High A20 expression negatively impacts survival in patients with breast cancer

Background A20 protein has ubiquitin-editing activities and acts as a key regulator of inflammation and immunity. Previously, our group showed that A20 promotes tumor metastasis through multi-monoubiquitylation of SNAIL1 in basal-like breast cancer. Here, we investigated survival outcomes in patients with breast cancer according to A20 expression. Patients and methods We retrospectively collected tumor samples from patients with breast cancer. Immunohistochemistry (IHC) with an A20-specific antibody was performed, and survival outcomes were analyzed. Results A20 expression was evaluated in 442 patients. High A20 expression was associated with advanced anatomical stage and young age. High A20 expression showed significantly inferior recurrence-free-survival and overall-survival (P<0.001 and P<0.001, respectively). Multivariate analysis showed that A20 was an independent prognostic marker for RFS (HRs: 2.324, 95% CIs: 1.446–3.736) and OS (HRs: 2.629, 95% CIs: 1.585–4.361). In human epidermal growth factor receptor 2 (HER2)-positive and triple negative breast cancer (TNBC) subtypes, high A20 levels were associated with poor OS. Conclusion We found that A20 expression is a poor prognostic marker in breast cancer. The prognostic impact of A20 was pronounced in aggressive tumors, such as HER2-positive and TNBC subtypes. Our findings suggested that A20 may be a valuable target in patients with aggressive breast cancer.

TNM stage was determined according to the American Joint Committee on Cancer (AJCC), 7 th edition. Tumor grade was determined using the modified Scarff-Bloom-Richardson grading system. Before February 1999, ER and PR status were assessed using a ligand binding assay, and tumors were considered positive if the score was >10 fmol/mg.
The study protocol was approved by the institutional review board of Gangnam Severance Hospital.

Ethics approval and consent to participate
Our study itself was conducted as human-specimen subject research and was approved by the institutional review board (IRB) review (Local IRB number: 3-2018-0067). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The need for informed consent was waived under the approval of the IRB due to the retrospective design.
For the molecular subtyping, the following definitions were used: i) Luminal/HER2 negative: ER positive and/or PR positive and HER2 negative; ii) HER2 positive: HER2 positive regardless of ER and PR status; and iii) TNBC: ER negative, PR negative, and HER2 negative.

Tissue microarray and IHC staining to evaluate A20 expression
Tissue microarray (TMA) paraffin blocks were generated as previously described using an Accu Max Array tissue-arraying instrument (Petagen, Inc., Seoul, Korea) [17]. For IHC, each TMA slide was stained with a rabbit monoclonal anti-A20 antibody (ab92324, 1:200; Abcam) and counter-stained with hematoxylin. After staining, A20 expression, assessed as cytoplasmic staining, was scored by an experienced pathologist (A.O.) using a microscope (400× magnification). Positive A20 expression in tumor cytoplasm was defined when the percentage of stained cells was equal to more than 60%. Finally, a score of 2+ or 3+ was defined as high A20 expression, whereas a score of 0 or 1+ was defined as low A20 expression (S1 Fig). The IHC results were interpreted blindly, without any information regarding clinical parameters or outcomes.

Statistical analysis
Continuous variables were compared using Student's t-test or Mann-Whitney test. Categorical variables were compared using the Chi-square test or Fisher's exact test. Overall survival (OS) was defined as the period from primary curative surgery to the last follow-up or death from any cause. Recurrence free survival (RFS) was defined the period from primary curative surgery to the date of any recurrence (loco-regional or distant metastasis), death due to any cause or the last follow-up. Data for patients who did not have an event of interest were censored at the date of the last follow-up. Kaplan-Meier survival curves were compared using the log-rank test. Clinicopathologic factors associated with survival outcomes were analyzed using the Cox proportional hazard model. The variables used in the multivariate analysis were those that showed statistical significance in the univariate analysis for OS and RFS.
To compare the predictive power of the Cox proportional hazard model with and without A20, we applied the Harrell's c-statistic to obtain concordance index (c-index) [19]. It is calculated to measure the concordance for time-to event data, in which increasing values between 0.5 and 1.0 indicated improved prediction. For model comparison, the bootstrapping method was used with resampling 1,000 times. We also performed net reclassification improvement (NRI) and integrated discrimination improvement (IDI) to evaluate improvement of discriminating ability when new factors were added in the survival model [20]. In NRI and IDI, a significant improvement is recognized in the prediction model when their values are greater than 0.
These analyses were performed using SPSS version 23 (SPSS; Chicago, IL, USA) and R version 3.4.2 (www.R-project.org). Statistical significance was defined as a P value less than 0.05, at a 95% confidence interval (CI) not including 1.

