Research Papers:

A novel integrative risk index of papillary thyroid cancer progression combining genomic alterations and clinical factors

Qing Cheng, Xuechan Li, Chaitanya R. Acharya, Terry Hyslop and Julie Ann Sosa _

PDF  |  HTML  |  Supplementary Files  |  How to cite

Oncotarget. 2017; 8:16690-16703. https://doi.org/10.18632/oncotarget.15128

Metrics: PDF 1642 views  |   HTML 2802 views  |   ?  


Qing Cheng1,4, Xuechan Li4, Chaitanya R. Acharya1, Terry Hyslop2,4, Julie Ann Sosa1,3,4

1Department of Surgery, Duke University Medical Center, Durham, NC 27710 USA

2Department of Biostatistics and Bioinformatics, Duke University Medical Center, Durham, NC 27710 USA

3Department of Medicine, Duke University Medical Center, Durham, NC 27710 USA

4Duke Cancer Institute, Duke University Medical Center, Durham, NC 27710 USA

Correspondence to:

Julie Ann Sosa, email: [email protected]

Qing Cheng, email: [email protected]

Keywords: papillary thyroid cancer, disease progression, genomic alterations, clinical and pathologic factors, risk index model

Received: September 02, 2016     Accepted: January 24, 2017     Published: February 06, 2017


Although the majority of papillary thyroid cancer (PTC) is indolent, a subset of PTC behaves aggressively despite the best available treatment. A major clinical challenge is to reliably distinguish early on between those patients who need aggressive treatment from those who do not. Using a large cohort of PTC samples obtained from The Cancer Genome Atlas (TCGA), we analyzed the association between disease progression and multiple forms of genomic data, such as transcriptome, somatic mutations, and somatic copy number alterations, and found that genes related to FOXM1 signaling pathway were significantly associated with PTC progression. Integrative genomic modeling was performed, controlling for demographic and clinical characteristics, which included patient age, gender, TNM stages, histological subtypes, and history of other malignancy, using a leave-one-out elastic net model and 10-fold cross validation. For each subject, the model from the remaining subjects was used to determine the risk index, defined as a linear combination of the clinical and genomic variables from the elastic net model, and the stability of the risk index distribution was assessed through 2,000 bootstrap resampling. We developed a novel approach to combine genomic alterations and patient-related clinical factors that delineates the subset of patients who have more aggressive disease from those whose tumors are indolent and likely will require less aggressive treatment and surveillance (p = 4.62 × 10–10, log-rank test). Our results suggest that risk index modeling that combines genomic alterations with current staging systems provides an opportunity for more effective anticipation of disease prognosis and therefore enhanced precision management of PTC.

Creative Commons License All site content, except where otherwise noted, is licensed under a Creative Commons Attribution 4.0 License.
PII: 15128