Prognostic significance of tumour infiltrating B lymphocytes in breast ductal carcinoma in situ

Tumour‐infiltrating lymphocytes (TILs) are an important component of the immune response to cancer and have a prognostic value in breast cancer. Although several studies have investigated the role of T lymphocytes in breast cancer, the role of B lymphocytes (TIL‐Bs) in ductal carcinoma in situ (DCIS) remains uncertain. This study aimed to assess the role of TIL‐Bs in DCIS.


Introduction
Although breast cancer is a heterogeneous disease, there is a great similarity between ductal carcinoma in situ (DCIS) and associated invasive carcinoma at histological and molecular levels. 1 Such similarity not only supports DCIS as a pre-invasive stage before progression to invasive duct carcinoma (IDC), but also indicates that they share molecular and behavioural features. 2 To date, neither histopathological features nor conventional breast cancer biomarkers can predict accurately whether DCIS lesions will progress to invasive disease or recur. 3 Evasion of immune surveillance and host immune response is a hallmark of carcinogenesis and cancer progression. 4 Furthermore, the intensity of tumour immune response influences the effectiveness of cancer therapy, and affects the clinical outcome positively in several solid tumours. 4 The association with patient outcome in the majority of studies have focused upon the role of tumour-infiltrating T lymphocytes (TILs). [5][6][7][8][9] However, there is also a suggestion of a critical role of tumour-infiltrating B lymphocytes (TIL-B) with patient survival. 10 B cells are activated commonly in cancer patients, supporting the possibility of a positive role in tumour immunity. 11 In breast tumours, TIL-Bs are present in approximately 25% and comprise up to 40% of the TIL population, [12][13][14] appearing early during breast tumorigenesis. 15 However, the assessment of TILs in cancer tissues remains challenging. 16 Despite reports implicating TIL-Bs in improving patient survival, the mechanisms, functional profiles and their allied antigens remain to be defined.
Of the common B cell antigens, CD20 is expressed on all mature B cells except plasma cells. 17 The T cell response is inhibited by resting B cells and facilitated by activated B cells. 18 CD19, another B cell marker, is a member of the immunoglobulin (Ig) superfamily that has a critical signal transduction function regulating the development, activation and differentiation of B lymphocytes. 19 The expression of CD19 is restricted only to B cells. Similar to CD20, it appears early during B cell maturation at the late pro-B cell stage and remains throughout maturation, but is lost when B cells differentiate into plasma cells. 20,21 CD138, another B lineage antigen, is a transmembrane heparin sulphate proteoglycan family member having several cellular functions, including proliferation and programmed cell death as well as a role as an extracellular matrix receptor. 22 It is expressed typically on the surface of mature epithelial cells and some stromal cells in developing tissues. 23 High levels of CD138 expression are detected in precursor B cells and plasma cell differentiation, 24 with monoclonal antibodies of the CD138 cluster being plasma cell-specific among haematopoietic elements. 24 This study aimed to determine the density and pattern of distribution of CD20/CD19-positive lymphocytes and CD138-positive plasma cells in patients with DCIS, including their prognostic significance.

S T U D Y P A T I E N T S
This retrospective study included cases of DCIS with or without an invasive component diagnosed from 1989 to 2000 at Nottingham City Hospital, Nottingham, UK who underwent conservative breast surgery with standard adjuvant treatment that was based on risk stratification. High-risk patients with pure DCIS received postoperative adjuvant radiotherapy. However, low-risk DCIS patients did not receive postoperative radiotherapy. Patients with IDC associated with DCIS were treated either with adjuvant hormonal, chemotherapy or radiotherapy or a combination based on hormone receptor status and the Nottingham Prognostic Index (NPI) of their invasive disease. Histological data for the DCIS type, including grade, presence or absence of necrosis, was recorded together with patient outcome. Recurrence-free survival (RFS) was calculated in months from the date of the first operation until the first recurrence. 25 For invasive cases associated with DCIS, histological data for lymph node status 26 and stage were recorded together with outcome data. Breast cancer-specific survival (BCSS) was defined as the time (in months) from the date of the primary surgical treatment to the time of death from breast cancer. 27

