Search Results

The DiaSorin LIAISON® TSH assay is an in vitro chemiluminescent immunoassay intended for the quantitative determination of thyroid stimulating hormone (TSH), also known as thyrotropin and thyrotropic hormone, in human serum. The test has to be performed on the LIAISON® XL Analyzer. Measurements of thyroid stimulating hormone produced by the anterior pituitary are used in the diagnosis of thyroid or pituitary disorders.

The DiaSorin LIAISON® Control Thyroid 1, LIAISON® Control Thyroid 2 and LIAISON® Control Thyroid 3 are intended for use as assayed quality control samples to monitor the accuracy and precision of the DiaSorin LIAISON® TSH assay.

Device Description

The DiaSorin LIAISON® TSH assay is an in vitro chemiluminescent immunoassay for the quantitative determination of thyroid stimulating hormone (TSH), also known as thyrotropin and thyrotropic hormone, in human serum. The test has to be performed on the LIAISON® XL Analyzer. Measurements of thyroid stimulating hormone produced by the anterior pituitary are used in the diagnosis of thyroid or pituitary disorders.

The method for the quantitative determination of TSH is a sandwich chemiluminescence immunoassay. A specific mouse monoclonal antibody is coated on the magnetic particles (solid phase); another monoclonal antibody is linked to an isoluminol derivative (isoluminol-antibody conjugate). All assay steps and incubations are performed by the LIAISON® XL Analyzer.

Results are determined by a 2 point calibration conversion of the master curve to a working curve. The light signal, and hence the amount of isoluminol-antibody conjugate, is measured by a photomultiplier as relative light units (RLU) and is inversely related to TSH concentration present in calibrators, patient samples or controls.

The LIAISON® Control Thyroid 1, 2 and 3 are prepared from human serum at target values of 0.5-0.78 mIU/L. 7.0-9.48 mIU/L and 42.7-57.7 mIU/L.

AI/ML Overview

The provided document describes the DiaSorin LIAISON® TSH assay, an in vitro chemiluminescent immunoassay for the quantitative determination of thyroid stimulating hormone (TSH) in human serum, and associated quality controls (LIAISON® Control Thyroid 1, 2, and 3). The submission aims to demonstrate substantial equivalence to a predicate device.

Here's an analysis of the acceptance criteria and the studies performed:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly present a table of "acceptance criteria" but rather describes various performance characteristics and their results, which serve as evidence of the device meeting certain performance standards for substantial equivalence. I will infer the acceptance criteria from the context of typical lab assay performance and the results presented.

Performance Characteristic	Acceptance Criteria (Inferred)	Reported Device Performance
Method Comparison	Good agreement with a commercially available immunoassay (predicate device or similar). Typically, R² near 1, slope near 1, and intercept near 0.	181 samples compared to a commercially available immunoassay. Passing & Bablok regression: y = 1.005x - 0.0030, R² = 0.9748. 95% CI for slope: 0.988 to 1.026. 95% CI for intercept: -0.0873 to 0.0411 mIU/L.
Reference Range	Establishment of a normal range for healthy individuals.	Established in 130 apparently healthy individuals: 0.357 – 4.789 mIU/L (Median: 1.438 mIU/L).
Reproducibility/Precision	Low coefficient of variation (%CV) for within-run and total precision across different concentrations (low, medium, high). Values typically less than 10-15% depending on concentration.	For 6 serum samples and 3 controls, total %CVs ranged from 0.3% to 5.5%. For example, control KC1 (0.6401 mIU/L) had a total %CV of 4.3%, and sample 6 (82.56 mIU/L) had a total %CV of 4.2%.
Linearity	Demonstration of a linear response across the measuring range. Typically, R² near 1, slope near 1, and intercept near 0 within the reportable range.	Linear from 0.02 mIU/L to 90 mIU/L. Passing & Bablok regression: y = 0.9807x + 0.0013, R² = 0.9997.
Limit of Blank (LoB)	LoB value established.	0.014 mIU/L
Limit of Detection (LoD)	LoD value established.	0.02 mIU/L
Limit of Quantitation (LoQ)	LoQ established, typically at < 20% CV.	0.02 mIU/L (defined as inter-assay precision of < 20% CV).
Interfering Substances	Non-significant interference (≤10% difference) with common endogenous and exogenous substances.	No significant interference (≤10% difference) reported for various substances including Triglycerides (500 mg/dL), Hemoglobin (250 mg/dL), Bilirubin (conjugated & unconjugated 5 mg/dL), Albumin (6000 mg/dL), and various hormones (LH, FSH, hGH, hCG, HAMA).
Traceability	Traceable to an international reference material.	Traceable to the WHO reference material: 2nd IRP WHO 80/558 standard.
Control Stability	Lyophilized controls stable until expiration. Reconstituted controls stable for specified period.	Lyophilized controls stable until expiration date. Reconstituted controls stable for up to 48 hours at 2-8°C, or up to 1 month frozen at -20°C.

