Specificity differences, suboptimal sensitivity, hook effects, and an inability to reliably detect interfering TgAb compromise the clinical utility of current Tg and TgAb methods. All of the IMA methods were prone to underestimate serum Tg in the presence of TgAb, whereas the RIA methods appeared resistant to TgAb interference.

Ethnic differences in TSH were not present when populations with the same relative frequency of thyroid antibodies were compared. TSH upper reference limits may be skewed by TPOAb-negative individuals with occult autoimmune thyroid dysfunction.

Serum thyroglobulin (Tg) measurements are used as a tumor marker for monitoring patients with differentiated thyroid carcinoma. The clinical utility of six different Tg methods [RIA or immunometric assay (IMA)] currently used in Europe and the US was evaluated, with focus on methodologic standardization, sensitivity, interassay precision across the typical clinical monitoring interval (6 to 12 months), "hook" effects (IMA methods), and Tg autoantibody interference. The methods evaluated were: DYNOtest Tg (Henning), OptiQuant Tg (Kronus), SELco Tg (Medipan), Thyroglobulin IRMA (Pasteur), Nichols Chemiluminescent ICMA (Corning Nichols), and an RIA developed by us (USC Endocrine Services Laboratory). The clinical impact of the current methodologic problems on the use of serum Tg measurements is reviewed. Optimal performance goals are recommended for manufacturers developing and laboratories and physicians selecting a serum Tg method to use for serial long-term monitoring of thyroid cancer patients.

Current medical needs dictate that laboratories offer thyrotropin (TSH) assays that can reliably measure low TSH concentrations--a prerequisite for using the more cost-effective TSH-centered strategy currently recommended by the American Thyroid Association. This study reviews the functional performance of the TSH immunometric assay methods currently used in clinical practice. Methodological differences between methods, the rationale for using the 20% interassay CV as the functional sensitivity limit for patient reports, and both TSH-related and non-TSH-related specificity problems are reviewed. We recommend that manufacturers and clinical laboratories use a clinically relevant standard protocol for functional sensitivity assessments. In this protocol, human serum pools are analyzed in random order 10 or more times across a clinically realistic time span (which approximates to 6-8 weeks for TSH measurements used in an outpatient setting). Laboratories should independently establish and periodically check their functional sensitivity by the standard protocol and enlist the manufacturer's help to accomplish this if necessary. Manufacturers should (a) develop promotional material that realistically projects the assay's functional sensitivity expected in a clinical laboratory setting, (b) ensure that new clinical laboratory users achieve the projected functional sensitivity target when using the standard protocol, and (c) focus on the typical functional sensitivity obtained by clinical laboratory users rather than the assay's "generation" achieved under ideal conditions. If manufacturers and laboratories collaborate to solve the sensitivity and specificity problems discussed here, clinical laboratories should be better able to consistently deliver reliable serum TSH measurements across the full range of TSH concentrations encountered in clinical practice.

The prevalence of circulating thyroid autoantibodies (TgAb or antithyroid peroxidase) was increased nearly 3-fold in patients with differentiated thyroid cancers (DTC) compared with the general population (40% vs. 14%, respectively). Serum TgAb (with or without antithyroid peroxidase) was present in 25% of DTC patients and 10% of the general population. Serial postsurgical serum TgAb and serum Tg patterns correlated with the presence or absence of disease. Measurements of serum Tg were made in 87 TgAb-positive sera by a RIA and two immunometric assay (IMA) methods to study TgAb interference. TgAb interference, defined as a significant intermethod discordance (>41.7% coefficient of variation) between the Tg RIA and Tg IMA values relative to TgAb-negative sera, was found in 69% of the TgAb-positive sera. TgAb interference was characterized by higher Tg RIA vs. IMA values and was, in general, more frequent and severe in sera containing high TgAb concentrations. However, some sera displayed marked interference when serum TgAb was low (1-2 IU/mL), whereas other sera with very high TgAb values (>1000 IU/mL) displayed no interference. An agglutination method was found to be too insensitive to detect low TgAb concentrations (1-10 IU/mL) causing interference. Exogenous Tg recovery tests were an unreliable means for detecting TgAb interference. Specifically, the exogenous Tg recovered varied with the type and amount of Tg added and the duration of incubation employed. Further, recoveries of more than 80% were found for some sera displaying gross serum RIA/IMA discordances. The measurement of serum Tg in DTC patients with circulating TgAb is currently problematic. It is important to use a Tg method that provides measurements that are concordant with tumor status. IMA methods are prone to underestimate serum when TgAb is present, increasing the risk that persistent or metastatic DTC will be missed. The RIA method used in this study provided more clinically appropriate serum Tg values in the group of TgAb-positive patients with metastatic DTC. Furthermore, as serial serum TgAb measurements paralleled serial serum Tg RIA measurements, TgAb concentrations may be an additional clinically useful tumor marker parameter for following TgAb-positive patients. Disparities between serial serum Tg and TgAb measurements might alert the physician to the possibility of TgAb interference with the serum Tg measurement and prompt a more cautious use of such data for clinical decision-making.

Clinically relevant interassay precision profiles for thyrotropin (thyroid-stimulating hormone; TSH) were constructed with human serum pools measured over 4-8 weeks by six immunometric assays, in at least two different reagent lots. Functional sensitivities (the concentration at which the interassay CV is < or = 20%) were determined in four to eight clinical laboratories plus the respective manufacturer's laboratory. These studies revealed that the manufacturer's stated functional sensitivity limit is rarely duplicated in clinical practice. Loss of specificity (indicated by artifactually high values) was seen with some methods when used to measure certain unrefrigerated low-TSH sera. Measurement of TSH in four human serum pools (TSH < 0.05-0.25 mIU/L) by 16 different methods (each in at least eight UK or US laboratories) showed that some methods could not reliably distinguish subnormal from normal TSH values. Better pool rankings and fewer misclassifications of low-TSH sera as "normal" were seen with use of assays capable of "third-generation" functional sensitivity (0.01-0.02 mIU/L) than with assays with "second-generation" functional sensitivity (0.1-0.2 mIU/L). Because inter- and intramethod differences in functional sensitivity negatively impact the diagnostic accuracy and cost-effectiveness of a TSH-centered thyroid-testing strategy, laboratories should independently establish an assay's functional sensitivity by a clinically relevant protocol. Moreover, manufacturers should assess functional sensitivity more realistically and improve the robustness of assays to ensure that their performance potential is consistently met in clinical practice.