The GS-TSH-R Mc4 cell line used in the Turbo TSI bioassay in the present study were obtained from QuidelOrtho Corporation, San Diego, CA. The GS-TSH-R Mc4 cells are Chinese hamster ovary (CHO) cells genetically engineered to express a chimeric form of the human TSH-R and a modified luciferase enzyme, Glosensor (GS), that is active only in the presence of cyclic AMP (15). Frozen cells are thawed immediately prior to use in the bioassay.

The Thyretain ^® TSI bioassay was performed according to the manufacturer’s product insert and as previously described (14, 16).

Briefly, TSI standards, controls, and test samples are added to a white 96-well white plate in 5 µL per well. A freezer vial containing Turbo TSI cells at a concentration of 4x10⁶ cells/mL is thawed in a 37°C water bath. These cells are then mixed with 5 mL of cAMP reagent with luciferase substrate (Promega, Madison, WI, USA) and placed in a reagent reservoir. The cell suspension is then dispensed into the wells of the plate at 50 µL/well. The plate is then incubated at ambient temperature for 60 minutes. At the end of the incubation, the plate is read using the GloMax Navigator luminometer (Promega, Madison, WI). The resulting relative light units (RLU) are then converted to TSI quantity in IU/L using the Thyretain ^® /Turbo TSI Data Analysis Tool.

The Turbo TSI assay quantifies results in IU/L using the Thyretain Turbo TSI Data Analysis Software (DAS) workbook. The assay’s Reportable Range is from 0.021 IU/L to 11 IU/L. Samples with results less than 0.021 IU/L will be reported as “< 0.021 IU/L”. Samples with results greater than 11 IU/L will be reported as “> 11 IU/L”. The dilution of samples with results greater than 11 IU/L is not supported.

Normal serum samples were obtained from Interstate Blood Bank, Inc. (Memphis, TN, USA) or from the serum bank at the Molecular Thyroid Research Laboratory, Johannes Gutenberg University (JGU) Medical Center, Mainz, Germany subsequent to written informed consent and approval of the Ethical Committee of the Rhineland-Palatinate Medical Association, number 837.214.10 (7223). All normal serum samples were tested for TSH, free T4, and free T3 and for autoantibodies to thyroperoxidase, thyroglobulin, and TSH-R. All normal serum samples were shown to be from euthyroid patients and were negative for autoantibodies to thyroid antigens. Serum samples from patients with autoimmune thyroid disease (AITD) and TSI-positive serum samples were obtained from the serum bank at the JGU Thyroid Lab.

The limits of blank (LoB), detection (LoD), and quantitation (LoQ) were determined according to the Clinical and Laboratory Standards Institute (CLSI)-approved guidelines (17, 18). LoB and LoD were calculated according to the following formulas: LoB = Mean _blank + 1.64 x SD _blank and LoD = LoB + 1.64 × mean SD _{low concentration sample}. The Turbo TSI LOD was verified using WHO IS spiked into normal serum at concentrations of 0.010, 0.012, 0.016 and to 0.018 IU/L. Twenty-six samples at each concentration were tested. The LoQ was determined using World Health Organization (WHO) International Standard (IS) spiked samples at concentrations between 0.02 and 0.03 IU/L. Five spiked samples were tested in 12 replicates each, across two lots of Turbo TSI bioassay over three days. The LoQ was determined from the lowest sample concentration meeting accuracy criteria (%TE ≤ 45%) where Total Error = Bias + 1.96 x Precision (%CV).

The Turbo TSI cut-off was determined using 198 serum samples that were classified as positive or negative for TSI by the Thyretain ^® TSI bioassay.

Twenty-one blood components, hormones and therapeutic drugs were tested at supraphysiologic concentrations. Serum obtained from patients with 12 disease states were tested (182 total samples).

The same 21 blood components, hormones and therapeutic drugs that were tested for cross reactivity were tested for interference of the Turbo TSI assay. Each substance was mixed with a TSI-positive serum (0.233 IU/L) sample and tested in the Turbo TSI assay. The percent difference between the measurement with and without the substance was determined. Fifty-eight serum samples from patients with various disease states were also tested for Interference in Turbo TSI assay by mixing them with a low TSI positive (0.178 IU/L) serum.

