INTRODUCTION

Breast cancer (BC) is the second most common malignancy after lung cancer worldwide. Moreover, BC is the most common cause of death due to cancer in women in a certain age group (40-49 years old), so this feature makes BC highly significant among other malignancies. Since most women with BC are non-metastatic at their time of diagnosis, very promising results have been reported with multidisciplinary management.¹ The management of BC depends on many different factors. The age, pathological stage of the BC, the biological characteristics of the BC [such as hormone receptor status, human epidermal growth factor receptor-2 (HER-2) status], and lymphatic or vascular invasion are just some prior determinants for treatment strategy. Axillary lymph node involvement and the number of metastatic lymph nodes (NMLN) were assumed to be the most important prognostic factors in the decision of adjuvant radiotherapy.² Under current clinical guidelines, pathological examination of at least 10 harvested lymph nodes with level 1-2 axillary dissection is accepted for the correct staging of the axillary lymph node stage.³ However, axillary dissection cannot be surgically performed with the same intensity in every patient due to reasons such as the experience of the surgeon or the pathologist, the patient’s age, the patient’s anatomical structure, and/or concomitant diseases. The count of metastatic lymph nodes in the axillary dissection material is divided into three groups according to the N-stage of the American Joint Committee on Cancer (AJCC) 8^th edition staging system. Under this classification, the total count of the harvested lymph nodes or the number of negative lymph nodes (NNLN) is not taken into account. As patients are classified by the number of “positive lymph nodes only”, a heterogeneous group is actually formed. Thus, axillary lymph node staging with the current N classification may change, result in inadequate treatment, and/or may be insufficient in predicting the prognosis.⁴ Therefore, it may be appropriate to consider not only the number of positive lymph nodes, but also the total number of lymph nodes and the number of NNLN in order to determine a more reliable prognosis for BC patients. There are new studies showing that the total lymph node count is a better prognostic parameter than the metastatic lymph node count.^5-7 Also, it has been supported by various studies that the pN classification, which evaluates the ratio lymph node ratio (LNR) of the NMLN to the number of harvested lymph nodes (NHLN), is more successful in determining the prognosis of BC.^4,5,8,9 However, as a limitation of LNR, the prognostic power of this value decreases in those patients with LNR values of “0 or 1”.^10,11 Therefore, the “modified lymph node ratio (mLNR)” was calculated by modifying the LNR classification (by adding 0.5 to both the numerator and the denominator) thus eliminating the possibility of a mLNR value of 0.¹¹

Another more complicated lymph node classification is the “log odds of positive lymph nodes (LODDS)”, calculated as the logarithm of the odds ratio (OR) between positive and NNLN. There are studies in the literature reporting that LNR, mLNR, and LODDS for BC patients have better prognostic value than the pN classification made with only the number of lymph nodes with metastasis.^10-13 There have been different studies investigating the importance of one or more of the NHLN, the number of NNLN, LNR, mLNR and LODDS values in the prognosis of BC. However, the existing literature did not include studies with all these values as covariates in determining the prognosis of operated BC in a large series including different molecular subtypes, and long-term follow-up in our country. Therefore, this study aimed to compare NHLN, NNLN, LNR, mLNR, and LODDS against the standard AJCC N stage for the prognosis of BC patients.

MATERIALS AND METHODS

Design

The study was designed retrospectively. Our study population consisted of patients treated and followed up for breast carcinoma between the years of 1989-2021 in University of Health Sciences Türkiye, İzmir Bozyaka Training and Research Hospital. Permission was obtained from the necessary places for data sharing.

Eligibility Criteria

Patients with bilateral BC, male BC cases, those who received neoadjuvant therapy, those cases without follow-up, or those with missing data were excluded. During the examination of the records, 1,873 patients were investigated. The final number of patients according to the eligibility criteria was determined to be 807 patients. All patients, in the final analysis, underwent breast-conserving surgery or mastectomy with axillary dissection. After surgery, all patients were administered adjuvant chemotherapy and/or radiotherapy and/or endocrine therapy according to NCCN guidelines.

