Clinical and Pathological Characteristics of Young Female Breast Cancer

This is an single-center, retrospective cohort study which adopted 2,142 women ≤ 41 years who were diagnosed with stage I-III invasive breast cancer. Patients were grouped into hormone receptor-positive and -negative groups. Variance of common characteristics between the two groups were compared using Chi-square test, Fisher-exact test and Wilcoxon rank sum test. Cox proportional hazards regression was employed for survival estimation and Kaplan–Meier curves were used to graphically present the survival data. Propensity score matching was used to balanced covariates between patients who received or not received the same treatment. is gradually increasing, and these women have worse survival outcomes than their elder counterparts. HR-positive disease predicted worse long-term survival outcomes. Adjuvant chemotherapy and adjuvant radiotherapy were required for all of the young patients. Young women with HR-positive disease can benet from endocrine therapy. No clear benet was seen from neoadjuvant chemotherapy in young women with early-stage breast cancer.


Introduction
Breast and cervical cancer have become the two most malignant tumors with the highest incidence in young adult women [1]. According to the latest breast cancer (BC) statistics data from America, the estimated BC incidence in females < 40 years old is approximately 4%, while the mortality rate is 3% [2].
The proportion of BC patients younger than 35 years was reported even up to 9.5-12% in Asian and has continually increased [3]. The occurrence of BC in patients younger than 40 years in China is 15.2%, much higher than in America, which is 5.6%, according to 2012 data of GLOBOCAN (http://globocan.iarc.fr/old/age-speci c_table_r.asp). As the optimization of diagnosis and treatment strategies increases, overall mortality from BC has decreased rapidly. However, the rate of decline of mortality in patients younger than 50 years old has decreased compared with female patients older than 50 since 2007 [4]. Young BC patients have a longer life expectancy after diagnosis, and their requirement for childbearing, work, and social intercourse is higher.
Previous studies reported that BC in young patients demonstrates a higher tumor histological grade and a higher proportion of triple negative and Human epidermal growth factor receptor 2 (HER2) positive subtype, resulting in a worse prognosis [5]. Differences in survival outcomes between young and older patients are more signi cant in the luminal A subtype, which may be caused by lower compliance to endocrine therapy in young women [6]. In addition, fewer young patients undergo menopause after chemotherapy, leading to a higher recurrence risk [7]. HR-positive expression had been reported as an independent risk factor for BC-speci c survival (BCSS) in premenopausal females with advanced BC [8].
Additionally, a higher proportion of endocrine therapy resistance has been reported in young BC patients [9]. The incidence of de novo metastatic breast cancer (dnMBC) is approximately 2.6%, and the metastasis rate of primary early-stage or locally advanced BC is 27.1% in young patients [10]. Research validating the Oncotype DX Recurrence Score enrolled only 4.4% patients younger than 41 years [11], and a study on the 70-gene signature enrolled 1.8% patients under age 35 [12]. Resulting relatively insu cient evidence to extend the outcomes to young patients.
We enrolled young females (< 41 years) diagnosed with BC from a breast cancer registration database. We hypothesized that HR status plays a key role in impacting the prognosis of young women with BC. We aimed to identify epidemiologic and pathological characteristics of BC occurring in young women who were diagnosed during different time periods. Survival outcomes and effectiveness of different treatment patterns were analyzed to identify prognostic factors and suitable treatments for young BC women.

Database
The Breast Cancer Information Management System (BCIMS) is a patient registration database of the West China Hospital, Sichuan University [13]. BC patients treated in the West China Hospital of Sichuan University signed informed consents and were included in this database for registration. Data adopted in this study were from BCIMS and approved by the Clinical Test and Biomedical Ethics Committee of the West China Hospital, Sichuan University. Researchers regularly monitored the quality of data to ensure the completeness and accuracy of data collection. Regular follow-up was initiated according to the NCCN and CSCO BC patient follow-up guidelines. Follow up with patients generally occurred at least every 4 Page 4/28 months during the rst 3 years after diagnosis, every 6-12 months during the third to fth year, and every year after 5 years. Female patients under age 41 who were diagnosed with primary anatomic-stage I-III invasive BC via  pathological examination at the West China Hospital of Sichuan University, received surgical treatment and systematic therapy, and had complete follow-up data were included in this study. We referred to the de nition of the third international consensus guidelines of the European Society for Medical Oncology (ESMO) and chose female breast patients under age 41 as young patients in this study [14]. Records missing HR status, anatomic stage, or follow-up information were excluded. Pathological diagnosis was determined by a pathologist via puncture biopsy samples or surgically resected samples. Estrogen receptor (ER) and progestin receptor (PR) status was determined according to the American Society of Clinical Oncology ER and PR immunohistochemical testing guidelines [15]. HER2-positive cancer is de ned as pathological immunohistochemical staining, showing HER2 "3+" or uorescence in situ hybridization (FISH) suggesting HER2 gene ampli cation [16]. Based on the HR expression status as determined by immunohistochemistry, patients were divided into an HR-positive group and an HRnegative group. Common demographic and baseline characteristics of patients enrolled in this research are displayed in Table 1.

