A nomogram for predicting prognosis of internal mammary lymph node metastasis in breast cancer patients with neoadjuvant chemotherapy

Background: Breast cancer patients with internal mammary lymph node (IMLN) metastasis have been shown to have poor prognosis, and the overall survival of patients was signicantly different. The nomogram was developed to predict prognosis of IMLN metastasis in breast cancer patients who received neoadjuvant chemotherapy and divide the patients into different risk groups. Methods: This retrospective study included 218 breast cancer patients with IMLN metastasis who underwent neoadjuvant chemotherapy. Based on the multivariate Cox regression model, we determined independent prognostic factors in breast cancer patients with IMLN metastasis and developed a predictive nomogram using these factors. The discrimination and performance of the nomogram was assessed by C ‐ index, the calibration curves and decision curve analyses. Patients were stratied into the high- risk group and low-risk group based on the best cut-off risk score. The Kaplan-Meier analyses were performed between the two different risk groups to assess the predictive ability of the nomogram. Results: Menopausal status, clinical T stage, pathological Complete Response, axillary lymph node metastasis and Ki-67 were independent prognostic factors in breast cancer patients with IMLN metastasis. These prognostic factors were determined for the development of the nomogram. The C ‐ index of the nomogram was 0.77, which showed the nomogram provided good discernment. The calibration curves demonstrated optimal agreement between prediction by the nomogram and actual observation. The decision curve analyses further demonstrated that the nomogram had the best net benets. The patients with IMLN metastasis were divided into high-risk group (total score ≥ 206.9733) and low-risk group (total score(cid:0)206.9733), and the risk group stratication conrmed that the nomogram had great capacity for distinguishing the prognosis. Conclusions: The nomogram was developed and validated to predict prognosis of IMLN metastasis in breast cancer patients with neoadjuvant chemotherapy. It might help clinicians to identify patients who need individualized treatment and active post-operative surveillance. However, it is necessary to further mine the unknown prognostic HR hazard ratio, SLN supraclavicular lymph node, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple-negative breast cancer, NAC neoadjuvant chemotherapy, BCS breast conserving surgery, BRS breast reconstruction, IMLN Internal mammary lymph node, pCR pathological complete response, ALNM axillary lymph node metastasis.


Background
Breast cancer is the most common cancer in women, and has surpassed lung cancer as the most common cancer worldwide [1]. Internal mammary lymph node (IMLN) and axillary lymph node belong to the " rst station" lymph node of breast cancer lymphatic drainage, and IMLN is also one of the important metastatic pathways of breast cancer [2]. A Meta-analysis of IMLN metastasis in breast cancer found that approximately 23% of patients developed IMLN metastasis [3]. With the development of imaging diagnostics and new diagnostic techniques, the incidence of IMLN metastasis will be further increased.
IMLN metastasis is not only a major component of breast cancer staging, but also an independent prognostic indicator of breast cancer [4]. A study has shown that breast cancer patients with IMLN metastasis have a higher risk of recurrence and a worse prognosis [5]. Although some studies found that the 5-year overall survival (OS) of patients with IMLN metastasis received neoadjuvant chemotherapy (NAC) was 75.5%-84.2%, there were corresponding differences in the prognosis of different patients [6][7][8][9]. However, TNM staging may not be able to predict the prognosis of all breast cancer patients [10], and there is currently no nomogram that can predict long-term survival in patients with IMLN metastasis. In the era of precision medicine, a more effective method to predict the survival of breast cancer patients with IMLN metastasis is needed to identify patients with long-term survival risk so as to provide optimized treatment plans.
Therefore, the aim of the present study was to identify clinicopathologic characteristics and to develop and validate a nomogram to predict the prognosis of patients with IMLN metastasis. The patients are divided into high-risk group and low-risk group by using the nomogram, which has important guiding signi cance for clinicians to accurately evaluate the prognosis of patients and formulate personalized treatment strategies.

Patients
We retrospectively analyzed breast cancer patients who received treatment at The Fourth Hospital of Hebei Medical University between 1 August 2010 and 31 August 2020. Among a total of 18025 breast cancer patients, 279 patients who pathologically con rmed IMLN metastasis without distant metastasis were enrolled. The inclusion criteria were as follows: (a) Eastern Cooperative Oncology Group performance status of 0 or 1; (b) histopathological con rmation of invasive breast cancer before ANC; (c) IMLN metastasis con rmed by histopathology or cytopathology; (d) intact clinicopathological and follow-up data; and (e) patients who received at least two cycles of NAC and surgical treatment and postoperative radiotherapy. The exclusion criteria were as follows: (a) distant metastasis; (b) previous or concurrent other malignant disease; (c) bilateral breast cancer; (d) male patients. 279 patients were screened, and nally, 218 patients ful lling all inclusion and exclusion criteria were included in the study.

