Differentiation of lung neoplasms with lepidic growth and good prognosis from those with poor prognosis using computer-aided 3D volumetric CT analysis and FDG-PET. (June 2014)

• Record Type: Journal Article
• Authors: Morimoto, Daisuke; Takashima, Shodayu; Sakashita, Naohiro; Sato, Yoshinobu; Jiang, Binghu; Hakucho, Tomoaki; Miyake, Chie; Takahashi, Yoshiyuki; Tomita, Yasuhiko; Nakanishi, Katsuyuki; Hosoki, Takuya; Higashiyama, Masahiko
• Is Part Of: Acta radiologica. Volume 55:Number 5(2014:Jun.)
• Page Start: 563
• Page End: 569
• Publication Date: 2014-06
• Subjects: Lung cancer -- adenocarcinoma -- computer-aided diagnosis -- PET/CT -- SUVmax; Radiology, Medical -- Periodicals; Radiography, Medical -- Periodicals; Radiotherapy -- Periodicals; 616.0757
• DOI: 10.1177/0284185113502336
• Languages: English
• ISSNs: 0284-1851

Abstract

Background: Many studies have reported that transverse computed tomography (CT) imaging findings correlate with prognosis of patients with small peripheral lung neoplasm with lepidic growth. However, no studies have examined this correlation with the aid of three-dimensional (3D) CT data. Purpose: To determine the most efficacious imaging factor for differentiation of lepidic growth type lung neoplasms with good prognosis from those with poor prognosis. Material and Methods: We evaluated CT findings, nodule patterns, SUVmax on FDG-PET/CT, as well as nodule volume and ratios of solid parts to nodule volume that were semi-automatically measured on CT images of 64 pulmonary nodules of ≤2 cm in 60 consecutive patients (24 men and 36 women; mean age, 65 years). For logistic modeling, we used all of the significant factors observed between the neoplasms with good and with poor prognosis as independent variables to estimate the statistically significant factors for discriminating invasive adenocarcinomas with lepidic growth (lesions with poor prognosis, n = 42) from the other neoplasms, including preinvasive lesions (lesions with good prognosis, n = 22), resulting in a recommendation for the optimal criterion for predicting lesions with poor prognosis. Results: The logistic regression model identified the ratio of the solid part to the whole volume of a pulmonary nodule as the only significant factor ( P = 0.04) for differentiating lepidic growth type lung neoplasms with good prognosis from those with poor prognosis. A ratio of 0.238 or more showed the highest discriminatory accuracy of 84% with 91% sensitivity and 76% specificity. Conclusion: Computer-aided analyses of pulmonary nodules proved most useful for establishing the optimal criterion for differentiation of lepidic growth type lung neoplasms with good prognosis from those with poor prognosis.