2. Sample Size Used for the Test Set and Data Provenance

Method Comparison: 181 samples. Data provenance is not explicitly stated (e.g., country of origin, retrospective/prospective), but it is implied to be clinical human serum samples.
Reference Range/Expected Values: 130 apparently healthy human serum samples. Data provenance is not explicitly stated.
Reproducibility/Precision: A coded panel of 6 frozen serum samples (low, medium, high), plus 3 levels of LIAISON® Control Thyroid. The study was performed at DiaSorin Inc., suggesting a prospective in-house study.
Linearity: One high TSH sample serially diluted to generate 7 concentrations.
LoB/LoD/LoQ: Six serum specimens at low TSH doses, assayed in 72 determinations.
Interfering Substances: Not explicitly stated how many samples per substance, but implied to be controlled laboratory experiments.

The provenance (country of origin) is not provided for clinical samples. The studies for reproducibility, linearity, LoB/LoD/LoQ, and interfering substances appear to be prospective laboratory studies.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

For this type of in vitro diagnostic device (quantitative immunoassay), the "ground truth" for the test set is established by the reference methods or established values, not human expert consensus.

Method Comparison: The predicate or commercially available immunoassay serves as the reference for comparison. No human experts are used for ground truth for individual samples; the accuracy of the comparator assay is assumed.
Reference Range: "Apparently healthy test subjects" were used. Their health status would have been determined by clinical criteria, likely by medical professionals (e.g., physicians) but specific numbers or qualifications of experts are not relevant or provided in this context.
Reproducibility/Precision, Linearity, LoB/LoD/LoQ, Interfering Substances, Traceability: These performance characteristics are evaluated against known standards, spiked samples, or by repeated measurements, not through expert consensus on individual results. The "ground truth" for these studies is typically defined by the experimental setup and reference materials.

4. Adjudication Method for the Test Set

Adjudication methods (e.g., 2+1, 3+1) are typically used in studies where human readers are part of the ground truth establishment (e.g., interpretation of medical images). This is not applicable to a quantitative immunoassay where numerical values are measured and compared against established analytical performance goals or a predicate device. The "adjudication" is based on statistical methods (e.g., Passing & Bablok regression, %CV calculations) comparing the device's numerical output to expected values or reference methods.

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

An MRMC study is not applicable as this is an in vitro diagnostic immunoassay, not an AI-assisted diagnostic imaging or interpretation device that involves human readers. Therefore, no such study was performed, and no effect size for human reader improvement with AI assistance is provided.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was done

This device is an automated immunoassay system (LIAISON® XL Analyzer) that quantitatively determines TSH levels. Its performance is inherently "standalone" in the sense that the analyzer yields a numerical result without human interpretation of raw signals; human involvement is in sample loading, calibration, quality control, and result review. The studies described (Method Comparison, Precision, Linearity, LoB/LoD/LoQ, Interference) all evaluate the performance of the automated assay system in generating these numerical results.

7. The Type of Ground Truth Used

Method Comparison: The results obtained from a "commercially available immunoassay" (predicate or similar) served as the comparative truth.
Reference Range: Established by measuring TSH levels in "apparently healthy test subjects."
Reproducibility/Precision: Determined by repeated measurements of samples (frozen serum, controls) with expected concentrations.
Linearity: Established by serially diluting a high TSH sample, where the concentration of each dilution is known relative to the original sample.
LoB/LoD/LoQ: Established through statistical methods on samples with very low or zero analyte concentrations.
Interfering Substances: Established by comparing results from samples spiked with known interfering substances against unspiked control samples.
Traceability: Traceable to the WHO reference material: 2nd IRP WHO 80/558 standard.

In essence, the ground truth is based on reference methods, known concentrations of reference materials, and established clinical characteristics (e.g., healthy individuals).

8. The Sample Size for the Training Set

This document describes performance validation studies for a diagnostic assay, not a machine learning or AI model that typically has a distinct "training set." Therefore, a "training set" in that context is not applicable. The development of the assay itself would have involved extensive R&D and optimization, but those details are not provided as specific "training set" sizes here.

9. How the Ground Truth for the Training Set Was Established

As noted above, the concept of a "training set" in the context of an AI/ML model with corresponding ground truth establishment is not directly applicable to this immunoassay device approval. The "ground truth" for the various analytical performance characteristics (e.g., linearity, precision, interference) are established via standard analytical chemistry and immunoassay validation practices using reference materials, spiked samples, and comparative methods, as described in point 7.

Ask a Question

Ask a specific question about this device

Page 1 of 1