Clinicopathologic characteristics
The clinical characteristics include age at diagnosis, menopausal status, tumor location, supraclavicular lymph node (SLN) status and clinical T staging. In this study, all patients were clinically staged according to the 2010 American Joint Committee on Cancer (AJCC) staging system for breast cancer. Clinical T stages were clinically determined according to physical examination and imaging at initial diagnosis. Estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2) and Ki67 status were determined by immunohistochemistry (IHC). Positive staining for ER or PR was de ned as ≥ 1% of tumor cell nuclei staining, HER2 positive was de ned as 3+ receptor overexpression on IHC staining or gene ampli cation on uorescence in situ hybridization. Ki-67 high expression was de ned as Ki-67 levels above 30%, and low expression as levels below 30%. For pathological evaluation of primary lesion and axillary lymph node metastasis after breast cancer NAC, the results of routine para n section examination were used. Pathological Complete Response (pCR) was de ned as the complete disappearance of all invasive tumor cells from breast tissue regardless of the presence of residual ductal carcinoma in situ in the breast (ypT0/is).

Comprehensive Treatment
All patients received at least two cycles of NAC. NAC regimens were as follows: anthracycline plus taxane, anthracycline-based, and taxane-based. Accurately grasped the indications and contraindications of breast conserving surgery (BCS) and breast reconstruction surgery (BRS), patients deemed eligible for BCS and BRS were carefully selected. All patients also underwent axillary lymph node dissection, but IMLN dissection could be selected according to the surgeon's preference and experience. All patients received radiotherapy to the breast or chest wall and regional lymph nodes (axillary node and IMLN) while supraclavicular radiotherapy was performed for patients with SLN metastasis suggested by imaging or pathologically con rmed. All patients who were hormone receptor positive received adjuvant endocrine therapy. All HER2 positive patients received anti-HER2 targeted therapy, except those who refused to apply targeted medicine.

Follow-up
Clinical follow-ups were conducted by medical record review, out-patient medical record review or telephone interview. The last follow-up date was 12 September 2021, and the median follow-up duration was 50.3 (range, 2.6-131.2) months. 205 breast cancer patients received complete follow-up among of 218 patients, with a follow-up rate of 94%. OS was de ned as the period from the date of surgery to the date of death or the last follow-up.

Statistical analysis
Statistical analysis was performed using the SPSS 25.0 (IBM Corporation, Armonk, NY, USA) and R 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). A P-value <0.05 was considered statistically signi cant. All statistical tests were two-sided.
Continuous and categorical variables were presented as mean±standard deviation and counts(percentages), respectively. Multivariate analysis was performed using the Cox regression model for variables which were found to be signi cant in univariate analysis, and the prognostic factors of OS were determined. The nomograms based on the multivariate Cox regression models were developed to estimate 3-year and 5-year OS probabilities. The discriminatory performance of the OS nomogram was evaluated by concordance index (C-index). In addition, the calibration curves were constructed with the bootstrap resampling (1000 bootstrap samples) method and compared whether the prediction probability and observation probability in the queue have good consistency. The clinical value of the predictive models was tested using decision curve analyses (DCAs). The scores of each variable were calculated using the nomogramEx package in R. On the basis of the scores of each variable, the total scores for each patient could be calculated. The "surv_cutpoint" function in R package "survminer" was used to nd the best cut-off risk score. According to the cut-off of the risk-score, patients were assigned to the low-risk and high-risk groups. The Kaplan-Meier method was carried out to produce the survival curves of the two different risk groups.

Results
Clinicopathologic characteristics of patients A total of 218 patients were nally enrolled in this study and the clinicopathologic characteristics of the patients are shown in Table  .008) and > 3 axillary lymph node metastasis (HR=3.336, 95% CI:1.236-9.005, P=0.017) were independent prognostic factors ( Table 2).

Development and validation of the nomogram
Based on the independent risk factors that were statistically signi cant in multivariate analysis, the nomogram was developed (Fig.1). The prediction of 3-year and 5-year OS probability of patients with IMLN metastasis can be calculated by this nomogram.
The C-index of the nomogram was 0.77 (95% CI: 0.73-0.80), which showed the nomogram provided good discernment. Additionally, the acceptable agreement between the predictions of nomogram and actual observations was illustrated by calibration curves for 3-year and 5-year OS ( Fig.2A-B). Therefore, the prediction performance of the nomogram for OS probability showed good predictive value. In this study, the 3-year and 5-year OS DCA curves indicated that the nomogram yielded clinical net bene ts (Fig.3A-B).
Performance of the nomogram in risk strati cation of patients The total score of each patient was calculated based on the nomogram, and patients were divided into two subgroups: low-risk group (total score < 206.9733) and the high-risk group (total score ≥ 206.9733).
In the whole study population, the 5-year survival rates of patients in low-risk group and high-risk group were 82.8% and 43.3% respectively. As shown in Fig.4, OS of the low-risk group and the high-risk group is different (P-value <0.0001).