I M M U N O H I S T O C H E M I S T R Y
Haematoxylin and eosin (H&E)-stained full-face sections representative of each case were examined to confirm the diagnosis and assess the suitability of the tissue block for immunohistochemistry (IHC). IHC staining was performed using the Novocastra Novolink Polymer Detection Kit (Leica Microsystems, Newcastle upon Tyne, UK), following the manufacturer's guidelines. Tissue sections were stained by optimized monoclonal anti-human CD20 antibody (clone L26; Dako, Glostrup, Denmark, dilution 1:300), monoclonal anti-human CD19 antibody (clone LE-CD19; Dako, dilution 1:75) and monoclonal anti-human CD138 antibody (clone MI15; Dako, dilution 1:40). CD19, CD20 and CD138 were applied on sequential tissue sections from the same paraffin block and not on the same slide. All markers were included in the analysis. Paraffin sections of normal human tonsil were used as a positive control, which showed positive cells distributed mainly in the germinal centres and mantle zone (B cell area), with some scattered interfollicular-positive cells. Negative controls were included with staining runs by omitting the primary antibody. The number of B lymphocytes, marked by CD20and/or CD19-positive cells, and plasma cell count evidenced by CD138 positivity, was counted in each tissue section using a Nikon Eclipse 80i microscope (Nikon, Tokyo, Japan). For the purpose of the analysis, cells showing membranous labelling with CD20 and/or CD19 were considered as TIL-Bs. For CD138, only cells with morphology consistent with plasma cells were considered during cell counting. If CD138 additionally showed membrane and cytoplasmic staining of DCIS and invasive tumour cells, this epithelial staining pattern was not considered in the analysis. Slide scanning was also performed using the Panoramic digital slide scanner (Panoramic 250 Flash II, 3DHISTECH Ltd, Budapest, Hungary), followed by viewing the slide using Panoramic Viewer software (developed by 3DHISTECH Ltd, Budapest, Hungary). Marking the zones of interest to be scored was performed digitally (0.5, 1 and 2 mm).
Pure DCIS cases and the in-situ component of the mixed cases were assessed, clearly defined and marked to delineate the tumour edge. TIL-Bs were identified within DCIS (TIL cells) and at the periphery of DCIS (stromal TIL-Bs). All foci of DCIS were evaluated with an average of one to six foci. The highest density focus (hot-spot) was then selected as the final result for analysis. Presence and distribution of positively stained cells with morphological features of lymphocytes and/or plasma cells were reported. Stromal TIL-Bs were divided into: (i) cuffing (in direct contact with DCIS), (ii) peritumoral (less than 0.5 mm distance from the DCIS profile margin), (iii) paratumoral TIL-Bs that were quantified by counting positive cells within the marked tumour area and not in direct/close contact with tumour cells (more than 0.5 mm and up to 1 mm distance away from the DCIS profile margin) and (iv) TIL-Bs present in up to 2 mm away from the DCIS profile margin. The mean number of TIL-Bs was then calculated for each compartment. Within the marked area, each case was assigned a qualitative stromal TIL-B density score: a score of 1 referred to the low density of TIL-Bs (less than or equal to 25% surrounding the duct circumference). A score of 2 (>25% and <50% TIL-Bs surrounding the duct circumference) and score of 3 referred to a diffuse/ marked infiltration of more than 50% (TIL-Bs surrounding most of the duct circumference).
In the cases with mixed DCIS and invasive tumours, TIL-Bs were assessed separately in both components to indicate the difference in the density and pattern of distribution between components and between DCIS associated with invasion and pure DCIS. For TIL-Bs surrounding/infiltrating the DCIS components in these cases, the same approach was used as in pure DCIS cases. Conversely, TIL-Bs were assessed in the invasive component according to the previously published guidelines 28,29 in three locations; (i) intratumoral compartment (defined as TIL-Bs in tumour nests that had cell-to-cell contact with no intervening stroma and directly interacting with carcinoma cells), (ii) within stroma away from the tumour (defined as TIL-Bs located dispersed in the stroma, more than one tumour cell diameter and among the carcinoma cells but not directly interfacing carcinoma cells) and (iii) peritumoral stroma (defined as TIL-Bs within one tumour cell diameter of the tumour). In this study, the total number of TIL-Bs was determined by adding the counts for the three tumour compartments. TIL-Bs in areas with crush artefacts, necrosis, inflammation around biopsy sites or extensive central regressive hyalinization and adjacent normal lobules were not included. 28,29 In addition to the presence, density and location of TIL-Bs, the presence and distribution of lymphoid follicles or aggregates were assessed and their relation to DCIS and invasive disease, when present, were recorded. Lymphoid follicles [tertiary lymphoid structures (TLSs)] were considered as aggregates of lymphocytes with a germinal centre, while lymphoid aggregates were considered as a collection of lymphocytes without germinal centre formation.
This study was approved by the Nottingham Research Ethics Committee 2 (REC C202313) under the title: 'Development of a molecular genetic classification of breast cancer'.

