Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

High performance plasma amyloid-β biomarkers for Alzheimer’s disease

Abstract

To facilitate clinical trials of disease-modifying therapies for Alzheimer’s disease, which are expected to be most efficacious at the earliest and mildest stages of the disease1,2, supportive biomarker information is necessary. The only validated methods for identifying amyloid-β deposition in the brain—the earliest pathological signature of Alzheimer’s disease—are amyloid-β positron-emission tomography (PET) imaging or measurement of amyloid-β in cerebrospinal fluid. Therefore, a minimally invasive, cost-effective blood-based biomarker is desirable3,4. Despite much effort3,4,5,6,7, to our knowledge, no study has validated the clinical utility of blood-based amyloid-β markers. Here we demonstrate the measurement of high-performance plasma amyloid-β biomarkers by immunoprecipitation coupled with mass spectrometry. The ability of amyloid-β precursor protein (APP)669–711/amyloid-β (Aβ)1–42 and Aβ1–40/Aβ1–42 ratios, and their composites, to predict individual brain amyloid-β-positive or -negative status was determined by amyloid-β-PET imaging and tested using two independent data sets: a discovery data set (Japan, n = 121) and a validation data set (Australia, n = 252 including 111 individuals diagnosed using 11C-labelled Pittsburgh compound-B (PIB)-PET and 141 using other ligands). Both data sets included cognitively normal individuals, individuals with mild cognitive impairment and individuals with Alzheimer’s disease. All test biomarkers showed high performance when predicting brain amyloid-β burden. In particular, the composite biomarker showed very high areas under the receiver operating characteristic curves (AUCs) in both data sets (discovery, 96.7%, n = 121 and validation, 94.1%, n = 111) with an accuracy approximately equal to 90% when using PIB-PET as a standard of truth. Furthermore, test biomarkers were correlated with amyloid-β-PET burden and levels of Aβ1–42 in cerebrospinal fluid. These results demonstrate the potential clinical utility of plasma biomarkers in predicting brain amyloid-β burden at an individual level. These plasma biomarkers also have cost–benefit and scalability advantages over current techniques, potentially enabling broader clinical access and efficient population screening.

This is a preview of subscription content

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Figure 1: The peptide and biomarker values in each study site.
Figure 2: High performance of the plasma biomarkers.
Figure 3: Plasma biomarkers are significantly correlated with brain Aβ burden and CSF-Aβ1–42 level.

References

  1. 1

    Sevigny, J. et al. The antibody aducanumab reduces Aβ plaques in Alzheimer’s disease. Nature 537, 50–56 (2016)

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  2. 2

    Sperling, R., Mormino, E. & Johnson, K. The evolution of preclinical Alzheimer’s disease: implications for prevention trials. Neuron 84, 608–622 (2014)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  3. 3

    Henriksen, K. et al. The future of blood-based biomarkers for Alzheimer’s disease. Alzheimers Dement. 10, 115–131 (2014)

    PubMed  Article  PubMed Central  Google Scholar 

  4. 4

    O’Bryant, S. E. et al. Blood-based biomarkers in Alzheimer disease: Current state of the science and a novel collaborative paradigm for advancing from discovery to clinic. Alzheimers Dement. 13, 45–58 (2017)

    PubMed  Article  PubMed Central  Google Scholar 

  5. 5

    Rembach, A. et al. Changes in plasma amyloid β in a longitudinal study of aging and Alzheimer’s disease. Alzheimers Dement. 10, 53–61 (2014)

    PubMed  Article  PubMed Central  Google Scholar 

  6. 6

    Swaminathan, S. et al. Association of plasma and cortical amyloid beta is modulated by APOE ε4 status. Alzheimers Dement. 10, e9–e18 (2014)

    PubMed  Article  PubMed Central  Google Scholar 

  7. 7

    Lövheim, H. et al. Plasma concentrations of free amyloid β cannot predict the development of Alzheimer’s disease. Alzheimers Dement. 13, 778–782 (2017)

    PubMed  Article  PubMed Central  Google Scholar 

  8. 8

    Kaneko, N. et al. Novel plasma biomarker surrogating cerebral amyloid deposition. Proc. Jpn. Acad., Ser. B, Phys. Biol. Sci. 90, 353–364 (2014)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  9. 9

