Internal Mammary Lymphadenopathy Does Not Impact Oncologic Outcomes in Patients Treated with Neoadjuvant Chemotherapy: Results from the I-SPY2 Clinical Trial.
Written by: Mara A PiltinPeter NorwoodVelle LadoresRita A MukhtarCandice A SauderMehra GolshanJulia TchouRoshni RaoMarie Catherine LeeJennifer SonChantal ReynaKelly HewittHenry KuererGretchen AhrendtIan GreenwaltJennifer TsengLauren PostlewaitMarissa Howard-McNattNora JaskowiakLaura J EssermanJudy C Boughey, ISPY2 Locoregional Working Group
Find these authors on Roon
JS
Dr. Jennifer Son
Breast Surgical Onc · Medstar Montgomery Medical Center
CR
Dr. Chantal Reyna
Breast Surgical Onc · Foster G. McGaw Hospital
KH
Dr. Kelly Hewitt
Breast Surgical Onc · UUHC
Introduction
Significant improvements in systemic therapy for breast cancer have continued to drive re-evaluation of locoregional treatment techniques and considerations. Historically, patients with internal mammary (IM) chain lymphadenopathy have been identified as having worse prognosis compared to those with the same disease biology, tumor size (cT) and clinical nodal category (cN) without IM involvement.1,2,3,4 The three most common lymphatic draining nodal basins of the breast are well described and include the ipsilateral axilla, internal mammary (IM) chain, and supraclavicular regions.5,6 Utilizing lymphatic mapping, the ipsilateral axilla has been clearly defined as the dominant draining basin for breast cancer and thus remains the location for sentinel lymph node (SLN) biopsy when surgically staging the regional nodes in a patient with primary breast cancer. In addition to the axilla, sentinel lymphatic drainage to the IM chain occurs in approximately 13–37% of cases.4,7,8,9 However, it is far less common to identify isolated IM node involvement in the absence of concomitant axillary nodal involvement.4,10,11,12 Surgical resection of the IM lymph nodes has been explored historically but has not shown to provide oncologic benefit; therefore, it has been largely abandoned.13,14,15 Current National Comprehensive Cancer Network (NCCN) recommendations support locoregional treatment of involved IM nodes with adjuvant radiation strategies and does not endorse routine surgical resection.16 For these reasons, attention to IM nodal involvement in the literature has been limited.Cross sectional imaging with breast Magnetic Resonance Imaging (MRI) and Positron Emission Tomography – Computed Tomography (PET-CT) are frequently employed in the workup and staging algorithm for breast cancer,17,18 especially when neoadjuvant chemotherapy is being considered. Percutaneous biopsy of IM nodes is not commonly employed and thus, IML detected by imaging is utilized as a surrogate for metastatic involvement by breast cancer. The presence or suspicion of abnormal IM nodes on imaging can raise a patient’s clinical N category from N0 to N2b in the absence of axillary disease or from N1 or N2 to N3b in the presence of concomitant axillary involvement.19,20 This ultimately impacts the patient’s clinical stage, treatment recommendations, prognosis, and outcomes. In the context of modern systemic therapy utilizing neoadjuvant treatment approaches, prognosis and adjuvant therapy recommendations are largely impacted by response to treatment. We sought to evaluate the frequency of detection of IM nodes on imaging and the impact of IML on oncologic outcomes in a large cohort of patients treated with neoadjuvant chemotherapy on a prospective clinical trial.
