Japanese Journal of Clinical Oncology, 2015, 45(5) 411 415 doi: 10.1093/jjco/hyv018 Advance Access Publication Date: 10 February 2015 Original Article Original Article Prognostic importance of ultrasound BI-RADS classification in breast cancer patients Ju-Yeon Kim, Eun Jung Jung*, Taejin Park, Sang-Ho Jeong, Chi-Young Jeong, Young-Tae Ju, Young-Joon Lee, Soon-Chan Hong, Woo-Song Ha, and Sang-Kyung Choi Department of Surgery, School of Medicine, Gyeongsang National University, Jin-ju, South Korea *For reprints and all correspondence: Eun Jung Jung, Department of surgery, School of Medicine, Gyeongsang National University, Jin-ju, South Korea. E-mail:drjej@gnu.ac.kr Received 10 December 2014; Accepted 16 January 2015 Abstract Objective: We investigated the prognostic importance of pre-operative Breast Imaging Reporting and Data System classification in ultrasound imaging. Methods: Histopathological differences and disease-free survival were analyzed in Breast Imaging Reporting and Data System classification subgroups. Univariate and multivariate analyses were used to identify the prognostic factors. Results: We identified 531 invasive breast cancer patients eligible for this study. Most patients classified as Breast Imaging Reporting and Data System 5 had large tumors and a higher rate of lymph node metastasis. However, hormonal receptor or HER-2 status did not differ according to Breast Imaging Reporting and Data System classification. During a median post-operative follow-up of 42.0 months, 43 patients were diagnosed with a disease-specific event. Disease-free survival was significantly lower in patients with Breast Imaging Reporting and Data System 5 than in patients with Breast Imaging Reporting and Data System 3 4. Subgroup analysis of patients with invasive breast cancer of Stage I showed that Breast Imaging Reporting and Data System 5 was an independent negative prognostic indicator of disease-free survival (hazard ratio 9.195; 95% confidence interval, 1.175 71.955; P = 0.035). Conclusions: Breast Imaging Reporting and Data System classification might be considered as prognostic factors especially in Stage I breast cancer. Further confirmatory studies are needed. Key words: breast cancer, BI-RADS, ultrasound Introduction The Breast Imaging Reporting and Data System (BI-RADS) introduced by the American College of Radiology has standardized the reporting of mammography findings, clarified their interpretation and facilitated communication among clinicians (1). Although mammography is recognized as the optimal screening method, ultrasound has been proposed as a supplemental screening test in women with dense breast tissue and in high risk women who cannot undergo magnetic resonance imaging for any reason (2,3). Since the establishing of BI-RADS for sonography in 2003, the sonographic BI-RADS lexicon or classification has shown reliability and good performance in evaluating the likelihood of malignancy (4 6). Ultrasound imaging has therefore played a key role in the clinical evaluation of breast lesions, especially in Korean women with dense breast tissue. Breast cancer is recognized as a heterogeneous disease. Although, the predictive and prognostic abilities of many new techniques, such as gene profiling assays, have been studied, these assays are unlikely to become widely used in daily practice due to their costs and lack of insurance coverage. Ultrasound associated features of breast masses, including their shape, orientation, margins, posterior acoustic features The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com 411
