If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for correspondence: Dan-Dan Chen, Shanghai Institute of Biological Products Co.,Ltd., No.350 Anshun Road, Changning District, Shanghai, 200051, China
Triple-negative breast cancer (TNBC) was a particularly aggressive subtype of breast cancer associated with poor prognosis. This retrospective study was conducted to investigate the clinical features, prognostic factors, and benefits of surgery of patients with TNBC.
Methods
From 2010 to 2015, 33654 female patients with TNBC were obtained from the Surveillance, Epidemiology, and End Results (SEER) database. The patients were randomly divided into the training and validation cohorts. Univariate and multivariable cox regression were performed to identify prognostic factors, based on which a nomogram was constructed. Validation of the nomogram was assessed by concordance index (c-index) and calibration curves. Survival curves were plotted according to metastatic burdens and risk groups differentiated by nomogram.
Results
Patients of younger age (<65 years old), white race, married status, lower grade, lower TNM stage and primary tumor surgery tended to have better outcome. The C-index and calibration curves displayed high discrimination in the training and validation sets (C-index 0.794 and 0.793, respectively), indicating suitable external performance of the nomogram model. Patients of bone-only metastases as well as bone and liver metastases showed superior cancer-specific survival (CSS) time if surgery of primary tumor was performed. Besides, patients of all risk groups showed better CSS when receiving surgery.
Conclusion
This study provided population-based prognostic analysis in patients with TNBC and constructed a predicting nomogram with a robust discrimination. The findings of potential benefit of surgery to CSS would shed some lights on the treatment tactics of patients with TNBC.
Breast cancer is the most commonly diagnosed cancer and ranks the fifth cause of cancer death among females worldwide, with an estimated 2.3 million new cases (11.7%) and 685 thousand deaths (6.9%) in 2020.
Triple-negative breast cancer (TNBC), defined by the absence of expression of both estrogen and progesterone receptor (ER/PR), as well as human epidermal growth factor receptor 2 gene (HER2), accounting for 10% to 20% of all invasive breast cancers.
Because TNBC has no explicit molecular markers, targeted treatment of TNBC remains a major challenge in clinical practice. In recent years, great progress has been made in the treatment of TNBC, more patients could get longer and better survival benefits.
Due to the high heterogeneity of TNBC, personalized therapy is more favored with factors such as patients' age, race, surgery and others to be considered.
At present, there are relatively few reports on the prognostic model of TNBC. The purpose of this study was to investigate the prognostic factors of female patients with TNBC, which may more accurately predict the survival of patients with TNBC and provide evidence for individualized treatment.
Materials and Methods
Data Collection
The data were extracted from the Surveillance, Epidemiology, and End Results (SEER) database, which was derived from 18 population-based cancer registries across the United States (U.S.) and covers approximately 27.8% of the U.S. population (based on 2010 census).
A case listing was exported from the SEER database using the SEER Stat v.8.4.0.1 statistical software (Calverton, MD). The female patients diagnosed of TNBC from 2010 to 2015 with active follow-up, valid survival time, known American Joint Committee on Cancer (AJCC) system stage, known histologic grade, known surgery of the primary site, and known cancer-specific of death were included in this study. The patients with unknown age, race or marital status were excluded from the analysis. In addition, the patients diagnosed only in autopsy and death certification were also excluded.
Before initiating this study, we submitted a data-use agreement to the SEER program and were officially granted access to the database. The variables extracted were age at diagnosis (<65 and ≥65 years old), race (white, black, and other), marital status at diagnosis (married and unmarried), derived American Joint Committee on Cancer (AJCC) T stage (T1, T2, T3, and T4), derived AJCC N stage (N0, N1, N2, and N3), derived AJCC M stage (M0, M1), grade (I, II, III, and IV), SEER cause-specific death classification (alive or dead of other cause and dead attributable to this cancer), vital status (alive and dead), survival time, and surgery information of primary site.
