(247 days)
The Vantera® Clinical Analyzer is an automated laboratory test analyzer which measures the 400 MHz proton nuclear magnetic resonance (NMR) spectrum of clinical samples to produce signal amplitudes, converting these signal amplitudes to analyte concentration. The device includes a 400 MHz NMR spectrometer and software to analyze digitized spectral data. This instrumentation is intended to be used with NMR based assays to detect multiple analytes from clinical samples.
The NMR LipoProfile® test, when used with the Vantera® Clinical Analyzer, an automated NMR spectrometer, measures lipoprotein particles to quantify LDL particle number (LDL-P), HDL cholesterol (HDL-C), and triglycerides in human serum and plasma using nuclear magnetic resonance (NMR) spectroscopy. LDL-P and these NMR-derived concentrations of HDL-C and triglycerides are used in conjunction with other lipid measurements and clinical evaluation to aid in the management of lipoprotein disorders associated with cardiovascular disease.
The Vantera Clinical Analyzer is a clinical laboratory analyzer that employs nuclear magnetic resonance spectroscopic detection to quantify multiple analytes in biological fluid specimens, specifically blood plasma and serum. The Vantera Clinical Analyzer system design is divided into 3 major subassemblies: a sample handling assembly, an NMR subassembly, and an enclosure. The Vantera Clinical Analyzer control system is distributed across three separate computers: The Host (1U) controls user interface, data handling, results calculation, system startup and shutdown. The Process Control (4U) schedules and manages all activities required to process a sample, controls all hardware in the sample handling subsystem, and manages remote access to the system. The NMR Control Computer controls all magnet operations. Two of these computers are contained within the Sample Handling Subassembly (1U and 4U) and one in the NMR Subassembly (NMR Console).
The NMR LipoProfile test involves measurement of the 400 MHz proton NMR spectrum of a plasma/serum sample, deconvolution of the composite signal at approximately 0.8 ppm to produce signal amplitudes of the lipoprotein subclasses that contribute to the composite plasma/serum signal, and conversion of these subclass signal amplitudes to lipoprotein subclass concentrations.
The provided 510(k) summary focuses on the analytical performance of the Vantera® Clinical Analyzer and the NMR LipoProfile® test compared to predicate devices, establishing substantial equivalence rather than providing explicit acceptance criteria as would be typical for a novel device. The study described primarily demonstrates that the proposed device performs comparably to its predicate devices in terms of analytical accuracy and precision.
Here's an analysis of the acceptance criteria and study information provided:
1. A table of acceptance criteria and the reported device performance
The document does not explicitly state acceptance criteria in a pass/fail format. Instead, it demonstrates performance by comparing the analytical results of the Vantera® Clinical Analyzer with the NMR LipoProfile® test to its predicate device (NMR Profiler for the assay) across various metrics. The unstated acceptance criteria for each analytical performance metric would be that the proposed device's performance must be comparable to or better than the predicate device's performance.
| Metric | Acceptance Criteria (Implied) | Reported Proposed Device Performance (Vantera® Clinical Analyzer) | Predicate Device Performance (NMR Profiler) |
|---|---|---|---|
| LDL-P | |||
| LoB | Comparable to predicate | 0 nmol/L | 0 nmol/L |
| LoD | Comparable to predicate | 40.7 nmol/L | 41 nmol/L |
| LoQ | Comparable to predicate | 132 nmol/L | 157 nmol/L |
| Measuring Range | Comparable to predicate | 300-3500 nmol/L | 300-3500 nmol/L |
| Linearity Regression (Y=mX+b) | R² comparable to predicate | y=1.02x+7.82, R²=0.9949 | y=0.99x-22.37, R²=0.9979 |
| Within-Run Precision (CV%) | Comparable to predicate | Level 1: 5.8%, Level 2: 3.0%, Level 3: 2.7% | Level 1: 5.0%, Level 2: 4.3%, Level 3: 3.7% |
