Search Results

This in vitro diagnostic assay is intended to quantitatively measure a-fetoprotein (AFP) in human serum on the Bayer Immuno 1TM system as an aid in the management of nonseminomatous testicular cancer. AFP values obtained using the Bayer Immuno 1™ AFP assay method must be interpreted in conjunction with all other available clinical and laboratory data before a medical decision is determined.

The Bayer Immuno ITM AFP Assay utilizes a well-established immunoassay technology in which one monoclonal AFP antibody is conjugated to fluorescein (designated Reagent 1, or R1) and a second monoclonal AFP antibody is conjugated to alkaline phosphatase (Reagent 2, or R2). The R1 and R2 conjugates are reacted with patient sample, calibrator, or control and are incubated at 37°C on the system. An Immuno 1 Magnetic Particle coated with an anti-fluorescein antibody is then added and a second incubation occurs during which the antibody complex is bound. The magnetic particles complexed with the immunological sandwich are then washed to separate unbound molecules, and a colorimetric substrate is added. The rate of conversion of substrate to a compound with absorbance at 405 and 450 nm is measured and the measured rate is proportional to the concentration of AFP antigen in the sample. A cubic-through-zero curve fitting algorithm is used to generate standard curves.

Here's an analysis of the provided text regarding the Bayer Immuno 1™ AFP Assay, structured according to your requirements:

Bayer Immuno 1™ AFP Assay Acceptance Criteria and Study Details

1. Table of Acceptance Criteria and Reported Device Performance

The document primarily focuses on demonstrating substantial equivalence to a predicate device (Abbott IMx AFP Assay) rather than pre-defined acceptance criteria with specific thresholds for all metrics. However, performance metrics are presented in comparison to the predicate and generally found to be "excellent" or "comparable."

• Triglycerides (900 mg/dL): 99.9% Recovery
• IgG (0.053 g/mL): 98.2% Recovery
• Hemoglobin (10.0 mg/mL): 102.1% Recovery
• Heparin (0.15 mg/mL): 101.5% Recovery
• Bilirubin (0.25 mg/mL): 99.7% Recovery
• Albumin (0.065 g/mL): 104.0% Recovery
• CEA (10,000 ng/mL): 99.9% Recovery
  Conclusion: None demonstrated significant interfering effects. |
  | | Exogenous Interferents (Chemotherapeutic drugs, OTC drugs, Vitamins):
• All tested showed % Recovery ranging from 95.2% to 104.8%.
  Conclusion: None demonstrated significant interfering effects. |
  | | Heterophilic Antibodies (Rheumatoid Factor, HAMA):
• RF (high titers): 96.4-103.2% Recovery
• HAMA (high titers): 93.5-102.4% Recovery
  Conclusion: Lack of significant heterophilic interference. |
  | Minimum Detectable Concentration (MDC) | Range: 0.04 ng/mL to 0.09 ng/mL across three sites.
  Claim: 0.1 ng/mL. (Deemed acceptable for an assay of this type). |
  | Imprecision (Total CV%) | Not greater than 3.6% for all tested concentrations (e.g., Calibrator 2 mean 5.0 ng/mL: Total CV 2.5%; Calibrator 6 mean 398.2 ng/mL: Total CV 2.6%).
  Conclusion: Highly reproducible. |
  | Linearity (Recovery%) | 95.7% to 102.8% of expected value over the entire calibration range (0-400 ng/mL).
  Conclusion: Clearly demonstrates linearity. |
  | Hook Effect | No "hook effect" observed until a concentration of 204,268 ng/mL AFP was reached.
  Conclusion: Lack of hook effect at AFP assay values ≤ 160,000 ng/mL. |
  | Spiked Recovery (%) | 96.7% to 99.3% in five patient samples spiked with ~300 ng/mL AFP.
  Conclusion: Accurate quantitation of spiked and recovered AFP values. |
  | Parallelism (Recovery%) | 92.7% to 111.0% in diluted patient samples. Linear regression showed no deviation from linearity.
  Conclusion: Further illustrates linearity and acceptability of 0 ng/mL Calibrator as diluent. |
  | Clinical Performance (Method Comparison with Abbott IMx) | |
  | Normal Range Cut-off (97.5th percentile) | Immuno 1: 8.9 ng/mL (based on male subjects from DOCRO).
  Abbott IMx: 8.1 ng/mL (based on male subjects from DOCRO).
  Conclusion: Similar to predicate device and manufacturer's suggested range (400 ng/mL: r=0.992, Slope=1.046 (OLS); r=0.992, Slope=0.985 (Passing-Bablok)
  MSKCC (N=318):
• 0-400 ng/mL: r=0.997, Slope=0.931 (OLS); r=0.997, Slope=1.024 (Passing-Bablok)

• 400 ng/mL: r=0.999, Slope=0.939 (OLS); r=0.999, Slope=0.931 (Passing-Bablok)
  Conclusion: Excellent comparison with slopes near 1.0 and r > 0.99. |
  | Clinical Sensitivity (Nonseminomatous Testicular Ca.) | Immuno 1: 64.0%
  Abbott IMx: 64.0%
  Conclusion: Comparable to predicate and previous studies (50-75% for nonseminomatous and mixed GCT). |
  | Clinical Sensitivity (Mixed Germ Cell Tumor) | Immuno 1: 56.5%
  Abbott IMx: 54.3%
  Conclusion: Comparable. |
  | Clinical Specificity (Healthy Individuals) | Immuno 1: 98.3%
  Abbott IMx: 98.6%
  Conclusion: Consistent with predicate and normal population. |
  | Clinical Specificity (Other Benign Diseases) | Immuno 1 & Abbott IMx: 88%-100% (for cirrhosis, hepatitis, benign GU, other non-malignant diseases).
  Conclusion: Consistent with predicate and literature. |
  | Longitudinal Monitoring (Patient Course Tracking) | Immuno 1 AFP Assay demonstrated clinical utility for 78.8% (41/52) of all longitudinal patients (testicular and hepatocellular cancer) and 76.6% (36/47) of testicular cancer patients.
  Conclusion: Identical trends observed with both Immuno 1 and IMx, demonstrating equivalent clinical utility. |

