Abstract
-
Purpose
- This study aimed to evaluate the impact of postoperative adjuvant chemotherapy (AC) on survival outcomes in breast cancer (BC) patients who have already undergone neoadjuvant chemotherapy (NAC) followed by surgery.
-
Materials and Methods
- Data from a population-based cohort (2010-2020) were analyzed for BC patients treated with NAC and surgery. Univariate and multivariate Cox regression identified prognostic factors for overall survival (OS), and a nomogram was developed and validated. Personalized scores from the nomogram were used for risk stratification to assess the effect of postoperative AC.
-
Results
- A total of 15,921 BC patients were analyzed, with 11,144 in the training cohort and 4,777 in the validation cohort. The key prognostic indicators for OS included age, race, marital status, histological grade, BC subtype, T category, N category, type of surgery, and response to NAC (all p < 0.05). The nomogram effectively predicted individualized OS rates and stratified patients into various risk categories. Postoperative AC was found to significantly enhance OS in the high-risk subgroup (p=0.011 in the training cohort, p=0.012 in the overall population). However, for the low-risk subgroup, there was no significant survival benefit from postoperative AC (p=0.130 for the training cohort, p=0.588 for the overall population), suggesting that some patients might safely forgo unnecessary postoperative AC.
-
Conclusion
- This study efficiently differentiates between varying levels of risk, enabling clinicians to identify patients unlikely to benefit from postoperative AC and thus reduce the likelihood of overtreatment.
-
Key words: Breast neoplasms, Neoadjuvant therapy, Adjuvant chemotherapy, Prognosis, Nomograms, Risk stratification
Introduction
Breast cancer (BC) remains the most common malignancy and a leading cause of cancer-related mortality among women worldwide [1]. Chemotherapy is a cornerstone in the treatment of BC, particularly for patients with aggressive tumor features or those at high risk of recurrence. Neoadjuvant chemotherapy (NAC), administered prior to surgery, aims to shrink tumors, enhance the feasibility of breast-conserving surgery, and improve overall survival (OS) outcomes [2,3]. This approach has gained widespread adoption as it also allows for the assessment of tumor response to chemotherapy, which is crucial in guiding subsequent treatment decisions.
While NAC has demonstrated significant benefits, particularly for patients with locally advanced disease or high-risk early-stage BC [4], the question of whether adjuvant chemotherapy (AC) is necessary after NAC and surgery continues to provoke debate. Some research suggests that patients who achieve a complete pathological response (pCR) after NAC may not derive additional survival benefits from further chemotherapy, potentially sparing them from the adverse effects and financial costs associated with AC [5,6]. Conversely, patients with residual disease after NAC may still be at considerable risk for recurrence, and AC could offer further protection against disease progression [7-9]. The challenge lies in identifying which patients truly benefit from postoperative AC and which patients could safely avoid this additional therapy without compromising their survival outcomes. With the growing emphasis on individualized treatment in oncology, there is a clear need to tailor therapy according to each patient’s unique risk profile, taking into account factors such as tumor size, lymph node involvement, and tumor biology. However, the role of AC following NAC remains a contentious issue, particularly for patients who have shown significant response to neoadjuvant therapy.
The increasing awareness of potential overtreatment in oncology has prompted a reevaluation of AC’s role in these patients. For instance, patients with favorable tumor characteristics or those achieving pCR may have a lower risk of recurrence, raising the question of whether they derive any substantial benefit from additional chemotherapy [10,11]. Balancing the risk of overtreatment with the potential benefits of AC is critical in optimizing patient outcomes and reducing unnecessary toxicity. The evolving paradigm of personalized medicine emphasizes the need to tailor treatments based on individual patient characteristics, including tumor biology, stage, and treatment response. Understanding the interplay between neoadjuvant and adjuvant therapies is crucial for optimizing treatment strategies and enhancing patient outcomes. Current clinical guidelines do not provide definitive recommendations regarding the use of AC following NAC, highlighting the importance of further research to clarify this relationship.
Therefore, in this study, we examined a large cohort of patients who underwent NAC and surgery to comprehensively evaluate clinicopathological characteristics, treatment patterns, and survival outcomes. Additionally, we developed a nomogram to facilitate individualized risk assessment and guide treatment decisions. By identifying prognostic factors and stratifying patients into risk categories, our goal is to offer insights that can inform clinical practice and reduce the risk of overtreatment in low-risk populations.
Materials and Methods
1. Study population
This study analyzed data from the Surveillance, Epidemiology, and End Results (SEER) database, focusing on patients diagnosed with BC who received NAC followed by surgery between 2010 and 2020.
The analysis included demographic information, clinicopathological features, treatment regimens, and survival outcomes. Fig. 1 illustrates the selection process for eligible patients, who met the following criteria: (1) pathologically confirmed BC (ICD-O-C50/3); (2) staging according to the American Joint Committee on Cancer (AJCC) TNM classification, limited to T1-4N0-3M0; (3) NAC and surgery of primary site was performed; (4) availability of clear data on the efficacy of neoadjuvant treatment; and (5) female sex. Patients were excluded if they presented with: (1) duplicate medical records; (2) ambiguous T, N, or M categories; (3) metastatic (M1) disease; or (4) unknown histological grade or BC subtype.
As the data for this population-based analysis were obtained from the publicly accessible database with all patient information anonymized, approval from an institutional ethics committee and written informed consent were not necessary. The work has been reported in line with the STROCSS criteria [12].