Patient characteristics based on A20 expression
This study included 442 patients, with a median age of 47 (22-86) years. Of the 442 patients, 390 (88.2%) had low A20 expressing tumors, and 52 (11.8%) had high A20 expressing tumors. Clinical characteristics were compared based on A20 expression in Table 1. High A20 expression was associated with anatomical tumor stage; A20 was highly expressed in advanced T stage (P = 0.006), N stage (P = 0.045), and AJCC stage tumors (P = 0.030). In addition, high A20 expression was associated with younger age at surgery (P = 0.011). However, it was not related to tumor grade (P = 0.299) or receptor status (ER; P = 0.765, PR; P = 0.412, HER2; P = 0.837). In addition, there was no statistical difference in adjuvant treatment, such as chemotherapy (P = 0.724), radiotherapy (P = 0.763) or endocrine therapy (P = 0.288).
We classified patients into three subtypes based on ER, PR, and HER2 status; 125 patients were of the luminal/HER2-negative subtype, 132 were of the HER2-positive subtype, and 154 were of the TNBC subtype. There were no significant associations between A20 expression and breast cancer subtype (P = 0.503).

Prognostic impact of A20 expression
At a median follow-up time of 92.5 months (1-258), 88 mortality events had occurred. There were a total of 110 recurrence events, of which 29 were locoregional recurrences, 97 were distant metastases, and 16 were both locoregional recurrences and distant metastases. Of 110 recurrence cases, there were 72 mortality events from breast cancer.
High A20 expression was significantly predictive of decreased RFS (P < 0.001, log-rank test; Fig 1A). Patients with high A20 expression also showed reduced OS (P < 0.001, log-rank test; Fig 1B). Younger age at surgery (�35 years old), advanced T stage, advanced N stage, LVI and A20 (HR 2.672, 95% CI 1.710-4.175, P<0.001) were significant factors in the univariate analysis of RFS (Table 2). A20 (HR 2.906, 95% CI 1.801-4.690, P<0.001) was still a significant variable in the univariate analysis of OS (Table 3), too. To evaluate an improvement of predicting ability between multivariate models with and without A20, we calculated c-index, NRI, and IDI. The addition of A20 to base model increased c-indices for both OS and RFS without a statistical significance (0.733 to 0.755 for OS; 0.689 to 0.698 for RFS; Table 4). At a median follow-up 93 months, the addition of A20 improved a discriminating ability for OS because NRI and IDI were 0.230 and 0.047 (P = 0.028 and P = 0.004, respectively; Table 4). At a median follow-up 83 months, a model with A20 showed a superior discrimination by IDI (0.030, P = 0.008) whereas a NRI for this model was greater than 0 (0.151) but not significant (P = 0.076).