Outcome Parameters

The patient’s demographics, clinical and pathological factors, and treatment modalities (types of surgery, adjuvant therapy, or hormone therapy) were identified. The tumor characteristics including the histologic type of tumor (invasive ductal carcinoma, invasive lobular carcinoma, mixed carcinoma, and special types), the histologic grade, the tumor size, the histologic features, and the presence of lymphovascular (LVI) were determined. Estrogen receptor (ER) and progesterone receptor (PR) status (positive, negative, or unknown), and HER-2 status (positive, negative, or unknown) were determined. Finally, tumor molecular subtypes were classified as luminal A (ER-positive and/or PR positive/HER-2 negative), luminal B (ER-positive and/or PR positive/HER-2 positive), HER-2 overexpressing (ER-negative/PR negative/HER-2 positive) or triple negative (ER-negative/PR negative/HER-2 negative). The pT stages and pN stages were determined according to the TNM classification of the relevant diagnostic pathology report of the AJCC 8^th edition. The NHLN and the number of NNLN were also recorded in detail. The LNR was defined as the ratio of NMLN to NHLN. mLDR was calculated with the formula [LD (+) + 0.5]/[LDT + 0.5]. LODDS was determined by taking the logarithm of the ratio as follows: Log [LD (+) + 0.5]/[LD (-) + 0.5]. Optimal cut-off points were analyzed for all these pN staging parameters, and their sensitivity and specificity were determined. Cancer-related mortalities (CRM) and cancer recurrences during follow-up were determined. Lymph node parameters were compared for disease-free survival (DFS) and overall survival (OS).

Statistical Analysis

Statistical analysis was performed by the SPSS 22.0 program (IBM Corp., Armonk, NY, USA). Nominal variables were compared with the chi-squared test. Scale variables were tested for normality distribution by the Kolmogorov-Smirnov test. Scale variables between two groups were compared using the t-test or the Mann-Whitney U test. Receiver operating characteristic (ROC) analysis was used to determine the significant cut-off value for mLNR. Kaplan-Meier survival analysis with log-rank comparisons was performed in groups consisting of lymph node-related parameters. Also, significant variables for recurrence and cancer-related deaths according to univariate analysis underwent a Cox regression model for DFS and OS.

RESULTS

Eight hundred seven women who underwent surgery for BC were included in this study according to the eligibility criteria. Their mean age was 53.98±13.14 (range: 23-99) years. While the most common BC histological type was invasive carcinoma-NOS (n=554, 69.2%), according to the molecular classification, the patients were mainly in the Luminal A group (n=338, 54.3%). The mean follow-up period was 113.34±74.85 (range: 6-378) months. The histopathological findings of the tumors are given in Table 1. The NHLN was 21.24±9.22, the NMLN was 4.85±7.38, the NNLN was 16.39±9.48, the LNR was 0.23±0.29, the mLNR was 5.38±7.38 and the LODDS was -0.74±0.80 on average (Table 2).

During the follow-up period, 42 (5.2%) patients experienced local recurrence and 188 (23.3%) had distant metastases, resulting in 252 (31.2%) deaths due to BC. The one-year overall survival (OS) rate was 0.984, the 3-year rate was 0.926, the 5-year rate was 0.849, and the 10-year rate was 0.708 for all patients. The one-year DFS rate was 0.981, the 3-year rate was 0.911, the 5-year rate was 0.842, and the 10-year rate was 0.741 for all patients. While the NHLNs in patients with CRM and cancer recurrence were similar, conversely, in those patients who did not develop recurrence and survived, NMLN, LNR, mLNR, and LODDS were found to be significantly lower compared to those with cancer recurrence or CRM (Table 3). In contrast to this, the number of NNLN was found to be significantly lower in those patients with cancer recurrence and CRM (p<0.001). The rate of cancer recurrence and CRM were significantly higher in those patients with mLNR >2.52 [Table 1, OR: 2.55, 95% confidence interval (CI) for OR: 1.84-3.55 and OR: 2.12, 95% CI for OR: 1.55-2.91]. Conversely, cancer recurrence (NNLN >13.0) and CRM (NNLN >15.5) were significantly lower in those patients with NNLN 14 and above (Table 1, OR: 2.40, 95% CI for OR: 1.74-3.32 and OR: 1.81, 95% CI for OR: 1.33-2.48). According to Cox regression analysis, increased NNLN was significantly related to a lower risk of cancer recurrence in non-metastatic patients (TNM N0) (p<0.001, HR: 15.87, 95% CI: 3.78-66.67). Also, in N0 patients, increased NNLN was significantly related to a lower risk of CRM (p<0.017, HR: 3.58, 95% CI: 1.26-10.21). However, in N0 patients, no significant cut-off value was found for cancer recurrence and CRM in the ROC analysis of the NNLN (log-rank: 0.963 and 0.609).