2. Variable extraction and definition
Data for this study were drawn from the SEER database, which provided detailed information on key variables such as age at the time of diagnosis, racial and marital status, primary tumor location, histological grade, and staging according to the AJCC system (pathological T category, N category, and TNM classification after NAC). Additionally, information on BC subtypes, the surgical procedures performed, and whether patients underwent radiation or chemotherapy was included. The data on response to neoadjuvant therapy were extracted from the “Response to Neoadjuvant Therapy Recode (2010+)” variable, which categorizes responses as complete response, partial response, or no response. The primary outcome analyzed was OS, measured from the date of diagnosis to the time of death from any cause.
3. Statistical analyses
OS was assessed using the Kaplan-Meier method to generate survival curves, and the log-rank test was employed to evaluate statistical differences between various patient groups. To further explore the impact of various factors on OS, multivariate Cox proportional hazards regression analysis was performed, enabling the identification of independent prognostic variables that influence survival outcomes. These key factors were subsequently used to construct a nomogram aimed at predicting individualized survival probabilities.
The nomogram’s performance and accuracy were rigorously evaluated through internal validation. The model’s discriminative ability was quantified using the concordance index (C-index), which reflects how well the model can differentiate between patients with different survival times. Time-dependent receiver operating characteristic (ROC) curves were generated to assess the model’s predictive accuracy at various time intervals, with the area under the curve (AUC) serving as a comprehensive measure of the model’s performance. Calibration plots were also created to visually compare the predicted survival probabilities with the actual observed outcomes, ensuring the robustness and reliability of the model’s predictions. Additionally, decision curve analysis (DCA) was performed to evaluate the clinical utility of the model by considering the net benefit of the model at different threshold probabilities, providing a more practical and patient-centered assessment of its effectiveness in real-world settings.
All statistical analyses were conducted with a two-sided approach, and a p-value of less than 0.05 was considered statistically significant. Statistical analyses were performed using a combination of software tools, including SPSS ver. 23.0 (IBM Corp.), RStudio ver. 1.4 (RStudio), and X-tile ver. 3.6.1.
Results
1. General characteristics and treatment modalities
In this study, a total of 15,921 women diagnosed with BC were included, with a median age at diagnosis of 54 years. The cohort was randomly assigned to two groups in a 7:3 ratio, yielding 11,144 patients in the training cohort and 4,777 in the validation cohort. The general characteristics and treatment modalities of the cohort are summarized in Table 1.
The majority of patients were White (n=11,958, 75.1%) and married (n=9,550, 60.0%). The most common location for the primary tumor was the upper-outer quadrant of the breast (n=5,598, 35.2%), followed by overlapping breast lesions (n=3,676, 23.1%). Most patients were diagnosed at the T2 category (n=8,105, 50.9%), and lymph node involvement was observed in 59.7% of cases. Additionally, more than half of the patients were classified as histological grade III (n=9,462, 59.4%). Regarding BC subtypes, the predominant category was human epidermal growth factor receptor 2 (HER2)–negative and hormone receptor (HR)–positive (HER2–/HR+), comprising 39.1%, while HER2-negative and hormone receptor-negative (HER2–/HR–) cases accounted for 25.4%. In terms of therapeutic interventions, 60.0% of patients (n=9,555) underwent mastectomy, while the remaining 40.0% received breast-conserving surgery. Radiotherapy was administered to over half of the cohort (61.6% vs. 38.4%), and 51.6% (n=8,217) received postoperative AC.
2. Survival and prognosis in enrolled cohorts
The median follow-up duration was 70 months. For the entire cohort, the median OS was not reached, and the 3- and 5-year OS rates were 89.6% and 82.9%, respectively (S1A Fig.). Survival analysis was conducted to evaluate the impact of various treatment modalities on prognosis, revealing no significant differences between the partial mastectomy plus postoperative AC group and the non-AC group (p=0.256). The 3-year and 5-year OS rates for patients receiving surgery combined with AC were 88.6% and 82.1%, respectively, which were comparable to those of patients undergoing surgery alone (88.6% vs. 90.7%, 82.1% vs. 83.7%) (S1B Fig.).
3. Univariate and multivariable analysis in the training cohort
In the training cohort, univariate and multivariable analyses were conducted on patients treated with NAC combined with surgery, as detailed in Table 2.
The univariate analysis identified several factors significantly associated with OS. Patients older than 60 years exhibited a higher hazard ratio (hazard ratio, 1.521; 95% confidence interval [CI], 1.397 to 1.656, p < 0.001) compared to those aged 60 years or younger. Black patients displayed a significantly increased risk (hazard ratio, 1.487; 95% CI, 1.335 to 1.657; p < 0.001) compared to White patients, while those classified as “Other” had a lower risk (hazard ratio, 0.784; 95% CI, 0.677 to 0.909; p=0.001). Unmarried patients were at an elevated risk of mortality compared to their married counterparts (hazard ratio, 1.360; 95% CI, 1.252 to 1.479; p < 0.001). Higher tumor grades, particularly grade III (hazard ratio, 1.579; 95% CI, 1.297 to 1.924; p < 0.001), and advanced stages (T4: hazard ratio, 3.611; 95% CI, 3.115 to 4.187; p < 0.001; N3: hazard ratio, 5.051; 95% CI, 4.429 to 5.761; p < 0.001) were associated with significantly worse outcomes. Moreover, mastectomy, compared to breast-conserving surgery, was linked to an increased risk of mortality (hazard ratio, 1.801; 95% CI, 1.640 to 1.978; p < 0.001).