Study Design
Using a computer-generated randomization list, the patients who met the inclusion and exclusion criteria were randomly divided into 2 groups with a 1:1 allocation ratio. One group was used to construct the predictive model (training cohort, n = 16,827), while the other group was used to validate the predictive model (validation cohort, n = 16,827) (Supplemental Figure 1). Cancer-specific survival (CSS) was defined as the interval from the date of diagnosis to the date of death due to TNBC.
Statistical Analysis
Statistical analysis was performed using the SAS statistical software version 9.4 (SAS Institute, Cary, NC, USA) and R software version 3.6.1 (Lucent Technologies). The frequency and proportion of the baseline characteristics of the training and validation cohorts were described by Chi-square test or or Fisher's exact test. Univariate cox proportional hazard model was performed with variables including age, race, marital status, AJCC T stage, N stage, M stage, grade, and surgery. Variables with an unadjusted P-value of <.05 were considered as potential risk or protective factors, which were taken into the multivariable cox analysis to determine the independent prognostic factors of patients with TNBC. A nomogram model based on the statistically significant factors in multivariate analysis was plotted to predict a patient of specific characteristic. The total score obtained by adding the corresponding scores of each variable together was used to predict the 3-year and 5-year survival rate in patients with TNBC. A concordance index (c-index) was calculated to evaluate the performance of the nomogram. Besides, the calibration curves (resampling B = 1000) were used to further validate the predictive model internally and externally for the evaluation of consistency between the predicted and actual CSS at 3 and 5 years, respectively. Kaplan-Meier curves and log-rank test were conducted to analyze the CSS in both cohorts. The cutoff values for risk stratification were generated by X-title software version 3.6.1 (Yale University). A 2-tailed P < .05 was considered statistical significant.
Results
Baseline Characteristics
From January 2010 to December 2015, a total of 33,654 female patients with TNBC were included in the final analysis. The demographic and clinical characteristics of training and validation cohorts were outlined in Table 1. Overall, the 2 groups had no significant differences (P > .05) in terms of age, race, marital status, histologic grade, AJCC T stage, N stage, M stage and surgery.
Table 1Demographic and Clinical Characteristics of Patients with TNBC in the Training and Validation Cohorts (N = 33,654)
Using data from 16,827 female patients with TNBC in the training cohort, we explored predictors of death due to TNBC. In univariate analysis, it was found that the patients of older age, black race, unmarried status, more advanced grade tumors, higher T stage, higher N stage, higher M stage and no primary tumor surgery displayed worse CSS (Table 2). These statistically significant factors were incorporated into a multivariate cox regression model. Patients older than 65 years old (HR = 1.48, 95% CI: 1.38-1.59, P < .0001), black race (HR = 1.15, 95% CI: 1.06-1.25, P .0010), more advanced grade (Ⅱ vs. Ⅰ: HR = 2.15, 95% CI: 1.41-3.26, P = .0004; Ⅲ+Ⅳ vs. Ⅰ: HR = 2.43, 95% CI: 1.61-3.67, P < .0001), higher T stage (T2 vs. T1: HR = 2.04, 95% CI: 1.85-2.24, P < .0001; T3 vs. T1: HR = 3.23, 95% CI: 2.86-3.65, P < .0001; T4 vs. T1: HR = 4.00, 95% CI: 3.52-4.56, P < .0001), higher N stage (N1 vs. N0: HR = 1.98, 95% CI: 1.81-2.16,P < .0001; N2 vs. N0: HR = 3.23, 95% CI: 2.88-3.62, P < .0001; N3 vs. N0: HR = 3.29, 95% CI: 2.92-3.72, P < .0001), higher M stage (M1 vs. M0: HR = 3.70, 95% CI: 3.31-4.31, P < .0001) were significantly related to worse CSS. Compared with unmarried patients, married patients showed improved CSS (HR = 0.86, 95% CI = 0.80-0.92, P < .0001). Moreover, primary tumor surgery was associated with better CSS (HR = 0.42, 95% CI = 0.37-0.46, P < .0001).
