TIM3 expression on TILs is associated with poor response to neoadjuvant chemotherapy in patients with locally advanced triple-negative breast cancer

Background The expression of immune checkpoint receptors (ICRs) on tumor-infiltrating lymphocytes (TILs) is associated with better response to immunotherapies via immune checkpoint inhibitors. Therefore, we investigated various ICR expressions on TILs in patients with locally advanced triple-negative breast cancer (TNBC) after neoadjuvant chemotherapy (NAC). Methods Expressions of ICRs were examined immunohistochemically in surgical specimens (n = 61) using monoclonal antibodies for PDL-1, PD-1, TIM-3, LAG-3, and CTLA-4. Positivity was defined as staining > 1% on TILs. Results The median age was 49 (24–76) years. The majority of patients were clinically T3–4 (n = 31, 50.8%) and clinically N1–3 (n = 58, 95.1%) before NAC. Of those, 82% were found to have CTLA-4 positivity, whereas PD1, PDL-1, LAG3, and TIM-3 expressions on TILs were 62.3, 50.9, 26.2, and 68.9%. A high expression of CTLA-4 was found to be associated with a better chemotherapy response (OR = 7.94, 95% CI: 0.9–70.12, p = 0.06), whereas TIM-3 positivity was contrarily associated with a worse chemotherapy response (OR = 0.253, 95% CI: 0.066–0.974, p = 0.047) as measured by the MDACC Residual Cancer Burden Index. At a 47-month follow-up, ypN0 (DFS; HR = 0.31, 95% CI: 0.12–0.83, p = 0.02 and DSS; HR = 0.21, 95% CI: 0.07–0.62, p = 0.005) and CTLA-4 high expression on TILs (DFS; HR = 0.38, 95% CI: 0.17–0.85, p = 0.019 and DSS; HR = 0.34, 95% CI: 0.15–0.78, p = 0.01) were found to be associated with improved survival. Conclusions These findings demonstrate that CTLA-4, PD-1, PDL-1, and TIM-3 were highly expressed in TNBC. Based on these high expression patterns, further studies directed towards combined therapies are warranted in advanced TNBC in future.


Background
As the interaction between the immune system and tumor progression is better understood, new treatment strategies have been developed [1,2]. A better understanding of this relationship has changed the treatment strategy for many cancers with the development of checkpoint inhibitors [3,4]. Many recent studies have demonstrated that the expression of immune checkpoint receptors (ICRs) on tumor-infiltrating lymphocytes (TILs) is associated with better response to immunotherapies, including immune checkpoint inhibitors [4][5][6][7][8].
Cytotoxic T Lymphocyte antigen 4 (CTLA-4), Programmed Cell Death 1 (PD-1), and Programmed Cell Death Ligand 1 (PD-L1, also known as B7-H1 or CD274) pathways are the best known immune checkpoint pathways for immune responses within the tumor microenvironment. Although considerable research has been devoted to CTLA-4, PD-1, and PD-L1, less attention has been paid to T cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) and Lymphocyteactivation gene 3 (LAG-3) [9]. The CTLA-4 antibody ipilimumab was the first checkpoint inhibitor for use in advanced melanoma [7]. Subsequently, the Food and Drug Administration (FDA) approved inhibitors of PD-1 and/or PD-L1 in different types of cancer. On the other hand, research on TIM-3-and LAG-3-based treatment options or combined checkpoint inhibitor therapy is still ongoing [10][11][12][13][14].
Triple-negative breast cancer (TNBC) constitutes about 15-20% of all diagnosed breast cancers [15][16][17]. Although TNBC is typically characterized by a high degree of aggressiveness and poor prognosis, there are no approved molecularly targeted therapies for it [18,19]. A recent meta-analysis reported an improved outcome in early TNBC with pathologic complete response (pCR) receiving neoadjuvant chemotherapy (NAC) [20]. Sequential administration of anthracyclines followed by taxanes has been the most common neoadjuvant approach in TNBC, and addition of carboplatin has also been considered to increase the pCR especially in those with BRCA mutations [21]. On the other hand, recent literature stresses that TNBC is more immunogenic than other breast cancer types [22][23][24]. Expression of ICRs on TILs has been associated with a better response to immunotherapies including immune checkpoint inhibitors (ICIs) as shown in advanced metastatic breast cancer by using combined modalities with chemotherapy [25]. Based on these encouraging results, recent studies focused on intensive investigations in neoadjuvant setting in combinations of ICIs with chemotherapeutic agents including anthracyclines, taxanes and platinum [26].
While early-phase studies of the immune checkpoint inhibitors in TNBC have shown evidence of activity, it is still a matter of debate as to which immune checkpoint marker will be used in treatment or which immune checkpoint marker has a prognostic value. Therefore, we investigated the expression of various new emerging ICRs, including TIM-3 and LAG-3, along with the best known PDL-1, PD-1, and CTLA-4 in patients with TNBC receiving NAC to explore their predictive and prognostic significance.