Baseline characteristics
A total of 2,142 female patients aged ≤ 40 who were diagnosed with anatomical stage I-III in the West China hospital, Sichuan University were enrolled. Speci cally, we collected data from 13,399 females with primary invasive BC, 10,778 women over 40 years old were excluded, 6 patients were excluded for recurrent disease, 257 cases were deleted for missing HR information, 27 cases were eliminated for missing clinical stage information, and 135 cases were eliminated for having a follow-up period less than 0.5 months. Additionally, 9 patients were also excluded due to being diagnosed before January 2000 and another 45 patients were deleted for having stage IV disease (Fig. 1). The median follow-up period was 7.1 years. Common characteristics of the whole cohort are listed in Table 1.
To explore the change of BC baseline characteristics in young women from 2000 to 2017, we analyzed the age of patients at diagnosis, HR status, HER2 status, anatomic stage, and surgery pattern according to diagnosis year. During the three time periods before 2008, from 2008 to 2012, and since 2013, the number of women younger than 30 when diagnosed with early breast cancer (EBC) has been increasing. The proportion of women aged 30-35 diagnosed with EBC was slightly decreased, while this percentage uctuated in the three phases in women equal to or older than 35 years. In all of the three time phases, the percentage of the HER2 negative subtype was the highest, while the percentage of the HER2 positive subtype continuously rose from 2000 to 2017. More patients were diagnosed with HR-positive tumors than HR-negative tumors since 2013 (36.6% vs. 28.0%), while a lower proportion of women were diagnosed with HR-positive BC than HR-negative BC between 2008 and 2012 (37.4% vs. 46.1%, P < 0.001). Interestingly, the proportion of clinical stage I-II increased from 2000 to 2017, and fewer patients were diagnosed with stage III disease. Regrettably, women with clinical stage 0 disease signi cantly decreased overall. Further, the number of breast conserving surgeries was remarkably increased in the last ten years and the percentage of radical/extensive radical surgeries decreased (Fig. 2). In the whole cohort, most women suffered from HR-positive disease in the three diagnosis time periods.

Intergroup differences in clinicopathological traits between HR-positive and HR-negative breast cancer
To explore the role of HR in young BC patients, we rst analyzed intergroup differences in clinical and pathological characteristics between the HR-positive and HR-negative groups. Patients were divided into an HR-positive group (N = 1,606) and an HR-negative group (N = 536) based on HR expression status. The median age of both groups was 37 years old (P = 0.99). In addition, 30.5% patients suffered pT1 and 5.1% had pT4 disease in the HR-positive group, while only 27.6% patients suffered pT1 and 3.5% had pT4 stage tumors in the HR-negative group (P < 0.001). Additionally, there were fewer patients with pT3 disease in the HR-positive group than the HR-negative group (5.5% vs. 8.4%, P < 0.001). There were more patients without axillary lymph node metastasis in the HR-negative group than in the HR-positive group (49.4% vs. 46.2%, P < 0.001). The proportion of pN2 disease in the HR-positive group was 12.2%, and in the HR-negative group it was 9.5%. Furthermore, 68.6% of HR-positive patients suffered clinical stage II-III disease and in the HR-negative group, 65.5% had stage II-III disease. Fewer patients with stage 0-I disease (31.5%) were in the HR-positive group compared with the HR-negative group (34.5%, P < 0.001). Ki67 expression and HER2 status were also clearly different between the two groups. A higher percentage of HR-negative women had tumors with Ki67 ≥ 14% (76.5%) and HER2-positive expression (27.0%), while these proportions in the HR-positive group were only 64.2% and 16.1%, respectively (P < 0.001, Table 2).  First stratum of every factor was the reference.

