A Novel Approach to the Differentiation of Intrabulbar Tumors in Color Doppler Imaging.

AIM OF THE STUDY: The similarity of many benign intrabulbar lesions to a malignant tumor requires detailed differential diagnostics. However, none of the known methods can be used as the only one to determine the type of lesions. The aim of this study was to determine color Doppler imaging (CDI) markers characteristic of choroidal melanoma and metastatic intrabulbar tumors, increasing the diagnostic value and giving a new insight into the use of this method.

MATERIALS AND METHODS: CDI was performed in 44 patients with malignant tumors and in 49 patients with benign tumors. Patients with malignant tumors were divided into melanomas (n = 28) and metastatic tumors (n = 16). Univariate analysis with the logistic regression method and multivariate stepwise logistic regression analysis to create models testing tumor malignancy and differentiating melanoma from metastatic lesions were used. Model sensitivity and specificity were evaluated by the receiver-operating characteristic curve. A K-fold validation was performed.

RESULTS: Arterial blood flow, regular tumor surface, and tumor location in peripheral choroid were found significant for tumor malignancy. Mixed blood flow increased the accuracy of the test (p > 0.05). Model sensitivity and specificity were 83.7% and 75.7%. A regular tumor surface and hypoechoic or isoechoic tumor mass differentiated melanoma and metastatic masses in the regression model, with a sensitivity of 85.2% and a specificity of 75.0%. The area under curve (AUC) for both the models was 0.851 SE (standard error) 0.041 and 0.853 SE 0.063, respectively. AUC in five-fold cross-validation was 0.80 SE 0.0477 and 0.743 SE 0.094, respectively.

CONCLUSIONS: Arterial or mixed blood flow, regular tumor surface, and tumor location in peripheral choroid may be characteristic of malignant tumors. Regular tumor surface and echogenicity of tumor mass could differentiate melanoma from metastatic tumor.