Objective:To explore the RNA expression and alterations in brain structure in individuals who have experienced stressful life events (SLE), as well as the correlation patterns between them and their association with the occurrence of depression.Methods:Prospectively, a total of 80 SLE subjects were recruited from the psychiatry and psychology clinic of the Jiangsu University Affiliated Yixing Hospital between January 2021 and December 2022, with 16 normal controls (NC) enrolled concurrently. The 17 items Hamilton depression scale (HAMD-17) and social readjustment rating scale (SRRS) were used to assess depressive symptoms and stress levels. RNA sequencing information of peripheral blood and imaging data at baseline were collected. Based on whether depression occurred during the 2-year follow-up period, SLE subjects were divided into the SLE-depression group ( n=15) and the SLE-non-depression group ( n=65). Differentially expressed genes (DEGs) were screened using differential analysis and protein-protein interaction (PPI) networks. Fractional anisotropy (FA) of white matter tracts and gray matter volume (GMV) were extracted using tract-based spatial statistics and voxel-based morphometry.Using analysis of variance compared inter-group differences in gene expression, GMV and white matter FA values. Partial correlation analysis was used to explore correlations between DEGs, altered GMV and white matter microstructure. Gene set enrichment analysis (GSEA) was performed on key genes to identify potential biological pathways. Propensity score matching constructed sensitivity subgroups to verify result robustness. Results:The SLE-depression group showed significantly higher SRRS and HAMD-17 scores at baseline and at the end of follow-up compared to the SLE-non-depression group and the NC group ( H=47.773, 35.427, 41.114, all P<0.05). Expression levels of IL-10 (2.12±0.28, 2.43±0.44), EZH2 (2.11±0.43, 2.45±0.51), NCAM1 (3.60±0.30, 3.03±0.39), CD3E (4.95±0.37, 4.57±0.48), CCK (3.29±0.28, 3.02±0.42), and CX3CR1 (5.55±0.40, 5.91±0.34) were significantly different between the SLE-depression group and SLE-non-depression group( F=5.549~28.371, all P<0.05). Compared with the SLE-non-depression group, the SLE-depression group exhibited significantly lower FA values in the genu of the corpus callosum (0.29±0.04, 0.31±0.04) and the left uncinate fasciculus (0.31±0.02, 0.33±0.02), as well as significantly smaller GMV in the right hippocampus (0.29±0.07, 0.33±0.06), bilateral middle frontal gyrus (left: 0.27±0.05, 0.31±0.05; right: 0.28±0.06, 0.32±0.06), right insula (0.36±0.03, 0.38±0.04), and left precentral gyrus (0.19±0.04, 0.24±0.05) ( F=4.593-12.064, all P<0.05, FDR correction). GMV in the right anterior cingulate and paracingulate gyri was significantly larger than that in the SLE-non-depression group (0.34±0.05, 0.29±0.06) ( F=6.704, P=0.034, FDR correction). Partial correlation analysis revealed significantly stronger correlations between hub DEGs and altered brain regions in the SLE-depression group ( r=0.017-0.801) compared to the SLE-non-depression group ( r=0.002-0.382), with a statistically significant difference ( U=629, P<0.001; Cliff's Delta=0.454). GSEA indicated that the aforementioned genes were primarily involved in pathways including the ribosome, spliceosome, ribosome biogenesis in eukaryotes, and neuroactive ligand-receptor interaction. Sensitivity analysis confirmed that the above results remained statistically significant after balancing sample sizes (all P<0.05). Conclusion:The SLE-depression group showed specific RNA expression and brain structure alterations compared to the SLE-non-depression group, and the correlation between RNA and brain structure was significantly enhanced in the SLE-depression group. This suggests that the correlation between genes and brain structure in the SLE population may be related to their susceptibility to depression.