Methods
We performed a retrospective analysis of patients enrolled on the I-SPY2 multi-center clinical trial.21 This is an ongoing randomized controlled trial which utilizes adaptive randomization to test novel systemic agents in the neoadjuvant setting. Subjects are stratified by risk as determined by 70-gene signature (Mammaprint) testing on the diagnostic core biopsy. All patients proceed to surgical intervention after neoadjuvant treatment. Enrolled patients are ≥ 18 years of age, tumor size ≥ 2.5 cm by clinical examination or ≥ 2 cm by imaging and have an Eastern Cooperative Oncology Group performance status of 0 or 1. Eligible tumor subtypes include luminal breast cancer if high risk by 70-gene signature (Mammaprint), triple negative, or human epidermal growth factor receptor 2 (HER2)+ disease. All patients enrolled on the trial have a baseline dynamic contrast enhanced MRI prior to neoadjuvant chemotherapy, and utilization of PET-CT or other systemic staging techniques such as CT of the chest, abdomen and pelvis plus bone scan is at the discretion of the providers. After the completion of neoadjuvant chemotherapy, all patients proceed to surgical treatment in the form of breast conserving surgery or mastectomy with sentinel lymph node surgery and/or axillary lymph node dissection. All patients must achieve negative surgical margins and radiation delivery is at the discretion of the treating providers. The primary endpoint of the trial is the rate of pathologic complete response (pCR) in each arm and secondary endpoints are event-free survival (EFS) and distant recurrence free survival (DRFS).21We evaluated all patients enrolled on the I-SPY2 clinical trial from March, 2010 – October, 2022. We utilized radiographic IML as a surrogate for involvement of IM nodes by malignancy, as is standard in current clinical practice. IML was defined as abnormal nodes by MRI, abnormal uptake on PET-CT, or both as these were the only two imaging modalities with IM node assessment captured on the trial. This was obtained from a baseline assessment form submitted for each patient enrolled at the time of diagnosis. Abnormal is not explicitly defined with a size measurement, but determination is at the discretion of the treating facility. Tumor biology was categorized as hormone receptor (HR) positive if estrogen receptor (ER) and/or progesterone receptor (PR) were positive (≥1% positively staining cells by immunohistochemistry respectively). HER2+ was defined as expression of HER2 staining intensity of IHC 3+, or IHC 2+ and positive by fluorescence in situ hybridization (FISH). Approximated biologic subtypes were categorized as HR+/HER2 negative (HR+/HER2−), HER2+ and HR−/HER2−. We evaluated the position of the tumors within the breast and defined inner breast location as those positioned in the upper inner quadrant, lower inner quadrant, 12 o’clock position, 6 o’clock position and left breast 9 o’clock position or right breast 3 o’clock position. Percutaneous sampling of axillary nodes was performed in the setting of suspicious axillary nodes. Pathology from breast and nodal surgery was captured to assess for response and extent of any residual disease, with Residual Cancer Burden (RCB) calculated by centrally trained study pathologists. From imaging studies at presentation, the size of largest abnormal IM node was collected when available. We also identified performance of PET-CT, the occurrence of radiographically abnormal IM node(s) and SUV max of IM node(s) when available. Rates of pCR were recorded in each group. We then evaluated locoregional recurrence (LRR), distant recurrence (DR), and estimated cumulative event free survival (EFS) in those with IML compared to those without IML overall as well as within the group who achieved a pCR and within the group who did not achieve a pCR separately. An event impacting EFS was defined as any locoregional, DR or death from any cause. LRR was defined as the presence of invasive disease involving the breast, regional lymph nodes including axilla, chest wall and/or skin ipsilateral to the primary diagnosis. DR was any recurrence except LRR or contralateral new breast cancer. The date of treatment consent was utilized for survival and patients without an event were censored at the last follow-up date.Statistical analyses compared distributions of key variables between the IM nodal cohorts. For non-survival distributions, p-values were generated from Fisher’s Exact Test for categorical variables and the Student’s t-test for continuous variables. The log-rank test was used for survival distributions; associated survival curve figures were also presented. R22 was used for all statistical analyses, the survival package23,24 was used to run log-rank tests, and the ggsurvfit package25 was used to generate survival figures.
