ADNI Biomarker Core Report Leslie M Shaw & John Q Trojanowski Magdalena Korecka Michal Figurski Teresa Waligorska Magdalena Brylska Leona Fields Sarah Pan

ADNI Biomarker core 2013-early 2014 Studies reported in the ADNI/LONI website & studies newly approved by RARC/NIA/ADNI using ADNI biofluids Biofluid report update 2014 ADNI II batch analyses of CSF underway mrm/tandem mass spectrometry reference method for Ab1-42, progress report Review of anchoring to 2007 ADNI BL & cutpoints for CSF biomarkers in ADNI Planning for ADNI III

ADNI Biomarker core 2013-early 2014: Add-on studies reported on the ADNI/LONI website & add-on studies newly approved by RARC/NIA/ADNI using ADNI biofluids Reported mrm/tandem mass spectrometry of 567 tryptic peptides associated with 221 proteins, Caprion/FNIH/ADNI/PPSB in ADNI I CSF Approved/shipped or to be shipped YKL40 & Vilip1 in longitudinal ADNI CSF, WashU, Anne Fagan, 4th qtr 2013 Metabolic Pathways & Networks in ADNI I BL serum, DukeU, Rima Kaddurah-Douk, 2nd qtr 2014 DDE* in ADNI serum, EmoryU, Allan Levey, 2nd qtr 2014 Neurogranin in BASELINE CSF of ADNI I subjects, UGothenberg, Sweden, Kaj Blennow, 2nd qtr 2014 Tau in plasma of ADNI I BASELINE subjects, UGothenberg, Sweden, Kaj Blennow, 2nd qtr 2014 Phosphorylated a-synuclein in ADNI I BL CSF, UWash, Jing Zhang, 2nd qtr 2014 *DDE-a major breakdown product of the insecticide DDT that is highly fat soluble, very long biological half-life, potential risk factor for neurodegeneration. JAMA Neurology 2014.

Subjects providing CSFs at Baseline in Each ADNI Study Phase

ADNI GO and 2 Total (CSF) Total (PLA + SER) Grand Total as of 4/1/2014 Plasma Through March 2014 CSF Through March 2014 N* 3629 1077 Average Time (min) 72.14 43.62 Median Time (min) 55.00 28.00 95% CI (min) 70.14 - 74.14 40.19 - 47.04 *N is the samples for which collection time and freezing time data is available as of April 1, 2014 . ADNI GO and 2 Total (CSF) Total (PLA + SER) Grand Total as of 4/1/2014 Grand Total as of 3/11/2013 Number of Biofluids Collected as of April 1, 2014 1,178 7,422 8,600 5,329 Number of Aliquots in Bank 35,510 109,932 145,442 94,290   All ADNI 2,140 15,061 17,201 13,930 65,568 221,852 287,420 236,268

Aliquot counts by study subject Updated as of 4/14 2014 Aliquot counts for plasma, serum, CSF, urine Total volume and # of pristine aliquots available Aliquots utilized for Biomarker Core measurements Aliquots sent to investigators who had RARC/NIA/ADNI-approved studies

2014 batch analyses of ADNI II CSFs N=406 ADNI II CSFs + 15 pristine randomly selected replicates assayed using the AlzBio3 RUO immunoassay 253 BASELINE samples; 153 have a paired BL(2013) + 24 month sample Results will be ready for reporting on ADNI/LONI website end of May. QC data to date in next ppt.

2014 batch analyses of ADNI II CSFs (Interim analysis CSF pool qc data) 2013 Sample N Mean ± SD 2014 t-tau CSF abnormal pool #53 25 128±10.4 7 120±4.9 CSF normal pool #54 65.9±6.0 58.0±3.3 Ab1-42 148±9.4 147±6.4 236±18 241±4.7 p-tau181 26.5±1.4 27.7±0.7 19.1±1.2 19.7±1.1