Objective:To investigate the predictive value of multimodal magnetic resonance imaging (MRI) techniques in assessing symptom remission of post-traumatic stress disorder (PTSD) of bereaved parents who lost their only child.Methods:In this prospective study, 34 parents with PTSD resulting from the loss of the only child were followed-up for 2 years. Based on the PTSD diagnostic status at the end of the follow-up, participants were divided into the remission group and the persistent group.R 3.6.1 and SPSS 20.0 software were used for statistical analysis.Baseline clinical data and neuroimaging findings were compared between the two groups. Logistic regression and LASSO regression analyses were used to identify independent predictors of PTSD symptom remission. The predictive performance of these factors was evaluated by receiver operating characteristic (ROC) curve analysis.Results:Initial screening with univariate Logistic regression and LASSO regression revealed that regional homogeneity (ReHo) in the left middle temporal gyrus, the combined predictive value based on ReHo, and the integrated predictive value combining gray matter volume (GMV) and ReHo (GMV-ReHo predictor) were significant factors influencing symptom remission (all P<0.05). Multivariate Logistic regression further demonstrated that the GMV-ReHo predictor retained independent predictive significance ( P<0.05), with ROC curve analysis showing an area under the curve (AUC) of 0.979 (95% CI=0.935-0.996, P<0.001) for its ability to predict PTSD remission. Notably, a combined model incorporating both the scores of the clinician administered PTSD scale (CAPS) and the GMV-ReHo predictor achieved an enhanced predictive performance, yielding an AUC of 0.984 (95% CI=0.952-0.998, P<0.001). Conclusion:The GMV-ReHo predictor effectively identifies symptom remission in PTSD resulting from the loss of the only child.

Objective:To investigate the differences in hippocampal subfield volumes and structural covariance network between participants with subjective cognitive decline (SCD) and healthy individuals, and to analyze the correlations of the volumes of the different subfields and altered covariance brain regions with memory function.Methods:A total of 57 SCD individuals(SCD group) and 44 normal controls(NC group) participants were assessed for memory function using composite scores from the auditory verbal learning test (AVLT) and the Wechsler memory scale visual reproduction (VR) test from June 2022 to October 2023.T1-weighted structural magnetic resonance imaging (MRI) data were collected from all participants, and hippocampal subfields, cortical regions, and subcortical nuclei were segmented using FreeSurfer to measure the gray matter volume of each structure. A structural covariance network was constructed based on the correlation of gray matter volumes across regions. Statistical analysis was performed using R 4.3.1 software. Inter-group differences in hippocampal subfield volumes were compared using multivariate analysis of covariance. Differences in structural covariance connectivity between groups were assessed using Z-test, while network topology differences were compared through permutation testing. Finally, partial correlation analysis was used to examine correlation of the volumes of the differential hippocampal subfields and covariance brain regions with memory function. Results:The SCD group exhibited significantly lower years of education, AVLT-immediate score, AVLT-delayed score, VR-immediate score, VR-delayed score, and memory function Z-score compared to the NC group ( t=2.064, 3.888, 2.622, 3.222, 4.761, 5.184, all P<0.05). The volumes of the right subiculum((387.75±55.20)mm 3, (352.70±70.25)mm 3), left presubiculum((263.12±38.52)mm 3, (239.79±46.02)mm 3), left subiculum((388.12±49.34)mm 3, (351.74±67.30)mm 3) and left CA1((571.01±80.01)mm 3, (526.51±98.80)mm 3) in the SCD group were smaller than the corresponding volumes in NC group ( F=9.139, 8.039, 11.207, 7.266, all P<0.05, FDR correction). Differences in structural covariance connectivity were found between the SCD and NC groups in the following pairs: right CA1-right subiculum, right CA1-left subiculum, right CA3-left parasubiculum and right hippocampus-amygdala transition area-left subiculum ( Z=-3.848, -3.896, -3.597, -3.895, all P<0.05, FDR correction).Partial correlation analysis revealed that in the SCD group, the volume of the left subiculum ( r=0.359, P=0.007), left CA1 ( r=0.430, P=0.001), right entorhinal cortex ( r=0.296, P=0.029), right middle temporal gyrus ( r=0.361, P=0.007), right parahippocampal gyrus ( r=0.313, P=0.021)were positively correlated with the total memory function score. Conclusion:Hippocampal subfields atrophy, as well as alterations in structural covariance network, have been found in SCD individuals. Furthermore, the decline in memory function may be closely associated with atrophy in hippocampal subfields and structurally covariant regions.

