Objectives: THRA mutations cause Resistance to Thyroid Hormone α (RTHα), an underdiagnosed disorder with hypothyroid features but near-normal thyroid function tests (TFTs). We developed a pathway, combining molecular analyses, new biomarkers and physiological measurements, to better diagnose and treat this disorder.
Methods: Structural and functional analyses of THRA variants, discovered by next generation sequencing in specific projects (eg 100K Genome, Deciphering Developmental Delay, Genetics of Obesity study) or unbiased investigation of patients, identified an RTHα cohort (n=32). In this cohort, we measured plasma metabolites or proteins and analysed facial images using artificial intelligence (AI) to differentiate RTHα from controls. We measured resting energy expenditure (REE) during thyroxine therapy of the disorder.
Results: 17 different, heterozygous THRA variants, in individuals investigated for diverse causes (growth retardation, developmental delay, autism, dysmorphic facies) localised to the hormone binding domain of TRα1, with 14 being homologous to THRB mutations causing RTHβ. Varying transcriptional impairment or morphological and skeletal abnormalities when variants were expressed in mammalian cells or developing zebrafish and reduced KLF9 expression in variant-containing, patients blood cells, led to their classification as pathogenic. 12 novel TRα mutations (R228C, R266L, D268N, Δ268-272, T275M, G278R, V282L, L287P, I299T, H381Q, P399S, L400Tfs*7) were identified. Mutations occurred de novo in 20/32 patients, including at a mutation hotspot (G291S) in five, unrelated, cases. With TFTs being near-normal (concentrations in reference range: TSH 100%; FT4 85%; RT3 70%; FT3 50%) in patients, omics technologies identified plasma metabolites or proteins, whose relative levels distinguish RTHα cases from controls with 95% accuracy. Validating this, plasma concentrations of the most important metabolites and protein differed significantly (RTHα vs Controls: Assymetric dimethylamine, (P=3.7E-05), pregnenolone sulphate (P=1.98E-07), or Factor XIII (P=1.1E-11). AI-guided scores of facial features in RTHα cases and controls differed significantly, generating a classifier with a receiver operating characteristic of 0.966. Thyroxine therapy, in TSH-suppressive dosage, raised REE from low (Z scores -3.58 to -0.02) to higher levels.
Conclusions: In silico analyses of THRA variants of unknown significance, identifies TRα mutations whose loss-of-function is confirmed using transcriptional, zebrafish model and patient cell-based assays. New biomarker and AI-guided dysmorphic feature analyses in individuals with mutant genotypes diagnoses RTHα, enabling thyroxine therapy to correct subnormal energy expenditure.
10 Sep 2022 - 13 Sep 2022