Considerations for Interpretation of Research Use Only (RUO)-Generated Data Relative to the FDA-Cleared Lumipulse® G pTau217/β-Amyloid 1-42 Plasma Ratio In Vitro Diagnostic (IVD) Test

Data generated using RUO versions of the Lumipulse® G pTau217 Plasma assay and the Lumipulse® G β-Amyloid 1-42-N Plasma assay should not be interpreted using the thresholds established for the FDA-cleared Lumipulse® G pTau217/β-Amyloid 1-42 Plasma Ratio IVD test. These thresholds were derived and validated within the context of the test system (assay components and platform), its analytical performance characteristics, and the intended use population defined in the Lumipulse G pTau217/β-Amyloid 1-42 Plasma Ratio instructions for use.

Application of thresholds established for the Lumipulse® G pTau217/β-Amyloid 1-42 IVD Plasma Ratio to RUO-generated data, particularly from sample sets that may not fully reflect the intended use population and conditions of use, may result in misclassification and does not represent the clinical performance characteristics of the FDA cleared test.

During the recent product hold and re-release process, the underlying issue affecting the Lumipulse G pTau217 Plasma assay component of the ratio test was identified and corrected. In this context, results generated using RUO configurations of the Lumipulse G pTau217 Plasma assay may not be directly comparable to those obtained with the current FDA-cleared Lumipulse G pTau217/β-Amyloid 1-42 Plasma Ratio IVD test, and caution is warranted when interpreting such data.

Importantly, ongoing and previously generated datasets using RUO assays continue to represent valuable, scientifically meaningful, and internally consistent research resources within their intended use context.

Note: This announcement applies to the dataset labeled, “UPENN – Plasma Biomarkers (AB42, AB40, ptau217, NfL, GFAP) measured by Fujirebio & Quanterix” in the ADNI-LONI repository.

A bug has been identified in a fraction of de-faced Magnitude DICOM series for multi-echo GRE (ME-T2starw) MRI sequences, wherein volumes were assembled in the incorrect order with incorrect echo times. We have included a list of affected images. For ADNI, this was 140 image series, roughly 16% of MEGRE magnitude series available. Users who have downloaded affected image IDs, or QSM analyses containing them, should delete their copies and any downstream analyses. We are in the process of removing all affected images from LONI and uploading corrected DICOM to replace them, in the next 1-2 weeks. These replacements will have new LONI ImageIDs. We apologize for not catching this error, and we thank the Michael Ewers lab at LMU Munich for bring it to our attention. Thank you for your understanding!

To obtain a listing of removed image series, please download from the IDA, the dataset labeled “Data Advisory: Listing of removed multi-echo GRE (ME-T2starw) images.

Summary:

Data collected from 78 ADNI participants has been removed from the IDA.

Description:

A recent review of ADNI data raised a serious data quality concern.

ADNI leadership reviewed the data in question, and after extensive analyses and discussion, made the decision to remove the data from PTIDs with the format 381_S_#####.

Of particular concern was the clinically acquired cognitive and functional assessment data. Due to the critical nature of those results for interpreting ADNI biomarker data, the decision at this time has been to remove all data associated with those PTIDs.

All affiliated data from those 78 PTIDs have been removed from the LONI-IDA ADNI data repository.

Recommendations:

ADNI leadership strongly advises that data with these PTIDs not be used in any analysis, submission of abstracts, or publications. 

Furthermore, the ADNI leadership requests that any abstracts or papers which have been submitted using these data be appropriately revised due to concerns with the data quality.  Finally, a reminder that the ADNI Data Use Agreement states that no efforts can be made to identify the ADNI participants or ADNI site locations.