Objective To investigate the impact of intergenerational caregiving on the brain structure and function of grandparents,and to analyze its correlation with caregiving factors.Methods Healthy adults(66 with grandchildren,24 without grandchildren)were recruited as study subjects,and clinical and MRI data were collected.Resting-state brain functional degree centrality(DC)and surface-based morphometry(SBM)methods were used to compare the differences in brain structure and function between the groups with and without grandchildren.The correlation between the differences in brain regions and △ values with grandchild's age,number,and time spent in childcare were assessed,respectively.Results Compared to the group without grandchildren,the group with grandchildren showed reduced surface area and cortical volume in the left middle temporal gyrus,as well as decreased DC values in the left medial superior frontal gyrus,bilateral orbital superior frontal gyrus,and left anterior cingulate and paracingulate gyrus(P＜0.05),respectively.In the grandchildren group,DC values and △ values in the left orbital superior frontal gyrus,left anterior cingulate and paracingulate gyrus were significantly positively correlated with time spent in childcare.Conclusion The brain structures and functions of grandparents related to empathy and motivation are changed in intergenerational caregiving,which may reveal the neuroplasticity after caring for their grandchildren.

Objective:To explore the RNA expression and alterations in brain structure in individuals who have experienced stressful life events (SLE), as well as the correlation patterns between them and their association with the occurrence of depression.Methods:Prospectively, a total of 80 SLE subjects were recruited from the psychiatry and psychology clinic of the Jiangsu University Affiliated Yixing Hospital between January 2021 and December 2022, with 16 normal controls (NC) enrolled concurrently. The 17 items Hamilton depression scale (HAMD-17) and social readjustment rating scale (SRRS) were used to assess depressive symptoms and stress levels. RNA sequencing information of peripheral blood and imaging data at baseline were collected. Based on whether depression occurred during the 2-year follow-up period, SLE subjects were divided into the SLE-depression group ( n=15) and the SLE-non-depression group ( n=65). Differentially expressed genes (DEGs) were screened using differential analysis and protein-protein interaction (PPI) networks. Fractional anisotropy (FA) of white matter tracts and gray matter volume (GMV) were extracted using tract-based spatial statistics and voxel-based morphometry.Using analysis of variance compared inter-group differences in gene expression, GMV and white matter FA values. Partial correlation analysis was used to explore correlations between DEGs, altered GMV and white matter microstructure. Gene set enrichment analysis (GSEA) was performed on key genes to identify potential biological pathways. Propensity score matching constructed sensitivity subgroups to verify result robustness. Results:The SLE-depression group showed significantly higher SRRS and HAMD-17 scores at baseline and at the end of follow-up compared to the SLE-non-depression group and the NC group ( H=47.773, 35.427, 41.114, all P<0.05). Expression levels of IL-10 (2.12±0.28, 2.43±0.44), EZH2 (2.11±0.43, 2.45±0.51), NCAM1 (3.60±0.30, 3.03±0.39), CD3E (4.95±0.37, 4.57±0.48), CCK (3.29±0.28, 3.02±0.42), and CX3CR1 (5.55±0.40, 5.91±0.34) were significantly different between the SLE-depression group and SLE-non-depression group( F=5.549~28.371, all P<0.05). Compared with the SLE-non-depression group, the SLE-depression group exhibited significantly lower FA values in the genu of the corpus callosum (0.29±0.04, 0.31±0.04) and the left uncinate fasciculus (0.31±0.02, 0.33±0.02), as well as significantly smaller GMV in the right hippocampus (0.29±0.07, 0.33±0.06), bilateral middle frontal gyrus (left: 0.27±0.05, 0.31±0.05; right: 0.28±0.06, 0.32±0.06), right insula (0.36±0.03, 0.38±0.04), and left precentral gyrus (0.19±0.04, 0.24±0.05) ( F=4.593-12.064, all P<0.05, FDR correction). GMV in the right anterior cingulate and paracingulate gyri was significantly larger than that in the SLE-non-depression group (0.34±0.05, 0.29±0.06) ( F=6.704, P=0.034, FDR correction). Partial correlation analysis revealed significantly stronger correlations between hub DEGs and altered brain regions in the SLE-depression group ( r=0.017-0.801) compared to the SLE-non-depression group ( r=0.002-0.382), with a statistically significant difference ( U=629, P<0.001; Cliff's Delta=0.454). GSEA indicated that the aforementioned genes were primarily involved in pathways including the ribosome, spliceosome, ribosome biogenesis in eukaryotes, and neuroactive ligand-receptor interaction. Sensitivity analysis confirmed that the above results remained statistically significant after balancing sample sizes (all P<0.05). Conclusion:The SLE-depression group showed specific RNA expression and brain structure alterations compared to the SLE-non-depression group, and the correlation between RNA and brain structure was significantly enhanced in the SLE-depression group. This suggests that the correlation between genes and brain structure in the SLE population may be related to their susceptibility to depression.

