(27 days)
Elecsys ß-Amyloid (1-42) CSF II and Elecsys Total-Tau CSF are in vitro electrochemiluminescence immunoassays for the measurement of the ß-Amyloid (1-42) (Abeta42) and Total-Tau (tTau) concentrations in cerebrospinal fluid (CSF) from adult patients aged 55 years and older being evaluated for Alzheimer's disease (AD) and other causes of cognitive impairment to generate a tTau/Abeta42 ratio value.
A negative result, defined as tTau/Abeta42 ratio value below cut-off or an Abeta42 value above the measuring range, is consistent with a negative amyloid position emission tomography (PET) scan result. A negative result reduces the likelihood that a patient's cognitive impairment is due to AD. A postive result, defined as tTau/Abeta42 ratio value above cut-off, is consistent with a positive amyloid PET scan result does not establish a diagnosis of AD or other cognitive disorder. The tTau/Abeta42 ratio result is used as an adjunct to other clinical diagnostic evaluations.
Limitations of Use
The performance of the tTau/Abeta42 ratio has not been established for:
· Predicting development of dementia or other neurologic conditions
· Monitoring responses to therapies
The Elecsys ß-Amyloid (1-42) II immunoassay makes use of a two-step, double antigen sandwich principle using a biotinylated monoclonal β-Amyloid (1-42)-specific and monoclonal ß-Amyloid (1-42) antibodies labeled with a ruthenium complex. The Elecsys ß-Amyloid (1-42) II immunoassay is intended for the in vitro quantitative determination of ß-Amyloid (1-42) in human CSF.
Similarly, the Elecsys Total-Tau immunoassay makes use of a two-step, double antigen sandwich principle using two biotinylated monoclonal Tau-specific antibodies (5.28.464 and 4.35.411) and a monoclonal Tau-specific (PC1P6) antibody labeled with a ruthenium complex. The Elecsys Total-Tau immunoassay is intended for the in vitro quantitative determination of Total-Tau in human CSF.
The assays are indicated for use with adult subjects being evaluated for Alzheimer's disease and other causes of cognitive impairment.
The results of the measurement of a CSF sample for each analyte in pg/mL is used in a ratio (tTau/Abeta42). The assays manually calculated ratio result is compared to a cutoff to determine if the test result indicates an amyloid PET positive or negative result. The result does not establish a diagnosis of AD or other cognitive disorders. Additionally, the result does not predict development of dementia or other neurologic disorders, nor does the result monitor responses to therapies. Both assays are intended for use on the cobas e 601 immunoassay analyzer.
Results for each assay are determined by an analyte specific calibration curve generated by 2point calibration and a master curve provided via the reagent barcode or e-barcode.
Here's an analysis of the provided text to extract information about the acceptance criteria and the study proving the device meets them:
Device: Elecsys B-Amyloid (1-42) CSF II and Elecsys Total-Tau CSF (used to generate a tTau/Abeta42 ratio value).
Intended Use: In vitro electrochemiluminescence immunoassays for the measurement of Abeta42 and tTau concentrations in CSF from adult patients aged 55 years and older being evaluated for Alzheimer's disease (AD) and other causes of cognitive impairment. The ratio result is used as an adjunct to other clinical diagnostic evaluations, consistent with amyloid PET scan results.
1. Table of Acceptance Criteria and Reported Device Performance
The document describes two separate phases of evaluation for the tTau/Abeta42 ratio: a BioFINDER cutoff setting cohort (using first-generation assays) and a clinical validation in an ADNI cohort (using a specific version of the assays, either first or second generation, with adjustments). The acceptance criteria are implicitly derived from the reported performance in the ADNI validation study which is stated to have "met the acceptance criteria."
Acceptance Criteria and Reported Device Performance (Clinical Validation in ADNI cohort):
| Acceptance Criterion (Implicit) | Reported Device Performance (ADNI Cohort) |
|---|---|
| PPA (Positive Percent Agreement) with amyloid PET classification at pre-specified cut-off | 85.0% (95% CI: 80.9% - 88.4%) at a pre-specified ratio-cutoff of 0.33. |
| NPA (Negative Percent Agreement) with amyloid PET classification at pre-specified cut-off | 94.0% (95% CI: 90.7% - 96.2%) at a pre-specified ratio-cutoff of 0.33. |
| Overall Percent Agreement (OPA) with amyloid PET classification | 89.2% (576 of 646 individuals had concordant predictions for amyloid status). (While OPA is a commonly reported metric and can be inferred, the document explicitly states PPA and NPA as primary results for the validation meeting acceptance criteria and provides the raw concordant count. For the BioFINDER cutoff setting cohort, OPA was explicitly 89.9% (95% CI: 85.7% to 93.2%) for a cutoff of 0.26). |
| Performance with Abeta42 values above the measuring range | A negative result is defined as tTau/Abeta42 ratio value below cut-off or an Abeta42 value above the measuring range, consistent with a negative amyloid PET scan result. This is an acceptance criterion by definition of valid output range, not a performance metric itself from the clinical study results. |
| LoB, LoD, LoQ, Linearity, Hook Effect, HAMA, Endogenous Interferences, Exogenous (drugs) Interferences, Cross Reactivity / Analytical Specificity, Stability Studies (Lot Calibration, On-Board Calibration, Reagent Stability after first opening, Reagent On-Board Stability, Open Rackpack On-Board Stability, Reagent Shelf-life Stability) for both individual assays | All samples/compounds met the predetermined acceptance criteria and/or specification. (Details are given for each in the "Non-Clinical Tests Summary" section for Elecsys B-Amyloid (1-42) CSF II and Elecsys Total-Tau CSF separately. These are non-clinical, analytical performance criteria rather than acceptance criteria for the clinical utility of the ratio.) |
2. Sample Size Used for the Test Set and Data Provenance
-
Test Set (Clinical Validation in ADNI cohort):
- Sample Size: 646 participants.