Methods
The study included 61 patients with triple-negative locally advanced breast cancer (LABC) who underwent surgery following NAC at the Istanbul University, Faculty of Medicine, Department of General Surgery, from July 2002 to January 2018. The study was approved by the ethical committee of Istanbul University, Istanbul Medical Faculty.

Patients and therapy
Patients with residual disease and pCR, who had received NAC due to the locally advanced TNBC were included in the study. The majority of patients (n = 57) received anthracycline and taxane containing NAC protocols, whereas 4 (6.6%) had an anthracycline-based regimen and 5 (6.6%) received additional carboplatine or cisplatine with/without gemcitabine or capecitabine. Of those, 45 (73.8%) underwent mastectomy and 26 (26.2%) had breast-conserving surgery. Furthermore, 53 patients (86.9%) underwent axillary dissection with/without sentinel lymph node biopsy (SLNB), whereas only 8 patients (13.1%) had SLN alone following completion of NAC. All patients received adjuvant radiotherapy to the chest wall for mastectomy and the whole breast for breastconserving surgery with axillary (level I-II-III) and supraclavicular lymph node region irradiation following surgery.
All patients with TNBC had estrogen receptor and progesterone receptor positivity < 1%, and HER2 negativity was determined according to the ASCO/CAP guidelines. The absence of a residual invasive tumor in the resected breast specimen and in the regional lymph nodes after completion of NAC was defined as pCR. The chemotherapy response was evaluated using the "MD Anderson Cancer Center (MDACC) Residual Cancer Burden Index" (www3.mdanderson.org/app/medcalc/ index.cfm.pagename = jsconvert) [27]. The residual cancer burden index was calculated from the parameters in pathology reports, including primary tumor bed area size, overall cancer cellularity, and percentage of in situ cancer. The MDACC Residual Cancer Burden Index classification implies a worsening chemotherapy response from "Class 0" to "Class 3". "Class 0" was considered a pathological complete response, whereas "Class 3" was considered chemotherapy-resistant. Briefly, response to chemotherapy is considered good if Class 0 or 1 and not as good if Class 2 or 3. Clinicopathologic data included age, tumor characteristics, operation type, clinical and pathological stage, histologic grade, and lymph node status, and any local and/or systemic recurrences and follow-up time were analysed.
The intensity and staining percentage were noted for each immune checkpoint receptor showing membranous staining. The intensity was scored as weak, moderate, or strong. Different staining percentages from 1 to 20% depending on the median values for each marker with or without staining intensity were tested for any significant associations with prognosis. CTLA-4 and PD-L1 positivity were defined as membranous staining > 1% on either tumor or TILs, whereas LAG-3, TIM-3, and PD-1 positivity on TILs was defined as membranous staining > 1% on TILs. All micrographs were taken by using an integrated digital camera (Olympus DP71, Japan) on a light microscope (Olympus BX51, Japan).