In the multivariable Cox regression analysis, many associations remained statistically significant after adjusting for potential confounders. Advanced age continued to predict poorer OS (hazard ratio, 1.408; 95% CI, 1.291 to 1.536; p < 0.001), and Black patients maintained an elevated risk compared to White patients (hazard ratio, 1.322; 95% CI, 1.183 to 1.477; p < 0.001). Unmarried status remained a significant risk factor (hazard ratio, 1.149; 95% CI, 1.055 to 1.252; p=0.001). Tumor grade and stage were also highly prognostic, with Grade III (hazard ratio, 1.961; 95% CI, 1.594 to 2.413; p < 0.001) and T4 category (hazard ratio, 2.145; 95% CI, 1.834 to 2.509; p < 0.001) continuing to predict adverse outcomes. Additionally, N3 category (hazard ratio, 3.637; 95% CI, 3.154 to 4.195; p < 0.001) and mastectomy (hazard ratio, 1.172; 95% CI, 1.058 to 1.298; p=0.002) remained associated with poorer survival, underscoring their independent prognostic significance.
4. Establishment and validation of the nomogram
A nomogram model was constructed to predict 3- and 5-year survival probabilities by incorporating the key prognostic factors identified through multivariable analysis. This visual tool, presented in Fig. 2, allows for individualized survival prediction by integrating various clinical and pathological variables, offering an intuitive method to estimate patient outcomes.
The nomogram underwent rigorous internal validation, demonstrating strong discriminatory ability. This is reflected in the concordance index (C-index) values, where the nomogram achieved a C-index of 0.760, notably higher than the conventional TNM staging system (C-index, 0.642) and the well-established clinical marker pCR (C-index, 0.621). Similarly, in the testing set, the nomogram continued to outperform the TNM stage and pCR, with a C-index of 0.749 compared to 0.630 and 0.620, underscoring its robustness and reliability across different patient subsets. Further, the AUC analysis reinforced the superior predictive performance of the nomogram. In the training cohort, the 3- and 5-year AUC values were 0.798 and 0.775, respectively, substantially higher than the corresponding values for the TNM staging system (0.664 and 0.647) and pCR (0.627 and 0.628) (Fig. 3A and B). The validation cohort mirrored these results, with the nomogram achieving 3- and 5-year AUC values of 0.791 and 0.758, again surpassing both the TNM stage (AUC, 0.646 and 0.632) and pCR (AUC, 0.628 and 0.621) (Fig. 3C and D). These findings emphasize the nomogram’s enhanced ability to discriminate between patients with differing survival probabilities over both short- and long-term follow-ups. In addition to its superior discriminatory power, the nomogram demonstrated excellent accuracy, as reflected in the calibration curves for both the training and validation cohorts. The curves showed a strong alignment with the 45-degree reference line, indicating that the predicted survival probabilities closely matched the actual outcomes (Fig. 3E-H). This suggests that the model is not only capable of accurately predicting survival but also reliably estimates risk in a clinically meaningful manner. Furthermore, the DCA further demonstrates the superiority of the model developed in this study, showing better performance compared to traditional TNM staging and the clinical marker pCR. The DCA reveals that, across various threshold probabilities, the model consistently provides higher net benefits, indicating its enhanced value in clinical decision-making. By evaluating the model’s ability to balance false-positives and false-negatives, we found that it offers a more comprehensive prognostic assessment than relying solely on TNM staging or pCR, thus supporting more individualized treatment strategies in clinical practice. These findings underscore the model’s clinical utility and its advantages over traditional approaches (Fig. 3I and J).
5. Risk stratification and adoption of AC
Using the developed nomogram, prognostic scores were calculated for each patient, with an optimal cutoff established for risk stratification. Patients with BC were classified into distinct risk subgroups: low risk (≤ 275.36) and high risk (> 275.36). Patients in the low-risk subgroup exhibited significantly improved OS compared to those in the high-risk subgroup within the training cohort (p < 0.001) (Fig. 4A). Additionally, Kaplan-Meier curves for the overall population demonstrated significant separation, indicating that our model effectively distinguishes patients with advanced cancer at high risk of mortality (Fig. 4D). The 3- and 5-year OS rates for the low-risk subgroup were 93.1% and 87.2%, respectively, while the corresponding rates for the high-risk subgroup were 59.3% and 45.0%.
As previously noted, the clinical utility of postoperative AC for patients undergoing NAC and surgical intervention remains uncertain, necessitating further discussion regarding the requirement for postoperative AC in all patients. Therefore, additional analyses were performed based on risk stratification to evaluate the clinical significance of postoperative AC in BC patients receiving NAC and surgical therapy. In the training cohort, no significant difference in OS was observed between the low-risk subgroup of patients receiving postoperative AC and those who did not (p=0.130) (Fig. 4B). However, in the high-risk subgroup, a significant difference in OS was noted between patients who received postoperative AC and those who did not (p=0.011) (Fig. 4C). Similar trends were observed in the overall population. In the low-risk subgroup, there was no additional survival benefit from receiving postoperative AC (p=0.588) (Fig. 4E). Conversely, patients receiving AC in the high-risk subgroup exhibited significantly longer OS compared to those without AC (p=0.012) (Fig. 4F). Specifically, patients who received postoperative AC experienced an absolute increase in 3-year OS rates of 7.7% and 5-year OS rates of 7.6%, respectively (3-year OS, 56.0% vs. 63.7%; 5-year OS, 41.7% vs. 49.3%).