| Within-Lab Precision (CV%) | Comparable to predicate | Level 1: 5.3%, Level 2: 4.0%, Level 3: 3.9% | Level 1: 7.6%, Level 2: 4.5%, Level 3: 4.3% |
| Method Comparison (Correlation R) | Comparable to predicate | R=0.978 | R=0.973 |
| Interference Study | No significant interference for tested substances | Salicylic acid (≥ 1.3mmol/L) and Clopidogrel hydrogensulfate (≥ 95.7 µmol/L) determined to interfere. | No interference found for 5 endogenous & 22 exogenous substances. |
| Specimen Stability (Refrigerated) | Comparable to predicate | 6 days | 5 days |
| Triglycerides | |||
| LoB | Comparable to predicate | 1.1 mg/dL | 1.4 mg/dL |
| LoD | Comparable to predicate | 2.4 mg/dL | 2.6 mg/dL |
| LoQ | Comparable to predicate | 4 mg/dL | 2.6 mg/dL |
| Measuring Range | Comparable to predicate | 5-1100 mg/dL | 5-1100 mg/dL |
| Linearity Regression (Y=mX+b) | R² comparable to predicate | y=1.008x-0.3979, R²=0.9999 | y=0.95x-12.21, R²=0.999 |
| Within-Run Precision (CV%) | Comparable to predicate | Level 1: 2.3%, Level 2: 2.1%, Level 3: 1.2% | Level 1: 2.6%, Level 2: 1.8%, Level 3: 1.3% |
| Within-Lab Precision (CV%) | Comparable to predicate | Level 1: 2.3%, Level 2: 2.4%, Level 3: 2.7% | Level 1: 3.6%, Level 2: 2.6%, Level 3: 2.5% |
| Method Comparison (Correlation R) | Comparable to predicate | R=0.998 | R=1.00 |
| Interference Study | No significant interference for tested substances | No interference found for 7 endogenous & 23 exogenous substances. | Ibuprofen may interfere with TG measurement at and above 210µg/mL for 5 endogenous & 22 exogenous substances. |
| Specimen Stability (Refrigerated) | Comparable to predicate | 6 days | 10 days |
| HDL-C | |||
| LoB | Comparable to predicate | 2.7 mg/dL | 4.3 mg/dL |
| LoD | Comparable to predicate | 3.5 mg/dL | 5.2 mg/dL |
| LoQ | Comparable to predicate | 4 mg/dL | 5.2 mg/dL |
| Measuring Range | Comparable to predicate | 7-140 mg/dL | 7-140 mg/dL |
| Linearity Regression (Y=mX+b) | R² comparable to predicate | y=1.049x-0.3459, R²=0.9961 | y=1.004x-0.5956, R²=0.9998 |
| Within-Run Precision (CV%) | Comparable to predicate | Level 1: 4.0%, Level 2: 2.8%, Level 3: 2.6% | Level 1: 2.0%, Level 2: 1.9%, Level 3: 0.9% |
| Within-Lab Precision (CV%) | Comparable to predicate | Level 1: 2.8%, Level 2: 2.0%, Level 3: 1.8% | Level 1: 3.3%, Level 2: 2.0%, Level 3: 1.8% |
| Method Comparison (Correlation R) | Comparable to predicate | R=0.989 | R=0.999 |
| Interference Study | No significant interference for tested substances | No interference found for 7 endogenous & 23 exogenous substances. | No interference found for 5 endogenous & 22 exogenous substances. |
| Specimen Stability (Refrigerated) | Comparable to predicate | 6 days | 10 days |
Study Proving Acceptance Criteria:
The study conducted was an analytical validation comparing the performance of the Vantera® Clinical Analyzer with the NMR LipoProfile® test to its predicate device (NMR Profiler for the assay) across various analytical parameters. The overall conclusion is that the new device is "as safe and effective as its predicate device."
2. Sample size used for the test set and the data provenance
- Test Sets (Analytical Studies):
- Analytical Sensitivity (LoB, LoD, LoQ): Five serum pools for low concentrations tested in replicates of 4 for 3 days. Non-lipoprotein specimens analyzed 60 consecutive times for 3 days for LoB.
- Assay Precision (Within-run & Within-Laboratory): 20 replicates of three patient serum pools in the same run and in 20 different runs over 20 days. Reproducibility study used 5 levels of serum panels tested for 5 days, 6 runs per day, 2 replicates per run at 3 sites (n=60 per panel per site, total N=177-180 across all sites for each panel).
- Linearity: Three serum pools prepared from patient specimens mixed and diluted to produce eleven (for LDL-P) or twelve (for TG and HDL-C) different samples, with four replicates of each pool analyzed.
- Method Comparison:
- LDL-P: n=1483 serum samples.
- HDL-C: n=1518 serum samples.
