Search Results

This test is to be used as an aid in diagnosis of diabetes and as an aid in identifying patients who may be at risk for developing diabetes. The cobas c 501 Tina-quant HbA1cDx Gen.3 assay is an in vitro diagnostics reagent system intended for quantitative determination of mmol/mol hemoglobin Alc (IFCC) and % hemoglobin A1c (DCCT/NGSP) in whole blood on the Roche/Hitachi cobas c 501 clinical chemistry analyzer.

Whole blood samples are placed on the analyzer. The anticoagulated whole blood is hemolyzed on board the analyzer prior to determination of HbA1c by this turbidimetric inhibition immunoassay. Liberated hemoglobin in the hemolyzed sample is converted to a derivative having a characteristic absorption spectrum and measured bichromatically. The instrument first measures hemoglobin (Hb) and glycohemoglobin (HbA1c) in terms of either g/dL or mmol/L, then calculates the % HbA1c from the HbA1c/Hb ratio according to a user-selected protocol, either IFCC or NGSP protocols.

The Roche cobas c 501 Tina-quant HbA1cDx Gen.3 assay is an in vitro diagnostic reagent system intended for the quantitative determination of HbA1c.

Here's an analysis of the acceptance criteria and the study that proves the device meets them:

1. Table of Acceptance Criteria and Reported Device Performance

The submission does not explicitly state formal acceptance criteria thresholds in terms of specific performance metrics (e.g., "Total Error must be <= X%"). Instead, it presents various analytical performance characteristics and compares them to the predicate device, or indicates "no significant interference." The overall conclusion from Roche is that the device is "substantially equivalent to the predicate device" based on these characteristics.

However, based on the data presented and common analytical performance expectations for such devices, we can infer some criteria and compare the reported performance:

2. Sample Sizes Used for the Test Set and Data Provenance

• Precision Study:

  • Test Set: Multiple human samples (4 distinct samples covering a range of HbA1c levels) and 2 control samples. Each sample was run in duplicate, twice a day for 21 days on 3 analyzers, using 3 reagent lots. This results in significant numbers of individual measurements per sample (2 samples/day * 21 days * 3 analyzers * 3 reagent lots = 378 data points per sample type, then additional replicates per run).
  • Data Provenance: Not explicitly stated, but typically these types of analytical validation studies are conducted internally by the manufacturer or by contracted clinical research organizations, likely in a controlled laboratory setting. It is prospective data collection for the specific purpose of validating the device.

• Method Comparison Study:

  • Test Set: 141 variant-free samples.
  • Data Provenance: Not explicitly stated, but "variant-free samples" suggests specific selection, likely from a diverse patient population to cover the reported HbA1c range. This would be prospective data collected for the method comparison.

• Endogenous Interference Study:

  • Test Set: 6 endogenous substances. For each substance, 2 HbA1c levels were tested, each with a dilution series (varying concentrations of the interferent). Each sample in the dilution series was tested 10-fold.
  • Data Provenance: Whole blood sample pools were spiked with interferents. This is an experimental, prospective design in a laboratory setting.

• Drug Interference Study:

  • Test Set: 16 drugs. For each drug, 2 HbA1c levels were tested, each at 2 concentrations of the drug. Additionally, a reference sample with no drug. Total of 64 samples prepared. Each sample was tested 10-fold.
  • Data Provenance: Whole blood samples were spiked with drugs. Experimental, prospective design.

• Hemoglobin Derivatives Cross-Reactivity Study:

  • Test Set: 6 hemoglobin derivatives. For each derivative, 2 HbA1c levels, with dilution series. Each sample tested 10-fold.
  • Data Provenance: Whole blood sample pools. Experimental, prospective design.

• Hemoglobin Variants Interference Study:

  • Test Set: 116 samples containing one of six common hemoglobin variants (S, C, E, D, F, A2).
  • Data Provenance: Samples categorized by Hb variant and their concentration ranges. Not explicitly stated regarding country or retrospective/prospective, but given the specific variant types, these are likely clinical samples collected from populations with higher prevalence of these variants, and used prospectively for this study.

