Background Tumour-infiltrating lymphocytes (TILs) have a strong prognostic significance, particularly in triple-negative breast cancer (TNBC). One important source of TILs in breast cancer is tertiary lymphoid structures (TLSs).
Objective To carry out a histological analysis of surgically resected TNBC to identify the location of TLSs, the relationship between TLSs and TILs and their prognostic significance in TNBC.
Methods We retrospectively analysed 769 patients with TNBC.
Results TILs were defined as the percentage of stroma of invasive carcinoma infiltrated by lymphocytes. TLSs were mainly present within adjacent terminal duct lobular units and around in situ components. TNBC with higher levels of TILs showed a higher nuclear grade, lower lymphovascular invasion, less accompanying in situ component, a homogeneous growth pattern, necrosis in invasive areas, low levels of tumour stroma, high levels of peritumoral lymphocytic infiltration and moderate to abundant TLSs in adjacent tissue. TILs, the degree of peritumoral lymphocytic infiltration and adjacent TLSs were prognostic factors for disease-free and overall survival. Although the TIL level did not have a prognostic value in stage I, it added significant prognostic information for stages II and III. Conversely, patients with high levels of TILs did not show prognostic differences according to the pTNM stage. Patients with high levels of TILs (>60%) and moderate to abundant TLSs had significantly better disease-free survival than those with high levels of TILs but none or few TLSs.
Conclusions TLSs are frequently present in TNBC and are closely associated with TILs. TILs provide additional prognostic information in patients with TNBC with a higher pTNM stage.
- BREAST CANCER
- TUMOUR IMMUNITY
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The importance of tumour-infiltrating lymphocytes (TILs) in adjuvant and neoadjuvant settings has been determined for breast cancer. In general, increased lymphocytic infiltration in tumours is inversely correlated with oestrogen receptor (ER) and/or progesterone receptor (PR) expression, but is positively correlated with the pathological complete response rate and increased patient survival.1–6 Although different methods have been used for assessing the presence of TILs in various studies, TILs have a strong prognostic significance, particularly in triple-negative breast cancer (TNBC). TNBC has no definite targeted treatment but a growing body of evidence suggests that immunotherapy has great potential for combating the disease.7 Therefore, a better understanding of TILs and related features and the mechanism of TIL influx might facilitate the development of efficient immunotherapeutic approaches.
Recently, the presence of tertiary lymphoid structures (TLSs) has been identified in solid organ tumours, including breast, colorectal, lung, ovarian and renal cell carcinomas and malignant melanoma.8 ,9 TLSs are ectopic lymph node-like structures characterised by lymphoid aggregation with high endothelial venules (HEVs). Similar to secondary lymphoid organs, T and B lymphocytes in TLSs are present in separate areas: the T cell-rich zone consists of T cells and mature dendritic cells, whereas the B cell-rich area consists of B cells, follicular helper T cells, follicular dendritic cells and macrophages and may contain a germinal centre.9 HEVs are specialised blood vessels with plump, cuboidal endothelial cells. The endothelial cells of HEVs express peripheral node addressin (PNAd), and the PNAd ligand CD62L is expressed in lymphocytes. Hence, the expression of PNAd in HEVs facilitates the extravasation of lymphocytes from the blood vessel to the tissue.10 Martinet et al8 showed that high densities of tumour HEVs correlated with increased naïve, central memory and activated effector memory T cell infiltration and T cell cytotoxicity in human primary solid tumours, including breast, colon, ovarian and lung carcinomas and melanomas. In addition, these authors reported a significantly longer disease-free survival in patients with breast cancer with high-density HEVs than in those with low-density HEVs. However, TLS location and the relationship between TILs and TLSs in TNBC have not been clearly elucidated.
Advances in molecular biology have highlighted the significance of tumour microenvironments in tumorigenesis.11 Stromal cells can promote tumour growth, angiogenesis, invasion and metastasis by various mechanisms.12 The prognostic significance of tumour stroma has been reported in breast cancer.13–16 Gujam et al16 also showed a correlation between a high percentage (>50%) of tumour stroma and low-grade tumour inflammatory infiltration. However, those authors only evaluated the relationship between tumour stroma and peritumoral lymphocytic infiltration without evaluation of TILs in the interior of invasive tumours.
To identify the location of TLSs and the relationship among TLSs, TILs and tumour stroma in TNBC, we performed the first comprehensive histological analysis of surgically resected TNBC from a large series of Asian patients. We compared clinicopathological factors among tumours with different amounts of stromal TILs and analysed the correlations among various factors. We found a prognostic significance of TILs and related factors in TNBC.