Discussion
The current study tackles a significant gap in the management of BC, focusing on the need for AC following NAC. While NAC is widely utilized in clinical practice, the role of AC post-surgery remains inadequately explored [13]. By analyzing data from 15,921 BC patients who underwent NAC followed by surgery, our findings contribute to a more nuanced understanding of personalized treatment strategies. This study is significant as it highlights the balance between avoiding overtreatment and ensuring optimal patient outcomes.
One of the primary strengths of our research lies in the development and implementation of a nomogram, which serves as a practical and individualized tool for clinicians to estimate survival probabilities based on specific patient characteristics. In our study, patients demonstrated a pCR rate of approximately 40%, which aligns with recent trends in BC treatment, where advancements in large-scale studies and newer regimens have contributed to improved response rates. Factors such as population differences, sample selection, and the nature of the SEER database may also influence the reported rates, reflecting ongoing progress in the field. Our study evaluates the performance of the model not only in comparison to the traditional TNM staging but also in relation to pCR, offering a comprehensive assessment of its predictive value. The results indicate that our model performs well across various metrics, including C-index values, ROC curves, AUC, and DCA, supporting the clinical utility of our nomogram. This is primarily due to the fact that pCR measures the response of the primary tumor alone, while cancer often involves systemic disease with micro-metastases. Effective treatment may not always lead to pCR, and pCR achieved through experimental therapies may only occur in patients who would have been cured with standard treatments. By comparing the predictive value of pCR with that of our nomogram, we emphasize the clinical relevance of our model as a more comprehensive tool for assessing patient outcomes.
More importantly, this nomogram aids in effective risk stratification, allowing for more tailored approaches to BC management. High-risk patients—those presenting with larger tumors, extensive lymph node involvement, and aggressive histological features—demonstrated a higher likelihood of recurrence and thus are more likely to benefit from postoperative AC. These results align with previous findings in the literature. For example, studies by von Minckwitz et al. [14] and Spring et al. [15] emphasize that patients with residual disease after NAC have a higher risk of recurrence and may derive substantial benefit from additional AC. Conversely, low-risk patients, particularly those who achieve a CR to NAC or exhibit favorable tumor biology, do not appear to derive significant survival benefits from AC. This observation is supported by Zhang et al. [16], who demonstrated that the omission of AC in low-risk patients following a strong response to NAC did not compromise OS. The adverse effects of chemotherapy, such as gastrointestinal distress, bone marrow suppression, peripheral neuropathy, and cardiotoxicity, may outweigh the minimal potential benefits in this group [17-19]. This highlights the importance of patient selection, as these toxicities can significantly impact the quality of life without offering meaningful improvements in prognosis.
The variation in AC efficacy between high-risk and low-risk groups likely stems from the biological behavior of their tumors [20]. High-risk patients with more aggressive tumor biology often require additional adjuvant therapies to reduce the risk of recurrence. This is consistent with findings from several studies that suggest that triple-negative BC and HER2-positive subtypes are more likely to benefit from intensified treatment regimens [21-23]. Conversely, low-risk patients with indolent tumors may have already achieved optimal treatment outcomes with NAC alone. In this subset of patients, further chemotherapy could represent overtreatment, as supported by findings from studies such as the CREATE-X trial, where those achieving a pathological CR demonstrated excellent outcomes without further systemic therapy [9].
Our results also emphasize the broader implications of incorporating individualized risk assessments into BC management. For high-risk patients, postoperative AC remains a vital treatment component, significantly reducing recurrence risk and improving overall survival. This aligns with the growing body of literature advocating for the use of precision oncology, where treatment decisions are guided by molecular profiling and other individualized assessments [24,25]. However, for low-risk patients, avoiding AC can help reduce treatment-related toxicity, preserve quality of life, and minimize healthcare costs associated with overtreatment. This approach is consistent with current efforts to de-escalate therapy in patients who may not benefit from aggressive treatments [26]. Another critical aspect of our findings is the role of shared decision-making between patients and healthcare providers. Tools like the nomogram can facilitate more informed discussions regarding the risks and benefits of postoperative AC based on an individual’s specific clinical scenario. Studies have shown that shared decision-making in oncology improves patient satisfaction, increases adherence to treatment plans, and enhances the overall quality of care [27]. Engaging patients in these discussions fosters a sense of partnership, leading to better long-term outcomes and higher satisfaction with care. Moreover, involving patients in decisions about their treatment aligns with the principles of personalized medicine, which aims to optimize therapeutic outcomes while minimizing unnecessary interventions. Although our study primarily focused on the role of postoperative AC in patients with BC, we fully recognize the importance of other adjuvant treatment regimens, such as capecitabine for triple-negative breast cancer and T-DM1 for HER2-positive patients, which are widely used in clinical practice and may significantly impact clinical decision-making. Given the potential of these emerging therapies to enhance the model’s comprehensiveness and better address diverse clinical needs, future predictive tools should integrate a broader range of treatment options, helping clinicians make more precise decisions and ultimately improve patient outcomes.
