Plasma pyroglutamate-modified amyloid beta differentiates amyloid pathology.

INTRODUCTION: Pyroglutamate-modified amyloid β (AβpE3) could be a biomarker for Aβ plaque pathology in the brain. An ultra-high-sensitive assay is needed for detecting AβpE3-40.
METHODS: Immunomagnetic reduction was used for quantification of AβpE3-40 in plasma from 46 participants. The concentrations of AβpE3-40 of these subjects were compared with 18F-florbetapir positron emission tomography (PET) images.
RESULTS: AβpE3-40 concentration was 44.1 ± 28.2 fg/mL in PET- (n = 28) and 91.6 ± 54.6 fg/mL in PET+ (n = 18; P < .05). The cutoff value of AβpE3-40 for discriminating PET- from PET+ was 55.5 fg/mL, resulting in a sensitivity of 83.3%, a specificity of 71.4%. The concentration of AβpE3-40 showed a moderate correlation (r = 0.437) with PET standardized uptake value ratio. DISCUSSION: We did not enroll pre-clinical AD subject with normal cognition but Aβ PET+. It would be an important issue to explore the feasibility of using AβpE3-40 for screening pre-clinical subjects.
CONCLUSION: These results reveal the feasibility of detecting Aβ pathology using quantification of a plaque-derived Aβ molecule in plasma.

INTRODUCTION

Amyloid β (Aβ) aggregation in the brain is the pathological hallmark of Alzheimer's disease (AD). , , , Toxic effects of these Aβ aggregates are correlated with the predominance of N‐terminally truncated species over the full‐length Aβ. , , Using mass spectrometry, various types of N‐terminally truncated species of Aβn‐40/42 are found in AD brain tissue, including N‐terminally truncated Aβ3‐40/42 that have been further catalyzed by glutaminyl cyclase to form pyroglutamate Aβ cyclization (AβpE3) variants. , , The particular Aβ form has high toxicity, high resistance to proteolytic degradation, increased hydrophobicity, and faster aggregation. , , , Thus, AβpE3 may be an important culprit during AD initiation and progression.

AβpE3 is evidenced as a major constituent of intra‐/extracellular and vascular Aβ deposits in AD brain tissue. , , In addition to post mortem human brain tissues, the abnormal levels of AβpE3 in the brain and the co‐localization of AβpE3 with Aβ plaques were found in different animal models, such as transgenic mice, canines, and Caribbean vervets. , , These results suggest that AβpE3 is a potential seeding species and may play an important role in the formation of pathological Aβ aggregates in the brain. , It could also be a biomarker specific for Aβ plaque pathology in the brain.

So far, the reported evidence for finding abnormal amyloidosis by AβpE3 in AD is tissues of animals or post mortem human brains. The difficulties of obtaining human brain samples seriously limit the exploration of AβpE3 in clinical cohorts. It is believed that the measurement AβpE3 in body fluids such as plasma would be important to further explore the relevance of AβpE3 in AD pathogenesis, and plasma AβpE3 may also have a potential as a diagnostic tool in the clinic. However, the concentration of AβpE3 in human body fluid is extremely low. An ultra‐high‐sensitive assay technology is needed for detecting AβpE3 in human body fluids.

Immunomagnetic reduction (IMR) is an ultra‐sensitive technology for assaying biomarkers at pg/mL or lower. , In addition, the correlation between these plasmas biomarkers and their concentration in cerebrospinal fluid (CSF), and their relation to neuroimaging measures such as Aβ positron emission tomography (PET) have been clarified. , The results reveal the reliabilities of assaying ultra‐low‐concentrated biomarkers using IMR. In this work, IMR was used to develop the quantitative detection of AβpE3‐40 in human plasma. Moreover, 28 subjects with negative Aβ PET (PET‐) and 18 subjects with positive Aβ PET (PET+) were enrolled. The measured concentrations of plasma AβpE3‐40 of these subjects were compared with amyloid PET. Moreover, the plasma Aβ1‐40 of all subjects were assayed using an IMR Aβ1‐40 kit to explore the roles of Aβ1‐40 and AβpE3‐40 in discriminating Aβ PET status.

