Chemically treated plasma Aβ is a potential blood-based biomarker for screening cerebral amyloid deposition

Plasma β-amyloid (Aβ) is a potential candidate for an Alzheimer’s disease (AD) biomarker because blood is an easily accessible bio-fluid, which can be collected routinely, and Aβ is one of the major hallmarks of AD pathogenesis in the brain. However, the association between plasma Aβ levels and AD diagnosis is still unclear due to the instability and inaccurate measurements of plasma Aβ levels in the blood of patients with AD. If a consistent value of plasma Aβ from the blood can be obtained, this might help determine whether plasma Aβ is a potential biomarker for AD diagnosis. We predicted the brain amyloid deposit by measuring the plasma Aβ levels. This cross-sectional study included 353 participants (215 cognitively normal, 79 with mild cognitive impairment, and 59 with AD dementia) who underwent Pittsburgh-compound B positron emission tomography (PiB-PET) scans. We treated a mixture of protease inhibitors and phosphatase inhibitors (MPP) and detected plasma Aβ42 and Aβ40 (MPP-Aβ42 and MPP-Aβ40) in a stable manner using xMAP technology. MPP-Aβ40 and MPP-Aβ42/40 (MPP-Aβs) were significantly different between subjects with positive amyloid deposition (PiB+) and those with negative amyloid deposition (PiB–) (P < 0.0001). Furthermore, MPP-Aβ40 (P < 0.0001, r = 0.23) and MPP-Aβ42/40 ratio (P < 0.0001, r = –0.23) showed significant correlation with global PiB deposition (standardized uptake value ratio). In addition, our integrated multivariable (MPP-Aβ42/40, gender, age, and apolipoprotein E genotypes) logistic regression model proposes a new standard for the prediction of cerebral amyloid deposition. MPP-Aβ might be one of the potential blood biomarkers for the prediction of PiB-PET positivity in the brain.

Efforts to identify the available and effective biomarkers for AD have been made in many countries. In the neuroimaging field, specific regional brain atrophy can be observed by magnetic resonance imaging (MRI), and Aβ deposition in the brain can be detected by positron emission tomography (PET) using specific radioactive ligands including Pittsburgh compound B (PiB) [12]. However, these are costly procedures for AD diagnosis [13]. Additionally, although molecular biomarkers such as tau and Aβ can be measured in the cerebrospinal fluid (CSF) [14], the collection of CSF is a highly invasive procedure [15].
On the other hand, the blood is an easily accessible bio-fluid which can be routinely collected and analyzed to detect or track the disease [15]. Hence, if we utilize blood proteins as a biomarker for AD, this might help recognize the disease earlier because of the simple method and high frequency of blood collection. Several studies reported the efflux/influx of Aβ across the blood-brain barrier and identified various Aβ transporters; hence, many researchers believe that plasma Aβ might reflect brain amyloid deposition [16]. However, the association between plasma Aβ levels and AD diagnosis is still unclear. Several cross-sectional studies showed that low plasma Aβ42 levels or Aβ42/40 ratios are associated with AD [17,18]. Some prospective studies have also suggested that a decrease in plasma Aβ42/40 ratio is related to cognitive impairment [19,20]. However, these results are in contrast with other reports, which indicate an increase in Aβ42 with cognitive decline [21], no association between plasma Aβ levels and AD [22], and a relation between low plasma Aβ40 and AD [23]. Thus, evidence for the effectiveness of measurement of plasma Aβ42, Aβ40, and the Aβ42/40 ratios in the diagnosis of AD is lacking. Such controversy on the relationship between plasma Aβ levels and AD may result from the inaccuracy in both plasma Aβ measurement and AD clinical diagnosis. In addition, plasma Aβ levels show unpredictable fluctuations in their values because of numerous factors. Human serum albumin (HSA) and transthyretin (TTR) interrupt the detection of Aβ [24][25][26][27] because both of them bind to Aβ, and their interaction status varies depending on blood dynamics [28,29]. Furthermore, protease inhibitors and phosphatase inhibitors in the blood might play a role in decreasing Aβ degradation because the Aβ sequence contains many possible proteases and phosphorylation sites [30,31]. After blood collection, both protease inhibitors and phosphatase inhibitors are randomly activated, which affects Aβ degradation and the interaction between Aβ and other proteins in the blood [31,32], resulting in the fluctuation of the measurable Aβ concentration in the blood. Moreover, about 12% of patients with clinically diagnosed AD dementia (ADD) do not exhibit AD pathology, and 23% of cognitively normal (CN) individuals show AD pathology [3,33]. Hence, the blood samples from previous studies that did not have PiB-PET data for the participants were not homogeneous samples in the view of brain amyloid pathology, which might have yielded the conflicting results for plasma Aβ level.
In this study, we aimed to investigate the potential of plasma Aβ level as a diagnostic or screening marker for AD. To resolve the instability or inaccuracy in assessment, we applied two distinct approaches. First, all participants underwent PiB-PET to quantify the cerebral Aβ deposition. This approach allowed us to test the effectiveness of plasma Aβ as a biomarker using pathological AD diagnosis, instead of clinical AD diagnosis, as the gold standard. Second, we developed a novel mixture (MPP, a mixture of protease inhibitors and phosphatase inhibitors) and treated the plasma with MPP to detect plasma Aβ levels in a stable and accurate manner using xMAP technology. Under this new but simple method, we determined that MPP-Aβ might be a reliable blood biomarker for screening cerebral amyloid deposition.

