A total of 134 Chinese Han participants (72 patients with AD and 62 controls) were recruited from the Second Hospital of Shandong University, Shandong, China between May 2017 and July 2019. The patient's age range was 53 to 86 years, and the ratio of male to female patients was 0.89:1. All subjects received a cranial MRI and/or CT scan in order to assess the degree of brain atrophy and to exclude patients with other diseases which may cause cognitive impairment, including acute cerebrovascular disease, hydrocephalus and brain tumors. No cerebrospinal fluid (CSF) samples were tested.

All participants underwent neuropsychological tests including the Montreal Cognitive Assessment Beijing Version (31,32), the Chinese version of the Mini Mental State Examination (MMSE) (33), the Clinical Dementia Rating (CDR) scale (34), the Activities of Daily Living scale (35), and the Hamilton Depression (HAMD) scale (36). Probable AD dementia was diagnosed according to the National Institute on Aging and the Alzheimer's Association criteria (2011) (23). Participants were excluded from the study if they had: i) A history of stroke; ii) psychosis; iii) severe depression (HAMD score ≥24 points); iv) other nervous system diseases, such as Lewy body dementia, Parkinson's disease, encephalitis, brain tumors and traumatic brain injury; or v) some systemic diseases, such as thyroid dysfunction, severe anemia, syphilis or HIV infection (37).

The present study was performed in accordance with the recommendations of the Medical Research Ethics Committee of the Second Hospital of Shandong University. Written informed consent was obtained from all controls and guardians of patients with AD.

A total of 5 ml whole blood samples were collected from all participants in EDTA tubes and were centrifuged at 3,000 × g at 4°C for 15 min to obtain the platelet-free plasma, which was subsequently aliquoted and stored at −80°C. Exosomes were isolated from the plasma using the ExoQuick Plasma prep and Exosome Precipitation kit (EXOQ5TM-1; System Biosciences, LLC). The primary experimental steps were: 0.5 ml plasma was added to 4 µl thrombin (611 U/ml) and incubated at room temperature for 5 min with agitation, then subsequently centrifuged at 14,000 × g for 5 min at 4°C. The serum-like supernatant was collected and further centrifuged at 3,000 × g for 15 min at 4°C to remove cells and debris and then the appropriate volume of ExoQuick Exosome Precipitation Solution was added according to the manufacturer's protocol. After refrigerating the mixture overnight at 4°C, samples were centrifuged at 1,500 × g for 30 min at 4°C. The pellets were resuspended in PBS and stored at −80°C (38).

Western blotting was performed to identify the marker proteins of exosomes. Exosome pellets were cleaved using radioimmunoprecipitation assay lysis buffer (Beyotime Institute of Technology) at 4°C, and protein concentration was determined using a bicinchoninic acid protein assay kit (cat. no. P0012, Beyotime Institute of Technology). Equal quantities of protein (20 µg) were loaded onto a 10% SDS-gel and resolved using SDS-PAGE. Subsequently, the resolved proteins were transferred to PVDF membranes (EMD Millipore) and incubated overnight at 4°C with either apoptosis-linked gene 2-interacting protein X (ALIX; 1:5,000, cat. no. ab186429, Abcam) or cluster of differentiation (CD)63 (1:2,000, cat. no. ab217345, Abcam) antibodies. The membranes were incubated for 1 h at room temperature with the goat anti-rabbit IgG/horseradish peroxidase (1:5,000, cat. no. ab6721, Abcam). Signals were visualized using chemiluminescence reagent (EMD Millipore) and densitometry analysis was performed using ImageJ version 1.46 (National Institutes of Health) (39).

The fresh exosomes were resuspended in PBS. After placing on copper grids coated with a layer of formvar/carbon, the specimens were fixed with 3% glutaraldehyde for 5 min at room temperature. Subsequently, they were stained negatively with 1% uranyl acetate for 10 min at room temperature. Finally, the exosomes were observed using a JEOL 1200EX electron microscope at ×10,000 magnification (JEOL, Ltd.) at 80 kV as described previously (40).

Nanoparticle tracking analysis (NTA) was performed using Nanoparticle Tracking Analysis (ZetaView; Particle Metrix) to measure the size distribution of particles. Samples were diluted in PBS at a ratio of 1:6,000 to achieve a particle count of 1×10⁸ particles per ml, the desired concentration. Then, 3 cycles were performed by scanning 11 cell positions for every measurement and capturing 60 frames per position. All pictures were processed and analyzed using the built-in ZetaView Software (41).

Previous studies have demonstrated that expression levels of lncRNA BACE1-AS, 51A, BC200 and BACE1 mRNA are upregulated in the brains of patients with AD and may be detected in the peripheral blood (18,19,22). To expand on these studies, the expression levels of these RNAs in peripheral blood exosomes were examined. Of the 134 recruited participants, the expression levels of BACE1-AS were assessed in all subjects, but the expression of BACE1 mRNA, BC200 and 51A were detected in only 123 subjects due to the limitation of blood specimens.

