Abstract Background: Invasive lobular carcinoma (ILC), representing 15% of all invasive breast cancers, is characterized by loss or dysfunction of the adhesion molecule E-cadherin. This leads to an infiltrative growth pattern that is unlikely to disrupt the normal architecture of breast tissue. As a result, ILC is often non-mass forming and imaging is limited in correctly diagnosing stage I-III ILC. Mammography is believed to underestimate ILC, while MRI tends to overestimate the extent of ILC. The aim of this retrospective study was to correlate radiological with pathological findings. Methods: All patients diagnosed with stage I-III pure (i.e. not mixed) ILC between January 2000 and December 2020 who underwent primary surgery in University Hospitals Leuven, were included. Data on patient characteristics, preoperative imaging and pathology were collected from the patient files. Imaging and pathology measurements were compared by use of Pearson correlation coefficient, Whitney U test or Chi-square test. Weighted kappa statistics were used to assess agreement. Since the techniques of mammography and ultrasound have evolved in the past 20 years, different time periods were considered in sub-analyses (2000-2004, 2005-2009, 2010-2014, 2015-2020). Results: In total 1029 patients were included. Median age at diagnosis was 61.0 years (range 32.0 – 95.0 years). The breast density score determined on mammogram was Bi-RADS type A for 52 (5.1%) patients, B for 238 (23.1%) patients, C for 235 (22.8%) patients and D for 141 (13.7%) patients. Density score was missing from the reports of the remaining patients (n=363, 35.3%). Contrast enhanced breast magnetic resonance imaging (MRI) was performed in 709 (68.9%) patients. An increase in preoperative use of MRI was seen over the years. In comparison to tumor size reported by pathology, all imaging techniques underestimated the tumor size. The mean difference in largest tumor size compared to pathology was 14.64 ± 21.15 mm for mammography, 18.19 ± 21.66 mm for ultrasound and 9.93 ± 20.63 mm for MRI. A higher breast density level and a larger tumor size were significantly associated with a larger difference between diameter on pathology versus mammography (ρ=0.102, p-value 0.025 and ρ=0.530, p-value < 0.001 respectively). Changes over different time periods are shown in Table 1 for mammography and ultrasound. There was an agreement on unifocality versus multifocality between pathology and mammography for 81.2% of the patients. In 52.1% of the 236 cases where multifocality was reported on MRI, only 1 lesion was reported by the pathologist. Multifocality was seen on pathology in 12.0% of the 460 cases that were seen as unifocal lesions on MRI. Considering adenopathies, the false positive rate of ultrasound was 3.0%. However, the false negative rate was 68.3%. Conclusions: The local extent of ILC is underestimated by conventional imaging techniques. Unlike previous reports, our results suggest that tumor size of ILC is also underestimated by MRI. Ultrasound was inferior to mammography and MRI in estimating tumor size in our series. It was confirmed that MRI tends to overestimate the number of foci which might lead to unnecessary secondary ultrasounds and biopsies. The presurgical underestimation of lymph node involvement might increase the need of secondary surgeries. It is crucial to address these limitations in imaging of ILC and to prioritize the development of enhanced imaging techniques to improve diagnostic accuracy for these patients. Table 1: difference between imaging techniques and pathology by year of diagnosis Citation Format: Kaat Van Herck, Karen Van Baelen, Chantal Van Ongeval, Valerie Celis, Helen De Boodt, Machteld Keupers, Renate Prevos, Giuseppe Floris, Ines Nevelsteen, Ann Smeets, Thaïs Baert, Sileny Han, Hans Wildiers, Annouschka Laenen, Christine Desmedt, Patrick Neven. Early stage invasive lobular breast cancer is underestimated on conventional imaging including MRI [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-07-09.
Read full abstract