- Triglycerides: n=1520 serum samples.
- Interfering Substances: 7 endogenous agents and 23 drugs were screened.
- Reference Range: Serum samples (n=452) from apparently healthy men (n=158) and women (n=294).
- Data Provenance: The document does not specify the country of origin. The studies were described as "analytical validations" and included testing using "patient specimens" and "serum pools." There is no explicit mention of the data being either retrospective or prospective, but the nature of the analytical studies suggests controlled laboratory environments rather than a large-scale clinical trial with patient follow-up. For the reference range, it states "serum samples... were analyzed from apparently healthy men and women," which implies a prospective collection for this specific purpose or a well-characterized existing cohort. The MESA (Multi-Ethnic Study of Atherosclerosis) is mentioned for the predicate device's reference range, suggesting a US context for that, but it's not explicitly stated for the proposed device's reference population.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
The document does not describe the use of human experts to establish "ground truth" for the test set in the context of interpretation or diagnosis. This device is an automated laboratory analyzer for quantifying analytes. The "ground truth" for its analytical performance studies (e.g., precision, linearity, method comparison) is established by comparing its measurements to a reference method or known concentrations, or by assessing consistency internally.
4. Adjudication method for the test set
Not applicable. There was no clinical study involving human interpretation or diagnosis that would require an adjudication method. The testing involved direct analytical measurements.
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 (IVD) device for quantifying analytes (LDL-P, HDL-C, Triglycerides), not an AI-assisted diagnostic imaging or interpretation device that would involve human readers.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
Yes, the primary performance studies presented are "standalone" in the sense that they assess the device's analytical performance (algorithm + instrument) in quantifying the specified analytes without human-in-the-loop performance for diagnosis or interpretation. The device's output is explicit numerical analyte concentrations.
7. The type of ground truth used
The ground truth used for verifying the analytical performance of the device was:
- Reference Methods/Known Concentrations: For analytical sensitivity (LoB, LoD, LoQ), linearity, and precision, the "ground truth" was established through precisely prepared serum pools and non-lipoprotein specimens with known or target concentrations, or through statistical determination methods (e.g., EP17-A).
- Comparison to Predicate Device: For method comparison studies, the "ground truth" or reference was the measurements obtained from the legally marketed predicate device (NMR Profiler) using patient samples. The goal was to show high correlation and similar results between the two devices.
- CLSI Guidelines: Standardized guidelines (e.g., EP5-A2, EP6-A, EP7-A2, EP9-A2, EP17-A) from the Clinical and Laboratory Standards Institute (CLSI) were referenced for establishing protocols for these analytical validations.
8. The sample size for the training set
The document does not explicitly mention a "training set" in the context of machine learning or AI models. This device is an automated NMR spectrometer that measures signals and converts them to concentrations based on specified deconvolution analysis models. The development of these deconvolution models would have involved a form of "training" or optimization, but the document does not detail the dataset size or methodology used for this prior model development. The document focuses on the analytical validation of the manufactured device.
9. How the ground truth for the training set was established
As there is no explicit mention of a "training set" in the conventional AI/ML sense, this question cannot be directly answered from the provided text. However, the assay description mentions:
"The NMR signals from the various lipoprotein subclasses have unique and distinctive frequencies and lineshapes, each of which is accounted for in the deconvolution analysis model. Each subclass signal amplitude is proportional to the number of subclass particles emitting the signal, which enables subclass particle concentrations to be calculated from the subclass signal amplitudes derived from the spectral deconvolution analysis."
This suggests that the deconvolution analysis model was developed using a "ground truth" based on the established biophysical properties of lipoprotein subclasses and their NMR spectral characteristics. This likely involved:
- Carefully characterized lipoprotein samples with known subclass concentrations.
- Expert knowledge of NMR spectroscopy and signal processing.
- Calibration against established reference methods for lipoprotein analysis.
The document does not provide details on the specific data sets or expert consensus used for the initial development and establishment of this deconvolution model.
§ 862.1705 Triglyceride test system.
(a)
Identification. A triglyceride test system is a device intended to measure triglyceride (neutral fat) in serum and plasma. Measurements obtained by this device are used in the diagnosis and treatment of patients with diabetes mellitus, nephrosis, liver obstruction, other diseases involving lipid metabolism, or various endocrine disorders.(b)
Classification. Class I (general controls). The device is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to § 862.9.