Predictive value of improvement in the immune tumour microenvironment in patients with breast cancer treated with neoadjuvant chemotherapy

Background Tumour-infiltrating lymphocytes (TILs) can be used to monitor the immune tumour microenvironment (iTME) and predict treatment response and outcome in breast cancer. We evaluated the prognostic significance of the levels of CD8+ TILs and forkhead box protein (FOXP3)-positive TILs before and after neoadjuvant chemotherapy (NAC). Patients and methods We examined 136 patients with breast cancer treated with NAC. The number of CD8+ TILs and FOXP3+ TILs in biopsy specimens and residual tumours was evaluated by immunohistochemistry. Results Patients with a high rate of change in the CD8/FOXP3 ratio (CFR) had significantly better recurrence-free survival (RFS) (p<0.001, log-rank). In multivariate analysis, the rates of change in the CD8+ TIL levels and the CFR were independent predictors for RFS (HR=2.304, p=0.036 and HR=4.663, p<0.001). In patients with triple-negative and hormone receptor-positive breast cancer, the rate of change in the CFR was an independent predictor for RFS (HR=13.021, p=0.002 and HR=4.377, p=0.003). Conclusion Improvement in the iTME following NAC is correlated with good outcome. The rate of change in the CFR may be a useful biomarker to predict prognosis of patients treated with NAC.


AbstrAct
Background Tumour-infiltrating lymphocytes (TILs) can be used to monitor the immune tumour microenvironment (iTME) and predict treatment response and outcome in breast cancer. We evaluated the prognostic significance of the levels of CD8 + TILs and forkhead box protein (FOXP3)positive TILs before and after neoadjuvant chemotherapy (NAC). Patients and methods We examined 136 patients with breast cancer treated with NAC. The number of CD8 + TILs and FOXP3 + TILs in biopsy specimens and residual tumours was evaluated by immunohistochemistry. Results Patients with a high rate of change in the CD8/ FOXP3 ratio (CFR) had significantly better recurrence-free survival (RFS) (p<0.001, log-rank). In multivariate analysis, the rates of change in the CD8 + TIL levels and the CFR were independent predictors for RFS (HR=2.304, p=0.036 and HR=4.663, p<0.001). In patients with triple-negative and hormone receptor-positive breast cancer, the rate of change in the CFR was an independent predictor for RFS (HR=13.021, p=0.002 and HR=4.377, p=0.003). Conclusion Improvement in the iTME following NAC is correlated with good outcome. The rate of change in the CFR may be a useful biomarker to predict prognosis of patients treated with NAC.

IntRoduCtIon
Neoadjuvant chemotherapy (NAC) is the gold standard of care for breast cancer and increases the options for breast-conserving surgery. [1][2][3] Pathological complete response (pCR) after NAC is currently acknowledged as an indicator of good outcome, especially in triple-negative breast cancer (TNBC) and human epidermal growth factor receptor 2 (HER2)-enriched breast cancer (HER2BC). 4 One previous study indicated that residual cancer cells after NAC may be more aggressive or have enhanced metastatic potential. 5 However, some patients who fail to achieve pCR after NAC have a relatively good outcome. Therefore, novel prognostic markers in residual tumours are needed to identify high-risk patients.
► Recently, the importance of regulating and improving the immune tumour microenvironment (iTME) has been reported to play an important role in predicting outcomes. ► Tumour-infiltrating lymphocytes (TILs) can be used to monitor the iTME and predict treatment response and outcome in breast cancer. ► CD8/FOXP3 ratio (CFR) in biopsy specimens before neoadjuvant chemotherapy (NAC) is a useful biomarker to predict treatment response to chemotherapy.
What does this study add?
► The predictive value of changes in lymphocytic subpopulations after NAC in all breast cancer subtypes has not been discussed sufficiently. ► The present study investigated the clinical significance and value of changes in the levels of CD8 + TILs and FOXP3 + TILs and the CFR before and after NAC in all breast cancer subtypes. ► To our knowledge, this is the first study to demonstrate the prognostic role of changes in CD8 + TIL levels, FOXP3 + TIL levels and the CFR in patients failing to achieve pathological complete response following NAC in all breast cancer subtypes.
How might this impact on clinical practice?
► These results suggest that by further evaluating the changes in other TILs, such as programmed death-1-positive TILs, along with those in CD8 + TILs and FOXP3 + TILs in patients treated with NAC, more accurate identification of patient-specific immune mechanisms and prediction of prognosis may be possible. ► Improvement in the iTME following NAC is correlated with good outcome. ► The rate of change in the CFR may be a useful biomarker to predict prognosis of patients treated with NAC.
important role in predicting outcomes. 6 7 Tumour-infiltrating lymphocytes (TILs) can be used to monitor the tumour microenvironment and are important in predicting treatment efficacy and clinical outcomes in many types of cancer, including breast cancer. [8][9][10] Various cells of the immune system can play varying roles in tumour progression; for instance, cytotoxic T cells (CD8 + T cells), natural killer cells, dendritic cells and macrophages are associated with improved clinical outcomes, whereas regulatory T (Treg) cells and myeloid-derived suppressor cells suppress antitumour immunity. 11 Specific TIL subsets, such as CD3 + , CD8 + and forkhead box protein 3 (FOXP3)-positive TILs, have been reported to be clinically significant and reliable in predicting treatment response. [12][13][14] In addition, since Treg cells suppress the induction of cytotoxic T cells in response to cancer cells, the CD8/FOXP3 ratio (CFR) has been reported to be associated with high pCR rates. 15 16 We have also suggested that the CFR in biopsy specimens before NAC is a useful biomarker to predict treatment response to chemotherapy in TNBC and HER2BC. 17 Chemotherapy enhances the immune activity or the reversal of immunosuppression. 7 18 Some studies revealed that changes in the levels of CD8 + or FOXP3 + TILs induced by chemotherapy can be used as a prognostic marker in aggressive breast cancer subtypes, such as TNBC. 15 19 20 However, the predictive value of changes in lymphocytic subpopulations after NAC in all breast cancer subtypes has not been discussed sufficiently.
The present study investigated the clinical significance and value of changes in the levels of CD8 + TILs and FOXP3 + TILs and the CFR before and after NAC in all breast cancer subtypes. To our knowledge, this is the first study to demonstrate the prognostic role of changes in CD8 + TIL levels, FOXP3 + TIL levels and the CFR in patients failing to achieve pCR following NAC in all breast cancer subtypes.