Despite the strengths of this study, several limitations must be acknowledged. Firstly, the retrospective nature of our analysis may introduce biases, particularly in patient selection. The SEER database, while extensive, does not capture all relevant clinical variables, such as detailed molecular characteristics of tumors or specific comorbidities that may influence treatment decisions. Moreover, the database does not include information on the exact regimens used, such as chemotherapy, endocrine therapy, or HER2-targeted therapies, limiting our ability to analyze the outcomes under different treatment protocols. Additionally, our findings may not be generalizable to all patient populations, as the demographics represented in the SEER database may not reflect the broader population of BC patients. Future research should focus on conduct prospective studies that validate our findings in diverse cohorts and explore the underlying biological mechanisms governing treatment responses. Furthermore, incorporating detailed clinical treatment regimen data will provide more precise insights into optimal strategies for different BC subgroups. It would also be beneficial to investigate the cost-effectiveness of personalized treatment strategies in this context.
In conclusion, our study provides valuable insights into the role of postoperative AC in BC patients who have undergone NAC and surgery. The development of a nomogram for individualized risk stratification highlights the potential to enhance clinical decision-making and minimize overtreatment in low-risk patients. As we move towards a more personalized approach in oncology, understanding the nuances of treatment responses will be crucial for optimizing therapeutic strategies and improving patient outcomes.
Electronic Supplementary Material
Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).
NOTES
-
Ethical Statement
Due to all data in our analysis can be searched from public SEER database with patient anonymity, institutional ethics committee approval and written consent were not required.
-
Author Contributions
Conceived and designed the analysis: Zhang D, Zhou Q.
Collected the data: Zhang D, Zhou Q.
Contributed data or analysis tools: Zhang D, Zhou Q.
Performed the analysis: Zhang D, Yang L, Zheng Y, Zhou Q.
Wrote the paper: Zhang D, Yang L, Zheng Y, Zhou Q.
-
Conflicts of Interest
Conflict of interest relevant to this article was not reported.
-
Acknowledgments
The authors acknowledge the open access to the database from SEER.
Fig. 1.The flowchart. AJCC, American Joint Committee on Cancer; BC, breast cancer; ICD, International Classification of Diseases; NAC, neoadjuvant chemotherapy; OS, overall survival; SEER, Surveillance, Epidemiology, and End Results.
Fig. 2.Nomogram in patients with breast cancer undergoing neoadjuvant chemotherapy combined with surgery. OS, overall survival.
Fig. 3.Area under curve and Calibration plot for the prediction of 3- and 5-year overall survival (OS) in the training cohort and the validation cohorts. (A) Receiver operating characteristic (ROC) curves of the nomogram versus TNM staging for 3-year OS prediction in the training cohort. (B) ROC curves of the nomogram versus TNM staging for 5-year OS prediction in the training cohort. (C) ROC curves of the nomogram versus TNM staging for 3-year OS prediction in the validation cohort. (D) ROC curves of the nomogram versus TNM staging for 5-year OS prediction in the validation cohort. (E) Calibration plot for 3-year OS prediction in the training cohort. (F) Calibration plot for 5-year OS prediction in the training cohort. (G) Calibration plot for 3-year OS prediction in the validation cohort. (H) Calibration plot for 5-year OS prediction in the validation cohort. (I, J) Decision curves for the training and validation cohorts. AUC, area under the curve; pCR, pathological response; SEER, Surveillance, Epidemiology, and End Results.
