Determinants of Prognosis in Triple-Negative Breast Cancer: Report from a Large Breast Cancer Registry

1MBreast Diseases Research Center, Shiraz University of Medical Sciences, Shiraz, Iran 2Department of Surgery, Karbala University of Medical Sciences, Karbala, Iraq 3Surgical Oncology Division, General Surgery Department, Shiraz University of Medical Sciences, Shiraz, Iran 4Department of Foreign Languages and Linguistics, Shiraz University, Shiraz, Fars, Iran 5Non-communicable Disease Research Center, Fasa University of Medical Sciences, Fasa, Iran 6Department of Biostatistics and Epidemiology, Faculty of Health, Isfahan University of Medical Sciences, Isfahan, Iran 7Department of Internal Medicine, Shiraz University of Medical Science, Shiraz, Iran


Introduction
Triple-negative breast cancer (TNBC) represents a subgroup of tumors that mainly lack specific surface antigens including hormone receptors (which includes estrogen [ER] and progesterone receptors [PR]) and the HER-2/neu or the human epidermal growth factor receptor 2. 1 TNBC comprises an estimated 10%-20% of all newly diagnosed early breast cancers (BC) and is often considered a basal-like BC. 2,3 Literature has shown that patients with TNBC have an overall worse clinical outcome and a higher recurrence rate compared to BC patients who are hormone receptors positive. 4,5 TNBC patients show a unique pattern of recurrence, with high rates of recurrence during the first five years, followed by a significant decrease and plateau in recurrence rates after this period. 6,7 These patients experience more frequent distant recurrences in visceral organs, including the lungs, brain, liver, and less frequently in the bone. 8 In addition, the survival rate of patients with TNBC decreases after recurrence of the disease compared to BC patients who test positive for hormone receptors. 3,5 Prognostic factors and determinants of outcome in patients with TNBC differ significantly from other groups of BC patients. Several epidemiological studies have demonstrated that age and ethnicity are important determinants of outcome in patients with TNBC. 9,10 Clinicopathological studies have also demonstrated that some clinical characteristics and tumor characteristics are considered important prognostic factors in patients with TNBC. 7,11 These include tumor size, lymphovascular invasion and distant metastasis. 12 Survival and outcome among patients with TNBC are highly dependent on demographic factors and ethnicity. 5,7,13 Thus, understanding prognostic factors in TNBC in each geographical region and ethnic group provides further understanding which will eventually provide a personalized management plan to tailor a multimodality treatment option for a subtype of BC which has no definite therapeutic target. In addition, data from Iran is scarce and only few studies have reported the clinicopathological characteristics of the disease. 11,13,14 Considering that TNBC is a unique clinical entity, we aimed to evaluate the clinicopathology of TNBC among women and to determine prognostic factors in this population using data from one of the largest BC registries in our region.

Study Population
This study is part of the Shiraz Breast Cancer Registry (SBCR), which is a tertiary referral cancer center in southern Iran, affiliated with Shiraz University of Medical Sciences. This is the largest BC registry in Iran and includes data from patients with BC from 1995 to the current date. Specifics on the registry have been reported elsewhere. 13 This study included all patients from the database and patients with a pathologically proven TNBC who had complete clinical, pathological and follow-up data from February 2001 to January 2019. Patients with ductal carcinoma in situ and recurrent BCs were excluded from the study. Furthermore, those with incomplete medical records and those who were lost to follow-up were also excluded.

TNBC Definition
TNBC was defined as lack of surface expression of ER, PR, and HER2/ErbB2. The status of ER, PR and HER2 was determined by immunohistochemical staining (IHC) and fluorescence in situ hybridization (FISH) at the pathology department of Shiraz University of Medical Sciences. The formalin-fixed paraffin-embedded histological sections were reviewed and the diagnoses were confirmed by two dedicated breast pathologists. IHC analysis was performed to determine ER/PR status using standard procedures on 4-μm sections of paraffinembedded tissues stained with monoclonal antibody for ER and PR. Assessment of ER and PR status was carried out according to the Reiner Score. ER and PR status was considered negative with less than 1% positive tumor cells. 15 HER2 was assessed by means of IHC and/or FISH. IHC was scored on a qualitative scale from 0 to 3+ based on interpretation of membranous staining intensity, where 0 and 1+ were classified as negative, 2+ as borderline, and 3+ as positive. HER2 (++) tissues were re-evaluated by FISH and if the HER2 gene amplification copy-to-CEP17 ratio was greater than 2.0, that sample was considered HER2 positive.

