Search Results

The Invader® Factor II test is an in vitro diagnostic test intended for the detection and genotyping of a single point mutation (G to A at position 20210) of the human Factor II gene in isolated genomic DNA obtained from whole blood potassium EDTA samples from patients with suspected thrombophilia.

The Invader Factor II test consists of the following components: Factor II Oligo Mix, Universal Buffer, Universal Enzyme Mix, No DNA Control, Factor II Wild Type Control, Factor II Heterozygous Control, Factor II Mutant Control, Invader Call Reporter™ Software, Invader® Factor II Software.

This document describes the Invader® Factor II test, an in vitro diagnostic test for detecting a specific mutation (G20210A) in the human Factor II gene, associated with thrombophilia.

Here's an analysis of the acceptance criteria and supporting studies:

1. Table of Acceptance Criteria and Reported Device Performance

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

• External Reproducibility (Study #1):
  • Sample Size: 9 unique leukocyte-depleted whole blood samples (3 WT, 3 HET, 3 MUT), tested in duplicate over 5 days by 2 operators at 3 sites. Total tests: 9 samples * 2 duplicates * 5 days * 2 operators * 3 sites = 540 initial tests.
  • Data Provenance: Not explicitly stated, but implied to be from internal and external validation sites. Likely retrospective as they were "spiked cell lines."
• Lot-to-Lot Reproducibility (Study #9):
  • Sample Size: 4 genomic DNA samples (3 WT, 1 HET), tested in quadruplicate using 3 different kit lots. Total tests: 4 samples * 4 replicates * 3 lots = 48 tests.
  • Data Provenance: Not explicitly stated. Likely retrospective as they are "genomic DNA samples."
• Real-Time Stability Study (Study #5):
  • Sample Size: 3 controls (WT, HET, MUT) and 4 gDNA samples (WT, WT, HET, MUT), tested in quadruplicate at each time point (T0, T4, T7) for 3 lots. Total tests are not explicitly summed, but 7 sample types * 4 replicates * 3 time points * 3 lots = 252 tests.
  • Data Provenance: Not explicitly stated. Likely retrospective (genomic DNA and controls).
• Reagent Freeze-Thaw Stability Study (Study #6):
  • Sample Size: 3 Controls (WT, HET, MUT) and 3 gDNA samples (WT, HET, MUT) with varying numbers of replicates tested at each of 15 freeze/thaw cycles. Total tests: 255.
  • Data Provenance: Not explicitly stated. Likely retrospective (genomic DNA and controls from cell lines).
• Analytical Sensitivity and Normal Range (Study #3):
  • Sample Size: 2 genomic DNA samples (1 HET, 1 WT), each diluted to 8 concentrations and tested in 40 replicates. Total tests: 2 samples * 8 concentrations * 40 replicates = 640 tests.
  • Data Provenance: Whole blood collected in potassium EDTA, then extracted and diluted. Retrospective for the samples, but the dilution and testing process is controlled.
• Analytical Specificity (Interfering Substances) (Study #4):
  • Sample Size: 4 whole blood samples differing genotype (3 WT, 1 HET), each with 9 different interfering substances/conditions. Total tests: 4 samples * 2 (with/without substance) * number of replicates not specified, but results given as "8 of 8" for percent agreement, suggesting at least 8 replicates per condition/sample. Total presented data points: 8 conditions * 8 replicates = 64 tests.
  • Data Provenance: Whole blood samples. Retrospective.
• Pre-Analytical Equivalency Study (Study #7):
  • Sample Size: 30 human whole blood samples and 10 leukocyte-depleted whole blood samples (total 40 samples), extracted using 4 different methods. Each resulting DNA analyzed in singlicate. Total tests: 40 samples * 4 extraction methods = 160 tests.
  • Data Provenance: Human whole blood samples. Implied retrospective for the samples themselves.
• Instrument Equivalency (Study #8):
  • Sample Size: 29 human whole blood samples and 10 leukocyte-depleted whole blood samples (total 39 samples), extracted (using 2 methods). These extracts tested with the device on 3 thermal cyclers and raw data acquired on 3 fluorometers. Total presented for agreement: 78 tests per thermal cycler/fluorometer combination (likely 39 samples tested in duplicate, or 39 samples * 2 extraction methods = 78 sample preparations). Total 78 across all combinations * 3 thermal cyclers * 3 fluorometers = 702 measurements.
  • Data Provenance: Human whole blood samples. Implied retrospective.
• Secondary Polymorphism Impact (Study #10):
  • Sample Size: 6 samples (1 homozygous normal, 1 heterozygous, 4 homozygous normal each with a known secondary polymorphism). 40 replicates for each sample. Total tests: 6 samples * 40 replicates = 240 tests.
  • Data Provenance: Samples with known genotypes and secondary polymorphisms. Likely retrospective.
• Method Comparison (Study #2):
  • Sample Size: 336 human whole blood samples.
  • Data Provenance: Human whole blood samples. Retrospective.

