FDA clearance means the test may be sold for the use written on its label. It is not a
promise that the result will change what happens to you. Most of the figures further down this page come from
research assays that are not cleared for clinical use.
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217; developed with Eli Lilly)
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
Compared against: amyloid-PET
Who was studied: Individuals >=55 y with signs, symptoms or complaints of cognitive decline (clearance study details not disclosed in release)
Fourth FDA-cleared AD blood test; first single-biomarker test cleared for both rule-in and rule-out with the same cutoffs across primary and specialty care; three result categories (positive/intermediate/negative). Press release gives NO performance numbers, only 'high agreement with amyloid PET imaging'. Not stand-alone. See separate row for the prototype-assay clinical-cohort study (Hibar 2026). OpenFDA 510(k) record: K261686 Elecsys Phospho-Tau (217P) Plasma, decision date 2026-08-19 (Roche announced 2026-08-24).
Source: FDA 510(k) clearance (Roche investor update, 24 Aug 2026) · 2026
C2N PrecivityAD2 (APS2: HRMS %p-tau217 + Aβ42/40), FDA-cleared IVD version
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 92%
- Specificity 98%
- PPV 97.6%
- NPV 93.1%
Compared against: amyloid-PET · 1,142 people
Who was studied: Pooled Validation Cohort, 3 cohorts, n=1,142 adults >=40 with cognitive symptoms (21.2% SCD, 47.3% MCI, 31.5% dementia); amyloid prevalence 52.7%
Third FDA-cleared AD blood test; only one indicated from age 40; first HRMS-based. Reference = amyloid PET visual read or CSF biomarkers. Dual cutoffs: Positive result PPV 97.6%, Negative result NPV 93.1%; a 'Likely Positive' intermediate category occurred in 17.3% of subjects with PPV 77.3%. Sensitivity/specificity/AUC not disclosed. Performance reported as consistent across 12 chronic comorbidities incl. CKD. FDA-cleared version expected available later in 2026; CLIA LDT version already offered. OpenFDA 510(k): K253240 PrecivityAD2 Test, C2N Diagnostics, LLC, decision date 2026-08-19 (C2N announced 2026-08-20). FROM 510(k) SUMMARY K253240 Table 53.24 (reference = amyloid PET visual read n=209 or Lumipulse CSF Aβ42/40 n=933): Positive 413 present/10 absent (PPV 97.6%, 95.7-98.7; 37.0% of results; LR+ 37.05); Likely Positive 153/45 (PV 77.3%; 17.3% of results); Negative 36/485 (PV 6.9% => NPV 93.1%; 45.6% of results; LR- 0.07). Sensitivity 92.0% (413/449) and specificity 98.0% (485/495) are DERIVED here from the FDA 2x2 excluding Likely Positive; FDA reports predictive values, not sens/spec. Performance similar across SCD (PPV 90.3%, prevalence 24.0%), MCI (97.9%) and dementia (98.5%). PET-visual-read-only subset (n=209): PPV 90.5%, Likely Positive PV 57.5%, NPV 93.7%; CSF-only subset (n=933): PPV 99.1%, NPV 93.0%. Note 82% of the reference standard was CSF, not PET.
Source: FDA 510(k) clearance clinical validation study (C2N press release, 20 Aug 2026) · 2026
Roche Elecsys Phospho-Tau (217P) plasma immunoassay (prototype)
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 98.98%
- Specificity 29.17%
- AUC 0.878
- NPV 92.51%
Compared against: amyloid-PET · 538 people
Who was studied: Cognitively impaired from 5 cohorts (n=538: 394 A+, 144 A-)
Cognitively impaired: AUC 0.878 (0.840-0.915); at rule-out cutoff <0.189 pg/mL sensitivity 98.98%, specificity 29.17%, NPV 92.51%. Note this AUC is lower than most academic p-tau217 reports. Five cohorts incl. A4, SKYLINE, AIBL, CREAD/CREAD2 (trial screening populations). Roche-authored. Rule-out cutoff deliberately trades specificity for sensitivity. FDA clearance K261686 (2026-08-19) applies to the commercial assay; this study used the prototype.
Source: Hibar et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 41537338 · doi:10.1002/alz.71009
Roche Elecsys Phospho-Tau (217P) plasma immunoassay (prototype)
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 95.54%
- Specificity 50.72%
- AUC 0.907
- NPV 98.6%
Compared against: amyloid-PET · 1,610 people
Who was studied: Cognitively unimpaired from 5 cohorts (n=1610: 224 A+, 1386 A-)
Cognitively unimpaired: AUC 0.907 (0.885-0.929); sensitivity 95.54%, specificity 50.72%, NPV 98.60% at <0.189 pg/mL. Five cohorts incl. A4, SKYLINE, AIBL, CREAD/CREAD2 (trial screening populations). Roche-authored. Rule-out cutoff deliberately trades specificity for sensitivity. FDA clearance K261686 (2026-08-19) applies to the commercial assay; this study used the prototype.
Source: Hibar et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 41537338 · doi:10.1002/alz.71009
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 93.1%
- Specificity 92%
- PPV 88.5%
- NPV 95.3%
Compared against: amyloid-PET · 958 people
Who was studied: Full analysis population, 958 with amyloid PET (age 55-85, mean 71.1; MMSE 20-30; 24.8% primary care, 75.2% secondary/tertiary); PET prevalence 39.5%
Overall 2x2: Positive 285 PET+/37 PET- (PV 88.5%, 84.9-91.4; 33.6% of results); Intermediate 72/119 (19.9% of results; PV 37.7%); Negative 21/424 (46.5% of results; PV for positivity 4.7% => NPV 95.3%). Derived sens 93.1% (285/306), spec 92.0% (424/461) excluding intermediates. LR+ 11.82, LR- 0.08. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 94.7%
- Specificity 93.5%
- PPV 78.3%
- NPV 98.6%
Compared against: amyloid-PET · 238 people
Who was studied: Primary care subgroup (n=238); PET prevalence 21.4%
PRIMARY CARE 2x2: Positive 36 PET+/10 PET- (PPV 78.3%, 66.1-87.0; 19.3% of results); Intermediate 13/34 (19.7%); Negative 2/143 (60.9% of results; NPV 98.6%). Derived sens 94.7% (36/38), spec 93.5% (143/153) excluding intermediates. FDA text: PPV 78.3%, NPA 94.7% at upper cutoff; NPV 98.6% at lower cutoff. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary. FDA-defined single-cutoff agreements: lower cutoff PPA 96.1% (49/51), upper cutoff NPA 94.7% (177/187); 19.7% intermediate.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 92.9%
- Specificity 91.2%
- PPV 90.2%
- NPV 93.7%
Compared against: amyloid-PET · 720 people
Who was studied: Secondary/tertiary care subgroup (n=720); PET prevalence 45.4%
SECONDARY/TERTIARY 2x2: Positive 249/27 (PPV 90.2%, 86.5-93.0; 38.3% of results); Intermediate 59/85 (20.0%); Negative 19/281 (41.7%; NPV 93.7%). Derived sens 92.9% (249/268), spec 91.2% (281/308) excluding intermediates. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary. FDA-defined: lower cutoff PPA 94.2%, upper cutoff NPA 93.1%; 20.0% intermediate.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
C2N PrecivityAD2 (APS2), FDA-cleared IVD version
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- Sensitivity 91.8%
- Specificity 92.7%
- PPV 90.5%
- NPV 93.7%
Compared against: amyloid-PET · 209 people
Who was studied: Subset of pooled validation cohort with amyloid PET visual read as reference (n=209; prevalence 45.9%)
Against PET alone the numbers are lower than the pooled headline: Positive 67 PET+/7 PET- (PPV 90.5%, 81.7-95.3); Likely Positive 23/17 (PV 57.5%; 19.1% of results); Negative 6/89 (PV 6.3% => NPV 93.7%). Derived sens 91.8% (67/73), spec 92.7% (89/96). Compare CSF-comparator subset (n=933): PPV 99.1%, NPV 93.0%, Likely Positive PV 82.3%.
Source: FDA 510(k) Summary K253240, Table 53.25 (amyloid PET visual read comparator subset) · 2026
C2N PrecivityAD2 (APS2), FDA-cleared IVD version
Measured
FDA cleared 2026-08-19
- Measures p-tau217
- Sample: plasma
- PPV 90.3%
- NPV 96%
Compared against: amyloid-PET · 242 people
Who was studied: SCD subgroup of pooled validation cohort (n=242; amyloid prevalence 24.0%)
Earliest symptomatic stage: Positive PV 90.3% (28/31; 75.1-96.7), Likely Positive PV 67.6% (23/34), Negative PV 4.0% (7/177) => NPV 96.0%; 14.0% Likely Positive. MCI (n=540, prev 53.9%): PPV 97.9%, NPV 91.4%. Dementia (n=360, prev 70.3%): PPV 98.5%, NPV 91.9%. Reference = PET visual read or CSF (pooled).