Research in Context

Pyroglutamate‐modified amyloid β (AβpE3) is a modified Aβ peptide that co‐oligomerizes with Aβ42 and deposited in the Alzheimer's disease (AD) brain. AβpE3 may act as a seed for misfolding of Aβ at a primary step in AD. The concentration of AβpE3 in body fluid is extremely low. Therefore, an ultra‐sensitive assay such as immunomagnetic reduction assay is developed to detect the level of AβpE3 in plasma.

We developed a new analysis method to measure the concentration of AβpE3. This study demonstrates the plasma AβpE3‐40 showed a correlation with Aβ positron emission tomography (PET) status and standardized uptake value ratio, which may be of value for screening and diagnosis as well as for applications in longitudinal clinical research studies and to monitor treatments in clinical trials.

Detection of plasma AβpE3‐40 in early stages could be a potential strategy for early diagnosis of AD. However, more participants should be enrolled for validating the correlation between AβpE3‐40 and Aβ PET status.

METHODS

IMR reagent for assaying AβpE3‐40

Antibodies against AβpE3‐40 were developed by Biogen Inc. According to the results via direct binding enzyme‐linked immunosorbent assay (ELISA), the antibodies show strong reactivity to AβpE3‐40 and Aβ3‐40 (>0.3 nM), and weak reactivity to Aβ1‐40, but not AβpE11‐40. There is not any available data concerning the reactivity with AβpE3‐42. Antibodies against AβpE3‐40 were covalently bound to dextran‐coated Fe3O4 nanoparticles (MF‐DEX‐0060, MagQu) via the chemical reactions in Yang et al.27

IMR measurements

The IMR analyzer (XacPro‐S, MagQu) was used to detect the reduction in the magnetic signals of reagent due to the association between antibody‐functionalized magnetic nanoparticles and AβpE3‐40 molecules. The ratio of the reduction to the alternative‐current magnetic signal of reagent before incubation is referred as IMR signal, as expressedwhere χ ac,o and χ ac,ϕ are the alternative‐current magnetic signals of reagent before and after incubation. For each reported IMR (%) in this work, an averaged value of duplicated IMR measurements was used.

Recruitment of subjects

All subjects were recruited in the study of the Alzheimer's Disease Neuroimaging Initiative at Taipei Veterans General Hospital (Taipei VGH), Linkou Chang Gung Memorial Hospital (CGMH), and Kaohsiung CGMH in Taiwan (T‐ADNI). The T‐ADNI study was approved by the ethics committees of the three hospitals. All participants were asked to complete written informed consents for this study.

Enrolled subjects were required to be aged >55 years. Magnetic resonance imaging (MRI) was used to exclude subjects with major neuropathologies such as tumors, strokes, severe white matter disease, or inflammation, but MRI was not used to diagnose dementia. Subjects with history of major brain trauma, stroke, brain tumor, epilepsy, major psychiatric illness, alcoholism, or other systemic diseases that might affect cognitive function were ruled out in this study.

The study included a battery of neuropsychological tests including the Geriatric Depression Scale, a Mini‐Mental State Examination (MMSE), and the Chinese version of the Wechsler Memory Scale‐III (WMS‐III). A Clinical Dementia Rating Scale (CDR) score was performed for each enrolled subject. Patients with AD and amnestic mild cognitive impairment (aMCI) patients follows criteria of National Institute on Aging and the Alzheimer's Association (NIA‐AA).

Collection and preparation of human plasma samples

Each enrolled subject provided a 9 mL non‐fasting venous blood sample (K3 ethylenediaminetetraacetic acid [EDTA], lavender‐top tube). Blood samples were centrifuged at 1500 to 2500 × g at room temperature for 15 minutes within 1 hour after blood draw. Plasma was then aliquoted into cryotubes (0.5 mL aliquots) and stored at ‐20°C.