412 Prognostic importance of BI-RADS classification and vascularity, may reflect tumor-specific characteristics. We therefore investigate the prognostic significance of BI-RADS classifications of pre-operative ultrasound imaging in breast cancer patients. Patients and methods The medial records of all patients who underwent curative surgery for newly diagnosed invasive breast cancer at a single institution between January 2005 and September 2013 were retrospectively evaluated. Patients were included if pre-operative ultrasound images were available and if they had been classified according to BI-RADS. Patients who received neoadjuvant chemotherapy, those with synchronous bilateral breast cancer or those with metastasis at the time of diagnosis were excluded. We identified a total of 531 patients with invasive breast cancer eligible for this study. Following surgery, patients were recommended to receive adjuvant therapy and periodic surveillance, according to the St. Gallen and/or NCCN guidelines. The prognostic importance of BI-RADS classification was assessed in patients who underwent surgery between January 2005 and December 2010 and were followed-up for >6 months postoperatively. Disease-specific events included local recurrence, axillary lymph node recurrence and distant metastasis. Disease-free survival (DFS) was defined as the time period from the date of breast surgery to that of first diagnosis with a disease-specific event or last follow-up. Continuous variables are expressed as means ± standard deviation (SD). χ 2 Test and Student s t-tests were used to compare the variables of the BI-RADS subgroup. DFS was assessed by the Kaplan Meier method, with curves compared using the log rank test. A Cox regression model was utilized to identify significant independent factors related to diseasespecific events. Significance was defined as P < 0.05. All statistical analyses were performed using SPSS (version 19.0 SPSS Inc., Chicago, IL, USA). The study protocol including the use of the database was approved by the institutional review board and met the guideline of the responsible governmental agencies. Results Of the 531 patients eligible for this study, 466 (87.8%) had invasive ductal carcinomas, 18 (3.4%) had invasive lobular carcinomas and 12 (2.3%) had medullary carcinomas. The median tumor size was 2.00 cm (mean ± SD = 2.04 ± 1.44 cm, range 0.1 11 cm). Axillary lymph node (LN) metastasis was identified in 186 (35.0%) patients. On pre-operative ultrasound imaging, eight patients were seen with only duct irregularity and two patients were seen with only parenchymal abnormality without mass lesion. Three hundred fourteen patients were presented with palpable mass lesion on physical examination. Two patients (0.4%) were classified as BI-RADS 3, 193 (36.3%) as BI-RADS 4 and 336 (63.3%) as BI-RADS 5. The clinical and histopathological characteristics of these patients are shown in Table 1. Most patients classified as BI-RADS 5 had large tumors and the rate of lymph node metastasis was higher in this group than in patients classified as BI-RADS 3 and 4. However, hormone and HER-2 receptor status did not differ by BI-RADS classifications. We analyzed the prognostic effect of BI-RADS classification on 524 patients. During a median post-operative follow-up of 42.0 months (range, 6 104 months), 43 patients were diagnosed with a diseasespecific event, including 8 with locoregional recurrence, 2 with contralateral recurrence and 33 with distant metastases. Rates of diseasespecific events were 4.7% (9/192) in the BI-RADS 3 4 group and 10.2% (34/332) in the BI-RADS 5 (P = 0.031). Kaplan Meier survival Table 1. The clinical and histopathological