Aplicación de la técnica de árboles de clasificación y regresión en la valoración ecográfica de los nódulos tiroideos / Using Classification and Regression Trees in the ultrasound evaluation of thyroid nodules

Objetivo: Evaluar la capacidad de los árboles de clasificación y regresión (CART) en la valoración ecográfica de los nódulos tiroideos malignos. Materiales y métodos: Se realizaron 404 punciones aspiración con aguja fina (PAAF) a 384 pacientes. De los nódulos analizados, se registraron las características ecográficas (localización, tamano, morfología, contorno, consistencia, ecoestructura, ecogenicidad, calcificaciones y vascularización) y los resultados de la elastografía. Se aplicó el CART para investigar la relación entre los hallazgos ecográficos, la elastografía y el cáncer de tiroides. Resultados: El análisis CART determinó que la realización de una elastografía no aporta datos relevantes y que las zonas homogéneas pueden clasificar a los nódulos tiroideos en: 1.ª zona) caracterizada por ausencia de áreas de degeneración coloide e hipoecogenicidad asociada a malignidad; 2.ª zona) diferenciada por presencia de áreas de degeneración coloide combinada con ausencia de microcalcificaciones, constituyendo un indicador fiable de benignidad de los nódulos tiroideos; y 3.ª zona) la ausencia de hipoecogenicidad y una lesión menos alta que ancha deparan un indicador fiable de benignidad. El árbol de clasificación alcanzó una sensibilidad del 87,5% y un valor predictivo negativo del 98,8%. Discusión: El CART muestra una elevada capacidad de predicción de los nódulos malignos frente a otras técnicas lineales. Conclusión: La utilización de los árboles de clasificación provee una herramienta simple para la toma de decisión clínica con el objeto de reducir las PAAF innecesarias, con una elevada sensibilidad.

Objective: To evaluate the use of Classification and Regression Trees (CART) in the ultrasound evaluation of malignant thyroid nodules. Materials and methods: A study was performed on 404 fine needle aspirates (FNA), with biopsies being performed on 384. The information collected about the thyroid nodules was: ultrasound features (location, size, morphology, contour, consistency, echo-structure, echogenicity, calcification, and vascularisation) and elastography results. The CART technique was used to investigate the relationship between ultrasound findings and the thyroid cancer. Results: The CART analysis showed that elastography does not provide any relevant data, and that the homogeneous areas could classified the thyroid nodules into: 1st area) characterised by the absence of colloid degeneration areas and a hypo-echogenicity associated with malignancy; 2nd area) differentiated by the presence of colloid degeneration areas combined with absence of microcalcifications, constituting a reliable indicator of benign thyroid nodules; and a 3rd area) the absence of hypo-echogenicity and a lesion wider than it is long that provides a reliable indicator of being benign. The optimum tree produced a sensitivity of 87.5% and negative predictive value of 98.8%. Discussion: The CART technique demonstrated a high predictive capacity for malignant nodules compared to other linear techniques. Conclusion: The use of classification trees provides us with a simple tool for clinical decision making, in order to reduce unnecessary FNA biopsies, as well as achieving a high sensitivity.

Utility of 18F-FDG-PET/CT in patients suspected of paraneoplastic neurological syndrome: importance of risk classification.

PURPOSE: To assess the diagnostic impact of 18F-FDG-PET/CT in patients suspected of paraneoplastic neurological syndrome (PNS) based on our own pre-test risk classification (PRC). METHODS: A multicenter retrospective longitudinal study was conducted from 2006 to 2014. We designed a seven-point scoring system using the clinical syndrome characteristics [classical (CS) and non-classical syndromes (NCS)] and its location (central, peripheral, in the neuromuscular junction or combined), onconeural antibodies and tumor markers. Patients were classified as low (score 0-2), intermediate (3-4) and high (5-7) pre-test risk of PNS. FDG-PET/CT was classified as negative or positive. Final diagnosis according Graus' criteria (definite, possible or no PNS) was established. Relations between clinical and metabolic variables with the final diagnosis were studied. RESULTS: 73 patients were included, with a follow-up time of 33 months. Eleven (15 %) patients were finally diagnosed with neoplasm (8 invasive cancers). Ultimately, 13 (18 %) and 24 (33 %) subjects were diagnosed as definite or possible PNS. All the patients with final diagnosis of neoplasm had a CS (p = 0.005). PET/CT was helpful to diagnose 6/8 (75 %) invasive cancers. PET/CT findings were associated with the final diagnosis of neoplasm (p = 0.003) and the diagnosis of PNS attending to Graus' criteria (p = 0.019). PRC showed significant association with the final diagnosis of neoplasm and PET/CT results. A majority of patients (10/11) diagnosed of neoplasm had intermediate/high-risk. CONCLUSIONS: Our PRC seems to be a valid tool to select candidates for PET/CT imaging in this setting. PET/CT detected malignancy in a significant proportion of patients with invasive cancer.