PatIents and metHods ethics
This study was conducted at the Osaka City University Graduate School of Medicine, Osaka, Japan according to the Reporting Recommendations for Tumour Marker Prognostic Studies guidelines and following a retrospectively written research proposal, pathological evaluation and statistical plan. 21 This study conformed to the provisions of the Declaration of Helsinki. All patients were informed of the investigational nature of this study and provided their written informed consent.

Patient background
A total of 214 patients with resectable, early-stage, primary infiltrating ductal breast cancer who were treated with NAC between 2007 and 2015 were included. Tumour stage and T and N factors were stratified based on the TNM Classification of Malignant Tumours, The Union for International Cancer Control Seventh Edition. 22 Tumours were classified into subtypes according to the immunohistochemical expression of oestrogen receptor (ER), progesterone receptor (PgR), HER2 and Ki-67. neoadjuvant therapy regimen and surgery All patients received a standardised NAC protocol consisting of four courses of FEC100 (500 mg/m 2 fluorouracil, 100 mg/m 2 epirubicin and 500 mg/m 2 cyclophosphamide) every 3 weeks, followed by 12 courses of 80 mg/ m 2 paclitaxel administered weekly. 23 24 Patients with HER2BC were additionally administered weekly (2 mg/ kg) or tri-weekly (6 mg/kg) with trastuzumab during paclitaxel treatment. 25 Clinical end points Therapeutic antitumour effects were assessed according to the Response Evaluation Criteria in Solid Tumors criteria. 26 The pCR was defined as the complete disappearance of the invasive compartment of the lesion with or without intraductal components, including the lymph nodes. 27 Overall survival (OS) was the period from the date of primary surgery to the date of death from any cause. Recurrence-free survival (RFS) was defined as the period from the date of primary surgery until the date of disease recurrence.

Immunohistochemical scoring
Immunohistochemical scoring was performed by two breast pathologists (MOhs and YK). The cut-offs for ER and PgR positivity were both >0% positive tumour cells with nuclear staining. Tumours with 3+HER2 on immunohistochemical staining were considered to have HER2 overexpression, tumours with 2+HER2 were analysed further by fluorescence in situ hybridisation and tumours with HER2/Centromere 17 ≥2.0 were also considered to exhibit HER2 overexpression. 28 Tumours with ≥14% Ki-67 nuclear staining were determined to be positive. 29 The assessment of unstained TILs was based on the criteria described by Salgado and colleagues. 11 TILs were evaluated within the stromal compartment close to the invasive tumour, and the proportion of stromal TILs was reported. In this study, unstained TILs were considered high if TILs occupied >10% of the stroma area and low when TILs occupied ≤10%. To evaluate CD8 and FOXP3 expression, four fields of view (FOV) in darkly stained areas were selected, and the number of TILs in the stroma surrounding the stained cancer cells in each FOV was measured microscopically at 400× magnification (figure 1). The mean number of CD8 + or FOXP3 + lymphocytes in each FOV was counted. The CFR was defined as the number of CD8 + TILs divided by the number of FOXP3 + TILs.