Statistical analysis
The statistical analyses were performed using the statistical software program SPSS 25 (Statistical Package for Social Sciences; SPSS, IBM Corp., Armonk, NY, USA). A p-value equal to or less than 0.05 was considered statistically significant. Categorical variables were evaluated by Chi Fisher's exact tests including Pearson chi-square, continuity correction, and Fisher's exact tests in twotailed univariate analyses. Binary logistic regression analysis was used to estimate the significant associations linked to chemotherapy response. Furthermore, the Pearson correlation test was used to assess the correlations of continuous variables.
Disease-free and disease-specific survival rates were calculated by Kaplan-Meier analyses, and factors associated with survival were tested by the log-rank test. Disease-free survival (DFS) was analysed by considering local and systemic metastases, and disease-specific survival (DSS) rates were analysed by considering breastcancer-related mortality. A log-rank test was used to compare the prognostic effect of different variables. Variables that were found to be significant in univariate analyses were further evaluated in a Cox regression model to calculate the hazard ratio of factors associated with poor prognosis.
Response to chemotherapy and histopathological analysis was evaluated in 60 patients by the "MDACC Residual Cancer Burden Index". One patient with a residual tumor in the lymphovascular area was excluded from this analysis. The median score was 2.72 (0-5.1). According to the MDACC Residual Cancer Burden classification, 26 patients (43.3%) had Class III as nonresponders, 21 patients (35%) had Class II as a partial response, 4 (6.7%) had Class I as a near-complete response, and 9 (15%) had Class 0 with a pathologic complete response.
Associations of ICR positivities and response to chemotherapy were tested by using the MDACC Residual Cancer Burden Index. A high expression of CTLA-4 (p = 0.047) and TIM-3-negativity (p = 0.003) were found to be significantly associated with pCR ( Table 2). When the MDACC Residual Cancer Burden Index "Class 0&1" vs "Class 2&3" were compared, high expression of CTLA-4 on TILs was found to be associated with a better chemotherapy response (OR = 7.94, 95% CI: 0.9-70.12, p = 0.06), whereas TIM-3 positivity was contrarily associated with a worse chemotherapy response (OR = 0.253, 95% CI: 0.066-0.974, p = 0.047). No significant associations could be found between other ICR expressions and response to NAC.

