Neuropsychological and morphometric biomarkers of poor prognosis in patients with mild cognitive impairment

Abstract

Abstract. Despite a high prevalence of mild cognitive impairment (MCI), there are no accepted algorithms of diff erentiating the syndrome and the prognosis evaluation of later cognitive decline at this time. Objective. To identify biomarkers of poor prognosis in the various MCI types by optimizing neuropsychological examination in combination with MRI morphometry of brain structures. Patients and methods. We examined 45 patients (9 men, 36 women, mean age 72 ± 6.7 years) with MCI according to the modifi ed Petersen’s criteria and the DSM-5 criteria. All patients underwent the MMSE scale, the Detailed Neuropsychological Testing (DNT), which included a Ten Words Test (TWT), a “Double Test” (DT), a visual acuity test, a high-fi eld magnetic resonance imaging (MRI) of the brain with morphometry of cerebral structures (FreeSurfer, FSL). Results. According to the MMSE score, MCI were found in 26 (58%) patients. During the DNT, depending on the state of memory, 14 participants of the study identifi ed a non-amnestic type of MCI (na-MCI), 15 — an amnestic variant with impaired reproduction (ar-MCI), and 16 people — an amnestic type with a primary memory defect (apm-MCI). Volume changes of the anterior corpus callosum segment (CCA) were signifi cantly associated with the Immediate Recall after 4th reading and the Delayed Recall in the general MCI group (rho = 0.58; 0.58; p < 0.05) and the apmMCI group (rho = 0.6; 0.56; p < 0.05). Kruskal–Wallis Test showed that there were signifi cant group diff erences in the volumes of the CCA, right caudate nucleus, left cerebellar hemisphere cortex, posterior corpus callosum segment and left thalamus. At the same time, the fi rst three structures were combined into a set of informative features for differentiating the type of MCI based on the results of Forward stepwise Discriminant Analysis with a 77.3% accurate classifi cation rate (Wilks’s Lambda: 0.35962; approx. F (6.78) = 8.678, p < 0.001). ROC-analysis established the threshold values of the CCA volumes of ≤ 0.05% and the right caudate nucleus volumes of ≤ 0.23% (81.25% sensitivity in both cases; 62.1% and 60.7% specifi city; AUC 0.787 and 0.767; 95% CI 0.639–0.865 and 0.615–0.881; OR 7.1 and 6.7 (95% CI 1.6–30.6 and 1.6–29), associated with a memory defect in persons with MCI, while the ORs are 7.1 and 6.7 (95% CI 1.6–30.6 and 1.6–29), respectively. When both cerebral structures were included in the logit model, 88.6% classifi cation accuracy, 92.6% sensitivity, and 82.4% specifi city of the method were achieved. Conclusion. It has been demonstrated that classifying patients into the various types of MCI based on the data of memory function refl ected by the DNT and supplemented with MRI morphometry of the brain areas may be used as a sensitive and specifi c instrument for determining the category of patients with a high risk of Alzheimer’s disease. A neuropsychological profi le with a defect in primary memory, atrophic changes in anterior segment of the corpus callosum and the right caudate nucleus have been proposed as biomarkers of poor prognosis. Further longitudinal studies are necessary to clarify the proposed biomarkers of poor prognosis information and to detail the mechanisms of the neurodegenerative process.