• LoB/LoD Study:

  • Test Set: One analyte-free sample (for LoB) and five low-analyte samples (for LoD).
  • Data Provenance: Experimental, prospective design.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of those Experts

• For this type of in vitro diagnostic device (chemical analyzer for HbA1c), the "ground truth" for the test set is established by reference methods and analytical specifications, rather than expert human interpretation of images or clinical cases.
• Method Comparison: The ground truth for the 141 variant-free samples was established using the Tosoh HPLC, a secondary NGSP reference laboratory method. The NGSP (National Glycohemoglobin Standardization Program) standardizes HbA1c test results to those of the Diabetes Control and Complications Trial (DCCT), involving rigorous certification for laboratories and methods.
• Hemoglobin Variants Interference Study: Ground truth was established using an NGSP reference method "that is known to be free from the hemoglobin interference being tested." This implies a certified and validated method capable of accurately measuring HbA1c in the presence of specific variants.
• Other studies (Precision, Interference, LoB/LoD, Linearity): The ground truth for these studies is the inherent analytical characteristic of the sample itself (e.g., true concentration, absence/presence of interferent), measured or prepared precisely according to established laboratory standards and protocols (e.g., CLSI guidelines). No human "experts" are involved in establishing this ground truth value itself, but rather experts in analytical chemistry and laboratory medicine design and execute these studies to rigorous standards.

4. Adjudication Method for the Test Set

• For an in vitro diagnostic device like this, "adjudication" in the sense of comparing multiple human expert interpretations is not applicable.
• The "adjudication" is instead the comparison of the device's results against established reference methods or known concentrations/parameters of samples, usually with statistical analyses (e.g., bias calculation in method comparison).
• For the interference studies, a 7% deviation from an initial value (uninterfered sample) was the threshold for "significant interference." This acts as a predefined decision rule.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• No, a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not done. This type of study is relevant for imaging devices or other diagnostic tools that require human interpretation (e.g., radiologists reading images) and where AI is used to assist the human reader.
• This device is an automated in vitro diagnostic system that produces a quantitative numerical result for HbA1c. There is no human "reader" in the diagnostic interpretation loop in the way there would be for an imaging study.

6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study

• Yes, the entire study focuses on the standalone performance of the cobas c 501 Tina-quant HbA1cDx Gen.3 assay.
• All reported performance metrics (precision, method comparison, interference, linearity, detection limits) reflect the analytical performance of the device itself, without any human intervention in interpreting the direct HbA1c result.

7. Type of Ground Truth Used

• The ground truth used primarily falls under reference method comparison and defined analytical characteristics of known samples:
  • Reference Method Comparison: For method comparison and Hb variant interference, the Tosoh HPLC and other NGSP-certified reference methods served as the ground truth.
  • Defined Analytical Characteristics: For precision, interference (endogenous, drug, Hb derivatives), LoB/LoD, and linearity studies, the ground truth was based on precisely prepared samples with known (or expected) physiological concentrations of analytes and interferents, or the inherent analytical properties of the samples, measured against established analytical standards.

8. Sample Size for the Training Set

• This submission describes a validation study for a new generation of an already existing assay/device, not the development or training of a de novo AI algorithm.
• Therefore, the concept of a "training set" in the context of machine learning (where an algorithm learns from labeled data) is not directly applicable here. This is a traditional analytical validation for an in vitro diagnostic assay. The instrument's internal "algorithms" for calculation are based on established chemical principles and calibration, not machine learning from a large, labeled dataset in the modern AI sense.

9. How the Ground Truth for the Training Set was Established

• As explained in point 8, there isn't a "training set" in the common AI/machine learning sense for this device.
• The device's operational parameters (e.g., calibration curves, reaction kinetics) would have been established during its development phase using calibrators and control materials traced to international reference standards (e.g., IFCC and DCCT/NGSP), ensuring accuracy and consistency. These are foundational for any in vitro diagnostic test.

Ask a specific question about this device