Prognostic implications of stromal tumor-inltrating lymphocytes and programmed death ligand 1 expression in HER2-positive invasive breast cancer

Background: Tumor inltrating lymphocytes (TILs) are known to be an important prognostic factor of specic breast cancer subtypes. However, to date, PD-L1 (cid:0) SP142 (cid:0) expression and tumor inltrating lymphocytes (TILs) has been only minimally reported in HER2-positive breast cancer cases. Methods: In this study, CD8,CD4 and PD-L1 (cid:0) SP142 (cid:0) were performed on tissue microarrays (TMA) representing 156 pretreatment cases of HER2-positive invasive breast carcinoma. Clinical data were collected, all cases were with complete follow-up data (23 recurrence and 2 deaths). The concordance rate in the tumor immune microenvironment(TIME) classication based on the PD-L1 (cid:0) SP142 (cid:0) expression status and CD8/CD4+TILs count was analyzed. Results: PD-L1 expression was identied in 49 cases (31.4%) in immune cells. We found there was no association was found between PD-L1 expression and DFS, but TIME classication was associated with DFS. Conclusion


Background
Breast cancer is the major cause of cancer-related death among females worldwide [1] .Precise treatment based on molecular typing of breast cancer signi cantly improves the prognosis of patients.
Approximately 10% to 15% of BCs have the human epidermal growth factor receptor 2 (HER2) gene ampli cation or protein overexpression. Anti-HER2 therapy, such as trastuzumab, represented a breakthrough for patients with HER2-positive disease. However, Cancer cells and tumor-associated cells are phenotypically and functionally heterogeneous as a result of genetic and non-genetic sources, thus causing both primary and secondary resistances to anti-HER2 targets in about 50% of HER2-postive patients [2] .
In recent years, immune therapy based on PD-1 (programmed death 1) or PD-L1 (programmed deathligand 1) identi cation by immunochemistry had become a hot topic [3] . Serving as an powerful immune checkpoint in tumor microenvironment, PD-1/ PD-L1 biomarkers had shown signi cant theropy e cacy in a variety of tumor types [4] .
Previously, most studies have implied that PD-L1 expression was associated with a worse prognosis in triple negative breast cancer, while tumor in ltrating lymphocytes (TILs) predicted favorable clinical outcomes, especially a abundant of CD8+ T cells both in triple negative and HER2-positive breast cancer [5][6] . Recently, Vihervuori, et al. [7] suggested that each 10% decrease in stromal TILs resulted in 20% increased risk of mortality. And, another research [8] of 1318 BC patients in European also suggested that tumor-in ltrating lymphocytes (TILs) density was remarkably related with PD-1 and PD-L1 expression in immune cells .The importance of both PD-L1 expression and TILs prediction roles in TNBC of clinical outcome has been con rmed, but it reminds clinical research data in HER2-postive patients.
Recently, a study [9] of lung adenocarcinoma patients which were divided into four subgroups based on the PD-L1 expression status and CD8+TIL count in the tumors reported that the outcomes of patients with low PD-L1 expression levels and high CD8 +TIL counts were signi cantly better than those with high PD-L1 expression levels and low CD8+TIL counts in the tumors, and this immunophenotyping were applied to breast cancer, shall we acquired the same result, we had analysed the result in our research.
The Food and Drug Administration (FDA) has recently granted an accelerated approval for atezolizumab, a monoclonal antibody drug targeting PD-L1, for the treatment of adults with PD-L1-positive TNBC, while SP142 is as the companion test for selecting TNBC patients, However, Limited data have been reported on the expression of PD-L1(SP142) in immune cells in a pure cohort of HER2-positive breast cancer, In the current study, we evaluated PD-L1(SP142) expression of immune cells and the relationship between TILs and PD-L1 or prognosis in HER2-positive cases. Moreover, an immune Classi cation of breast tumors into 4 groups in the light of their PD-L1 status and CD4/CD8 status has already been proposed in breast cancer for the rst time.

Patients and specimens
A total of 156 con rmed cases of HER2-positive breast cancer para n specimens were collected from the Fourth Hospital of Hebei Medical University from Jan. 2010 to Sep. 2015.All cases were treated with standard adjuvant chemotherapy and HER2-blocking therapy following surgery. HER2 status was determined by HER2 immunohistochemistry (IHC) and/or HER2 uorescence in situ hybridization (FISH).
All cases with complete follow-up data with 23 recurrence and 2 deaths cases.

Antibodies and Tissue microarrays
Tissue microarray (12*8 array) combined with immunohistochemistry was used to evaluate the expression of PD-L1(SP142) in tumor in ltrating lymphocytes (IC),3 cores size of 2mm were collected as representative of primary resected tumor specimen.
Tissue microarray model was RP-20 (2.0mm 8*12 points, Servicebio Company). IHC was performed for PD-L1 (SP142, Roche, Shanghai). Para n slices were baked in oven at 56℃, dewaxed with xylene, hydrated with gradient ethanol, washed with phosphate buffer (PBS) and washed with distilled water. 3% of hydrogen peroxide solution blocked the endogenous peroxidase 10min and EDTA antigen repair 15min. A rst antibody was added and incubated at 4℃ for the night. Rinse with PBS, add second anti drops, incubate 20-30min at room temperature, rinse with PBS solution. Diaminobenzidine (DAB), hematoxylin re-staining, gradient ethanol dehydration, xylene transparency, neutral gum seals, observed under microscope.

