Prognosis classification of breast cancer with distant lymph node metastasis IIIc or M1 category


 Background: In the American Joint Commission on Cancer (AJCC) staging manual, ipsilateral supraclavicular lymph node metastases (ISLM) were defined as N3c, while contralateral supraclavicular and subclavian lymph node, contralateral internal mammary lymph node, contralateral axillary lymph node, and cervical lymph node as distant lymph nodes metastasis (DLM) were classified as M1, stage IV. Herein, we used the information recorded in the Surveillance, Epidemiology, and End Results (SEER) database to compare patients' overall survival with DLM, ISLM, and distant organ metastasis. Methods: A total of 376 patients with ISLM, 562 patients with DLM, 5,069 patients with IIIc stage breast cancer, and 7,540 patients with distant organ metastasis from the SEER database (2004-2016) were included in the present study. R package was used to perform the Kaplan–Meier survival analysis among these four groups, and propensity score weighting (PSW) was used to minimize the differences in clinicopathological characteristics among groupsResults: After PSW, the median survival of ISLM, DLM, IIIc stage, and distant organ metastasis was 42 months, 50 months, 55 months, and 31months, respectively. DLM achieved significantly better overall survival than distant organ metastasis (p<0.001, HR=0.646, 95%CI:0.561-0.737), while there was no significant difference in long-term survival between DLM and ISLM (p=0.467, HR=0.915, 95%CI:0.719-1.163) or DLM and IIIc stage (p=0.959, HR=0.995, 95%CI:0.837-1.183). Surgery (p<0.001, HR=0.502, 95%CI:0.375-0.674) and chemotherapy (p=0.007, HR=0.701, 95%CI:0.527-0.932) could significantly improve the OS for patients with DLM. When compared with chemotherapy alone, a combination of chemotherapy, surgery, and radiotherapy could achieve better survival, with the increased median survival from 26 months to 82 months (p<0.001, HR=0.366, 95%CI:0.253-0.529).Conclusion: The prognosis of breast cancer patients with DLM was similar to ISLM and IIIc stage, but much better than that of patients with distant organ metastasis. Also, combined-modality therapy significantly improved long-term survival. Consequently, it seems more reasonable to classify DLM as IIIc category instead of M1.


Introduction
Anatomy staging of breast cancer is critical to determine prognosis and treatment choice; however, cancer staging is not an exact science. Following newly available information on etiology and various diagnosis and treatment methods, the classi cation and staging of cancer also tend to change. [1] In the fth version of AJCC TNM staging, ipsilateral supraclavicular lymph node metastasis (ISLM) was de ned as M1, stage IV [2]. However, related studies have con rmed that ISLM was more consistent with N3, stage III, in terms of survival prognosis and tumor biological behavior. [3,4] Therefore, in the sixth edition of TNM, ISLM was rede ned as N3c, stage IIIC [5], which changed patients' treatment decisions with ISLM in clinical practice. [6,7] It appears that distant lymph node metastases (DLM) face the same situation.
The de nition of DLM in breast cancer is based on the existence of at least one of the following conditions: contralateral internal mammary lymph node metastasis (CILM), contralateral axillary lymph node metastasis (CAM), contralateral supraclavicular, subclavian lymph node metastasis (CSLM), and cervical lymph node metastasis (CLM). [8][9] Information on DLM is lacking due to its low incidence. [10][11][12] As far as we know, there are no cohort studies to date on the treatment outcomes of DLM; thus, there is no consensus on how to manage these patients. DLM may have synchronous or metachronous presentation [10,13,14], where the former implies that DLM was found at the rst visit, while the latter means that the DLM was not found at the rst visit but was found during the follow-up time.
In the AJCC staging system, DLM of breast cancer is classi ed as M1, stage IV. Nevertheless, there are controversies regarding the clinical treatment decision. Retrospective studies with a small sample size have observed that the survival prognosis of patients with DLM is more similar to patients classi ed as N3c or IIIC and is signi cantly different from the patients with distant organ metastasis, such as bone, liver, and lung. [8][9] However, whether DLM should be considered as an advanced locoregional stage or distant metastasis has not yet been answered and should be addressed by prospective trials, which are di cult to conduct since the low incidence of DLM.
The aim of this study was to use the Surveillance, Epidemiology, and End Results (SEER) database to compare the overall survival of DLM, ISLM, IIIc stage, and distant organ metastasis. Furthermore, we also aimed to analyze the impact of different treatment modalities on survival, hoping it could offer some implication to clinical practice.