Table 2Univariate and Multivariate Cox Regression of Cancer-specific Survival of Patients with TNBC
Development and Validation of a 3-Year and 5-Year CSS Predicting Nomogram
On the basis of factors independently associated with CSS, a nomogram, including age, race, marital status, T stage, N stage, M stage, grade, and surgery, was developed to predict a 3-year and 5-year CSS. A total nomogram score was generated for a specific patient, which was corresponded to a predicted 3- and 5-year survival (Figure 1). The nomogram demonstrated medium accuracy in predicting the CSS, with internally validated C-index of 0.794 (95% CI: 0.786-0.802) and externally validated C-index of 0.793 (95% CI: 0.785-0.801), respectively. In addition, the calibration curves suggested that the predictive outcomes had quite good accordance with the actual 3- and 5-year CSS in both cohorts (Figure 2).
Figure 1Nomogram to predict the 3-year and 5-year survival rate in patients with TNBC
Figure 2Calibration curves compare predicted and actual 3-year and 5-year CSS rates in the training set (A-B) and validation set (C-D). Probability of survival based on the nomogram is listed on the x-axis, while the actual probability of survival is listed on the y-axis. The calibration curves suggested that the predictive outcomes had quite good accordance with the actual 3- and 5-year CSS
Kaplan Meier Survival Analysis by Nomogram Risk Category
Overall, a higher nomogram score accounted for a worse CSS in both cohorts (HR = 1.01, 95% CI: 1.01-1.01, P < .0001), when the nomogram score was treated as continuous variable. A risk stratification model was established to calculate the cutoff values by X-tile. Then all the patients were divided into three risk groups: low-risk patients (training 10,548, 62.68%; validation 10,392, 61.76%; total points <=167), intermediate-risk patients (training 4534, 26.94%; validation 4581, 27.22%; total points 168-250), high-risk patients (training 1745, 10.37%; validation 1854, 11.02%; total points >=251). The median CSS of high-risk group were 22 months (95% CI: 21-24) in the training cohort and 23 months (95% CI: 22-25) in the validation cohort, separately (Figure 3).
Figure 3Survival of female patients with TNBC according to different risk groups. (A) CCS in nomogram-based low-, intermediate-and high-risk subgroups in the training set. (B) CSS in nomogram-based low-, intermediate-and high-risk subgroups in the validation set
Benefits of Primary Tumor Surgery in Patients Subdivided by Metastatic Sites and Nomogram Risk Category
In the whole cohort, primary tumor surgery could prolong CSS (HR = 0.17, 95% CI = 0.16-0.18, P < .0001). In this study, there were 1593 (4.7%) women involved in tumor metastasis, with 20.3% (323/1593) bone only metastasis, 1.4% (22/1593) bone and brain metastasis, 6.8% (108/1593) bone and liver metastasis, 6.3% (100/1593) bone and lung metastasis, respectively. In terms of metastatic burden, bone-only metastasis as well as bone and liver metastasis patients could benefit from surgery (bone-only metastasis: HR = 0.46, 95% CI = 0.35-0.59, P < .0001; bone and liver metastasis: HR= 0.53, 95% CI= 0.33-0.84, P < .01) (Figure 4). In addition, we also found that brain-only metastasis, liver-only metastasis, lung-only metastasis, as well as liver and lung metastasis could benefit from surgery (Supplemental Figure 2). The Kaplan-Meier curves showed that surgery of the primary site could prolong CSS in all risk subgroups (low-risk group: HR = 0.14, 95% CI = 0.13-0.15, P < .0001; intermediate-risk group: HR = 0.76, 95% CI = 0.62-0.94, P = .01; high-risk group: HR = 0.59, 95% CI = 0.53-0.67, P < .0001), while low-risk patients tended to get best benefit from surgery (Figure 5).