Pathologic assessment
Tumor specimens from the 101 included patients were immunohistochemically evaluated to detect the expression of PD-L1 (SP142), CD4 and CD8. The stained tissue sections were independently scored by two pathologists who were blinded to the patients' clinical characteristics and outcomes.
PD-L1(SP142) positive expression was de ned as any staining in more than 1% of immune cells (IC) and negative when less than 1% in the tumor area (Figure1). CD4 and CD8 expression on lymphocytes was reported as the proportion of positive cells among all nucleated cells in the stromal compartments of each core, and scoring was recorded as negative (<10%) or positive (≥10%) ( TILs are divided into intratumoral and stromal TILs. intratumoral TILs directly contact with tumor cells and there is no stromal tissue between cells while stromal TILs are distributed between tumor nests. Stromal TILs in ltration has been turned out more clinically signi cant. Therefore, this study mainly evaluates the in ltration of TILs in stromal tissue. The quantity of TILs in the sample was determined by the percentage of TILs covering the interstitial tissue area, and was de ned as 1, lymphocyte in ltration < 10%; 2, 10-40%; 3, 40% (Figure 2 A, TILs<10%; B,10%≤ TILs <40%; C, TILs≥40%.).
The TIME classi cation into four groups based on the PD-L1 and CD4/CD8 expression status has been proposed, as follows: type-I (PD-L1 positive expression with CD4/CD8 positive expression), type-II (PD-L1 negative expression with CD4/CD8 negative expression), type-III (PD-L1 positive expression with CD4/CD8 negative expression), and type-IV (PD-L1 negative expression with CD4/CD8 positive expression). And, the relationship between TIME and DFS was also analysed.

Statistical analysis
Data were analyzed by using SPSS 24.0 software. Qualitative variables were compared using Chi-square test and Fisher's exact test. Logistic regression was used for multivariate analysis. DFS was analysed rstly by Kaplan-Meier and log-rank test and later with Cox regression to adust for covariates. P-value were two sides and a P<0.05 was considered statistically signi cant.

Patients Characteristics
A total of 156 surgically resected FFPE primary HER2-positive breast carcinoma patients. The mean age at surgery was 49 years (range, 28-73years). The mean tumor size was 2.5cm (range, 0.5-7.0cm) The majority of tumors (63.5%) were grade 3, while 50% of patients had lymph node metastasis and 45.5% of patients had lymphovascular invasion .Patients with a medium or high level of TILs accounted for 46.2%.There was a median follow up of 57 months (range 36-104 months).(Table1). Assessment of PD-L1 expression and immunereaction in 156 HER2-positive breast carcinomas Among all 156 cases, PD-L1 expression was identi ed in 49 cases (31.4%). PD-L1 expression was positively associated with tumor size, lymphovascular invasion ,TILs, CD4+ cells and CD8+ cells ( Table   2); obtained using Logistic multivariate regression analysis the tumor size, the presence or absence of lymphovascular invasion, the expression of TILs and CD8 were independent factors affecting PD-L1 expression (Table 3).  (Table 4). Kaplan-Meier curves of DFS were also plotted with several signi cantly associated factors including with tumor size, TILs, lymphovascular invasion, CD4+ cells, CD8+ cells, among them big tumor size is associated with poor prognosis, lymphovascular invasion positive is associated with poor prognosis, lymph node metastasis positive is associated with poor prognosis, low TILs is associated with poor prognosis, CD4 positive expression is associated with poor prognosis, CD8 negative expression is associated with poor prognosis (Figure 3). Moreover, in TIME-CD4 patients with CD4-/PD-L1-had a better prognosis while those with CD4+/PD-L1expression had a worse prognosis; however in TIME-CD8 patients with CD8+/PD-L1-had a better prognosis while those with CD8-/PD-L1+ had a worse prognosis ( Figure 4). Multivariate analysis with Cox regression model indicated that had lymphovascular invasion, TILs, TIME-CD4 and TIME-CD8 were independent factor affecting DFS (Table 5).

