic fluency (mean z‐score ‐1.3±1.9 vs. ‐2.3, p<0.05), verbal learning (mean z‐score ‐2.0±2.2 vs. ‐2.93±2.6, p<0.05) and less severe score in the medial temporal lobe atrophy rating scale (median score 1 vs. 2, p<0.001). The presence of abnormal phosho‐Tau levels did not mediate the observed differences. Conclusion These findings suggest that patients in the borderline group are at least partly similar to the typical AD group, as well as the low Aß 42 group. Further information from longitudinal data to assess risk progression and cross‐validation in other cohorts is needed to better translation of the AT(N) classification and assist clinical decision regarding borderline cases.