Mastectomy (p<0.001), positive HER-2 (0.028), LVI invasion (p<0.001), advanced T-stage (p<0.001), advanced N-stage (p<0.001), LNR >0.140 (p<0.001), mLNR >2.52 (p<0.001) and LODDS >-0.728 (p<0.001) were significantly related with cancer recurrence. Also, mastectomy (p<0.001), positive PR (0.032), LVI invasion (0.048), advanced T-stage (p<0.001), advanced N-stage (p<0.001), LNR >0.117 (p<0.001), mLNR >2.52 (p<0.001) and LODDS >-0.805 (p<0.001) were significantly related with CRM (Table 4).

According to ROC analysis, LNR, mLNR, and LODDS were found to be significant variables for both cancer recurrence and CRM, but the sensitivity and specificity for the calculated cut-off values were low (Table 5, Figure 1, 2). According to Kaplan-Meier survival analysis, both DFS and OS differed significantly in LNR groups determined according to the cut-off value and four LNRs (p<0.001, Figure 3, 4, 5, 6). Also, the LODDS and mLNR groups determined according to the cut-off value were significant for DFS and OS according to survival analysis (p<0.001, Figure 7, 8, 9, 10). Similarly, AJCC N staging was found to be significant for DFS and OS according to the survey analysis (Figure 11, 12). According to Cox regression analysis, among the lymph node parameters, only LODDS were found to be significant independent risk factors for OS [p=0.014, HR: 3.78, 95% CI for HR: 1.30-10.94, (Table 6)].

DISCUSSION

The AJCC pN staging classification is based only on the absolute number of positive lymph nodes. This classification does not take into account the density of axillary lymph node dissection in surgical dissection, and it is instead dependent on the number of lymph nodes detected in the postoperative axillary dissection material. This is particularly relevant when the number of lymph nodes detected in axillary dissection is very low and thus not able to be accurately evaluated pathologically. However, the lower limit of the number of lymph nodes to be evaluated in axillary lymph node dissection materials is not clearly defined. As a guideline, it is recommended to assess at least 10 dissected lymph nodes for pN staging.¹⁴ A mathematical model of axillary lymph node involvement was tested in a large series of 1,446 patients with invasive BC, and it was shown that pN staging can be achieved with 90% accuracy by evaluating at least 10 lymph nodes which have been dissected.¹⁵ The mean NHLN found in our study was much higher than the number suggested by the literature. NHLN was not a significant factor for cancer recurrence, CRM, DFS, or OS in this study, but as the number of NNLNs increased, recurrence and CRM decreased, and DFS and OS increased. NHLN was more than double the number recommended in the literature, which is sufficient for NMLN and NNLN-dependent staging. For this reason, a classification which includes or combines NMLN and NNLN variables may provide more information.

Additionally, there are also studies investigating the predictive values of the NNLN number in the survival of BC. The predictive value of the intact lymph node count in BC patients remains uncertain.^16,17 In our study, 239 patients did not have lymph node metastases. A significant cut-off value for NNLN could not be determined in these patients, but it was shown that both DFS and OS increased significantly with increasing NNLN. Among all patients, an NNLN of 13 or more for cancer recurrence and 15.5 or more for CRM was determined as a good prognostic factor. Similarly, Kuru¹⁸ indicated that an NNLN number over 15 was significantly associated with a better prognosis. In another study, NNLN was found to affect survival in BC with 4 or more metastatic lymph nodes.¹⁹ In another case series of 455 cases in which the NNLN cut-off value was determined as 5, it was shown that DFS and OS were better in those with NNLN numbers of 5 and above. However, when multivariate analysis was performed in that same study, no difference was found in DFS and OS.²⁰ In our study, only the LODDS variable for OS was found to be significant among the lymph node parameters in multivariate analysis. However, the ACOSOG Z0011 randomized trial demonstrated that the extension of axillary lymph node dissection did not improve the survival of BC patients compared to negative or less than three positive sentinel lymph nodes after surgery. Also, expanded axillary lymph node dissection was recommended to be avoided. They showed that radical axillary lymph node dissection followed by axillary radiotherapy was associated with higher morbidity.²¹

LNR staging is recommended as another lymph node staging in BC patients. There are different reasons for this. Firstly, LNR has been shown to be more advantageous over pN stage, especially in those patients with low NHLN counts.⁵ Another factor is that LNR makes the staging system more comparable between different oncological managements.²² Finally, it partially prevents pN deviations.^23,24 The importance of LNR in BC is increasing currently, but the cut-off values recommended for LNR varies widely in the literature.^25,26 Although LNR is generally divided into groups according to different threshold values in studies, there is no general agreement. The most accepted classification was proposed by Vinh-Hung et al.¹² This recommended classification was based on 1,829 patient results. In that study, LNR rates were divided into three risk groups (low, ≤0.20; intermediate, 0.21-0.65; and high, >0.65). There have been studies using this classification²⁷. In our study, we tested this classification and analyzed a cut-off value in our own population. Both the classification reported in the literature and the dual classification according to the cut-off values determined in this study were significant for DFS and OS. However, the cut-off values in this study were in the low-risk group according to the classification in the literature and were partially compatible (0.140 for recurrence and 0.117 for CRM). A sufficient number of original studies and meta-analysis studies are needed for a universally used LNR classification.