Source: FDA 510(k) Summary K253240, Table 53.27 (subjective cognitive decline subgroup) · 2026
Fujirebio Lumipulse G pTau217 and pTau217/Aβ42 ratio
Measured
FDA cleared 2025-05-16
- Measures p-tau217
- Sample: plasma
- Sensitivity 83%
- NPV 93%
Compared against: amyloid-PET · 191 people
Who was studied: Mayo Clinic Florida cognitively normal participants: 57 Black, 46 Hispanic White, 88 non-Hispanic White (n=191)
Preclinical AD (amyloid-PET >=25 Centiloids) in cognitively normal, ethnoracially diverse sample. Cutoffs derived in symptomatic patients gave sensitivity only 57% (p-tau217) and 63% (ratio); cohort-optimized cutoffs raised sensitivity to 83% and 77%; NPV >93% (recorded as 93, lower bound). Evidence that symptomatic-population cutoffs under-detect preclinical amyloid. FDA clearance applies to the ratio in symptomatic patients >=55, not this preclinical use.
Source: Piura et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 42619350 · doi:10.1002/alz.71763
Fujirebio Lumipulse G pTau217/β-Amyloid 1-42 Plasma Ratio
Measured
FDA cleared 2025-05-16
- Measures p-tau217
- Sample: plasma
- Sensitivity 97.6%
- Specificity 90.8%
- PPV 91.8%
- NPV 97.3%
Compared against: amyloid-PET · 499 people
Who was studied: Multi-center clinical study, 499 plasma samples from cognitively impaired adults >=55 y (specialized care setting)
First FDA-cleared blood test for AD (510(k), Breakthrough Device; predicate = Lumipulse CSF Aβ42/40 ratio). Reference = amyloid PET or CSF. PPV 91.7%, NPV 97.3%; <20% of 499 patients received an indeterminate result (PPV/NPV exclude indeterminates). Sensitivity/specificity/AUC not stated in press release. Intended for symptomatic patients >=55 in specialized care; not a screening or stand-alone test. OpenFDA 510(k) record: K242706, decision date 2025-05-16, applicant Fujirebio Diagnostics, Inc. Predicate Lumipulse G β-Amyloid Ratio (1-42/1-40) CSF was De Novo DEN200072 (2022-05-04). FROM 510(k) SUMMARY K242706 pivotal 2x2 (n=499, PET or CSF reference, prevalence 51.1%): Positive 201 PET/CSF+ vs 18 neg (PV 91.8%, 95% CI 87.8-94.6); Indeterminate 49/49 (19.6% of results); Negative 5 pos vs 177 neg (PV of negative for positivity 2.7%, i.e. NPV 97.3%). Sensitivity 97.6% (201/206) and specificity 90.8% (177/195) are DERIVED here from the FDA 2x2 after excluding indeterminates; FDA summary itself reports predictive values and likelihood ratios (LR+ 10.68, LR- 0.03), not sens/spec. Cohorts: AriBio Polaris-AD phase III screening, BioFINDER-2 (Malmö), Bio-Hermes-001, WRAP. Mean age 72, 84.8% white; 58.2% clinically AD dementia, 40.7% SCD/MCI. Cutoffs derived in separate 208-sample set (PV+ 94.6%, prevalence 54.8%).
Source: FDA 510(k) clearance clinical study (FDA press release, 16 May 2025) · 2025
Roche Elecsys pTau181 plasma (cobas; developed with Eli Lilly)
Measured
FDA cleared 2025-10-08
- Measures p-tau181
- Sample: plasma
- Sensitivity 92.7%
- Specificity 51.3%
- PPV 22.4%
- NPV 97.9%
Compared against: amyloid-PET · 312 people
Who was studied: Multicenter non-interventional study, 312 participants >=55 y with cognitive complaints; early-stage, low-prevalence population reflective of primary care
Second FDA-cleared AD blood test; first indicated for primary care; RULE-OUT only. Press release states NPV 97.9% and nothing else (no sensitivity, specificity, PPV, AUC, or reference-standard detail beyond 'Alzheimer's-related amyloid pathology'). NPV in a low-prevalence population is inherently high; PPV was not disclosed. Also CE-marked (IVDR). OpenFDA 510(k) record: K252163 Elecsys Phospho-Tau (181P) Plasma, decision date 2025-10-08 (Roche announced 2025-10-13). FROM 510(k) SUMMARY K252163 (Roche study RD006263, prospective, 8 sites: 6 US, 2 EU; 95.8% US): n=312, mean age 69.1 (55-80), 59% white, 34% Black; 41% SCD, 56% MCI, 1% mild dementia. Reference = amyloid PET visual read (3-reader majority); PET prevalence only 13.1% (41/312). At cutoff 0.722 pg/mL: PPA 92.7% (38/41; 80.6-97.5), NPA 51.3% (139/271; 45.4-57.2), TPA 56.7%, PPV 22.4% (38/170), NPV 97.9% (139/142; 94.5-99.3), LR+ 1.90, LR- 0.143, rule-out rate 45.5%. KEY CAVEAT: the headline 97.9% NPV arises mostly from the 13% prevalence; specificity is 51% and PPV 22%, so a positive result is more often wrong than right.
Source: FDA 510(k) clearance clinical study (Roche investor update, 13 Oct 2025) · 2025
Roche Elecsys CSF p-tau181/Aβ42 ratio (fully automated)
Measured
FDA cleared 2022-12-07
- Measures p-tau181
- Sample: CSF
- AUC 0.94
Compared against: amyloid-PET · 923 people
Who was studied: BioFINDER (n=277, flutemetamol PET) for cutoff definition; ADNI (n=646, florbetapir PET) validation
BENCHMARK CSF test. BioFINDER: overall percent agreement with PET 90%, AUC 94% (recorded); ADNI validation with predefined cutoffs: OPA 89-90%, AUC 96%. Elecsys CSF p-tau181/Aβ42 later FDA-cleared (2022; date not verified here). Plasma %p-tau217 was judged clinically equivalent to this test in Barthélemy 2024. OpenFDA 510(k): K221842 Elecsys β-Amyloid (1-42) CSF II / Phospho-Tau (181P) CSF, decision 2022-12-07.
Source: Hansson et al. 2018, Alzheimer's & Dementia · 2018 · PubMed 29499171 · doi:10.1016/j.jalz.2018.01.010
Florbetapir F18 amyloid PET (Amyvid), visual read
Measured
FDA cleared
- Measures amyloid-PET
- Sample: plasma
Compared against: autopsy · 29 people
Who was studied: End-of-life patients from hospice/long-term care (n=29 primary autopsy cohort; 35 autopsied) plus 74 young controls
BENCHMARK for the 'approaches amyloid-PET accuracy' claim. Visual PET read agreed with postmortem amyloid status in 96% of 29 individuals; correlation with immunohistochemistry rho 0.78. Sensitivity/specificity not reported in abstract (not given). The sample type is not applicable (imaging); set to plasma only to satisfy schema enum. Amyvid FDA approval (2012) date not verified here.
Source: Clark et al. 2011, JAMA (AV45-A07 autopsy study) · 2011 · PubMed 21245183 · doi:10.1001/jama.2010.2008
[18F]flortaucipir tau PET (Tauvid), visual read
Measured
FDA cleared
- Measures tau-PET
- Sample: plasma
- Sensitivity 92.3%
- Specificity 52%
Compared against: autopsy · 64 people
Who was studied: A16 primary cohort: 64 terminally ill patients >50 y with autopsy (mean age 82.5; 77% dementia), 28 sites
BENCHMARK. Prediction of B3 (Braak V/VI) tau pathology across 5 readers: sensitivity 92.3-100% and specificity 52.0-92.0% (lower bounds recorded). High ADNC: sensitivity 94.7-100%, specificity 50.0-92.3%. Shows that even PET 'gold standards' have reader-dependent specificity. The sample type not applicable (imaging). Tauvid FDA approval (2020) date not verified here.
Source: Fleisher et al. 2020, JAMA Neurology (A16 autopsy study) · 2020 · PubMed 32338734 · doi:10.1001/jamaneurol.2020.0528
Roche cobas e801 plasma p-tau217 (RUO, fully automated)
Measured
FDA cleared
- Measures p-tau217
- Sample: plasma
- AUC 0.94
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
AUC 0.94 (0.91-0.96) in full interim sample; 0.92 (0.89-0.96) in cognitively unimpaired. (Commercial Elecsys pTau217 later FDA-cleared K261686, 2026-08-19.) Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Fujirebio Lumipulse G1200 plasma p-tau217/Aβ42 ratio
Measured
FDA cleared
- Measures p-tau217
- Sample: plasma
- AUC 0.91
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
AUC 0.91 (0.89-0.94) for ratio vs 0.90 (0.86-0.93) for p-tau217 alone. Note: in this largely cognitively unimpaired community cohort the FDA-cleared ratio scored lowest of the four platforms. Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Fujirebio Lumipulse G pTau217 and pTau217/Aβ42 ratio
Measured
cleared
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 83%
- NPV 93%
- Cut-off 0.132 pg/mL (p-tau217, preclinical-optimized); ratio cutoff 0.0059
Compared against: amyloid-PET · 191 people
Who was studied: Mayo Clinic Florida cognitively normal participants: 57 Black, 46 Hispanic White, 88 non-Hispanic White (n=191)
Strength of the evidence: Moderate (small, cutoffs optimized in same cohort = optimistic)
Preclinical AD (amyloid-PET >=25 Centiloids) in cognitively normal, ethnoracially diverse sample. Cutoffs derived in symptomatic patients gave sensitivity only 57% (p-tau217) and 63% (ratio); cohort-optimized cutoffs raised sensitivity to 83% and 77%; NPV >93% (recorded as 93, lower bound). Evidence that symptomatic-population cutoffs under-detect preclinical amyloid. FDA clearance applies to the ratio in symptomatic patients >=55, not this preclinical use.
