| Objective:Late-onset depression(LOD)is defined as major depressive disorder with first onset at the age of 55 years or older,which presents with more severe cognitive and somatization symptoms in addition to the typical manifestations of depressed mood.According to research,LOD can increase the risk of developing Alzheimer’s disease(AD)by nearly 50%.Amnestic Mild Cognitive Impairment(a MCI)is the most common subtype of AD in the preclinical stage of the disease.a MCI has mild cognitive impairment,mainly in the form of memory function and attention decline.In addition,approximately 35-85% of patients with a MCI have neuropsychiatric symptoms.LOD and a MCI are high-risk groups for AD,and the two have been found to share many common risk factors for AD,predicting that LOD and a MCI may be a potential clinical unifier.However,there are fewer studies on the mechanisms underlying the differences in cognitive impairment patterns between the two in general,and in neurophysiology specifically.In this study,a case-control study was conducted to explore the differences in cognitive impairment and neurophysiological mechanisms between LOD and a MCI patients based on ERP/EEG technology,and to try to elucidate the relationship between the two,so as to provide a scientific basis for early clinical identification and treatment of cognitive impairment in the elderly.Methods:A total of 65 subjects were included in this study based on inclusion and exclusion criteria,including 14 patients with LOD,34 patients with a MCI,and 17 healthy controls(HC).The Hamilton Depression Scale 24(HAMD-24)and the Geriatric Depression Scale(GDS)were used to assess the depressive mood of the patients.The Hamilton Anxiety Rating Scale(HAMA)is used to assess patients’ anxiety;the Mini-mental State Examination(MMSE)and the Montreal Cognitive Assessment(Mo CA)were used to assess patients’ cognitive function;the Pittsburgh Sleep Quality Index Scale(PSQI)was used to assess patients’ sleep.A multidimensional cognitive assessment tool was also used to assess subjects’ cognitive functions(including: Clock Drawing Test;Verbal Fluency;Connecting the Dots Test;Digit Breadth Test;Digit Symbol Substitution Test;Hopkins Word Learning);Stroop Test;Line Orientation Judgment Test;Silhouette Test;Complex Eye Area Emotion Recognition Task;Boston Naming Test;Logical Memory Test),and all neuropsychological assessments were integrated into 5 cognitive domains(memory,information processing speed,visuospatial function,executive function,and verbal learning ability),and raw scores were converted to the corresponding composite Z-scores to assess the function of each cognitive domain.One-way analysis of variance(ANOVA)or chi-square test was used to statistically analyze age,years of education,MMSE,Mo CA,GDS,PSQI,Z-values of each cognitive domain and Z-values of each subtest for the three groups of subjects,and if the statistical results suggested significant differences between the three groups,Bonferroni’s post hoc test was used to analyze differences between the two pairs.Comparisons of categorical variables(e.g.,gender)were analyzed using chi-square tests.All subjects were subjected to EEG/ERP data collection,where EEG collection consisted of resting data with eyes open and eyes closed for 5 minutes,and ERP experimental paradigm used the classical n-back paradigm(0-back,1-back,2-back),during which the subjects’ correct rates and reaction times in each test were collected.All EEG and ERP data were preprocessed by MATLAB software and the EEGLAB toolkit.EEGLAB was applied to compare the differences between the three groups in different frequency bands(δ: 1-4 Hz;θ: 4-8 Hz;α:8-13 Hz;β: 13-30 Hz;γ: 30-45 Hz)and total energy after preprocessing EEG data with FDR correction and topographic maps,and statistical differences were considered at P < 0.05,and the three groups with significant differences in frequency bands were analyzed two-by-two to compare the differences.The energy of the different frequency bands of each channel of the EEG was extracted using MATLAB software and the average energy of each frequency band of the whole brain was calculated and collated to SPSS 26.0 for statistical analysis,using one-way ANOVA for statistics if it satisfied the normal distribution,otherwise nonparametric tests were used,statistically significant differences were considered at P<0.05.Manual peak detection was performed using MATLAB for all n-back conditions(0-back,1-back,and 2-back)and for each mean waveform for each participant to ensure that the correct peak is recorded.The P200 peak is the most positive peak at 100-220 ms intervals after stimulation,the P300 peak is the most positive peak at 350-500 ms intervals after stimulation,and the time corresponding to the maximum peak is its respective latency.The data were collated and extracted to SPSS26.0 for statistical analysis in the same way as EEG.Also,the correct rates and response time recorded by E-prime3.0 software for each subject in different n-back conditions were collated to SPSS26.0 software for statistical test analysis.Finally,the correlation analysis was performed between the differential band energy and ERP components of the LOD and a MCI groups and the MMSE,Mo CA,GDS,PSQI,Z-values of each cognitive domain and Z-values of each subtest.Results:1.HAMD(P<0.001),HAMA(P<0.001),GDS(P<0.001)scores and PSQI(P<0.001)were significantly higher in the LOD group than in the a MCI group.There were significant differences in MMSE(P<0.001)and Mo CA(P<0.001)between the three groups,and MMSE and Mo CA scores were significantly lower in the LOD and a MCI groups than in the HC group(P<0.05),while there was no statistical difference between LOD and a MCI(P>0.05).Significant differences were found between the three groups of subjects in four cognitive domains of information processing speed(P=0.001),visuospatial function(P<0.001),language