Patients, materials and methods
Patients and tissue specimens
A total of 769 patients with TNBC who underwent surgery for primary breast cancer between 2004 and 2010 at Asan Medical Center and who had formalin-fixed, paraffin-embedded, tissue samples for analysis were included in this study. All patients were preoperatively chemotherapy and radiotherapy naïve and all but one underwent adjuvant treatment. Of the 769 patients, 518 were treated with four cycles of adjuvant anthracycline and cyclophosphamide (AC; Adriamycin (doxorubicin) 60 mg/m2 and cyclophosphamide 600 mg/m2). The remaining 250 patients were treated with either four cycles of AC followed by four cycles of paclitaxel (175 mg/m2) or four cycles of AC followed by four cycles of docetaxel (75 mg/m2). Most patients treated with AC (95.9%) had no lymph node metastasis, whereas most patients treated with AC and taxane (98.4%) had lymph node metastasis. Radiotherapy was performed in 78.5% of patients. Clinicopathological information was obtained from the patients' medical records and surgical pathological reports.
Expressions of standard biomarkers including ER, PR and HER2 (human epidermal growth factor receptor 2) were reviewed in whole sections that were immunohistochemically stained at the time of diagnosis. ER and PR levels were regarded as positive if there was at least 1% positive tumour nuclei staining.17 HER2-overexpressing tumours were defined as those with scores of 3+ by immunohistochemistry or gene amplification by fluorescence in situ hybridisation (FISH) or silver in situ hybridisation (SISH).18 FISH or SISH was performed for samples with an immunohistochemical score of 2, and HER2 amplification was not seen in any case. Exemption from informed consent after de-identification of information was approved by the institutional review board of Asan Medical Center.
The H&E-stained slides were reviewed by two pathologists (HJL and GG). Slides were histopathologically analysed for (i) TILs (defined as the mean percentage of stroma of invasive carcinoma infiltrated by lymphocytes and plasma cells in 10% increments; if <10% of stroma was infiltrated by TILs, 1% or 5% criteria were used; all available full sections were evaluated);1 ,19 (ii) the amount of inflammatory cells evaluated at the tumour's invasive margin (according to Klintrup criteria; score 0, no inflammatory cells at the invasive margin; score 1, mild and patchy increase of inflammatory cells; score 2, increased inflammatory cells forming a band-like infiltration at the invasive margin; score 3, prominent inflammatory reaction forming a cup-like zone at the invasive margin);6 ,20 (iii) aggregation of lymphoid cells in terminal duct lobular units (TDLUs; within 5 mm and >5 mm from the invasive or in situ carcinoma); (iv) the amount of TLSs in adjacent tissue including the in situ component (none, no TLS formation in the area adjacent to the tumour; little, TLSs occupying an area of <10% of the circumference of the tumour; moderate, 10–50%; or abundant, >50%; shown in figure 1); (v) the presence or absence of a germinal centre in TLSs; (vi) the percentage of tumour stroma in the most invasive tumour area (high percentage of tumour stroma, >50%; low percentage of tumour stroma, ≤50%);16 (vii) heterogeneity of the tumour (heterogeneous: invasive tumour mixed with normal breast parenchyma; homogeneous: invasive tumour mass without normal breast parenchyma); (viii) histological subtype and grade, percentage of in situ component, tumour size, pT stage, pN stage and lymphovascular invasion. Lymphoid aggregation with vessels showing HEV features (plump, cuboidal endothelial cells) with or without germinal centres was considered a TLS. Histological type was defined based on the 2012 WHO classification criteria and histological grade was assessed using the modified Bloom–Richardson classification.21
To compare the degrees of TLSs assessed histologically and immunohistochemically, 20 consecutive whole tumour sections were evaluated with an automated immunohistochemical staining device (Benchmark XT; Ventana Medical Systems, Tucson, Arizona, USA). Antibody to MECA-79 (1:200 dilution; Santa Cruz Biotechnology, Santa Cruz, California, USA), which recognises sulfate-dependent carbohydrate epitopes of PNAd expressed in endothelial cells of HEVs, was used. The presence of MECA-79-positive blood vessels was considered a positive indicator of HEVs. To test the specificity of MECA-79 for HEVs, we immunostained CD31 (1:200 dilution; NOVO, Newcastle, UK) and MECA-79 on the lymph node, haemangioma and normal TDLU (see online supplementary figure S1). To find basal-type TNBC, antibodies to epidermal growth factor receptor (EGFR; EGFR pharmDxTM; DAKO, Glostrup, Denmark) and CK5/6 (1:200 dilution; DAKO) were also used. For cytokeratin 5/6 and EGFR, any positive cytoplasmic and membranous staining was considered as positive.22
All statistical analyses were performed using SPSS statistical software (V.18; SPSS, Chicago, Illinois, USA). Kruskal–Wallis one-way analysis of variance, χ2 test, linear-by-linear association test, Cohen's κ coefficient, Pearson's correlation, log-rank test and Cox proportional hazards regression model were used as appropriate. All tests were two-sided and statistical significance was set at 5%.
