UNDERSTANDING THE COMPLEX RELATIONSHIP BETWEEN TSH AND FREE THYROXINE

Abstract

OBJECTIVES
Guidelines suggest screening for thyroid disease using TSH first before FT4 due to their inverse log-linear relationship. The signal favours TSH over FT4 in a ratio of >100:2. We reassessed this FT4 to TSH relationship following the third-generation TSH assay from Roche.

METHODOLOGY
Consecutive paired TSH/FT4 from ambulatory non- hospitalized subjects assayed on Cobas e801 (Roche) between October 2020 to July 2022 were retrieved from our Laboratory Information System. For duplicate/multiple values from the same individual, the earliest values were retained. Reference intervals for the Roche assays were TSH 0.4-4.0 mIU/L and FT4 10-20 pmol/L. Descriptive statistics, linear and nonparametric (kernel) regression and cross- validation of the results (n = 7484) were performed using the statistical software packages in R 4.2.0.

RESULTS
Linear regression of logTSH against FT4 gave an r = -0.26. Nonparametric regression revealed two negative sigmoid curves merging into the euthyroxinemic region. Three different segments were determined: hypothyroxinemic (FT4<10): logTSH = 1.75-0.15fT4 (n = 144; r = -0.34, P<0.0001),
euthyroxinemic (FT4 10-20): logTSH = 0.51-0.02fT4 (n = 6886, r = -0.10, p<0.0001), and hyperthyroxinemic (FT4>20): logTSH = 1.89-0.09fT4 (n = 454, r = -0.45, P<0.0001).

CONCLUSION
The relationship between FT4 and TSH is complex. The widely accepted view of a constant gradient of logTSH against FT4 over the whole range of thyroid test values is inaccurate. While the TSH-first strategy is applicable for thyroid dysfunction, two tests may be preferred for euthyroxinemic states in a hospital-based practice.