(90 days)
The cobas® Cdiff Nucleic acid test for use on the cobas® Liat® System is an automated, qualitative in vitro diagnostic test, that utilizes real-time polymerase chain reaction (PCR), for the detection of the toxin B (tcdB) gene of toxigenic Clostridium difficile in unformed (liquid or soft) stool specimens suspected of having C. difficile infection (CDI). The cobas® Cdiff Nucleic acid test for use on the cobas® Liat® System is intended for use as an aid in the diagnosis of CDI in humans in conjunction with clinical and epidemiological risk factors.
The cobas® Cdiff Nucleic acid test for use on the cobas® Liat® System (cobas® Cdiff ) is a rapid, automated in vitro diagnostic test for the qualitative detection of C. difficile DNA in human stool specimens. The cobas® Liat® System is for in vitro diagnostic use. The system is designed to identify and/or measure the presence of genetic material in a biological sample. The system automates all nucleic acid amplification test (NAAT) processes, including reagent preparation, target enrichment, inhibitor removal, nucleic acid extraction, amplification, real-time detection, and result interpretation in a rapid manner. The cobas® Cdiff test uses silica magnetic particle-based nucleic acid extraction and TaqMan probe-based real-time PCR amplification and detection. The cobas® Liat® Analyzer automates and integrates sample purification, nucleic acid amplification and detection of the target sequence in biological samples. Other than adding the sample to the cobas® Cdiff assay tube, no reagent preparation or additional steps are required. The cobas® Liat® System consists of an cobas® Liat® Analyzer with integrated software for running tests and analyzing the results, and a single-use disposable cobas® Cdiff assay tube that holds all of the sample purification and PCR reagents and hosts the sample preparation and PCR processes specific for the Cdiff analyte. The test uses the assay tube as both the sample and reaction vessel. The assay tube comprises flexible tubing containing all required unit dose reagents pre-packed in tube segments, separated by pressure-sensitive seals, in the order of reagent use. During the testing process, multiple sample processing actuators of the analyzer compress the cobas® Cdiff assay tube to selectively release reagents from tube segments, move the sample from one segment to another, and control reaction conditions such as reaction volume, temperature, pressure, and incubation time. Precise control of all these parameters provides optimal conditions for assay reactions, allowing the test to achieve high performance similar to or better than that of currently available molecular assays. The cobas® Liat® Analyzer software controls and coordinated these actions to perform all required assay processes, including sample preparation, nucleic acid extraction, target enrichment, inhibitor removal, nucleic acid elution, and real-time PCR. All assay steps are performed within the closed and self-contained cobas® Cdiff assay tube, thereby eliminating the potential for cross-contamination between samples. The collected data are automatically analyzed and the result is displayed in the assay report on the integrated LCD touch screen of the cobas® Liat® Analyzer.
Here's an analysis of the acceptance criteria and study details based on the provided text:
Acceptance Criteria and Device Performance for cobas® Cdiff Nucleic acid test for use on the cobas® Liat® System
1. Table of Acceptance Criteria and Reported Device Performance
The document does not explicitly state formal "acceptance criteria" in a separate section with numerical targets like a typical clinical trial report for an AI device. Instead, the performance metrics are presented as the results of the studies conducted. I will infer the acceptance criteria from the reported performance, implying that the reported values met the internal standards set by the manufacturer for substantial equivalence.
