ESPE Abstracts (2015) 84 P-1-163

Paediatric Thyroid Nodule Score: Derivation and Validation of a Predictive Score for Thyroid Nodule Assessment in Children

Danielle van der Kaaya, Jonathan Wassermana, Nadia Akseera, Michael Roskiesb, Frederique la Liberteb, Lily Nguyenb & Pramod Puligandlab


aThe Hospital for Sick Children, Toronto, Canada; bThe Montreal Children’s Hospital, Montreal, Canada


Background: Differentiated thyroid carcinoma has an incidence of 15.2 per 100 000 in adolescents. The clinical challenge is identifying nodules requiring further intervention. Current modalities, in isolation, have poor ability to reliably differentiate benign from malignant nodules.

Objective and hypotheses: To derive and validate a predictive score that integrates clinical, laboratory, radiological and cytopathological parameters to define malignancy risk of a pediatric thyroid nodule.

Method: With research ethics approval, a retrospective analysis of patients <18 years with histopathologically confirmed diagnoses of papillary/follicular thyroid carcinoma (88 patients) and benign thyroid nodular disease (37 patients) between 1987 and 2014 was performed at tertiary care institution A. Relevant clinical, laboratory, ultrasound and cytopathological variables were retrieved from hospital records. Bivariate analyses identified variables (P<0.2) that will undergo logistic regression to derive a predictive score. Variance inflation factors will be consulted for multicolinearity. Construct and predictive validity and reliability will be independently confirmed in a second cohort from institution B.

Results: Patients with malignant and benign nodules had similar ages at presentation (12.8±2.8 vs 11.7±3.5 years) and female patients predominated (70%) in both groups. Variables identified to undergo logistic regression analysis for possible incorporation in the score: i) clinical: palpable thyroid nodule (P=0.09); palpable lymph nodes (P=0.02); ii) laboratory: TSH level (P<0.01); iii) ultrasound: nodule size (P=0.02), calcifications (P<0.01), focality (P=0.14), echogenicity (P=0.13), margins (P=0.12), >50% cystic (P<0.01), lymphadenopathy (P<0.01); iv) cytopathology (P<0.01).

Conclusion: We hope to derive a clinical tool to predict the likelihood of malignancy of a thyroid nodule based on integration of clinical, biochemical, radiological and cytopathological features. This tool has the potential to improve diagnostic accuracy.

Funding: This work is supported by a $3 500 microgrant from the Rare Disease Foundation and the BC Children’s Hospital Foundation (no grant number has been assigned).

Article tools

My recent searches

No recent searches.