Introduction

Neoadjuvant therapy (NAT) is the standard of care for patients with clinically axillary lymph node-positive breast cancer [1]. For patients with ALN pCR after NAT, omission of ALND can reduce morbidity and complications, such as lymphedema, numbness, axillary web syndrome, and upper-extremity range of motion [2]. Sentinel lymph node biopsy (SLNB) is an alternative surgical method for staging the axilla after NAT in patients with clinically node-positive breast cancer [3]. However, several large prospective trials have shown that the overall false-negative rates (FNR) of SLNB were 12.6% (in the Z1071 trial) to 14.2% (in the SENTINA trial) [4,5,6], which exceeds the clinically accepted cutoff of 10% [7, 8]. Thus, the appropriate use of SLNB in the NAT setting remains controversial [9, 10].

With the evolution of ALN management, medical imaging in the NAT setting has become of great significance. According to American College of Radiology, or ACR, Appropriateness Criteria, the most accurate imaging modality in the evaluation of residual ALN disease after NAT is ultrasound [11]. The morphologic ultrasound features of ALN showing cortical thickness of more than 3 mm, loss of fatty hilum, oval shape, or peripheral nonhilar blood flow were associated with residual ALN disease [12,13,14]. Unfortunately, axillary ultrasound alone cannot accurately predict ALN pCR preoperatively, with a false-negative rate (FNR) of up to 29% [15,16,17]. Promisingly, previous studies reported that combining negative post-NAT AUS findings with SLNB could decrease the FNR, from 12.6 to 9.8% in the Z1071 trial [18] and 8.4 to 2.7% in the SN FNAC trial [19]. However, the rather wide 95%CI in the SN FNAC trial and an FNR close to 10% in the Z1071 trial could not determine whether negative post-NAT AUS decreased the FNR of SLNB, which deserves further study. Furthermore, these studies did not perform subgroup analysis by the number of sentinel lymph nodes (SLNs) removed, molecular subtypes, and clinical N stage. These parameters could be associated with the FNR for SLNB [20,21,22].

This study aimed to further determine whether negative post-NAT AUS could reduce the FNR for SLNB. We also performed subgroup analysis to identify the appropriate patient for AUS combined with SLNB.

Methods

Study design and patients

This study was approved by the ethics committee of Guangdong Provincial People’s Hospital. Informed consent was waived due to the retrospective nature of the study. After a review of the electronic medical record, patients with biopsy-proven lymph node-positive breast cancer who received NAT followed by SLNB and then ALND between July 2014 and July 2021 were initially included. Exclusion criteria were as follows: (1) patients missing clinical information (n = 5); (2) SLNB failure (failure to identify sentinel lymph node, n = 3); (3) patients with prior breast cancer (n = 2); (4) patients treated in other institutions (n = 7) (Fig. 1). The NAT regimens were based on the current National Comprehensive Cancer Network guideline [23].

Fig. 1
figure 1

The flowchart of the study. SLNB sentinel lymph node biopsy; ALND axillary lymph node dissection; NAT Neoadjuvant therapy; AUS axillary ultrasound

Full size image

Axillary ultrasound examination

After NAT and within 1 month before surgery, some of the patients underwent an AUS examination. The AUS evaluation was performed by one of the two sonographers (YL and MX. Y with 5 and 10 years of experience, respectively) by using a 5-18MHZ linear array transducer. The morphologic ultrasound features of ALN showing cortical thickness of more than 3 mm, loss of fatty hilum, oval shape, or peripheral nonhilar blood flow were defined as positive [12,13,14]. ALN was classified as negative if the sonographer did not see any ALN on AUS or judged the ALN was normal in morphologic appearance after NAT.

Pathological evaluation

The status of ALN pCR was determined by surgical pathology within 1 month after NAT, which was defined as a complete absence of micrometastases and macrometastases in ALN. Isolate tumor cells were considered as ALN pCR (ypN0) [24]. Immunohistochemistry (IHC) was used if nodes were negative on hematoxylin and eosin stains.

The status of estrogen receptor (ER), progesterone receptor (PR), HER2, and Ki-67 was determined by IHC. Patients with a Ki-67 proliferation index less 30% were classified as low proliferation, and high proliferation otherwise [25]. The status of ER and PR was regarded as positive if the tumor showed at least 1% of positive cells on nuclear staining [26]. HER2-positive was defined as IHC 3+ or IHC 2+ and amplified by fluorescence in situ hybridization (FISH). HER2-negative was defined as IHC 0 or IHC 1+ or IHC 2+ and FISH-negative [27]. The molecular subtypes were classified as HR-positive/HER2-positive, HR-negative/HER2-positive, HR-positive/HER2-negative, and HR-negative/HER2-negative.