Methods
This study was part of the Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer's Disease (KBASE), an ongoing prospective cohort study aimed at searching for new biomarkers for AD and elucidating various life experiences contributing to AD-related brain changes. This work was approved by the Institutional Review Board (IRB) of the Seoul National University Hospital, South Korea, and the subjects or their legal representatives provided their written informed consent. Additional file 1 shows the experimental flow chart for this study.

Participants
Overall 353 middle-aged or old-aged subjects with age ≥ 55 years, including 215 CN individuals, 79 individuals with mild cognitive impairment (MCI), and 59 individuals with ADD, participated in the study. All participants underwent comprehensive clinical and neuropsychological assessments, neuroimaging examinations including structural MRI and PiB-PET, and comprehensive laboratory blood tests. CN participants had a Clinical Dementia Rating (CDR) score of 0 [34] and were without diagnoses of MCI or dementia. Individuals with MCI met the following criteria: (a) memory complaint corroborated by self, an informant, or a clinician; (b) objective memory impairment for age, education, and gender; (c) largely intact functional activities; and (d) not demented. All individuals with MCI had a global CDR score of 0.5. In terms of criterion (b), all participants with MCI had a performance score at least 1.5 standard deviation (SD) below the respective age-specific, education-specific, and gender-specific mean for at least one of the four episodic memory tests included in the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) neuropsychological battery (namely, Word List Memory, Word List Recall, Word List Recognition, and Constructional Recall test) [35]. Patients with ADD met both the criteria for dementia (in accordance with the Diagnostic and Statistical Manual, fourth edition (DSM-IV)) [36] and the criteria for probable AD set in accordance with the National Institute of Aging and Alzheimer's Association (NIA-AA) guidelines [37]. The exclusion criteria for all participants were: any present serious medical, psychiatric, and neurological disorders that could affect mental function; the presence of severe communication problems that would make a clinical examination or brain scan difficult; contraindications for MRI scan (e.g., pacemaker, claustrophobia); absence of a reliable informant; and illiteracy.

Clinical and neuropsychological assessment
Participants were administered standardized clinical assessments based on the KBASE clinical assessment protocol, which incorporated the Korean version of the Consortium to Establish a Registry for Alzheimer's Disease Assessment Packet (CERAD-K) [38], by trained psychiatrists. They were also administered the KBASE neuropsychological assessment protocol incorporating the CERAD neuropsychological battery [35], which included the Mini-Mental State Examination (MMSE), by trained neuropsychologists. In this study we used the MMSE z score, with consideration for age, gender, and education levels, because the basic demographic information of subjects such as gender, age, and education is known to have an effect on the change in MMSE scores [39].

