Background
Neoadjuvant systemic therapy (NST) is increasingly used for the treatment of operable breast cancer in patients [
1]. NST has been shown to increase the rate of breast conservation without compromising survival [
2,
3]. The increased use of NST has given rise to some controversial issues such as the optimal method of determining the residual extent of tumor and the use of sentinel node biopsy after NST [
4‐
6]. In this study, we raise another clinically important issue for breast cancer patients who receive NST: the issue of initial axillary staging. The decision for post-NST axillary management and adjuvant radiation therapy often relies on the initial axillary nodal status. As more patients with early breast cancer receive NST, the importance of accurate initial axillary nodal staging is increasing.
In breast cancer patients who undergo primary surgery, the number of metastatic lymph nodes is a major prognostic factor, and the risk of recurrence is proportional to the degree of disease burden in the axillary lymph nodes [
7,
8]. In contrast, it is often difficult to obtain an accurate estimation of the extent of nodal involvement in patients who undergo NST. The current TNM staging system recommends physical examination to determine the N stage in patients receiving NST, based on the presence of the fixed nature or locations of the palpable nodes [
9]. Unfortunately, studies have reported limited accuracy of physical examination in determining clinical N stage in patients with breast cancer, with a sensitivity around 30% [
10‐
13]. Furthermore, clinical N staging with physical examination is a highly subjective method, and the accuracy may vary among surgeons.
In this study, we aimed to develop an objective staging system based on initial computed tomography (CT) images that can provide improved prognostic information for patients who receive NST.
Methods
The medical records of patients who underwent NST and surgery for invasive breast cancer at Seoul National University Hospital (SNUH) between Jan 2006 and Dec 2011 were reviewed retrospectively. In our institution, patients who undergo neoadjuvant systemic therapy often received chest CT to rule out visceral metastasis and to assess the degree of nodal enlargement prior to the administration of the systemic treatment. For the present study, we excluded the patients who underwent palliative surgery or had a history of breast cancer. Also, patients who did not have chest CT images at diagnosis or who had poor-quality images were excluded. Patients were divided into two cohorts, in the ratio of two to one, for development and validation of a CT-based nodal staging system (development cohort and validation cohort, respectively). Patients’ data including clinicopathologic, treatment-related, and survival information were obtained from SNUH Breast Cancer Center database, which is a prospectively maintained web-based database [
14]. In all individuals, CT scanning was performed at end-inspiration following hyperventilation. CT imaging was performed using the following scanners: Genesis Hispeed and LightSpeed Ultra; GE Healthcare, Milwaukee, WI, USA; Somatom Plus-4 and Sensation-16; Siemens Medical Systems, Erlangen, Germany; Brilliance-64; Phillips Medical Systems, the Netherlands. Two board-certified radiologists (mean 13 years of experience) who were blind to the clinical staging and treatment outcome information evaluated the CT scans independently. On the CT scans, lymph node status was evaluated based on the size and location (axillary levels I, II, and III, supraclavicular lymph node (SCN) or internal mammary lymph node (IMN)). At each location, the number of lymph nodes larger than 1 cm and larger than 2 cm was separately recorded. Discordant data from two radiologists were reevaluated by the same two radiologists to reach a consensus, and the final integrated results were recorded.
All patients were treated with anthracycline- and/or taxane-based chemotherapy regimens. Among the patients who had a human epidermal growth factor receptor 2 (HER2)-amplified tumor, 15% (17/113) received a HER2-related targeted therapy preoperatively. All patients received postoperative hormonal or radiation therapy, if indicated. Postoperative follow-up with the clinic was done at least every 6 months and included a routine physical examination, laboratory testing, breast ultrasonography, mammography and chest X-ray. Additional examinations were done at the physician’s discretion.
In this study, we used two additional groups of patients to assess the negative predictive value of CT evaluation and to estimate the hormone receptor (HRc)-dependent prognostic impact of nodal status. To assess the negative predictive value of the axillary nodal status as evaluated by chest CT, the medical records of breast cancer patients who received primary surgery as their initial treatment between January 2014 and December 2014 were reviewed for CT findings and pathological N staging (supplementary cohort A). To demonstrate the HRc-dependent prognostic effect, we selected consecutive patients who received primary surgery between July 2005 and June 2008 in whom adequate survival information was available (supplementary cohort B).