Study Protocol
Baseline information including age at diagnosis, history of BC and physical activity; obstetric and gynecological information including age at start of menstruation, age at first pregnancy, number of pregnancies, number of abortions, number of children, duration of breast feeding, oral contraceptive use, hormone replacement therapy, history of benign breast diseases, and age at menopause; tumor characteristics such as size, stage, grade, nodal involvement, invasion status; treatment; and pathology related characteristics including type of surgery, grade of nucleus, tumor necrosis, in situ component in histological evaluation, histopathological subtype and type of management of axillary lymph nodes were registered. Invasion status was considered positive if present on either biopsy or surgical pathology. All patients were followed according to the registry protocol. 13 Recurrence in the ipsilateral treated breast and/or chest wall or ipsilateral nodal basin was considered loco-regional recurrence. Any recurrence at a distant site was considered distant metastasis. All study outcomes were evaluated in the last follow-up.

Statistical Analysis
All statistical analyses were conducted using the Statistical Package for Social Sciences (SPSS Inc., Chicago, Illinois, USA) version 22.0. For comparison of quantitative data with normal distribution between two groups, the independent t test, and for comparison of qualitative data between groups, the chi-square test was used.
Initially, the Kaplan-Meier test was used to evaluate overall survival (OS) and disease-free survival (DFS). DFS was considered as the period from the last day of treatment to confirmation of recurrent disease in the ipsilateral breast, regional, or distant site, and/or death during which the patient was symptom-free. For patients who remained alive and recurrence-free, OS was considered as the period from the last day of treatment to death or the last followup. The logrank test was used to compared OS and DFS between subgroups.
To assess the independent relationship between each variable and survival and recurrence, we used a multivariate Cox regression analysis considering the date of treatment as the start point and the date of event as either death or recurrence as the endpoint. Recurrence was considered any type of local, regional and/or distant metastasis. A stepwise method was used for variable insertion and a P value cutoff entry level of 0.10 was set for variable selection for the Cox regression model.
For evaluation of predictors of recurrence and survival, those with stage 4 BC's were excluded, as these individuals already have metastatic disease. The OS and DFS were assessed for the whole population and for individuals with diagnosis of BC more than five years separately.
A two-sided P value of less than 0.05 was considered statistically significant.

Results
During the study period, a total of 6145 women with primary BC confirmed by pathological examination were included in the SBCR. A total of 523 women were diagnosed with TNBC. The mean age of patients at first diagnosis was 46.52 ± 11.42 (ranging from 24 to 81) years. Most of the patients were in histological grade III (48.8%) and the majority of patients had a tumor size between 2 and 5 cm (66.7%). Invasive ductal carcinoma and medullary carcinoma were the most common types of BC (77.7% and 14.1%, respectively). During our study period, in total, 17.8% of patients experienced recurrence. Nodal involvement was recorded in one third of patients (35.5%). Most of the patients (58.9%) were treated with breast conserving surgery and auxiliary node dissection was carried out in the majority of patients (62.5%). The clinical and histopathological characteristics of patients are summarized in Table 1.
The DFS rate was 80.6% and the mean DFS was 157.76 ± 9.48 months among patients with TNBC. The OS rate was 90.1% and the mean OS was 182.73 ± 3.28 months ( Figure 1).
The Cox regression analysis showed that for overall death, stage 3 BC (compared to stages 0 and 1 as base) showed a higher odds of earlier death (HR: 4.191, 95% CI = 1.392 -12.621; P = 0.011). Similarly, among individuals who had a diagnosis of BC for more than five years, stage 3 BC showed higher odds of earlier death (HR: 4.210, 95% CI = 1.393 -12.719; P = 0.011) ( Table 3).
For recurrence, stage 3 BC (HR: 1.044, 95% CI = 1.209 -6.673; P = 0.017) (compared to stages 0 and 1 as base) showed significantly higher odds for developing earlier recurrence. Moreover, those who had invasive ductal carcinoma (compared to other types of BCs) had higher odds for developing earlier recurrence (HR: 3.307, 95% CI = 1.191-0.724; P = 0.012). Among those who had a diagnosis of BC for more than five years, only invasive ductal carcinoma (compared to other types of BC's) showed significantly higher odds of developing earlier  Figure 2).

