Daly Abigail E, Anderman Kyle J, Holt Logan R, Shern Tyler P, Bahl Manisha, Gadd Michelle A, Specht Michelle C, Verdial Francys C, Kwait Rebecca, Smith Barbara L, Ozmen Tolga
Breast Section, Division of Gastrointestinal and Oncologic Surgery, Massachusetts General Hospital, MGH Center for Breast Cancer, Boston, MA, USA.
Department of Radiology, Massachusetts General Hospital, Boston, MA, USA.
Ann Surg Oncol. 2025 Jun 16. doi: 10.1245/s10434-025-17663-5.
Accurate preoperative imaging of breast tumor size is essential, as small measurement differences can influence the treatment strategy. This study evaluates the accuracy of tumor size estimation by mammography, ultrasound, and magnetic resonance imaging (MRI) compared with pathology and examines factors influencing imaging performance.
We retrospectively analyzed patients with breast cancer treated from 2019 to 2024. Measurements were considered concordant if they fell within ±20% of the pathological size. Statistical analyses performed include paired t-tests, chi-squared tests, Lin's concordance correlation coefficient (CCC), and logistic regression. Python-based framework was used to identify the most accurate weighted formula.
We included 460 patients with a median age of 57 (26-90) years. MRI had the highest concordance (62%), outperforming mammography (57%) and ultrasound (53%) (p = 0.004). Our alternative weighted average formula (0.66 × MRI size + 0.35 × US size) yielded the highest concordance rate (65.2%, CCC = 0.785). Average tumor size on pathology was 17.31 mm. MRI slightly overestimated the size (18.38 mm, p = 0.482), while mammography (14.8 mm, p = 0.06) and ultrasound (14.31 mm, p = 0.019) underestimated. MRI demonstrated the highest accuracy in T-stage classification (89%). Concordance was highest for masses without non-mass enhancement (NME) (CCC = 0.834) and declined with NME (CCC = 0.635). MRI accuracy improved in tumors > 15 mm (OR 2.47) and high-grade tumors (OR 1.75) but declined in extremely dense breasts (OR 0.42) and lobular histology (OR 0.46).
MRI demonstrated the highest concordance with tumor size and T stage. Its accuracy improved in larger and high-grade tumors but decreased with dense breasts, NME, and lobular histology. A combined imaging approach using MRI and ultrasound may enhance preoperative size estimation.
准确的乳腺肿瘤术前大小成像至关重要，因为微小的测量差异可能会影响治疗策略。本研究评估了乳腺钼靶、超声和磁共振成像（MRI）与病理检查相比在肿瘤大小估计方面的准确性，并探讨了影响成像性能的因素。
我们回顾性分析了2019年至2024年接受治疗的乳腺癌患者。如果测量值在病理大小的±20%范围内，则认为测量结果一致。进行的统计分析包括配对t检验、卡方检验、林氏一致性相关系数（CCC）和逻辑回归。使用基于Python的框架来确定最准确的加权公式。
我们纳入了460例患者，中位年龄为57岁（26 - 90岁）。MRI的一致性最高（62%），优于乳腺钼靶（57%）和超声（53%）（p = 0.004）。我们的替代加权平均公式（0.66×MRI大小 + 0.35×超声大小）产生了最高的一致率（65.2%，CCC = 0.785）。病理检查中肿瘤的平均大小为17.31毫米。MRI略微高估了大小（18.38毫米，p = 0.482），而乳腺钼靶（14.8毫米，p = 0.06）和超声（14.31毫米，p = 0.019）则低估了。MRI在T分期分类中显示出最高的准确性（89%）。对于无非肿块强化（NME）的肿块，一致性最高（CCC = 0.834），随着NME的出现而下降（CCC = 0.635）。在肿瘤>15毫米（OR 2.47）和高级别肿瘤（OR 1.75）中，MRI的准确性提高，但在极度致密的乳腺（OR 0.42）和小叶组织学类型（OR 0.46）中准确性下降。
MRI在肿瘤大小和T分期方面显示出最高的一致性。其准确性在较大和高级别肿瘤中提高，但在致密乳腺、NME和小叶组织学类型中下降。使用MRI和超声的联合成像方法可能会提高术前大小估计的准确性。