Treatment patterns and chemotherapy regimens
The main surgery types in this study included breast conserving surgery, modi ed radical surgery, and radical or extensive radical surgery. There was no signi cant difference in the surgery pattern between the HR-positive group and the HR-negative group. Differences in chemotherapy between the two groups were not obvious either. There were marginally more women free from radiotherapy during the whole disease process in the HR-negative group compared with their counterparts (63.2% vs. 56.6%, P = 0.06). In both groups, the percentage of patients receiving anthracycline combined with taxanes as neoadjuvant or adjuvant chemotherapy was the highest. Anthracycline was the second highest frequently used chemotherapy regimen. Taxanes were used more frequently in the adjuvant phase than in the neoadjuvant phase. Over 80% of women never received neoadjuvant chemotherapy, while approximately 20% of women were free from adjuvant chemotherapy in both the HR-positive group and the HR-negative group. In addition, 11.9% of women received surgical castration after a diagnosis of HR-positive BC, and 2.4% of women underwent surgical castration during the treatment process in the HR-negative group (  Fig. 4A). After additionally adjusting independent prognosis relative factors and HR status, the latter demonstrated remarkable relativity with OS (HR = 1.51, 95% CI: 1.03-2.21, P = 0.04). Besides, more advanced pT and pN, Ki67 ≥ 14%, adjuvant radiotherapy, and endocrine therapy also independently prognosed worse OS in multivariate cox regression. The three-year and ve-year iDFS were 90.0% and 86.6% in the HR-negative group, and the three-/ ve-year iDFS in the HR-positive group were 92.3% and 87.7%, respectively (Log-rank P = 0.85, Fig. 3B). Univariate Cox hazards proportional regression showed that age, time-period diagnosed, surgical castration, tumor histologic grade, pT and pN, anatomic clinical stage, Ki67 expression intensity in tumor tissue, chemotherapy and radiotherapy patterns, as well as endocrine therapy were all iDFS relative factors (P < 0.05, Fig. 4B).
Factors that were signi cant in the univariate analysis were included in the multivariate Cox hazards proportional regression combined with HR status. The HR-positive status predicted worse iDFS in the multivariate analysis (HR = 1.47, 95% CI: 1.05-2.05, P = 0.02). Diagnosis after 2008 and receiving endocrine treatment with SERMs were independent favorable outcome predictors, while higher pT and pN stage, surgical castration, and Ki67 ≥ 14% were independent risk factors in multivariate cox regression analysis.
Considering the important role of HR status in young female BC, we further analyzed factors related to OS and iDFS in HR-positive and HR-negative subgroups. Factors which were associated with OS and iDFS (P < 0.1, Supplementary Fig. 1/2) in univariate cox regression were further adopted in multivariate analysis.
Higher gravidity, more advanced N stage, Ki67 ≥ 14% and radical mastectomy were independent risk factors while endocrine therapy was independent protective factor in HR-positive subgroup for OS (Table 4). In HR-negative subgroup, higher T stage and N stage independently predicted worse OS while modi ed mastectomy prognosed better OS. Multivariate analysis also showed that later diagnosis period and endocrine therapy independently prognosed better iDFS and more advanced T stage, N stage and histological grade related to worse iDFS outcomes in HR-positive subgroup (Table 4). Only women who diagnosed with BC at older ages prognosed better iDFS in HR-negative subgroup (Table 4)  Propensity score matching (PSM) There were signi cant differences in baseline clinicopathological characteristics between patients receiving or not receiving chemoradiotherapy. To investigate the effects of these treatments, PSM was rst conducted to achieve post-randomization. Before matching, variables including diagnosis timeperiod, pT, pN, Ki67, histological grade, and HER2 status were imbalanced between patients receiving or not receiving all three treatments. HR status was also signi cantly different between patients receiving or not receiving adjuvant radiotherapy (Table 5). We matched patients based on their propensity to receive neoadjuvant chemotherapy, adjuvant chemotherapy, and adjuvant radiotherapy. The propensity score (PS) was the conditional probability of receiving an exposure given a group of covariates that can be used to adjust intergroup bias to estimate the effects of different treatment patterns. The intergroup difference of all variables had been eliminated after PSM (Table 5), and the balance of covariates of the three treatments in Table 5 was substantially improved after matching (Fig. 5). We did not conduct PSM for surgery because most of the patients received modi ed radical surgery (77.7%, Table 1). Age was excluded from PSM for patients receiving or nor receiving adjuvant chemotherapy because an ideal balance could not achieve when age was included. # Only diagnosis time period pT, pN, and HR status were included in PSM for patients receiving or nor receiving adjuvant radiotherapy. Covariates included age, Ki67, HER2 status, WHO grade were balanced between the two groups and ideal balance could not be acquired when add any of these four covariates.