Total RNA was extracted from the plasma exosomes using RNAiso Plus (Takara Bio, Inc.). The RNA was reverse transcribed using PrimeScript™RT reagent kit (Takara Bio, Inc.) in a total volume of 20 µl with the following conditions: 15 min at 37°C, and 5 sec at 85°C. The expression of RNA was detected using quantitative PCR using TB Green Premix Ex Taq (Takara Bio, Inc) on a Bio-Rad CFX 96 Real-Time system (Bio-Rad Laboratories, Inc.). All reactions were performed in triplicate. The thermocycling conditions were: 95°C for 10 min; followed by 40 cycles of 95°C for 15 sec and 60°C for 60 sec. GAPDH was used as the internal control. The expression of RNA was calculated using the 2^−ΔΔCq method (42). The primer sequences are presented in Table I.

Of all the participants, there were 22 patients with AD and 26 controls who received brain MRI scans that met the analysis criteria. The 3.0T magnetic resonance (Discovery MR750W; GE Healthcare) was used for image acquisition. T1-weighted MRI scans were acquired using a 3D-BRAin VOlume (BRAVO) sequence with the following parameters: Repetition time (TR)=8.2 msec, echo time (TE)=.2 msec, flip angle=12°, inversion time (TI)=450 msec, matrix=256 × 256, slices=140, and slice thickness=1.0 mm (43).

In order to extract effective analytical features from MR brain T1 structure images, a collaborating company (QED Technique Co., Ltd.) developed a set of automated analysis tools for brain structure for the present study. Hybrid processing algorithms which assisted in the complete removal of non-brain tissue, subcortical volume segmentation and cortical reconstruction without human intervention were used. The specific steps were: i) Tissue segmentation; all MR brain T1 images were registered to a standard ICBM-152 template which then calculated the maximum probabilities and corrected intensity non-uniformities, created an effective brain mask (GM, WM, CSF), and completed tissue segmentation associated with AD, including caudate, putamen, hippocampus, amygdala, ventricles and other regions; ii) cortical morphometric analyses; all T1 spatially normalized images were registered nonlinearly to a surface unbiased group template using surface-based with a spherical warping algorithm. Subsequently, cortical surfaces between WM-GM and GM-CSF were estimated, and then cortical thickness was calculated between these surfaces (WM-GM/GM-CSF). In previous studies, convex (gyri) and concave (sulci) surfaces had different anatomical and connective characteristics; therefore, these units were also analyzed separately to calculate local curvature and surface area. These features were used to distinguish between patients with AD and control subjects (44,45).

Statistical analysis was performed using SPSS Statistics version 24.0 (IBM Corp.) and GraphPad Prism Version 5.0 (GraphPad Software, Inc.). A Kolmogorov-Smirnov or Shapiro-Wilk test was used to check the Gaussian distribution. Gaussian distributed numerical data are presented as the mean ± standard deviation and skewed distribution variables are expressed as the median and interquartile range. Categorical variable data are expressed as a frequency and percentage. The correlation between lncRNA and clinical features was analyzed using a Spearman's rank or Pearson correlation tests. Logistic regression was used to compute probabilities, the corresponding 95% confidence interval (CI) and the odds ratio. Quantitative values were compared using a Mann-Whitney U test or Student's t-test. A χ² test was used to assess differences in categorical data. Receiver operating characteristic (ROC) curves and the areas under curve (AUC) were used to assess the sensitivity and specificity of the predictive ability of using plasma exosomal lncRNAs and MRI measurement parameters as biomarkers for AD. Multi-index combined detection for diagnosis was based on series-parallel test and logistic regression model. P<0.05 was considered to indicate a statistically significant difference.

A total of 72 patients with AD and 62 control subjects were recruited in the present study. The two groups were matched based on sex, age and education level. As shown in Table II, the MMSE scores of patients with AD was significantly lower compared with the controls (P<0.001). According to the CDR score, there were 9 mild, 36 moderate and 27 severe AD patients. In order to examine potential biomarkers, other clinical characteristics were analyzed, including blood pressure, blood lipid, blood glucose, hemoglobin (Hb), thyroid hormone levels, heart disease, history of anesthesia, and smoking and drinking status. Results demonstrated a significant decrease in triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) levels, in patients with AD, and in Hb levels in male patients with AD (P<0.05), and none of the other assessed characteristics were significantly different (P>0.05).

