A 32-year-old female patient presented complaining of increased flushing, perspiration and heat intolerance of 3 months’ duration. Medical history included idiopathic thrombocytopenic purpura of 4 years’ duration, which had been treated by splenectomy after failed immunosuppression with prednisone and azathioprine and was currently in remission. On examination, she was found to be clinically euthyroid without a goiter. She had also developed a diffuse erythematous papular rash on the face and back, with bullous lesions on the chest. Immunofluorescent antibody studies performed on a punch biopsy of skin were positive for several autoantibodies, leading to a diagnosis of subacute cutaneous lupus erythematosus. This diagnosis was further characterized by positive titers of nuclear and double-stranded DNA autoantibodies in the serum. Thyroid function testing on an Advia Centaur® Immunoassay System (Siemens Medical Solutions Diagnostics) revealed an increased concentration of serum free thyroxine (FT4)1 of 90.1 pmol/L (6.97 ng/dL) (reference range 11.5–22.7 pmol/L), a nonsuppressed thyroid-stimulating hormone (TSH) concentration of 1.8 mIU/L (1.8 μIU/mL) (reference range 0.35–5.5 mIU/L), and normal free triiodothyronine (FT3) concentration of 4.2 pmol/L (0.33 ng/dL) (reference range 3.5–6.5 pmol/L). Repeat investigation 1 and 2 months later revealed a progressive increase in FT4 to 125.3 pmol/L (9.7 ng/dL) and ≥155 pmol/L (≥12.0 ng/dL), respectively. TSH and FT3 remained within reference intervals, as did total T4 by RIA, at 155 nmol/L (12 μg/dL) (reference range 58–161 nmol/L). Furthermore, she tested positive for antithyroperoxidase antibodies at 110 IU/L (reference range <37 IU/L) and antithyroglobulin antibodies at 149 IU/L (reference range <98.1 IU/L). An investigation of her disconcordant thyroid function tests was initiated.
Increased serum FT4 concentrations, with a nonsuppressed TSH, are most often seen with erratic thyroxine replacement therapy. Other less common causes include antibody interference in the FT4 assay causing falsely increased FT4, thyroid hormone resistance syndromes, TSH-secreting pituitary adenoma, familial dysalbuminemic hyperthyroxinemia, amiodarone therapy, and primary hyperthyroidism with antibody interference in the TSH assay causing falsely increased TSH results. More recently, patients have been described with type 2 deiodinase deficiency due to defective selenoprotein synthesis; these patients presented with increased FT4 and normal FT3 and TSH (1).
In this case, antithyroxine autoantibody interference in the FT4 assay appeared likely, as the FT3 was within reference limits, in keeping with a clinically euthyroid state. The same patient sample yielded euthyroid FT4 when measured using 2 other immunoassay platforms, namely 14.5 pmol/L and 13.4 pmol/L on the AxSYM and Architect platforms (Abbott Diagnostics), respectively. This discrepancy could be explained by the fact that the Advia Centaur design uses a 1-step analog-based FT4 immunoassay. In this assay, the analog is a high-molecular-weight IgG-T4 complex labeled with acridinium ester (2). Serum and labeled FT4 analog are introduced into the reaction cuvette simultaneously, and sample FT4 and analog T4 compete for solid-phase antibody. Unbound material is then washed out, and the remaining bound analog T4 is measured by chemiluminescence. Increased luminescence therefore relates to decreased FT4 and vice versa. During this single incubation, anti-T4 antibodies in the sample may bind the analog T4, preventing its binding to the solid-phase antibody, leading to reduced luminescence signal and a falsely increased FT4 result. In contrast, both the Abbott AxSYM fluoroimmunoassay and the Abbott Architect chemiluminescence immunoassay for FT4 use a 2-step design in which labeled analog T4 is introduced only after unbound material from the sample has been removed in a wash step, thereby precluding interaction between sample antibodies and hormone analog. The fact that there was not marked interference in the total T4 assay in this case might be because FT4 assays sequester a much smaller fraction of the total T4 in serum to minimize changes in the binding equilibrium of T4. As such, it is likely that higher titers of antithyroxine autoantibodies would be required to cause interference in a total T4 assay than in a comparative 1-step analog FT4 assay.
Antibody interference in thyroid hormone immunoassays are variously ascribed to thyroid hormone autoantibodies (THAAbs), which bind endogenous thyroid hormones directly; heterophile antibodies, which bind to animal antibodies used in immunoassays; or rheumatoid factors, which also bind to animal antibodies used in immunoassays (3). The prevalence of THAAbs has been reported to be 1.8% in the general population and as high as 7% in autoimmune thyroid disease, although the frequency of significant assay interference by THAAbs is considerably less (4)(5). Furthermore, most patients with THAAbs also produce thyroglobulin autoantibody and thyroperoxidase autoantibody, reflecting an underlying thyroid autoimmune process (6). Based on the type of interference seen in this case, it was most likely that antithyroxine autoantibodies were responsible for the discrepant results obtained.