statistical analyses
Statistical analysis was performed using the JMP13 software programme (SAS Institute). The associations between levels of CD8 + TILs and FOXP3 + TILs and clinicopathological variables were analysed using Χ 2 tests or Fisher's exact tests, as appropriate. OS and RFS were estimated using the Kaplan-Meier method and compared using the log-rank test. Univariate and multivariate HRs were computed for the study parameters with 95% CIs using a Cox proportional hazards model and used in a backward stepwise method for variable selection in multivariate analyses. A p value <0.05 was considered significant.
Changes in Cd8 + tIls and FoXP3 + tIls before and after naC and their association with prognosis The mean rates of change in CD8 + TIL levels, FOXP3 + TIL levels and the CFR before and after NAC were as follows: CD8 + TILs, 1.2; FOXP3 + TILs, 0.5 and CFR, 2.3. Of 129 patients, 56 (43.4%) had a high rate of change in CD8 + TILs, 82 (63.6%) had a low rate of change in FOXP3 + TILs and 68 (52.7%) had a high rate of change in the CFR. In addition, 62 (48.1%) patients had a high rate of change in unstained TILs. Younger patients (≤56 years) were significantly more likely to have a high rate of change in CD8 + TIL levels than older patients (>56 years) (p=0.013). Patients with TNBC had a significantly higher rate of change in FOXP3 + TIL levels than patients with other subtypes (p=0.014) (table 1). Patients with a high rate of change in unstained TILs were significantly higher in patients with a high rate of change in CD8 + TIL levels (p<0.001) and a high rate of change in FOXP3 + TILs (p=0.003), but there was no significant correlation between the rate of change in unstained TILs and the CFR (p=0.479). Patients with a high rate of change in CD8 + TIL levels had significantly better RFS and OS than those with a low rate of change (p=0.005, p=0.032, log-rank, respectively). Patients with a low rate of change in FOXP3 + TIL levels had significantly better RFS and OS than those with a high rate of change (p=0.044, p=0.025, log-rank, respectively). Patients with a high rate of change in the CFR also had significantly better RFS and OS than those with a low rate of change (p<0.001, log-rank, both end points) (figure 2).
In univariate analysis, pathological response (HR=6. 33 3). Since the number of patients with HER2BC was small (n=20), it could not be analysed.