characteristics of the included patients. analysis showed that DFS was significantly poorer in patients with BI-RADS 5 than in patients with BI-RADS 3 4 (P = 0.036) (Fig. 1A). Cox proportional hazard analysis showed that BI-RADS classification was a significant prognostic factor on univariate analysis (hazard ratio [HR] 2.15; 95% confidence interval [CI] 1.033 4.491), but not on multivariate analysis. The prognostic factors of DFS on multivariate analysis included age at diagnosis, estrogen receptor (ER) negativity, HER-2 positivity, lymphatic invasion and LN metastasis (Table 2). We also analyzed the prognostic effect of BI-RADS classification on 244 patients with Stage I, early breast cancer. During follow-up, 11 of the 262 patients (4.5%) experienced a disease-specific event. Rates of disease-specific events were 1.6% (2/128) in the BI-RADS 3 4 group and 7.6% (9/116) in the BI-RADS 5 (P = 0.028). Kaplan Meier analysis of these patients showed that DFS was significantly poorer in patients classified as BI-RADS 5 than in patients classified as BI-RADS 3 4 (P = 0.034)(Fig. 1B). However, BI-RADS classification was not associated with DFS in Stages II and III) (Fig. 1C and D). A Cox proportional hazard model showed that BI-RADS 5 was an independent negative prognostic indicator of DFS (HR 9.195; 95% CI, 1.175 71.955; P= 0.035) (Table 3). However, BI-RADS classification was not associated with DFS in Stages II and III. Discussion BI-RADS 3 4 (195 cases) BI-RADS 5 (336 cases) P-value Age (years) 49.35 ± 11.09 51.67 ± 11.17 0.021 Tumor size (cm) 1.62 ± 1.21 2.31 ± 1.50 <0.001 T stage <0.001 T1 151 (77.4%) 176 (52.4%) T2 4 44 (22.6%) 160 (47.6%) N stage <0.001 N0 156 (80.0%) 189 (56.2%) N1 26 (13.3%) 85 (25.3%) N2 5 (2.6%) 33 (9.8%) N3 8 (4.1%) 29 (8.6%) Multiplicity 22 (11.3%) 55 (16.4%) 0.126 Lymphatic invasion 2 (1.0%) 14 (4.1%) 0.062 Histologic grade 0.121 Grades 1 and 2 110 (56.4%) 172 (51.2%) Grade 3 62 (31.8%) 133 (39.6%) Unknown 23 (11.8%) 31 (9.2%) Estrogen receptor 0.772 Negative 61 (31.3%) 112 (33.3%) Positive 130 (66.7%) 222 (66.1%) Unknown 4 (2.0%) 2 (0.6%) Progesterone receptor 0.710 Negative 71 (36.4%) 130 (38.7%) Positive 120 (61.5%) 204 (60.7%) Unknown 4 (2.1%) 2 (0.6%) HER-2 receptor 0.272 Negative 147 (75.4%) 240 (71.4%) Positive 37 (19.0%) 78 (23.2%) Unknown 11 (5.6%) 18 (5.4%) BI-RADS, Breast Imaging Reporting and Data System. We have shown here that a BI-RADS 5 classification was associated with poorer DFS in breast cancer patients, especially in early breast
Jpn J Clin Oncol, 2015, Vol. 45, No. 5 413 Figure 1. The Kaplan Meier survival curves for disease-free survival according to BI-RADS classification. The disease-free survival was evaluated in the 524 patients (A), and according to TNM stage (B) Stage I, (C) Stage II and (D) Stage III. Table 2. Univariate and multivariate analyses of prognostic factors including BI-RADS classification for disease-free survival. Univariate analysis Multivariate analysis HR (95%CI) P-value HR (95%CI) P-value Age (Cont.) 1.042 (1.012 1.073) 0.006 1.045 (1.010 1.080) 0.010 Tumor size > 2 cm 1.701 (0.874 3.309) 0.118 Node metastasis 2.299 (1.181 4.472) 0.014 2.120 (0.985 4.562) 0.055 ER negative 2.376 (1.221 4.626) 0.011 1.348 (0.617 2.945) 0.454 PR negative 1.597 (0.819 3.114) 0.170 HER-2 positive 3.060 (1.544 6.065) 0.001 2.416 (1.149 5.078) 0.020 Lymphatic invasion 7.327 (2.575 20.848) <0.001 5.702 (1.745 18.637) 0.004 Histologic Grade 3 2.981 (1.445 6.148) 0.003 2.303 (1.029 5.153) 0.042 BI-RADS 5 2.354(1.028 5.689) 0.043 1.201 (0.492 2.931) 0.687 ER, estrogen receptor; PR, progesterone receptor; HR, hazard ratio; CI, confidence interval.