Analysis of apolipoprotein E (APOE)genotypes

APOE genotyping was performed for each enrolled subject by polymerase chain reaction (PCR) amplification of a 500 base‐pair fragment of the APOE gene spanning the bases coding for amino acid positions 112 and 158, followed by direct DNA sequencing. Subjects with either one or two ε4 alleles were regarded as ε4 carriers.

Aβ PET imaging

All PET images were acquired from Linkou Chang Gung Memorial Hospital. The details of Aβ PET imaging were described in Lin et al. and Hsiao et al. , The 18F‐florbetapir PET scan comprised a 10‐minute acquisition period (acquired in 2 × 5minute frames) beginning 50 minutes following 10 mCi injection of the 18F‐florbetapir tracer. Imaging was performed on a Biograph mCT PET/CT scanner (GE Healthcare, Milwaukee, WI, USA). Structural MRI scans were acquired using a uniform scanning protocol that minimized and accounted for between‐site differences in MRI systems.

All PET image data were processed and analyzed using PMOD image analysis software (version 3.7, PMOD Technologies Ltd., Zurich, Switzerland), including MR‐based spatial normalization to the Montreal Neurological Institute (MNI) MRI template. Seven volumes of interest (VOIs), the frontal, anterior cingulate, posterior cingulate, precuneus, parietal, occipital, and temporal areas, were selected, and the regional standardized uptake value ratio (SUVR) using the whole cerebellum as the reference region was calculated for each VOI. Moreover, the average SUVR from these seven cerebral cortical VOIs was computed to yield an estimate of global cortical SUVR for further analysis.

The PET images were interpreted blindly by an experienced nuclear medicine physician (Kun‐Ju Lin). A five‐point visual scale was used to classify the amyloid loading, from 0, indicated no tracer retention in cortical gray matter, to 4, indicated high levels of cortical amyloid accumulation. Visual rating scores of 2 to 4 were considered indicative of amyloid PET+ brains and ratings of 0 to 1 were considered negative for amyloid PET.

Assay of plasma Aβ1‐40

IMR reagent for Aβ1‐40 (MF‐AB0‐0060, MagQu) with an aid of IMR analyzer (XacPro‐S, MagQu) were used to detect the concentrations of plasma Aβ1‐40 of subjects. Duplicated measurements were done for each sample. The reported concentration of the plasma Aβ1‐40 is the mean value of the duplicated measurements. The variations of measured concentrations of control solutions with respect to the known concentrations should be <15%.

Statistical methods

Continuous variables for each measurement are presented as means ± standard deviations. Continuous variables were compared using a t‐test, and P‐values were determined. Pearson correlation r was done with GraphPad Prism. Negative, positive, overall percentage agreements were calculated to quantify the consistency between plasma‐biomarker diagnosis and clinical diagnosis.

RESULTS

AβpE3‐40 concentration‐dependent IMR(%)

Phosphate buffered saline (PBS) solutions spiked with various concentrations of AβpE3‐40 from 0.1 fg/mL to 1000 pg/mL were assayed with IMR. For each concentration, duplicated IMR measurements were performed. The averaged IMR(%) of the duplicated measurement was used to establish the relationship between IMR(%) and AβpE3‐40 concentration. Figure 1A shows the relationship between IMR(%) and spiked AβpE3‐40 concentration. The error bars with each data point are attributed from the duplicated measurements. It was observed in Figure 1A that IMR(%) increased from 2.69% to 4.42% as spiked βpE3‐40 concentration increases from 0.1 fg/mL to 1000 pg/mL. The relationship in Figure 1A follows the logistic functionwhere A, B, ϕ o, and γ are fitting parameters; ϕ AβpE3‐40 denotes the spiked AβpE3‐40 concentration in PBS. By fitting Equation (2) to the experimental data points in Figure 1A, the values for the parameters A, B, ϕ o, and γ were found to be 2.69, 4.46, 1.67, and 0.628, respectively. A in Equation. (2) is the IMR(%) as the spiked AβpE3‐40 concentration approaches zero. Thus, A denotes the noise level for IMR(%). B in Equation. (2) is the IMR(%) as the spiked AβpE3‐40 concentration approaches infinity. Thus, B denotes the upper‐limit signal for IMR(%).