dIsCussIon
In the present study, the proportion of unstained TILs, the number of CD8 + TILs and the CFR increased, and the number of FOXP3 + TILs decreased in breast tumours after NAC. Those results indicated that a regimen of FEC followed by paclitaxel±trastuzumab enhances antitumour immunity and reversal of immunoescape in cancer cells. This improvement in the immune microenvironment following NAC was significantly correlated with prognosis. TILs are mononuclear immune cells in the tumour microenvironment. Infiltration of TILs before NAC is a useful biomarker to predict treatment response in patients with TNBC and HER2BC, two subtypes of highly malignant breast cancer. 11 30-33 In these subtypes, high TILs group before NAC is significantly associated with higher pCR rate, good prognostic factor. In our study, investigating patients with non-pCR after NAC, the proportion of unstained TILs alone was not a useful predictor of outcome (online supplementary figure 3). Therefore, more detailed evaluation of TILs becomes necessary. In a previous study, Ladoire et al examined changes in the levels of CD8 + TILs and FOXP3 + TILs after NAC in 56 patients with breast cancer and reported that a high rate of change in the CFR was associated with pCR. 20 Miyashita et al also analysed 78 patients with TNBC and reported that high rates of change in the level of CD8 + TILs and the CFR were significantly correlated with good RFS and OS. 15 However, there have been few studies stratifying by intrinsic subtype of breast cancer. Our study is the first to indicate that a high rate of change in the CFR is an independent prognostic factor for good outcome in patients with TNBC and HRBC who do not achieve pCR after NAC. Figure 2 Analysis of the rate of changes in CD8 + TILs, FOXP3 + TILs and the CFR and RFS and OS in patients with all breast cancer subtypes. Patients with a high rate of change in CD8 + TIL levels had significantly better RFS (A) and OS (B) than those with a low rate of change (p=0.005, p=0.032, log-rank, respectively). Patients with a low rate of change in FOXP3 + TIL levels had significantly better RFS (C) and OS (D) than those with a high rate of change (p=0.044, p=0.025, log-rank, respectively). Patients with a high rate of change in the CFR also had significantly better RFS (E) and OS (F) than those with a low rate of change (p<0.001, log-rank, both end points). Goto et al. 17 CFR, CD8/FOXP3 ratio; FOXP3, forkhead box protein 3; OS, overall survival; RFS, recurrence-free survival; TIL, tumour-infiltrating lymphocyte. Goto  TILs are mononuclear immune cells in the tumour microenvironment. Infiltration of TILs is a useful biomarker to predict treatment response in patients with TNBC and HER2BC, two subtypes of highly malignant breast cancer. 11 30-33 These studies suggest that TNBC and HER2BC have high immunoactivity. However, based on the detailed subclassification of TILs as CD8 + TILs or FOXP3 + TILs, studies evaluating the prognostic significance of CD8 + TILs or FOXP3 + TILs or the CFR in the intrinsic molecular subtypes of breast cancer have shown conflicting results. 15 17 34-40 One possible explanation consistent with these discrepant findings is that HRBC is also considered to be associated with some kind of immunity.
The CFR reflects the interplay between CD8 + TILs and Treg cells in a tumour and indicates the activity of the immune microenvironment. A higher CFR has been shown to be significantly associated with better survival in hormone receptor-negative tumours. [15][16][17] In the present study, we focused on the rate of changes in the CFR induced by NAC and demonstrated that an increase in the CFR was significantly associated with improved clinical outcomes in not only TNBC but also HRBC. This result suggests that the change in the CFR after NAC may be a more accurate indicator of immune activity induced by chemotherapy. The recently identified immune checkpoint markers, programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1), are present in some breast cancers. 41 The PD-1/PD-L1 axis, a major immune checkpoint pathway, leads to a reduction in the immune response by inducing T cell tolerance. 42 In previous studies, Muenst et al reported that PD-1 + TILs are significantly associated with worse OS in the luminal B and basal-like subtypes. 43 Ali et al also reported that PD-L1 expression is significantly enriched in the basal-like subtype and is correlated to the presence of TILs. 44 In addition, one recent study indicated that PD-L1 expression in residual tumours is significantly associated with the levels of CD8 + TILs and FOXP3 + TILs and may be a useful prognostic marker in patients with breast cancer following NAC. 45 These results suggest that by further evaluating the changes in other TILs, such as PD-1 + TILs, along with those in CD8 + TILs and FOXP3 + TILs in patients treated with NAC, more accurate identification of patient-specific immune mechanisms and prediction of prognosis may be possible.
As a potential limitation, since the sample size of our study was small, we did not evaluate the relationship between HER2BC and the rate of change in the levels of CD8 + TILs and FOXP3 + TILs or the CFR. In addition, we analysed without dividing patients with HRBC into a HER2-positive and HER2-negative group. This is the first study to indicate that improvement in the immune microenvironment following NAC has a relationship with good outcome, and that a high rate of change in the CFR could be a potential prognostic marker in patients with TNBC and HRBC who do not achieve pCR after NAC .
acknowledgements The authors would like to thank Yuko Kuwae (YK) from the Department of Diagnostic Pathology, Osaka City University Graduate School of Medicine for providing assistance with immunohistochemical scoring. We also thank Yayoi Matsukiyo and Tomomi Okawa from the Department of Surgical Oncology, Osaka City University Graduate School of Medicine for their helpful advice regarding data management.
Contributors All authors were involved in the preparation of the manuscript. WG collected the data and wrote the manuscript. SK, YA, KTakad, KTakah, TH and TT performed the operation and designed the study. WG, SK and ST summarised the data and revised the manuscript. MOhs performed the pathological diagnosis. HM, KH and MOhi made substantial contribution to the study design, performed the operation and revised the manuscript. All authors read and approved the final manuscript.
Funding This study was supported in part by Grants-in Aid for Scientific Research (KAKENHI, nos 25461992, 26461957 and 17K10559) from the Ministry of Education, Science, Sports, Culture and Technology of Japan.
Competing interests Not declared.

Patient consent Obtained.
ethics approval The Ethics Committee of Osaka City University approved the study protocol (no 926).

Provenance and peer review Not commissioned; externally peer reviewed.
open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https:// creativecommons. org/ licenses/ by/ 4. 0/.