Figure 4Survival of patients with TNBC in different metastatic burdens according to primary surgery. (A) CSS in patients with bone-only metastasis; (B) CSS in patients with bone and brain metastasis; (C) CSS in patients with bone and liver metastasis; (D) CSS in patients with bone and lung metastasis
Figure 5Survival of patients with TNBC in different nomogram-based risk groups according to primary surgery. (A) CSS in nomogram-based low-risk group; (B) CSS innomogram-based intermediate-risk group; (C) CSS in nomogram-based high-risk group
With great diversity and heterogeneity, the prognosis and treatment tactics of TNBC should be tailored on the basis of their physiological characteristics and clinical features. Currently, nomograms have been proposed as a reliable tool to quantify risk by incorporating and illustrating the important factors for tumor prognosis. An objective nomogram was generated based on SEER database in this study, which could make a more accurate prediction of the 3-year and 5-year CSS of patients with TNBC, being of great significance to clinical decision-making for both doctors and patients.
In this analysis, it found that age <65 years old, white race, married status, lower grade, lower T stage, lower N stage, lower M stage and primary tumor surgery were associated with improved outcomes. Similar results were reported in the previous studies involving patients with TNBC.
combined significant clinical and laboratory parameters together, 7 independent prognostic factors, including family history of breast cancer, tumor location, the number of positive lymph nodes, histological grade, serum CEA, CA125, and CA153 were identified as independent prognostic factors. While in another study held in Guangdong province,
China, stromal tumor-infiltrating lymphocytes (TILs), tumor size, node status, and Ki67 index were significantly associated with disease-free survival and overall survival. In another SEER-based analysis,
it was claimed that age, tumor size, and the number of positive lymph nodes had strong correlations with the probability of death. However, these studies did not incorporate the risk factors of TNM stage and surgery information, therefore, we considered these factors in the construction of nomogram.
The age of patients at the time of diagnosis is an important factor affecting the prognosis of breast cancer.
Our study showed that the prognosis of elderly patients (≥65 years old) was worse than that of younger patients (< 65 years old) (P < .001). While in a previous study of patients with TNBC, patients with age at diagnosis less than 40 years had poorer survival despite more aggressive systemic therapy.
Therefore, if the age was further grouped, there may be subtle alteration of prognosis outcomes for more delicate evaluation and prediction. In addition, similar to previous report,
the African American women with TNBC had worse clinical outcomes than white women, which may contribute to socioeconomic status, geographical environment, genomics, proteomics, epigenetics, and other factors. However, the white women did have a worse outcome when compared with other races. Similarly, Parise et al. also reported that the reduced risk of TNBC was seen in Asian/Pacific Islander (OR = 0.84; 95% CI = 0.79-0.90) compared to whites.
Limited data suggested genetic differences in immune responses by race, which favor a stronger Thr type 2 (Th2) immune response leading to dysregulated immunity among African American women than white women.
found that the mutational landscape of TNBC is similar between African Americans and Caucasians. Basically, further exploration was needed to systematically examine racial differences in molecular characteristics.
Marital status was also an important prognostic factor for patients with TNBC. Our study showed that married women had lower risk for mortality of TNBC (HR: 0.86, 95% CI: 0.80-0.92). Previous studies had confirmed that married patients could obtain more mental and financial support, have better compliance, which could help them to be diagnosed at an early stage and get more appropriate treatments, ultimately prolonging their survival.
A recent study demonstrated that marital status had no influence on risk of mortality for either black or Hispanic women, and only white and Asian/Pacific Islanders women with TNBC have a marital advantage.
Predicting the survival of triple-negative breast cancer in different stages: a SEER population based research referring to clinicopathological factors.
marital status may have different effects on the prognosis of patients with TNBC in each stage, being meaningless in stage III.
It was concluded that histological grade and TNM stage were the critical factors negatively correlated with the prognosis of patients with TNBC in our study. Notably, determining the TNBC phase was very important for clinical diagnosis and treatment. In general, the more advanced histological grade and TNM stage, the poorer the prognosis of the patients. These were the essential indicators for prognostic evaluation of TNBC, which provided an objective direction for the treatment. For metastatic burden, the current study indicated that apart from patients with bone and brain metastasis as well as bone and lung metastasis, patients of other metastatic patterns could benefit from surgery. Actually, bone, lung, liver, and brain were generally accepted as the preferred primary target sites of breast cancer metastasis.
Of these, brain metastasis was leading cause of death due to the limited permeability of the blood-brain barrier and consequent lack of effective treatments.
Given the fact that lung metastasis usually elicited little or no symptoms until the lungs had been vastly occupied with metastatic tumor masses, it was of paramount importance to early diagnoses of breast cancer by virtue of advanced technologies such as detection of circulating tumor cells (CTCs).
Progress in the treatment of TNBC remained an important challenge. Our study showed that patients undergoing surgical treatment at primary site had prolonged survival. There were many surgical methods for breast cancer, but the trend was focused on minimally invasive therapy and aesthetics to achieve minimal wound and aesthetic effect, with breast conserving surgery (BCS) being the most representative.
While Lironne Wein and Sherene Loi pointed out that chromosomal instability, genetic heterogeneity, neoadjuvant chemotherapy, immune microenvironment, the Ras-MAPK pathway and immune evasion may lead to chemotherapy resistance.
In recent years, the antibody-drug conjugates (ADCs) designed based on new concepts were not only antibody-targeted, but also maintained the high antitumor activity of cytotoxic drugs, which brought new treatment options to patients with TNBC.
As a retrospective study based on the database, there are several limitations though. Firstly, the SEER database covering about 30% of USA population represents a general situation, while it may be immature for application on Chinese population considering ethics difference. Secondly, due to much unknown status of radiation and chemotherapy in SEER database, these 2 factors are not considered in this study, which may greatly affect the applicability of the study in real-world cases. Thirdly, given that the diagnosis and treatment of TNBC is a multi-stage and systemic process, if other confounding prognostic factors like tumor markers, family history, menstrual history, fertility, genome status could be included, the prediction accuracy of the model may be further improved. Although a validation cohort is used to test and verify the prediction of nomogram, external validation outside the SEER database is needed to further enforce the reliability.
Conclusion
In conclusion, the current study identified the potential prognostic factors in predicting survival in female patients with TNBC. The constructed nomogram provided a quantitative evidence support for survival prediction in different risk subgroups, which would help clinicians with reasonable risk management and to choose long-term survival strategies for patients with TNBC.
Clinical Practice Points
•
Triple-negative breast cancer (TNBC) accounts for approximately 15% to 20% of the breast cancer cases with a higher recurrence rate and a poorer prognosis. Due to the high heterogeneity of TNBC, targeted treatment of TNBC remains a major challenge in clinical practice, and personalized therapy is more favored. Therefore, our study constructed a predictive nomogram based on age, race, marital status, TNM stage, grade, and surgery, which was proved to have good accuracy. Besides, patients of all risk groups showed better cancer-specific survival (CSS) when receiving surgery. The findings of this study could provide guidance on the personalized therapy of patients with TNBC.
Disclosure
The authors declare no conflict of interest.
Data Availability Statement
The datasets generated and/or analyzed during the current study are available in the SEER repository (https://seer.cancer.gov/).
Author Contributions
H-LZ and D-DC contributed to the study's conception and design. H-LZ and D-DC carried-out the study. H-LZ analyzed the data. H-LZ wrote the paper. D-DC provided important suggestions in analysis and the writing of the paper.
Ethics approval and consent to participate
Not applicable.
Patient consent for publication
Not applicable.
Acknowledgments
This work was supported by the Science and Technology Innovation Plan of Shanghai Science and Technology Commission [grant numbers 21S11909900].
Predicting the survival of triple-negative breast cancer in different stages: a SEER population based research referring to clinicopathological factors.
00001-0&id=gr1.jpg){kind=link}
00001-0&id=gr2.jpg){kind=link}
00001-0&id=gr3.jpg){kind=link}
00001-0&id=gr4.jpg){kind=link}
00001-0&id=gr5.jpg){kind=link}