Fig. 4.Overall survival (OS) of patients in different risk subgroups and comparison of OS between postoperative adjuvant chemotherapy (AC) and non-adjuvant chemotherapy (non-AC) groups. (A) Risk stratification of patients in the training cohort based on the nomogram. (B) OS comparison between AC and non-AC in the low-risk subgroup of the training cohort. (C) OS comparison between AC and non-AC in the high-risk subgroup of the training cohort. (D) Risk stratification of patients in the overall cohort based on the nomogram. (E) OS comparison between AC and non-AC in the low-risk subgroup of the overall cohort. (F) OS comparison between AC and non-AC in the high-risk subgroup of the overall cohort.
Table 1.Patient demographics and clinical characteristics
|
Characteristic |
Training cohort (n=11,144) |
Validation cohort (n=4,777) |
Total (n=15,921) |
p-value |
|
Age at diagnosis (yr)
|
|
|
|
|
|
≤ 60 |
7,601 (68.2) |
3,231 (67.6) |
10,832 (68.0) |
0.479 |
|
> 60 |
3,543 (31.8) |
1,546 (32.4) |
5,089 (32.0) |
|
|
Race
|
|
|
|
|
|
White |
8,332 (74.8) |
3,626 (75.9) |
11,958 (75.1) |
0.150 |
|
Black |
1,504 (13.5) |
591 (12.4) |
2,095 (13.2) |
|
|
Other |
1,308 (11.7) |
560 (11.7) |
1,868 (11.7) |
|
|
Marital status
|
|
|
|
|
|
Married |
6,660 (59.8) |
2,890 (60.5) |
9,550 (60.0) |
0.386 |
|
Unmarried |
4,484 (40.2) |
1,887 (39.5) |
6,371 (40.0) |
|
|
Primary site
|
|
|
|
|
|
Upper-outer quadrant of breast |
3,938 (35.3) |
1,660 (34.8) |
5,598 (35.2) |
0.304 |
|
Lower-outer quadrant of breast |
845 (7.6) |
363 (7.6) |
1,208 (7.6) |
|
|
Upper-inner quadrant of breast |
1,244 (11.2) |
531 (11.1) |
1,775 (11.2) |
|
|
Lower-inner quadrant of breast |
521 (4.7) |
256 (5.4) |
777 (4.9) |
|
|
Central portion of breast, nipple |
545 (4.9) |
228 (4.8) |
773 (4.9) |
|
|
Axillary tail of breast |
61 (0.6) |
34 (0.7) |
95 (0.6) |
|
|
Overlapping lesion of breast |
2,546 (22.9) |
1,130 (23.7) |
3,676 (23.1) |
|
|
Breast, NOS |
1,444 (13.0) |
575 (12.0) |
2,019 (12.7) |
|
|
Grade
|
|
|
|
|
|
I |
689 (6.2) |
311 (6.5) |
1,000 (6.3) |
0.685 |
|
II |
3,805 (34.1) |
1,622 (34.0) |
5,427 (34.1) |
|
|
III |
6,630 (59.5) |
2,832 (59.3) |
9,462 (59.4) |
|
|
IV |
20 (0.2) |
12 (0.3) |
32 (0.2) |
|
|
Breast subtype
|
|
|
|
|
|
HER2–/HR+ |
4,349 (39.0) |
1,871 (39.2) |
6,220 (39.1) |
0.811 |
|
HER2+/HR+ |
2,571 (23.1) |
1,073 (22.5) |
3,644 (22.9) |
|
|
HER2+/HR– |
1,401 (12.6) |
619 (13.0) |
2,020 (12.7) |
|
|
HER2–/HR– |
2,823 (25.3) |
1,214 (25.4) |
4,037 (25.4) |
|
|
TNM stage
|
|
|
|
|
|
I |
1,196 (10.7) |
476 (10.0) |
1,672 (10.5) |
0.315 |
|
II |
6,142 (55.1) |
2,638 (55.2) |
8,780 (55.2) |
|
|
III |
3,806 (34.2) |
1,663 (34.8) |
5,469 (34.4) |
|
|
T category
|
|
|
|
|
|
T1 |
2,058 (18.5) |
864 (18.1) |
2,922 (18.4) |
0.755 |
|
T2 |
5,685 (51.0) |
2,420 (50.7) |
8,105 (50.9) |
|
|
T3 |
2,143 (19.2) |
952 (19.9) |
3,095 (19.4) |
|
|
T4 |
1,258 (11.3) |
541 (11.3) |
1,799 (11.3) |
|
|
N category
|
|
|
|
|
|
N0 |
4,500 (40.4) |
1,923 (40.3) |
6,423 (40.3) |
0.285 |
|
N1 |
4,549 (40.8) |
1,994 (41.7) |
6,543 (41.1) |
|
|
N2 |
1,240 (11.1) |
533 (11.2) |
1,773 (11.1) |
|
|
N3 |
855 (7.7) |
327 (6.8) |
1,182 (7.4) |
|
|
Surgery method
|
|
|
|
|
|
Breast-conserving surgery |
4,416 (39.6) |
1,950 (40.8) |
6,366 (40.0) |
0.159 |
|
Mastectomy |
6,728 (60.4) |
2,827 (59.2) |
9,555 (60.0) |
|
|
Radiotherapy
|
|
|
|
|
|
No |
4,315 (38.7) |
1,805 (37.8) |
6,120 (38.4) |
0.266 |
|
Yes |
6,829 (61.3) |
2,972 (62.2) |
9,801 (61.6) |
|
|
Response to NAC
|
|
|
|
|
|
PD |
1,312 (11.8) |
554 (11.6) |
1,866 (11.7) |
0.592 |
|
CR |
4,357 (39.1) |
1,909 (40.0) |
6,266 (39.4) |
|
|
PR |
5,475 (49.1) |
2,314 (48.4) |
7,789 (48.9) |
|
|
AC
|
|
|
|
|
|
No |
5,710 (51.2) |
2,507 (52.5) |
8,217 (51.6) |
0.151 |
|
Yes |
5,434 (48.8) |
2,270 (47.5) |
7,704 (48.4) |
|
Table 2.Univariable and multivariable Cox regression model for overall survival in the training cohort
|
Characteristic |
Univariate
|
Multivariate
|
|
Hazard ratio (95% CI) |
p-value |
Hazard ratio (95% CI) |
p-value |
|
Age at diagnosis (yr)
|
|
|
|
|
|
≤ 60 |
1 |
|
1 |
|
|
> 60 |
1.521 (1.397-1.656) |
< 0.001 |
1.408 (1.291-1.536) |
< 0.001 |
|
Race
|
|
|
|
|
|
White |
1 |
|
1 |
|
|
Black |
1.487 (1.335-1.657) |
< 0.001 |
1.322 (1.183-1.477) |
< 0.001 |
|
Other |
0.784 (0.677-0.909) |
0.001 |
0.805 (0.694-0.933) |
0.004 |
|
Marital status
|
|
|
|
|
|
Married |
1 |
|
1 |
|
|
Unmarried |
1.360 (1.252-1.479) |
< 0.001 |
1.149 (1.055-1.252) |
0.001 |
|