FIGURE 1

A, Amyloid β (AβpE3‐40) concentration‐dependent IMR signals and (B) relationship between spiked concentration of AβpE3‐40 (x axis) and measured concentration of AβpE3‐40 (y axis) using immunomagnetic reduction

Lower limit of detection of the AβpE3‐40 assay

The lower limit of detection is usually defined as the concentration showing the IMR(%) higher than the noise level by triple standard deviation for IMR signals at low concentrations. According to the results in Table 1, the standard deviation for IMR(%) at low concentrations, such as the spiked 0.001‐pg/mL AβpE3‐40, was found to be 0.06%. Thus, the lower limit of detection in terms of IMR(%) was (2.69 + 3 × 0.06)% = 2.88%. Via the logistic function in Equation (2), the lower limit of detection for assaying AβpE3‐40 using IMR was around 0.005 pg/mL, that is, 5 fg/mL.

Table 1

Interference tests of Aβ3‐40, Aβ1‐42, and Aβ1‐40 for assaying AβpE3‐40 using immunomagnetic reduction

		Measured Aβ pE3‐40 concentration
Sample No.	Intuition	Mean (pg/mL)	SD (pg/mL)	CV (%)	Recovery rate (%)
1	Plasma	0.0229	0.0061	26.46	‐
2	Plasma + AβpE3‐40 (0.05 pg/mL)	0.0741	0.0051	6.83	(Ref.)
3	Plasma + AβpE3‐40 (0.05 pg/mL) +Aβ3‐40 (20 pg/mL)	0.0781	0.0105	13.47	105.46
4	Plasma + AβpE3‐40 (0.05 pg/mL) +Aβ1‐42 (20 pg/mL)	0.0734	0.0064	8.72	99.06
5	Plasma + AβpE3‐40 (0.05 pg/mL) + Aβ1‐40 (80 pg/mL)	0.0776	0.0089	11.45	104.73
6	Plasma + AβpE3‐40 (0.05 pg/mL) + Aβ1‐40 (100 pg/mL)	0.0842	0.0023	2.76	113.66

Abbreviations: Aβ, amyloid β; CV, coefficient of variation; SD, standard deviation

Measurement range of the AβpE3‐40 assay

The measured IMR(%) in Figure 1A was converted to the measured AβpE3‐40 concentration via Equation (2). The correlations between measured AβpE3‐40 concentrations and spiked AβpE3‐40 concentrations were investigated. The results are shown in Figure 1B. Notably, one of the duplicated measurements of IMR(%) for 0.1 fg/mL AβpE3‐40 PBS sample was <2.69%, which was the noise level of IMR(%). The measured AβpE3‐40 concentration was not available for this IMR(%) <2.69%. Hence, the measured AβpE3‐40 concentration for the spiked 0.1 fg/mL AβpE3‐40 PBS sample was not counted in Figure 1B. Meanwhile, one of duplicated measurements of IMR(%) for 1000‐pg/mL AβpE3‐40 PBS sample was > 4.46%, which was the upper‐limit signal of IMR(%). The measured AβpE3‐40 concentration was not available for this IMR(%) >4.46%. Hence, the measured AβpE3‐40 concentration for the spiked 1000‐pg/mL AβpE3‐40 PBS sample was not counted in Figure 1B either. The measured AβpE3‐40 concentrations versus the spiked AβpE3‐40 concentrations from 1 fg/mL to 100 pg/mL was plotted in Figure 1B. The slope of the linearity in Figure 1B was found to be 1.07. According to the Clinical & Laboratory Standards Institute (CLSI) guideline EP06‐A2, the acceptable range of the slope is from 0.9 to 1.1 for demonstrating the linearity between the measured concentrations and spiked concentrations. Therefore, by taking the results in Figures 1A and 1B into account, the measurement range of AβpE3‐40 using IMR was from 5 fg/mL to 100 pg/mL.

