(603 days)
ADVIA Centaur® Enhanced Liver Fibrosis (ELF™) is for in vitro diagnostic use in the determination of an ELF score based on the combined quantitative measurements of hyaluronic acid, amino-terminal propeptide of type III procollagen, and tissue inhibitor of matrix metalloproteinase 1 in human serum using the ADVIA Centaur XP system.
ADVIA Centaur ELF is indicated as a prognostic marker in conjunction with other laboratory findings and clinical assessments in patients with advanced fibrosis (F3 or F4) due to non-alcoholic steatohepatitis (NASH), to assess the likelihood of progression to cirrhosis and liver-related clinical events.
The ADVIA Centaur ELF test contains the following:
Reagents: Lite Reagent and Solid Phase reagents for HA, PIIINP and TIMP-1, and Ancillary Well reagents for HA and PIIINP are contained within ReadyPack® reagent packs. Each ReadyPack reagent cartridge has a barcode label used to automatically transfer information to the instrument when loaded into the instrument.
Calibrators: ADVIA Centaur ELF Calibrator kit contains 2 levels of calibrators (Low and High). Calibration is performed using Low and High Calibrators of known value (as per the Calibrator Assigned Value card provided in the ELF Calibrator kit).
Controls: The ADVIA Centaur ELF QC kit contains 3 controls (Level 1, Level 2, Level 3).
The provided text describes the acceptance criteria and a study proving the device's performance, but it does not pertain to a device utilizing Artificial Intelligence (AI). Instead, it describes an in vitro diagnostic immunoassay called ADVIA Centaur Enhanced Liver Fibrosis (ELFTM) which measures specific biomarkers (hyaluronic acid, amino-terminal propeptide of type III procollagen, and tissue inhibitor of metalloproteinase 1) to calculate an ELF score.
Therefore, many of the requested elements related to AI (e.g., sample size for test set and training set specifically for AI model, number of experts for ground truth for AI, MRMC study for AI assistance, standalone AI performance) are not applicable to this document. The document focuses on the analytical and clinical performance of a laboratory test.
However, I can provide the information that is applicable based on the provided text, reformulating some points to fit the nature of this diagnostic device:
1. Table of Acceptance Criteria and Reported Device Performance (Focus on Clinical Prognosis):
The document does not explicitly state "acceptance criteria" in a numerical, pre-defined manner for the clinical performance. Instead, it presents the observed clinical performance in terms of risk stratification based on ELF scores in different patient populations. The risk thresholds (cut-offs) were determined in a separate clinical study not fully detailed here.
| Acceptance Criteria (Implied Clinical Performance) | Reported Device Performance (Pooled Data) |
|---|---|
| Cirrhotic Population (NASH F4): | |
| Ability to stratify risk of Liver Related Event (LRE) progression | ELF Score 11.3: 21.0% Risk of LRE (95% CI: 12.1-29.9%) (n=81) |
| Bridging Fibrosis Population (NASH F3): | |
| Ability to stratify risk of cirrhosis progression | ELF Score 11.3: 44.4% Risk of Cirrhosis (95% CI: 21.5-67.4%) (n=18) |
Note on Analytical Performance Acceptance Criteria: The document references CLSI guidelines (e.g., EP05-A3 for precision, EP06-A for linearity, EP17-A2 for detection capability) which inherently define acceptance criteria for analytical performance studies. The reported values (e.g., SD and %CV for repeatability and reproducibility, LoQ/LoD values) indicate that these criteria were met.
2. Sample Size and Data Provenance for Clinical Studies (Analogous to Test Set):
- Sample Size for Clinical Studies:
- Cirrhotic population (NASH F4): Total 305 subjects (47 in 11.3 group). Data from 3 trials (Trial 1, 2, 3), specifically placebo arms.
- Bridging fibrosis population (NASH F3): Total 212 subjects (105 in 11.3 group). Data from 2 trials (Trial 4, 5), specifically placebo arms.
- Data Provenance: The data was gathered from (implied multi-center/international) clinical trials intended to assess investigational therapeutic substances. The specific country of origin is not explicitly stated, but the context of "US blood donors" for expected values suggests a US presence. The data is retrospective in the sense that it was re-analyzed from existing placebo arms of therapeutic trials.
3. Number of Experts and Qualifications for Ground Truth:
- Not Applicable: This is an in-vitro diagnostic test measuring biomarkers. The "ground truth" for clinical performance is defined by clinically observed outcomes (e.g., "Liver Related Event" progression to cirrhosis, all-cause mortality, liver transplantation, esophageal variceal bleeding). This does not involve expert readers establishing ground truth for images or similar data.
