Factors Associated With Preoperative Radiological Tumor Size Underestimation in Clinical T1–2 Breast Cancer Patients

Authors:

Abstract
Background: Accurate preoperative tumor size measurement is essential for determining optimal surgical margins in breast cancer patients. Thus, this study aimed to evaluate the factors associated with preoperative radiological tumor underestimation in clinical T (cT) Stage 1–2 breast cancer patients. Methods: We retrospectively reviewed the data of 365 cT1–2 breast cancer patients. Radiological tumor size was defined as the larger dimension on ultrasonography or magnetic resonance imaging. A pathological‐to‐radiological tumor size ratio > 1.2 or a tumor size discrepancy (pathology minus radiology) ≥ 5 mm was considered indicative of radiological underestimation. Preoperative variables, including age, body mass index (BMI), cT stage, maximum standardized uptake value of tumor, molecular subtype, and histologic subtype and grade, were analyzed to identify associated factors. Tumor size discrepancy (pathology minus radiology) was compared across subgroups, and the difference between pathological and radiological tumor size measurements was evaluated in each subgroup. Results: A BMI ≥ 25 kg/m 2 ( p = 0.006) and invasive lobular carcinoma (ILC) histology ( p = 0.030) were associated with a pathological‐to‐radiological tumor size ratio > 1.2. A BMI ≥ 25 kg/m 2 ( p = 0.015), ILC histology ( p = 0.022), and cT stage ( p = 0.027) were associated with a tumor size discrepancy (pathology minus radiology) ≥ 5 mm. Significant differences in tumor size discrepancy between patients with a BMI ≥ 25 kg/m 2 and those with a BMI < 25 kg/m 2 ( p = 0.003) and between patients with an ILC and those with non‐ILC ( p = 0.041) were observed. Bland–Altman analysis showed radiological underestimation in ILCs (4.3 mm). Conclusion: In cT1–2 breast cancer patients, a BMI ≥ 25 kg/m 2 and/or an ILC are predictive of radiological tumor underestimation, warranting supplemental imaging and intraoperative margin assessment.

This publication has 44 references indexed in Scilit: