Search Results

The CYFRA 21-1 EIA kit is intended for the quantitative determination of soluble cytokeratin 19 fragments in human serum. The assay is to be used as an aid in monitoring disease progression during the course of disease and treatment in lung cancer patients. Serial testing for patient CYFRA 21-1 assay values should be used in conjunction with other clinical methods used for monitoring lung cancer.

The CYFRA 21-1 EIA is a solid phase, non-competitive immunoassay based on two monoclonal antibodies (derived from mice) directed against two separate antigenic determinants of soluble fragments of cytokeratin 19. Calibrators, controls and patient samples are incubated together with biotinylated Anti-CYFRA 21-1 MAb and horseradish peroxidase (HRP) labeled Anti-CYFRA 21-1 MAb in streptavidin coated micro strips. After washing, buffered Substrate/Chromogen reagent (hydrogen peroxide and 3, 3', 5, 5' tetramethylbenzidine) is added to each well and the enzyme reaction is allowed to proceed. During the enzyme reaction a blue color will develop if antigen is present. The intensity of the color development is proportional to the amount of CYFRA 21-1 present in the samples. The color intensity is determined in a microplate spectrophotometer at 620 nm (or optionally at 405 nm after addition of Stop Solution). Calibration curves are constructed for each assay by plotting absorbance value versus the concentration for each calibrator. The CYFRA 21-1 concentrations of patient samples are then read from the calibration curve.

Here's an analysis of the provided text regarding the CYFRA 21-1 EIA Kit, focusing on acceptance criteria and the supporting study:

Acceptance Criteria and Device Performance

The document does not present explicit, pre-defined "acceptance criteria" in a typical numerical format (e.g., "Sensitivity must be >X%"). Instead, it describes performance characteristics, and the "acceptance" is implied by the FDA's clearance based on demonstrated substantial equivalence to a predicate device and adherence to relevant guidelines.

However, based on the studies presented, we can infer performance metrics that were deemed acceptable for clearance. The key study provided is a clinical study for monitoring disease progression in lung cancer.

Table of Performance Metrics

Study Details

The primary study mentioned proving the device meets its intended use is a retrospective clinical study focused on monitoring the course of disease in lung cancer patients.

1. Sample size used for the test set and the data provenance:

   • Sample Size: 100 lung cancer patients, yielding a total of 314 observation pairs of serial serum samples. An average of 4.1 observations per patient.
   • Data Provenance: Retrospective clinical study. Patient samples were collected from a tertiary cancer center. The country of origin is not explicitly stated, but the submission is to the US FDA, implying the study was likely conducted in the US or under comparable standards.

2. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:

   • The document does not specify the number of experts or their qualifications used to establish the ground truth for disease progression/no progression. It states that "Changes in CYFRA 21-1 levels in serial serum samples... were compared to changes in disease status." This implies clinical assessment of disease status was the ground truth, likely determined by treating physicians or a review board at the tertiary cancer center, but no details are provided.

3. Adjudication method for the test set:

   • The document does not describe an explicit adjudication method (like 2+1 or 3+1) for establishing disease status. The "changes in disease status" used for ground truth would typically be based on standard clinical practices (imaging, biopsy, clinical examination), but the specific process for confirming or adjudicating these statuses for the study is not outlined.

4. If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance:

   • No, an MRMC comparative effectiveness study was not performed. This device is an in vitro diagnostic (IVD) assay (a blood test), not an imaging-based AI system that would typically involve human readers. Therefore, the concept of "human readers improve with AI vs without AI assistance" does not apply here.

5. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:

   • Yes, a standalone performance was done in the sense that the assay results (CYFRA 21-1 concentrations) were directly compared to the established disease status without an intermediate "human-in-the-loop" interpretation of the assay values themselves for the purpose of the study. The device provides a quantitative measurement, and thresholds were applied to this measurement to determine its correlation with disease progression. However, the intended use states it is an "aid" in monitoring and "should be used in conjunction with other clinical methods," implying human interpretation within a broader clinical context. The performance metrics (sensitivity, specificity etc.) presented in the tables are for the assay's output as a standalone indicator of change.

6. The type of ground truth used (expert consensus, pathology, outcomes data, etc.):

   • The ground truth for the clinical study was based on "changes in disease status." This likely encompasses a combination of clinical assessments, imaging results, and possibly pathology reports over time, interpreted by clinicians at the tertiary cancer center. The specific methods used to define "disease progression" and "no progression" within the study are not detailed beyond "changes in disease status."

7. The sample size for the training set:

   • The document does not provide information on a specific training set size. The clinical study described is the test set for assessing the final device's performance for monitoring. For an immunoassay, the "training" analogous to machine learning would be the development and optimization of the assay itself (antibody selection, reagent concentrations, calibration methods), which doesn't typically rely on a "training set" of patient data in the same way an AI algorithm does.

8. How the ground truth for the training set was established:

   • As there's no explicitly defined "training set" in the context of an AI/ML algorithm development, this question is not directly applicable. The assay's analytical performance (precision, linearity, LoD, LoQ, interference) and the clinical utility (monitoring study) are reported, which are evaluated against established analytical standards and clinical outcomes, respectively.

Ask a specific question about this device