3. Number of Experts and Qualifications for Ground Truth

• General Ground Truth Method: The primary and most frequently cited ground truth method across all studies is bi-directional DNA sequencing.
• Number of Experts: The document does not specify the number of experts used for establishing ground truth via bi-directional DNA sequencing. Sequencing is typically a highly automated process with results interpreted by trained molecular biologists or laboratory staff, rather than a consensus of "experts" in the clinical sense (like radiologists).
• Qualifications of Experts: Not specified. It's assumed that standard molecular biology laboratory practices for sequencing and interpretation were followed.

4. Adjudication Method for the Test Set

• Not Applicable in the traditional sense. For diagnostic tests like this, adjudication by multiple human experts (e.g., radiologists) is not a typical part of ground truth establishment.
• The ground truth is established by a definitive molecular method (bi-directional DNA sequencing).
• In the external reproducibility study (Study #1), two initial "No Call" results were resolved by retesting, which then agreed with the sequencing results. This implies a retesting/re-evaluation process for indeterminate results, rather than expert adjudication of the initial "No Call."

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

• No, an MRMC comparative effectiveness study was not done.
• This device is an in vitro diagnostic test for genotyping, where the output is a molecular result (genotype). It does not involve human readers interpreting images or data where AI assistance would directly improve human performance in the same way as, for example, a radiology AI.
• The "Invader Call Reporter™ Software" mentioned is for data analysis and conversion of raw fluorescence data into genotype calls, not for assisting human interpretation that would typically be evaluated in an MRMC study.

6. Standalone Performance Study

• Yes, a standalone performance study was done.
• The entire set of analytical performance studies (Reproducibility, Stability, Analytical Sensitivity, Analytical Specificity, Extraction Equivalency, Instrument Equivalency, Secondary Polymorphism Impact) and the method comparison study against bi-directional sequencing demonstrate the standalone performance of the algorithm/device system.
• The device takes raw fluorescence data and, through its software ("Invader Call Reporter™ Software" and "Invader® Factor II Software"), outputs a genotype call. The studies rigorously validate that these automated calls are accurate and reproducible compared to the gold standard of sequencing. The system is designed to provide automated calls, not to serve as an aid to a human reader's interpretation in a separate step.

7. Type of Ground Truth Used

• The primary ground truth used across all key performance studies is bi-directional DNA sequencing. This is considered a gold standard molecular method for determining specific genetic mutations.
• For controls, "expected genotype" based on the composition of the control material (e.g., cell lines) is used.

8. Sample Size for the Training Set

• The document does not explicitly describe a separate "training set" for the device's algorithm.
• For molecular diagnostic assays like this, the algorithm (e.g., for converting fluorescence data to genotype calls) is typically developed based on known statistical models and pre-defined thresholds related to the chemistry, rather than being "trained" on a large dataset of classified samples in the machine learning sense.
• The various controls (No DNA, WT, HET, MUT) are used within each run to validate performance and inform signal-to-noise calculations, serving as internal calibration rather than an external training set for an AI model.

9. How Ground Truth for the Training Set Was Established

• As a formal "training set" is not explicitly mentioned or implied in the context of supervised machine learning, the process of establishing ground truth for such a set is not described.
• However, the underlying principles for the chemistry and software's interpretive rules would have been developed using well-characterized samples with ground truth established by methods like DNA sequencing during the research and development phase before formal validation studies.

Ask a specific question about this device