Development of Novel mRNA Classifiers to Stratify Preoperative Thyroid Tumor Risk.

OBJECTIVE

Develop a preoperative risk classifier for guiding the extent of surgery in differentiated thyroid carcinomas (DTCs) under 4 cm.

STUDY

DESIGN

In silico study, validation with retrospective cohort.

SETTING

Genomic data repositories, academic tertiary medical center.

METHODS

Prognostic transcriptional groups and American Thyroid Association (ATA) risk groups in The Cancer Genome Atlas (TCGA) were used to create novel Clinico-TRanscriptomic (CTR) groups. Differential gene expression (DGE) was performed. Consensus clustering was used to assess the reproducibility of genomic and histologic features in: (1) Bethesda V/VI fine-needle aspirates (FNA) from the Afirma Genomic Sequencing Classifier (GSC) database (GC cohort); (2) thyroid carcinomas annotated by ATA histopathologic risk group (surgical cohort); (3) independent surgical cohort.

RESULTS

Within TCGA (n = 451), CTR-Low was enriched with BRAFV600E mutations. CTR-Intermediate was enriched in follicular cancers, while CTR-High had high rates of TERT promoter (TERTp) and BRAFV600E mutations. A 144 gene classifier (PreOperative INdicator of Thyroid cancer progression [POINT]) was developed based on inter-CTR DGEs. Clustering of the GC cohort using POINT demonstrated similar genomic features with 3 groups characterized by BRAFV600E and ALK/NTRK/RET fusions, RAS mutations, or TERTp/BRAFV600E alterations. Agreement of POINT with ATA risk group in the surgical cohort (n = 129) showed fair agreement (weighted κ = 0.25), NPV 0.89 for ruling out high ATA risk, and NNT 5.3 to avoid 1 completion thyroidectomy.

CONCLUSION

The POINT classifier is predictive of genomic alterations and ATA risk in independent data sets. This study advances precision oncology, shifting the paradigm from size-based toward individualized, biology-driven care.