The exosome marker proteins, ALIX and CD63, were detected using western blotting (Fig. 1A). To further confirm that the sediment extracted from plasma was exosomes, TEM and NTA analyses were performed. Using electron microscopy, it was shown that the vesicles were spherical and had a diameter ranging between 30–150 nm. The arrow in Fig. 1B indicates a vesicle that was ~100 nm in diameter. The particle diameters ranged between 16.5 and 478.3 nm. The majority of the diameters were 98.9 nm, accounting for 99.7% of all particles (Fig. 1C). Therefore, the extracellular vesicles extracted matched the characteristics of exosomes (10).

The relative expression of lncRNA BACE1-AS, 51A, BC200 and BACE1 mRNA was determined (Table III). As shown in Fig. 2, BACE1-AS levels in patients with AD were significantly higher compared with the controls (Fig. 2A; P<0.005). There were no significant differences in the expression levels of BACE1 mRNA (Control, 55; AD, 68; Fig. 2B; P=0.327), BC200 (Control, 57; AD, 66; Fig. 2C; P=0.354) and 51A (Control, 57; AD, 66; Fig. 2D; P=0.451) between the two groups.

As mentioned above, the blood TG, TC and HDL-C levels in patients with AD were significantly lower compared with controls (Table II). In order to determine whether the diagnostic value of BACE1-AS was independent of other clinical characteristics, correlation analyses between BACE1-AS and blood levels of TG, TC and HDL-C and age in both groups, as well as onset time and MMSE scores in patients with AD, was performed. As shown in Fig. 3, there were no significant correlations between these indices and BACE1-AS in both the AD and control group (Fig. 3A-D, F and G; all P>0.05). As previously mentioned, the blood levels of Hb in male AD patients were significantly lower compared with the females. The correlation between BACE1-AS and Hb levels in men and women was assessed separately, but no significant correlations were observed (Fig. 3E; P>0.05).

As shown in Fig. 2, the BACE1-AS levels of the patients with AD were significantly higher compared with the controls. ROC curve analysis was performed to determine the diagnostic value of BACE1-AS; the AUC was 0.761 (95% CI, 0.675–3.848; P<0.001). Although there were no statistically significant differences in expression levels of BACE1 mRNA, BC200 or 51A between the two groups, ROC curve analyses were also performed to evaluate their diagnostic value. For 51A, the AUC was 0.540 (95% CI, 0.436–3.643; P=0.451), the AUC of BC200 was 0.541 (95% CI, 0.436–3.645; P=0.441) and the AUC of BACE1 mRNA was 0.540 (95% CI, 0.437–3.644; P=0.444; Fig. 4).

There were 22 patients with AD and 26 controls who underwent 3D-BRAVO sequence MRI scans and met the automated analysis criteria. The hippocampus and entorhinal cortex MRI parameters in patients with AD and in the control group were analyzed, including the bilateral hippocampus volume, entorhinal cortex volume, entorhinal cortex mean thickness, entorhinal cortex surface area and entorhinal cortex mean curvature. As shown in Fig. 5F, the 3 images show the axial, coronal and sagittal position of a brain MRI of a subject from left to right, with the yellow region showing the hippocampus and the green region showing the entorhinal cortex. The results demonstrated that the right entorhinal cortex volume (P=0.015) and right entorhinal cortex mean thickness (P=0.022) of patients with AD were significantly lower compared with the controls (Table IV; Fig. 5A and B). There were no significant correlations between the right entorhinal cortex volume (r=−0.247, P=0.094), mean thickness and BACE1-AS in both groups (r=−0.246, P=0.096; Fig. 5C and D).

To evaluate the potential of the right entorhinal cortex volume and mean thickness as novel biomarkers for AD, ROC curves were plotted (Fig. 5E). The AUC was 0.688 (95% CI, 0.527–3.850; P=0.015) for the right entorhinal cortex volume, the sensitivity was 59.1%, and the specificity was 84.6% (cutoff point, 1,142.5). The AUC was 0.689 (95% CI, 0.536–3.843; P=0.025) for the right entorhinal cortex mean thickness, the sensitivity was 80.8%, and the specificity was 59.1% (cutoff point, 2.705; Table V).

As shown in Table V, none of the 3 indicators (BACE1-AS, right entorhinal cortex volume and mean thickness) had a high sensitivity and specificity at the same time for diagnosis of AD. Therefore, the cumulative performance of the 3 indices together was assessed using a series-parallel test and binary logistic regression models. The series-parallel test of BACE1-AS, right entorhinal cortex volume and right entorhinal cortex mean thickness raised the specificity and sensitivity to 96.15 and 90.91%, respectively. The combination of the 3 indices using a logistic regression model offered improved sensitivity and specificity simultaneously, particularly when adjusting for age and sex (AUC, 0.819; sensitivity, 81%; specificity, 73.1%; accuracy, 76.6%; Table V).