Data Extraction
This study was approved by the Institutional Review Board by Chinese Center of Disease control and prevention. Data were acquired from the SEER database between January 1, 2004, and December 31, 2015, as the 6th AJCC TNM staging system was adopted by SEER in 2004. Data were extracted for all cases that were initially diagnosed as malignant primary breast cancer, while multiple primary malignant tumors were excluded. Breast cancer patients who lack information on tumor staging, therapy, or prognosis were excluded. After these steps, we excluded patients who were male and with bilateral breast cancer. The follow-up time was from breast cancer diagnosis until death or the end of the follow-up period. Finally, 376 patients with ISLM but no distant metastasis classi ed as N3cM0, 562 patients with DLM classi ed as M1-LN, 7,540 patients with distant organ metastasis classi ed as M1, and 5,069 patients with IIIc stage classi ed according to the 6th edition of AJCC were included in the study ( Figure  1).

Data Availability
All data generated or analyzed during this study are included in this published article and its supplementary les. The original data were downloaded from the SEER Web site server (https://seer.cancer.gov/data/) via the SEER*Stat software version 8.3.5 in the client-server model. [15] Propensity score weighting (PSW) This was a retrospective and observational study. Some signi cant covariates of the patients were heterogeneous and possibly affected the outcomes. Propensity score weighting (PSW) was applied to minimize the selection bias and to approximately balance the covariates among groups. [16] The propensity score for each patient was estimated by a logistic regression model in the N3cM0 group, IIIc group, M1-LN group, and M1 group according to age at diagnosis, histologic type based on the ICD-O-3, tumor differentiation grade based on the ICD-O-2, T or N stage based on the Derived AJCC Stage (6th), estrogen receptor (ER) status, progesterone receptor (PR) status, Her2 status, surgery, chemotherapy and radiotherapy. The survival rate among the four groups and different treatments in M1-LN were further compared.

Statistical analysis
Overall survival (OS) was de ned as the time from the date of diagnosis to the date of death due to any cause or the date at the last follow-up. The survival probability of the four groups was calculated by the Kaplan-Meier method and compared by log-rank test before and after PSW. To evaluate the impact of treatment modalities (surgery, radiotherapy, chemotherapy, comprehensive therapy) on survival, Kaplan-Meier survival analysis, and log-rank test were also used before and after PSW. The de nition of surgery is presented in Supplemental Table 1. Baseline characteristics were compared using the χ2 test or Fisher's exact test before and after PSW ( Table 1, Table S2-5). Univariate and multivariate analyses were performed using a Cox regression proportional hazard model in the M1-LN group. The hazard ratio (HR) and corresponding 95% con dence interval (CI) were also calculated. All statistical analyses were performed in RStudio (version 1.3.1056, R Foundation for Statistical Computing, Vienna, Austria) using the packages "survival" (version 2.41-3), "nnet" (version 7. [3][4][5][6][7][8][9][10][11][12][13][14], and "IPWsurvival" (version 0.5). All P values were calculated as two-sided, with P< 0.05 being considered as statistical signi cance.

Baseline characteristics of the patients
The clinical and pathological features of the primary tumors and the treatments of these cohorts are summarized in Table 1 and Table S2

Survival analysis
For the survival analyses, patients' numbers in each background factor were adjusted by weighting propensity score. After the adjustment, the clinical and pathological factors no longer differed among cohorts (Table 1, Table S2-5). Kaplan-Meier survival curves and people's number at risk at different follow-up times were performed before and after PSW (Figure 2A & B In order to evaluate the impact of treatment strategies on OS, Kaplan -Meier survival curves were performed before and after PSW in the M1-LN group according to whether chemotherapy, radiotherapy, or surgery was done. Patients in the M1-LN group achieved signi cant survival bene t if they underwent either surgery (p<0.001) or chemotherapy (p<0.05) before and after PSW ( Figure 3A&B, E&F), while radiotherapy led to some survival bene ts (p<0.001, HR=0.467, 95%CI:0.361-0.604) before PSM, but not after PSW (p=0.117, HR=0.74, 95%CI:0.527-1.038, Figure 3C&D). When compared with chemotherapy alone, comprehensive therapy (surgery, chemotherapy, and radiotherapy) led to better survival before and after PSW (p<0.001, Figure 3G&H), with the median survival increasing from 26 months to 82 months.

Prognostic factors
Univariate and multivariate analysis was performed ( Table 2). Under multivariate analysis by Cox regression model, surgery and chemotherapy resulted as independent favorable factors as well as PR positive and HER2 positive. In contrast, lobular carcinoma, T4, and poor grade resulted as independent risk factors for prognosis.