Discussion
With the continuous development and progress of breast cancer diagnosis and comprehensive treatment, the prognosis of these patients has been signi cantly ameliorated. 4 Although patients with IMLN metastasis are a clinical subgroup with particularly poor prognosis, the 5-year OS has increased to nearly 85.0% with the improvement of comprehensive treatment in recent years. 9 However, patients may still experience early disease recurrence and death. Nomogram that accurately predict OS in breast cancer patients received comprehensive treatment can guide more aggressive surveillance and intensive treatment of high-risk patients. At present, there are still limitations in the accuracy of the TNM staging system. Besides the TNM staging system, the application of nomogram in individual risk prediction has been widely recognized in many cancers [11][12][13][14]. As far as we know, there are no reports on the nomogram predicting prognosis of breast cancer patients with IMLN metastasis. In this study, we attempted to develop and validate a model to predict the prognosis of breast cancer patients with IMLN metastasis.
In multivariate analysis, menopausal status, clinical T staging, Ki67, breast pCR and axillary lymph node metastasis were important independent factors for predicting OS in breast cancer patients with IMLN metastasis. Clinical T staging and the number of axillary lymph node metastasis are important reference indicators for TNM staging to assess the prognosis of breast cancer patients, and are widely used in clinical practice. Previous studies have shown that breast pCR was an important prognostic indicator in breast cancer patients [15][16]. In a population-based study, Nathalie LeVasseur et al. [15] found that early breast cancer patients who received NAC and achieved pCR had better prognosis than non-pCR patients. Similarly, María Gion et al. [16] also suggested that the correlation between pCR and long-term prognosis is strongest for breast cancer patients (TNBC or HER2 positive / hormone receptor negative). In our report, we observed that Ki-67 in breast cancer patients before receiving NAC was highly predictive of OS. OS decreased signi cantly in patients with Ki-67 high expression (HR=2.354, 95% CI: 1.145-4.840, p=0.020). Haeyoung Kim et al. [17] found that high expression of Ki-67 was an important risk factor for DFS (HR=2.8, 95% CI: 1.  [19]. Ki67 may be a reliable marker for optimizing the 8th Edition AJCC breast cancer prognostic staging evaluation system. Although Ki-67 has prognostic signi cance for breast cancer patients, its threshold has great variability and limits the clinical application of Ki-67 [20]. Therefore, it is important to further de ne the threshold that are better correlated with clinical outcomes.
It may be bene cial to predict which patients are at risk of early death in breast cancer with IMLN metastasis. The potential clinical application of this nomogram is a better risk strati cation for breast cancer patients with IMLN metastasis. This enables clinicians to identify high-risk patients who might bene t from more aggressive post-operative surveillance and individualized intensive treatment, and identify low-risk patients who might be safely treated with less intensive treatment regiments.
Currently, there is no consensus on adjuvant chemotherapy for patients with IMLN metastasis who received NAC, regardless of the degree of tumor regression or other clinicopathologic features. Julie Labrosse et al. [21] found that adjuvant chemotherapy after NAC was not bene cial in OS outcomes compared with patients treated with NAC alone. However, Julia foldi et al. [22] believed that non-PCR patients after NAC had a higher risk of recurrence, and these high-risk patients tended to improve their prognosis after adjuvant chemotherapy. The CREATE-X trial randomly assigning 910 patients with (T-DM1) in patients with residual invasive disease after NAC plus HER2-targeted therapy [25]. Therefore, it is of great importance to risk stratify breast cancer patients received NAC and individualize treatment for high-risk patients.
There are several limitations to our study. Firstly, the nomogram was developed and validated based on a retrospective analysis. Due to the inherent biases from the study design, such as selection bias and recall bias, these results need to be further validated by prospective studies. Secondly, some potential predictors are not included in the nomogram. It is necessary to further explore the biomarkers or diagnostic modalities related to the prognosis of patients with IMLN metastasis. Thirdly, due to the data are relatively small and from a single institution, these are some limitations on the applicability and generalizability of the nomogram. In order to con rm the reliability and applicability of the nomogram, it is necessary for more other institution external validations.

Conclusions
This study rst developed and validated the nomogram for predicting OS in patients with IMLN metastasis of breast cancer. Menopausal status, clinical T stage, Ki67, breast pCR and axillary lymph node metastasis were signi cant independent predictors of OS. Our prognostic nomogram accurately and reliably predicts 3-year and 5-year OS of IMLN metastasis in breast cancer patients treated with ANC, which provides an effective clinical tool to identify high-risk patients who may bene t from individualized treatment and more aggressive post-operative surveillance. However, it is necessary to further explore the unknown prognostic factors to optimize the nomogram, and multi-institution external validation is still needed.   Figure 1 The nomograms that are used to predict the 3-year and 5-year OS of breast cancer patients with IMLN metastasis. For each patient, we calculated the points of the clinicopathologic characteristics, and summed up the points to obtain the total points. The predicted the 3-year and 5-year OS can be estimated based on the total points of each patient.