ABSTRACT

Background: In the past few decades, due to complications of axillary surgery there has been a drive to de-escalate axillary surgery for secondary involvement of axillary lymph nodes with breast cancer.

Objective: This study aims to assess the diagnostic ability of different imaging modalities for accurately predicting axillary lymph node

involvement in breast cancer with a particular focus on identifying N2 axillary nodal disease from N1.

Methods: A retrospective cohort of all axillary dissections for breast cancer performed in Basildon University Hospital in the last 3 years (2021- 2024) were assessed. Axillary lymph node status on pre-operative imaging was compared with histological lymph node yield. A subgroup of patients without neo-adjuvant chemotherapy was also assessed. The ability of imaging modalities to predict lymph nodes positive for metastasis, especially N2 disease (number of metastatic nodes of more than 3) was assessed using the ROC curve. Sensitivity and specificity were also computed.

Results: There was a total of 133 patients, 131 females (98.5%) and 2 males (1.5%). 36 patients had neoadjuvant chemotherapy (NACT). The false negative rate of identification of any lymph node metastasis was as follows: US (n=40/133, 30.1%), CT (n=48/125, 38.4%), MRI (60/98, 61.2%), and PET CT (n=11/34, 32.3%). The sensitivity for identifying the N2 axillary nodal stage for US, CT, MRI, and PET CT was 36.4%, 48.7%, 17.6%, and 53.3% respectively. The specificity for identifying those without N2 axillary nodal stage for US, CT, MRI, and PET CT was 88.2%, 78.4%, 93.3%, and 60.0% respectively. The area under the curve (AUC) for accuracy in the prediction of N2 nodal disease for US, CT, MRI, and PET CT was 0.62, 0.64, 0.55, and 0.58 respectively.

Conclusion: None of the imaging modalities had high accuracy in the prediction of the N2 axillary nodal stage. CT and PET CT had higher sensitivity in predicting the N2 axillary nodal stage, and MRI scans had the highest specificity in identifying those patients who did not have N2 axillary nodes.

Keywords: Breast cancer, axillary lymph nodes, CT scan, ultrasound for axillary lymph nodes, imaging versus histology for axillary lymph

INTRODUCTION

Breast cancer is one of the most common cancers in females [1], accounting for 15% of cancer-related deaths. Expected 5-year survival is 98% in patients with disease confined to the breast; however, it drops to 85% in those with axillary node involvement [2].

Axillary lymph node (ALN) involvement is the most important factor in predicting survival and risk of recurrence. The nodal status will determine the type of surgery, adjuvant therapy, type of reconstruction, and post-op radiotherapy [1]. Hence ALN status is a key factor for staging breast cancer and for assessing the prognosis and therefore deciding on appropriate treatment [2]. The nodal staging in breast cancer is based on several positive nodes identified in the axilla and abnormal internal mammary nodes. Usually, up to 3 nodes with macro-metastasis in the axilla are regarded

^*Corresponding Author: Nida Bashir, Basildon and Thurrock University

Hospital, Basildon, UK, Email: Nidajillani85@nhs.net

Received: August 25, 2024; Revised: October 16, 2024; Accepted:

October 17, 2024

DOI: https://doi.org/10.37184/nrjp.3007-5181.1.7

as N1 stage and 4 to 9 nodes are considered N2 stage without abnormal nodes in the internal mammary chain, or only the presence of internal mammary nodes in the absence of ALN [2].

Ultrasound has been used as a primary modality to predict nodal status in breast cancer [2, 3]. CT scans and MRIs are advantageous as they are less operator-dependent and can assess axillae with reasonable accuracy irrespective of the patient’s body habitus [4].

Axillary lymph node dissection has been traditionally used as a standard procedure to stage and treat axilla in node-positive breast cancer patients [5]. However, axillary lymph node dissection is associated with significant morbidity: lymphoedema, paraesthesia, and functional problems to name a few [6, 7]. De-escalation of axillary surgery has been a primary focus in breast cancer management [1, 3]. Studies have also shown axillary node clearance to have no further survival benefit in patients with low axillary nodal burden [3]. It is mainly used to stage the axilla to dictate the adjuvant therapy [3].