Clinicopathological characteristics of the study population
All but one of the 769 patients were women and their median age at diagnosis was 47 years (range 23–76 years). Invasive tumour sizes ranged from 0.3 to 10 cm (median 2.2 cm). There were 332 pT1 tumours, 407 pT2 tumours, 29 pT3 tumours and 1 pT4 tumour. In lymph node analysis, 502 tumours were pN0 stage, 163 were pN1, 58 were pN2 and 46 were pN3.
Comparison of tumour characteristics according to the amount of TILs
To identify the clinicopathological significance of TILs, we divided the tumours into four groups: ≤10% TILs (188 cases, 24.4%), 20–30% TILs (171 cases, 22.2%), 40–60% TILs (160 cases, 20.8%) and 70–100% TILs (250 cases, 32.5%). A summary of the clinicopathological characteristics of each group according to the amount of TILs is shown in table 1. Patients with higher levels of TILs were younger than those with lower levels of TILs. Adenoid cystic carcinoma, carcinoma with apocrine differentiation, metaplastic carcinoma and invasive lobular carcinoma were more closely associated with lower levels of TILs, whereas carcinoma with medullary features and invasive micropapillary carcinoma showed higher levels of TILs. Histological grade was higher in tumours with higher levels of TILs. The presence of lymphovascular invasion and percentage of in situ component were lower in tumours with higher levels of TILs. The higher degree of peritumoral lymphocytic infiltration (higher score of Klintrup criteria), a homogeneous growth pattern and necrosis in the invasive tumour area were closely associated with higher levels of TILs. High levels of tumour stroma were more frequently seen in tumours with lower levels of TILs. Most cases (96.6%) showed lymphoid aggregation in TDLUs located within 5 mm of the carcinoma. However, only 26/769 cases (3.4%) also showed lymphoid aggregation in TDLUs located more than 5 mm from the carcinoma. To increase the reliability of this analytical assessment, we dichotomised the patients by the date of surgery. Carcinoma with medullary features, higher histological grade, higher score of Klintrup criteria and presence of necrosis in invasive area were more frequently seen in tumours with higher levels of TILs in both cohorts (see online supplementary table S1 and S2).
Association between TILs and TLSs
TLSs are mainly present in adjacent TDLUs and around carcinoma in situ (figure 2). Among the total 769 cases, 56 tumours (7.3%) showed no TLSs, while the remaining tumours had various amounts of TLSs: little, 132 tumours (17.2%); moderate, 278 tumours (36.2%); abundant, 303 tumours (39.4%, figure 1). The degrees of TLSs in the adjacent area assessed histologically and immunohistochemically showed excellent agreement (κ value=0.831, p=0.005, online supplementary table S3). Larger amounts of TLSs in adjacent tissue, and the presence of germinal centres in adjacent tissue, were closely associated with higher levels of TILs (table 1). Abundant TLSs and presence of germinal centres in adjacent tissue also showed significant association with higher levels of TILs in both dichotomised patients groups according to the date of surgery (see online supplementary table S1 and S2).