| Performance Metric | Acceptance Criteria (Inferred from Predicate/Study Goals) | Reported Device Performance (cobas® Cdiff Test) |
|---|---|---|
| Analytical Sensitivity (LOD) | To detect C. difficile at a concentration to yield a Hit Rate of at least 95%, where all higher concentrations also have a Hit Rate of ≥ 95%. | ATCC 43255: 90 CFU/swab |
| R12087 (CD196): 45 CFU/swab | ||
| Inclusivity | To detect 100% of tested toxigenic strains at 3xLOD. | All 37 toxigenic strains detected with 100% hit rate. |
| Analytical Specificity (Cross-reactivity) | No cross-reactivity with non-toxigenic C. difficile strains, human genomic DNA, or other common stool microorganisms, and no interference with toxigenic C. difficile detection. | No cross-reactivity with 149 tested non-toxigenic Cdiff strains, human epithelial cells, or other microorganisms. No interference with detection of toxigenic Cdiff at 3x LOD. |
| Interference | No interference from common medications and fecal substances at specified concentrations. | Fecal Fat: No interference up to 39% (w/v) |
| Whole blood: No interference up to 100% (v/v) | ||
| Mucin: No interference up to 50% (w/v) | ||
| Various other medications and substances tested and confirmed for no interference at specified concentrations (see Table 5). | ||
| Reproducibility (~1xLOD) | Overall percent agreement for low positive samples (~1xLOD) to demonstrate consistency across sites, operators, and lots. | 98.5% (262/266) with a 95% CI of (96.2%, 99.4%) |
| Reproducibility (~3xLOD) | Overall percent agreement for moderate positive samples (~3xLOD) to demonstrate consistency across sites, operators, and lots. | 99.3% (268/270) with a 95% CI of (97.3%, 99.8%) |
| Reproducibility (Negative) | Overall percent agreement for negative samples to demonstrate consistency across sites, operators, and lots. | 100.0% (262/262) with a 95% CI of (98.6%, 100.0%) |
| Clinical Sensitivity (vs. Combined Culture) | High sensitivity for detecting toxigenic C. difficile in clinical samples. | 87.2% (156/179; 95% CI: 81.5% to 91.3%) |
| Clinical Specificity (vs. Combined Culture) | High specificity for detecting toxigenic C. difficile in clinical samples. | 98.1% (818/834; 95% CI: 96.9% to 98.8%) |
| Clinical PPV (vs. Combined Culture) | High Positive Predictive Value. | 90.7% (156/172; 95% CI: 85.4% to 94.2%) |
| Clinical NPV (vs. Combined Culture) | High Negative Predictive Value. | 97.3% (818/841; 95% CI: 95.9% to 98.2%) |
| Initial Invalid Rate | Low initial invalid rate. | 1.4% |
| Initial Failed Rate | Low initial failed rate. | 0.2% |
| Final Invalid Rate (after retest) | Very low final invalid rate. | 0.1% |
| Final Failed Rate (after retest) | Zero final failed rate. | 0% |
2. Sample size used for the test set and the data provenance
- Analytical Sensitivity (LOD): Not specified as a per-sample-size, but replicates of each 5-member test panel were tested with 2 different lots of assay tubes. Additional replicates were tested for confirmation.
- Inclusivity: 3 replicates per strain (37 toxigenic strains) were tested.
- Analytical Specificity (Cross-reactivity): Not specified as a per-sample-size, but 149 non-toxigenic organisms/cells were tested.
- Interference: Not specified as a per-sample-size, but 38 commonly used medications, fecal fat, whole blood, and mucin were tested.
- Reproducibility: 818 total tests were performed, with 798 valid for analysis across 3 panel members (Negative, ~1xLOD, ~3xLOD). 3 replicates per panel member, run on 5 nonconsecutive days by 2 operators at 3 sites for 3 reagent lots.
- Clinical Performance: 1,013 evaluable fresh remnant stool specimens were prospectively collected.
- Data Provenance:
- Country of Origin: Not explicitly stated, but "Nine (9) study sites from geographically diverse locations participated in this study." This suggests a multi-center study, likely within a single country (such as the US, given the FDA submission) or potentially across a few countries, but a specific list is not provided.
- Retrospective/Prospective: The clinical performance study was prospective. "Evaluable fresh remnant specimens were prospectively collected from 1,013 patients."
- Data Provenance:
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
- Analytical (LOD, Inclusivity, Specificity, Interference): Not applicable for these studies as the "ground truth" is based on engineered samples (known concentrations of specific strains, or presence/absence of interfering substances).
