USING LYMPH NODE SIZE AND STANDARDIZED UPTAKE UNITS TO DETERMINE DIAGNOSTIC ACCURACY OF PET-CT IN DETECTING MALIGNANT MEDIASTINAL AND HILAR DISEASE IN A REGION ENDEMIC WITH HISTOPLASMOSIS

Yousef Ahmad; Miles Hagner; Arjan Flora

doi:10.1016/j.chest.2019.08.986

PURPOSE: Proper staging of the mediastinal and hilar lymph nodes is key to determining the best treatment options for patients with suspected lung cancer. Previously, two broad methods have been used in predicting malignant potential based on size and standardized uptake units (SUV) on positron emission tomography with computed tomography scans (PET-CT). Using a lymph node short axis size cut off greater than or equal to 10 mm alone may risk missing early or partially malignant nodes. While, an SUV of >2.5 has been generally accepted as being indicative of malignancy. However, this second predictive criterion is limited as lymphadenopathy secondary to inflammatory and infectious processes can also be PET-avid. Herein, we attempt to describe a method of analysis allowing greater distinction between inflammatory and malignant processes. METHODS: We retrospectively reviewed patients who had undergone endobronchial ultrasound with transbronchial needle aspiration (EBUS-TBNA), mediastinoscopy, or video assisted thoracoscopy in the heart of the Ohio River Valley Basin, a region endemic for Histoplasmosis. We identified 215 biopsied lymph node stations from 69 patients (average of 3.2 stations per patient, range 1 – 8 stations), with an average size of 14.88 mm and an SUV of 1.46 (Not avid to 14.4). 116 lymph node stations had both an SUV and their short axis measured. RESULTS: A PET avid parenchymal nodule was associated with 98.7% of the biopsied nodules. The diagnostic yield in which malignancy was found was as low as 29.3%. Using a short axis size of ≥10 mm alone, the sensitivity of lymph node size for malignancy is 91.43% (95% CI 76.94-98.2%), specificity 25.93% (95% CI 16.82-36.86%), positive predictive value 34.78% (95% CI 31.16-38.54%), negative predictive value 87.5% (95% CI 69.06-95.64%), with an accuracy of 45.69% (95% CI 36.41-55.19%). Using an SUV of ≥2.5 alone, the sensitivity for lymph node metastasis through PET avidity is 67.65% (95% CI 49.47-82.61%), specificity 79.52% (95% CI 69.24-87.59%), positive predictive value 57.50% (95% CI 45.48-68.69%), negative predictive value 85.71% (95% CI 78.48-90.80%), with an accuracy of 76.07% (95% CI 67.30-83.47%). Finally, using a combination of both a short axis ≥10 mm and an SUV ≥2.5, the sensitivity is 64.71% (95% CI 46.49-80.25%), specificity 84.15% (95% CI 74.42-91.28%), positive predictive value 62.86% (95% CI 49.23-74.71%), negative predictive value 85.19% (95% CI 78.32-90.15%), with an accuracy of 78.45% (95% CI 69.85-85.54%). Of the false positive lymph nodes, 53.85% had normal lymphoid tissue, reactive/inflammatory tissue, granulomatous disease and identifiable Histoplasmosis on GMS stain each made up 23.1% of the remainder. CONCLUSIONS: Diagnostic predictive value is increased when using the above methods. CLINICAL IMPLICATIONS: Our study addresses concerns regarding PET-CT specificity for diagnosis of inflammatory versus malignant lesions.

USING LYMPH NODE SIZE AND STANDARDIZED UPTAKE UNITS TO DETERMINE DIAGNOSTIC ACCURACY OF PET-CT IN DETECTING MALIGNANT MEDIASTINAL AND HILAR DISEASE IN A REGION ENDEMIC WITH HISTOPLASMOSIS

View Online

Abstract

Details

Metrics