In order to increase the prognostic power of the LNR value over time, especially for those patients with LNRs of “0 or 1”, mLNR has been suggested and it is thought to be more powerful. However, there are limited studies in the literature on this subject. Wen et al.¹¹ recommended groupings as 0.5 and below vs. above 0.5 for the mLNR ratio in a large series (n=3,339). In their study, it was shown that mLNR is an independent parameter in cancer-specific survival and is a much stronger prognostic factor than classical pN staging, especially in those patients with limited lymph node counts. Similarly, the importance of mLNR was supported by a much larger number of BC patients (n=264,096) and two cut-off values for mLNR were recommended in that study; 0.20 and 0.50 were suggested.¹⁰ In the present study, the cut-off value of mLNR was found to be 2.52, with this value being higher than the previously recommended values. Despite the high number of NHLNs in our study, this finding may be due to large number of axillary dissections or heterogeneity in the patient groups.

The LODDS is a similar parameter derived from NMLN and NNLN and it is discussed in the literature with different cut-off values. In the literature, the LODDS classification has been shown to be a convenient prognostic factor in determining survival in different cancers.^28,29 It was also an independent prognostic factor in BC and it was superior to pN staging.¹² In some studies, similar LODDS classifications were used, but their effects on survival were found to be different. This may be due to the small number of BC cases in these studies.³⁰ The cut-off values of LODDS for cancer recurrence (-0.728) and CRM (-0.805) in our study provided a useful distinction for DFS and OS. Also, according to multivariate analysis, LODDS was reported to be an independent risk factor for OS among all lymph node staging systems.

Study Limitations

The present study has a list of limitations which should be considered. Firstly, the retrospective design of this study may have caused data to be lacking. Additionally, the sampling of this study was from a single center which may have caused selection bias despite the large sample size. Also, the time of initial diagnosis of some cases goes back to 1989, so HER-2 status could not be accurately identified in some patients. However, this situation was not an obstacle to our primary purpose in this study. Thirdly, the changes in treatment options over time may have affected outcomes. Therefore, we cannot apply detailed therapy categories to the prognostic models.

CONCLUSION

In the present study, we assessed the survival of BC patients in Türkiye in order to determine different parameters of lymph node status (NHLN, NNLN) and the prognostic value of some different lymph node staging methods (AJCC N-stage, LNR, mLNR, LODDS). Until now, there had been no study comparing the different parameters of lymph node status and the N-stage for predicting BC outcomes with surgery in the Turkish population. The results showed that LODDS seems to be a better option compared to pN classification for OS, which is consistent with previous studies. The present study demonstrated that LODDS has greater usefulness in determining BC patients with distant metastasis compared with the AJCC pN classification.

MAIN POINTS

• In this article study, we assessed the survival of Turkish patients with breast cancer to determine different parameters of lymph node status (NHLN, NNLN) and the prognostic value of different lymph node staging methods (AJCC N-stage, LNR, mLNR, LODDS.

• Until now, no study comparing different parameters of lymph node status and the lymph node staging methods for predicting outcome in BC patients with mastectomy has been reported in Turkish population.

• The results indicated that LODDS is superior to pN classification for OS.

• We can say that the LODDS has obvious advantages in discriminating patients in non-distant metastatic BC compared with the AJCC pN classification.

Ethics

Ethics Committee Approval: The study was designed retrospectively. Permission was obtained from the necessary places for data sharing.

Informed Consent: Informed consent was obtained from each patient.

Peer-review: Externally peer-reviewed.

Authorship Contributions

Surgical and Medical Practices: N.Ö.E., A.İ., B.Z., Concept: N.Ö.E., İ.B., B.Z., Design: N.Ö.E., A.İ., B.Z., Data Collection or Processing: N.Ö.E., A.İ., Analysis or Interpretation: N.Ö.E., İ.B., A.İ., Literature Search: N.Ö.E., İ.B., Writing: N.Ö.E., İ.B.

Conflict of Interest: The authors declared that they have no conflict of interest.

Financial Disclosure: The authors declared that this study received no financial support.