(622 days)
The Lp(a) Ultra assay is intended for in vitro diagnostic use in the immunoturbidimetric quantitation of lipoprotein (a) [Lp(a)] in human serum and plasma using an automated analyzer. The measurement of Lp(a) is useful in evaluating lipid metabolism disorders and assessing atherosclerotic cardiovascular disease in specific populations, when used in conjunction with clinical evaluation.
For In Vitro Diagnostic use.
Lp(a) Ultra assay is composed by 2 ready to use liquid reagents (Reagent 1and Reagent 2) that are supplied in the following configuration: Reagent 1 fill volume 18 mL in a 20 mL wedge and Reagent 2 fill volume 9 mL in a 20 mL wedge, 1 wedge of each/kit.
The kit contains one plastic (HDPE) vial of Reagent 1 and one plastic (HDPE) vial of Reagent 2, which allows the customer to perform 86 tests (on AU680 automatic analyzer).
The provided text describes the performance testing of the Lp(a) Ultra assay, an in vitro diagnostic device, and its acceptance criteria. This is a lab-based assay, not an AI/ML medical device, and therefore the concepts of human readers, AI assistance, ground truth experts, and training/test sets as understood in AI/ML are not directly applicable in the same way. However, I can extract and present the information in a table format that parallels the requested structure for acceptance criteria and performance against those criteria.
Key takeaway: This document describes a traditional in-vitro diagnostic device, not an AI/ML based device. Therefore, many of the requested fields (e.g., number of experts, adjudication methods, MRMC studies, AI assistance) are not relevant.
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criterion (Test Performed) | Target/Requirement (Implicit) | Reported Device Performance |
|---|---|---|
| Precision (Repeatability/Reproducibility) | Inter-Assay (Total Imprecision): Good precision across the concentration range. | |
| Intra-Assay (Within Run): Calculated CV% lower than 5%. | Inter-Assay: Total %CV ranged from 2.3% to 5.3% across 5 concentration levels (20 mg/dL to 100 mg/dL). | |
| Intra-Assay: Total %CV ranged from 0.5% to 4.8% across 5 concentration levels and 3 runs. All calculated CV% were lower than 5%. | ||
| Linearity (Analytical Measuring Range - AMR) | Demonstrate linearity up to 100 mg/dL and establish AMR. | Linear range found as 8.4 to 105.5 mg/dL (Lot 00507) and 8.4 to 103.8 mg/dL (Lot 10208). The assay is linear up to 100 mg/dL. Claimed AMR: 10 mg/dL to 100 mg/dL (based on LoQ and linearity). |
| Analytical Sensitivity/Detection Limit | Limit of Blank (LoB): Highest measurement for a blank sample within reasonable limits. | |
| Limit of Detection (LoD): Lowest analyte concentration reliably distinguished from LoB. | ||
| Limit of Quantitation (LoQ): Lowest amount quantifiable with stated accuracy. | LoB: 0.7 mg/dL (for both reagent lots). Supports LoB claim of 0.7 mg/dL. | |
| LoD: 1.6 mg/dL (Lot 90266), 1.9 mg/dL (Lot 90530). Supports LoD claim of 1.9 mg/dL. | ||
| LoQ: 2.5 mg/dL (Lot 90266), 3.0 mg/dL (Lot 90530). Supports LoQ claim of 3.0 mg/dL. | ||
| Interference (Endogenous Substances) | No significant interference from common endogenous substances (e.g., Intralipid, Triglycerides, Bilirubin, Rheumatoid Factor, Hemoglobin, Ascorbic Acid) at specified concentrations. | No interference observed for: |
- Intralipid® Sterile Fat Emulsion: up to 1000 mg/dL
- Conjugated bilirubin: up to 60 mg/dL
- Unconjugated bilirubin: up to 60 mg/dL
- Rheumatoid Factor: up to 500 UI/mL
- Hemoglobin: up to 1000 mg/dL
- Ascorbic Acid: up to 180 mg/dL
- Triglycerides: Tested up to 1000 mg/dL; no significant interference from 461 to 715 mg/dL. |
| Stability (On Board/Calibration) | Demonstrate stated on-board and calibration stability claims. | Calibration Stability: 15 days on AU 680 Analyzer.
On-board stability: 30 days on AU680 analyzer.
%bias for levels 1-5 within acceptable ranges (-4.8% to 5.1%). |
| Prozone Effect | No high-dose hook effect (prozone) observed within the claimed measuring range. | No Prozone effect observed up to 500.0 mg/dL. No prozone effect claimed up to the upper limit of the measuring range. |
| Method Comparison vs. Predicate Device | Demonstrate substantial equivalence in performance to the predicate device. | Correlation Coefficient (r): 0.995 (Passing & Bablok fit and Linear fit).
Slope: 0.9850 (0.9706 - 1.0000) for Passing & Bablok; 0.9771 (0.9612-0.9929) for Linear fit. |
| Matrix Comparison (Serum vs. Plasma) | Demonstrate correlation between serum and various plasma types (Lithium Heparin, Sodium Heparin, Di-Potassium EDTA, Tri-Potassium EDTA). | Lithium Heparin: r=0.997, Slope=0.99 (0.95-1.03).