Blood sampling
Blood samples were obtained via venipuncture in the morning (around 9:00 am) after an overnight fast and collected several hours before the injection of PET tracer in K2 ethylenediaminetetraacetic acid (EDTA) tubes (BD Vacutainer Systems, Plymouth, UK). The samples were stabilized and centrifuged at 700 × g for 5 min at room temperature (RT) to obtain the plasma supernatants in 15-ml centrifuge tubes (SPL Life Sciences Co., Gyeonggi-do, Korea). To obtain the samples with high purity, the plasma supernatants were further centrifuged under the same conditions, and the collected pure plasma supernatants were aliquoted and stored immediately at -80°C.

PiB-PET
Participants underwent simultaneous three-dimensional PiB-PET imaging and T1-weighted MR using a Biograph mMR scanner (Siemens, Washington, DC, USA) with the manufacturer's approved guidelines. A 30-min emission scan was acquired after 40 min of intravenous administration of 555 MBq of 11 C-PiB (range, 450-610 MBq). The PiB-PET data collected in the list mode were processed for routine corrections such as uniformity, ultrashort echo time (UTE)-based attenuation, and decay corrections and were reconstructed into a 256 × 256 image matrix using iterative methods (six iterations with 21 subsets). T1-weighted scans (repetition time = 1670 ms; echo time = 1.89 ms; field of view = 250 mm; 256 × 256 matrix with 1.0 mm slice thickness) were acquired in the sagittal orientation. The following image preprocessing steps were performed using Statistical Parametric Mapping 8 (SPM8) implemented in MATLAB 2014a (MathWorks, Natick, MA, USA). Static PiB-PET images were coregistered to the individual T1 structural images and transformation parameters for spatial normalization of the individual T1 images to a standard Montreal Neurological Institute (MNI) template were calculated. Utilizing Individual Brain Atlases using Statistical Parametric Mapping Software (IBASPM) software, the inverse transformation parameters were used to transform the coordinates from the automatic anatomic labeling (AAL) 116 atlas [40] into an individual space for each subject (resampling voxel size = 1 × 0.98 × 0.98 mm 3 ). The nongray matter portions of the atlas were individually masked using the cerebral gray matter segment image from each subject. The mean regional 11 C-PiB uptake values from the cerebral regions were extracted using the individual AAL116 atlas from the T1-coregistered PiB-PET images. The cerebellar gray matter was used as the reference region due to its relatively low Aβ deposition for the quantitative normalization of cerebral 11 C-PiB uptake values. To measure the 11 C-PiB uptake in the cerebellar gray matter regions, a probabilistic cerebellar atlas (Institute of Cognitive Neuroscience, UCL; Cognitive Neuroscience Laboratory, Royal Holloway) was transformed into an individual space in the same manner as already described. Of the 28 anatomical structural regions in the cerebellar atlas, all of the cerebellar lobular regions except for the vermis were included to extract the mean cerebellar uptake values. The AAL algorithm and a region-combining method [20] were applied to determine the regions of interest (ROIs) to characterize the 11 C-PiB retention level in the frontal, lateral parietal, posterior cingulate-precuneus (PC-PRC), and lateral temporal regions, where prominent 11 C-PiB retention has been reported [13].
The standardized uptake value ratio (SUVR) values for each ROI were calculated by dividing the mean value for all voxels within each ROI by the mean cerebellar uptake value in the same image. Each participant was classified as PiB-positive (PiB+) if the SUVR value was >1.4 in at least one of the four ROIs (i.e., frontal, lateral temporal, lateral parietal, and PC-PRC) or PiB-negative (PiB-) if the SUVR values of all four ROIs were ≤1.4 [17]. A global cortical ROI consisting of the four ROIs was also defined, and a global PiB deposition value (SUVR) was generated by dividing the mean value for all voxels of the global cortical ROI by the mean cerebellar uptake value of the same image.