Discussion
In the AJCC staging system, ISLM is de ned as N3, while contralateral supraclavicular and subclavian lymph node, contralateral internal mammary lymph node, contralateral axillary lymph node metastasis (CAM), and cervical lymph node metastasis (CLM) are de ned as DLM and classi ed as M1, stage IV. There are limited data and a number of controversies on DLM staging and clinical treatment decisions. Our study used the information recorded in the SEER database to address these issues.
The available literature is quite exiguous, and it mainly consists of small retrospective series and a few case reports. CAM is the most studied DLM. From 1995-2020, three relatively larger sample-sized studies reported on a total of 154 patients with CAM [8,17]. In one of these studies, 47 patients with CAM were suspected of having N3 disease, showing better response than at other sites of distant dissemination, when subjected to surgical and systemic treatments with curative intent [8]; while the other two studies draw a contrary conclusion from 28 and 23 patients, respectively that CAM was most likely to be stage IV. [10,17] KROG 18 − 02 study that included 78 patients indicated that patients with breast cancer who had ipsilateral CLM without other distant metastasis might bene t from locoregional treatment of the ipsilateral breast and systemic therapies, just as those with N3c disease. [9] It remains unclear whether DLM should be considered an advanced locoregional stage or distant metastasis. Accordingly, prospective trials should address this issue, which is very di cult to conduct due to the low incidence of DLM. To the best of our knowledge, our study is the largest sample-sized study conducted to investigate the survival outcomes of patients with DLM.
In our study, we found no signi cant difference in long-term survival between DLM, N3c, and IIIc stage. Surgery and chemotherapy could signi cantly improve the OS for patients with DLM, where the multivariate analysis also showed that surgery and chemotherapy were independent favorable prognostic factors. Compared with chemotherapy alone, a combination of chemotherapy, surgery, and radiotherapy could achieve better survival. Accordingly, it is reasonable to reconsider DLM as a loco-regional disease and adopt a curative approach.
DLM is more common in patients with previous surgery or radiotherapy, resulting in disrupted primary lymphatic drainage. Repeated sentinel node biopsy in patients with locally recurrent breast cancer showed about 30% aberrant regional lymphatic drainage, including CAM. [18,19] The possible explanations for DLM of aberrant regional lymphatic drainage also support DLM as advanced stage III disease but do not provide support for hematogenous spread (stage IV).
DLM can happen synchronously and metachronously. In a retrospective study conducted at Mayo Clinic, most patients (18/23, 78.3%) had a metachronous CAM presentation, whereas 5 (21.7%) developed CAM synchronous with the primary breast tumor [17]. Synchronous CAM seemed to have more aggressive primary tumor features: 4/5 (80%) had in ammatory breast cancer, a high-grade disease requiring neoadjuvant chemotherapy, clinical ndings of a large tumor (11-20 cm), and N2-N3 axillary disease. A literature review showed that among 84 patients with isolated CAM (65% metachronous and 35% synchronous), patients who developed synchronous CAM typically had LABC or in ammatory breast cancer at the time of initial diagnosis [20]. Our results were similar with previous studies, as all the patients in our study had synchronous DLM, of which 39.5% wereT4 and 38.1% were N3, thus implying that disease burden in the breast and in the ipsilateral axilla led to the diversion of lymphatic ow to DLM.
Interestingly, Chkheidze's study reported that in most synchronous CAM cases (6/7), CAM was incidentally detected at the time by contralateral SLN at the time of contralateral prophylactic mastectomy.
[20] A large cohort study from Milan showed that contralateral or bilateral lymph drainage could be seen in T1-2 N0 early breast cancer patients, where the incidence rate was quite low, less than 1%.[21] It is possible that DLM is not that rare, especially in patients with higher T and N stage. In our study, the incidence of DLM in IIIC patients was at least 4.1%, and the incidence in N3c patients was at least 22.3% (data wasn't showed). Furthermore, we were curious about the starting point of metachronous DLM. In addition to local recurrence and aberrant drainage, it was important to learn whether it could result from undetected synchronous DLM. If so, lymphoscintigraphy may be necessary at the initial evaluation. Although it is not cost-effective from an SLNB perspective, it may be necessary for axillary staging of locally advanced patients such as in ammatory breast cancer or N3. Advanced imaging methods, such as PET-CT, could be used as an optional choice.
There are some limitations to the present study. Due to the characteristics of the SEER database, we were unable to distinguish the speci c lymph node location of DLM. Also, the treatment details were not available. The extent of the radiation elds is unknown, and it is unclear whether the radiation included the distant lymph node. Regarding the analysis of molecular traits, our ndings were limited by the incompleteness of the molecular subtyping and the information on HER2 status. Our study was a retrospective analysis, and despite the fact that PSW was performed, it was di cult to completely eliminate the bias. Therefore, reported results need to be further veri ed by future randomized control trials.

Conclusion
The prognosis of breast cancer patients with DLM was similar to those at ISLM and IIIc stage, but much better than in those with distant organ metastasis. Combined-modality therapy in accordance with the curable principle of locally advanced breast signi cantly improved long-term survival. Accordingly, it seems more reasonable to classify DLM as IIIc category instead of M1.

Declarations
Authors' contributions: Jinbo Wu and Taobo Hu participated in the study design and coordination and participated in the data analysis and written the draft. Shu Wang conceived the idea and design the study. All authors approved the nal manuscript.
Fundings: This study was sponsored by Minister of Science and Technology of People's Republic of China (2016YFC0901302).
Con ict of interest: The authors declare that they have no con ict of interest.

Tables
Due to technical limitations, the tables are only available as a download in the supplemental les section. Figure 1