Objective To construct a predictive model for pathological complete response (pCR) following neoadjuvant chemotherapy (NAC) for Human epidermal growth factor receptor 2 (HER2)-positive breast cancer and to analyze the relationship between pCR and prognosis. Methods A total of 183 HER2-positive breast cancer patients were retrospectively categorized into a case group (85 cases achieved pCR following NAC treatment) and a control group (98 cases did not achieve pCR following NAC treatment). The Student's t-test, Mann–Whitney U test, and Chi-square test were employed to contrast the general clinicopathological data of the two groups. Multivariate logistic regression was utilized to analyze the independent influences on pCR following NAC and to develop a predictive model for predicting pCR. The Hosmer–Lemeshow test and receiver operating characteristic curve (ROC) were used to assess the predictive effectiveness of the prediction model and to externally validate the model. The 3-year disease-free survival (DFS) of the two groups was analyzed using the Kaplan–Meier method with log-rank tests. Results There were statistically significant differences in chemotherapy regimen, tumor cT and cT grade, tumor Nottingham grade, Ki-67 percentage and menstruation status between the two groups of patients (all P < 0.05). Multivariate analysis identified cT stage (Ⅳ vs. Ⅱ, OR = 0.139, P = 0.021), cN stage (Ⅰ vs. 0, OR = 0.368, P = 0.038), cN stage (Ⅱ vs. 0, OR = 0.278, P = 0.015), cN stage (Ⅲ vs. 0, OR = 0.148, P = 0.012), Nottingham stage (Ⅲ vs. Ⅰ, OR = 9.894, P = 0.012) and Ki-67(%) (OR = 1.022, P = 0.020) were independent predictive factors of pCR. The expression of the established prediction model was Logit (P) = − 2.611–0.691X1−1.932X2−1.011X3−1.231X4−2.012X5 + 1.695X6 + 2.295X7 + 0.022X8 + 0.669X9. The P value for the predictive model Hosmer–Lemeshow test was 0.728, the area under the ROC curve of the predictive model was 0.809 (95% CI: 0.750–0.869). The external validation results show that the predictive model has an area under the ROC curve of 0.820 (95% CI: 0.770–0.871), a sensitivity of 82.4%, a specificity of 69.4%, and a Yoden index of 0.518. There was a statistically significant difference in the 3-year DFS rate between case and control groups (82.4% vs. 69.4%, P = 0.039). Conclusion The logistic regression prediction model demonstrates high predictive efficacy and is a valuable tool for clinicians to predict the likelihood of achieving pCR following NAC in patients with HER2-positive breast cancer. HER2-positive breast cancer who achieved pCR following NAC had a longer DFS than those who did not, and pCR could be an independent predictor of a favorable prognosis in HER2-positive breast cancer.

Construction of a Predictive Model for Pathological Complete Response Following Neoadjuvant Chemotherapy for HER2-Positive Breast Cancer and Analysis of the Correlation Between Pathological Complete Response and Prognosis

期刊论文
发表日期：

2025年9月18日

查看全文

Yongpeng Ouyang,
Na Tang,
Kunjian Xia

摘要

Objective

To construct a predictive model for pathological complete response (pCR) following neoadjuvant chemotherapy (NAC) for Human epidermal growth factor receptor 2 (HER2)-positive breast cancer and to analyze the relationship between pCR and prognosis.

Methods

A total of 183 HER2-positive breast cancer patients were retrospectively categorized into a case group (85 cases achieved pCR following NAC treatment) and a control group (98 cases did not achieve pCR following NAC treatment). The Student's t-test, Mann–Whitney U test, and Chi-square test were employed to contrast the general clinicopathological data of the two groups. Multivariate logistic regression was utilized to analyze the independent influences on pCR following NAC and to develop a predictive model for predicting pCR. The Hosmer–Lemeshow test and receiver operating characteristic curve (ROC) were used to assess the predictive effectiveness of the prediction model and to externally validate the model. The 3-year disease-free survival (DFS) of the two groups was analyzed using the Kaplan–Meier method with log-rank tests.

Results

There were statistically significant differences in chemotherapy regimen, tumor cT and cT grade, tumor Nottingham grade, Ki-67 percentage and menstruation status between the two groups of patients (all P < 0.05). Multivariate analysis identified cT stage (Ⅳ vs. Ⅱ, OR = 0.139, P = 0.021), cN stage (Ⅰ vs. 0, OR = 0.368, P = 0.038), cN stage (Ⅱ vs. 0, OR = 0.278, P = 0.015), cN stage (Ⅲ vs. 0, OR = 0.148, P = 0.012), Nottingham stage (Ⅲ vs. Ⅰ, OR = 9.894, P = 0.012) and Ki-67(%) (OR = 1.022, P = 0.020) were independent predictive factors of pCR. The expression of the established prediction model was Logit (P) = − 2.611–0.691X₁−1.932X₂−1.011X₃−1.231X₄−2.012X₅ + 1.695X₆ + 2.295X₇ + 0.022X₈ + 0.669X₉. The P value for the predictive model Hosmer–Lemeshow test was 0.728, the area under the ROC curve of the predictive model was 0.809 (95% CI: 0.750–0.869). The external validation results show that the predictive model has an area under the ROC curve of 0.820 (95% CI: 0.770–0.871), a sensitivity of 82.4%, a specificity of 69.4%, and a Yoden index of 0.518. There was a statistically significant difference in the 3-year DFS rate between case and control groups (82.4% vs. 69.4%, P = 0.039).

Conclusion

The logistic regression prediction model demonstrates high predictive efficacy and is a valuable tool for clinicians to predict the likelihood of achieving pCR following NAC in patients with HER2-positive breast cancer. HER2-positive breast cancer who achieved pCR following NAC had a longer DFS than those who did not, and pCR could be an independent predictor of a favorable prognosis in HER2-positive breast cancer.