Reagents
The mixture of protease inhibitors and phosphatase inhibitors (MPP) was composed of protease inhibitor cocktail (PI), phenylmethanesulfonyl fluoride (PMSF, a serine protease inhibitor; Sigma-Aldrich, CA, USA), and phosphatase inhibitor cocktail I and II (PPI I and II; A. G. Scientific, Inc., CA, USA). They were mixed in the same proportion. Aβ peptide was purchased from Bachem Americas, Inc. (Torrance, CA, USA) and prepared as described previously [41]. HSA (Sigma-Aldrich) was used for mimicking the plasma sample condition because it is the most abundant protein in the human plasma [25,42,43].

INNO-BIA plasma Aβ forms assay using xMAP technology
To simultaneously determine the concentrations of plasma Aβ42 and Aβ40, the INNO-BIA plasma Aβ forms kit (Innogenetics, Gent, Belgium) was used in accordance with the manufacturer's guidelines. In brief, the samples were diluted 3-fold in the MPP-treated plasma diluent buffer (final concentration, 4% MPP solution in plasma diluent buffer) or MPP nontreated plasma diluent buffer and incubated for 30 min at RT. After washing the filter plate, the diluted bead mix was transferred to the wells of the plate. The plate was dried gently, washed, and 25 μl of conjugate 1 working solution A and 75 μl of standards, blanks, controls, and diluted plasma samples were added. The plate was incubated overnight at 4°C, and the next day each well was washed with 100 μl of the detection solution added to the mixture. After 1 h, the plates were washed again, and the reading solution was added to each well. The levels of plasma Aβ were measured using xMAP technology (Bioplex 200 systems; Bio-Rad, Hercules, CA, USA).

Proof-of-concept experiments Gel electrophoresis
We prepared the synthetic Aβ42 (diluted in 1× phosphate buffered saline (PBS), 2 μM) to visualize the change in Aβ form against MPP treatment. Aβ42 solution was equally divided (20 μl) among eight 1.  Fig. 2b, lanes 5-8) was added to the Aβ42 solution (final concentration: 1 μM of Aβ42). The samples were incubated for 30 min at RT and then mixed with 4× sample buffer (without boiling) and loaded equally on 4-12% NuPAGE bis-tris gel (Thermo Fisher Scientific, Waltham, MA, USA) for gel electrophoresis. Next, the gel was transferred to a polyvinylidene difluoride (PVDF) membrane for 60 min at 70 V, and the membrane was blocked with 5% skim milk in 1× Tris buffered saline with Tween 20 (TBST) for 1 h. After blocking, the membrane was incubated with anti-6E10 Aβ antibody (primary antibodies overnight at 4°C) and the following day with secondary antibodies for 1 h at RT. The protein bands on the PVDF membrane were visualized using a bio-imaging analyzer (LAS-3000; Fujifilm Corporation, Tokyo, Japan) with a chemiluminescence detection solution (Ab Frontier Co., Seoul, Korea). The images were imported from a Multi-Gauge program (Fujifilm Corporation), and the band intensities (ratio, %) were measured (monomeric Aβ,~4 kDa; oligomeric Aβ, 8-14 kDa) [44].

xMAP technology
We repetitively measured the synthetic Aβ and plasma Aβ levels using xMAP technology. For synthetic Aβ levels, 4% MPP solution (or absent), 0.5% HSA (or absent), and Aβ42 (200 pg/ml) (aliquots from the same pool, in separate 1.5-ml tubes; -MPP or + MPP, each n = 6) in Bioplex sample diluent buffer were incubated for 30 min at RT, followed by Aβ42 measurement using xMAP technology. For plasma Aβ levels, samples (aliquots from the same pooled plasma, in separate 1.5-ml tubes, n = 5) were diluted 3-fold in MPP-treated plasma diluent buffer (final concentration, 4% MPP solution in plasma diluent buffer) or MPP nontreated plasma diluent buffer, incubated for 30 min at RT, and then measured using xMAP technology (-MPP or + MPP, each n = 5).