Discussion
In this study, we investigated the determinants of prognosis in a large series of women with TNBC. In our univariate analysis, we found that those who developed recurrence had a higher rate of individuals with higher stages, more individuals with invasive ductal carcinomas (in comparison to other subtypes), more lymphovascular and perineural invasion, more involved lymph nodes, higher rates of breast conserving surgery, and higher rates of axillary node dissection compared to other axillary management modalities. In our Cox regression analysis, we found that regarding death, stage was a predictor of earlier death and regarding recurrence, stage and histopathological subtype were significant determinants of earlier recurrence. These findings are consistent with previous reports from other ethnic groups. [16][17][18][19][20][21][22] In our series, lymphovascular and perineural invasion were found to be significantly higher among patients with TNBC who showed recurrence. Previously, lymphovascular invasion was shown to be associated with increased risk of recurrence in patients who either had breast conservation therapy or mastectomy. 23 Lymphovascular invasion has been further associated with worse BC-specific survival and distant metastasis-free survival. 24 Moreover, we found that about 40% of the patients with TNBC in our series had lymphovascular invasion which is high compared to previous series. 16,[22][23][24] Various studies have addressed prognostic factors in TNBC patients in different ethnic groups. Rakha et al 7 studied prognostic factors among a sample of TNBC patients from the UK. In their report, among the total 1726 cases of invasive breast carcinomas whom they studied, 282 were TNBC's. They found that nodal status, tumor size, and androgen receptor expression were the most important prognostic factors. We found no association between prognosis and tumor size among our TNBC patients. They also demonstrated that in the lymph node-negative group, basal phenotype was the sole marker that showed prognostic value whereas other specifics, including patients' age, tumor size, and androgen receptor expression, were not significant predictors. 7 In our multivariate model, we only found the stage of cancer to be a predictor of survival and both stage and histopathological subtype to be predictors for recurrence. The different findings could be attributed to multiple factors: first, the different variables which were included in our regression models, as the mentioned study mostly focused on molecular determinants. Moreover, the mentioned study had a relatively small sample size (compared to that of our study), which may have affected the results of the regression analysis, although this is expected considering the overall low number of patients with TNBC.
Recently, Kashi et al 22 reported survival rates and the determinants of outcome in a series of patients with BC. They evaluated 1910 BC patients, among whom 180 (9.4%) patients had TNBC. They reported that age (≥40 years), grade and stage III at first diagnosis (compared to grade and stage 1), and visceral recurrence were significant predictors of outcome. 22 In another study, Mirzania et al 14      different. 25 Conversely, Kim et al 26 found that patients with TNBC have similar results with regard to loco-regional recurrence using breast conserving surgery compared to those without TNBC. Consequently, they concluded that breast conserving surgery is a good treatment choice among these patients. 26 In a previous comprehensive report, the authors found age at diagnosis, number of dissected lymph nodes, number of involved lymph nodes, in situ component, grade, tumor necrosis, history of breast disease, smoking, type of axillary management, radiotherapy and stage of cancer to be involved in recurrence among BC patients (not specific to TNBC) using a machine learning algorithm. 27 Although a different statistical approach was used, these findings were different when compared to patients with TNBC as we found only stage and histopathological subtype to be significant predictors of recurrence in this population. This difference in findings could be attributed to the different subtype of BC which was studied, and further studies are needed to directly compare the determinants of outcome between different subtypes of BC. Another point is that some previous studies had entered factors such as tumor size as a categorical variable in their multivariable models and perhaps these factors are significant for prediction of prognosis at a specific cut-off point; thus, they did not appear as significant in our final model.
Our findings indicate that patients with TNBC who present with a higher stage would perhaps benefit from earlier screening programs for evaluation of recurrence in their follow-up visits as stage is a predictor of earlier recurrence in this specific population. This could also be considered for patients who show invasion (lymphovascular or perineural) on their assessment.
This study was not without limitations. We did not include demographic and social variables in our analysis as they had a lot of missing data, although the role of demographic/ethnical and social factors are well known in determining the outcome of patients with TNBC. 16 Although the rates of TNBC are low by nature, we used data from one of the largest databases to overcome this issue; however, some subgroups did still have low numbers of individuals. Due to the rarity of TNBC patients, compared to the whole number of patients with BC, we included all individuals in our regression models. Some patients who were included in our models and who have been recently diagnosed may not have had the chance to present outcomes such as death or recurrence and may have caused some bias in our results. Some of the total 6145 BC patients that were included in our current report may have had missing data regarding HER2 status, ER or PR receptor expression status; thus, the exact rate of TNBC among our total population is not measurable from the current report. For recurrence, we classified histopathology into two groups of IDC and "others". This classification was mainly done due to the small number of patients who would classify as "others", and those with IDC may not necessarily have had better conditions regarding recurrence compared to all subtypes of BC's as shown in our multivariate analysis. Furthermore, although we separated those with stage four BC, to eliminate the