Interference tests of assaying AβpE3‐40

Six human plasma samples were prepared for the interference tests, as tabulated in Table 1. Sample No. 1 was native human plasma. Sample No. 2 contained spiked AβpE3‐40 of 0.05 pg/mL. In addition to 0.05 pg/mL AβpE3‐40, Sample No. 3‐6 contained difference spiked Aβ3‐40 (AS‐61029, Anaspec), Aβ1‐42 (A9810, Sigma), or Aβ1‐40 (A1075, Sigma). It has been reported that the concentration of plasma Aβ1‐42 is around 10∼20 pg/mL, while plasma Aβ1‐40 is 30 to 60 pg/mL. , , , The spiked Aβ1‐42 and Aβ1‐40 concentrations in Samples No. 4 through 6 were thus reasonable. The measured AβpE3‐40 concentrations for each sample are listed in Table 1. The measured AβpE3‐40 concentration of Sample No. 2 was used as a reference. The recovery rates in the measured AβpE3‐40 concentrations for Samples No. 3 through 6 were calculated as the ratio of mean concentrations of duplicated measurements for each sample to that of Sample No. 2. The results are shown in Table 1. It was found that the recovery rates for Samples No. 3 through 5 were within the range from 90% to 110%, which means that there was no significant interference to the assay of AβpE3‐40 in Samples No. 3 through 5. However, the recovery rate of Sample No. 6 was > 110%, which revealed that the 100‐pg/mL Aβ1‐40 contributed significantly to false signal for assaying AβpE3‐40 in human plasma. Fortunately, the measured Aβ1‐40 concentrations in human plasma were < 100 pg/mL in both healthy controls and AD patients, that is, 30 to 60 pg/mL. , Hence, for real human plasma, there would be no significant interferences by Aβ3‐40, Aβ1‐42, or Aβ1‐40 to the assay of AβpE3‐40 using AβpE3‐40 IMR reagent.

Demographic characteristics of enrolled subjects

Forty‐six human plasma samples from Taiwan Alzheimer's Disease Neuroimaging Initiative (T‐ADNI) were assayed with AβpE3‐40 IMR reagent. The demographic characteristics, including sex, age, education, APOE ε4 status, CDR, MMSE, global SUVR, and measured plasma AβpE3‐40 and Aβ1‐40 concentrations, are shown in Table 2. In Table 2, the demographic characteristics of various diagnostic groups, that is, normal controls (NC), aMCI, and AD, in Aβ PET‐ and Aβ PET+ are also listed. The combined includes NC, aMCI, and AD. The typical Aβ PET images of enrolled subjects are shown in Figure 2. Figures 2A–2C are for Aβ PET‐ and Figures 2D and 2E are for Aβ PET+.

Table 2

Demographic information of enrolled subjects

Amyloid PET	Negative (Aβ PET‐)				Positive (Aβ PET+)
Dx(n)	NC(5)	aMCI(18)	AD(5)	Combined(28)	aMCI(7)	AD(11)	Combined(18)
Female/male	3/2	7/11	1/4	11/17	7/0	4/7	11/7
Age (years)	60.4 ± 2.8	71.2 ± 9.8	75.6 ± 11.5	70.1 ± 10.3	72.7 ± 5.9	71.2 ± 9.2	71.8 ± 7.92
Education (years)	14.2 ± 2.9	10.8 ± 3.6	13.0 ± 3.0	11.8 ± 3.6	12.7 ± 5.6	13.5 ± 3.1	13.2 ± 4.1
ApoEε4 allele frequency	30%	2.78%	0%	7.69%	7.14%	36.36%	25%
CDR	0	0.5	0.6 ± 0.2	0.43 ± 0.22	0.5	0.5	0.50 ± 0.00
MMSE	29.6 ± 0.89	27.28 ± 2.03	24.0 ± 3.54	27.11 ± 2.74	26.14 ± 2.41	22.82 ± 2.04	24.11 ± 2.70a
Global SUVR	1.08 ± 0.06	1.05 ± 0.20	1.11 ± 0.15	1.06 ± 0.17	1.57 ± 0.18	1.50 ± 0.17	1.53 ± 0.17a
Plasma AβpE3‐40 (fg/mL)	31.00 ± 16.88	44.59 ± 30.28	55.41 ± 28.4	44.09 ± 28.19	65.64 ± 18.56	108.2 ± 63.9	91.62 ± 54.60a
Plasma Aβ1‐40 (pg/mL)	44.65 ± 9.64	49.39 ± 7.18	49.53 ± 8.15	48.57 ± 7.71	54.28 ± 6.17	50.28 ± 6.91	51.84 ± 6.75
Plasma AβpE3‐40‐to‐Aβ1‐40 ratio	(0.072 ± 0.048)%	(0.092 ± 0.063)%	(0.108 ± 0.041)%	(0.091 ± 0.057)%	(0.122 ± 0.040)%	(0.215 ± 0.122)%	(0.179 ± 0.107)%a

