The application of PET/CT with 18F-FDG in the differential diagnosis of lung solitary lesions

Abstract

Purpose of the study. The study was aimed to investigate the effectiveness of PET / CT with 18 fluoro‑2 deoxy-D-glucose (18F-FDG) in the differential diagnosis of focal lung neoplasms.Materials and methods. Patients (n = 108) with newly diagnosed lung focal lesions were enrolled in the study. All patients underwent PET / CT with 18fluoro‑2deoxy-D-glucose in the "whole body" mode in accordance with the standard protocol. The next step was surgical treatment with morphological verification. According to the results of the morphological conclusion, malignant neoplasm was not diagnosed in 11 (10.2 %) patients, while post-inflammatory changes were detected in 7 patients (SUVmax from 2.3 to 15.15), hamartoma was verified in 3 patients (SUVmax from 1.1 to 4.2) and anthracosis was detected in 1 patient.Results. The median radiopharmaceutical accumulation (SUVmax) was 6.0 (ICR 3.9–8.4, n = 108). In turn, in patients with diagnosed lung cancer this indicator was 7.0 (ICR 5.8–10.9, n = 60), in patients with metastatic lesions 4.3 (ICR 2.5–7.1, n = 37). The threshold for SUVmax was 5.4 for the detection of malignant tumors. PET/CT with 18F-FDG demonstrated high data variability regarding the size of lung focal lesion. An error within 35 % was observed in 76 % of cases, underestimating small lesions (up to 40 mm) but overestimating the major neoplasms. Application of a linear model for adjustment of neoplasm size assessment allows to estimate the actual size of neoplasms with parameters 5.862 + 0,817 × х (х – PET / CT size) in 84.5 % of cases with an error of 50 %. The optimal diagnosing size for metastatic lesions is in the range between 16.4 and 19 mm.Conclusion. Taken together the results of the study show that PET / CT with 18F-FDG gives a relatively accurate estimation of the tumor size. Application of the linear model corrects a radiological size measurements and helps to predict an actual size of a neoplasm in 84.5 % of cases with an error of 50 %. The prospective threshold for SUVmax was at least 5.4 for the detection of malignant neoplasms.