Supplementary Material for: BASELINE PREDICTORS OFLONGITUDINAL COGNITIVE OUTCOMES IN PERSONS WITH MILD COGNITIVE IMPAIRMENT (MCI)

Introduction To explore longitudinal cognitive outcomes and to ascertain predictors of conversion to dementia in a hospital-based MCI cohort classified according to the neuropsychological phenotype at baseline Materials & Methods Subjects aged >55 years who had a clinical diagnosis of MCI at initial visit between 2010-2018, with at least one formal neuropsychological assessment at baseline and follow up of a minimum of 2 years were included. The prospective study was completed based on evaluation at last follow up to gauge conversion to dementia, quantification of performance on activities of daily living and where available longitudinal neuropsychological test scores. Results Ninety-five patients with MCI met the inclusion criteria with a mean age of 68.4 ±6.4yrs at baseline and a mean duration of follow-up for 6.4 ±3.2 yrs. The cumulative conversion rate to dementia was 22.2% (21/95) and the annualized conversion rate (ACR) was 3.3 % per year of follow up. The majority of subjects who had converted had multidomain MCI (66%). Only white-matter changes on MRI brain revealed correlation with baseline neuropsychology tests. The multivariate logistic regression analysis revealed the utility of lower baseline list recognition (adjusted odds ratio-0.735[95% confidence interval-0.589-0.916]; p-0.006), lower immediate logical memory (0.885[0.790-0.990]; p-0.03) and high perseverative error scores on set-shifting (3.116[1.425-6.817]; p-0.004) as predictors of conversion. A model score of +2.615 could predict conversion with sensitivity of 72% and specificity of 98% over 6.4 years follow up. Conclusion There was a higher risk of conversion associated with multidomain MCI. Logistic regression-based estimations of dementia risk utilizing domain-based neuropsychology test scores in MCI have high specificity for diagnosis at baseline.