Comments on: (1) “anchoring” & (2) cutpoint determination Anchoring ADNI 2008 & later batch datasets to 2007 BASELINE dataset: 2007 ADNI BASELINE data: generated at the same time as non-ADNI UPenn ADRC AD(autopsy-based) & living controls using the same lot # of immunoassay reagents Cutpts determined in the ADNI-independent cohort;ROC(balanced sens/spec) & mixture modeling used The anchoring process used a linear regression best fit eqn – no effect on interpretation of the data using cutpts determined in the “anchored”-to-2007 datasets vs the corresponding “RAW” datasets Cutpoint determination: In the ADNI-independent cohort with pathologic diagnosis in the AD group: ROC, with balanced sensitivity/specificity/test accuracy, and mixture modeling in this cohort very similar cutpt values Diagnosis-independent mixture modeling, first shown to provide a cutpoint in the ADNI 1(NC, LMCI, AD) population that was equivalent to that in two ADNI-independent path-based AD populations, the UPenn autopsy(AD) & Belgian autopsy(AD) cohorts. This confirmed the 192 pg/mL cutpoint for Ab1-42 If data not anchored to 2007, ie, if “RAW” data used, 192 pg/mL corresponds to ~250 pg/mL in 2008 onward for prospective studies the choice of cutpoint very much depends on: specificity vs sensitivity needs(eg, if goal in a treatment trial is to miss as few subjects with amyloid plaque burden then greater sensitivity and less specificity may be required=higher cutpt for Ab1-42; if goal in a more clinical setting is to provide confirmation of a high level of clinical likelihood of AD diagnosis, greater specificity may be required=lower cutpt for Ab1-42) Analytical method selected and any biases associated with the method, such as precision performance, accuracy, lot-to-lot variance Reference population, eg, clinically based AD compared to pathologically based AD diagnoses The addition of t-tau or p-tau181 to Ab1-42 results in improved detection of AD pathology

Anchoring 2008 CSF BASELINE dataset to 2007 N=328 BL 2007 CSF pristine aliquots analyzed In 2008 as part of the ADNI grant-defined longitudinal dataset, BL+yr1. *a modification of the assay buffer in 2008 resulted in a modest linear difference across measured concentrations of Ab1-42. Linear regression* Model for anchor MM cutpt=241pg/mL MM cutpt=202 pg/mL No effect of “anchoring” on the distribution of the data, but this “rescales” the dataset to the 2007 baseline dataset. 217 AD + NC 217 AD + NC “RAW” data “transformed”data Linear regression Model for anchor *Data and regression analyses as described, 4/27/2008, Seattle ADNI SC meeting.

No bias produced by “anchoring”on interpretation of CSF Ab1-42 ADNI 2012/13, anchored to 2007; CSF Ab1-42 concordance with FBP using Mixture Modeling (MM) cutpt concord discord FBP-/Ab- FBP+/Ab+ FBP-/Ab+ FBP+/Ab- concordance NC n=143 116 27 78 38 23 4 0.811 LMCI n=135 123 12 34 89 9 3 0.911 EMCI n=243 209 115 94 25 0.860 AD n=79 75 71 0.949 All n=600 523 77 231 292 61 16 0.872 ●●ADNI subjects are all 2012/13 AD+NC who had both lp & FBP scan●● ADNI 2012/13, RAW data; CSF Ab1-42 concordance with FBP using MM cutpt concord discord FBP-/Ab- FBP+/Ab+ FBP-/Ab+ FBP+/Ab- concordance NC n=143 116 27 78 38 23 4 0.811 LMCI n=135 123 12 34 89 9 3 0.911 EMCI n=243 209 115 94 25 0.860 AD n=79 75 71 0.949 All n=600 523 77 231 292 61 16 0.872

High conc Guanidine HCl(5 M) to release Ab1-42 from aggregates, oligomers Mixed bed ion exchange 96 well format for 1st step sample cleanup Use of a surrogate matrix with equivalent performance to CSF as a calibration matrix: defined all constituents and sources of these Use high quality Ab1-42 standard and cross-checked performance of several lots of this material; use uniformly N15-labelled IS; cross-check performance against an international standard preparation Waters ACUITY 2D HPLC + API 5000 tandem mass spectrometer Employ quality controls: aCSF (4 mg/mL BSA + electrolytes + Ab1-42) & CSF pools throughout Evaluated all major analytical parameters as defined in US FDA Guidance and CLSI guidelines Analyzed AD(autopsy-confirmed) & control CSF samples provided by the UPenn ADRC