Statistical analysis

Demographic and clinicopathological variables between groups were compared using t test or Wilcoxon rank sum test for continuous variables and a chi-square test or Fisher’s exact test for categorical variables. For the post-NAT AUS alone, a false-negative event was defined as patients with negative nodes in the post-NAT AUS who had a residual disease in either SLNB or ALND, or both. For SLNB alone, a false-negative event was defined as patients with negative sentinel nodes who had a residual disease in ALND. For the combined post-NAT AUS with SLNB, a false-negative event was defined as patients with negative nodes in the post-NAT AUS and SLNB who had a residual disease in ALND. The FNR was calculated as the number of false-negative events divided by the total number of patients with residual disease (in either SLNB or ALND, or both). 95% confidence interval (CI) was calculated using exact (Clopper-Pearson) confidence limits for binominal proportion. Fisher’s exact test was used to compare the FNRs between groups. All P values were two-sided tests, and a P value less than 0.05 was considered significant. The data were analyzed with SPSS version 23.0 (SPSS, Chicago, IL, US) and R software version 3.5.0 (Vienna, Austria).

Results

Patients

After exclusion criteria were applied, a total of 220 patients were finally enrolled in this study (Fig. 1). The overall ALN pCR rate was 45.5% (100/220). ALN pCR rates according to molecular subtypes were in 61.2% (38/62) HR-positive/HER2-positive group, 66.7% (28/42) in HR-negative/HER2-positive group, 51.9% (14/27) in HR-negative/HER2-negative group, and 29.0% (20/69) in HR-positive/HER2-negative group.

Three patients (1.4%) received an anthracycline-based regimen without a taxane, 104 patients (47.2%) received a taxane/anthracycline-based combination, 102 patients (46.4%) received a taxane-based regimen without an anthracycline, and 11 patients (5.0%) received a no taxane/no anthracycline-based regimen. Of the 104 human epidermal growth factor receptor 2 (Her2)-positive patients, 52 patients (50.0%) received trastuzumab, 43 patients (41.3%) received trastuzumab and Pertuzumab, and 9 patients (8.7%) did not receive anti-HER2 regimen because of financial burden.

FNR of SLNB in the entire cohort

The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of SLNB after NAT were 84.2, 100, 100, 84.0, and 91.4%, respectively. Overall, the FNR for SLNB alone was 15.8% (95%CI 9.2 to 22.5%). Among 102 patients with less than three SLNs removed, the FNR was 24.5% (95%CI 12.6 to 36.5%). While the FNR decreased to 9.0% (6/67, 95%CI 1.9 to 16.0%) in patients with at least three SLNs removed (P = 0.02, Supplementary Figure S1).

Comparison of clinicopathological variables

Only the number of sentinel lymph nodes (SLNs) removed, and tumor histology showed significant differences between patients with and without post-NAT AUS results ((P < 0.05, Table 1). In addition, there was a higher proportion of cN2, PR-positive, HER2-negative, and HR-positive/HER2-negative in the positive post-NAT AUS group (P < 0.05, Table 2).

Table 1 Comparison of clinicopathological characteristic between patients with post-NAT AUS available and patients with post-NAT AUS not available.
Full size table
Table 2 Comparison of clinicopathological characteristic between patients with negative AUS and patients with positive AUS
Full size table

Post-NAT AUS results and pathologic ALN status

Post-NAT AUS results were significantly associated with pathologic ALN status (P < 0.001) (Supplementary Figure S2). Patients with positive AUS findings were more likely to have a greater number of positive SLNs and ALNs than those with negative AUS findings (range: 0–10 vs. 0–5 and 0–24 vs. 0–16, respectively, P < 0.001). The sensitivity, specificity, PPV, NPV, and accuracy of post-NAT AUS were 63.3, 84.3, 82.7, 66.0, and 72.9%, respectively. The FNR of AUS was 36.7% (36/98, 95%CI 27.0 to 46.4%) (Supplementary Figure S2).

FNR of SLNB when combined with negative post-NAT AUS results

In this study, using negative post-NAT AUS results to select patients for SLNB, the FNR was 7.5% (95%CI 2.4 to 12.7%). The sensitivity, specificity, PPV, NPV, and accuracy of post-NAT AUS combined with SLNB were 91.8, 84.3, 87.4, 89.7, and 88.4%, respectively. 13 patients with ALN pCR were subjected to an unnecessary ALND. 8 patients with negative SLNB had residual ALN disease who would have been undertreated had they undergone SLNB alone. The majority of patients (n = 160) underwent appropriate axillary surgery (Fig. 2).