Results
A total of 2,096 patients were identified from the I-SPY2 clinical trial from March, 2010 – October, 2022 with a median follow up of 3.72 years (range 0.4–10.2). 199 patients (9.5%) were found to have abnormal IM lymph nodes by MRI and/or PET-CT. One patient later classified as M1 was excluded due to progression to metastatic disease while on neoadjuvant therapy. This resulted in 198 (9.5%) patients with IM lymphadenopathy (IM+) in our study cohort of 2,095 total patients. The remaining 1,897 patients (90.5%) did not have abnormal IM nodes by either imaging modality (IM-). Mean age at diagnosis was 48.4 years, with the IM+ patients being younger (mean age 46.0 years vs 48.6 years) (p = 0.001). There was no difference between the IM+ and IM- groups in race, ethnicity, tumor laterality or approximated biologic subtype (Table 1). Patients with IM+ nodes were more likely to have larger clinical T category at presentation (p<0.001) and have grade III disease (60.1% vs 48.6%, p = 0.026). Of the 198 patients with IM+ disease, 47% (n=94) were inner breast tumors. 107 IM+ patients had axillary lymph node imaging abnormalities and were biopsied, with 89 resulting as positive for malignancy, for a rate of concomitant axillary and IM nodal involvement of 45%. Use of adjuvant radiation therapy was higher (68.2% vs 54.1%, p<0.001) in patients with IM+ disease; however, there is a high rate of missing radiation data (31.3% overall), higher in the IM- group (Table 1).Evaluation of the breast and axillary surgical procedures performed, rate of pCR, residual cancer burden (RCB) class, and pathologic tumor and nodal categories was performed. There were no statistically significant differences between IM+ and IM- in any of these measures, except for pCR rate. The IM+ patients had higher rate of pCR (41.4% vs 34% respectively, p=0.031) (Table 2).Of the 198 IM+ patients, 154 patients had IML by MRI only, 11 by PET-CT alone and 33 by both modalities (Table 3). Since all patients had an MRI, the rate of IM+ on MRI was 188/2095 (8.97%). A total of 505 patients of the entire study cohort underwent PET-CT as a component of their workup and the rate of IM+ on PET-CT was 44/505 (8.71%). Use of PET-CT was higher in patients that were IM+ (43.3%, n=86) than in patients that were IM- (22.1%, n=419).On MRI, the majority of IM+ patients had one abnormal IM node (n=112, 56.6%) and the mean size of the largest IM lymph node was 7.52 mm (SD 3.18). The average size increased when both imaging modalities detected IML (8.37 mm vs 7.34 mm). Among IM+ cases by PET-CT, the mean SUV max of the IM node(s) was 4.39 (SD 3.8) (Table 3).Evaluating oncologic outcomes, we found no differences between IM+ and IM- patients with respect to LRR (5.6% vs 3.8%, p=0.251), DR (9.1% vs 7.9%, p=0.581), or EFS (61.6% vs 57.2%, p=0.481) (Table 4). Figure 1A shows the Kaplan-Meier estimates of the probability of EFS by the presence or absence of IML. Looking specifically at patients who achieved a pCR, there was also no significant difference between the IM+ and IM- groups in oncologic outcomes (Figure 1B). While these event rates became quite small, the group who achieved pCR had LRR rates for IM+ vs IM- of 2.4% vs 0.5% (p=0.14), DR of 4.9% vs 2.6% (p=0.282) respectively and EFS of 70.7% vs 68.4% (p=0.1) (Table 5). Similarly, for patients who did not achieve a pCR, we saw no statistical differences in LRR, DR or EFS (Table 6, Figure 1C).