Time-dependent alterations in plasma Aβ42 and Aβ40 levels
Plasma Aβ42 and Aβ40 levels were quantified in a timedependent manner using xMAP technology. We used individual human plasma samples (n = 4) and also diluted them 3-fold in MPP-treated plasma diluent buffer (final concentration, 4% MPP solution in plasma diluent buffer) or MPP nontreated plasma diluent buffer, and incubated at RT followed by measurement at each time point (0, 6, 12, 24 h) and quantification (-MPP or + MPP, each n = 4).

Effects of MPP on discrimination between subject groups
Plasma samples (20 CN-, 12 MCI+, 23 ADD+, total of 55 subjects;and +, PiB-PET positivity) were diluted 3-fold in MPP-treated plasma diluent buffer (final concentration, 4% MPP solution in plasma diluent buffer) or MPP nontreated plasma diluent buffer and incubated for 30 min at RT followed by measurement using xMAP technology. We compared the intergroup differences in MPP-treated plasma Aβ (MPP-Aβ) with those of non-MPP-treated plasma Aβ (nMPP-Aβ).

Statistics
Statistical analysis was performed using GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA) and Medcalc (Medcalc Software, Ostend, Belgium). All data are presented as mean ± standard error of mean (SEM). An unpaired t test was used to assess the quantitative differences between two groups. For the tests of differences between three groups or more, multifactorial analyses of variance (ANOVA) followed by Tukey's multiple comparisons tests were performed. We used a bivariate correlation analysis to test the association between variables. Furthermore, we conducted logistic regression analysis followed by receiver operating characteristic (ROC) curve analysis on these values using Medcalc software. Cutoff criteria (optimal cutoff point) were determined by the Youden index [45] with an appropriate balance between sensitivity and specificity. Pearson's chi-square test (χ 2 ) was carried out to evaluate the intergroup differences of two-category variables (gender, apolipoprotein E ε4 carrier status) using Medcalc software. To compare the variance between -MPP and + MPP in the repetitive measurement assay, an F test was performed (GraphPad Prism 5). Squared deviation ((valuemean value) 2 ) indicates how far each point is from the mean value. Because the y axis of the graphs indicates the squared deviation in Fig. 2a-d, the highest points of each bar graph equates to 'Variance': where X is an individual data point, m is the mean of data points, and N is the total number of data points. Statistically significant results were shown as the appropriate P value.

Demographic data of subjects
A total of 353 subjects were included in this study. Subjects were classified according to their pathological (brain Aβ deposition) and/or clinical diagnosis (Fig. 1, Table 1, and Additional file 2). Two hundred and fiftythree subjects with negative amyloid deposition (PiB-) and 100 subjects with positive amyloid deposition (PiB+) were included; out of which 187 were PiB-CN (CN-), 50 were PiB-MCI (MCI-), 16 were PiB-ADD (ADD-), 28 were PiB+ CN (CN+), 29 were PiB+ MCI (MCI+), and 43 were PiB+ ADD (ADD+). Figure 1b shows the representative PiB-PET images of the study cohort (n = 353). The colors of the rainbow spectrum show the degree of Aβ deposition (PiB retention, SUVR; red → purple, high → low, respectively).

Proof-of-concept experiments for the effects of MPP on Aβ quantification
To verify the effect of MPP on Aβ quantification, we used the synthetic Aβ42 and plasma samples for the proof-ofconcept experiments (POC) (Fig. 2). When the same concentration of synthetic Aβ42 (final concentration: 1 μM of Aβ42) in the presence or absence of MPP was electrophoresed using western blotting, the Aβ42 samples without

POC for the effects of MPP on the distinction among the subject groups
To examine the correlation between the plasma Aβ concentration and brain Aβ deposition, plasma samples from 55 subjects (20 CN-, 12 MCI+, 23 ADD+; see Additional file 4 for more detail) who received PiB-PET imaging were used for further analysis (Fig. 3a, b). nMPP-Aβ42 and nMPP-Aβ42/40 levels showed no significant intergroup differences among all groups. In Pearson's correlation) were more correlated with brain amyloid deposit than nMPP-Aβ42 and nMPP-Aβ42/40 ( Fig. 3b; nMPP-Aβ42, P < 0.05, r = -0.29; nMPP-Aβ42/40, P = 0.12, r = -0.21; Pearson's correlation). Therefore, we suggest that the treatment of the plasma with MPP provides an improved detection method for stable and reliable plasma Aβ level, and plasma Aβ concentration might reflect the brain amyloid deposit.