Sodium Heparin: r=0.999, Slope=0.99 (0.95-1.01).
Di-Potassium EDTA: r=0.997, Slope=0.97 (0.95-1.01).
Tri-Potassium EDTA: r=0.998, Slope=0.99 (0.95-1.03). |
2. Sample size used for the test set and the data provenance:
- Precision (Repeatability/Reproducibility):
- Inter-Assay: 5 samples at different concentrations. 2 replicates per sample, 2 runs per day, for 28 testing days. (Total of 280 measurements per sample level over the entire study for 5 levels? The specific number of measurements is implied but not explicitly stated as 'n=X' for the total dataset.)
- Intra-Assay: 5 samples at different concentrations. 20 replicates per sample, run on 3 different runs. (Total of 300 measurements over the entire study: 5 samples x 20 replicates x 3 runs).
- Linearity: 2 different reagent lots. No specific sample count is given, but "dilution series" are implied.
- Analytical Sensitivity (LoB, LoD, LoQ):
- LoB: 4 saline samples (zero-analyte) for two reagent lots, tested in 5 replicates in 3 different runs. (Total of 60 measurements per lot).
- LoD & LoQ: The guidelined CLSI EP17-A2 was followed, but specific sample counts were not explicitly stated for these sections beyond the general method.
- Interference: "Low: ~30 mg/dL" and "High: ~50 mg/dL" Lp(a) concentrations. Two aliquots of serum pool per concentration. Tested in different replicates for paired difference and then diluted. Specific numbers of individual patient samples tested for interference effects for each substance are not provided.
- Stability (On Board Calibration): 5 samples at different concentrations (20 mg/dL to 100 mg/dL).
- Prozone Study: Last calibrator level used to reach a high concentration (approx. 500.0 mg/dL). Sample diluted in saline.
- Method Comparison vs. Predicate Device: Not explicitly stated but usually involves a significant number of patient samples covering the assay range to demonstrate correlation. The CLSI document EP09c 3rd Edition was followed.
- Matrix Comparison:
- Lithium Heparin: 57 serum and Lithium Heparin plasma samples "derived from the same patients", tested in duplicate.
- Sodium Heparin: 58 serum and Sodium Heparin plasma samples "derived from the same patients", tested in duplicate.
- Di-Potassium EDTA: 57 serum and Di-Potassium EDTA plasma samples "derived from the same patients", tested in duplicate.
- Tri-Potassium EDTA: 56 serum and Tri-Potassium EDTA plasma samples "derived from the same patients", tested in duplicate.
Data Provenance: The document does not explicitly state the country of origin of the data or whether the studies were retrospective or prospective, beyond stating that they used "human serum and plasma" samples. Given it's a 510(k) submission from an Italian company for an in vitro diagnostic device, the samples are typically human biological samples obtained under ethical guidelines, often from healthy volunteers or patient populations relevant to the assay's indication.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience):
Not applicable for this type of IVD device. The "ground truth" for quantitative assays is established by reference methods, traceable calibrators, and robust analytical procedures, not through expert consensus like in image interpretation. The performance is assessed by direct measurement and comparison to established analytical performance metrics and standard reference materials.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set:
Not applicable. Adjudication methods are typically used in clinical trials involving human interpretation of medical images or symptoms where there might be inter-reader variability. For an IVD assay, performance is determined by instrumental measurements and statistical analysis against predefined acceptance criteria.
5. 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:
Not applicable. This is an in vitro diagnostic device, not an AI/ML-driven analysis tool for human readers. No human interpretation or AI assistance is involved in its direct operation or intended use.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
Not applicable in the context of an AI algorithm. The device, Lp(a) Ultra, is a standalone reagent (assay) used on an automated analyzer (Beckman Coulter AU680). Its performance is inherently "standalone" in that it measures analyte concentration without human interpretation of raw signals; humans perform the testing and interpret the numerical results from the analyzer.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc):
The "ground truth" for this in vitro diagnostic assay is established through:
- Reference standards and calibrators: Used to define known concentrations of Lp(a).
- Defined analytical methods: Following recognized guidelines (e.g., CLSI EP05-A3, EP15-A3, EP06, EP17-A2, EP07, EP25-A, EP09c) for precision, linearity, sensitivity, interference, etc.
- Comparison to a legally marketed predicate device: "Ground truth" for demonstrating substantial equivalence is the performance of the predicate device (Diazyme Lipoprotein (a) Assay).
- Known concentrations in spiked samples or characterized biological samples: Used for studies like linearity and interference.
8. The sample size for the training set:
Not applicable. This is not an AI/ML device, so there is no "training set" in that sense. The device is a chemical reagent kit with defined analytical characteristics.
9. How the ground truth for the training set was established:
Not applicable. See point 8.
§ 866.5600 Low-density lipoprotein immunological test system.
(a)
Identification. A low-density lipoprotein immunological test system is a device that consists of the reagents used to measure by immunochemical techniques the low-density lipoprotein in serum and other body fluids. Measurement of low-density lipoprotein in serum may aid in the diagnosis of disorders of lipid (fat) metabolism and help to identify young persons at risk from cardiovascular diseases.(b)
Classification. Class II (performance standards).