Discussion
Targeting immune checkpoints in cancer immunotherapy has made substantial progress in the treatment of certain solid cancers, including breast cancer, focusing on TNBC as the most immunogenic and aggressive subtype. However, long-term durable responses were obtained in only 10 to 30% of patients, while some patients develop resistance to the current immunotherapy regimens and may benefit from further combinatorial blockade targeting potential new immune checkpoints, such as LAG-3 or TIM-3 [28]. Furthermore, a significant proportion of patients suffer from immune-related adverse effects when treated with CTLA-4, PD-1, or PD-L1 inhibitors. The hypothesis regarding whether combining the currently known inhibitors targeting PD-1, PD-L1, or CTLA-4 with the promising new inhibitors against LAG-3 and TIM-3 may enhance therapeutic efficacy while reducing the side effects is currently under investigation in clinical trials [14]. In this study, the predictive and prognostic significance of various ICRs including PD-L1, PD-1, and CTLA-4, along with the new emerging receptors TIM3 and LAG3, was tested in locally advanced TNBC. We briefly report here that CTLA-4 was the only biomarker as ICR associated with a better chemotherapy response and better outcome, whereas TIM-3 expression was negatively correlated with response to NAC. Multiple ICRs, including PD-1, CTLA-4, TIM-3, and LAG-3, were found to be upregulated in breast cancer tissues as compared to normal breast tissue [29]. Despite some inconsistent results in the literature, the majority of the studies indicated an improved prognosis and better response to immunotherapy in patients with TNBC having PD-1 or PD-L1 expression on TILs [30,31]. In a recent meta-analysis including 47 studies with a total of 14, 367 patients who had breast cancer, Huang et al. [31] reported that PD-L1+ expression on TILs was 1.14-2.20, p = 0.006). In our study, we could not find any significant difference in DFS or DSS between patients with or without PD-L1 or PD-1 expression on TILs. Though there was a trend towards a better prognosis in patients with PD-L1 expression on TILs, it did not reach statistical significance.
Furthermore, we found a significant correlation between tumoral PD-L1 expression and PD-1 expression on TILs similar to the findings of the metaanalysis of Huang et al. [31]. The first inhibitory immune checkpoint receptor studied in the clinic, CTLA-4, was previously shown to be present as a surface molecule on solid tumor cell lines including breast carcinomas and osteosarcomas [32]. Though the exact function of expression of CTLA-4 on tumor cells is unknown, incubating the cells with CTLA-4 ligands CD80 or CD86 induced apoptosis in human osteosarcoma cell line   [34]. However, TIM-3, which has been studied as a newly emerging and promising coinhibitory molecule expressed on TILs, was contrarily found to be associated with a worse chemotherapy response (OR = 0.253, 95% CI: 0.066-0.974, p = 0.047) as measured by the MDACC Residual Cancer Burden Index in our study. Using an anti-TIM-3 antibody increased the response to paclitaxel chemotherapy in MMTV-PyMT transgenic mice as studied in models of triple-negative and luminal B disease, with no evidence of toxicity [35]. Yashinka et al. further suggested that the TIM-3-galectin-9 pathway may be involved in the immune escape of cancer cells [36]. They reported higher expression of both galectin-9 and TIM-3 in breast cancer tissues compared to healthy breast tissues of the same patients by  [37]. In multivariate analysis, the expression of TIM-3, PD-1, or LAG-3 on TILs was found to be an independent favorable prognostic factor among ER-negative patients. Similarly, Byun et al. reported a better prognosis in patients with TIM-3 expression on TILs in TNBC [38]. These studies also demonstrated that the presence of TIM-3 on TILs was significantly associated with the expression of other ICRs, including PD-1 and LAG-3 on TILs and PD-L1 on tumors. Our findings have shown that the majority of patients (69%) had TIM-3 positivity, and similarly indicated significant correlations between TIM-3 expression on TILs and LAG-3 or PD-1 or PD-L1 on TILs. Even though TIM-3 expression was negatively correlated with response to NAC in our study, we could not demonstrate any prognostic significance of these ICRs in our cohort of patients with TNBC following NAC presenting with more advanced disease compared to other studies. The prognostic significance of TIM-3 expression should be therefore studied in cases with TNBC including different stages in future.
Another novel coinhibitory receptor, LAG-3, which negatively regulates T-cell activation, has been reported to be present at a lower expression rate (18%) in early-stage TNBC as compared to PD-1 expression on TILs (30%), whereas coexpression of both receptors was found in 15.4% of breast-tumorassociated TILs [39]. Similarly, expressions of PD-1 and LAG-3 and concurrent expressions of PD-1 and LAG-3 on TILs were shown to be 62, 26, and 23%, respectively, in our study. However, the biological significance of the concurrent expression of inhibitory receptors such as PD-1 and LAG-3 is unknown. No prognostic significance could be found in patients with LAG-3 expression alone or concurrent LAG-3 and PD-1 expression in the study of Bottai et al. in concordance with our study. However, a recent meta-analysis of 15 studies including different cancers indicated that LAG-3 expression was associated with better OS (HR = 0.81, 95% CI: 0.66-0.99, p = 0.04), with a greater magnitude in early-stage malignancies (HR = 0.73, 95% CI: 0.60-0.88) and an improved DFS in breast cancer in a subgroup analysis (HR = 0.64, 95% CI: 0.42-0.98) [40]. Finally, Burugu et al. [41] studied LAG-3 expression in 4322 breast cancer specimens by immunohistochemistry and reported an improved outcome in patients with LAG-3 expression on TILs with a breast-cancerspecific survival (HR = 0.71, 95% CI: 0.56-0.90), particularly among estrogen receptor-negative patients (HR = 0.50, 95% CI: 0.36-0.69). Furthermore, concurrent LAG-3 expression with PDL1 and PD1 was 53 and 61%, respectively. These values seem to be higher than our findings and Bottai's study [39], which might be due to the different clinicopathological characteristics and stages of patients in the study cohorts.
We recently have studied the expressions of immune checkpoints receptors including PD-1, LAG-3, TIM-3, TIGIT and CTLA-4 on CD8 T lymphocytes and Natural Killer (NK) cell subsets obtained from TILs of patients with LABC following NAC by flowcytometry [42]. Our results suggested that HER2+ patients were more likely to have increased TIM-3 expressions on cytotoxic CD8-T cells, and patients with a younger age and advanced tumor burden are more likely to express ICRs on TILs compared to others. More studies are warranted to understand the precise role of LAG-3 or TIM-3 coexpressions on TILs by using different techniques including flow-cytometry or multiplex immunohistochemistry stainings in larger cohorts of TNBC with different stages.

Conclusion
Our findings indicate that novel immunological biomarkers, including TIM-3 and LAG-3, were highly expressed in TNBC after NAC similar to CTLA-4, PD-1, and PD-L1 showing different predictive and prognostic features. Based on these high expression patterns, combined immune checkpoint inhibitor therapies via CTLA-4 and/or PD-1 or PDL-1 along with TIM-3 or LAG-3 are to be investigated in future studies, which may improve the prognosis and pathologic complete response rates of patients undergoing chemotherapy [6, 9-12, 14, 43, 44].

Availability of data and materials
The datasets during and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate The study was approved by the ethical committee of Istanbul University, Istanbul Medical Faculty. in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration. A written informed consent was obtained from all individual participants included in the study.

Consent for publication
Not applicable.