Objective:To investigate the predictive value of multimodal magnetic resonance imaging (MRI) techniques in assessing symptom remission of post-traumatic stress disorder (PTSD) of bereaved parents who lost their only child.Methods:In this prospective study, 34 parents with PTSD resulting from the loss of the only child were followed-up for 2 years. Based on the PTSD diagnostic status at the end of the follow-up, participants were divided into the remission group and the persistent group.R 3.6.1 and SPSS 20.0 software were used for statistical analysis.Baseline clinical data and neuroimaging findings were compared between the two groups. Logistic regression and LASSO regression analyses were used to identify independent predictors of PTSD symptom remission. The predictive performance of these factors was evaluated by receiver operating characteristic (ROC) curve analysis.Results:Initial screening with univariate Logistic regression and LASSO regression revealed that regional homogeneity (ReHo) in the left middle temporal gyrus, the combined predictive value based on ReHo, and the integrated predictive value combining gray matter volume (GMV) and ReHo (GMV-ReHo predictor) were significant factors influencing symptom remission (all P<0.05). Multivariate Logistic regression further demonstrated that the GMV-ReHo predictor retained independent predictive significance ( P<0.05), with ROC curve analysis showing an area under the curve (AUC) of 0.979 (95% CI=0.935-0.996, P<0.001) for its ability to predict PTSD remission. Notably, a combined model incorporating both the scores of the clinician administered PTSD scale (CAPS) and the GMV-ReHo predictor achieved an enhanced predictive performance, yielding an AUC of 0.984 (95% CI=0.952-0.998, P<0.001). Conclusion:The GMV-ReHo predictor effectively identifies symptom remission in PTSD resulting from the loss of the only child.

Objective:To investigate the differences in hippocampal subfield volumes and structural covariance network between participants with subjective cognitive decline (SCD) and healthy individuals, and to analyze the correlations of the volumes of the different subfields and altered covariance brain regions with memory function.Methods:A total of 57 SCD individuals(SCD group) and 44 normal controls(NC group) participants were assessed for memory function using composite scores from the auditory verbal learning test (AVLT) and the Wechsler memory scale visual reproduction (VR) test from June 2022 to October 2023.T1-weighted structural magnetic resonance imaging (MRI) data were collected from all participants, and hippocampal subfields, cortical regions, and subcortical nuclei were segmented using FreeSurfer to measure the gray matter volume of each structure. A structural covariance network was constructed based on the correlation of gray matter volumes across regions. Statistical analysis was performed using R 4.3.1 software. Inter-group differences in hippocampal subfield volumes were compared using multivariate analysis of covariance. Differences in structural covariance connectivity between groups were assessed using Z-test, while network topology differences were compared through permutation testing. Finally, partial correlation analysis was used to examine correlation of the volumes of the differential hippocampal subfields and covariance brain regions with memory function. Results:The SCD group exhibited significantly lower years of education, AVLT-immediate score, AVLT-delayed score, VR-immediate score, VR-delayed score, and memory function Z-score compared to the NC group ( t=2.064, 3.888, 2.622, 3.222, 4.761, 5.184, all P<0.05). The volumes of the right subiculum((387.75±55.20)mm 3, (352.70±70.25)mm 3), left presubiculum((263.12±38.52)mm 3, (239.79±46.02)mm 3), left subiculum((388.12±49.34)mm 3, (351.74±67.30)mm 3) and left CA1((571.01±80.01)mm 3, (526.51±98.80)mm 3) in the SCD group were smaller than the corresponding volumes in NC group ( F=9.139, 8.039, 11.207, 7.266, all P<0.05, FDR correction). Differences in structural covariance connectivity were found between the SCD and NC groups in the following pairs: right CA1-right subiculum, right CA1-left subiculum, right CA3-left parasubiculum and right hippocampus-amygdala transition area-left subiculum ( Z=-3.848, -3.896, -3.597, -3.895, all P<0.05, FDR correction).Partial correlation analysis revealed that in the SCD group, the volume of the left subiculum ( r=0.359, P=0.007), left CA1 ( r=0.430, P=0.001), right entorhinal cortex ( r=0.296, P=0.029), right middle temporal gyrus ( r=0.361, P=0.007), right parahippocampal gyrus ( r=0.313, P=0.021)were positively correlated with the total memory function score. Conclusion:Hippocampal subfields atrophy, as well as alterations in structural covariance network, have been found in SCD individuals. Furthermore, the decline in memory function may be closely associated with atrophy in hippocampal subfields and structurally covariant regions.