Discussion
For the infrequent finding of IML identified on imaging in the clinical workup of breast cancer, we utilized a large cohort of patients from a multicenter clinical trial, I-SPY2, to show the impact of IM+ nodes in patients treated with neoadjuvant chemotherapy in a modern context. Prior studies have identified patient and tumor factors that are associated with higher likelihood of IM nodal involvement, including a study evaluating 1,697 patients who underwent extended radical mastectomy in the absence of pre-operative systemic therapy from 1956–2003. They found that >4 positive axillary nodes, medially located T2 tumors, medial tumors of any size with concomitant axillary involvement, T2 tumors in any location with concomitant axillary involvement, and age <35 with a T3 tumor were all associated with a >20% risk of having IM nodal involvement pathologically at surgery.26 In a study published in 2023 by Qiu et al., predictive modeling was used to identify factors associated with IM+ metastasis. They found in their validated nomogram that tumor size, location, presence of lymphovascular invasion, and number of involved axillary nodes all had a significant relationship with IM nodal metastasis on multivariate analysis.27 We similarly identified that our IM+ patients were more likely to be younger, have larger tumors, and higher grade (all p<0.05). Of the IM+ patients in our study, 47.5% of the tumors were located in the inner breast (n=94/198) and 45% of the IM+ patients had biopsy proven axillary nodal involvement (n=89/198).Due to the reduced frequency of surgical resection of IM nodes, as well as minimal instances of percutaneous sampling in current clinical practice, we utilized abnormal IM lymph nodes detected by MRI and/or PET-CT as a surrogate for involvement by metastatic malignancy. This approach has been utilized in prior studies.28,29,30 We identified 9.5% of the total study population had abnormal IM nodes by imaging, which is similar to previous reports.28,31,32 Since utilization of PET-CT in this study was at the discretion of the treating physicians, we were unable to perform a head-to-head analysis of MRI and PET-CT for the detection of IML. However, all patients underwent MRI per study protocol. Prior publications have shown that the performance of MRI in detecting IML is not inferior to PET-CT,17,33,34 although head-to-head analyses are limited. Interestingly, we found that use of PET-CT was higher in patients that were IM+, which may be related to the baseline disease characteristics or may be secondary to findings from an MRI and could be further explored in future studies.Historically, IM nodal involvement has indicated worse prognosis and outcomes for patients. This was largely investigated prior to the significant utilization of neoadjuvant chemotherapy and cross-sectional imaging modalities have continued to improve over time thus identifying small and possibly less clinically significant disease. We sought to identify if the clinical detection of IML in a modern cohort of high-risk patients treated with neoadjuvant therapies remained a poor prognostic predictor. In our cohort, we found that oncologic outcomes were not negatively impacted by the presence of IML. This finding remained statistically true when evaluating patients with or without IML who achieved a pCR and who did not achieve a pCR. Interestingly, we identified a higher rate of pCR within the IM+ group. This could be attributed to a higher proprtion of grade III disease within the IM+ group, resulting in a better response to NAC.35,36 We saw no difference in the types of breast or axillary surgery performed. Importantly, we did note a significantly higher use of radiation therapy in the IM+ group, as expected. This illustrates that the presence of IML is influencing treatment strategies effectively. Of note, at least 6% of the patients with IML in our cohort did not receive radiation, which would be recommended by the NCCN guidelines. We suspect that this is a combination of shared decision making between patients and providers, or potentially that the IML documented on imaging at diagnosis was not ultimately felt to be clinically meaningful by the treating team after completion of neoadjuvant chemotherapy and surgery.While our study utilized a large cohort of patients to obtain relatively high numbers of an uncommon clinical finding, the number of patients with involved IM nodes remains limited. Further follow-up is necessary to comment on long-term oncologic outcomes with confidence, specifically in patients with luminal breast cancer. Additionally, on the ISPY-2 trial, utilization of PET-CT was at the discretion of the treating team and other systemic staging techniques such as CT chest were not captured. Additionally, there was no standardized definition of abnormal IM node such as minimum size criteria on MRI or SUV max on PET. While this introduces some heterogeneity in the patient population, this is reflective of real world practice. This evaluation would benefit significantly from more complete information on the delivery, dose and treatment plans of radiation therapy.As the multidisciplinary care of breast cancer evolves, improvements in imaging techniques, systemic therapy, surgery and radiation have changed the way we diagnose and treat IML. This study presents a valuable addition to the literature as an update on the oncologic impact of IML in the context of a rapidly advancing field. Historically, IML was associated with poor prognosis and here we show that with our current clinical practice, other factors such as response to neoadjuvant therapy likely have a stronger impact on oncologic outcomes.
Conclusion
While IM nodal involvement is a relatively uncommon clinical finding in the staging workup of breast cancer that impacts treatment decisions, we report on a large cohort of patients treated on a randomized controlled trial over >10 years of enrollment and identify no prognostic impact from the presence of IML at a median follow-up of 3.72 years.
JS
Dr. Jennifer Son
Surgery · Breast Surgical OncologyMedstar Montgomery Medical Centerjenniferson