We performed the following studies to demonstrate antithyroxine autoantibody interference in the FT4 assay system.
peg precipitates radiolabeled t4
Aliquots of patient and control serum were incubated with radioactive 125I-labeled T4 solution (Coat-A-Count® Total T4 RIA; Siemens Medical Solutions Diagnostics) that is manufactured with a blocking agent to prevent thyroxine from interacting with thyroid hormone–binding proteins (7). Saline was substituted for serum to calculate nonspecific binding. All samples were mixed with equal volumes of 25% polyethylene glycol (PEG-6000), and total radioactivity was determined. After centrifugation to precipitate immunoglobulin, the supernatant was discarded and residual radioactivity in the immunoglobulin pellet was measured. Nonspecific binding was subtracted from all sample readings, and the final percentage of [125I]T4 binding was calculated. In this case, the patient sample demonstrated a 71% [125I]T4 binding capacity, compared with control samples at 7.6% and 8.7%. This simple assay therefore demonstrated that an immunoglobulin with specific T4-binding capacity was present in the serum.
peg precipitation eliminates interference
We analyzed 54 unaffected serum samples [FT4 range 10.1–61.9 pmol/L (0.78–4.80 ng/dL)] with appropriate TSH values. Each sample was split into 2 aliquots; the first was subjected to PEG precipitation as described, and the second was kept at room temperature. After precipitation, FT4 was measured in the undiluted serum and in the PEG-treated supernatant using the Advia Centaur (8). Linear regression yielded the regression equation: y = 0.614x − 0.46, where y = post-PEG FT4 and x = pre-PEG FT4 in pmol/L (y = 0.614x − 0.035 ng/dL). The distribution around the regression line demonstrated little random error, with Sy·x of 0.73 pmol/L (0.056 ng/dL), correlation coefficient (r) = 0.99. The relative distribution plot yielded a mean relative difference between pre- and post-PEG FT4 values of 41.4% (range 33.6%–54.8%).
We added patient antithyroxine autoantibody serum to 8 unaffected serum samples [FT4 range 10.7–28.7 pmol/L (0.83–2.22 ng/dL)] to yield false FT4 values of 23.3–52.2 pmol/L (1.8–4.05 ng/dL) and subjected them to the same procedure. This was done to ascertain whether PEG precipitation could differentiate interference within this range. All 8 samples exceeded 3SD predictive intervals on the linear regression, difference, and residual plots (Fig. 1⇓ ). Using the regression equation, we were also able to predict the patients’ true FT4 value based on the post-PEG FT4 value, and this correlated well with results obtained on other analyzers (Table 1⇓ ). In this case, it was possible to demonstrate thyroxine-autoantibody interference in the Advia Centaur FT4 immunoassay by PEG precipitation of serum and subsequent comparison of the difference between pre- and post-PEG FT4 results with those of controls.
The data suggest the presence of thyroid hormone autoantibodies. The patient was not treated for thyroidal illness and remained clinically euthyroid 6 months later.
POINTS TO REMEMBER
Investigation of immunoassay interference should include repeat analysis, determination of nonlinearity with dilution, testing by alternative method, and removal of interfering antibodies or blocking studies.
Changes in measured concentration after removal of immunoglobulins or addition of blocking agents are indicative of antibody interference, but no effect does not rule out interference.
Results obtained after dilution, immunoglobulin removal, or blocking studies should not be reported, as they may not reflect true concentrations.
Clear communications between clinicians and the laboratory are required to minimize the impact of assay interference on clinical decision making.
Grant Funding/Support: G. van der Watt was funded entirely by a grant from the National Health Laboratory Service of South Africa.
Financial Disclosures: None declared.
Acknowledgments: We thank Dr. Carel Meyer for practical assistance. Consultant experts: Graham Beastall, Glasgow Royal Infirmary, Glasgow, UK; Krishna Chatterjee, University of Cambridge, UK; James Faix, Stanford University, CA.
1 Predicted FT4 derived from post-PEG FT4 and the linear regression equation: y = 0.614x – 0.46 pmol/L.
↵1 Nonstandard abbreviations: FT4, free thyroxine; TSH, thyroid-stimulating hormone; FT3, free triiodothyronine; THAAb, thyroid hormone autoantibody.
- © 2008 The American Association for Clinical Chemistry