    Ovod, V. et al. Amyloid β concentrations and stable isotope labeling kinetics of human plasma specific to central nervous system amyloidosis. Alzheimers Dement. 13, 841–849 (2017)

    PubMed  PubMed Central  Article  Google Scholar 

  10. 10

    Gelfanova, V. et al. Quantitative analysis of amyloid-β peptides in cerebrospinal fluid using immunoprecipitation and MALDI-Tof mass spectrometry. Brief. Funct. Genomic. Proteomic. 6, 149–158 (2007)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  11. 11

    Kaneko, N., Yamamoto, R., Sato, T. A. & Tanaka, K. Identification and quantification of amyloid b-related peptides in human plasma using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Proc. Jpn. Acad., Ser. B, Phys. Biol. Sci. 90, 104–117 (2014)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  12. 12

    Ellis, K. A. et al. The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int. Psychogeriatr. 21, 672–687 (2009)

    PubMed  Article  PubMed Central  Google Scholar 

  13. 13

    Vandenberghe, R. et al. 18F-flutemetamol amyloid imaging in Alzheimer disease and mild cognitive impairment: a phase 2 trial. Ann. Neurol. 68, 319–329 (2010)

    PubMed  Article  PubMed Central  Google Scholar 

  14. 14

    Wong, D. F. et al. In vivo imaging of amyloid deposition in Alzheimer disease using the radioligand 18F-AV-45 (florbetapir F 18). J. Nucl. Med. 51, 913–920 (2010)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. 15

    Landau, S. M. et al. Amyloid PET imaging in Alzheimer’s disease: a comparison of three radiotracers. Eur. J. Nucl. Med. Mol. Imaging 41, 1398–1407 (2014)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  16. 16

    Mormino, E. C. et al. Amyloid and APOE ε4 interact to influence short-term decline in preclinical Alzheimer disease. Neurology 82, 1760–1767 (2014)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  17. 17

    Villemagne, V. L. et al. En attendant centiloid. Adv. Res. 2, 723–729 (2014)

    Article  Google Scholar 

  18. 18

    Fagan, A. M. et al. Comparison of analytical platforms for cerebrospinal fluid measures of β-amyloid 1-42, total tau, and p-tau181 for identifying Alzheimer disease amyloid plaque pathology. Arch. Neurol. 68, 1137–1144 (2011)

    PubMed  PubMed Central  Article  Google Scholar 

  19. 19

    Irwin, D. J. et al. Comparison of cerebrospinal fluid levels of tau and Aβ 1-42 in Alzheimer disease and frontotemporal degeneration using 2 analytical platforms. Arch. Neurol. 69, 1018–1025 (2012)

    PubMed  PubMed Central  Article  Google Scholar 

  20. 20

    Jagust, W. J. et al. Relationships between biomarkers in aging and dementia. Neurology 73, 1193–1199 (2009)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  21. 21

    Li, Q. X. et al. Alzheimer’s disease normative cerebrospinal fluid biomarkers validated in PET amyloid-β characterized subjects from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study. J. Alzheimers Dis. 48, 175–187 (2015)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  22. 22

    Shaw, L. M. et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann. Neurol. 65, 403–413 (2009)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  23. 23

    Wang, J., Gu, B. J., Masters, C. L. & Wang, Y. J. A systemic view of Alzheimer disease - insights from amyloid-β metabolism beyond the brain. Nat. Rev. Neurol. 13, 612–623 (2017)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  24. 24

    Janelidze, S. et al. Plasma β-amyloid in Alzheimer’s disease and vascular disease. Sci. Rep. 6, 26801 (2016)

    ADS  CAS  PubMed  PubMed Central  Article  Google Scholar 

  25. 25

    Rembach, A. et al. Plasma amyloid-β levels are significantly associated with a transition toward Alzheimer’s disease as measured by cognitive decline and change in neocortical amyloid burden. J. Alzheimers Dis. 40, 95–104 (2014)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  26. 26

    Jarrett, J. T., Berger, E. P. & Lansbury, P. T., Jr. The C-terminus of the β protein is critical in amyloidogenesis. Ann. NY Acad. Sci. 695, 144–148 (1993)

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  27. 27

    Rogers, M. B. Are CSF Assays Finally Ready for Prime Time? Alzforumhttps://www.alzforum.org/news/conference-coverage/are-csf-assays-finally-ready-prime-time (2017)