learning ability(P<0.001)and memory(P<0.001)and total cognitive function Z-score(P<0.001),and no significant differences were seen in executive function(P=0.155).In terms of information processing speed,no significant differences were seen between the LOD and a MCI groups(P=0.063),both of which were significantly lower than the HC group(P<0.05).In terms of visuospatial function,the LOD group was significantly lower than the HC group(P<0.001)and the a MCI group(P=0.008),and no statistical differences were seen between the HC and a MCI groups(P=0.130).In terms of language learning ability,the LOD group was significantly lower than the HC group(P<0.001)and the a MCI group(P=0.004),and no significant difference was seen between the a MCI and HC groups(P=0.061).In memory function,both the LOD group(P=0.005)and the a MCI group(P<0.001)were significantly lower than the HC group,and no significant difference was seen between the two groups(P=1.000).In terms of total cognitive function,the LOD group(P<0.001)and the MCI group(P<0.001)were both significantly lower than the HC group,and the LOD group was significantly lower than the a MCI group(P=0.046).2.The subjects in the three groups had statistically different energy values in the β-band(P=0.001)and γ-band(P=0.023),but no statistically significant differences were seen in the remaining bands(P>0.05).The energy in the γ-band was significantly higher in the LOD group than in the HC group(P=0.020),and no significant differences were seen in the a MCI group compared with the HC(P=0.494)and LOD groups(P=0.156).The energy in the β-band of the LOD group was significantly higher than that of the HC group(P=0.001)and the a MCI group(P=0.016),while no statistically significant difference was found between the a MCI and HC groups(P=0.227).3.Three groups of subjects were statistically different in 0-back reaction time(P=0.033)and 1-back correct rate(P=0.005).1-back correct rate was significantly lower in the LOD than in the HC group(P=0.001),while there was no significant difference between the HC and a MCI groups(P=0.061)and the LOD and a MCI groups(P=0.079).The LOD group had a significantly longer reaction time in the 0-back group(P=0.010),while there was no significant difference between the HC and a MCI groups(P=0.279)and the LOD and a MCI groups(P=0.068).P200 latency was not statistically different between the three groups in the 0-back and 1-back tasks(P>0.05).And in the 2-back task the P200 latency was significantly longer in both the LOD and a MCI groups compared to the HC group(P<0.05),and there was no significant difference between the LOD and a MCI groups(P>0.05).There was a statistical difference(P<0.05);P300 amplitude was not statistically different in 0-back,1-back and 2-back tasks(P>0.05).In the 0-back task,the P300 latencies in the Fz and Cz channels were significantly prolonged in the LOD and a MCI groups compared with the HC group(P<0.05),and there was no significant difference between the LOD and a MCI groups(P>0.05).In the 2-back task,the P300 latency in the Fz and Cz channels was significantly prolonged in both the LOD and a MCI groups compared with the HC group(P<0.05),and there was no significant difference between the LOD and a MCI groups(P>0.05).4.In the LOD group,Mo CA(r=0.763,P=0.001),memory Z-value(r=0.558,P=0.048),and executive function Z-value(r=0.646,P=0.017)were significantly and positively correlated with years of education,and MMSE(r=-0.655,P=0.015)and executive function Z-value(r=-0.581,P=0.048)were significantly and negatively correlated with PSQI.In a MCI,memory Z-value was significantly and negatively correlated with PSQI(r=-0.462,P=0.023),memory Z-value was significantly and negatively correlated with PSQI(r=-0.462,P=0.023),memory Z-value(r=-0.468,P=0.005),information processing speed Z-value(r=-0.484,P=0.004),executive function Z-value(r=-0.461,P=0.006)and total cognitive function Z-value(r=-0.565,P<0.001)were significantly and negatively correlated with age.In the 0-back task,the P300 latency of the LOD group was significantly negatively correlated with its memory function and total cognitive function Z values(Z memory: Fz: r=-0.793,P=0.006,Cz: r=-0.637,P=0.048;Z total cognitive function: Fz: r=-0.641,P=0.034,Cz: r=-0.666,P= 0.025);P300 latency in the Fz channel of the a MCI group was significantly and positively correlated with PSQI(r=0.581,P=0.015)and with Mo CA(r=-0.445,P=0.028),information processing speed(r=-0.531,P=0.010),memory(r=-0.447,P=0.028),and total cognitive function Z values(r=-0.458,P=0.049)were significantly and negatively correlated.The P300 latency in the LOD group in the 2-back task was significantly and positively correlated with HAMA(Fz: r=0.872,P=0.024;Cz: r=0.888,P=0.018).the P200 latency in the LOD group in the 2-back task was significantly and negatively correlated with the Zvalue of information processing speed(Fz: r=-0.731,P=0.040;Cz: r=-0.802,P=0.017).Conclusion:In this study,we analyzed the cognitive function,EEG and ERP of LOD and a MCI,and found that the pattern and degree of cognitive impairment of LOD and a MCI were not identical,and there was a more extensive cognitive decline,among which visuospatial function and language learning ability were probably the more characteristic cognitive impairment domains of LOD.However,the pattern of neurophysiological changes in MCI and LOD were basically the same compared to HC.The present study suggests that the cognitive symptoms of LOD may exist independently of the emotional symptoms and that the combined effect of the two causes the characteristic neurophysiological changes in LOD.We consider that the cognitive impairment in LOD patients may be related to neurodegenerative changes as in a MCI,and that depressive symptoms may be neuropsychiatric symptoms that emerge during the progression of neurodegeneration,further increasing the risk of conversion to AD. |