Univariate analysis, using the Pearson chi-square test, was performed to compare clinicopathologic features of the development and validation cohorts. Univariate survival analysis for development and validation of the CT-based nodal staging system was performed using Kaplan-Meier survival analysis and log-rank tests. Multivariate survival analysis was conducted using a Cox proportional hazards regression model. Prognostic factors with statistical significance in the univariate analysis were entered in the Cox regression model. Disease-free survival (DFS) was defined as the time from start of neoadjuvant systemic therapy to the date of breast cancer recurrence, death from any cause or final outpatient clinic visit. Breast cancer recurrence was defined as locoregional recurrence or distant metastasis, and contralateral breast recurrences were excluded.
Discussion
In the present study, we developed and validated an objective clinical N staging method using initial chest CT images of patients undergoing NST. When the patients were classified according to the number of axillary lymph nodes larger than 1 cm, we observed a significant and proportional increase in risk of recurrence. Our data suggest that a CT-based objective axillary staging using the number of enlarged nodes can be a useful alternative to conventional axillary staging done by a physical examination.
Many previous studies have attempted to explore the value of imaging studies, including contrast-enhanced breast magnetic resonance imaging, ultrasonography, and positron emission tomography (PET) scan, in determining nodal status in patients with breast cancer [
15]. However, most studies have focused on the role of imaging studies in predicting the presence of nodal metastasis rather than evaluating the quantitative burden of nodal disease. Axillary ultrasound is an imaging modality associated with low cost and risk and is also reported to have high sensitivity in excluding or predicting heavy nodal burden [
16‐
18]. However, quantification of nodal disease using ultrasound is difficult and operator dependent, which compromises the objectivity of the imaging test. The practicality of a PET-CT scan as a nodal staging modality has also been demonstrated in many studies recently [
19‐
24]. For patients undergoing NST, Koolen et al. have demonstrated the role of PET-CT scan for quantification of axillary nodal status and evaluation of extra-axillary nodal involvement [
19,
21]. A PET-CT scan performs with high sensitivity and specificity, but its lack of a standard cutoff level for standardized uptake values (SUVmax) and its dependency on the SUVmax of the main tumor are obstacles for objective axillary nodal classification [
20,
21,
24]. Moreover, considering the cost and availability of this technique, PET-CT may be best reserved as an adjunct for indeterminate lesions [
25]. Compared to these imaging modalities, axillary staging done by chest CT imaging has the advantages of a fairly low cost and high availability. Also, the staging system developed in this study enables physicians to quantify axillary nodal status through an objective and reproducible method for women undergoing NST.
It has been reported that extra-axillary lymph node metastasis, such as SCN and IMN, is associated with poor outcome in patients with breast cancer [
9,
26‐
29]. In our study, we could not demonstrate an independent prognostic value of extra-axillary lymph node enlargement. While patients with enlarged SCN showed worse outcomes, they often had more than three enlarged axillary nodes. In patients with more than three enlarged axillary nodes, the presence of SCN did not confer a significant prognostic difference. Our data suggest that the known prognostic importance of extra-axillary nodal involvement can be the consequence of the degree of axillary metastatic burden rather than an independent prognostic factor. Indeed, Olivotto et al. [
30] showed similar overall survival between patients with SCN metastasis and stage IIIB tumors, and Chen et al. [
31] showed comparable outcomes between patients with SCN metastasis and N3 stage tumors.
The ability of our staging system to classify patients according to their risk of recurrence was more pronounced in HRc-negative tumors. This phenomenon was also seen in the pathologic nodal staging system for 1702 primary breast cancer patients who received surgery as their initial treatment (Supplementary cohort B). This can be explained by the effect of a higher baseline risk of recurrence for HRc-negative tumors compared to that of HRc-positive tumors, despite a similar increase in the relative risk [
32,
33]. On the other hand, the prognostic impact of the degree of nodal involvement may differ according to the molecular characteristics of the breast cancer [
34]. The relationship between nodal metastatic burden and the risk of recurrence according to the molecular subtypes of breast cancer should be examined further with a larger dataset.
The retrospective nature of this study is a major limitation. Various CT systems were used for evaluation of enlarged lymph nodes, causing minor differences in slice thickness, resolution, and image quality. A prospective validation study with a standardized protocol is needed to strengthen the value of our nodal staging system. Also, our patient cohort was mainly composed of stage III breast cancer patients (71.2%); therefore, applying our results to early breast cancer patients may have some limitations. Other limitations include the lack of pathologic confirmation of extra-axillary node involvement and the lack of subtype-specific analysis due to the limited number of patients.