Thyroid nodules are a very frequent condition reaching up to 30-40% of the adult
population. Although most thyroid nodules have little clinical significance, in many
cases a fine needle aspirate (FNA) biopsy will be performed to determine its nature. In
70% of cases, a FNA will be reported as benign and in 10% of cases as cancer. However,
the remaining 20% of cases the thyroid nodule will be reported as indeterminate. The
latter patients have a risk of malignancy ranging from 15 to 25%, and in most cases the
patient will undergo thyroid lobectomy or total thyroidectomy to determine the final
pathology, resulting in an unacceptable number of unnecessary surgeries. This has a major
health impact, including surgical risks and permanent hormonal supplementation, as well
as unwarranted health costs estimated at 1.6 billion USD. Therefore, there is a need for
diagnostic tools in order to improve the diagnostic accuracy of the FNA and avoidance of
the high rate of unnecessary surgeries.
GeneproDx has developed a gene expression signature to improve the diagnostic accuracy of
FNA biopsy of thyroid nodules reported as indeterminate. The ThyroidPrint diagnostic
measures the expression of 10 genes in a FNA sample. It combines the results of the 10
biomarkers using a proprietary algorithm to predict benign thyroid nodules. This assay is
classified as multi-analyte algorithm assays (MAAA).
The biomarkers consist of multiplex TaqMan® gene expression assays run on Qiagen's
Rotor-Gene Q MDx RT-PCR IVD Platform instrument, which is a FDA cleared instrument. The
following 10 genes comprise the biomarker panel: CXCR3, CCR3, CXCL10, CK19, TIMP1, CLDN1,
CAR, XB130, HO-1 and CCR7. Each gene run in a multiplex configuration with two reference
genes. Each assay is performed with Research Use Only (RUO) kits and reagents on a FDA
cleared instrument.
ThyroidPrint has been developed using two different cohorts of samples, a training set
and a testing set. Using linear discriminant analysis, the training set identified the
final biomarker panel including; CXCR3, CCR3, CXCL10, CK19, TIMP1, CLDN1, CAR, XB130,
HO-1 and CCR7. In brief, the biomarkers have the following significance. CCR3 and CCR7
are chemokine receptors that are highly expressed in papillary thyroid cancer tumor
cells. CXCR3 is also a chemokine receptor and along with its receptor, CXCL10, are
detected in thyroid autoimmune disease. CAR is a G-couple receptor and has been shown to
be involved in cancer and has decreased expression in parathyroid adenoma. CK19 is a
keratin and has been used in thyroid tumors to recognize papillary carcinomas. CLDN1 is a
structural protein and has been shown to be differentially expressed in tumors compared
to normal tissue and has increased mRNA levels in papillary thyroid carcinoma. XB130 is
also a structural protein and its expression has been demonstrated in papillary thyroid
carcinoma. TIMP1 is a protease inhibitor with mRNA levels increased in advanced stages of
thyroid carcinoma. HO- 1 is an oxygenase and its expression has correlates with tumor
aggressiveness in thyroid cancer. In the final classifier, the expression of each gene
was ̈weighted ̈ based on its individual relative classifying ability. The cutoff score
was chosen in the ROC curve generated in the training set based on a minimum Sensitivity
of 92% to guarantee a high Negative Predictive Value (>95%). This cutoff score offered a
Specificity of 83%.An independent testing set of samples reproduced the diagnostic
performance observed in the training set and showed consistent results in FNA samples.
The assay has proven to accurately predict benign nodules in thyroid FNA samples with a
Negative Predictive Value of 96% and Specificity of 83% in both cohorts. The definitive
validation of an MAAA requires a final validation set, which analyzes samples that will
be used in the routine clinical setting; in this case indeterminate thyroid nodules
samples. In addition, in order to show clinical validity, the validation set must be
performed, as a statistically powered multi-institutional trial to assure that the data
is applicable to a broad population spectrum and has appropriate confidence intervals. A
first, statistically powered, multi-institutional trial is currently underway in Chile to
prove Clinical Validity. This trial includes 8 sites and will recruit approximately 3000
FNA to be completed by December 2017.
