[Thyroglobulin assay ambiguities].

Thyroglobulin immunometric "sandwich" assays (IMA) have taken over competitive radioimmunoassays, but this assay remains problematic. A human thyroglobulin reference material (CRM 457) has been prepared but is not widely used. That constitutes the main cause of very marked between kit variability of thyroglobulin results. High-dose hook effect, which can falsely decrease the result of a serum with high concentration of thyroglobulin, is not exceptional in one-step assays and should be systematically checked. Selection of monoclonal antibodies with no cross-reactivity with anti-thyroglobulin autoantibodies or of polyclonal antibodies with very high affinity, have reduced the frequency of interference due to autoantibodies, but did not abolish it. Recovery test is used to detect such interference, but with insufficient sensitivity. In fact, recovery determination can be influenced by the nature of thyroglobulin added to the serum (not identical to endogenous thyroglobulin), by the delay of incubation of exogenous thyroglobulin and serum autoantibodies and by the amount of added thyroglobulin. In addition, recovery is often wrongly expressed as the observed/theoretic ratio of final concentrations instead of added concentrations. In untreated Graves' disease patients and despite normal recovery test, thyroglobulin measured by IMA is lower when anti-thyroglobulin autoantibodies are present. Consequently, thyroglobulin result should be interpreted in function of presence or absence of autoantibodies. Development of total (free and autoantibody bound) thyroglobulin assay would be useful to evaluate assay and recovery test performances.