(132 days)
Assay for the in vitro quantitative determination of free thyroxine in human serum and plasma. Measurements obtained by this device are used in the diagnosis and treatment of thyroid disease.
The electrochemiluminescence immunoassay "ECLIA" is intended for use on the cobas e 411 immunoassay analyzer.
The Elecsys FT4 III immunoassay is used for the in vitro quantitative determination of free thyroxine in human serum and plasma. It is intended for use on the cobas e 411 immunoassay analyzer. The cobas e family of analyzers uses electrochemiluminescence immunoassay "ECLIA" technology. The reagent working solutions include: Rackpack (kit placed on analyzer) M: Streptavidin-coated microparticles. R1: Anti-T4-Abbiotin. R2: Anti-T4-AbRu(bpy).
This document describes the premarket notification for the Elecsys FT4 III device, an immunoassay for the quantitative determination of free thyroxine. The submission details the non-clinical performance evaluation of the device as it relates to its substantial equivalence to a predicate device.
Here's an analysis of the provided information regarding acceptance criteria and supporting studies:
1. Table of Acceptance Criteria and Reported Device Performance
The document provides a comparative assay table between the Elecsys FT4 II (predicate device) and Elecsys FT4 III (candidate device), and also presents detailed studies for various performance characteristics. I will synthesize the acceptance criteria (often implied by the "same" or within a stated range compared to the predicate, or by specific criteria in the study descriptions) and the reported performance as presented in the document.
| Feature | Acceptance Criteria (Implied/Stated) | Reported Device Performance (Elecsys FT4 III) |
|---|---|---|
| Intended Use | Same as predicate. | "Assay for the in vitro quantitative determination of free thyroxine in human serum and plasma. Measurements obtained by this device are used in the diagnosis and treatment of thyroid disease. The electrochemiluminescence immunoassay "ECLIA" is intended for use on the cobas e 411 immunoassay analyzer." (Matches predicate, with specified analyzer). |
| Detection Method | Same as predicate. | Same: Electrochemiluminescent Assay. |
| Applications/Test Time | Same as predicate. | Same: 18 minute application. |
| Sample/Reagent Ratio | Same as predicate. | Same: 15 µL. |
| Sample Type/Matrix | Same as predicate. | Same: Serum and plasma (Li-heparin, K2-EDTA, K3-EDTA). |
| Calibration Method | Same as predicate. | Same: 2-point calibration. |
| Calibration Interval | Same as predicate. | Same: Once per reagent lot (renewed after 1 month/28 days when using same lot, after 7 days using same kit, or as required by QC). |
| Controls | Same as predicate. | Same: PreciControl Universal. |
| Reagent Stability | Same as predicate: Unopened up to expiration date; After opening at 2-8°C, 84 days (12 weeks); On the analyzers, 28 days (4 weeks); or 56 days (8 weeks) with alternative storage, total onboard not exceeding 120 hours. | Studies executed to support these claims. Study 1: "Reagent stability after first opening at 2-8°C (84 days)"; Study 2: "On board reagent stability (28 days)"; Study 3: "Alternate storage in the refrigerator / on the analyzer (56 days, not exceeding 120 hours onboard)". Real-time stability study ongoing for shelf-life claim. Performance for early time-points supports stability. |
| Measuring Range | Same as predicate. | Same: 0.101-7.77 ng/dL (1.3-100 pmol/L). |
| LoB (Limit of Blank) | Values within acceptable limits compared to predicate (predicate: 0.03 ng/dL). | Reported: 0.02 ng/dL (0.3 pmol/L). (Better than predicate). |
| LoD (Limit of Detection) | Values within acceptable limits compared to predicate (predicate: 0.05 ng/dL). | Reported: 0.04 ng/dL (0.5 pmol/L). (Better than predicate). |
| LoO (Limit of Quantitation) | Same as predicate. | Same: 0.101 ng/dL (1.3 pmol/L). |
| **Analytical Specificity | ||
| (Cross-reactivity)** | Cross-reactivity percentages for various substances should be comparable to or better than the predicate. | The reported cross-reactivity percentages for Elecsys FT4 III are identical to those of the predicate device for L-T3, D-T3, rT3, 3-iodo-L-tyrosine, 3,5-diiodo-L-tyrosine, 3,3',5-triiodothyracetic acid, and 3,3',5,5'-tetraiodothyroacetic acid, at similar tested concentrations. |
