Introduction
Neoadjuvant chemotherapy (NAC) is a standard treatment for advanced breast cancer (BC) patients with the aim to decrease the extent of surgery [
1]. Moreover, it is possible to evaluate the efficacy of NAC in a comparatively short time via therapeutic response, which lets tumor response to chemotherapeutic agents be monitored by this approach [
2]. Based on the 2011 St Gallen consensus, there are four subtypes of BC: luminal, luminal–HER2, HER2-rich, and triple-negative on the basis of the immunohistochemistry results of ER, PR, and HER2 [
3]. Many studies have revealed effective predictors of the response to NAC with different molecular subtypes [
4‐
6], but some of these conclusions remain controversial.
ki67 level was related to tumor cell proliferation, which is the first immunohistochemical (IHC) marker that calls for a precise quantity [
7]. A great number of studies have shown that Ki67 was regarded as the marker that provides prognosis for BC patients who have undergone NAC [
8‐
10], and it has been observed that Ki67 is a factor that can predict the response to NAC [
7,
11,
12]. However, despite the increasing evidence showing the predictive value according to molecular subtype, it is not clear whether Ki67 is identically helpful for predictive approaches in various subtypes, especially the changes of Ki67 during NAC.
In this clinical practice, we have evaluated whether pretreatment Ki67 levels are able to calculate the effectiveness of chemotherapy among molecular subtypes with the same chemotherapy regimen. Furthermore, we investigated whether the change of Ki67 between the needle biopsy and the residual tumor can be used as a predictor for NAC in different subtypes with a relatively large number cohort. In addition, the clinical and pathological response of NAC was also correlated to the conventional clinicopathological factors in this research.
Discussion
NAC, which downstages the disease and reduces tumor volume, has been resulted in improved success rates of breast-conserving operations. Furthermore, the patients who attain pCR after NAC have a very low risk of relapse and death regardless of the earlier stage and molecular subtype [
4,
15]. In the latest years, it has been accepted that BC can be categorized into multiple subtypes by IHC analysis of ER, PR, HER-2, and Ki67, and several studies revealed the response to NAC in different molecular subtypes [
1,
3,
5,
6]. However, some of these conclusions remain controversial [
11,
16].
A great number of studies have demonstrated a positive connection between Ki67 expression and chemotherapy response [
2,
8,
17,
18]. In some studies, Ki67 is thought to predict NAC response only in ER-positive BCs [
2,
18‐
20]. However, some clinical trial with relatively large cases suggested that Ki67 independently improved the prediction of treatment response in luminal tumors as well as triple-negative tumors [
8,
17,
21]. Furthermore, although some studies revealed that Ki67 changes play different predictive roles in ER-positive and ER-negative patients during NAC, the limited number of cases limits further typing analysis [
12,
22].
In this study, we first assessed the usefulness of commonly applied tumor characteristics to forecast clinical response and pCR after NAC. In the clinical response assessment, pretreatment Ki67 (≥ 14%), tumor size ≥ 4 cm, and positive clinical nodal were significantly associated with tumor size reduction, whereas Age, ER, PR, and HER2 status did not show a statistically significant response to NAC. In addition, the menstrual status had marginal P values in this analysis. Furthermore, patients with negative ER, PR, and smaller tumor size (< 4 cm) were more probable to attain pCR, while the Age, HER2 status, menstrual, and clinical nodal status did not show statistically significant response in pathological response assessment. Moreover, the response to NAC of patients with HR 1–10% positive were also evaluated because recent research indicated that weakly HR expression was a poor correlate of luminal subtype [
23], and the therapeutic effect of hormone therapy of patients with HR 1–10% positive was poor [
24]. According to our results, patients with HR 1–10% positive did not demonstrate a significant response to NAC in statistics compared with HR ≥ 10% or HR– patients. However, the number of patients with HR 1–10% positive in our study was relatively small (
n = 31), so it is necessary to further confirm the above relationship through large sample studies.
Most of the above results were in line with previous literature [
12,
16,
19,
25]. However, the HER2 status did not show a statistically significant response in clinical and pathological response in our analysis, which is not consistent with the existing literature. The possible explanation is that only 3% HER2 (+) patients accept trastuzumab before surgery in this study, while this phenomenon also confirms the importance of using trastuzumab in HER2 (+) patients during NAC [
26,
27].
The overall pCR rate in our study was 13%, which was relatively low compared with some large studies (15.8–27.1%) [
9,
28,
29]. We speculated that this phenomenon may be due to relatively short cycles of NAC in this study (89.2% of patients experienced 4 cycles of chemotherapy) compared with the 8 cycles of NAC in other studies [
9,
29]. Another possible reason for this is that the proportion of patients with clinical tumor size ≥ 4 cm in our study reached 41%, which was significantly higher than the 28% of NSABP B-27 trial [
15] because NAC may aid in shrinking the size of the tumor instead of directly leading to pCR status [
30].
Next, we focused on evaluating the predictive role of Ki67 expression in different molecular subtypes for response to NAC. According to our findings, the pretreatment Ki67 could be used as a predictor of NAC only in luminal-type BCs. With a ROC curve, we identified 25.5% as the best cut-off value of pretreatment Ki67 for predicting response to NAC with an optimal sensitivity of 46.6% and a specificity of 69.9%. To assess the impact of Ki67 changes on response to treatment, we first investigated the changes in Ki67 after NAC in different molecular subtypes. As expected, a decrease in Ki67 had been found in almost all molecular subtypes after NAC (the triple-negative type BCs had a borderline significant), which was consistent with previous research [
18]. However, when considering the distribution of responders and non-responders, the reduction in Ki67 may induce a positive response only in luminal and luminal–HER2 subtypes.
Our results show that Ki67 has been demonstrated to forecast NAC response only in ER-positive BCs, which are consistent with several studies [
2,
12,
17,
18,
20]. Furthermore, our research also revealed that the reduction in Ki67 could not be used as a predictor of NAC in HER2-rich and triple-negative subtypes, which was in line with previous study [
12]. It has been shown that the post-treatment, but not pretreatment Ki67 indication levels identify a group of patients at great danger for relapse in patients who do not achieve pCR [
9]. Therefore, patients with HER2-rich and triple-negative subtype who exhibit marked tumor shrinkage in NAC but still have high levels of Ki67 might be good candidates for more aggressive adjuvant systemic treatments, such as extension of postoperative chemotherapy, proper radiotherapy, and administration of Capecitabine, which was recently shown to prolong the survival of non-pCR patients [
31]. Another interesting finding in this study is that the response to NAC in patients with luminal–HER2 subtype can be predicted by the reduction in Ki67, but not by pretreatment Ki67. Possible explanations for this phenomenon are tumor heterogeneity [
32], different signaling pathways caused by chemotherapy [
33], and the presence of more molecular subtypes.
The major limitation of this study is that our data come from a retrospective study in one single center. Therefore, it is necessary to perform prospective study to obtain evidence supporting our results.