Abbreviations: Aβ, amyloid β; AD, Alzheimer's disease; aMCI, amnestic mild cognitive impairment; CDR, clinical dementia ranking; Dx, clinical diagnosis; MMSE, mini‐mental state examination; NC, normal controls; PET, positron emission tomography; SUVR, standardized uptake value ratio.

* P < .05 between Aβ PET‐ and Aβ PET+ **: P < .001 between Aβ PET‐ and Aβ PET+

FIGURE 2

Typical amyloid β (Aβ) positron emission tomography (PET) images of the enrolled subjects with (A) normal controls (NC) and Aβ PET‐, (B) amnestic mild cognitive impairment (aMCI) and Aβ PET‐, (C) Alzheimer's disease (AD) and Aβ PET‐, (D) aMCI and Aβ PET+, and (E) AD and Aβ PET+

The comparisons in the demographic characteristics between Aβ PET‐ and Aβ PET+ individuals, that is, demographic characteristics in combined columns, were made. The age, education years, and CDR between Aβ PET‐ and Aβ PET+ individuals were matched. There was no significant difference in CDR between Aβ PET‐ and Aβ PET+ individuals. The frequency of APOE ε4 allele is much higher in Aβ PET+ (25%) as compared to Aβ PET‐ (7.69%) individuals. Meanwhile, significantly higher scores of MMSE (P < .001), higher values of global SUVR (P < .001), and higher levels of measured plasma AβpE3‐40 (P < .05) were found in Aβ PET+ individuals. As to plasma Aβ1‐40 concentrations, there is no significant difference between Aβ PET‐ and Aβ PET+ individuals.

Plasma AβpE3‐40 for discriminating Aβ PET status

The measured plasma AβpE3‐40 concentrations in Aβ PET‐ and in Aβ PET+ individuals are plotted in Figure 3A. The error bar of each data point in Figure 3A is attributed from the duplicated measurements of plasma AβpE3‐40 concentrations. Concentrations of plasma AβpE3‐40 was 44.1 ± 28.2 fg/mL in Aβ PET‐ subjects, as compared with 91.6 ± 54.6 fg/mL in the Aβ PET+ group (P = .012). The analysis of the receiver operating characteristic (ROC) curve was performed for the data shown in Figure 3A. The ROC curve is shown Figure 3B. The cutoff value of plasma AβpE3‐40 concentration for discriminating Aβ PET‐ from Aβ PET+ individuals was 56.3 fg/mL, as plotted with the gray dashed line in Figure 3A. The corresponding clinical sensitivity and specificity was 79.0% and 71.4%, respectively. The area under the curve (AUC) was 0.808.

FIGURE 3

(A) Measured concentrations of plasma amyloid β (AβpE3‐40) in Aβ‐positron emission tomography (PET)‐ and Aβ‐PET+ subjects using immunomagnetic reduction and (B) receiver operating characteristic (ROC) curve for differentiating PET‐ from PET+, (C) correlation between measured concentrations of plasma AβpE3‐40 and global standardized uptake value ratio

Correlation between plasma AβpE3‐40 and SUVR

The relationship between measured plasma AβpE3‐40 concentration and global SUVR is shown in Figure 3C. Through Pearson correlation analysis, the correlation coefficient r was found to be 0.450 (P < .05), as guided with the gray dashed line in Figure 3C.