  28. 28

    Boccardi, M. et al. Assessment of the incremental diagnostic value of florbetapir F 18 imaging in patients with cognitive impairment: the incremental diagnostic value of amyloid PET with [18F]-florbetapir (INDIA-FBP) Study. JAMA Neurol. 73, 1417–1424 (2016)

    PubMed  Article  PubMed Central  Google Scholar 

  29. 29

    Caselli, R. J. & Woodruff, B. K. Clinical impact of amyloid positron emission tomography—is it worth the cost? JAMA Neurol. 73, 1396–1398 (2016)

    PubMed  Article  PubMed Central  Google Scholar 

  30. 30

    O’Bryant, S. E. et al. Guidelines for the standardization of preanalytic variables for blood-based biomarker studies in Alzheimer’s disease research. Alzheimers Dement. 11, 549–560 (2015)

    PubMed  Article  PubMed Central  Google Scholar 

  31. 31

    Albert, M. S. et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7, 270–279 (2011)

    PubMed  PubMed Central  Article  Google Scholar 

  32. 32

    McKhann, G. M. et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging–Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7, 263–269 (2011)

    PubMed  PubMed Central  Article  Google Scholar 

  33. 33

    Rowe, C. C. et al. Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging. Neurobiol. Aging 31, 1275–1283 (2010)

    PubMed  Article  PubMed Central  Google Scholar 

  34. 34

    Bourgeat, P. et al. Comparison of MR-less PiB SUVR quantification methods. Neurobiol. Aging 36, S159–S166 (2015)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  35. 35

    Clark, C. M. et al. Use of florbetapir-PET for imaging β-amyloid pathology. J. Am. Med. Assoc. 305, 275–283 (2011)

    ADS  CAS  Article  Google Scholar 

  36. 36

    Lundqvist, R. et al. Implementation and validation of an adaptive template registration method for 18F-flutemetamol imaging data. J. Nucl. Med. 54, 1472–1478 (2013)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  37. 37

    Pannee, J. et al. A selected reaction monitoring (SRM)-based method for absolute quantification of Aβ38, Aβ40, and Aβ42 in cerebrospinal fluid of Alzheimer’s disease patients and healthy controls. J. Alzheimers Dis. 33, 1021–1032 (2013)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  38. 38

    Patterson, B. W. et al. Age and amyloid effects on human central nervous system amyloid-beta kinetics. Ann. Neurol. 78, 439–453 (2015)

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  39. 39

    Manzoni, C. et al. Overcoming synthetic Aβ peptide aging: a new approach to an age-old problem. Amyloid 16, 71–80 (2009)

    CAS  Article  Google Scholar 

  40. 40

    Schlenzig, D. et al. N-Terminal pyroglutamate formation of Aβ38 and Aβ40 enforces oligomer formation and potency to disrupt hippocampal long-term potentiation. J. Neurochem. 121, 774–784 (2012)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  41. 41

    Toombs, J., Paterson, R. W., Schott, J. M. & Zetterberg, H. Amyloid-beta 42 adsorption following serial tube transfer. Alzheimers Res. Ther. 6, 5 (2014)

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  42. 42

    Tanaka, S. et al. Mass++: A visualization and analysis tool for mass spectrometry. J. Proteome Res. 13, 3846–3853 (2014)

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  43. 43

    Mattsson, N. et al. CSF biomarker variability in the Alzheimer’s Association quality control program. Alzheimers Dement. 9, 251–261 (2013)

    PubMed  PubMed Central  Article  Google Scholar 

  44. 44

    Shapiro, S. S. & Wilk, M. B. An analysis of variance test for normality (complete samples). Biometrika 52, 591–611 (1965)

    MathSciNet  MATH  Article  Google Scholar 

  45. 45

    Youden, W. J. Index for rating diagnostic tests. Cancer 3, 32–35 (1950)

    CAS  Article  Google Scholar 

  46. 46

    DeLong, E. R., DeLong, D. M. & Clarke-Pearson, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44, 837–845 (1988)

    CAS  MATH  Article  Google Scholar 

  47. 47

    Pencina, M.J., D’Agostino, R.B. Sr., D’Agostino, R.B. Jr. & Vasan, R.S. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat. Med. 27, 157–172 (2008)