Source: Piura et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 42619350 · doi:10.1002/alz.71763
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217; developed with Eli Lilly)
Measured
cleared
- Sample: plasma
- Platform: Elecsys
Compared against: amyloid-PET
Who was studied: Individuals >=55 y with signs, symptoms or complaints of cognitive decline (clearance study details not disclosed in release)
Strength of the evidence: Low for numbers (none disclosed); regulatory clearance itself verified
Fourth FDA-cleared AD blood test; first single-biomarker test cleared for both rule-in and rule-out with the same cutoffs across primary and specialty care; three result categories (positive/intermediate/negative). Press release gives NO performance numbers, only 'high agreement with amyloid PET imaging'. Not stand-alone. See separate row for the prototype-assay clinical-cohort study (Hibar 2026). OpenFDA 510(k) record: K261686 Elecsys Phospho-Tau (217P) Plasma, decision date 2026-08-19 (Roche announced 2026-08-24).
Source: FDA 510(k) clearance (Roche investor update, 24 Aug 2026) · 2026
C2N PrecivityAD2 (APS2: HRMS %p-tau217 + Aβ42/40), FDA-cleared IVD version
Measured
cleared
- Sample: plasma
- Platform: mass-spec
- Sensitivity 92%
- Specificity 98%
- PPV 97.6%
- NPV 93.1%
- Cut-off 66.5 APS2 score: Positive >66.5; Negative <27.5; Likely Positive 27.5-66.5
Compared against: amyloid-PET · 1,142 people
Who was studied: Pooled Validation Cohort, 3 cohorts, n=1,142 adults >=40 with cognitive symptoms (21.2% SCD, 47.3% MCI, 31.5% dementia); amyloid prevalence 52.7%
Strength of the evidence: Moderate (company press release; PPV/NPV depend on study prevalence, not disclosed)
Third FDA-cleared AD blood test; only one indicated from age 40; first HRMS-based. Reference = amyloid PET visual read or CSF biomarkers. Dual cutoffs: Positive result PPV 97.6%, Negative result NPV 93.1%; a 'Likely Positive' intermediate category occurred in 17.3% of subjects with PPV 77.3%. Sensitivity/specificity/AUC not disclosed. Performance reported as consistent across 12 chronic comorbidities incl. CKD. FDA-cleared version expected available later in 2026; CLIA LDT version already offered. OpenFDA 510(k): K253240 PrecivityAD2 Test, C2N Diagnostics, LLC, decision date 2026-08-19 (C2N announced 2026-08-20). FROM 510(k) SUMMARY K253240 Table 53.24 (reference = amyloid PET visual read n=209 or Lumipulse CSF Aβ42/40 n=933): Positive 413 present/10 absent (PPV 97.6%, 95.7-98.7; 37.0% of results; LR+ 37.05); Likely Positive 153/45 (PV 77.3%; 17.3% of results); Negative 36/485 (PV 6.9% => NPV 93.1%; 45.6% of results; LR- 0.07). Sensitivity 92.0% (413/449) and specificity 98.0% (485/495) are DERIVED here from the FDA 2x2 excluding Likely Positive; FDA reports predictive values, not sens/spec. Performance similar across SCD (PPV 90.3%, prevalence 24.0%), MCI (97.9%) and dementia (98.5%). PET-visual-read-only subset (n=209): PPV 90.5%, Likely Positive PV 57.5%, NPV 93.7%; CSF-only subset (n=933): PPV 99.1%, NPV 93.0%. Note 82% of the reference standard was CSF, not PET.
Source: FDA 510(k) clearance clinical validation study (C2N press release, 20 Aug 2026) · 2026
Plasma p-tau217 on Simoa platforms, pooled (meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: Simoa
- Sensitivity 89.4%
- Specificity 87.5%
Compared against: amyloid-PET · 18,073 people
Who was studied: Pooled studies with amyloid-PET reference, Jan 2020-Feb 2026
Strength of the evidence: High (systematic review) but high heterogeneity
Simoa pooled sensitivity 0.894 (0.861-0.920), specificity 0.875 (0.845-0.900). Heterogeneity I2 63-86%. P-tau217/Aβ42 ratio roughly doubled DOR vs p-tau217 alone (relative DOR 1.96, 1.19-3.23) in 5 paired studies.
Source: Yuan et al. 2026, Neurological Sciences (meta-analysis, 25 studies, 18,073 participants, PROSPERO CRD420261371940) · 2026 · PubMed 42649434 · doi:10.1007/s10072-026-09322-8
Plasma p-tau217 on CLIA (chemiluminescence, e.g. Lumipulse) platforms, pooled (meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 88.6%
- Specificity 86.5%
Compared against: amyloid-PET · 18,073 people
Who was studied: Pooled studies with amyloid-PET reference, Jan 2020-Feb 2026
Strength of the evidence: High (systematic review) but high heterogeneity
CLIA pooled sensitivity 0.886 (0.860-0.907), specificity 0.865 (0.826-0.896). Heterogeneity I2 63-86%. P-tau217/Aβ42 ratio roughly doubled DOR vs p-tau217 alone (relative DOR 1.96, 1.19-3.23) in 5 paired studies.
Source: Yuan et al. 2026, Neurological Sciences (meta-analysis, 25 studies, 18,073 participants, PROSPERO CRD420261371940) · 2026 · PubMed 42649434 · doi:10.1007/s10072-026-09322-8
Plasma p-tau217 on ECLIA (electrochemiluminescence, e.g. Elecsys/MSD) platforms, pooled (meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: Elecsys
- Sensitivity 93.3%
- Specificity 79.1%
Compared against: amyloid-PET · 18,073 people
Who was studied: Pooled studies with amyloid-PET reference, Jan 2020-Feb 2026
Strength of the evidence: High (systematic review) but high heterogeneity
ECLIA pooled sensitivity 0.933 (0.918-0.946), specificity 0.791 (0.571-0.915) - highest sensitivity, lowest specificity, very wide CI. Heterogeneity I2 63-86%. P-tau217/Aβ42 ratio roughly doubled DOR vs p-tau217 alone (relative DOR 1.96, 1.19-3.23) in 5 paired studies.
Source: Yuan et al. 2026, Neurological Sciences (meta-analysis, 25 studies, 18,073 participants, PROSPERO CRD420261371940) · 2026 · PubMed 42649434 · doi:10.1007/s10072-026-09322-8
Plasma p-tau217 on LC-MS/MS platforms, pooled (meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: mass-spec
- Sensitivity 82.7%
- Specificity 93.8%
Compared against: amyloid-PET · 18,073 people
Who was studied: Pooled studies with amyloid-PET reference, Jan 2020-Feb 2026
Strength of the evidence: High (systematic review) but high heterogeneity
LC-MS/MS pooled sensitivity 0.827 (0.725-0.897), specificity 0.938 (0.692-0.990) - highest specificity; note the wide CIs. Heterogeneity I2 63-86%. P-tau217/Aβ42 ratio roughly doubled DOR vs p-tau217 alone (relative DOR 1.96, 1.19-3.23) in 5 paired studies.
Source: Yuan et al. 2026, Neurological Sciences (meta-analysis, 25 studies, 18,073 participants, PROSPERO CRD420261371940) · 2026 · PubMed 42649434 · doi:10.1007/s10072-026-09322-8
Plasma p-tau217, pooled across assays (Bayesian bivariate meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: other
- Sensitivity 87.3%
- Specificity 85.5%
Compared against: amyloid-PET · 19,652 people
Who was studied: Pooled biomarker-defined AD cohorts
Strength of the evidence: High (systematic review) but heterogeneous
Amyloid-positivity subgroup recorded (sens 87.3%, spec 85.5%). Overall biomarker-defined AD: sens 85.4% (81.4-88.7), spec 88.0% (85.1-90.6), PLR 7.13, NLR 0.167, DOR 42.7. Clinical AD diagnosis subgroup: sens 72.9%, spec 89.5%. Authors caution about optimized cutoffs and heterogeneity; position p-tau217 as a triage biomarker.