The Turbo TSI assay was run at 17 concentrations from 0.19 to 12.00 IU/L using 5 replicates per concentration.

Five concentrations of positive TSI samples from 0.047 to 9.807 IU/L were tested by 2 operators using two replicates over 20 days (n=80).

A panel of TSI-positive serum samples were tested in the Turbo TSI bioassay by two operators at three locations over 5 days.

TSI-positive samples at five concentrations were stored at the indicated temperatures for the indicated periods of time days before testing. Three different concentrations of TSI serum samples were tested by three lots of Turbo TSI kit once a month up to 16 months. The stability of the TSI kit was evaluated using the recommended approach in the CLSI-EP25 guidelines. A scatter plot containing a linear regression line was generated to visualize the calculated TSI concentration (IU/L) for each sample (y-axis) plotted against time over 16 months (x-axis).

Two-hundred and ninety-five serum samples from patients with AITD were tested with the Thyretain ^® TSI Bioassay and Turbo TSI bioassay at two sites.

The LOB was found to be 0.007 IU/L. The LOD was determined to be 0.016 IU/L ( Table 1 ). The LoQ was determined to be 0.021 IU/L ( Supplementary Table 1 ). The Turbo TSI bioassay cutoff was determined to be 0.0241 IU/L using a ROC curve analysis, which yielded the highest true positive rate (98.7% sensitivity) together with the lowest false positive rate (93.5% specificity) against results from the Thyretain ^® TSI bioassay ( Figure 1 ).

The TSI concentrations measured for each of the tested blood components, hormones and drugs were all below the Turbo TSI assay cutoff (0.0241 IU/L), indicating there was no cross-reactivity ( Table 2 ). Of the 182 serum samples from patients with various disease states and autoantibodies all but seven were below the Turbo TSI cut-off of 0.0241 IU/L (1/22 with type 1 diabetes; 2/50 celiac disease, and 4/21 thyroid cancer) ( Supplementary Table 2 ). Six of these seven positive samples were positive in the Thyretain ^® TSI bioassay (data not shown). The same blood components, hormones and drugs were tested in interference studies and the percent change for each tested substance was ≤15% ( Supplementary Table 3 ). None of the sera from patients with various disease states or autoantibodies interfered with the Turbo TSI assay ( Supplementary Table 4 ).

The Turbo TSI assay demonstrated linearity from 0.015 to 11.958 IU/L, as all samples showed ±10% deviation from the expected linear relationship ( Supplementary Table 5 ). The overall within-laboratory precision was ≤15%CV for all samples ( Table 3 ).

TSI serum samples stored at different temperatures (15°C, 23°C, and 30°C) are stable for up to 4 days ( Supplementary Table 6 ). TSI serum samples were stable at -70°C for up to at least 14 months. The samples also can be stored at -20°C for 3 months, or at 4°C for 3 weeks ( Supplementary Table 7 ).

Overall quantification (IU/L) for all serum samples measured over 16 months using three different kit lots showed a %CV ≤14% and, based on a regression analysis, no significant measurement drift was observed under the storage conditions for all Turbo TSI kits ( Supplementary Table 8 ).

The overall reproducibility of Turbo TSI assay was %CV ≤20 for five concentrations (0.06 to 5.16 IU/L) ( Supplementary Table 9 ).

A sample panel containing five clinical TSI positive specimens were tested at different temperatures. The Turbo TSI assay performed in the temperature range between 18°C and 26°C yielded consistent results with %CV ≤20% for all controls and samples ( Supplementary Table 10 ).

Turbo TSI bioassay demonstrated 95.2% positive percent agreement and 94.8% negative percent agreement when compared to the Thyretain ^® TSI Reporter bioassay using 295 serum samples ( Table 4 ). All samples were tested with a TRAb immunoassay (Kronus, Inc., Star, ID, USA) and there was 89% concordance between the TRAb assay and the two bioassays (data not shown).