Fig. 2
figure 2

The false-negative rate when selection of patients for SLNB based on post-NAT AUS results.NAT neoadjuvant therapy; AUS axillary ultrasound; ALND axillary lymph node dissection; SLNB sentinel lymph node biopsy

Full size image

Subgroup analysis results are shown in Table 3. Although no statistical significance was observed, the FNR was lower for negative post-AUS combined with SLNB when compared with SLNB alone (P > 0.05, Table 3).

Table 3 Subgroups analysis of the FNR when AUS was combined with SLNB
Full size table

Discussion

In this study, we observed that the strategy of combined negative post-NAT AUS and SLNB resulted in an FNR of 7.5%. Subgroup analysis showed that the FNR exceeded the clinically accepted threshold of 10% only in patients with less than three nodes removed, HER2-positive, and cN1 breast cancer. The study demonstrated that negative post-NAT AUS combined with SLNB could improve the ability to accurately restage the axilla.

The most essential finding of the present study was that combining negative post-NAT AUS and SLNB could decrease FNR, which is in agreement with previous studies [18, 19]. However, our work differs from their studies in two ways. First, this is a real-world example of the application of post-NAT AUS to guide SLNB outside a clinical trial. Second, the subgroup analysis of the study may be useful to guide individual treatment. Notably, the confidence intervals in this study are also wide due to a small sample size; but the study is still valuable because ALND for all pre-NAT node-positive patients has largely been abandoned, so it is difficult to replicate these data. Perhaps a meta-analysis is needed at this point. When applying negative post-NAT AUS and SLNB to the patients in our study, 13 patients with ALN pCR were subjected to ALND (false positive events, potentially overtreatment). In addition, 8 patients with negative SLNB had residual ALN disease (false-negative events, potentially undertreatment). The majority of patients (n = 160) underwent appropriate axillary surgery. We also attempt to analyze the results of performing ALND according to SLNB results for patients with positive post-NAT AUS to avoid potential overtreatment. Six patients with negative SLNB had residual ALN disease (false-negative events, potentially undertreatment), and the false-negative rate for AUS combined with SLNB was 14.3(14/98, evidence from Fig. 2). According to previous studies, SLNB alone after NAC is associated with low recurrence for patients with ALN pCR. However, SLNB alone still has an inferior prognosis for patients with residual axillary disease after NAC [28,29,30,31]. Thus, we only propose using negative post-NAT AUS to select patients for SLNB.

We also found that the FNR for negative post-NAT AUS combined with SLNB exceeded the threshold 10% in patients with less than three nodes removed. Previous studies reported a lower FNR when more SLNs were removed, similar with the trends with ours [4, 5]. As the accuracy of any sampling test to a large extent depends on the amount of material sampling, these results were not surprising. Thus, post-NAT AUS has the potential to guide surgeons in accurately staging the axilla by removing at least three SLNs. In addition, the FNRs were different across molecular types, which might be because of the different responses to NAT in different subtypes [32]. Further, in both HR-negative/HER2-positive and HR-positive/HER2-positive subgroups, the FNR also exceeded 10%. The reason may be because HER2-positive subgroup showed a stronger response to NAT than other molecular subtypes. Tumors that respond more strongly may undergo greater changes in the lymphatic drainage pattern [33]. Hence, there is a potential for clinicians to select SLNB alone more confidently in HER2-negative breast cancer with negative post-NAT AUS. Our results also showed that the FNR for negative post-NAT AUS combined with SLNB was highest in the cN1, followed by cN2 and cN3 subgroup. There are several reasons for this. Firstly, the higher ALN burden was more easily detected on AUS [15]. Secondly, the small sample size in the cN2 and cN3 subgroups (36 patients and 16 patients, respectively) is also undoubtedly a factor.

Our results were consistent with the previous studies [4,5,6], in which the FNR for SLNB alone was higher than the predetermined acceptable FNR. Therefore, improved methods for patient selection are needed to guide the use of SLNB. Previous studies reported that targeted axillary dissection and SLNB, dual tracer SLNB, using lymph node examination by immunohistochemistry can decrease the FNR [34, 35], which did not perform in our study. However, we provided a cost-effective and simple method for selecting patients for pursuing SLNB and obtained satisfied results that combined negative post-NAT AUS and SLNB could reduce the FNR and improve patient section for SLNB.

Our study had several limitations. Firstly, this is a retrospective design, causing an inevitable risk of selection bias and confounding. Secondly, we did not use special techniques (such as clipping nodes with I125 radioactive seed and wires) [35] in this study.

Conclusion

Using negative AUS may help to decrease the FNR and improve patient selection for SLNB. Further, post-NAT AUS has the potential to guide surgeons in accurately staging the axilla by removing at least three SLNs or in HER2-negative patients.