Source: Gebril et al. 2026, Molecular Neurobiology (meta-analysis, 27 studies, 19,652 participants, search to Jul 2025) · 2026 · PubMed 42071133 · doi:10.1007/s12035-026-05864-2
Roche Elecsys Phospho-Tau (217P) plasma immunoassay (prototype)
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- Sensitivity 98.98%
- Specificity 29.17%
- AUC 0.878
- NPV 92.51%
- Cut-off 0.189 pg/mL (rule-out cutoff; negative if below)
Compared against: amyloid-PET · 538 people
Who was studied: Cognitively impaired from 5 cohorts (n=538: 394 A+, 144 A-)
Strength of the evidence: Moderate-high (large, but industry-run and trial-screening cohorts)
Cognitively impaired: AUC 0.878 (0.840-0.915); at rule-out cutoff <0.189 pg/mL sensitivity 98.98%, specificity 29.17%, NPV 92.51%. Note this AUC is lower than most academic p-tau217 reports. Five cohorts incl. A4, SKYLINE, AIBL, CREAD/CREAD2 (trial screening populations). Roche-authored. Rule-out cutoff deliberately trades specificity for sensitivity. FDA clearance K261686 (2026-08-19) applies to the commercial assay; this study used the prototype.
Source: Hibar et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 41537338 · doi:10.1002/alz.71009
Roche Elecsys Phospho-Tau (217P) plasma immunoassay (prototype)
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- Sensitivity 95.54%
- Specificity 50.72%
- AUC 0.907
- NPV 98.6%
- Cut-off 0.189 pg/mL (rule-out cutoff; negative if below)
Compared against: amyloid-PET · 1,610 people
Who was studied: Cognitively unimpaired from 5 cohorts (n=1610: 224 A+, 1386 A-)
Strength of the evidence: Moderate-high (large, but industry-run and trial-screening cohorts)
Cognitively unimpaired: AUC 0.907 (0.885-0.929); sensitivity 95.54%, specificity 50.72%, NPV 98.60% at <0.189 pg/mL. Five cohorts incl. A4, SKYLINE, AIBL, CREAD/CREAD2 (trial screening populations). Roche-authored. Rule-out cutoff deliberately trades specificity for sensitivity. FDA clearance K261686 (2026-08-19) applies to the commercial assay; this study used the prototype.
Source: Hibar et al. 2026, Alzheimer's & Dementia · 2026 · PubMed 41537338 · doi:10.1002/alz.71009
Fujirebio Lumipulse G pTau217 plasma (routine clinical lab)
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 84.5%
- Specificity 87.5%
- Cut-off 0.22 ng/L (low <0.22; high >0.34; intermediate between)
Compared against: CSF · 121 people
Who was studied: Sahlgrenska University Hospital clinical neurochemistry lab, all routine plasma p-tau217 tests Aug 2025-Mar 2026 (n=1,352 measurements; 121 with paired CSF)
Strength of the evidence: Moderate (small paired subset, real-world)
REAL-WORLD: positive percent agreement 84.5% (72.6-92.7) and negative percent agreement 87.5% (73.2-95.8) vs CSF Aβ42/40 (recorded as sens/spec). Of 1,352 routine samples, 40.0% low, 16.9% intermediate, 43.1% high probability. Within-batch CV <7.3%. Routine-practice agreement is well below research-cohort AUCs.
Source: Arslan et al. 2026, Clinical Chemistry and Laboratory Medicine · 2026 · PubMed 42053135 · doi:10.1515/cclm-2026-0370
Plasma p-tau217 (platform not stated in abstract)
Measured
Research use only
- Sample: plasma
- Platform: other
- Sensitivity 88%
- Specificity 84%
Compared against: clinical-diagnosis · 738 people
Who was studied: Huashan Hospital (Shanghai) memory clinic, 738 consecutive patients with cognitive complaints
Strength of the evidence: Moderate (single center, retrospective)
Real-world retrospective: sensitivity 0.88, specificity 0.84 for AD (reference = tier-1 amyloid PET or CSF-informed diagnosis). Adding p-tau217 revised 30.2% of diagnoses and raised clinician confidence 65.1% to 79.4%. Assay platform not named in abstract.
Source: Huang et al. 2026, The Innovation · 2026 · PubMed 42254957 · doi:10.1016/j.xinn.2026.101272
Fujirebio Lumipulse G pTau217 plasma
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.916
- Cut-off 0.146 pg/mL lower cutoff; 0.486 pg/mL upper cutoff (97.5% sens/spec model)
Compared against: CSF · 185 people
Who was studied: Prospective, 7 memory clinics in Catalonia (only one with prior blood-biomarker experience); consecutive patients with cognitive impairment undergoing CSF testing (n=185)
Strength of the evidence: Moderate-high (prospective multicenter real-world)
AUC 0.916 vs CSF Aβ status; single-cutoff accuracy only 84.9%. With strict dual thresholds (97.5% sens/spec), 42.7% of patients fell in the intermediate zone and the remaining 57.3% were classified with 95.3% accuracy. Illustrates the accuracy-vs-indeterminate trade-off in real-world multicenter use.
Source: Massons et al. 2026, Annals of Clinical and Translational Neurology · 2026 · PubMed 41968970 · doi:10.1002/acn3.70387
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- Sensitivity 93.1%
- Specificity 92%
- PPV 88.5%
- NPV 95.3%
- Cut-off 0.378 pg/mL: Positive >0.378; Negative <=0.223; Intermediate 0.223-0.378
Compared against: amyloid-PET · 958 people
Who was studied: Full analysis population, 958 with amyloid PET (age 55-85, mean 71.1; MMSE 20-30; 24.8% primary care, 75.2% secondary/tertiary); PET prevalence 39.5%
Strength of the evidence: High (regulatory pivotal, large) but retrospective and PPV/NPV prevalence-dependent
Overall 2x2: Positive 285 PET+/37 PET- (PV 88.5%, 84.9-91.4; 33.6% of results); Intermediate 72/119 (19.9% of results; PV 37.7%); Negative 21/424 (46.5% of results; PV for positivity 4.7% => NPV 95.3%). Derived sens 93.1% (285/306), spec 92.0% (424/461) excluding intermediates. LR+ 11.82, LR- 0.08. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- Sensitivity 94.7%
- Specificity 93.5%
- PPV 78.3%
- NPV 98.6%
- Cut-off 0.378 pg/mL: Positive >0.378; Negative <=0.223; Intermediate 0.223-0.378
Compared against: amyloid-PET · 238 people
Who was studied: Primary care subgroup (n=238); PET prevalence 21.4%
Strength of the evidence: High (regulatory pivotal, large) but retrospective and PPV/NPV prevalence-dependent
PRIMARY CARE 2x2: Positive 36 PET+/10 PET- (PPV 78.3%, 66.1-87.0; 19.3% of results); Intermediate 13/34 (19.7%); Negative 2/143 (60.9% of results; NPV 98.6%). Derived sens 94.7% (36/38), spec 93.5% (143/153) excluding intermediates. FDA text: PPV 78.3%, NPA 94.7% at upper cutoff; NPV 98.6% at lower cutoff. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary. FDA-defined single-cutoff agreements: lower cutoff PPA 96.1% (49/51), upper cutoff NPA 94.7% (177/187); 19.7% intermediate.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
Roche Elecsys Phospho-Tau (217P) Plasma (pTau217), commercial assay
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- Sensitivity 92.9%
- Specificity 91.2%
- PPV 90.2%
- NPV 93.7%
- Cut-off 0.378 pg/mL: Positive >0.378; Negative <=0.223; Intermediate 0.223-0.378
Compared against: amyloid-PET · 720 people
Who was studied: Secondary/tertiary care subgroup (n=720); PET prevalence 45.4%
Strength of the evidence: High (regulatory pivotal, large) but retrospective and PPV/NPV prevalence-dependent
SECONDARY/TERTIARY 2x2: Positive 249/27 (PPV 90.2%, 86.5-93.0; 38.3% of results); Intermediate 59/85 (20.0%); Negative 19/281 (41.7%; NPV 93.7%). Derived sens 92.9% (249/268), spec 91.2% (281/308) excluding intermediates. Reference = amyloid PET visual read, 6 readers (2 vendors x 3), majority vote; inter-reader PPA only 83.6% (range 44.3-100%), NPA 91.3%. 16 sites US/EU; 91.4% US; 84.7% white. Roche did not report AUC in the summary. FDA-defined: lower cutoff PPA 94.2%, upper cutoff NPA 93.1%; 20.0% intermediate.
Source: FDA 510(k) Summary K261686 pivotal study RD007230 (retrospective; APPROACH cohort RD006263 + AACI phase III screening samples) · 2026
C2N PrecivityAD2 (APS2), FDA-cleared IVD version
Measured
cleared
- Sample: plasma
- Platform: mass-spec
- Sensitivity 91.8%
- Specificity 92.7%
- PPV 90.5%
- NPV 93.7%
- Cut-off 66.5 APS2 score: Positive >66.5; Negative <27.5; Likely Positive 27.5-66.5
Compared against: amyloid-PET · 209 people
Who was studied: Subset of pooled validation cohort with amyloid PET visual read as reference (n=209; prevalence 45.9%)
Strength of the evidence: Moderate (small PET subset)
Against PET alone the numbers are lower than the pooled headline: Positive 67 PET+/7 PET- (PPV 90.5%, 81.7-95.3); Likely Positive 23/17 (PV 57.5%; 19.1% of results); Negative 6/89 (PV 6.3% => NPV 93.7%). Derived sens 91.8% (67/73), spec 92.7% (89/96). Compare CSF-comparator subset (n=933): PPV 99.1%, NPV 93.0%, Likely Positive PV 82.3%.