ADNI 2: Biomarker Core mrm LC/MSMS method for Ab peptides in CSF Xevo TQ-S tandem mass spectrometer Ab1-40 Ab1-42 Eppendorf LoBind Tubes Oasis MCX mElution Plate Ab1-38 Injection position of switching valve - 2D chromatography ACQUITY UPLC Trapping column Analytical column UPLC BEH-300 C18 2.1x150mm, 1.7 µm Time(min) m/z 1129.0 m/z 1078.8 (Ab1-42 precursor [4+] ion) ( product [4+] ion) Further improved using a Xevo TQ-S tandem mass spectrometer; CSF aliquot size 100 mL Accuracy-based measurement of Ab1-42, Ab1-40 & Ab1-38 for all ADNI CSF samples Round Robin study completed(4 centers) in collaboration with Kaj Blennow, manuscript in press Will help answer questions about potential contribution of other metabolites, eg, Ab1-40 , Ab1-38 to utility of Ab1-42 and other APP species, tau fragments.

Method comparison in autopsy-proven AD subjects & living controls (n= 41 each) P=0.2229, deLong’s test

Support of standardization efforts ADNI-longterm commitment to standardization of all methods Alz Assn Global Biomarker Standardization Consortium Analytical methods standardization--strong support for improved performance of existing and new immunoassays for CSF biomarkers Support for mrm/tandem mass spectrometry for direct measurement of absolute Ab1-42 concentration IFCC/IRMM project to develop reference Ab1-42 peptide material and using mrm/msms and large pools of CSF with accurately measured Ab1-42 Need same for t-tau CAMD(Coalition Against Major Diseases) has made a substantial commitment to support use of HV and CSF AD biomarkers in treatment trials Hippocampal volume CSF AD biomarkers Close collaboration with Japan ADNI on a joint effort to standardize lab-to-lab method performance of AlzBio3 immunoassay Completed development and testing of a unified test procedure A follow up study now just underway for a final assessment in AD and normal CSF samples

Collaboration with Japan ADNI Completed first pilot study using a jointly developed “Unified” test procedure (U-SOP) that was described at the AAIC meeting in Boston, 2013 Will shortly complete the follow-up study that includes 20 AD and 20 controls, each run twice in separate runs. A total of 4 CSF qc pools are included and each run in quadruplicate in a total of 4 analytical runs (n=16 per pool). Plate 1 (Day 1)   1 2 3 4 5 6 7 8 9 10 11 12 A Blank ConA US-Pool_10 J-CSF 7 aliquot 1 US-CSF 9 aliquot 1 B St1 ConB US-CSF 1 aliquot 1 J-CSF 8 aliquot 1 US-CSF 10 aliquot 1 C St2 J-Pool_1 US-CSF 2 aliquot 1 J-CSF 9 aliquot 1 D St3 J-CSF 1 aliquot 1 US-CSF 3 aliquot 1 J-CSF 10 aliquot 1 J-Pool_2 E St4 J-CSF 2 aliquot 1 US-CSF 4 aliquot 1 US-Pool_15 F St5 J-CSF 3 aliquot 1 US-CSF 5 aliquot 1 US-CSF 6 aliquot 1 G St6 J-CSF 4 aliquot 1 US-CSF 7 aliquot 1 H St7 J-CSF 5 aliquot 1 J-CSF 6 aliquot 1 US-CSF 8 aliquot 1

