As is well known, reliable markers of chemosensitivity help select patients who most benefit from NCT. Additionally, the early identification of nonresponsive patients will protect these patients from unnecessary toxicity caused by ineffective chemotherapy and will provide alternative effective treatment schemes for tumor biological characteristics. Therefore, finding clinical or molecular markers that can predict the efficacy of NCT and then screening patients who can benefit from chemotherapy have been hot research topics in recent years.
However, identifying reliable predictive factors for NCT in breast cancer remains a challenge. In a previous study, not all patients with HER-2-positive disease were treated according to recent guidelines with the standard agent trastuzumab. Hwang HW et al performed a nomogram to predict the pCR of NCT in breast cancer patients and used a calibration plot to assess the agreement between the predicted and observed probabilities, but they did not use a bootstrap method to validate the model internally or externally . Rouzier et al. performed a nomogram according to clinical characteristics such as tumor size, patient age and hormonal status. However, due to the long history, preoperative chemotherapy regimens have been developed, and this nomogram fails to guide existing clinical treatment. Based on this, researchers are attempting to predict the efficacy of neoadjuvant chemotherapy through a multifactorial digital model. Due to the heterogeneity of the evaluated chemotherapy, the results have been inconsistent [22–25].
According to univariate and multivariate logistic regression analyses, we screened lymphovascular invasion, anemia, ER expression level, Ki67 expression level and NCT regimen as independent predictors, and then we constructed a nomogram to predict the probability of pCR in NCT patients. The AUC of the ROC curve was 0.758, indicating good predictive ability. We used a 5-fold cross validation model in a cohort of patients, and the final prediction accuracy was 76%. According to the Youden index and diagnostic odds ratios, we assigned an optimal cutoff value of 0.32. According to our results, patients with lymphovascular invasion, anemia (HB ≤ 120 g/L), ER positivity, and low Ki67 expression levels (≤ 60) were most likely associated with a lower pCR rate from NCT.
Lymphovascular invasion is an independent prognostic parameter for poor outcome of invasive breast cancer and is the main prerequisite for metastasis[26–27]. A previous study showed that lymphovascular invasion was significantly associated with chemoresistant breast cancer. Another recent study reported that the absence of lymphovascular invasion in post-NCT specimens correlated with pathologic response . Our research previously showed that lymphovascular invasion was related to a low pCR rate in NCT patients, which indicates that lymphovascular invasion may be an important molecular target in breast cancer. In addition, lymphovascular invasion was obtained before NCT treatment in our research, so we could know in advance whether the patient could reach pCR and then decide whether to administer neoadjuvant chemotherapy in these patients.
Previous studies have determined that anemia is an independent prognostic factor that adversely affects the survival of several cancer patients, including breast cancer patients [30–31]. The impaired survival observed among cancer patients with anemia has been mainly due to a decrease in oxygen transport capacity, leading to tumor hypoxia . Previous findings have suggested that anemia may result in worse treatment outcomes from breast cancer chemotherapy . Another study  evaluated the influence of hemoglobin levels in 144 patients receiving chemotherapy and found that hemoglobin levels in patients who responded to treatment (tumor size reduced by more than 50%) were higher than those in patients who did not respond (P < 0.01). The hemoglobin concentration of 12.5 g/dl provided a significant cutoff value below which no reaction was likely to occur. This is the same as our research, in which a HB level ≤ 120 g/L had a lower pCR rate in patients treated with NCT.
Some prospective studies have shown that patients with HR-negative diseases can more often obtain pCR than those with HR-positive diseases [35–36], which is the same as our results. Compared with patients with other subtypes, patients with estrogen receptor-positive tumors have a lower pCR rate from NCT. Most efficacy prediction models established in previous studies have also included the expression of HR. In addition, in previous studies, higher expression of PR was associated with a lower degree of response to adjuvant chemotherapy. In our research, the inclusion or exclusion of PR in multifactor analysis did not significantly affect the prediction accuracy of the model. Additionally, PR was often analyzed in breast cancer tumors but was rarely taken into account. Therefore, the final prediction model did not include PR status.
Tumor cell proliferation indexes, such as baseline Ki67, can be used to predict the efficacy of NCT in breast cancer. Previous studies have shown a significant correlation between gene expression markers related to cell proliferation or genetic grading and chemosensitivity in ER-positive and ER-negative breast cancer. ER-positive breast cancer can be further divided into different subtypes, namely, luminal A and luminal B subtypes, and these two subtypes have different prognoses. A transformation study based on randomized clinical trials also confirmed that the Ki67 index before neoadjuvant chemotherapy was not only a predictor of efficacy but also a prognostic factor of breast cancer. This study confirmed the importance of proliferation-related markers from the perspective of treatment and re-emphasized that the cell proliferation index should be included on the basis of clinical decision-making regarding breast cancer.
It has been well established that pCR varies depending on the treatment regimen and breast cancer subtype. In our research, the pCR rates of HER2-positive and triple-negative tumors were significantly higher than those of luminal B tumors, and the use of TCbH had a higher pCR rate than the use of TEC for chemotherapy. A study  showed that when NCT was combined with trastuzumab, patients with HER-2 overexpression had a higher pCR rate. The use of TCbH has shown encouraging results in the neoadjuvant environment . In HER2-positive breast cancer patients who received chemotherapy combined with trastuzumab, the increase in the pCR rate was directly related to an improvement in the survival rate and a reduction in disease recurrence and death risk [43–45]. As a result, trastuzumab has become an integral part of the global guidelines for breast cancer treatment [46–47].
Based on clinical biological factors, our nomogram has good predictive ability for the early efficacy of NCT for breast cancer. Compared with those of previous similar studies, the advantages of this study are specifically reflected by the following aspects. First, in addition to traditional clinicopathological factors, this study attempted to incorporate important biological factors into the predictive model, such as anemia and lymphovascular invasion. Chemotherapy sensitivity was assessed as accurately as possible by comprehensively evaluating multiple factors related to the efficacy of NCT. Second, in this study, patients received NCT with TEC or TCbH protocols, which complied with the latest version of the NCCN guidelines; the tested information of the enrollment was complete, and the research results were highly applicable and reliable. Finally, the pCR rate of NCT can be known only by routine inspection before operation, so we can quickly estimate the probability of individual NCT benefits and help doctors more effectively make clinical decisions.
Nevertheless, several limitations are worth noting in our present study. First, our observations were limited to retrospective studies from a single center, and our nomogram was not validated in an external cohort. Second, this model could be further improved by expanding its scope of application, such as HER2-positive breast cancer patients using trastuzumab and pertuzumab in combination. Finally, the sample size was relatively small, and the predictive ability of the model needs to be further verified in large-sample studies.