ζ Covariate of diagnosis time period was excluded from PSM for patients receiving or nor receiving neoadjuvant chemotherapy because an ideal balance could not achieve when this factor was included.
Multivariable Cox hazards proportional regression was done to explore the effectiveness of neoadjuvant chemotherapy, adjuvant chemotherapy and adjuvant radiotherapy in pre-matched and post-matched cohorts. The prognosis-related variables of OS and iDFS were adjusted in the regression models together with treatment status (Table 6) No association was seen between adjuvant radiotherapy and iDFS both before and after matching ( Table 6).    [17][18][19][20]. The proportions of the triple negative subtype (15.9%) and HER2-enriched subtype (HR-negative, 6.8%) were also similar to previous studies [21,22]. The luminal A subtype in a group of elder patients (median age 55.5) was 34.4%, and the proportion of luminal B, HER2-enriched, and basal-like subtypes were 17.5%, 20.8%, and 24.7%, respectively [23]. Combined with our data, this suggests that there is a higher proportion of HR-positive tumors in young patients. Notably, previous data showed that BC in young patients is more likely to be of the triple-negative and HER2-enriched subtypes [17,24,25]. However, the proportions of the two subtypes in our cohort were lower than in elder patients in previous studies [23].
HR-positive disease was an independent risk factor of OS and iDFS in young patients, which was validated in our study and others [20,21]. It is worthy of noting that endocrine therapy was de nitely protective factor in HR-positive subgroup for OS and iDFS (Supplementary table 1/2), suggesting that endocrine therapy is particularly relevant to the long-term prognosis to HR-positive BC. Tumor and lymph node stage were the main factors impacting the survival outcomes of young patients with both HRpositive and HR-negative BC. It had been reported HR positive was an independent factor for more later period BC-speci c death rate, especially in the subgroup of 20-40 years old patients [8]. Another research reported that HR positive only prognoses worse survival outcome in young women (< 60 or premenopausal women) rather in old women [26]. In view of a signi cant impact of HER2 on the prognosis of BC, we had adopted HER2 expression status into multivariate cox regression analysis. After combined adjusted HR and other prognosis related factors with HER2 status, HR-positive expression was still prognosed worse survival outcomes (OS: HR = 1.46, 95% CI 0.99-2.16, P = 0.057, DFS: HR = 1.41, 95% CI 1.01-1.98, P = 0.04). We also reviewed study which reported that HER2 did not have an prognostic in uence on HR-positive young BC [27]. Age has been de ned as an independent risk factor for survival in BC patients [9,28,29]. Xu et al. reported a 95.2% 5-year OS in patients under age 60 (mainly composed of post-menopausal women ) [30]. In the present research, the median iDFS of the whole cohort was 6.3 years and the 5-year OS was 67.7%, clearly inferior to that of the elder patients. Furthermore, studies had reported that BC in young patients was more prone to lymph node metastasis [31][32][33]. Our data displayed that 53% of patients had lymph node metastasis, while this rate in a study of patients aged 40-50 was 49.2% [22]. Few studies had explored the relationship between the HR status and node metastasis. There were slightly more patients (53.7%) suffering from lymph node metastasis in the HRpositive group than in the HR-negative group (50.6%, Table 2). Although logistic regression in this study did not show a de nite correlation between the HR status and the node metastasis (OR = 1.14, P = 0.20). We found that breast conserving surgery and modi ed radical surgery marginally improved OS in young BC patients, probably due to the majority patients received modi ed radical surgery. In addition, adjuvant chemotherapy and adjuvant radiation improved survival outcomes signi cantly, emphasizing the importance of adjuvant systematic cytotoxic treatment and adjuvant radiotherapy in young patients.

Conclusion
Although we did not nd that age act as independent prognostic factor, survival outcomes in this young cohort were clearly worse than in elder patients from previous studies. HR-positive status predicted worse survival outcomes. Adjuvant chemotherapy and adjuvant radiotherapy were required for all young patients. In addition, endocrine therapy was necessary for young women with HR-positive disease. No clear bene t was seen from neoadjuvant chemotherapy, and the role of neoadjuvant chemotherapy in young women with early-stage BC requires more exploration.

Con ict of Interest
The authors declare that the research was conducted in the absence of any commercial or nancial relationships that could be construed as a potential con ict of interest.

Data Availability Statement
The original datasets generated during analyzed are not publicly available due individual privacy could be compromised but are available from the corresponding author on reasonable request.