Advancements in technology have led to an increasing focus on finding nonoperative methods to assess axillary node involvement in cancer, including the use of CT scans and other imaging modalities to predict nodal status in breast cancer [6]. Despite ongoing research in this area, there is still no consensus on the best modality for predicting nodal metastasis in breast cancer. In this study, we compared the diagnostic abilities of ultrasound (US), CT, and MRI with histopathology for accurately predicting axillary lymph node involvement in breast cancer. The study aimed to identify the imaging modality that can be most helpful in identifying patients with N2 axillary nodal disease so that the patients who do not have N2 disease can have de-escalation of oncological therapy and axillary surgery.

METHODS

This is a single-centre retrospective cohort study conducted at Basildon University Hospital, Essex, UK. All patients who underwent axillary lymph node dissection (ALND) for breast cancer from January 1, 2021, to December 31, 2023, were included in this study. Patients who did not undergo ALND were excluded. Data was collected retrospectively from patients’ electronic records. We gathered information on the number of axillary lymph nodes (ALN) predicted by various radiological investigations and compared these predictions with the histologically confirmed lymph nodes post-ALND. Additionally, we included various clinical, biological, and histological parameters such as BMI, smoking status, menopause status, histological tumor type, genetic testing results, receptor status, and details of neo-adjuvant and adjuvant therapies, including hormonal therapy, chemotherapy, and radiotherapy.

Pre-operative prediction of the nodal status of the imaging modality was assessed. The number of positive or abnormal nodes estimated by the imaging modality was recorded in all patients who underwent breast cancer surgery. They either had primary axillary node clearance with breast surgery, wide local excision, or mastectomy, or they had sentinel node biopsy followed by completion axillary node clearance. ALND was used as a standard to assess the ability of imaging modalities to predict abnormal nodes. In patients who had neoadjuvant chemotherapy or neoadjuvant hormone manipulation, the treatment response could be evident in the histology of the axillary nodes in the form of fibrosis. In these patients, the total number of positive nodes after the neoadjuvant therapy included all nodes with macro-metastasis, micro-metastasis, and fibrosis. Descriptive statistics were carried out for different clinical parameters of the sample population. John

Hopkins Medicine ROC curve calculator was used to calculate the accuracy of each imaging modality along with their false negative rates. Diagnostic acuracy parameters including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) were also computed.

RESULTS

The study included a total of 133 patients, comprising 131 females (98.5%) and 2 males (1.5%). The mean age of the study population was 60.3 (SD 14.4) years, and the mean BMI was 29.3 (SD 6.6).

Among the participants, 46 patients (34.7%) were smokers, and 9 patients (6.8%) had a strong family history of breast cancer, with hereditary gene mutations identified in 2 patients (1.5%). There were 101 post- menopausal patients (75.9%). The histological outcomes of the patients in the study are described in Table 1.

Table 1: Histological characteristics of study population. IDC (invasive Ductal carcinoma), ILC (invasive lobular carcinoma), ER (Estrogen receptor), PR (Progesterone receptor).

A total of 36 patients received neoadjuvant chemotherapy (NACT), with 20 (55.6%) showing a partial response and 10 (27.8%) showing a complete response. Similarly,

8 patients received neoadjuvant endocrine therapy (NAET), with 4(50%) having a partial response and 4(50%) achieving a complete response. Post-operative radiotherapy to the breast was administered to 88 patients (66.2%).

A subset of 34 patients underwent sentinel node biopsy. The mean positive node yield was 1.14, and the mean total number of nodes excised was 2.51. In those who had a completed axillary node clearance (ANC), additional positive nodes were found in 10 patients (29.4%), and

3 patients (8.8%) had their nodal staging upgraded

Table 2: Ability of different imaging modalities to differentiate between N1 and N2 nodal disease.

Table 3: Subgroup analysis: Ability of different imaging modalities to differentiate between N1 and N2 disease in non NACT group.

from N1 to N2. The mean number of positive nodes for metastasis in the ANC was 2.9 (SD 3.7), and the mean total number of nodes excised was 12.6 (SD 6.5).