    MathSciNet  PubMed  Article  PubMed Central  Google Scholar 

Download references

Acknowledgements

The NCGG study group thank all participants of the study, clinicians who referred patients, and all the staff who supported the MULNIAD project. We thank N. Sugimoto for conducting statistical analyses, S. Niida and the NCGG Biobank members for the management of plasma samples and M. Kawakage for data monitoring. This study was supported by The Research Funding for Longevity Sciences (25-24 and 26-30) from the National Center for Geriatrics and Gerontology, and partially supported by Research and Development Grants for Dementia from the Japan Agency for Medical Research and Development, AMED. This study is registered under UMIN ID: 000016144. The AIBL study would like to thank all participants of the study and the clinicians who referred participants. The AIBL study (https://aibl.csiro.au/) is a consortium between Austin Health, CSIRO, Edith Cowan University and the Florey Institute, The University of Melbourne. Partial financial support was provided by the Alzheimer’s Association (US), the Alzheimer’s Drug Discovery Foundation, an anonymous foundation, the Cooperative Research Centre for Mental Health, CSIRO Science and Industry Endowment Fund, the Dementia Collaborative Research Centres, the Victorian Government Operational Infrastructure Support program, the McCusker Alzheimer’s Research Foundation, the National Health and Medical Research Council, and the Yulgilbar Foundation. Funding sources had no role in study design, data collection, data analyses or data interpretation.

Author information

Affiliations

Authors

Contributions

A.N., N.K., T.K., K.It., K.T., and K.Y. designed the study. A.N., N.K., V.L.V., C.L.M., and K.Y. wrote the manuscript and A.N., N.K., V.L.V., V.D., T.T., and K.M. made the figures. A.N., V.L.V., T.K., V.D., C.F., Q.-X.L., R.M., C.R., Ke.Is., Ka.Is., Y.A., and C.L.M. collected the data. A.N., N.K., V.L.V., T.K., J.D., V.D., C.F., Q.-X.K., R.M., C.R., Y.A., T.T., K.M., S.I., K.It., K.T., and C.L.M. analysed the data. All authors interpreted the data and critically revised the manuscript.

Corresponding author

Correspondence to Katsuhiko Yanagisawa.

Ethics declarations

Competing interests

N.K., S.I. and K.T. are employees of Shimadzu. NCGG and Shimadzu have patents pending that are related to this study: PCT/JP2016/076706 and PCT/JP2015/064386. Shimadzu has patents pending: JP 2015-140899 and US 15/209331. V.L.V. received grants from CSIRO and NHMRC during the study. J.D. received grants from CRC for Mental Health, during the study, and has patent TW8546/AU pending.

Additional information

Reviewer Information Nature thanks H. Federoff, R. Thomas and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Figure 1 The amino acid sequences of Aβ-related peptides and results of the pilot study.

a, Overview of the amino acid sequences of the Aβ-related peptides Aβ1–40, Aβ1–42 and APP669–711. b, ROC analyses of the blinded pilot study for 20 Aβ+ and 20 Aβ subjects (see Supplementary Information). The green, blue, and red curves indicate APP669–711/Aβ1–42, Aβ1–40/Aβ1–42 and the composite biomarker, respectively. The AUCs and the representative best values of sensitivity, specificity and accuracy for these biomarkers as determined by Youden’s index are as follows: APP669–711/Aβ1–42, AUC = 0.923, sensitivity = 0.850, specificity = 0.950, accuracy = 0.900; Aβ1–40/Aβ1–42, AUC = 0.930, sensitivity = 0.900, specificity = 0.900, accuracy = 0.900; composite biomarker, AUC = 0.975, sensitivity = 0.950, specificity = 0.950, accuracy = 0.950.