Source: FDA 510(k) Summary K253240, Table 53.25 (amyloid PET visual read comparator subset) · 2026
C2N PrecivityAD2 (APS2), FDA-cleared IVD version
Measured
cleared
- Sample: plasma
- Platform: mass-spec
- PPV 90.3%
- NPV 96%
- Cut-off 66.5 APS2 score: Positive >66.5; Negative <27.5; Likely Positive 27.5-66.5
Compared against: amyloid-PET · 242 people
Who was studied: SCD subgroup of pooled validation cohort (n=242; amyloid prevalence 24.0%)
Strength of the evidence: Moderate (subgroup)
Earliest symptomatic stage: Positive PV 90.3% (28/31; 75.1-96.7), Likely Positive PV 67.6% (23/34), Negative PV 4.0% (7/177) => NPV 96.0%; 14.0% Likely Positive. MCI (n=540, prev 53.9%): PPV 97.9%, NPV 91.4%. Dementia (n=360, prev 70.3%): PPV 98.5%, NPV 91.9%. Reference = PET visual read or CSF (pooled).
Source: FDA 510(k) Summary K253240, Table 53.27 (subjective cognitive decline subgroup) · 2026
Washington University / C2N mass-spec plasma %p-tau217
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- Sensitivity 91%
- Specificity 94%
- AUC 0.96
Compared against: amyloid-PET · 998 people
Who was studied: BioFINDER-2 (n=998, CU and CI, mean age 68.5); replicated in Knight ADRC/WashU (n=219)
Strength of the evidence: High (large head-to-head with PET reference, external replication)
Highest performer: accuracy 0.93, sensitivity 91%, specificity 94% for Aβ-PET status. AUC recorded as upper bound of the 0.91-0.96 range for all p-tau217 tests since abstract states %p-tau217 had significantly higher AUC than all immunoassays (Pdiff<0.007); exact point AUC is in full text, not abstract. Authors: MS %p-tau217 may serve as stand-alone confirmatory test; immunoassays better suited as triage tests.
Source: Warmenhoven et al. 2025, Brain (head-to-head of key plasma p-tau217 tests) · 2025 · PubMed 39468767 · doi:10.1093/brain/awae346
Eli Lilly MSD plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
Compared against: amyloid-PET · 998 people
Who was studied: BioFINDER-2 (n=998, CU and CI, mean age 68.5); replicated in Knight ADRC/WashU (n=219)
Strength of the evidence: High (large head-to-head with PET reference, external replication)
Immunoassays as a group: accuracy 0.83-0.88, sensitivity 84-87%, specificity 85-89% for Aβ-PET (assay-specific values not in abstract; null). Lilly and ALZpath had higher AUC than Janssen for Aβ-PET (Pdiff<0.006); Lilly outperformed ALZpath for tau-PET (Pdiff=0.025). All plasma p-tau217 tests AUC range 0.91-0.96.
Source: Warmenhoven et al. 2025, Brain (head-to-head of key plasma p-tau217 tests) · 2025 · PubMed 39468767 · doi:10.1093/brain/awae346
ALZpath pTau217 (Simoa)
Measured
Research use only
- Sample: plasma
- Platform: Simoa
Compared against: amyloid-PET · 998 people
Who was studied: BioFINDER-2 (n=998, CU and CI, mean age 68.5); replicated in Knight ADRC/WashU (n=219)
Strength of the evidence: High (large head-to-head with PET reference, external replication)
Immunoassays as a group: accuracy 0.83-0.88, sensitivity 84-87%, specificity 85-89% for Aβ-PET; assay-specific values not in abstract (not given). ALZpath AUC higher than Janssen for Aβ-PET; lower than Lilly for tau-PET.
Source: Warmenhoven et al. 2025, Brain (head-to-head of key plasma p-tau217 tests) · 2025 · PubMed 39468767 · doi:10.1093/brain/awae346
Janssen Simoa plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: Simoa
Compared against: amyloid-PET · 998 people
Who was studied: BioFINDER-2 (n=998, CU and CI, mean age 68.5); replicated in Knight ADRC/WashU (n=219)
Strength of the evidence: High (large head-to-head with PET reference, external replication)
Lowest AUC among the p-tau217 immunoassays for Aβ-PET status (significantly below Lilly and ALZpath, Pdiff<0.006). Assay-specific values not in abstract (not given); group range accuracy 0.83-0.88.
Source: Warmenhoven et al. 2025, Brain (head-to-head of key plasma p-tau217 tests) · 2025 · PubMed 39468767 · doi:10.1093/brain/awae346
Alamar NULISA plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: other
Compared against: amyloid-PET · 998 people
Who was studied: BioFINDER-2 (n=998, CU and CI, mean age 68.5); replicated in Knight ADRC/WashU (n=219)
Strength of the evidence: High (large head-to-head with PET reference, external replication)
Measured only in subsets of both cohorts; abstract states performance similar to the other immunoassays. No numbers in abstract (not given).
Source: Warmenhoven et al. 2025, Brain (head-to-head of key plasma p-tau217 tests) · 2025 · PubMed 39468767 · doi:10.1093/brain/awae346
Fujirebio Lumipulse G pTau217/β-Amyloid 1-42 Plasma Ratio
Measured
cleared
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 97.6%
- Specificity 90.8%
- PPV 91.8%
- NPV 97.3%
- Cut-off 0.007 pTau217/Aβ42 plasma ratio: Positive >=0.00738; Negative <=0.00370; Indeterminate 0.00371-0.00737
Compared against: amyloid-PET · 499 people
Who was studied: Multi-center clinical study, 499 plasma samples from cognitively impaired adults >=55 y (specialized care setting)
Strength of the evidence: High (regulatory pivotal study) but PPV/NPV depend on ~study prevalence
First FDA-cleared blood test for AD (510(k), Breakthrough Device; predicate = Lumipulse CSF Aβ42/40 ratio). Reference = amyloid PET or CSF. PPV 91.7%, NPV 97.3%; <20% of 499 patients received an indeterminate result (PPV/NPV exclude indeterminates). Sensitivity/specificity/AUC not stated in press release. Intended for symptomatic patients >=55 in specialized care; not a screening or stand-alone test. OpenFDA 510(k) record: K242706, decision date 2025-05-16, applicant Fujirebio Diagnostics, Inc. Predicate Lumipulse G β-Amyloid Ratio (1-42/1-40) CSF was De Novo DEN200072 (2022-05-04). FROM 510(k) SUMMARY K242706 pivotal 2x2 (n=499, PET or CSF reference, prevalence 51.1%): Positive 201 PET/CSF+ vs 18 neg (PV 91.8%, 95% CI 87.8-94.6); Indeterminate 49/49 (19.6% of results); Negative 5 pos vs 177 neg (PV of negative for positivity 2.7%, i.e. NPV 97.3%). Sensitivity 97.6% (201/206) and specificity 90.8% (177/195) are DERIVED here from the FDA 2x2 after excluding indeterminates; FDA summary itself reports predictive values and likelihood ratios (LR+ 10.68, LR- 0.03), not sens/spec. Cohorts: AriBio Polaris-AD phase III screening, BioFINDER-2 (Malmö), Bio-Hermes-001, WRAP. Mean age 72, 84.8% white; 58.2% clinically AD dementia, 40.7% SCD/MCI. Cutoffs derived in separate 208-sample set (PV+ 94.6%, prevalence 54.8%).
Source: FDA 510(k) clearance clinical study (FDA press release, 16 May 2025) · 2025
Plasma p-tau217, pooled across assays (systematic review/meta-analysis)
Modelled estimate
Research use only
- Sample: plasma
- Platform: other
- Sensitivity 82%
- Specificity 86%
Compared against: amyloid-PET
Who was studied: Pooled studies with PET reference
Strength of the evidence: High (systematic review) but heterogeneous assays/cutoffs
Pooled plasma p-tau217: sensitivity 82%, specificity 86% for amyloid PET; 83%/83% for tau PET. CSF p-tau217: 79%/91% amyloid. Accuracy lower in cognitively unimpaired than impaired. Pooled values are substantially below single-cohort headline AUCs of 0.95+.
Source: Khalafi et al. 2025, Alzheimer's & Dementia (meta-analysis, 30 studies, search to Aug 2024) · 2025 · PubMed 39711334 · doi:10.1002/alz.14458
Fujirebio Lumipulse G pTau217 plasma
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.92
Compared against: CSF · 148 people
Who was studied: Tallaght (Dublin) specialist memory service, 148 patients (101 MCI, 47 mild dementia), age 69.4
Strength of the evidence: Moderate (single center, small)
AUC 0.92 (0.87-0.97) for CSF-defined Aβ positivity; median p-tau217 >4-fold higher in Aβ+. Two-threshold approach could avoid >half of lumbar punctures.