| Endogenous Interferences | Criterion: Recovery of ≤ ± 0.6 pmol/L of initial value ≤ 6 pmol/L and ± 10 % of initial value > 6 pmol/L for Bilirubin, Hemoglobin, Intralipid, Rheumatoid factors, IgG, IgA, IgM. Biotin interference should be improved compared to the predicate, with a specific threshold. | Bilirubin: ≤ 701 µmol/L or ≤ 41 mg/dL; Hemoglobin: ≤ 0.621 mmol/L or ≤ 1000 mg/dL; Intralipid: ≤ 2000 mg/dL; Rheumatoid factors: ≤ 1200 IU/mL; IgG: ≤ 7 g/dL; IgA: ≤ 1.6 g/dL; IgM: ≤ 1 g/dL. |
| Biotin Interference: Specimens with biotin concentrations up to 104 ng/mL demonstrated ≤ 10% bias. Biotin concentrations > 104 ng/mL can lead to higher positive bias. This is an improvement from predicate's limitation of "> 5 mg/day" and "until at least 8 hours following the last biotin administration". The new device quantifies the effect and states "Do not test samples from patients who take biotin." | ||
| Special Thyroid Drugs | No interference with the assay found for specified drugs at given concentrations. Criterion: Recovery within ± 10 % of initial value. | No interference was found for Iodide, Carbimazole, Thiamazole, Propylthiouracil, Perchlorate, Propranolol, Amiodarone, Prednisolone, Hydrocortisone, Flurocortolone, Octreotide at the specified concentrations. (These are the same drugs and concentrations as the predicate, and "no interference" finding is consistent). |
| Precision | Meeting CLSI guideline EP05-A3 for repeatability and intermediate precision. | Precision was evaluated on a single cobas e 411 analyzer according to CLSI guideline EP05-A3. Results supporting this are expected to be presented in the full study report (not explicitly tabularized here, but stated that it was evaluated). |
| Sample Matrix Comparison | Acceptable comparison values between serum and various plasma types. | Comparison study performed between Serum, K2-EDTA plasma, and K3-EDTA plasma. Findings indicate acceptable performance, as it is presented as a completed study without issues. |
| Method Comparison to Predicate | "Assess the bias" between the two assays. Substantial equivalence to predicate expected. | 141 serum samples were measured. FT4 values ranged between 2.31 and 91.9 pmol/L for the Reference Method (X). The study aims to support substantial equivalence. (Specific statistics on bias are not provided in this summary, but the study execution implies findings were acceptable). |
| Lot Calibration Stability | Calibration of a reagent lot is recommended every 28 days (4 weeks), and fresh reagent kits of the same lot can use the Day 0 calibration curve within this period. Recovery of samples within acceptable limits after 36 days using a new kit of the same lot with Day 0 calibration curve. | Elecsys FT4 III was calibrated with a fresh reagent kit on day 0. After 36 days a new reagent kit of the same lot was used and recovery of samples was determined using the calibration curve of day 0. (The study implies acceptable recovery, as "Lot calibration study" is listed as successfully completed). |
| On-board Cal Stability | Reagent kits can be stored on board for up to 7 days without a new calibration. | Samples measured on Day 0 and Day 8 (on-board condition) using the Day 0 calibration curve. (The study implies acceptable results, as "On-board Calibration Stability" is listed as successfully completed). |
2. Sample Size Used for the Test Set and the Data Provenance
-
Precision (Repeatability and Intermediate Precision):
- Sample Size: Two levels of control (PreciControl Universal) and five human sera (native single donors, native serum pools, spiked serum pools). Tested in 2 replicates per run, 2 runs per day for 21 days.
- Data Provenance: Not explicitly stated, but "human sera" and "single donors" suggest human samples, likely retrospective pools or prospectively collected for the study. No country of origin mentioned.
-
Endogenous Interferences (Biotin):
- Sample Size: Three human serum samples containing low, mid, and high concentrations of fT4, tested in accordance with CLSI EP07-A2.
- Data Provenance: Human serum samples (single donors, native as well as spiked).
-
Sample Matrix Comparison:
- Sample Size: 53 serum/plasma pairs per sample material (serum, K2-EDTA plasma, K3-EDTA plasma).
- Data Provenance: Native single donors and pools, as well as spiked samples; human.
-
Method Comparison to Predicate:
- Sample Size: 141 serum samples (138 native human serum samples and 3 spiked sample pools).
- Data Provenance: Human serum samples.