Source data

Extended Data Figure 2 Peptide and biomarker values, and their distributions.

a, The Aβ-related peptide and biomarker values in each study site (NCGG and AIBL). Normalized intensity of each peptide (top), and values for each biomarker (bottom) in the NCGG (n = 121) and AIBL overall (n = 252) data sets. Composite biomarker values are the average of the normalized values of APP669–711/Aβ1–42 and Aβ1–40/Aβ1–42. Peptide values are arbitrary units. Means and 95% confidence intervals (CI) in parentheses; P values show statistical differences between the Aβ+ and Aβ groups (two-sided Student’s t-test or Welch’s t-test). Superscripts indicate statistically significant site differences (aP = 0.012, bP = 0.002, cP = <0.0001, two-sided). These site differences did not change when using analysis of covariance (ANCOVA) adjusted for semi-quantitative measures of Aβ-PET, using SUVR (PIB) and BeCKeT (FLUTE and FBP) values. b, Histograms of the biomarker value distributions for APP669–711/Aβ1–42, Aβ1–40/Aβ1–42 (top), and their inversions (Aβ1–42/APP669–711 and Aβ1–42/Aβ1–40) (bottom). Blue and red bars represent the distributions of Aβ and Aβ+ populations, respectively. P values represent the results of the Shapiro–Wilk test (see Methods).

Source data

Extended Data Figure 3 Adjusted ROC analyses corresponding to Fig. 2.

a, ROC analyses for each biomarker when predicting individual Aβ+/Aβ status for the discovery, validation, and combined data sets. Adjusted (age, gender, APOE4, and clinical category) analyses for the NCGG PIB discovery data (left), AIBL PIB validation data (middle), and AIBL overall (all tracers) validation data (right). See Extended Data Table 1a for detailed performance values. Data are from 121, 111 and 252 individuals for the NCGG PIB, AIBL PIB and AIBL overall data, respectively. b, Comparisons of biomarker performances within each analysis corresponding to the ROC curves in a. Each colour bar represents the AUC and 95% confidence interval. Statistically significant differences between two AUCs (DeLong test) and significant increments in predictive ability as assessed by NRI and IDI are indicated as in Fig. 2. All P values are two-sided, and Bonferroni corrected (multiplied by the number of comparisons, 6). Note that the NRI in the comparison between Aβ1–40/Aβ1–42 and the composite biomarker in NCGG data was negative (NRI = −0.382) indicating that the reclassification ability is lower in the composite biomarker. c, Adjusted (age, gender, APOE4 and clinical category) ROC analyses of the composite biomarker compared by different PET tracers; PIB (NCGG, n = 121, and AIBL, n = 111), flutemetamol (AIBL, n = 81), and florbetapir (AIBL, n = 60). See Extended Data Table 1b for detailed performance values. d, e, Adjusted (age, gender, APOE4) ROC curves of the composite biomarker within the AD and MCI (d) and cognitively normal (e) groups. Sample sizes are the same as those listed in Fig. 2d, e; see Extended Data Table 1c for detailed performance values.

Source data

Extended Data Figure 4 Correlations between plasma biomarkers and brain Aβ burden: additional data related to Fig. 3a.

ac, Biomarker values plotted against SUVR values from PIB-PET imaging; Aβ1-42 (a), APP669-711/Aβ1-42 (b) and Aβ1–42/Aβ1–40 (c). Data are from 121 (NCGG PIB, top) and 111 (AIBL PIB, bottom) individuals. Colours represent the clinical categories: AD, red; MCI, orange; cognitively normal, blue. The vertical dashed lines represent cut-off values of PIB-PET imaging (1.4), and horizontal dashed lines represent the common cut-off values of the plasma biomarkers estimated in Extended Data Fig. 7. d, A summary table for the correlation analyses. The sample sizes for each data set are; NCGG PIB, n = 121; AIBL PIB, n = 111; NCGG + AIBL PIB, n = 232; AIBL FLUTE, n = 81; AIBL FBP, n = 60; and NCGG + AIBL overall, n = 373. Pearson’s correlation coefficients (r) and their significance (two-sided P) are presented in the plots and the table.

Source data

Extended Data Figure 5 Reliability of the IP–MS methods.

