Sasaguri Kohei, Takahashi Naoki, Gomez-Cardona Daniel, Leng Shuai, Schmit Grant D, Carter Rickey E, Leibovich Bradley C, Kawashima Akira
1 Department of Radiology, Mayo Clinic, 200 First St SW, Rochester, MN 55905.
2 Present address: Department of Radiology, Faculty of Medicine, Saga University, Saga, Japan.
AJR Am J Roentgenol. 2015 Nov;205(5):999-1007. doi: 10.2214/AJR.14.13966.
The purpose of this study was to evaluate whether small (< 4 cm) oncocytomas can be differentiated from renal cell carcinomas (RCCs) on biphasic contrast-enhanced CT.
Forty-three patients with 53 oncocytomas and 123 patients with 128 RCCs (24 papillary subtype and 104 clear cell and other subtypes) who underwent biphasic contrast-enhanced CT were included in the study. Patient demographics and CT tumor characteristics were evaluated in each case. A multinomial logistic regression model was then constructed for differentiating oncocytoma from clear cell and other subtype RCCs, oncocytoma from papillary RCCs, and clear cell and other subtype RCCs from papillary RCCs. The probability of each group was calculated from the model. Diagnostic performance among three pairwise diagnoses and between oncocytoma and any RCC (clear cell and other subtypes and papillary) were assessed by AUC values.
Patient age, tumor CT attenuation values and skewness (i.e., histogram analysis of CT values) in both the corticomedullary and nephrographic phases, and subjective tumor heterogeneity were statistically significant variables in the multinomial logistic regression analysis. The logistic regression model using the variables yielded AUCs of 0.82, 0.95, 0.91, and 0.84 for differentiating oncocytomas from clear cell and other subtype RCCs, oncocytomas from papillary RCCs, clear cell and other subtype RCCs from papillary RCCs, and oncocytomas from any RCC (clear cell and other subtypes and papillary), respectively.
A combination of imaging features on biphasic CT, including tumor CT attenuation values and tumor texture (heterogeneity and skewness), can help differentiate oncocytoma from RCC.
本研究旨在评估在双期对比增强CT上，小（<4 cm）嗜酸细胞瘤能否与肾细胞癌（RCC）相鉴别。
本研究纳入了43例患有53个嗜酸细胞瘤的患者以及123例患有128个RCC的患者（24例乳头状亚型，104例透明细胞及其他亚型），这些患者均接受了双期对比增强CT检查。对每个病例的患者人口统计学和CT肿瘤特征进行了评估。然后构建了一个多项逻辑回归模型，用于区分嗜酸细胞瘤与透明细胞及其他亚型RCC、嗜酸细胞瘤与乳头状RCC，以及透明细胞及其他亚型RCC与乳头状RCC。根据该模型计算每组的概率。通过AUC值评估三种两两诊断之间以及嗜酸细胞瘤与任何RCC（透明细胞及其他亚型和乳头状）之间的诊断性能。
在多项逻辑回归分析中，患者年龄、肿瘤在皮质髓质期和肾实质期的CT衰减值及偏度（即CT值的直方图分析），以及主观肿瘤异质性是具有统计学意义的变量。使用这些变量的逻辑回归模型在区分嗜酸细胞瘤与透明细胞及其他亚型RCC、嗜酸细胞瘤与乳头状RCC、透明细胞及其他亚型RCC与乳头状RCC，以及嗜酸细胞瘤与任何RCC（透明细胞及其他亚型和乳头状）时，AUC分别为0.82、0.95、0.91和0.84。
双期CT上的成像特征组合，包括肿瘤CT衰减值和肿瘤质地（异质性和偏度），有助于鉴别嗜酸细胞瘤与RCC。
Zotero 插件