Objective:To investigate the alterations of individual metabolic brain network properties in patients with mild cognitive impairment (MCI) and their correlations with cognitive function.Methods:One hundred and five participants from Alzheimer's Disease Neuroimaging Initiative (ADNI) database enrolled from March 2012 to February 2016 were chosen, including 61 MCI patients and 44 normal controls (NC). Cognitive assessments, including mini-mental state examination (MMSE), auditory verbal learning test (AVLT), trail making test (TMT), and semantic verbal fluency (SVF) score, were performed in both groups; differences of above scores and clinical data between the participants from the two groups were compared. T1-weighted imaging and fluorodeoxyglucose positron emission tomography (FDG-PET) images were collected in both groups; individual metabolic brain networks were constructed based on differences in effect sizes between brain regions and network properties were calculated. Spatial correlation analysis was used to compare the correlations of metabolic brain networks at the individual and group levels. General linear model was employed to compare the differences in network properties between the two groups. Partial correlation analysis was used to examine the correlations of differential network properties with cognitive function in MCI patients. A support vector machine (SVM) classification model was constructed based on individual metabolic brain network properties, and receiver operating characteristic (ROC) curve was used to explore the diagnostic value of this SVM classification model in MCI.Results:(1) Compared with the NC group, the MCI group had significantly lower MMSE and AVLT-immediate recall scores, and longer TMT-A completion time ( P<0.05). (2) Spatial correlation analysis revealed a positive correlation between individual metabolic brain networks and group-level metabolic brain networks in patients of the MCI group ( r=0.825, P<0.001). No significant differences in global network properties were noted between the two groups ( P>0.05). Compared with the NC group, the MCI group significantly decreased degree centrality in the left A8vl, right A39c, and right V5/MT+ regions, increased degree centrality in the left anterior cuneus, decreased nodal efficiency in the left A8vl, right V5/MT+, and right caudal hippocampus regions, increased nodal shortest path length and nodal clustering coefficient in the left A8vl region ( P<0.05). (3) The degree centrality at the A8vl of ventral part of the left middle frontal gyrus and nodal efficiency in right caudal hippocampus region were positively correlated with AVLT-immediate recall scores ( r=0.331, P=0.010; r=0.282, P=0.030), nodal efficiency in the left A8vl region was negatively correlated with TMT-A completion time ( r=-0.470, P<0.001), and nodal efficiency in the left A8vl region was positively correlated with SVF score ( r=0.263, P=0.044). (4) Area under the curve of SVM classification model in diagnosing MCI was 0.880 (95% CI: 0.813-0.945, P<0.001), with an accuracy rate of 0.790. Conclusions:Patients with MCI have alterations in individual metabolic brain network properties, among which the degree centrality and nodal efficiency of some nodes are closely related to cognitive function changes. Models constructed based on individual metabolic brain network properties can help to effectively diagnose MCI.