a, Standard curves of Aβ1–42 (left), Aβ1–40 (middle), and APP669–711 (right) in PBS containing BSA. The standard curves were generated over a 2.5–40 pM range for Aβ1–42 and APP669–711, and a 10–160 pM range for Aβ1–40. The linearity was evaluated with the coefficient of determination (R2). The error bars indicate the standard deviations of normalized intensities obtained from four mass spectra. The normalized intensities (AU) and signal-to-noise ratios at the lowest concentration were 0.119 AU and 10.9 for Aβ1–42, 0.152 AU and 16.1 for APP669–711 and 1.56 AU and 165 for Aβ1-40, respectively. The lower limit of quantification referred to the lowest concentration at which Aβ1–42, APP669–711 and Aβ1–40 showed a signal-to-noise ratio greater than 10. Data reproducibility was confirmed by two additional experiments. b, Relationships between plasma dilution and normalized intensity of endogenous Aβ1–42, Aβ1–40, and APP669–711, which were contained in the human plasma. Normalized intensity indicates the mass spectrometry signal normalized with the signal for SIL-Aβ1–38. The linearity was evaluated with R2. The error bars indicate the standard deviations of normalized intensities obtained from four mass spectra. The data reproducibility was confirmed by one additional experiment. c, Signal linearity of plasma peptides spiked with synthetic Aβ1–42, synthetic Aβ1–40, and synthetic APP669–711. The plasma samples, which were spiked over a 2.5–40 pM range for Aβ1–42 and APP669–711, and a 10–160 pM range for Aβ1–40, were prepared and measured by the IP–MS method. The linearity was evaluated with R2. The error bars indicate the standard deviations of normalized intensities obtained from four mass spectra. The data reproducibility was confirmed by one additional experiment. d, Normalized signal intensity of Aβ1–42, Aβ1–40, and Aβ1–40/Aβ1–42 in 19 subjects measured by two methods; using common internal standard SIL-Aβ1–38 (x axis) and using corresponding SIL-peptides (y axis). Pearson’s correlation coefficients (r) and their significance (two-sided P) are presented in the plots. The experiments were performed once. e, ROC analyses for Aβ1–42 and Aβ1–40/Aβ1–42 to distinguish between Aβ+ and Aβ individuals (n = 19) of the two methods; using the common internal standard SIL-Aβ1–38 (left) and using the corresponding SIL-peptides (right). f, Tables showing the performance values corresponding to e, as determined by ROC analyses and Youden’s index.

Source data

Extended Data Figure 6 Cellular and molecular characteristics of APP669–711.

a, Results of additional experiment 1 (Supplementary Information). Aβ-related peptides produced from human neuroblastoma cell line. MALDI–TOF mass spectra of Aβ-related peptides in human plasma (top), BE(2)-C cell culture supernatant (middle) and medium without cell culture (bottom). Representative spectra from five experiments are shown. The theoretical m/z values of peptides are 4,330.9 for Aß1–40, 4,515.1 for Aß1–42, and 4,689.4 for APP669–711. SIL-Aβ1–38 was used as an internal standard for the normalization of mass spectra. bd, Results of additional experiment 2 (Supplementary Information). b, Kinetics of fibril formation. Aβ1-42 (15 μM, open circles) or APP669–711 (15 μM, closed circles) were incubated in PBS at 37 °C. Fibril formation was monitored using the thioflavin T spectroscopic assay. Data are mean ± s.d. from four (Aβ1–42) or five (APP669–711) experiments. c, Size exclusion chromatography. Aβ1–42 (15 μM, left) or APP669–711 (15 μM, right) were incubated in PBS at 37 °C, and the supernatants were centrifuged (10,000g for 10 min) and subjected to size-exclusion chromatography (Sephacryl S-300 HR) at 0 (black), 3 (red), 6 (blue), 12 (orange), and 24 h (green). The elution times for molecular mass standards (kDa) are indicated by arrows. d, Changes in secondary structure during peptide aggregation. Aβ1–42 (15 μM, left) and APP669–711 (15 μM, right) were incubated in PBS at 37 °C, and circular dichroism spectra were measured at 0 (black), 3 (red), 6 (blue), 12 (orange), and 24 h (green). Experiments in bd were each performed once.

Source data

Extended Data Figure 7 Common cut-off values are applicable for both NCGG and AIBL data sets.