- Data Provenance: Retrospectively collected CSF samples from the Alzheimer's Disease Neuroimaging Initiative (ADNI-GO and ADNI2) studies.
- Country of Origin: The ADNI study is primarily based in the United States and Canada.
- Subject Characteristics: Patients with significant memory concerns (SMC, N = 94), early MCI (N = 272), late MCI (N = 152), and Alzheimer's Disease (AD, N = 128). Average age 72 years (range 55-91), 46% female / 54% male, 50% ApoE4 carriers / 50% non-carriers.
-
BioFINDER Cohort (for initial cutoff setting):
- Sample Size: 277 participants with mild cognitive symptoms.
- Data Provenance: Retrospective samples from the Swedish BioFINDER1 study.
- Country of Origin: Sweden.
- Subject Characteristics: 120 subjective cognitive decline (SCD), 153 mild cognitive impairment (MCI), 4 not assigned.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications
- Number of Experts: 3 trained readers.
- Qualifications of Experts: They were "trained readers." No specific professional qualifications (e.g., "radiologist with 10 years of experience") are provided beyond being "trained." They were blinded to clinical information, diagnosis, and CSF biomarker measurements.
4. Adjudication Method for the Test Set
- Adjudication Method (for Amyloid PET scans): Majority voting was used to classify each image as amyloid positive or negative.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done
- No, a "multi-reader multi-case (MRMC) comparative effectiveness study" involving human readers improving with AI vs. without AI assistance was not done or described. This study focuses on the diagnostic performance of the device (a blood test generating a ratio) against an established ground truth (amyloid PET scan results). The "trained readers" mentioned were establishing the ground truth (PET read), not evaluating the device's impact on their own performance.
6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done
- Yes, this study primarily represents a standalone performance evaluation of the device. The device (Elecsys B-Amyloid (1-42) CSF II and Elecsys Total-Tau CSF assay system) produces a tTau/Abeta42 ratio. This ratio is then compared directly to the amyloid PET scan status. There is no human-in-the-loop component for the device's interpretation; it provides a direct output (ratio) that is then compared to a cut-off to determine a positive or negative amyloid status.
7. The Type of Ground Truth Used
- The primary ground truth used for the clinical validation of the tTau/Abeta42 ratio was amyloid Positron Emission Tomography (PET) scan results, classified as amyloid positive or negative via visual read and majority voting by 3 trained readers.
- For the initial cutoff setting (BioFINDER cohort), the ground truth was also amyloid PET scan results ([18F]-Flutemetamol PET) by visual read.
8. The Sample Size for the Training Set
The document does not explicitly describe a separate "training set" for an AI algorithm in the traditional sense. It describes:
- Initial Cutoff Setting Cohort (BioFINDER): This cohort of 277 participants was used to define the initial ratio cutoff (0.26) based on agreement with amyloid PET status. While not a "training set" for an ML model, it served a similar purpose in establishing initial parameters for the assay's interpretation.
- Bridging Studies: Two bridging studies were conducted (N=20/17 and N=25/24) to adjust the cutoff based on pre-analytical differences and assay updates. These are not typical "training sets" for an algorithm, but rather calibration/normalization studies for the assay system.
Therefore, there isn't a "training set" for an AI model in this context, as the device is an in-vitro diagnostic immunoassay system, not an AI algorithm learning from data. The cut-off and the adjustments resemble empirical derivation from study data ("training" the cut-off, not an AI).
9. How the Ground Truth for the Training Set Was Established
As noted above, there isn't a "training set" for an AI algorithm. However, for the data used to establish the cutoff for the assay:
- BioFINDER Cohort: The ground truth was established by amyloid PET scan results via visual read. The specifics of how many readers or their adjudication for this initial phase are not detailed, but it's implied to be based on clinical interpretation of PET scans.
- Conversion Factors/Bridging Studies: The ground truth for these studies involved paired CSF samples handled under different pre-analytical protocols or measured with different assay generations. The "ground truth" here is the comparative measurement of samples under controlled conditions to determine conversion factors.
N/A