Discussion
To our knowledge, the purpose of this cohort of HER2-positive breast cancer research was to evaluate the expression of PD-L1 in stromal l in ltrating immune cells and the in ltration level of tumor in ltrating lymphocytes (TILs) and the expression of other immunological biomarkers (CD4, CD8).Our study cohort included 156 patients who were HER2-positive. The expression rate of PD-L1 in 156 HER2-positive breast cancer mesenchymal immune cells in this study was 31.4% (49/156). In this study, it was found that the expression of PD-L1 in breast cancer with HER2-positive was signi cantly correlated with the tumor size, lymphovascular invasion ,TILs in ltration level, CD4+ cells and CD8+ cells TILs mainly includes different types of T lymphocytes, B lymphocytes and NK cells, among which T lymphocytes that mainly play an anti-tumor role are divided into CD4+T cells, CD8+T cells and regulatory T cells. The type and in ltration level of TILs are related to the prognosis of tumor patients [10][11] . In HER2-positive breast cancer, patients with high TILs in ltration levels often have a better prognosis [12] . Raphael et al. [13] reported that in breast cancer with HER2-positive, the in ltration level of TILs was positively correlated with the histological grade, that is, the higher the histological grade was, the higher the in ltration level of TILs was. It suggested that TILs in ltration level was closely related to tumor differentiation. This study found that TILs in ltration level was positively correlated with the expression of PD-L1. Some studies have indicated that TILs in ltration level affects the responsiveness of breast cancer patients with HER2-positive to neoadjuvant chemotherapy and trastuzumab [14] . Therefore, the study of related factors affecting TILs in breast cancer and intervention may help improve the anti-tumor immune function of patients, improve the clinical tumor treatment effect and improve the prognosis of patients.
For breast cancer patients receiving standardized treatment, whether PD-L1 expression is a favorable or unfavorable prognostic factor is a subject of con icting results in the literature. Most of the published studies have involved different subtypes of breast cancer, with the exception of several studies that speci cally targeted TNBC [15,16] . Reported inconsistencies may be related to evaluation methods, numerical thresholds, antibody clones, or differences in cohort composition. In our cohort, patients' DFS was signi cantly correlated with tumor size, TILs, lymphovascular invasion, CD4+ cells, CD8+ cells, TIME-CD4 and TIME-CD8, while PD-L1 expression was not signi cantly correlated with patients' DFS. Since PD-L1 mediates checkpoint immune escape by suppressing the immune response of tumor cells, it has been speculated that the expression of PD-L1 may be associated with poor prognosis. However, studies of breast cancer and many tumors, including lung cancer and melanoma, have shown that PD-L1 expression is associated with better prognosis [17][18][19] .Therefore, the expression of PD-L1 in most cases may re ect a strong primary immune response rather than a successful immune escape. One possible explanation for these ndings is that the expression of PD-L1 in immune cells in the tumors studied was an adaptive response to a cytotoxic immune attack on tumor neoantigens, rather than PD-L1 produced by tumor cells based on a gene activation pathway [20] . This will make these tumors more sensitive to driver pathway blocking therapy (such as HER2) or checkpoint immune pathway blocking therapy. The signi cance of different immune response patterns and intratumoral and periatumoral patterns for checkpoint blocking therapy must await data from immunotherapy trials. The biggest limitation of this study may lie in the use of tissue chips. TAMs may not be representative, increasing the possibility of false negative results.
In this study, we analyzed for the rst time the relationship between immune typing and progression-free survival in HER2-positive patients. The results showed that the type of CD8+/PD-L1+ had the best prognosis, while the type of CD8-/PD-L1+ had the worst prognosis. The TILs in ltration score was positively correlated with prognosis. There are several limitations to our study. First, the number of patients participating in this study is small, re ecting the di culty of obtaining the right number of matched samples for such studies. Second, for IHC analysis of PD-L1 expression and CD8+ TILs, we used a relatively standardized scoring system. However, different PD-L1 scoring schemes have been used to determine the treatment of each PD-L1 inhibitor available on the market. Although there is currently a standardized interstitial CD8+ TIL scoring system, there is no standardized CD8+ TIL scoring protocol for breast cancer.
In conclusion, our data indicate that the type of CD8+/PD-L1+ and TILs in ltration levels predict the prognosis of HER2-positive breast cancer treated with standard therapy. There are several shortcomings in this study: the number of cases was small, the tissue chip had some limitations, the heterogeneity of the tumor itself and the protein loss or insensitivity caused by the wax block stored for a long time. Some results have yet to be con rmed by big data. These ndings, which are supported by other studies, offer hope for predicting immune checkpoint therapy.

Conclusions
This is a retrospective study of HER2-positive breast cancer, the results of this study indicate the independent in uencing factors of disease-free survival in patients with HER2-positive breast cancer are TIME-CD4 and TIME-CD8. Among them, patients with CD4-/PD-L1-group had better prognosis in the TIME-CD4 and CD8+/PD-L1-group had better prognosis in the TIME-CD8. Hope this study can provide some reference for immunotherapy of HER2-positive breast cancer. Availability of data and materials The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Consent for publication
Not applicable.

Funding
The study was supported by grants from the fourth hospital of Hebei Medical University. The funding body had no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.  Effect of TIME on DFS in HER2-positive breast carcinomas Kaplan-Meier survival analysis: : CD4-/PD-L1with better prognosis; B: CD8+-/PD-L1+ is associated with better prognosis;