Source: Dyer et al. 2025, Alzheimer's & Dementia: DADM (TIMC-BRAiN) · 2025 · PubMed 39959359 · doi:10.1002/dad2.70080
Fujirebio Lumipulse G plasma p-tau217/Aβ42 ratio
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.963
Compared against: amyloid-PET · 391 people
Who was studied: Clinic cohort, China (n=391)
Strength of the evidence: Moderate-high (PET reference, two settings; Chinese population)
Abstract gives AUC 0.963-0.966 across the two cohorts for Aβ-PET (lower bound assigned to clinic; per-cohort assignment not stated in abstract). Tau-PET AUC 0.947-0.974. Ratio outperformed p-tau217 alone, Aβ42/40, p-tau181; equivalent to CSF. Intermediate zone 10.6% (ratio) vs 13.0% (p-tau217 alone).
Source: Wang et al. 2025, Alzheimer's & Dementia (TBRAIN, China) · 2025 · PubMed 40156286 · doi:10.1002/alz.70038
Fujirebio Lumipulse G plasma p-tau217/Aβ42 ratio
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.963
Compared against: amyloid-PET · 121 people
Who was studied: Community cohort, China (n=121)
Strength of the evidence: Moderate-high (PET reference, two settings; Chinese population)
AUC range 0.963-0.966 (lower bound recorded). Intermediate-zone fraction 16.5% for ratio vs 31.4% for p-tau217 alone in community, i.e. P-tau217 alone leaves a third of community participants indeterminate. Chinese-population cutoffs established.
Source: Wang et al. 2025, Alzheimer's & Dementia (TBRAIN, China) · 2025 · PubMed 40156286 · doi:10.1002/alz.70038
Roche cobas e801 plasma p-tau217 (RUO, fully automated)
Measured
cleared
- Sample: plasma
- Platform: Elecsys
- AUC 0.94
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
Strength of the evidence: Low-moderate (conference abstract, interim)
AUC 0.94 (0.91-0.96) in full interim sample; 0.92 (0.89-0.96) in cognitively unimpaired. (Commercial Elecsys pTau217 later FDA-cleared K261686, 2026-08-19.) Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Siemens Healthineers Atellica IM plasma p-tau217 (RUO)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.92
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
Strength of the evidence: Low-moderate (conference abstract, interim)
AUC 0.92 (0.89-0.95). Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Beckman Coulter DxI 9000 Access plasma p-tau217 (RUO)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.92
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
Strength of the evidence: Low-moderate (conference abstract, interim)
AUC 0.92 (0.89-0.95). Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Fujirebio Lumipulse G1200 plasma p-tau217/Aβ42 ratio
Measured
cleared
- Sample: plasma
- Platform: Lumipulse
- AUC 0.91
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
Strength of the evidence: Low-moderate (conference abstract, interim)
AUC 0.91 (0.89-0.94) for ratio vs 0.90 (0.86-0.93) for p-tau217 alone. Note: in this largely cognitively unimpaired community cohort the FDA-cleared ratio scored lowest of the four platforms. Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Fujirebio Lumipulse G1200 plasma p-tau217 alone
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.9
Compared against: amyloid-PET · 676 people
Who was studied: WRAP (Wisconsin Registry for Alzheimer's Prevention); interim 2,821 plasma samples of planned 5,574 from 1,982 participants (1,746 CU, 197 CI); 676 with amyloid PET
Strength of the evidence: Low-moderate (conference abstract, interim)
AUC 0.90 (0.86-0.93). Amyloid PET cutoff 24.1 Centiloids. All four RUO fully automated assays similar in cognitively impaired (AUC >0.94). Industry co-authors (Roche, Beckman Coulter, Siemens).
Source: Denkinger et al. 2025, Alzheimer's & Dementia (AAIC 2025 conference abstract; interim analysis) · 2025 · doi:10.1002/alz70861_108882
Fujirebio Lumipulse G pTau217 plasma (single predefined cutoff)
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- PPV 82%
- NPV 88%
Compared against: CSF · 548 people
Who was studied: Swedish primary care cohort (n=548)
Strength of the evidence: High (large, multicohort, predefined cutoffs, includes primary care)
PRIMARY CARE with single cutoff: accuracy 85%, PPV 82%, NPV 88% - notably lower than the 91-92% PPV/NPV of the mass-spec APS2 test in the same setting (Palmqvist 2024). AUC not given separately for primary care in abstract (overall range 0.93-0.96). Reference = AD pathology by abnormal CSF Aβ42:p-tau181. Predefined cutoffs. Total study n=1,767 (4 secondary-care cohorts Malmö/Gothenburg/Barcelona/Brescia + Swedish primary care). Two-cutoff approach raised accuracy to 92-94% while excluding 12-17% intermediate. Accuracy 83% in age >=80. Mass-spec %p-tau217 had higher accuracy in primary care and was unaffected by age.
Source: Palmqvist et al. 2025, Nature Medicine · 2025 · PubMed 40205199 · doi:10.1038/s41591-025-03622-w
Fujirebio Lumipulse G pTau217 plasma (single predefined cutoff)
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.93
- PPV 89%
- NPV 77%
Compared against: CSF · 1,219 people
Who was studied: Four secondary-care cohorts: Malmö (n=337), Gothenburg (n=165), Barcelona (n=487), Brescia (n=230)
Strength of the evidence: High (large, multicohort, predefined cutoffs, includes primary care)
Secondary care: AUC 0.93-0.96 (lower bound recorded), accuracy 89-91%, PPV 89-95% (lower bound), NPV 77-90% (lower bound recorded). Lower NPV bound reflects high-prevalence memory-clinic setting. Reference = AD pathology by abnormal CSF Aβ42:p-tau181. Predefined cutoffs. Total study n=1,767 (4 secondary-care cohorts Malmö/Gothenburg/Barcelona/Brescia + Swedish primary care). Two-cutoff approach raised accuracy to 92-94% while excluding 12-17% intermediate. Accuracy 83% in age >=80. Mass-spec %p-tau217 had higher accuracy in primary care and was unaffected by age.
Source: Palmqvist et al. 2025, Nature Medicine · 2025 · PubMed 40205199 · doi:10.1038/s41591-025-03622-w
Washington University / C2N mass-spectrometry plasma %p-tau217 (p-tau217/np-tau217 ratio)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.95
- PPV 89%
- NPV 89%
Compared against: amyloid-PET · 1,759 people
Who was studied: BioFINDER-2 (n=1,422) and Knight ADRC (n=337); cognitively impaired subcohorts BioFINDER-2 n=720, Knight ADRC n=50
Strength of the evidence: High (two cohorts, PET reference), but research settings
AUC for amyloid-PET status reported as 'between 0.95 and 0.97' (lower bound recorded); tau-PET AUC 0.95-0.98. In cognitively impaired subcohorts accuracy/PPV/NPV 89-90% for amyloid-PET (lower bound recorded), 87-88% for tau-PET; 95% with a two-cutoff approach (intermediate zone excluded). Judged clinically equivalent to FDA-approved CSF Aβ42/40 and p-tau181/Aβ42 immunoassays. Research cohorts, not primary care. %p-tau217 is the basis of C2N PrecivityAD2 (LDT).
Source: Barthélemy et al. 2024, Nature Medicine · 2024 · PubMed 38382645 · doi:10.1038/s41591-024-02869-z
C2N PrecivityAD2 (APS2: mass-spec %p-tau217 + Aβ42/40)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.97
- PPV 91%
- NPV 92%
Compared against: CSF · 307 people
Who was studied: Swedish primary care, single-batch analysis (n=307)
Strength of the evidence: High (prospective, predefined cutoffs, real-world primary care)
AUC 0.97 (95% CI 0.95-0.99), PPV 91% (87-96), NPV 92% (87-96). Reference standard = AD pathology defined by abnormal CSF Aβ42/40 and p-tau217 (not amyloid-PET). Predefined cutoffs from an independent cohort. Overall diagnostic accuracy APS2 90% (88-92) vs %p-tau217 alone 90% (88-91): adding Aβ42/40 did not improve accuracy. Prevalence of AD pathology 50% in both settings; PPV/NPV will differ at other prevalences.
Source: Palmqvist et al. 2024, JAMA · 2024 · PubMed 39068545 · doi:10.1001/jama.2024.13855
C2N PrecivityAD2 (APS2: mass-spec %p-tau217 + Aβ42/40)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.96
- PPV 88%
- NPV 87%
Compared against: CSF · 300 people
Who was studied: Swedish secondary care (memory clinic), single-batch analysis (n=300)
Strength of the evidence: High (prospective, predefined cutoffs, real-world primary care)
AUC 0.96 (0.94-0.98), PPV 88% (83-93), NPV 87% (82-93). Reference standard = AD pathology defined by abnormal CSF Aβ42/40 and p-tau217 (not amyloid-PET). Predefined cutoffs from an independent cohort. Overall diagnostic accuracy APS2 90% (88-92) vs %p-tau217 alone 90% (88-91): adding Aβ42/40 did not improve accuracy. Prevalence of AD pathology 50% in both settings; PPV/NPV will differ at other prevalences.
