Different cutoffs definitions to evaluate the concordance between CSF Ab42 and Ab42/Ab40 across neurodegenerative diseases

Abstract

AbstractBackgroundCerebrospinal fluid (CSF) Aβ1‐42 and the Aβ1‐42/Aβ1‐40 ratio are markers of amyloid pathology. We aimed to compare their agreement in different neurodegenerative disorders applying different cutoffs.MethodWe included 1564 participants from the Sant Pau Initiative on Neurodegeneration (SPIN), with AD, dementia with Lewy bodies (DLB), frontotemporal lobar degeneration‐related syndromes (FTLD), cognitively impaired but non‐neurodegenerative conditions (others), and cognitively normal individuals (CN). Cutoffs were defined as one‐sided 97.5%, 95% and 90% (Q97.5%, Q95% and Q90%, respectively) quantiles values in a middle‐aged cognitively normal population. These cutoffs for Aβ1‐42 (617 pg/mL, 637 pg/mL and 732 pg/mL, respectively) and Aβ1‐42/Aβ1‐40 (0.060, 0.070 and 0.082, respectively) were used to classify all participants as “positive” or “negative” according to each marker. We compared concordance of Aβ1‐42 and the Aβ1‐42/Aβ1‐40 ratio between groups through X2 test and correlations with Spearman test.ResultCorrelation between Aβ1‐42 and the Aβ1‐42/Aβ1‐40 ratio in CSF was strong (0.71, p<0.001) in the whole cohort, DLB and others groups (0.69 and 0.64, respectively, both p<0.0001); correlation was weak in AD, FTLD and CN groups (0.33, 0.52 and 0.51, respectively, all p<0.0001). Using Q97.5%, Q95% and Q90% cutoffs, Aβ1‐42 and ratio had an overall agreement of 76.6%, 78.3% and 82.9%, respectively in the whole sample. Both measures were normal (« negative ») in 46.1%, 41.6% and 35.8%, and both were abnormal (« positive ») in 30.5%, 36.7% and 47.1% of participants, respectively. Within each diagnostic category, the agreement ranged from 64% (AD) to 91.5% (CN) using Q97.5%, from 69.6% (AD) to 88.7% (CN) using Q95% and from 78.9% (DLB) to 85.9% (AD) using Q90%. Regardless of the cutoff definition, most discordant cases in AD and DLB were due to normal Aβ1‐42 and low ratio, while discordant cases in FTLD were more frequently due to low Aβ1‐42 and normal ratio.ConclusionAlthough Aβ1‐42 and Aβ1‐42/Aβ1‐40 ratio may reflect the same pathophysiological pathway, their agreement depends on cutoff definition. Our results confirm that both parameters show medium to high concordance, and that in case of discordance, Aβ1‐42 positivity with negative ratio is more associated to non‐AD dementias, independently of the cutoff used.