A novel gene panel for prediction of lymph-node metastasis and recurrence in patients with thyroid cancer

Surgery. 2020 Jan;167(1):73-79. doi: 10.1016/j.surg.2019.06.058. Epub 2019 Nov 9.


Background: Although well-differentiated papillary thyroid cancer may remain indolent, lymph node metastases and the recurrence rates are approximately 50% and 20%, respectively. No current biomarkers are able to predict metastatic lymphadenopathy and recurrence in early stage papillary thyroid cancer. Hence, identifying prognostic biomarkers predicting cervical lymph-node metastases would prove very helpful in determining treatment.

Methods: The database of the Cancer Genome Atlas included 495 papillary thyroid cancer samples. Using this database, we developed a machine learning model to define a gene signature that could predict lymph-node metastasis (N0 or N1). Kruskal-Wallis tests, univariate and multivariate logistic and Cox regression models, and Kaplan-Meier analyses were performed to correlate the gene signature with clinical outcomes.

Results: We identified a panel of 25 genes and constructed a risk score that can differentiate N0 and N1 papillary thyroid cancer samples (P < .001) with a sensitivity of 86%, a specificity of 62%, a positive predictive value of 93%, and a negative predictive value of 42%. This panel represents an independent biomarker to predict metastatic lymphadenopathy (OR = 8.06, P < .001) specifically in patients with T1 lesions (OR = 7.65, P = .002) and disease-free survival (HR = 2.64, P = .043).

Conclusion: This novel 25-gene panel may be used as a potential prognostic marker for accurately predicting lymph-node metastasis and disease-free survival in patients with early-stage papillary thyroid cancer.

MeSH terms

  • Adult
  • Biomarkers, Tumor / genetics*
  • Computational Biology
  • Disease-Free Survival
  • Feasibility Studies
  • Female
  • Humans
  • Lymphatic Metastasis / diagnosis*
  • Lymphatic Metastasis / genetics
  • Lymphatic Metastasis / prevention & control
  • Machine Learning
  • Male
  • Middle Aged
  • Models, Biological
  • Neoplasm Recurrence, Local / diagnosis*
  • Neoplasm Recurrence, Local / genetics
  • Neoplasm Recurrence, Local / prevention & control
  • Neoplasm Staging
  • Patient Selection
  • Predictive Value of Tests
  • Prognosis
  • RNA-Seq
  • ROC Curve
  • Thyroid Cancer, Papillary / diagnosis*
  • Thyroid Cancer, Papillary / genetics
  • Thyroid Cancer, Papillary / mortality
  • Thyroid Cancer, Papillary / surgery
  • Thyroid Neoplasms / genetics*
  • Thyroid Neoplasms / mortality
  • Thyroid Neoplasms / pathology


  • Biomarkers, Tumor