4. Adjudication Method for Clinical Outcomes (Analogous to Test Set):
- Not Explained: The document does not specify an adjudication method for the clinical outcomes. It states that outcomes were "evaluated" but doesn't detail how consistency or accuracy of these outcome assessments were ensured across different trial sites or by independent reviewers.
5. MRMC Comparative Effectiveness Study:
- No: This is an in-vitro diagnostic test. MRMC studies are typically performed for imaging or qualitative diagnostic tools where human interpretation is involved, often comparing human performance with and without AI assistance. This device provides a quantitative score.
6. Standalone Performance (Algorithm Only):
- Applicable, and the study details this: The device's performance is its "standalone" performance. The ELF score is calculated automatically by the instrument based on the measured concentrations of HA, PIIINP, and TIMP-1:
ELF score = 2.278 + 0.851*In(CHA) + 0.751*In(CpmNp) + 0.394*In(CTMP-1). The entire "Performance Characteristics" section (L) and "Clinical Studies" section (3) describe this standalone performance.
7. Type of Ground Truth Used:
- Outcomes Data/Clinical Events: For the clinical performance, the ground truth was observed clinical outcomes and events experienced by patients in the placebo arms of various clinical trials. These included:
- For cirrhotic population: all-cause mortality, liver transplantation, qualification for liver transplantation (MELD ≥ 15), esophageal variceal bleeding, clinically apparent ascites, hepatic encephalopathy, newly diagnosed varices, progression of varices.
- For bridging fibrosis population: development of cirrhosis.
8. Sample Size for Training Set:
- Not Directly Applicable in the AI Sense: This device is not an AI model that undergoes a "training" phase. The formula for the ELF score (
2.278 + 0.851*In(CHA) + 0.751*In(CpmNp) + 0.394*In(CTMP-1)) is a fixed algorithm. - However, "training" in a broad sense implies model development data. The document states: "The sponsor previously conducted a separate clinical study to determine the clinically relevant cut off ELF scores." This "separate clinical study" would be the closest analog to a "training set" or "development set" from which the algorithm's parameters (the coefficients in the ELF score equation) and the clinical cut-off values (9.8 and 11.3) were derived. The size of this previous study is not provided in this document.
9. How Ground Truth for "Training Set" Was Established:
- As above, for the determination of the ELF score formula and clinical cut-offs, the ground truth would have been established through clinical correlation studies, likely involving comparison of ELF scores to biopsy-proven fibrosis stages, or to long-term clinical outcomes/endpoints, similar to what was done for the "test set" (clinical studies described in section L.3.c). The exact methodology for this earlier study is not detailed in this document.
§ 862.1622 Prognostic test for assessment of liver related disease progression.
(a)
Identification. A prognostic test for assessment of liver related disease progression is intended to measure one or more analytes obtained from human samples as an aid in assessing progression of liver related disease. This device is not intended for diagnosis of any disease, for monitoring the effect of any therapeutic product, for assessing progression to hepatocellular carcinoma, or for assessing disease progression in individuals with viral hepatitis. It is also not intended for the detection of viruses, viral antigens, or antibodies to viruses.(b)
Classification. Class II (special controls). The special controls for this device are:(1) Design verification and validation must include clinical validation data providing:
(i) Information demonstrating clinical performance in a population of patients with liver disease for the different risk categories (
e.g., at lower risk, at higher risk) for progression of their disease using well characterized clinical specimens representing the intended use population collected from multiple intended clinical sites, or an alternative study design determined to be appropriate by FDA.(ii) Information demonstrating that the outcomes measured and the length of followup are clinically relevant for the progression of the specified liver disease.
(iii) Information demonstrating that the clinical criteria for determining whether the target disease is present and that the exclusion and inclusion criteria for subjects who have the target disease are appropriate.
(iv) Information demonstrating test performance of the complete test system, including any sample collection and processing steps.
(v) Information, provided or referenced, generated in samples from non-diseased individuals, that demonstrate the upper and lower reference intervals for the output provided by the device.
(2) The labeling required under 21 CFR 809.10(b) must include:
(i) A warning statement that test results are not intended to diagnose disease or for monitoring the effect of any therapeutic product.
(ii) A warning statement that test results are intended to be used in conjunction with other clinical and diagnostic findings, consistent with professional standards of practice, including information obtained by alternative methods, and clinical evaluation, as appropriate.
(iii) A warning statement that describes any limitations on the clinical interpretation(s) of the test results.
(iv) Detailed information on device performance, including any limitations to the data generated in the clinical study(ies) and information on device performance in relevant subgroups (
e.g., severity of liver disease at the beginning of the observation period) observed in the clinical study(ies).(v) Information on the analytical performance of the device, including demonstration of reproducibility across multiple sites and multiple reagent lots, or an alternative reproducibility study design determined to be appropriate by FDA.