Predicting Breast Tumor Size for Pre-operative Planning: Which Imaging Modality is Best?

Abstract

We sought to compare breast tumor size predicted by imaging modality to the actual pathologic size in order to determine which imaging modality is best at estimating tumor size. We identified 261 patients with biopsy-proven invasive ductal (IDC) and/or invasive lobular (ILC) carcinomas with documented tumor dimensions predicted by imaging and maximum dimensions determined by final pathology. Results of imaging-predicted dimension were correlated with final pathological size. Spearman correlations were calculated and compared by Zou's method and agreement was assessed by McNemar's test. There was no significant difference (p > 0.05) between correlations of pathologic size by ultrasound (r = 0.71) and magnetic resonance imaging (MRI) (0.76). The correlations between MRI or ultrasound and pathologic size are significantly stronger than the correlations between mammography or clinical breast exam and pathologic size (p < 0.05). MRI and ultrasound are both strongly correlated with pathologic size overall and within grades in both IDC and ILC. The correlations between MRI or ultrasound and pathologic size are significantly better than the correlations between mammography or clinical breast exam and pathologic size (p < 0.05).