Source: Palmqvist et al. 2024, JAMA · 2024 · PubMed 39068545 · doi:10.1001/jama.2024.13855
C2N PrecivityAD2 (APS2: mass-spec %p-tau217 + Aβ42/40)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.96
- PPV 88%
- NPV 90%
Compared against: CSF · 208 people
Who was studied: Swedish primary care, prospective biweekly analysis (n=208)
Strength of the evidence: High (prospective, predefined cutoffs, real-world primary care)
AUC 0.96 (0.94-0.98), PPV 88% (81-94), NPV 90% (84-96). Primary care physicians' clinical accuracy for AD was 61% (53-69) vs 91% (86-96) with APS2. Reference standard = AD pathology defined by abnormal CSF Aβ42/40 and p-tau217 (not amyloid-PET). Predefined cutoffs from an independent cohort. Overall diagnostic accuracy APS2 90% (88-92) vs %p-tau217 alone 90% (88-91): adding Aβ42/40 did not improve accuracy. Prevalence of AD pathology 50% in both settings; PPV/NPV will differ at other prevalences.
Source: Palmqvist et al. 2024, JAMA · 2024 · PubMed 39068545 · doi:10.1001/jama.2024.13855
C2N PrecivityAD2 (APS2: mass-spec %p-tau217 + Aβ42/40)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.97
- PPV 91%
- NPV 91%
Compared against: CSF · 398 people
Who was studied: Swedish secondary care (memory clinic), prospective biweekly analysis (n=398)
Strength of the evidence: High (prospective, predefined cutoffs, real-world primary care)
AUC 0.97 (0.95-0.98), PPV 91% (87-95), NPV 91% (87-95). Dementia specialists' clinical accuracy 73% (68-79) vs 91% (88-95) with APS2. Reference standard = AD pathology defined by abnormal CSF Aβ42/40 and p-tau217 (not amyloid-PET). Predefined cutoffs from an independent cohort. Overall diagnostic accuracy APS2 90% (88-92) vs %p-tau217 alone 90% (88-91): adding Aβ42/40 did not improve accuracy. Prevalence of AD pathology 50% in both settings; PPV/NPV will differ at other prevalences.
Source: Palmqvist et al. 2024, JAMA · 2024 · PubMed 39068545 · doi:10.1001/jama.2024.13855
ALZpath pTau217 (Simoa HD-X immunoassay)
Measured
Research use only
- Sample: plasma
- Platform: Simoa
- AUC 0.92
Compared against: amyloid-PET · 786 people
Who was studied: TRIAD (Montreal), WRAP (Wisconsin), SPIN (Barcelona); n=786, mean age 66.3
Strength of the evidence: Moderate-high (3 cohorts, but research volunteers)
AUC for elevated Aβ reported as range 0.92-0.96 across 3 cohorts (95% CI 0.89-0.99); lower bound recorded. Tau pathology AUC 0.93-0.97. Reference standard was PET or CSF depending on cohort. 3-range reference (intermediate zone) reduced confirmatory testing by ~80%. Research cohorts, mean age 66; not primary care.
Source: Ashton et al. 2024, JAMA Neurology · 2024 · PubMed 38252443 · doi:10.1001/jamaneurol.2023.5319
Eli Lilly MSD plasma p-tau217 (research assay)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- Sensitivity 94%
- Specificity 86%
- AUC 0.94
Compared against: amyloid-PET · 912 people
Who was studied: BioFINDER-2 participants with cognitive concerns (n=912; 80/20 train/test split); independent validation n=118
Strength of the evidence: Moderate-high
Test-set AUC 0.94 (95% CI 0.90-0.97) for Aβ positivity (Aβ-PET or CSF Aβ42/40). Sensitivity 0.94 (0.90-0.98) and specificity 0.86 (0.77-0.95) are for the 2-cut-point procedure where 17.1% ambiguous cases were resolved with CSF/PET, so they are not stand-alone blood-test values. Tau-PET high load among Aβ+: AUC 0.92 (0.86-0.97).
Source: Mattsson-Carlgren et al. 2024, JAMA Neurology · 2024 · PubMed 38048096 · doi:10.1001/jamaneurol.2023.4596
Fujirebio Lumipulse G pTau217 plasma (fully automated)
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.94
Compared against: CSF · 290 people
Who was studied: Sant Pau memory clinic, 290 consecutive patients with lumbar puncture (66 CU, 130 MCI, 94 dementia)
Strength of the evidence: Moderate (single center, consecutive real-world sample)
AUC 0.94 (95% CI 0.92-0.97) for CSF-defined amyloid positivity; 4-7% misclassification; two-threshold global accuracy >90% in symptomatic groups. CKD raised concentrations. Stand-alone p-tau217 assay, not the FDA-cleared p-tau217/Aβ42 ratio.
Source: Arranz et al. 2024, Alzheimer's Research & Therapy · 2024 · PubMed 38926773 · doi:10.1186/s13195-024-01513-9
Fujirebio Lumipulse G pTau217 and ALZpath pTau217 (two assays, Mayo Clinic clinical lab)
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 92%
- Specificity 96%
Compared against: amyloid-PET · 427 people
Who was studied: Mayo Clinic participants with MCI or mild dementia, n=427
Strength of the evidence: Moderate-high (clinical lab, PET reference)
IMPORTANT CAVEAT: with a single cutpoint NEITHER assay achieved >=90% in both sensitivity and specificity. The 92%/96% recorded are for a two-cutpoint approach that leaves 20% (Lumipulse) and 39% (ALZpath) of results indeterminate. Platform field set to Lumipulse but row covers both assays.
Source: Figdore et al. 2024, Alzheimer's & Dementia · 2024 · PubMed 39030981 · doi:10.1002/alz.14140
C2N PrecivityAD2 %p-tau217 (mass-spec)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- Sensitivity 88.5%
- Specificity 87.1%
- AUC 0.927
- PPV 86.7%
- NPV 88.8%
- Cut-off 4.06 % p-tau217 (positive if above)
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.927 (95% CI 0.900-0.955). Sensitivity/specificity recorded as PPA/NPA at cutoff 4.06%; overall accuracy 87.8%. Adding Aβ42/40 (APS2) gave AUC 0.929, not significantly better (p=0.54). 110/392 samples were below the assay's p-tau217 limit of detection. Cutoff 4.06% is Youden-optimal in this cohort (prevalence 48.7%).
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
C2N PrecivityAD2 APS2 (%p-tau217 + Aβ42/Aβ40 combined model)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- AUC 0.929
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.929 (0.902-0.956), highest of any test in the study; not superior to %p-tau217 alone (p=0.54).
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
C2N PrecivityAD2 p-tau217 concentration (mass-spec, non-ratio)
Measured
LDT-only
- Sample: plasma
- Platform: mass-spec
- Sensitivity 86.4%
- Specificity 87.1%
- AUC 0.916
- PPV 86.4%
- NPV 87.1%
- Cut-off 2.34 pg/mL (Youden cutoff)
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.916 (0.887-0.946). Ratio form (%p-tau217) performed slightly better. Table 4 (Youden cutoff 2.34 pg/mL): PPA 86.4%, NPA 87.1%, accuracy 86.7%, PPV 86.4%, NPV 87.1%.
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
Fujirebio Lumipulse p-tau217 + Aβ42/Aβ40 combined model
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.911
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.911 (0.882-0.940); not significantly different from C2N combined model (p=0.16). Note this is a p-tau217 + Aβ42/40 model, not the later FDA-cleared p-tau217/Aβ42 ratio.
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
Fujirebio Lumipulse G pTau217 plasma
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- Sensitivity 84.3%
- Specificity 82.1%
- AUC 0.896
- PPV 81.7%
- NPV 84.6%
- Cut-off 0.158 pg/mL (positive if above)
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.896 (0.864-0.928). At cutoff 0.158 pg/mL: PPA 84.3%, NPA 82.1%, accuracy 83.2%, PPV 81.7%, NPV 84.6%. Assay CV was 11% and 34% for high and low controls (higher imprecision than other assays). Table 4 accuracy 83.2%. Cutoff is Youden-optimal in this cohort.
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
ALZpath pTau217 (Quanterix Simoa)
Measured
Research use only
- Sample: plasma
- Platform: Simoa
- Sensitivity 84.3%
- Specificity 83.1%
- AUC 0.885
- PPV 82.6%
- NPV 84.8%
- Cut-off 0.444 pg/mL (Youden cutoff)
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.885 (0.851-0.920); significantly lower than C2N combined model (p=0.0006). Table 4: PPA 84.3%, NPA 83.1%, accuracy 83.7%, PPV 82.6%, NPV 84.8%.
