Mapping variants in thyroid hormone transporter MCT8 to disease severity by genomic, phenotypic, functional, structural and deep learning integration.

Summary

This integrated genomics-to-phenotype study maps MCT8 variant classes to survival and 24/32 clinical features, identifies a mild phenocopy in population cohorts, delineates seven functional domains, and delivers a pathogenicity–severity classifier with AUC 0.91/0.86 for 8,151 variants. Therapeutic effectiveness did not differ across loss-of-function categories, informing counseling and trial design.

Key Findings

• Genotype–phenotype relationships spanned survival and 24/32 disease features in MCT8 deficiency.
• Common MCT8 variants produced a mild phenocopy in ~400,000 population participants.
• Seven functional domains of MCT8 were delineated, advancing structural insights.
• An AI-based classifier predicted pathogenicity (AUC 0.91) and severity (AUC 0.86) for 8,151 variants.
• Therapeutic effectiveness did not differ across LoF categories.

Clinical Implications

Improves clinical interpretation of SLC16A2/MCT8 variants, enabling earlier diagnosis and standardized severity grading; supports use of the classifier in genetic pipelines and informs that treatment response may not depend on LoF class.

Limitations

• Rare disease context may limit immediate generalizability without broader ancestry validation
• Prospective clinical utility and health-economic impact of the classifier remain to be tested

Future Directions

Prospective integration of the classifier into diagnostic pipelines, validation across ancestries, and evaluation of treatment stratification and outcomes.