414 Prognostic importance of BI-RADS classification Table 3. Univariate and multivariate analyses of prognostic factors for disease-free survival in T1N0M0 breast cancer patients. Univariate analysis Multivariate analysis HR (95%CI) P-value HR (95%CI) P-value Age (Cont.) 1.006 (0.952 1.062) 0.838 Tumor size (Cont.) 1.454 (0.562 3.761) 0.441 ER negative 3.805 (1.114 12.659) 0.029 1.475 (0.414 5.258) 0.549 PR negative 1.723 (0.548 5.415) 0.351 HER-2 positive 0.206 (0.065 0.650) 0.007 4.474 (1.265 15.819) 0.020 Histologic Grade 3 3.830 (1.120 13.094) 0.032 2.469 (0.708 8.609) 0.156 BI-RADS 5 11.289 (1.457 87.462) 0.020 7.979 (1.008 63.187) 0.049 cancer patients. Similarly, BI-RADS classification on mammograms has been reported to be significantly associated with DFS in Taiwanese breast cancer patients, with significantly poorer 5-year overall survival and DFS rates in patients with BI-RADS 5 than BI-RADS 0 4 (7). Until recently, many studies of breast imaging focused on the ability to predict malignant lesions or to find additional lesions in patients scheduled for surgery, with fewer studies evaluating the relationships between imaging findings and tumor characteristics. For example, Irshad et al. (8) reported that the presence of posterior shadowing on ultrasound was strongly associated with an ER-positive and low-grade tumor, whereas posterior enhancement was strongly associated with a high-grade tumor and the risk of being receptor negative. Similarly, contrast-enhanced ultrasound kinetics was associated with the hormonal status of invasive breast cancer (9), and greater stiffness on shear-wave elastography correlated with poorer prognostic factors (10 12). However, the associations of imaging features with patient prognosis were not directly evaluated. We found that a BI-RADS 5 classification was associated with older age, larger tumor size, lymph node metastasis and lymphatic invasion, but not with the immunohistochemical profile of the tumor. BI-RADS 5 was associated with poorer DFS than BI-RADS 3 4, but this factor was not significant on multivariate analysis. Subgroup analysis of patients with Stage I breast cancer showed that BI-RADS 5 and HER-2 positivity were significantly prognostic. Few included patients had been treated with targeted anti-her-2 target, because of several reasons such as lack of insurance coverage. Thus, HER-2 positivity was the strongest prognostic factor. Multivariate analysis except for HER-2 receptor status showed that BI-RADS 5 was the only factor prognostic of DFS in T1N0M0 breast cancer patients. Since ultrasound BI-RADS was developed in 2003 by the American College of Radiology, most radiologists carrying out breast imaging have used this system in reporting and data acquisition. BI-RADS had many descriptive terms and examiner should choose the individual descriptor each time. Therefore, they might occasionally encounter some difficulties in their respective choices, leading to variable interpretations. Interobserver variabilty have been studied by many authors, and they have argued that the ultrasound lexicon has demonstrated good agreement between observers (5,6,13,14). The main difference between BI-RADS 4 and 5 lesions is that cancer is almost certain in the BI-RADS 5 lesion. However, the reference of Category 5 out of Category 4 has the lack of standard. In our institution, if the mass lesion had coexistence of microcalcification on the image of mammography and/or ultrasound, it has been considered the significant finding of Category 5. Moreover, mass lesion which had hypoechoic halo also has been considered as Category 5. It is unclear why BI-RADS was significantly associated with prognosis in early breast cancer patients. Conventionally, the risk of breast cancer recurrence is estimated based on histologic type and grade, receptor status and LN metastasis or by composite prognostic tools. In advanced breast cancer, these conventional prognostic factors were considered more effective for prediction than BI-RADS classification. However, in addition to these conventional factors, the other factors such as genetic variation or molecular changes might affect the morphology of cancer especially in early breast cancer. Certain genetic or molecular changes may manifest as altered tumor morphology. For example, loss of adhesion molecules, which have been associated with tumor progression and is predictive of survival, is thought to affect the histological appearance of tumors, resulting in loosely dispersed linear columns of cells and a typical discrete mass in invasive lobular carcinoma (15,16). In addition, the activation of phosphatidylinositol-3 kinase signaling is a frequent event in malignant tumors, and is associated with modulation of collagen cross-linking (17,18), which may be associated with changes in tumor morphology. Further research on the correlation between biophysical and imaging findings in breast cancer patients may broaden the clinical significance of our results. This study has several limitations. Due to its retrospective nature, there may have been a selection bias. Specifically, ultrasound images were not reinterpreted as part of the prospective design. Therefore, several radiologists performed the breast ultrasound examinations and assigned their BI-RADS classification. However, in >80% of all study cases, the ultrasound was performed and interpreted by just one special radiologist. In addition, the median follow-up of 42 months was not sufficient to assess the long-term prognostic significance of BI-RADS classification in breast cancer patients. Moreover, since the number of included patients was small, the number of disease-specific events was quite small. We concluded that BI-RADS classification in patients with early breast cancer might be considered as the independent prognostic factor. However, this result was interpreted carefully due to several limitations. Further confirmatory studies and biophysical studies are needed. Conclusion By assessing the prognostic significance of BI-RADS classification in patients with invasive breast cancer, we found that DFS was poorer in patients with BI-RADS 5 than in those with BI-RADS 3 4. In patients with T1N0M0 breast cancer, BI-RADS 5 was a significant independent prognostic factor. Conflict of interest statement None declared. References 1. Liberman L, Menell JH. Breast imaging reporting and data system (BI-RADS). Radiol Clin North Am 2002;40:409 30.