DISCUSSION

Although Aβ PET is approved for diagnosing AD in clinics, it is a very costly and not that accessible of an examination. It would be better to have a screening tool for evaluating the requirement of performing Aβ‐PET examination, such as a blood test. Many research groups have tried to develop methods to quantify plasma Aβ in a manner that correlates with Aβ PET. Table 3 lists some typical results showing the possibilities to discriminate Aβ PET‐ from PET+ using human plasma Aβ species, with Aβ1‐40 and Aβ1‐42 being the core biomarkers in these studies. Depending on the clinical diagnosis of enrolled subjects, assay methods and plasma biomarkers, the AUC of discriminating Aβ PET‐ and PET+ ranges from 0.66 to 0.969. However, studies on the role of plasma AβpE3‐40 in differentiating Aβ PET status are currently lacking. To our knowledge, this is the first study to report the feasibility of discriminating Aβ PET‐ and PET+ by using plasma AβpE3‐40 in humans. The AUC of 0.808 suggests that AβpE3‐40 in plasma is a promising test for detecting Aβ PET+, which may be of value for screening and diagnosis as well as for applications in longitudinal clinical research studies and to monitor treatments in clinical trials.

Table 3

Reported plasma Aβ biomarkers for differentiating Aβ PET‐ from PET+ in AD

					Aβ PET‐ vs. Aβ PET+				P
References	Dx	N	Assay	Plasma biomarker	Cutoff value	Sensitivity	Specificity	AUC
37	NC	76	Sandwich ELISA	Aβ1‐42/Aβ1‐40	—	71%	78%	0.79	‐
38	SCD	200	Sandwich ELISA	Aβ1‐42/Aβ1‐40	0.08	83.3%	51.9%	0.68	−
39	NC + MCI	39	Sandwich ELISA	Aβ1‐40	—	−	−	−	.04
				Aβ1‐42/Aβ1‐40	—	−	−	−	.02
40	SCD	69	SIMOA	Aβ1‐40	—	−	−	0.66	.018
				Aβ1‐42/Aβ1‐40	—	70%	78%	0.79	−
41	NC + MCI +AD	66	IP‐MS	Aβ1‐42	0.183 pg/mL	0.825	0.773	0.808	−
				Aβ1‐42/Aβ1‐40	0.009	0.750	0.773	0.798	−
				APP669‐711/Aβ1‐42	0.914	0.925	0.955	0.969	−
30	NC + MCI +AD	45	IMR	Aβ1‐42/Aβ1‐40	−	−	−	−	< .001
This work	NC + aMCI +AD	46	IMR	AβpE3‐40	55.45 fg/mL	83.3%	71.4%	0.808	.0012

Abbreviations: Aβ, amyloid β; AD, Alzheimer's disease; aMCI, amnestic mild cognitive impairment; AUC, area under curve; ELISA, enzyme‐linked immunosorbent assay; IMR, immunomagnetic reduction; IP‐MS, immunoprecipitation mass spectrometry; MCI, mild cognitive impairment impairment; NC, normal controls; PET, positron emission tomography; SCD, subjective cognition decline; SIMOA, single molecule array

The dependence of plasma AβpE3‐40 on APOE ε4 genotype was investigated. There was no significant difference (P > .05) in the levels of plasma AβpE3‐40 between APOE ε4 non‐carriers (56.58 ± 44.24 fg/mL) and carriers (80.00 ± 49.92 fg/mL), which resonates well with the fact that AβpE3‐40 is generated from Aβ1‐40 via truncation and pyroglutamation; a process not directly related to APOE genotype. There was no significant difference in levels of plasma Aβ1‐40 between Aβ PET‐ and PET+ individuals, but the levels of plasma AβpE3‐40 in Aβ PET+ subjects were higher, suggesting that AβpE3‐40 is more crucial than Aβ1‐40 to the formation of Aβ plaques in the brain. The result is corroborated by the plasma AβpE3‐40‐to‐Aβ1‐40 ratio, which was significantly higher (P < .05) in Aβ PET+ (0.179% ± 0.107%) as compared to Aβ PET‐ individuals (0.091% ± 0.057%).

