Background
Abnormalities in cerebrospinal fluid (CSF)-amyloid-beta (Aβ)42, CSF-Aβ40, CSF-Aβ38, CSF-soluble amyloid precursor proteins α and β, CSF-total-tau, CSF-phosphorylated-tau, CSF-neurofilament light protein (NF-L), CSF-neurogranin, plasma-Aβ42, plasma-Aβ40, plasma-total-tau, plasma-NF-L and, serum-S100B during affective episodes may reflect brain changes that could impact cognitive function in patients with bipolar disorder (BD). The study aimed to investigate the association between these biomarkers indicative of Alzheimer’s disease and those reflecting neurodegeneration alongside their impact on cognitive function in patients with BD and healthy control individuals (HC). The primary hypothesis was that GL and VL would increase with increasing levels of CSF-Aβ42 based on data from T0 and T3 in BD and HC jointly.

Methods
In a prospective, longitudinal case–control study euthymic patients with BD (N = 85) and HC (N = 44) were evaluated with clinical assessment and neuropsychological testing at baseline (T0) and during euthymia after a year (T3). Patients’ affective states were recorded weekly as euthymic, subthreshold level, major depression, or (hypo)mania. If an episode occurred during follow-up, the patient was also assessed in post-episode euthymia. Cognitive performance was measured as a global cognitive score (GL) for four cognitive domains including verbal learning and memory (VL).

Results
Estimated in a linear mixed model GL increased with 0.001 for each increase of 1 pg/ml of CSF-Aβ42 (97.5%, CI 0.00043–0.0018, adjusted-p = 0.0005) while VL increased by 0.00089 (97.5%, CI 0.00015–0.0018, adjusted-p = 0.045) in BD and HC jointly. The association was weak, however stronger in patients with BD compared to HC. Associations between other biomarkers including CSF-neurogranin, and cognitive domains were overall weak, and none remained significant after adjustment for multiple testing.

Limitations
Modest sample size. A complete data set regarding both CSF-AB-42 and cognitive test scores was obtained from merely 61 patients with BD and 38 HC individuals.

Conclusion
CSF-Aβ42 may be associated with cognitive dysfunction in patients with BD and HC individuals. The association appeared to be stronger in BD but with overlapping confidence intervals. Hence it remains uncertain whether the association is a general phenomenon or driven by BD.