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

• Total Samples for Method Comparison (Cross-sectional): 863 serum samples (545 from DOCRO + 318 from MSKCC).
  • DOCRO (Diagnostic Oncology CRO, Stratford, CT): 545 samples from Western States Plasma Company (Fallbrook, CA).
    • Clinical Classifications: Normal (male/female), various testicular cancers, hepatocellular carcinoma, genito-urinary cancer, gastrointestinal/lung/pancreatic cancer, hepatitis, cirrhosis, benign genito-urinary diseases, other non-malignant diseases.
  • MSKCC (Memorial Sloan-Kettering Cancer Center, New York, NY): 318 samples from MSKCC's own in-house specimen collection.
    • Clinical Classifications: Normal, various testicular cancers, prostate cancer, liver cancer, colon cancer, breast cancer, lung cancer, benign GU disease, and samples from longitudinal patients (counted in the 318 for cross-sectional and separately for longitudinal).
• Longitudinal Monitoring Test Set:
  • 314 longitudinal samples from 56 patients (51 testicular cancer, 5 hepatocellular cancer).
  • All these samples were obtained from MSKCC's in-house specimen collection.
• Normal Range Determination: 350 healthy subjects (100 males from MSKCC, 150 males from DOCRO, 100 females from DOCRO).
• Provenance: Data was retrospective, collected from two U.S. clinical trial sites (Diagnostic Oncology CRO and Memorial Sloan-Kettering Cancer Center) and a plasma company (Western States Plasma Company, CA).

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

The document does not explicitly state the number or specific qualifications of experts used to establish the ground truth for the test set beyond mentioning that "physician evaluations" were guides. The ground truth for longitudinal studies relied on "medical, diagnostic, and course of therapy information" and "diagnostic testing, therapeutic interventions, and physician evaluations." The evaluation of the Immuno 1 AFP Assay's correspondence with the clinical course was performed by "Bayer scientists." No specific qualifications (e.g., years of experience as a radiologist) are provided for these "Bayer scientists" or the "physicians" whose evaluations were used.

4. Adjudication Method for the Test Set

The document does not describe a formal adjudication method for discrepancies in the ground truth for the clinical test set. For the longitudinal studies, the Immuno 1 AFP results were "graphically compared to the patient's course of disease using diagnostic testing, therapeutic interventions, and physician evaluations as guides to describe the clinical picture." The judgment of correspondence was made by "Bayer scientists." This implies a single-party evaluation of the clinical course against the assay results, rather than a multi-expert adjudication method like 2+1 or 3+1.

5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was Done

No, an MRMC comparative effectiveness study was not done. This study is an in vitro diagnostic assay comparing a new device (Bayer Immuno 1™ AFP Assay) to a predicate device (Abbott IMx AFP Assay) directly, not assessing how human readers (e.g., radiologists, pathologists) improve with or without AI assistance. The "AI" in this context refers to the automated immunoassay system, not an independent AI algorithm assisting human interpretation.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done

Yes, the study presents the standalone performance of the Bayer Immuno 1™ AFP Assay. The results (e.g., AFP values, sensitivity, specificity, correlation, longitudinal trends) are generated directly by the Immuno 1 system and evaluated against clinical classifications or compared to the predicate device's standalone results. There is no human-in-the-loop component being evaluated or improved upon.

7. The Type of Ground Truth Used

The ground truth for the clinical studies relied on a combination of:

• Clinical Classifications: Based on patient diagnoses (e.g., "Normal," "Testicular Cancer," "Hepatocellular Carcinoma," "Hepatitis," "Cirrhosis").
• Absence/Presence of Disease: For sensitivity/specificity calculations.
• Clinical Course Information: For longitudinal studies, this included "documented medical, diagnostic, and course of therapy information," as well as "diagnostic testing, therapeutic interventions, and physician evaluations." Essentially, expert clinical judgment and patient outcomes data.

The ground truth for the non-clinical studies (e.g., specificity, imprecision, linearity) was based on known spiked concentrations or established reference materials/methods.

8. The Sample Size for the Training Set

The document does not explicitly state a separate training set size beyond implying that the Immuno 1 system and its assay were developed and optimized internally by Bayer. It describes the characterization of antigens and antibodies and the development of the assay method (e.g., cubic-through-zero curve fitting algorithm). However, there is no mention of a distinct "training set" of patient data used for machine learning or algorithm development in the way one might find for a modern AI device. The study primarily focuses on validation against the predicate and clinical performance.

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

Since a distinct "training set" for an AI algorithm is not described, the concept of ground truth establishment for such a set is not applicable here. The ground truth for assay development and calibration (implicitly the "training" phase for the assay itself) would have been established using known concentrations of AFP calibrators, controls, and characterized biological samples to define the standard curve and ensure assay performance characteristics like linearity and imprecision.

Ask a specific question about this device