Objective:To investigate the interactive effects of loss of the only child and childhood trauma on brain structure, function, and structure-function coupling, and to analyze their association with clinical symptom.Methods:A total of 112 parents who lost their only child and participated in the psychological aid project organized by Local Civil Affairs Department in Sunan aear of Jiangsu Province in China from April 2021 to July 2021 and 36 healthy controls recruited from the community during the same period were selected. Based on childhood trauma questionnaire scores, parents who had lost their only child were divided into those with childhood trauma (group A, n=55) and those without childhood trauma (group B, n=57); similarly, the healthy controls were divided into a group with childhood trauma (group C, n=12) and a group without childhood trauma (group D, n=24). All participants were evaluated by clinical scales such as Hamilton Depression Scale (HAMD), Hamilton Anxiety Scale (HAMA), and Mini-Mental State Examination (MMSE). MRI 3D-T1 structural images and resting-state functional magnetic resonance imaging data were collected; gray matter volume (GMV) and degree centrality (DC) were calculated by standardized image preprocessing procedure, and ratio of DC to GMV within each voxel was computed to obtain the structure-function coupling map. A two-factor analysis of variance was used to analyze the independent effect and interactive effect of loss of the only child and childhood trauma on GMV, DC, and DC/GMV coupling value. Spearman rank correlation analysis was used to evaluate the associations of above indicators in brain regions with significant difference in independent effect and interactive effect with clinical scale scores. Results:(1) Compared with the participants without childhood trauma (group B+group D), the participants with childhood trauma (group A+group C) showed significantly reduced GMV in the left middle temporal gyrus and right dorsolateral superior frontal gyrus (voxel-level P<0.01, cluster-level P<0.05, Gaussian random field [GRF] corrected). A significant interactive effect of loss of the only child and childhood trauma on GMV in the right precuneus was observed (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). (2) Compared with the healthy controls, parents who had lost their only child exhibited significantly increased DC in the left middle frontal gyrus (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). Compared with participants without childhood trauma, participants with childhood trauma showed significantly increased DC in the right thalamus (voxel-level P<0.01, cluster-level P< 0.05, GRF corrected). A significant interactive effect of loss of the only child and childhood trauma on DC in the left dorsolateral superior frontal gyrus was observed (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). (3) Compared with the healthy controls, parents who had lost their only child showed significantly decreased DC/GMV coupling value in the left middle frontal gyrus (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). Compared with participants without childhood trauma, participants with childhood trauma showed significantly increased DC/GMV coupling value in the right thalamus (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). A significant interactive effect of loss of the only child and childhood trauma on DC/GMV coupling value in the left dorsolateral superior frontal gyrus was observed (voxel-level P<0.01, cluster-level P<0.05, GRF corrected). (4) Correlation analysis revealed that GMV in the right precuneus with significant interactive effect of loss of the only child and childhood trauma was positively correlated with MMSE score ( r s=0.317, P=0.010, Bonferroni corrected). GMV in the left middle temporal gyrus with significant independent effect of childhood trauma was positively correlated with both HAMD score and HAMA score ( r s=0.362, P=0.006; r s= 0.349, P=0.008, Bonferroni corrected). Conclusion:Loss of the only child and childhood trauma can interact to jointly affect the brain structure, function, and structure-function coupling; and some of these brain structure alterations are closely associated with clinical symptoms.

