Biomarkers.

Background: Cognitive disorders are clinical diagnoses informed by history, cognitive testing and investigations including MRI, FDG-PET and CSF biomarkers. These investigations do not always correlate however. This study sought to investigate the concordance between MRI, FDG-PET and CSF biomarkers with final diagnosis in a real-world clinical context.

Method: This is a prospective longitudinal cohort study (CogNID, NCT03861884, IRAS 250525) of a referral cognitive neurology NHS service (n=345). MRI and FDG-PET at assessment were appraised for whether they were suggestive of the final diagnosis. An interim analysis was performed on a quality assured dataset of those aged ≥40 and <70 years old.

Result: Cohort characteristics: 40≤age<70 (n=265), MRI results available (n=245), FDG-PET available (n=129), CSF analysis available (n=104). Interim results (n=77): Kappa coefficient of concordance of 0.15 (95% CI 0.03-0.28) between MRI and FDG-PET for suggestion of final diagnosis, demonstrating a poor level of agreement. MRI at time of assessment was suggestive of final diagnosis in 48% of cases and FDG-PET in 56%. The positive predictive value for a final diagnosis of Alzheimer's Disease based on CSF analysis in keeping with Alzheimer's continuum was 64%, with a negative predictive value of 75%.

Conclusion: Investigations must be viewed holistically in combination with a clinical history and cognitive examinations. Initial results at time of assessment are often not indicative of the final diagnosis, and may be discordant. Follow-up and repeat investigations where appropriate are critical for improving diagnostic accuracy. *CG&MW are joint first authors.