SAT456 Validation And Optimization Of TSH And FT4 Reference Interval Definitions During Pregnancy

Disclosure: J. Osinga: None. A. Derakhshan: None. W. Consortium on Thyroid and Pregnancy: None. T. Korevaar: None. Objective: While international guidelines recommend to diagnose gestational thyroid disease according to trimester specific TSH and FT4 reference intervals (RIs), most centers still rely on non-pregnancy RIs in combination with a fixed upper limit for TSH or by a deduction from this upper limit. Both strategies are based on expert opinion and have not been properly validated. We aimed to quantify the diagnostic accuracy of the alternative methods, and to identify the optimal diagnostic strategy to diagnose thyroid disease using a non-pregnancy TSH and FT4 RI. Methods: We performed an individual participant data meta-analysis of prospective birth cohorts. RIs and disease prevalences were calculated according to 1) trimester specific RIs, 2) non-pregnancy RIs, 3) non-pregnancy RIs with a fixed upper limit for TSH of 4.0mU/L and 4) non-pregnancy RIs with a 0.5mU/L subtraction from the TSH upper limit. Sensitivity and false discovery rates (FDR) of the alternative methods were calculated as compared to method 1), which is currently accepted as the best practice. We also varied non-pregnancy RIs with differing subtractions to identify the optimal diagnostic method. Participants with a treatment indication were defined as either overt hypothyroidism, subclinical hypothyroidism with TSH>10 or with concomitant TPOAb positivity. Results: The study population comprised 58,815 participants in 19 prospective cohorts. For first trimester overt hypothyroidism, sensitivity of all alternative methods (non-pregnancy RIs, fixed limit approach, subtraction approach) was similarly poor (0.43, 0.46 and 0.46) with a moderate FDR (0.22, 0.28 and 0.29). For first trimester subclinical hypothyroidism, results were slightly better for all methods, except for the FDR of the subtraction method; sensitivity (0.44, 0.54 and 0.55) and FDR (0.23, 0.23 and 0.35). Using the trimester specific approach, 2.01% of all women had a treatment indication in the first trimester. For this group, both the fixed limit and subtraction approach had a sensitivity of 0.65 and a FDR of 0.11 and 0.17 respectively. Furthermore, we leveraged our dataset to identify how non-pregnancy TSH and FT4 RIs can be used to approach pregnancy-specific RIs. We identified that for a treatment indication, subtraction of 20-25% of the non-pregnancy upper limit of TSH and subtracting 5-10% of the non-pregnancy lower limit of FT4 yielded a higher sensitivity (0.72-0.77) with a comparable FDR (0.20-0.25). Conclusion: Currently recommended alternative methods for defining RIs for TSH and FT4 in pregnancy have poor sensitivity when compared to the currently accepted trimester specific approach, and lead to substantial overdiagnosis. A relative subtraction from the non-pregnancy upper limit of TSH and lower limit of FT4 could considerably improve the diagnostic rates and would be directly available in daily practice. Presentation: Saturday, June 17, 2023