The false negative rate of identification of any lymph node metastasis was as follows: US (n=40/133, 30.1%), CT (n=48/125, 38.4%), MRI (60/98, 61.2%), and

PET CT (n=11/34, 32.3%). The false negative rate of identification of any lymph node metastasis in the patients without NACT was as follows: US (n=29/90, 32.2%), CT (n=29/85, 34.12%), MRI (7/38, 18.4%), and PET CT (n=10/22, 45.4%).

The accuracy of different imaging modalities in differentiating between N1 and N2 is provided in Table 2.

Subgroup analysis was conducted to check the accuracy of different imaging modalities in differentiating between N1 and N2 in patients who did not have NACT. 97 patients did not have NACT. Among this group, 90 patients had US, 77 patients had CT scans, 33 patients had MRI and 22 patients had PET CT. The accuracy

of each imaging modality is shown in Table 3. ROC (Receiver operating characteristic) curves for US, CT, and MRI from our study sample are shown in Figs. (1-3) respectively.

DISCUSSION

This study has aimed to assess the predictive ability of different imaging modalities for axillary nodal staging when compared with histologically confirmed lymph node yield after axillary lymph node dissection. The false negative rate of different imaging modalities to detect axillary macro-metastasis in the axilla was more than 30%. The sensitivity to predict the N2 stage was higher for CT and PET/CT scans. On the other hand, MRI was most specific in identifying patients who did not have N2 axillary nodal metastasis. Similarly, in subgroup

analysis for the patients who did not have NACT, the sensitivity to detect the N2 nodal axillary nodal stage was higher for CT and PET CT. MRI was most specific in detecting those who do not have N2 disease. Overall, the accuracy of most imaging modalities was not higher than 0.70 on the AUC curve.

Few studies have compared radiologically predicted lymph node status with histologically confirmed axillary lymph node (ALN) involvement in breast cancer [8, 9].

A meta-analysis by Morwenn Le Boulc’h et al. reported the following sensitivities and specificities: ultrasound (US) had a sensitivity of 55% and a specificity of 99%; MRI had a sensitivity of 83% and a specificity of 85%;

have N2-3 axillary node metastasis [11]. This can be very important in clinical settings and de-escalated surgical treatment of axilla can only be considered after thorough radiological investigation or wire-guided SLNB [12].

In breast surgery, there have been efforts to avoid axillary node clearance due to its detrimental effects on quality of life (QOL) including chronic pain, numbness of the upper arm, limited mobility of the arm, and other physical, emotional, and sexual aspects, compared to sentinel lymph node biopsy (SLNB) or nonoperative management of the axilla [6, 7]. The ongoing SENOMAC and INSEMA trials provide valuable insights into how axillary surgery affects QOL and support the de- escalation of surgical management of the axilla [13]. Notably, the SENOMAC trial is unique in its inclusion of T3 tumors, whereas other trials primarily focus on early-stage tumors. The oncological outcomes of these trials are still pending [13].

It is important to mention the ACOSOG Z0011 metastatic axillary lymph nodes.

PET had a sensitivity of 49% and a specificity of 94% [10]. This meta-analysis and systematic review did not include node-positive patients or those who received neoadjuvant chemotherapy in their study. The findings of a study by Sameie et al. suggest that none of the imaging modalities are highly accurate in the detection of N2 axillary nodal involvement [11]. MRI scan was highly specific in detecting those patients who did not

trial, which significantly altered the management of metastatic axillary lymph nodes by demonstrating the non-inferiority of sentinel lymph node biopsy (SLNB) compared to axillary lymph node dissection (ALND) for node-negative T1 and T2 breast cancers in terms of overall survival [8]. The trial showed that patients with 1 or 2 positive sentinel lymph nodes could be effectively treated with breast-conserving surgery (BCS) and systemic adjuvant therapy, rather than undergoing ALND. Before this trial, ALND was considered the gold standard for achieving disease control in cases with positive sentinel lymph nodes [8, 9].

