Hashimoto Yaichiro, Tsujikawa Tetsuya, Kondo Chisato, Maki Masako, Momose Mitsuru, Nagai Atsushi, Ohnuki Takamasa, Nishikawa Toshio, Kusakabe Kiyoko
Division of Nuclear Medicine, Department of Radiology, School of Medicine, Tokyo Women's Medical University, Tokyo, Japan.
J Nucl Med. 2006 Mar;47(3):426-31.
Benign and malignant pulmonary lesions usually are differentiated by 18F-FDG PET with a semiquantitative 18F-FDG standardized uptake value (SUV) of 2.5. However, the frequency of malignancies with an SUV of <2.5 is significant, and pulmonary nodules with low 18F-FDG uptake often present diagnostic challenges.
Among 360 consecutive patients who underwent 18F-FDG PET to evaluate pulmonary nodules found on CT, we retrospectively analyzed 43 who had solid pulmonary lesions (excluding lesions with ground-glass opacity, infiltration, or benign calcification) with an SUV of <2.5. The uptake of 18F-FDG was graded by a visual method (absent, faint, moderate, or intense) and 2 semiquantitative methods (SUV and contrast ratio [CR]). Final classification was based on histopathologic findings or at least 6 mo of clinical follow-up.
We found 16 malignant (diameter, 8-32 mm) and 27 benign (7-36 mm) lesions. When faint visual uptake was the cutoff for positive 18F-FDG PET results, the receiver-operating-characteristic (ROC) analysis correctly identified all 16 malignancies and yielded false-positive results for 10 of 27 benign lesions. Sensitivity was 100%, specificity was 63%, and the positive and negative predictive values were 62% and 100%, respectively. When an SUV of 1.59 was the cutoff for positive 18F-FDG PET results, the ROC analysis revealed 81% sensitivity, 85% specificity, and positive and negative predictive values of 77% and 89%, respectively. At a cutoff for positive 18F-FDG PET results of a CR of 0.29, the ROC analysis revealed 75% sensitivity, 82% specificity, and positive and negative predictive values of 71% and 85%, respectively. The areas under the curve in ROC analyses did not differ significantly among the 3 analyses (visual, 0.84; SUV, 0.81; and CR, 0.82). Analyses of intra- and interobserver variabilities indicated that visual and SUV analyses were quite reproducible, whereas CR analysis was poorly reproducible.
These results suggested that for solid pulmonary lesions with low 18F-FDG uptake, semiquantitative approaches do not improve the accuracy of 18F-FDG PET over that obtained with visual analysis. Pulmonary lesions with visually absent uptake indicate that the probability of malignancies is very low. In contrast, the probability of malignancy in any visually evident lesion is about 60%.
良性和恶性肺部病变通常通过18F-FDG PET以半定量的18F-FDG标准化摄取值（SUV）2.5进行区分。然而，SUV<2.5的恶性肿瘤发生率显著，且18F-FDG摄取低的肺结节常带来诊断挑战。
在360例连续接受18F-FDG PET以评估CT发现的肺结节的患者中，我们回顾性分析了43例SUV<2.5的实性肺病变患者（不包括磨玻璃影、浸润或良性钙化的病变）。18F-FDG摄取通过视觉方法（无、淡、中或浓）和2种半定量方法（SUV和对比率[CR]）进行分级。最终分类基于组织病理学结果或至少6个月的临床随访。
我们发现16例恶性病变（直径8 - 32 mm）和27例良性病变（7 - 36 mm）。当淡视觉摄取作为18F-FDG PET阳性结果的临界值时，受试者操作特征（ROC）分析正确识别了所有16例恶性肿瘤，27例良性病变中有10例假阳性结果。敏感性为100%，特异性为63%，阳性和阴性预测值分别为62%和100%。当SUV为1.59作为18F-FDG PET阳性结果的临界值时，ROC分析显示敏感性为81%，特异性为85%，阳性和阴性预测值分别为77%和89%。当CR为0.29作为18F-FDG PET阳性结果的临界值时，ROC分析显示敏感性为75%，特异性为82%，阳性和阴性预测值分别为71%和85%。3种分析（视觉、SUV和CR）的ROC分析曲线下面积差异无统计学意义（视觉，0.84；SUV，0.81；CR，0.82）。观察者内和观察者间变异性分析表明，视觉和SUV分析具有相当好的可重复性，而CR分析的可重复性较差。
这些结果表明，对于18F-FDG摄取低的实性肺病变，半定量方法并不比视觉分析提高18F-FDG PET的准确性。视觉上无摄取的肺部病变表明恶性肿瘤的可能性非常低。相反，任何视觉上明显的病变中恶性肿瘤的可能性约为60%。