-
Reagent Stability (After First Opening, On-board, Alternate Storage):
- Sample Size: Six human serum (HS) samples (native single donor serum, native serum pools, and spiked serum pools) and two controls (PreciControl Universal). Tested in duplicate (for "first opening" and "alternate storage") or two-fold determination (for "on-board").
- Data Provenance: Human serum samples.
-
Reagent Real-time Stability:
- Sample Size: Three human serum samples.
- Data Provenance: Human serum samples (stored at -20°C).
-
Calibration Stability (Lot and On-board):
- Sample Size: Five human serum (HS) samples (single donor serum, native serum pools, and spiked serum pools) and two controls (PreciControl Universal). Tested in two-fold determination.
- Data Provenance: Human serum samples.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and the Qualifications of Those Experts
This document describes the performance evaluation of an in vitro diagnostic (IVD) device, specifically an immunoassay for free thyroxine. For such devices, "ground truth" is typically established by reference methods, consensus values for control materials, or by testing against well-characterized clinical samples in comparison to an established predicate device. It is not generally established by a panel of human experts reviewing individual cases in the same way an imaging AI algorithm might be.
Therefore, the concept of "number of experts used to establish the ground truth" as it applies to imaging or clinical decision support AI is not applicable here. The ground truth for these types of studies is based on biochemical measurements and analytical comparisons.
4. Adjudication Method for the Test Set
Again, for an IVD device evaluating a biomarker, traditional "adjudication" by human experts in the context of clinical interpretation of individual cases (e.g., 2+1, 3+1 for discordances in imaging) is not applicable. The performance is adjudicated against analytical measurements and statistical criteria defined in CLSI guidelines and internal validation plans.
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 (immunoassay), not an AI-powered clinical decision support or imaging interpretation tool that "assists human readers." Therefore, no MRMC study with human readers improving with AI assistance would be performed.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
The Elecsys FT4 III is an automated immunoassay intended for use on the cobas e 411 analyzer. Its performance is inherently "standalone" in the sense that the analyzer and reagents perform the measurement autonomously. A human laboratory professional loads samples and reagents, and reviews the results. The studies described (precision, interference, method comparison) are all evaluating the standalone analytical performance of this device.
Therefore, yes, the studies described inherently evaluate the "standalone" performance of the algorithm/device system in quantifying free thyroxine in samples.
7. The Type of Ground Truth Used
The ground truth for the various studies is established through:
- Reference methods / Predicate Device: For method comparison, the Elecsys FT4 II assay (K131244) served as the reference method ("X").
- Known concentrations: For interference studies (e.g., biotin, cross-reactivity), samples were spiked with known concentrations of interfering substances or analytes.
- Industry Standards / Guidelines: Adherence to CLSI guidelines (e.g., EP05-A3 for precision, EP07-A2 for interferences) implies that the ground truth for acceptability criteria is based on established laboratory practice and industry consensus for robust analytical performance.
- Controls and Calibrators: PreciControl Universal and CalSet FT4 III are used, which have established target values and ranges for quality control and calibration, respectively.
8. The Sample Size for the Training Set
This document describes the performance validation of a commercially manufactured in vitro diagnostic device, not the development of a machine learning algorithm. Therefore, the concept of a "training set" in the context of AI model development is not directly applicable here.
The device's underlying principles are based on established electrochemiluminescence immunoassay (ECLIA) technology and biochemically derived reagent formulations. While there is a development process that optimizes reagents and assay parameters, this is not typically referred to as "training" in the AI sense, nor does the document provide details about such a "training set" if it were analogous.
9. How the Ground Truth for the Training Set Was Established
Given that "training set" as understood in AI/ML model development is not directly applicable to this IVD device's validation data, this question is not applicable in the context of the provided document. The "ground truth" for the device's development (the underlying chemical and physical principles on which it operates) would have been established through extensive biochemical research, assay development, and optimization processes, rather than a discrete "training set" with established ground truth labels in the AI sense.
§ 862.1695 Free thyroxine test system.
(a)
Identification. A free thyroxine test system is a device intended to measure free (not protein bound) thyroxine (thyroid hormone) in serum or plasma. Levels of free thyroxine in plasma are thought to reflect the amount of thyroxine hormone available to the cells and may therefore determine the clinical metabolic status of thyroxine. Measurements obtained by this device are used in the diagnosis and treatment of thyroid diseases.(b)
Classification. Class II.