Prognostic significance of TILs and TLSs
Next, the prognostic significance of TILs and related features were analysed (table 2). In univariate analysis, higher pT, presence of lymphovascular invasion and lymph node metastasis, higher pTNM stage, chemotherapy regimen with AC and taxane and lower levels of TILs, peritumoral lymphocytic infiltration and TLSs in adjacent tissue were more negative prognostic factors for disease-free and overall survival outcomes. In multivariate analysis using conditional forward stepwise algorithm, pTNM stage, presence of lymphovascular invasion and TILs were found to be independent prognostic factors for disease-free survival: pTNM stage, HR=1.744, 95% CI=1.314 to 2.314, p<0.001; presence of lymphovascular invasion, HR=1.968, 95% CI 1.322 to 2.930, p=0.001; TILs, HR=0.982, 95% CI 0.975 to 0.989, p<0.001. All these were also independent prognostic factors for overall survival: pTNM stage, HR=1.630, 95% CI 1.187 to 2.238, p=0.003; presence of lymphovascular invasion, HR=1.940, 95% CI 1.238 to 3.038, p=0.004; TILs, HR=0.980, 95% CI 0.972 to 0.989, p<0.001. In addition, to exclude the possibility that adjuvant chemotherapy affected the patients' survival through different interaction with the immune compartment, univariate analysis of clinicopathological variables, including TILs and TLSs, affecting clinical outcomes was performed in each patient group according to the chemotherapy regimens (see online supplementary table S4 and S5). In both groups, TILs and TLS were established as prognostic factors both for disease-free and overall survival, showing that their prognostic effects are indeed independent of chemotherapy types.
We assessed the contribution of the TIL level to the prognostic power of each pTNM stage and vice versa. Although the TIL level did not have prognostic value in stage I, it did have significant prognostic information for stage II and III patients (figure 3). Conversely, pTNM stage had prognostic significance in patients with low to moderate levels of TILs, but patients with high levels of TILs did not show prognostic differences according to pTNM stage (figure 4).
Even though the TIL level and degree of adjacent TLSs correlated well (r=0.548, p<0.001), some tumours with moderate to abundant TLSs did not have high levels of TILs in the interior of the invasive cancer. Therefore, we divided tumours into four groups according to TIL level (TILs ≤60% vs TILs 70–100%) and degree of adjacent TLSs (none or few vs moderate to abundant). As shown in figure 5, patients with high levels of TILs and moderate to abundant TLSs had significantly better disease-free survival than those with low levels of TILs but moderate to abundant TLSs.
To the best of our knowledge, this study is the first large comprehensive report on the importance of TILs and TLSs in TNBC. We found that TLSs are mainly present in adjacent TDLUs and around carcinoma in situ. The presence of TLSs in breast cancer has been described by other groups. For example, Gu-Trantien et al23 found that “TLS and germinal centre were detected adjacent to the tumour bed at a higher incidence in extensively compared with minimally lymphocyte infiltrated tumours” and Martinet et al8 stated that “We found a preferential localization of HEVs around in situ tumour components”. We also found the presence of HEVs in the interior of the invasive tumour area by MECA-79 immunohistochemistry in tissue microarray sections.
The presence of TLSs was described in tissue with chronic inflammation that can be associated with infection, autoimmune disease and chronic allograft rejection.24 Tumour HEVs are reported to be structurally and phenotypically similar to the HEVs of chronic inflammation.8 In lymph nodes, CD11c+ dendritic cells play a critical role in modulating the phenotype of HEVs by expressing lymphotoxin.25 Martinet et al26 also found that dendritic cells and dendritic cell-derived lymphotoxin are closely associated with the formation of tumour HEVs in breast cancer. Since we found in our analyses that TLSs were mainly present in adjacent TDLUs, we speculate that HEVs might be transformed from the pre-existing blood vessels in TDLUs. Further studies that can explain the development and regulation mechanism of tumour HEVs might allow development of a manipulation technique to transform tumour vessels to tumour HEVs.26
TNBC is characteristically described as having pushing margins, central necrosis, high-grade features, high mitotic indices and scant stromal content.27 These features are well correlated with our present findings for TNBC with high levels of TILs, which showed a higher nuclear grade, lower lymphovascular invasion rate, less accompanying in situ component, homogeneous growth pattern, necrosis in invasive areas and low levels of tumour stroma. These findings suggest that tumours with high immunogenicity, which is characterised by high levels of TILs, might grow rapidly and result in lower levels of in situ component, a homogeneous tumour composition and higher rates of tumour necrosis. An alternative explanation of the characteristic features of TNBC might be that some TNBCs grow rapidly and therefore have frequent tumour necrosis, which can promote increased antigen presentation and subsequent influx of TILs. Recently, Brown et al28 analysed the RNA-seq data of glioblastoma multiforme, ovarian serous adenocarcinoma, lung squamous cell carcinoma, colorectal adenocarcinoma, breast