Prediction of cerebral amyloid deposition based on MPP-Aβ42/40 ratio
To apply these analyzed results (Fig. 4) to practical use, we performed a subsequent logistic regression and ROC curve analysis using independent variables (Fig. 5). Multiple variables (MPP-Aβs and ApoE genotype) and control variables (gender and age) were mixed by logistic regression analysis, and the predicted probabilities were used to assess the discrimination power (ROC curve analysis) for the prediction of PiB-PET positivity. The classification variable was ND-vs ND+ (Fig. 5a) or PiB-vs PiB+ (Fig. 5b). The combination of MPP-Aβ42/40 and control variables (gender and age) increased the area under curve (AUC) values (ND-vs ND+, 0.695; PiB-vs PiB+, 0.682) compared with MPP-Aβ42/40 alone (ND-vs ND+, 0.639; PiB-vs PiB+, 0.668) (Fig. 5 and Table 3). Furthermore, AUC values were further enhanced on adding the ApoE variable (ND-vs ND+, 0.783; PiB-vs PiB+, 0.799). Table 3 presents more detail about the results of logistic regression followed by ROC curve analysis.

Discussion
Plasma Aβ levels are believed unstable, and many plasma proteins such as HSA, TTR, and others are believed to interrupt the detection of Aβ [24,26]. Furthermore, the plasma contains diverse proteases released from neutrophils and phagocytes [40,47,48]. Although the plasma contains several protease inhibitors such as α 2 -macroglobulin, α 1 -protease inhibitor, and plasminogen activator inhibitor-1 (PAI-1) [49][50][51][52], these are not sufficient for the preservation of the integrity of plasma proteins during the storage of samples at 4°C or RT [53]. Moreover, Kumar et al. [31] reported that phosphorylation of Aβ at Ser-8 inhibited the degradation of monomeric Aβ. Accurate analysis with plasma requires a high quality of starting samples; however, there have been no previous attempts to stabilize the plasma Aβ by chemical treatment. We hypothesized that protease inhibitors and phosphatase inhibitors might have a role in reducing Aβ degradation because the Aβ sequence contains many possible proteases and phosphorylation sites [30,31]. We found that the variation in Aβ concentration through repetitive measurement (using western blot and Luminex system) was significantly reduced by MPP treatment in a variety of settings including Aβ with pure sample diluent buffer (1× PBS or Bioplex buffer)  Table 2 Detailed information of the correlation between brain amyloid deposition (global PiB deposition (SUVR)) and MPP-Aβs SUVR standard uptake value ratio, MPP-Aβ MPP-treated plasma β-amyloid, CN cognitively normal, MCI mild cognitive impairment, ADD Alzheimer's disease dementia, ND nondemented group (CN plus MCI), CI cognitively impaired group (MCI plus ADD), PiB-or PiB+ subjects with negative or positive amyloid deposition, All complete study cohort *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 and Pearson r values by Pearson's correlation analysis (two-tailed) # P < 0.10, trend toward significance (Fig. 2a, c), with plasma mimic buffer using HSA (Fig. 2c) [43], and with diluent buffer containing the original plasma sample (Fig. 2b, d). The purpose of repetitively measuring experiments (Fig. 2a- whereas non-MPP-treated plasma Aβ42 (nMPP-Aβ42) levels decreased rapidly, and nMPP-Aβ40 levels fluctuated with incubation times (Fig. 2e). These results indicate that MPP might have a stabilizing effect on the quantification of plasma Aβs. Hence, we next differentiated PiB-(CN-, MCI-, and ADD-) vs PiB+ (CN+, MCI +, and ADD+) subjects using MPP-Aβs. Interestingly, although