Sensitivity
Historically, the "quality" of a serum TSH method has been determined from a clinical benchmark - the assay's ability to discriminate euthyroid levels (~ 0.4 to 4.0 mIU/L) from the profoundly low (<0.01 mIU/L) TSH concentration typical of overt Graves' thyrotoxicosis. Most TSH methods now claim a detection limit of 0.02 mIU/L or less ("third generation" assays).[202]
Guideline 19. Definition of Functional Sensitivity
Functional Sensitivity should be used to determine the Lowest Detection Limit of the assay.
TSH assay functional sensitivity is defined by the 20% between-run coefficient of variation (CV) determined by the recommended protocol (see Guideline 20).
Guideline 20. Protocol for TSH Functional Sensitivity & Between-Run Precision
Measure human serum pools covering the assay range in at least 10 different runs. The lowest pool value should be 10% above the detection limit and the highest pool value should be 90% of the upper assay limit.
Carry-over should be assessed by analyzing the highest pool followed by the lowest pool.
Use the same test mode as for patient specimens (i.e. singlicate or duplicate).
The instrument operator should be blinded to the presence of test pools in the run.
Runs should be spaced over a clinically representative interval (i.e. 6 to 8 weeks for TSH in an outpatient setting).
Use at least two different lots of reagents and two different instrument calibrations during the testing period.
When running the same assay on two similar instruments, blind duplicates should be run on each instrument periodically to verify correlation.
Manufacturers have largely abandoned the use of the "analytical sensitivity" parameter for determining the sensitivity of a TSH assay because it is calculated from the within-run precision of the zero calibrator which does not reflect the sensitivity of the test in clinical practice.[126,127] Instead, a "functional sensitivity" parameter has been adopted.[202] Functional sensitivity is calculated from the 20% between-run coefficient of variation (CV) for the method and is used to establish the lowest reportable limit for the test.[202]
Functional sensitivity should be determined by strictly adhering to the recommended protocol (Guideline 20) that is designed to assess the minimum detection limit of the assay in clinical practice and ensure that the parameter realistically represents the lowest assay value that can be reliably reported. The protocol is designed to take into account a variety of factors that can influence TSH method imprecision in clinical practice. These include:
Matrix differences between patient serum and the standard diluent
Erosion of precision over time
Lot-to-lot variability in the reagents supplied by the manufacturer
Differences between instrument calibrations and technical operators
Carry-over from high to low specimens[205]
The use of the functional sensitivity limit as the lowest detection limit is a conservative approach to ensure that any TSH result reported is not merely assay "noise". Further, the 20% between-run CV approximates the maximum imprecision required for diagnostic testing ( Table 5 ).
Guideline 21. For Laboratories Performing TSH Testing
Functional sensitivity is the most important performance criterion that should influence the selection of a TSH method. Practical factors such as instrumentation, incubation time, cost, and technical support though important, should be secondary considerations. Laboratories should use calibration intervals that optimize functional sensitivity, even if re-calibration needs to be more frequent than recommended by the manufacturer:
Select a TSH method that has a functional sensitivity ≤0.02 mIU/L
Establish functional sensitivity independent of the manufacturer by using Guideline 20
There is no scientific justification to reflex from a less sensitive to a more sensitive test. (Insensitivity causes falsely high, not falsely low, values that are missed by reflex testing!)
© 2003 Mary Ann Liebert, Inc.
Cite this: Thyrotropin/Thyroid Stimulating Hormone (TSH) Measurement - Medscape - Jan 01, 2003.