invasive carcinoma and renal cell carcinoma from The Cancer Genome Atlas. These authors predicted immunogenic mutations based on the predicted autologous HLA-A-binding affinity of the peptide derived from the mutated gene. Those tumours with immunogenic mutations showed increased levels of CD8A gene expression and prolonged patient survival. Even though the total mutation count and overall survival were not associated (HR=0.91, 95% CI 0.68 to 1.23, p=0.55), tumours with high levels of mutations are more likely to contain immunogenic mutations than those with low levels of mutations. Since the overall mutation rate was highest in basal-like tumours in The Caner Genome Atlas data, the likelihood of having immunogenic mutations might be greater in TNBC.29 Algorithms that are currently in development that can predict immunogenic mutations might lead to a better understanding of the characteristics of TNBC.30
TILs showed prognostic value, particularly in patients with a higher pTNM stage. Conversely, if there were high levels of TILs, patient prognosis was not significantly different according to pTNM stage. The American Joint Committee on Cancer/Union International Contre le Cancer (AJCC/UICC) TNM classification is the most common cancer classification system for evaluating the degree of tumour progression.31 Many patients with cancer are treated using standard therapeutic guidelines based on the tumour's TNM stage (ie, http://www.nccn.org/professionals/physician_gls/f_guidelines.asp; http://www.esmo.org/Guidelines-Practice/Cl-inical-Practice-Guidelines). However, the TNM staging system gives incomplete prognostic information,32 possibly owing to a lack of consideration of the tumour microenvironment.33 We show here that the addition of TILs can provide better prognostic information for TNBC. However, since the microenvironment, including TILs, is as heterogeneous as the tumour cell itself, the development of a widespread standardised evaluation system of TILs is needed. Recently, the International TILs working group published an article on guidelines for TIL evaluation, which recommended the same TIL evaluation method that we used in our study.19 Therefore, using these guidelines might help to improve prognostic tools in breast cancer.
In our study, moderate to abundant TLSs in adjacent tissue of the invasive area, including carcinoma in situ, was seen in three-quarters of patients (n=581, 75.6%). However, of these cases, 59% of tumours had low to moderate levels of TILs (≤60%) in the interior of the invasive carcinoma. The prognosis of these patients was poorer than that of patients with high levels of TILs (>60%) and moderate to abundant TLSs. Consistent with our results, in colorectal cancer, combined evaluation of immune cell infiltration in the core of the tumour and in the invasive margin is reported to be associated with an improved accuracy of the prediction of survival than analysis of each region alone.34 In response to these observations, a scoring system (‘Immunoscore’) has been proposed for colorectal cancer.35 This score is based on the numeration of two lymphocyte populations (CD3/CD45RO, CD3/CD8, or CD8/CD45RO) in both the core of the tumour and the invasive margin.
The retrospective design of our study is a limitation. Nevertheless, our data provide a relevant basis for the evolution of an immunotherapy for TNBC. Further studies are necessary to confirm our results and elucidate the mechanism of immune cell infiltration in TNBC.
In conclusion, lymphoid aggregation in adjacent TDLUs was found to be present in most TNBCs, and TLSs were mainly present in TDLUs and around in situ components. TNBC with higher levels of TILs showed a higher nuclear grade, lower lymphovascular invasion, less accompanying in situ components, a homogeneous growth pattern, necrosis in invasive areas, low levels of tumour stroma, high levels of peritumoral lymphocytic infiltration and moderate to abundant TLSs in adjacent tissue. TILs and TLSs were found to be prognostic factors. In addition, the TIL level added significant prognostic information for stage II and stage III patients.
Take home messages
Tertiary lymphoid structures (TLSs) are mainly present in terminal duct lobular units and around carcinoma in situ components in triple-negative breast cancer (TNBC).
Large amount of TLSs in adjacent tissue and the presence of germinal centres in adjacent tissue are closely associated with higher levels of tumour-infiltrating lymphocytes (TILs) in TNBC.
Amount of TILs, degree of peritumoral lymphocytic infiltration and the presence of TLSs in adjacent tissue are prognostic factors for disease-free and overall survival outcomes of the patients with TNBC.
The level of TILs added significant prognostic information for pTNM stage II and III patients, while it did not have prognostic value in stage I. Conversely, pTNM stage had prognostic significance in patients with low to moderate levels of TILs, but patients with high levels of TILs did not show prognostic differences according to pTNM stage.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online tables
HJL and IAP are contributed equally.
Handling editor Cheok Soon Lee
Funding This study was supported by a grant (2013-0866) from the Asan Institute for Life Sciences, Asan Medical Center Seoul, Korea.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Institutinal review board of the Asan Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.