BIOMARKER CORE AIMS FOR THE ADNI-3 RENEWAL Leslie M. Shaw and John Q BIOMARKER CORE AIMS FOR THE ADNI-3 RENEWAL Leslie M. Shaw and John Q. Trojanowski It is important to take into account the heterogeneity of AD in the ADNI-3 Biomarker Core in ADNI3. Many studies emphasize this including ADNI data showing that that a very small minority (4/22) of ADNI subjects with clinical AD/MCI only had AD plaque and tangle pathology at autopsy, while 82% had plaques and tangles in addition to TDP-43 and/or alpha-synuclein (a-syn) inclusions as well as hippocampal sclerosis in some cases (Toledo et al, ANP Commun, 1:65, 2013). These findings are echoed in a larger study of non-ADNI Penn subjects (Toledo et al, ANP, 124:23-35, 2012). Further, we have used a CSF total a-syn assay in ADNI CSF samples that may enable detection of co-morbid LBs in MCI/AD subjects in ADNI3 (Toledo et al, ANP, 126:683-697, 2013) and we also have access to a phospho-a-syn immunoassay that could be incorporated into ADNI-3 (Wang et al, Sci Trans Med, 4:121-20, 2/15/2012). TDP-43 biomarkers are not yet available so we work with Hugo Vanderstichele at ADx and Andreas Jeromin at Quanterex on TDP-43 ELISA based assays, but it is not yet certain if a TDP-43 immuno-assay will be ready for use in ADNI-3. We also need to address the issue of co-morbid cerebrovascular disease (CVD), but information on CVD may come from imaging rather than chemical biomarker studies. With this as background, the Aims of the Biomarker Core in ADNI 3 are as follows:

Continue to collect, aliquot, store, curate and track all biofluid samples collected from subjects in ADNI-1, ADNI-GO, ADNI-2 and ADNI-3 and continue regular reconciliation & reviews with the clinical core at UCSD . Continue longitudinal studies of existing ADNI biomarkers that are most informative in ADNI studies including CSF t-tau, p-tau181 and Aβ1-42 as well as possibly total a-syn and phospho-a-syn while VILIP1 and YKL40 data should be available later this year from WashU. Implement other validated and promising new biomarker immunoassays such as those that measure other species of tau and Aβ in CSF, but especially in plasma, including fragments of tau and Aβ as well as oligomers thereof in addition to other biomarkers such as VILIP1 and YKL40. Ongoing studies will determine the likelihood of incorporating these new analytes into ADNI-3. Continue modeling assay data studies for interconversions, eg, lot-to-lot; analytical platform to platform Pending the outcome of ongoing studies, incorporate CSF and plasma proteomic, metabalomic and lipidomic biomarkers. Use our newly validated mrmMass Spec assay to identify and measure additional species of Aβ (e.g. Aβ 40, 38, etc) as well as other species/fragments of tau. Partner with the PPSB and other investigators who pursue additional approved Add-On studies of AD biomarkers in plasma and/or CSF in order to bring these assays on-line for use in ADNI 3  

Examples Of Analytes And Biomarker Platforms To Consider In ADNI 3 At the Keystone AD/PD Meeting, Dennis Selkoe described detection of Aβ oligomers in CSF and will apply to study ADNI CSF samples with this immunoassay; Mary Savage has recently described validation of a new Ab oligomer immunoassay & we plan to collaborate on studies using UPenn ADCC CSF from AD, PD, ALS, FTLD and controls. Kaddurah-Daouk will study ADNI serum with the platforms described here and Federoff plans further studies with his platform

It takes a great team effort! Margaret Knapik-Czajka Ravi Patel Pawel Zero William Hu Ju Hee Kang Jon Toledo Anne Fagan Uwe Christians Kaj Blennow Henrik Zetterberg John Q Trojanowski Magdalena Korecka Magdalena Brylska Teresa Waligorska Michal Figurski Leona Fields Sarah Pan Virginia M-Y Lee Chris Clark* Steve Arnold Hugo Vanderstichele *Deceased Holly Soares Adam Simon Robert Dean Eric Siemers Piotr Lewczuk William Potter Rand Jenkins Erin Chambers Supported by the NIH/NIA and families of our patients ADNI investigators include: (complete listing available at www.loni.ucla.edu\ADNI\ Collaboration\ADNI_Manuscript_Citations.pdf).