Objective:To investigate the differences in hippocampal subfield volumes and structural covariance network properties among patients with mild cognitive impairment (MCI) exhibiting different cognitive outcomes and normal controls (NCs), and to further evaluate the predictive value of these imaging indicators for cognitive deterioration in MCI patients.Methods:A total of 43 NCs, 65 stable MCI (sMCI), and 26 progressive MCI (pMCI) patients enrolled in the Alzheimer′s Disease Neuroimaging Initiative (ADNI) database between December 2012 and May 2016 were included in this study. Baseline demographic information and T 1-weighted magnetic resonance imaging scans were collected. Hippocampal subfield volumes were extracted using freesurfer software, and structural covariance networks of hippocampal subfields were constructed. Multivariate analysis of covariance was used to compare hippocampal subfield volumes among the 3 groups. A general linear model was applied to examine group differences in hippocampal subfield structural covariance network properties. Least absolute shrinkage and selection operator (LASSO)-Logistic regression was employed to identify imaging predictors associated with conversion to Alzheimer′s disease (AD), based on which structural, network-based, and combined predictive models were constructed. Model discrimination was evaluated using the area under the curve (AUC); internal validation was performed using Bootstrap resampling; model calibration was assessed with the Hosmer-Lemeshow test; and clinical utility was evaluated through decision curve analysis. Results:Significant differences in hippocampal subfield volumes (mm3) were observed among the 3 groups (all P<0.05, Bonferroni-corrected). Specifically, left parasubiculum (65.58±13.30, 61.96±17.56, 49.56±11.82, F=9.900), right parasubiculum (65.92±15.21, 59.45±16.65, 47.69±15.48, F=11.612), left presubiculum (277.09±39.85, 258.15±44.86, 224.05±45.05, F=14.513), right presubiculum (262.85±40.43, 247.41±43.27, 209.97±46.11, F=14.500), left subiculum (399.66±32.19, 374.25±55.83, 306.12±51.62, F=32.923), right subiculum (417.93±48.92, 376.59±51.01, 316.82±70.22, F=28.764), left cornu ammonis 1 (CA1) (592.10±83.87, 561.96±94.72, 490.06±86.89, F=13.352), right CA1 (632.15±100.09, 601.24±88.88, 531.05±110.29, F=10.579), left CA3 (191.58±30.08, 180.47±34.66, 155.08±37.82, F=12.182), right CA3 (210.42±28.92, 203.84±34.80, 176.69±41.47, F=9.597), left CA4 (224.61±28.94, 210.49±35.04, 183.98±36.89, F=16.521), right CA4 (238.49±28.14, 227.43±30.65, 200.23±42.74, F=13.702), left granule cell-molecular layer-dentate gyrus (GC-ML-DG) (259.96±36.76, 239.42±41.17, 207.61±41.84, F=19.831), right GC-ML-DG (273.98±35.12, 258.79±36.82, 227.81±49.07, F=14.204), left molecular layer (505.62±66.16, 468.58±75.17, 402.68±75.47, F=22.293), right molecular layer (527.39±72.39, 493.14±70.39, 423.81±88.09, F=19.588), left hippocampal amygdala transition area (HATA) (54.91±9.99, 49.52±9.93, 43.27±9.59, F=13.571), right HATA (58.43±9.83, 54.55±10.80, 47.12±12.54, F=10.037), left fimbria (69.94±25.04, 56.63±23.74, 40.58±19.83, F=14.846), right fimbria (68.61±26.24, 53.95±23.16, 45.25±17.04, F=10.424), left hippocampal tail (488.37±83.44, 463.54±80.33, 393.83±77.73, F=13.570), and right hippocampal tail (519.78±80.22, 498.84±81.68, 419.75±93.29, F=14.339) all showed significant group differences. Significant group differences were also observed in small-worldness metric γ (0.51±0.10, 0.51±0.08, 0.62±0.14, F=9.317), small-worldness metric λ (0.39±0.02, 0.39±0.02, 0.43±0.04, F=9.925), global efficiency (0.19±0.01, 0.20±0.01, 0.18±0.01, F=3.189), local efficiency (0.26±0.02, 0.26±0.01, 0.27±0.01, F=3.068), clustering coefficient (0.23±0.01, 0.23±0.01, 0.24±0.02, F=4.274), and characteristic path length (0.73±0.06, 0.72±0.06, 0.76±0.07, F=4.477) of the hippocampal subfield structural covariance network (all P<0.05). Specifically, the pMCI group exhibited higher γ ( t=3.773, P<0.001), λ ( t=4.060, P<0.001), local efficiency ( t=2.445, P=0.047), and clustering coefficient ( t=2.849, P=0.015) than the NCs group, and higher γ ( t=4.074, P<0.001), λ ( t=4.068, P<0.001), and characteristic path length ( t=2.986, P=0.010) but lower global efficiency ( t=-2.444, P=0.047) than the sMCI group. The AUC of the structural, network, and combined models based on LASSO-Logistic regression was 0.837, 0.861, and 0.899, respectively. After internal validation, the corrected AUC was 0.835, 0.855, and 0.889, respectively. All models demonstrated good calibration ( P>0.05), and decision curve analysis indicated favorable clinical net benefit across models. Conclusions:Both sMCI and pMCI patients exhibit widespread hippocampal subfield atrophy and altered global properties of hippocampal subfield structural covariance networks compared to NCs. The models constructed based on hippocampal subfield volumes and structural covariance networks show strong potential for predicting cognitive decline in MCI patients.