The roles of plasma AβpE3‐40 in determining cognitive‐disorder severity are investigated. As listed in Table 2, for Aβ PET‐ individuals, MMSE significantly decreases from NC to aMCI and AD (P < .001). An obviously different degree of severity in cognitive disorder among diagnostic groups was evidenced in Aβ PET‐ individuals. Although the mean value of plasma AβpE3‐40 concentrations increases from NC (31.00 fg/mL), aMCI (44.59 fg/mL) to AD (55.41 fg/mL) for Aβ PET‐ individuals, there is no significant difference among these diagnostic groups. This might be due to the limited range for plasma AβpE3‐40 concentrations in Aβ PET‐ individuals (44.09 ± 28.19 fg/mL). However, for Aβ PET+ individuals, the plasma AβpE3‐40 concentrations distribute much more heterogeneously (91.62 ± 54.60 fg/mL) as compared to that for Aβ PET‐ individuals, as shown in Figure 3A. It might be possible to find the significant difference in the plasma AβpE3‐40 concentrations between aMCI and AD in Aβ PET+ individuals. As tabulated in Table 2, AD with Aβ PET+ shows significantly higher levels of plasma AβpE3‐40 (108.2 ± 63.9 fg/mL) than that of aMCI (65.64 ± 18.56 fg/mL, P < .05) with Aβ PET+. These results reveal that plasma AβpE3‐40 level is not only promising to discriminate Aβ PET status, but also able to determine the severity of cognitive disorder in Aβ PET+ individuals.

As expected, the concentration of AβpE3‐40 in human plasma was very low. With the development of ultra‐sensitive assays like IMR, it becomes feasible to precisely detect such low concentrations of biomarkers in human plasma. More investigations in plasma AβpE3‐40 shall be explored using ultra‐sensitive assays in the future.

There are some limitations in this work. For example, the total number of enrolled subjects is relatively limited. More subjects should be enrolled for validating the cutoff value of plasma AβpE3‐40 to discriminate Aβ PET status in future studies. Moreover, we did not enroll any pre‐clinical AD subject, that is, subjects with normal cognition but Aβ PET+. It would be an important issue to explore the feasibility of using plasma AβpE3‐40 for screening pre‐clinical subjects.

CONCLUSION

Reagent for assaying plasma AβpE3‐40 by using immunomagnetic reduction were developed. The measurement range of assaying AβpE3‐40 was 5 fg/mL to 100 pg/mL. The levels of plasma AβpE3‐40 were found to be able to discriminate Aβ PET status. The cutoff value of plasma AβpE3‐40 for discriminating PET‐ from PET+ was 55.5 fg/mL, showing the sensitivity of 83.3%, specificity of 71.4% ,and area under curve of 0.808. Moreover, plasma AβpE3‐40 level is promising to determine the severity of cognitive disorder in Aβ PET+ individuals. As compared to the native primary structure of Aβ1‐40, the pyroglutamate modification was more closely related to Aβ pathology in the brain. It was also found that AβpE3‐40 is independent of ApoE genotype.

CONFLICTS OF INTEREST

Huei‐Chun Liu is an employee of MagQu Co., Ltd. Shieh‐Yueh Yang is an employee of MagQu Co., Ltd. and MagQu LLC. Shieh‐Yueh Yang is a share owner of MagQu Co., Ltd. Kaj Blennow has served as a consultant, at advisory boards, or given lectures and chairing symposia for Alector, Biogen, CogRx, Lilly, MagQu, Novartis, and Roche Diagnostics, and is a co‐founder of Brain Biomarker Solutions in Gothenburg AB, a GU Venture‐based platform company at the University of Gothenburg, all unrelated to the work presented in this paper.