Jpn J Clin Oncol, 2015, Vol. 45, No. 5 415 2. Smith RA, Cokkinides V, Brooks D, Saslow D, Brawley OW. Cancer screening in the United States, 2010: a review of current American Cancer Society guidelines and issues in cancer screening. CA Cancer J Clin 2010;60:99 119. 3. Berg WA. Beyond standard mammographic screening: mammography at age extremes, ultrasound, and MR imaging. Radiol Clin North Am 2007; 45:895 906. 4. Kim EK, Ko KH, Oh KK, Kwak JY, You JK, Kim MJ, et al. Clinical application of the BI-RADS final assessment to breast sonography in conjunction with mammography. AJR Am J Roentgenol 2008;190:1209 15. 5. Lazarus E, Mainiero MB, Schepps B, Koelliker SL, Livingston LS. BI-RADS lexicon for US and mammography: interobserver variability and positive predictive value. Radiology 2006;239:385 91. 6. Lee HJ, Kim EK, Kim MJ, Youk JH, Lee JY, Kang DR, et al. Observer variability of Breast Imaging Reporting and Data System (BI-RADS) for breast ultrasound. Eur J Radiol 2008;65:293 8. 7. Kuo YL, Cheng L, Chang TW. Clinical impact of BI-RADS classification in Taiwanese breast cancer patients: BI-RADS 5 versus BI-RADS 0 4. Eur J Radiol 2012;81:1504 07. 8. Irshad A, Leddy R, Pisano E, Baker N, Lewis M, Ackerman S, et al. Assessing the role of ultrasound in predicting the biological behavior of breast cancer. AJR Am J Roentgenol 2013;200:284 90. 9. Szabo BK, Saracco A, Tanczos E, Aspelin P, Leifland K, Wilczek B, et al. Correlation of contrast-enhanced ultrasound kinetics with prognostic factors in invasive breast cancer. Eur Radiol 2013;23:3228 36. 10. Youk JH GH, Son EJ, Kim JA, Jeong J. Shear-wave elastography of invasive breast cancer: correlation between quantitative mean elasticity value and immunohistochemical profile. Breast Cancer Res Treat 2013;138: 119 26. 11. Chang JM, Park IA, Lee SH, Kim WH, Bae MS, Koo HR, et al. Stiffness of tumours measured by shear-wave elastography correlated with subtypes of breast cancer. Eur Radiol 2013;23:2450 8. 12. Evans A, Whelehan P, Thomson K, McLean D, Brauer K, Purdie C, et al. Invasive breast cancer: relationship between shear-wave elastographic findings and histologic prognostic factors. Radiology 2012;263: 673 77. 13. Abdullah N, Mesurolle B, EI-Khoury M, Kao E. Breast imaging reporting and data system lexicon for US: interobserver agreement for assessment of breast masses. Radiology 2009;252:665 72. 14. Park CS, Lee JH, Yim WH, Kang BJ, Kim HS, Jung JI, et al. Observer agreement using the ACR Breast Imaging Reporting and Data System (BI-RADS)-ultrasound, First Edition (2003). Korean J Radiol 2007;8: 397 402. 15. Doyle S, Evans AJ, Rakha EA, Green AR, Ellis IO. Influence of E-cadherin expression on the mammographic appearance of invasive nonlobular breast carcinoma detected at screening. Radiology 2009;253:51 5. 16. Gastl G, Spizzo G, Obrist P, Dunser M, Mikuz G. Ep-CAM overexpression in breast cancer as a predictor of survival. Lancet 2000;356:1981 2. 17. Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT, et al. Matrix crosslinking forces tumor progression by enhancing integrin signaling. Cell 2009;139:891 906. 18. Carracedo A, Pandolfi PP. The PTEN-PI3K pathway: of feedbacks and cross-talks. Oncogene 2008;27:5527 41.