a, Unadjusted ROC analyses for each biomarker when predicting individual Aβ+/Aβ status for the NCGG + AIBL PIB (left, n = 232) and NCGG + AIBL overall (right, n = 373) data sets. b, Comparisons of biomarker performances within each analysis corresponding to the ROC curves in a. Each colour bar represents the AUC and 95% confidence. Statistically significant differences between two AUCs (DeLong test) and significant increments in predictive ability as assessed by NRI and IDI are indicated as in Fig. 2. All P values are two-sided, and Bonferroni corrected (multiplied by the number of comparisons, 6). c, Biomarker performances when applying the same cut-off values to each data set. For each biomarker, an optimal common cut-off value was determined by the Youden’s index of the ROC analysis for the NCGG + AIBL overall data set. The sensitivity, specificity and accuracy were then calculated at the common cut-off point for each biomarker in all data sets. d, Diagnostic performance plots for Aβ1–42, APP669–711/Aβ1–42, Aβ1–40/Aβ1–42 and the composite biomarker. Each row from top to bottom shows the plots for the NCGG PIB, AIBL PIB, AIBL overall, NCGG + AIBL PIB, and NCGG + AIBL overall data (unadjusted), respectively. Sensitivity (orange), specificity (blue) and accuracy (green) were plotted using the values of the biomarkers (x axis). The vertical dashed lines indicate the common cut-off values as shown in c. The blue and pink shaded squares indicate ranges in which a biomarker can perform with at least 80% and 85% accuracy, respectively.

Source data

Extended Data Figure 8 Possible clinical utility of the plasma biomarker.

a, b, Diagnostic performance plots for the composite biomarker in two specific settings (see Supplementary Discussion). a, Diagnostic performance plots for subjects with MCI in the AIBL PIB unadjusted data (n = 33) (left). The prevalence of Aβ-positivity for subjects with MCI was assumed to be 66%. Sensitivity (orange), specificity (blue), accuracy (green), PPV (red, dashed) and NPV (dark blue, dashed) were plotted against the values of the composite biomarker (x axis). The black vertical dashed line indicates a cut-off point as determined by the Youden’s index (y point) in the AIBL PIB data. At the y point, the sensitivity and specificity were 0.900 and 0.923, respectively. With these values, relationships between the prevalence and PPV or NPV were plotted (right). Note that these data do not correspond to the ROC analysis shown in Fig. 2d, because this diagnostic performance plot analysis does not contain subjects with AD. b, Diagnostic performance plots for cognitively normal subjects in the AIBL PIB unadjusted data (n = 63) (left). The prevalence of Aβ-positivity in general elderly people was assumed to be 30%. At the y point, the sensitivity and specificity were 0.880 and 0.868, respectively. With these values, relationships between the prevalence and PPV or NPV were also plotted (right). cg, Results of the additional analysis for subjects with and without AD (see Supplementary Discussion). c, Sample numbers for subjects with and without AD. d, Detailed clinical diagnoses of subjects without AD. e, ROC analyses for each plasma biomarker in the overall data (n = 51). f, ROC analysis of the composite biomarker in the overall (n = 51), AD (n = 31), and non-AD (n = 20) groups. g, Performance of the composite biomarker using the common cut-off value. The AUC values were computed from the ROC analysis and sensitivity, specificity, and accuracy were computed by applying the common cut-off value for the composite biomarker (0.376).

Source data

Extended Data Figure 9 Optimal generation of the composite biomarker.

Unadjusted ROC analyses of the composite biomarkers generated by different weightings (see Supplementary Discussion). a, Comparisons of the composite biomarkers generated by different weights for APP669–711/Aβ1–42 and Aβ1–40/Aβ1–42 normalized values (z-scores) in the discovery NCGG PIB data (n = 121, left) and validation AIBL PIB data (n = 111, right). The composite biomarker generated by the original weight (1:1) is coloured red, the weight estimated by the NCGG data (1.14:3.59) green, and the weight estimated by the AIBL data (3.04:1.95) blue. b, A comparison of the composite biomarkers generated by using different reference databases. The original composite biomarker (normalized by NCGG data) is coloured red, and the alternative composite biomarker normalized by AIBL data is orange. c, Summary of the ROC analyses. The AUCs and the representative best values of sensitivity, specificity and accuracy for these biomarkers as determined by the Youden’s index are shown.

Source data

Extended Data Table 1 Performance values of the plasma biomarkers

Supplementary information

Life Sciences Reporting Summary (PDF 86 kb)

Supplementary Information

This file contains Supplementary Text, Discussion, additional experiments and references. (PDF 1311 kb)

PowerPoint slides

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Nakamura, A., Kaneko, N., Villemagne, V. et al. High performance plasma amyloid-β biomarkers for Alzheimer’s disease. Nature 554, 249–254 (2018). https://doi.org/10.1038/nature25456

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing