Influence of breast cancer histology on the relationship between ultrasound and pathology tumor size measurements.

Establishing an accurate primary invasive breast cancer size is crucial for patient management. Although ultrasonographic measurement is reported to correlate reliably with the gold standard pathology measurement, few authors have examined the influence of histologic subtype on ultrasound measurement. The common subtypes of invasive breast carcinoma, ductal and lobular, have different growth patterns, which may influence the ability of ultrasound to predict pathologic size. For this analysis, ultrasound and pathology reports were retrospectively reviewed for 204 women with 210 invasive breast cancers, including 129 ductal, 41 lobular, and 40 mixed pattern ductal and lobular carcinomas. For each tumor, the largest pathology and ultrasound dimensions were compared using Pearson's correlations, linear regression, paired t-tests and Wilcoxon signed ranks tests, stratified by histologic subtype. The Hodges-Lehmann approach was used to obtain 95% confidence intervals (CI) for median difference of the sizes. Ultrasonography consistently underestimated pathologic tumor size; the overall median difference was 3.5 mm (CI: 2.5-4.0 mm) and for subtypes: 2.5 mm (CI: 1.5-3.5 mm) for ductal pattern; 3.0 mm (CI: 1.5-4.5 mm) for mixed pattern; and in contrast, 7.5 mm (CI: 5.0-13.5 mm) for lobular pattern tumors. Significant correlations of similar magnitude, were observed between size measurements for ductal, lobular, and mixed subtypes (r=0.816, 0.811 and 0.672, respectively; all P<0.001); however, linear regression models differed between subtypes. Although practical and widely available, ultrasonography tends to underestimate pathologic tumor size. The size difference may be large for lobular carcinomas, potentially influencing stage; differences are less pronounced for ductal and mixed subtypes. Pathologic tumor size can be estimated from the ultrasonographic measurement, particularly if the histologic tumor subtype is known. The results of this study underscore the continued benefit of pretreatment tumor histology.