- Reproducibility: Not applicable, as this assessed the consistency of the device's own results, not against an external ground truth.
- Clinical Performance:
- Number of experts: Not explicitly stated.
- Qualifications of experts: Not explicitly stated. The "ground truth" was established by "combined results of direct and enriched toxigenic culture using leftover, deidentified, unformed stool samples" performed at a "reference laboratory." The document mentions "Nineteen (19) operators distributed across these study sites performed testing with the cobas® Cdiff test," but these are operators, not necessarily the adjudicating experts for the ground truth. For microbiological culture, laboratory personnel with specific training in microbiology would establish the ground truth.
4. Adjudication method for the test set
- Analytical, Reproducibility: Not applicable.
- Clinical Performance:
- Initial Ground Truth: "Combined results of direct and enriched toxigenic culture" served as the primary comparative method.
- Discrepant Analysis: "All evaluable discordant specimens and an equal number of randomly selected and representative concordant samples were used for discrepant testing by an FDA-cleared comparator NAAT. Testing was performed at an external lab that was preselected and qualified by RMS." This implies a form of 2-method consensus with a third method for discrepancies. If direct and enrichment culture disagreed, or if the cobas test disagreed with the combined culture, a third (comparator NAAT) test was used for "discrepant analysis." However, the final clinical performance tables still directly compare the cobas test to the combined culture results, with the discrepant analysis results discussed qualitatively ("Of the 23 specimens with false-negative cobas® Cdiff test results relative to combined direct culture and enrichment culture, 19 were negative. 3 were positive and 1 was invalid by the second NAAT method.").
5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done
- No, an MRMC comparative effectiveness study was not done. This device is an in vitro diagnostic (IVD) test, not an imaging AI device that relies on human-in-the-loop interpretation. The clinical study compares the device's output to laboratory culture results, not to human readers' interpretations with or without AI assistance.
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done
- Yes, the performance presented is purely standalone (algorithm/device only). The cobas® Cdiff Nucleic acid test is an automated, qualitative in vitro diagnostic test. The device provides "result interpretation" directly on its integrated LCD touch screen (page 5). The clinical performance data reflects the device's ability to detect the target gene independently, compared against culture methods.
7. The type of ground truth used
- Analytical Studies (LOD, Inclusivity, Specificity, Interference): Lab-prepared spiked samples with known concentrations/compositions.
- Clinical Performance: Microbiological culture results. Specifically, "combined results of direct and enriched toxigenic culture." A "second NAAT method" was used for discrepant analysis.
8. The sample size for the training set
- Not applicable / Not explicitly stated. This is a molecular diagnostic test for Clostridium difficile, not a machine learning or AI algorithm in the typical sense that requires explicit "training data" for a model to learn from. The assay uses predefined primers and probes for real-time PCR detection (page 4). While the assay design and optimization would involve extensive internal development and testing, this is not characterized as a "training set" in the context of typical AI algorithm development.
9. How the ground truth for the training set was established
- Not applicable / Not explicitly stated. As mentioned above, this is not an AI/ML device that uses a "training set" in the conventional sense. The "ground truth" for developing the test was based on well-characterized isolates of C. difficile and other microorganisms to design and validate the specificity and sensitivity of the primers and probes.
§ 866.3130 Clostridium difficile toxin gene amplification assay.
(a)
Identification. AClostridium difficile toxin gene amplification assay is a device that consists of reagents for the amplification and detection of target sequences inClostridium difficile toxin genes in fecal specimens from patients suspected of havingClostridium difficile infection (CDI). The detection of clostridial toxin genes, in conjunction with other laboratory tests, aids in the clinical laboratory diagnosis of CDI caused byClostridium difficile. (b)
Classification. Class II (special controls). The special controls are set forth in FDA's guideline document entitled: “Class II Special Controls Guideline: Toxin Gene Amplification Assays for the Detection ofClostridium difficile; Guideline for Industry and Food and Drug Administration Staff.” See § 866.1(e) for information on obtaining this document.