Primary site
|
|
|
|
|
|
Upper-outer quadrant of breast |
1 |
|
|
|
|
Lower-outer quadrant of breast |
0.907 (0.760-1.083) |
0.282 |
|
|
|
Upper-inner quadrant of breast |
0.923 (0.794-1.073) |
0.296 |
|
|
|
Lower-inner quadrant of breast |
1.096 (0.897-1.340) |
0.369 |
|
|
|
Central portion of breast, nipple |
1.115 (0.916-1.358) |
0.279 |
|
|
|
Axillary tail of breast |
0.682 (0.340-1.370) |
0.282 |
|
|
|
Overlapping lesion of breast |
1.003 (0.894-1.124) |
0.963 |
|
|
|
Breast, NOS |
1.464 (1.296-1.654) |
< 0.001 |
|
|
|
Grade
|
|
|
|
|
|
I |
1 |
|
1 |
|
|
II |
1.178 (0.960-1.447) |
0.116 |
1.353 (1.100-1.665) |
0.004 |
|
III |
1.579 (1.297-1.924) |
< 0.001 |
1.961 (1.594-2.413) |
< 0.001 |
|
IV |
1.070 (0.394-2.905) |
0.894 |
0.975 (0.358-2.654) |
0.960 |
|
Breast subtype
|
|
|
|
|
|
HER2–/HR+ |
1 |
|
1 |
|
|
HER2+/HR+ |
0.460 (0.402-0.526) |
< 0.001 |
0.634 (0.552-0.727) |
< 0.001 |
|
HER2+/HR– |
0.627 (0.538-0.731) |
< 0.001 |
0.942 (0.801-1.107) |
0.467 |
|
HER2–/HR– |
1.299 (1.182-1.428) |
< 0.001 |
1.716 (1.542-1.910) |
< 0.001 |
|
T category
|
1 |
|
1 |
|
|
T1 |
|
|
|
|
|
T2 |
1.249 (1.090-1.431) |
0.001 |
1.064 (0.928-1.221) |
0.372 |
|
T3 |
2.089 (1.806-2.416) |
< 0.001 |
1.391 (1.197-1.617) |
< 0.001 |
|
T4 |
3.611 (3.115-4.187) |
< 0.001 |
2.145 (1.834-2.509) |
< 0.001 |
|
N category
|
|
|
|
|
|
N0 |
1 |
|
1 |
|
|
N1 |
1.812 (1.626-2.019) |
< 0.001 |
1.707 (1.526-1.910) |
< 0.001 |
|
N2 |
3.084 (2.706-3.514) |
< 0.001 |
2.208 (1.921-2.539) |
< 0.001 |
|
N3 |
5.051 (4.429-5.761) |
< 0.001 |
3.637 (3.154-4.195) |
< 0.001 |
|
Surgery method
|
|
|
|
|
|
Breast-conserving surgery |
1 |
|
1 |
|
|
Mastectomy |
1.801 (1.640-1.978) |
< 0.001 |
1.172 (1.058-1.298) |
0.002 |
|
Radiotherapy
|
|
|
|
|
|
No |
1 |
|
|
|
|
Yes |
0.987 (0.906-1.075) |
0.764 |
|
|
|
Response to NAC
|
|
|
|
|
|
PD |
1 |
|
|
|
|
CR |
0.192 (0.168-0.220) |
< 0.001 |
1 |
< 0.001 |
|
PR |
0.560 (0.505-0.622) |
< 0.001 |
0.614 (0.552-0.683) |
< 0.001 |
REFERENCES
- 1. Smolarz B, Nowak AZ, Romanowicz H. Breast cancer-epidemiology, classification, pathogenesis and treatment (review of literature). Cancers (Basel). 2022;14:2569.ArticlePubMedPMC
- 2. Teshome M, Hunt KK. Neoadjuvant therapy in the treatment of breast cancer. Surg Oncol Clin N Am. 2014;23:505–23. ArticlePubMedPMC
- 3. Liedtke C, Mazouni C, Hess KR, Andre F, Tordai A, Mejia JA, et al. Response to neoadjuvant therapy and long-term survival in patients with triple-negative breast cancer. J Clin Oncol. 2023;41:1809–15. Article
- 4. Peng J, Hong Y, Chen Q, Xu F, Zhang D, Yao J, et al. Comparison of neoadjuvant chemotherapy response and prognosis between HR-low/HER2-negative BC and TNBC: an exploratory real-world multicentre cohort study. Front Endocrinol (Lausanne). 2024;15:1347762.ArticlePubMedPMC
- 5. Chen AX, Chen X, Li XX, Guo ZY, Cao XC, Wang X, et al. Impacts of tumor stage at diagnosis and adjuvant therapy on long-term survival outcomes in patients with triple-negative breast cancer achieving pathologic complete response after neoadjuvant chemotherapy. Clin Breast Cancer. 2025;25:e30–9. ArticlePubMed
- 6. Foldi J, Rozenblit M, Park TS, Knowlton CA, Golshan M, Moran M, et al. Optimal management for residual disease following neoadjuvant systemic therapy. Curr Treat Options Oncol. 2021;22:79.ArticlePubMedPDF
- 7. Wang X, Wang SS, Huang H, Cai L, Zhao L, Peng RJ, et al. Effect of capecitabine maintenance therapy using lower dosage and higher frequency vs observation on disease-free survival among patients with early-stage triple-negative breast cancer who had received standard treatment: the SYSUCC-001 randomized clinical trial. JAMA. 2021;325:50–8. ArticlePubMedPMC
- 8. Joensuu H, Kellokumpu-Lehtinen PL, Huovinen R, Jukkola-Vuorinen A, Tanner M, Kokko R, et al. Adjuvant capecitabine in combination with docetaxel, epirubicin, and cyclophosphamide for early breast cancer: the randomized clinical FinXX trial. JAMA Oncol. 2017;3:793–800. ArticlePubMedPMC
- 9. Masuda N, Lee SJ, Ohtani S, Im YH, Lee ES, Yokota I, et al. Adjuvant capecitabine for breast cancer after preoperative chemotherapy. N Engl J Med. 2017;376:2147–59. ArticlePubMed