not all comparisons were statistically significant, most cases comparing PiB-vs PiB+ had the same patterns for MPP-Aβ40 level and MPP-Aβ42/40 ratio (MPP-Aβ40, CN-< CN+, MCI-< MCI+, and ADD-< ADD+; MPP-Aβ42/40, CN-> CN+, and ADD-> ADD+; Fig. 4a). Hence, we compared the subjects again as ND-vs ND+ and PiB-vs PiB+. The MPP-Aβ40 level was dramatically increased, and the MPP-Aβ42/40 ratio was significantly reduced in ND+ and PiB+ in comparison with ND-and PiB-subjects respectively (Fig. 4c, d). This indicates that MPP-Aβ might be a  Table 3.
PiB Pittsburgh compound B,or + PiB negativity or positivity, ND nondemented, ApoE apolipoprotein E, MPP-Aβ MPP-treated plasma β-amyloid, AUC area under curve ApoE Apolipoprotein E, AUC area under curve, MPP mixture of protease inhibitors and phosphatase inhibitors, CI confidence interval, ROC receiver operating characteristic,or + PiB-PET positivity, ND-PiB-PET-negative cognitively normal (CN) and PiB-PET-negative subjects with mild cognitive impairment (MCI), ND+ PiB-PET-positive CN and PiB-PET-positive MCI, PiB-grouped CN-, MCI-, and ADD-, PiB+ grouped CN+, MCI+, and ADD+ potential biomarker to predict Aβ deposition in the brain. Although the age of subjects who were PiB+ was relatively higher than that of PiB-subjects (Table 1), the significant difference of MPP-Aβs between PiB+ and PiB-subjects (MPP-Aβ40 increased and MPP-Aβ42/40 decreased in PiB+, compared with PiB-; Fig. 4c) was not due to the effects of aging because comparing young-middleaged controls (YC, cognitively normal subjects with age 20-55 years, n = 61) vs CN-subjects showed completely opposite trends to PiB-vs PiB+ subjects, regarding MPP-Aβs (see Additional files 5 and 6 for more detail). Even in cases where the MPP-Aβ42/40 ratio was dramatically different between PiB-and PiB+ subjects, it was insufficient to be used as a variable for ROC curve analysis to discriminate ND-vs ND+ or PiB-vs PiB+ subjects (ND-vs ND+, AUC 0.639 with 71.8% sensitivity and 51.5% specificity; PiB-vs PiB+, AUC 0.668 with 71.0% sensitivity and 52.2% specificity; Fig. 5 and Table 3). Using the combination of additional risk factors (gender, age, and ApoE types) reported previously [54,55], we conducted the logistic regression analysis followed by ROC curve analysis to establish the screening model for cerebral amyloid deposition. We showed that the combination of MPP-Aβ42/40 ratio, gender, and age had a stronger discrimination power (ND-vs ND+, AUC 0.695 with 79.0% sensitivity and 57.8% specificity; PiB-vs PiB+, AUC 0.682 with 71.0% sensitivity and 54.2% specificity; Fig. 5 and Table 3) than the MPP-Aβ42/40 level alone. Furthermore, we observed that AUC was increased further by ApoE variable (ND-vs ND+, AUC 0.783 with 80.7% sensitivity and 69.2% specificity; PiB-vs PiB+, AUC 0.799 with 78.0% sensitivity and 66.8% specificity; Fig. 5 and Table 3).
These results are in line with several previous reports that showed an interrelationship between plasma Aβ42/ 40 ratio and dementia [56,57]. However, this study is the first to stabilize the quantification of plasma Aβ by the treatment of chemicals and predict the PiB-PET positivity by classifying the subjects into two large groups, ND-vs ND+ subjects and PiB-vs PiB+ subjects, regardless of the stage of cognitive impairment. We showed a distinct relationship between MPP-Aβs and pathological AD-related Aβ burden in the brain. In conclusion, these results suggest that our model using MPP-Aβs and other factors could be utilized as a prescreening tool to predict cerebral Aβ deposition.