Discussion
In this study, we found that A20 expression was an independent prognostic factor for breast cancer. Compared to patients with low A20-expressing tumors, patients with high A20- . Although A20 expression was not associated with specific subtypes, the prognostic impact of A20 expression was more pronounced in aggressive subtypes, such as HER2-positive and TNBC subtypes. It might be implicated that A20 has clinical value as a novel target for improving treatment outcomes in patients with aggressive breast cancer, such as HER2-positive subtype.
Earlier studies on A20 in inflammatory and autoimmune diseases showed that A20 is associated with a complex regulator of ubiquitylation of canonical NF-κB and cell-survival signals A20: linking a complex regulator of ubiquitylation to immunity and human disease [2-10, 21, 22]. In contrast, the role of A20 in malignant disease has not been extensively explored. Several genetic studies have reported that deletion or mutation of A20 is associated with hematologic malignancies, such as multiple lymphoma, MALT lymphoma, Hodgkin's disease, and follicular lymphoma, which are characterized by NF-κB signal activation [23][24][25][26]. Other studies have suggested that high A20 expression was associated with tumor survival and growth in solid tumors such as malignant glioma and breast cancer [11][12][13][14]. Previous studies on the role of A20 in malignancies suggest that it acts as a tumor suppressor or enhancer depending on the cell type.
In addition to the above findings, A20 has also been reported to have a novel role in breast cancer; A20 promotes tumor progression and activates EMT signals, mainly by affecting SNAIL1 stabilization [16]. EMT signaling, which is associated with the capacity to migrate to distant organs [15], is frequently activated in TNBC [27,28]. A mechanism by which the EMT pathway is activated via A20 elevation [16] might partly explain our findings: poor survival outcomes in TNBC patients with high A20 expression. Further studies of EMT features, such as high vimentin expression or E-cadherin loss, in TNBC with high A20 expression levels are warranted.
Intriguingly, we also found that high A20 expression is a poor prognostic marker in patients with HER2-positive breast cancer. Several studies have provided evidence that upregulation of NF-κB signaling in HER2-positive breast cancer is associated with resistance to various therapies, including anti-HER2 therapy [29,30]. Although A20 is a negative regulator of NF-κB signaling in inflammation and immune response, a previous study by Lerebours et al. showed that A20 is involved in upregulation of the NF-κB pathway in inflammatory breast cancer (IBC) [31]. Although HER2 status was not reported in the study, it has been noted that 30~50% of IBCs have HER2 overexpression [32][33][34], suggesting that A20 expression might be associated with NF-κB pathway activation in HER2-positive breast cancer. Further studies on the association between A20 and NF-κB are warranted in HER2-positive cancer.
The biology underlying poor outcomes in patients with high A20-expressing HER2-positive breast cancer is also supported by a mechanism of A20-mediated TGF-β-activation. Previous studies have shown that HER2 stimulates TGF-β-mediated EMT, contributing to the aggressive behavior of HER2-positive breast cancer [35][36][37]. Our previous study also provided evidence that A20 elevates TGF-β signaling, suggesting that A20 could promote a TGF-βmeditated EMT pathway and tumor aggressiveness in HER2-positive breast cancer.
Other studies have reported that A20 overexpression was associated with endocrine therapy, chemotherapy, and radiotherapy treatment failure. A preclinical study of MCF7 cells demonstrated that tamoxifen-resistant cancer cells showed A20 overexpression [14]; however, there was no prognostic influence of A20 expression in patients with luminal/HER2-negative breast cancer. In addition, another study suggested that A20 reduced the efficacy of radiotherapy and chemotherapy via the regulation of anti-apoptotic mechanisms [38].
Our study had several limitations. First, it was a retrospective study with a small sample size, and the enrolled patients received uncontrolled adjuvant treatments. Particularly, most patients (88.5%) with HER2-positive tumors did not receive anti-HER2 therapy (S2 Fig) because most of the HER2-positive samples were collected prior to 2010 when the use of trastuzumab was not reimbursed by the national insurance system in Korea. In addition, there were higher proportions of HER2-positivie and TNBC subtypes in our study than there are in global breast cancer populations, because of selection bias during sample storage and preservation. Despite these limitations, our study comprehensively showed the clinical outcomes based on A20 expression among patients with breast cancer and demonstrated A20 expression as an independent prognostic factor. Furthermore, prognostic discrimination based on A20 expression was pronounced in aggressive tumors, such as HER2-positive and TNBC subtypes. Our findings suggest that A20 may be a valuable target in patients with aggressive breast cancer.

Conclusion
In conclusion, we found that A20 expression is a poor prognostic marker in breast cancer. The prognostic impact of A20 was pronounced in aggressive tumors, such as HER2-positive and TNBC subtypes. Our findings suggested that A20 may be a valuable target in patients with aggressive breast cancer Funding acquisition: Sung Gwe Ahn.