- 10. To YH, Gibbs P, Tie J, Loree J, Glyn T, Degeling K. Circulating tumour DNA guided adjuvant chemotherapy decision making in stage II colon cancer: a clinical vignette study. Cancers (Basel). 2023;15:5227.ArticlePubMedPMC
- 11. Sugawara T, Rodriguez Franco S, Sherman S, Kirsch MJ, Colborn K, Ishida J, et al. Association of adjuvant chemotherapy in patients with resected pancreatic adenocarcinoma after multiagent neoadjuvant chemotherapy. JAMA Oncol. 2023;9:316–23. ArticlePubMedPMC
- 12. Mathew G, Agha R, Albrecht J, Goel P, Mukherjee I, Pai P, et al. STROCSS 2021: strengthening the reporting of cohort, cross-sectional and case-control studies in surgery. Int J Surg. 2021;96:106165.ArticlePubMed
- 13. Krawczyk N, Fehm T, Ruckhaeberle E, Brus L, Kopperschmidt V, Rody A, et al. Post-neoadjuvant treatment in HER2-positive breast cancer: escalation and de-escalation strategies. Cancers (Basel). 2022;14:3002.ArticlePubMedPMC
- 14. von Minckwitz G, Blohmer JU, Costa SD, Denkert C, Eidtmann H, Eiermann W, et al. Response-guided neoadjuvant chemotherapy for breast cancer. J Clin Oncol. 2013;31:3623–30. ArticlePubMed
- 15. Spring LM, Fell G, Arfe A, Sharma C, Greenup R, Reynolds KL, et al. Pathologic complete response after neoadjuvant chemotherapy and impact on breast cancer recurrence and survival: a comprehensive meta-analysis. Clin Cancer Res. 2020;26:2838–48. ArticlePubMedPMCPDF
- 16. Zhang S, Liu Y, Liu X, Liu Y, Zhang J. Prognoses of patients with hormone receptor-positive and human epidermal growth factor receptor 2-negative breast cancer receiving neoadjuvant chemotherapy before surgery: a retrospective analysis. Cancers (Basel). 2023;15:1157.ArticlePubMedPMC
- 17. van Putten M, Lemmens V, van Laarhoven HWM, Pruijt HFM, Nieuwenhuijzen GAP, Verhoeven RHA. Poor compliance with perioperative chemotherapy for resectable gastric cancer and its impact on survival. Eur J Surg Oncol. 2019;45:1926–33. ArticlePubMed
- 18. Yano R, Konno A, Watanabe K, Tsukamoto H, Kayano Y, Ohnaka H, et al. Pharmacoethnicity of docetaxel-induced severe neutropenia: integrated analysis of published phase II and III trials. Int J Clin Oncol. 2013;18:96–104. ArticlePubMedPDF
- 19. Chuah B, Goh BC, Lee SC, Soong R, Lau F, Mulay M, et al. Comparison of the pharmacokinetics and pharmacodynamics of S-1 between Caucasian and East Asian patients. Cancer Sci. 2011;102:478–83. ArticlePubMed
- 20. Carey LA, Dees EC, Sawyer L, Gatti L, Moore DT, Collichio F, et al. The triple negative paradox: primary tumor chemosensitivity of breast cancer subtypes. Clin Cancer Res. 2007;13:2329–34. ArticlePubMedPDF
- 21. Gannon MR, Dodwell D, Miller K, Medina J, Clements K, Horgan K, et al. Survival following adjuvant trastuzumab-based treatment among older patients with HER2-positive early invasive breast cancer: a national population-based cohort study using routine data. Eur J Cancer. 2024;211:114309.ArticlePubMed
- 22. Chen YA, Lai HW, Su HC, Loh EW, Huang TW, Tam KW. Efficacy and safety of adjuvant therapies in older patients with breast cancer: a systematic review and meta-analysis of real-world data. Breast Cancer. 2024;31:739–53. ArticlePubMedPDF
- 23. Leon-Ferre RA, Goetz MP. Advances in systemic therapies for triple negative breast cancer. BMJ. 2023;381:e071674ArticlePubMed
- 24. Pereira B, Chin SF, Rueda OM, Vollan HK, Provenzano E, Bardwell HA, et al. The somatic mutation profiles of 2,433 breast cancers refines their genomic and transcriptomic landscapes. Nat Commun. 2016;7:11479.PubMedPMC
- 25. Anampa J, Makower D, Sparano JA. Progress in adjuvant chemotherapy for breast cancer: an overview. BMC Med. 2015;13:195.ArticlePubMedPMCPDF
- 26. Trapani D, Gandini S, Corti C, Crimini E, Bellerba F, Minchella I, et al. Benefit of adjuvant chemotherapy in patients with lobular breast cancer: a systematic review of the literature and metanalysis. Cancer Treat Rev. 2021;97:102205.ArticlePubMed
- 27. Elwyn G, Frosch D, Thomson R, Joseph-Williams N, Lloyd A, Kinnersley P, et al. Shared decision making: a model for clinical practice. J Gen Intern Med. 2012;27:1361–7. ArticlePubMedPMC
Citations
Citations to this article as recorded by
KCA
, Luo Yang2, Yuan Zheng1, Qi Zhou1
PubReader
ePub Link
Cite
XML Download