The integration of radiomics or artificial intelligence (AI) with imaging and clinicopathological parameters to predict axillary nodal metastasis appears promising and could transform surgical practice in breast cancer [1, 14]. Radiomics, a recent development leveraging AI, extracts quantitative data from traditional imaging modalities— such as ultrasound [14], MRI [15], CT, or PET/CT

[16] and uses this information to create models that objectively quantify lesion heterogeneity, particularly in oncology [14]. Radiomics has been successfully used to preoperatively assess lymph node metastasis in various cancers and shows potential in predicting axillary lymph node involvement in breast cancer, thereby offering insights into prognosis [14, 16].

Neoadjuvant oncological treatment is based on the initial nodal staging predicted by the imaging. If any of the imaging shows N2 disease in the axilla, the patient may be offered aggressive neoadjuvant chemotherapy [17], a higher dose of radiotherapy, or radiotherapy

to the chest wall and regional nodes, which in turn dictates the timing and type of reconstruction offered [18]. Patients with positive axillary nodes post neoadjuvant chemotherapy are treated with ALND or targeted ALND [19]. ATNEC trial is a vital study being conducted in UK. It offers sentinel node biopsy to patients with N1 disease on diagnosis and good response to neoadjuvant chemotherapy. Those with N2 disease are excluded from the study and cannot avoid axillary node clearance.

Our study has several limitations. The study is prone to selection bias as it is a non-randomized cohort study. The effect of neoadjuvant chemotherapy can turn macro-metastasis into fibrosis, however, if there is limited disease or micro-metastasis then this effect may not be conspicuous on histology [20]. It may not be relevant to this study as our study has focused on nodes with macro-metastasis. The number of patients receiving MRI and PET scans was relatively small due to the specialized nature of these investigations, which are typically used after neoadjuvant chemotherapy or for advanced or recurrent cancers. Additionally, this is a retrospective study with a relatively small sample size.

CONCLUSION

None of the imaging modalities were very accurate in predicting N2 nodal staging. The sensitivity to predict the N2 stage was higher for CT and PET/CT scans. On the other hand, MRI was most specific in identifying patients who did not have N2 axillary nodal metastasis.

ETHICS APPROVAL

This study/audit was approved by Audit Department of Mid and South Essex NHS Foundation Trust, registered as GSURG075. All procedures performed in studies involving human participants were following the ethical standards of the institutional and/ or national research committee and the Helsinki Declaration.

CONSENT FOR PUBLICATION

Not applicable.

AVAILABILITY OF DATA

All data is taken from electronic records of patients and is mentioned in the methods.

FUNDING

None.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ACKNOWLEDGEMENTS

Declared none.

AUTHORS’ CONTRIBUTION

Nida  Bashir:  Data  collection,  manuscript  writing,

editing, referencing and correspondence.

Ahsan Rao: Data analysis, results formulation, editing,

supervision and liaison. Danny Fraser: Data collection Fatema Rezai: Data collection Sanjith Navas: Data collection

REFERENCES

DOI: https://doi.org/10.2147/BCTT.S273039

DOI: https://doi.org/10.1148/radiol.2020192534

DOI: https://doi.org/10.1007/s10549-022-06672-7

2000; 39(3): 277-81.

DOI: https://doi.org/10.1080/028418600750013014

DOI: https://doi.org/10.1016/S1470-2045(10)70207-2

DOI: https://doi.org/10.1200/JCO.2010.28.2491

DOI: https://doi.org/10.1053/j.seminoncol.2020.09.001

DOI: https://doi.org/10.1001/jama.2017.11470

DOI: https://doi.org/10.1007/s10549-022-06699-w

DOI: https://doi.org/10.3390/jcm10071543

DOI: https://doi.org/10.1038/s41523-021-00233-9

DOI: https://doi.org/10.1016/j.breast.2022.02.013

DOI: https://doi.org/10.3389/fonc.2023.1217309

DOI: https://doi.org/10.3390/cancers13040757

DOI: https://doi.org/10.3748/wjg.v29.i19.2888

DOI: https://doi.org/10.3390/cancers14051246