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
Janssen LucentAD p-tau217 (Quanterix Simoa)
Measured
Research use only
- Sample: plasma
- Platform: Simoa
- Sensitivity 80.6%
- Specificity 83.1%
- AUC 0.882
- PPV 81.9%
- NPV 81.9%
- Cut-off 0.061 pg/mL (Youden cutoff)
Compared against: amyloid-PET · 392 people
Who was studied: ADNI, n=392 (median age 78.1; 48.7% amyloid-PET+; 49% cognitively impaired); Quanterix Neurology 4-Plex on subcohort n=342
Strength of the evidence: High (independent consortium, blinded head-to-head, PET reference); ADNI is a highly selected research cohort, 93% white
Unadjusted AUC for amyloid-PET status (Centiloid >20) from Table 3; AUC 0.882 (0.848-0.916); significantly lower than C2N (p=0.0006). Table 4: PPA 80.6%, NPA 83.1%, accuracy 81.9%, PPV 81.9%, NPV 81.9%.
Source: Schindler et al. 2024, Alzheimer's & Dementia (FNIH Biomarkers Consortium head-to-head of leading blood tests) · 2024 · PubMed 39394841 · doi:10.1002/alz.14315
Fujirebio Lumipulse G pTau217 plasma
Measured
Research use only
- Sample: plasma
- Platform: Lumipulse
- AUC 0.85
Compared against: CSF · 209 people
Who was studied: Population-based volunteers >55 y without cognitive impairment, Cantabria, Spain (n=209, mean age 64)
Strength of the evidence: Moderate (single center, small)
PRECLINICAL / cognitively unimpaired: AUC 0.85 for CSF amyloid positivity (p-tau217 + APOE4 as predictors). Median p-tau217 0.26 pg/mL in A+ vs 0.12 in A-. Markedly lower than the 0.94-0.97 AUCs in symptomatic memory-clinic cohorts.
Source: Martínez-Dubarbie et al. 2024, Journal of Prevention of Alzheimer's Disease (Valdecilla cohort) · 2024 · PubMed 39559871 · doi:10.14283/jpad.2024.152
Washington University mass-spectrometry plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: mass-spec
- AUC 0.947
Compared against: CSF · 135 people
Who was studied: BioFINDER-1 MCI patients (n=135; 71 Aβ-positive by CSF Aβ42/40; mean age 72.4)
Strength of the evidence: Moderate (single small MCI cohort; head-to-head design is the strength)
Detection of abnormal Aβ status (CSF Aβ42/40 reference): AUC 0.947, significantly better than all 9 other assays (Pdiff<0.015). AUC for progression to AD dementia 0.932. Plasma-CSF correlation R=0.891.
Source: Janelidze et al. 2023, Brain (head-to-head of 10 plasma p-tau assays) · 2023 · PubMed 36087307 · doi:10.1093/brain/awac333
Eli Lilly MSD plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.886
Compared against: CSF · 135 people
Who was studied: BioFINDER-1 MCI patients (n=135; 71 Aβ-positive by CSF Aβ42/40; mean age 72.4)
Strength of the evidence: Moderate (single small MCI cohort; head-to-head design is the strength)
Abstract gives AUC range 0.886-0.889 across the two outcomes (abnormal Aβ status; progression to AD dementia); lower bound recorded. Best of the immunoassays; not significantly different from Janssen p-tau217, ADx p-tau181, WashU p-tau181 (AUC 0.835-0.872).
Source: Janelidze et al. 2023, Brain (head-to-head of 10 plasma p-tau assays) · 2023 · PubMed 36087307 · doi:10.1093/brain/awac333
Janssen Simoa plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: Simoa
Compared against: CSF · 135 people
Who was studied: BioFINDER-1 MCI patients (n=135; 71 Aβ-positive by CSF Aβ42/40; mean age 72.4)
Strength of the evidence: Moderate (single small MCI cohort; head-to-head design is the strength)
Abstract reports only that Janssen p-tau217 AUC fell in the range 0.835-0.872 (grouped with ADx p-tau181 and WashU p-tau181) and was not significantly different from Lilly p-tau217; assay-specific AUC not in abstract (not given).
Source: Janelidze et al. 2023, Brain (head-to-head of 10 plasma p-tau assays) · 2023 · PubMed 36087307 · doi:10.1093/brain/awac333
Eli Lilly MSD plasma p-tau217 (model with age + APOE ε4)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.943
Compared against: amyloid-PET · 348 people
Who was studied: BioFINDER-1 (development) and BioFINDER-2 (validation) MCI patients, n=348 combined; TRIAD n=84 secondary
Strength of the evidence: Moderate-high (validated in independent cohort; MCI only)
AUC 0.893 in BioFINDER-1 (development), 0.943 in BioFINDER-2 (validation; recorded). Model includes age and APOE ε4, not p-tau217 alone. Two-step workflow overall accuracy 88.2/90.5/92.0% with lenient/moderate/stringent thresholds while reducing CSF tests by 85.9/72.7/61.2%.
Source: Brum et al. 2023, Nature Aging (two-step workflow) · 2023 · PubMed 37653254 · doi:10.1038/s43587-023-00471-5
Eli Lilly MSD plasma p-tau217 (research assay)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
Compared against: amyloid-PET · 1,329 people
Who was studied: Mayo Clinic Study of Aging (population-based, age 30-98)
Strength of the evidence: High (population-based)
KEY COUNTER-EVIDENCE to headline accuracy: in an unselected community sample, continuous unadjusted plasma p-tau181 and p-tau217 predicted abnormal amyloid-PET with AUROC only 0.81-0.86 (range given jointly for both analytes in abstract; AUC field left blank because the abstract does not give a p-tau217-specific point value). Tau-PET temporal ROI AUROC <0.70. Chronic kidney disease raised p-tau levels by an amount similar to the amyloid-positive vs -negative difference.
Source: Mielke et al. 2022, Nature Medicine · 2022 · PubMed 35618838 · doi:10.1038/s41591-022-01822-2
Eli Lilly MSD plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.91
Compared against: amyloid-PET · 146 people
Who was studied: UCSF Memory and Aging Center (n=443) and ARTFL consortium (n=150); total n=593, mean age 64
Strength of the evidence: Moderate-high (multicohort; young cohort mean age 64, FTLD-enriched)
Amyloid-PET positivity AUC 0.91 (95% CI 0.88-0.94), n=146; vs p-tau181 0.89 (Pdiff=0.049). AD syndromes vs CU: AUC 0.98; AD syndromes vs FTLD syndromes 0.93 (0.91-0.96).
Source: Thijssen et al. 2021, Lancet Neurology · 2021 · PubMed 34418401 · doi:10.1016/s1474-4422(21)00214-3
Eli Lilly MSD plasma p-tau217
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.96
Compared against: autopsy · 83 people
Who was studied: UCSF Memory and Aging Center (n=443) and ARTFL consortium (n=150); total n=593, mean age 64
Strength of the evidence: Moderate-high (multicohort; young cohort mean age 64, FTLD-enriched)
Pathology-confirmed AD (n=15) vs pathology-confirmed FTLD (n=68): AUC 0.96 (0.92-1.00) for p-tau217 vs 0.91 (0.82-1.00) for p-tau181 (Pdiff=0.22, not significant). Very small autopsy sample.
Source: Thijssen et al. 2021, Lancet Neurology · 2021 · PubMed 34418401 · doi:10.1016/s1474-4422(21)00214-3
Eli Lilly MSD plasma p-tau217 (research assay)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.89
Compared against: autopsy · 81 people
Who was studied: Arizona neuropathology cohort (Banner Sun Health; cohort 1: 34 AD, 47 non-AD)
Strength of the evidence: Moderate (small autopsy cohort, selected)
Antemortem plasma p-tau217 vs neuropathologically defined AD; AUC 0.89 (95% CI 0.81-0.97). Plasma p-tau181 and NfL AUC range 0.50-0.72 in same cohort. Mean age 83.5 y. Sensitivity/specificity not reported in abstract.
Source: Palmqvist et al. 2020, JAMA · 2020 · PubMed 32722745 · doi:10.1001/jama.2020.12134
Eli Lilly MSD plasma p-tau217 (research assay)
Measured
Research use only
- Sample: plasma
- Platform: immunoassay
- AUC 0.96
Compared against: clinical-diagnosis · 699 people
Who was studied: Swedish BioFINDER-2 (cohort 2: 301 CU, 178 MCI, 121 AD dementia, 99 other neurodegenerative)
Strength of the evidence: Moderate-high (large research cohort; selected participants)
Clinical AD dementia vs other neurodegenerative diseases: AUC 0.96 (95% CI 0.93-0.98). Not significantly different from CSF p-tau217, CSF p-tau181 or tau-PET (AUC 0.90-0.99). Abnormal vs normal tau-PET: AUC 0.93 (0.91-0.96). This is the origin of the 'approaches PET/CSF accuracy' claim; note the comparison is dementia-stage AD vs other neurodegenerative disease in a research cohort, not amyloid-PET status in an unselected population.
Source: Palmqvist et al. 2020, JAMA · 2020 · PubMed 32722745 · doi:10.1001/jama.2020.12134