S T A T I S T I C A L A N A L Y S E S
IBM-SPSS statistical software version 22.0 (SPSS, Inc., Chicago, IL, USA) was used to analyse the correlation between the number of CD20 + , CD19 + lymphocytes and CD138 + plasma cells and the various clinicopathological parameters. The optimal cut-off point for CD20 + , CD19 + lymphocytes and CD138 + plasma cells against patient survival was defined using X-tile bioinformatics software (Yale University, version 3.6.1). Kaplan-Meier curves and log-rank test were used for survival analyses. A P-value less than 0.05 (two-tailed) was considered statistically significant.

P A T I E N T C H A R A C T E R I S T I C S
Patient characteristics for the 80 patients on the study are shown in Table 1. Patients were aged 70 years or less (median: 55 years) with long-term follow-up (median follow-up: 266 months). No patients received neoadjuvant therapy. Thirteen patients received chemotherapy and 17 cases were treated hormonally for invasive disease. Of the two categories, 36 cases were of DCIS alone (none of the pure DCIS cases contained microinvasion), while DCIS was associated with an invasive component in 44 cases (all cases included invasive carcinoma measuring more than 1 mm). Most pure cases (28 of 36: 77.8%) were from postmenopausal women. Histological assessment for DCIS is summarized in Figure 1A. During the follow-up period, ipsilateral local recurrence in pure DCIS occurred in seven of 36 (19.4%) patients, in whom two (5.5%) cases recurred as invasive disease.
Forty-four female patients with DCIS had associated invasive disease; 27 of these 44 were postmenopausal. Figure 1B,C summarizes the histological data for the invasive disease, including tumour size, type grade, presence or absence of comedo necrosis, lymphovascular invasion 30 and lymph node status. 26 Median follow-up was 143 months (19-307 months), during which period recurrence of invasive disease occurred in 12 (27.3%) patients, seven (15.9%) patients developed ipsilateral local recurrence and eight (18.2%) cases progressed into distant metastasis.
To summarize the pathology parameters in this series, solid DCIS with necrosis (comedo type) was the predominant type. Most cases showed high nuclear grade (55%). The mean DCIS size was 24 mm. Positive hormone receptor status [oestrogen receptor/progesterone receptor (ER/PR)] and negative HER2 was the predominant pattern in both groups.
Assessment of ER, PR and HER2 assays were based on the American Society of Clinical Oncology/College of American Pathologists Clinical Practice Guideline Update (ASCO). 31,32 ER and PR were considered positive if there were at least 1% positive tumour nuclei. For HER2 status, intense complete/circumferential IHC membrane staining within more than 10% of tumour cells was considered positive (3 + ).

D C I S T I L -B S : F R E Q U E N C Y A N D L O C A L I Z A T I O N
Intratumoral TIL-Bs were extremely rare within DCIS compared with the periductal stromal TIL-Bs. Both diffuse and aggregate patterns of TIL-Bs were observed in the stroma. The density of CD19 was lower than that of CD20 and the number of follicles stained with CD19 was fewer than those identified by the CD20. Plasma cells did not show any specific pattern or distribution. Pure DCIS cases showed a significantly higher number of TLSs (maximum 20 lymphoid aggregates) than those cases associated with invasion (maximum 12 lymphoid aggregates) (P = 0.04). Tertiary lymphoid structures (TLSs) were localized mainly around the DCIS component of the mixed group.
Intratumoral, peritumoral, paratumoral and stromal lymphocyte results in the pure and mixed groups are summarized in Table 2. The mean count of peritumoral lymphocytes was 80.4 in pure DCIS cases, compared to 37.7 in cases associated with an invasive component (P = 0.002). In paratumoral TIL-Bs, the mean number was 108.1 in pure DCIs cases compared to 56.7 in DCIS cases mixed with an invasive component (P = 0.006). A high level of B cells as defined by CD20 and or CD19 positivity was observed around the DCIS component of the tumour in 65.9% of cases, and was observed around the invasive component in 27.3% (P = 0.01) ( Figure 1D-F). It was observed that the number of B lymphocytes around the DCIS foci was higher in the pure group when compared to the DCIS foci in the mixed cases (P < 0.001). Pure DCIS cases showed a higher number of plasma cell count (mean: 91) than mixed cases (mean: 44), although this does not achieve statistical significance (P = 0.4). Stromal distribution was the detected pattern of plasma cell distribution. No intratumoral plasma cells were found.

A S S O C I A T I O N W I T H C L I N I C O P A T H O L O G I C A L V A R I A B L E S
Overall associations of TIL distribution with clinicopathological variables are summarized in Table 3. In pure DCIS cases, increased numbers of TLSs and dense peri-and paratumoral TIL-Bs were associated significantly with larger tumour size (P = 0.016), hormone receptor (ER/PR)-negative tumours (P = 0.008) and HER2-positive status (P = 0.01). No association between plasma cells and the clinicopathological parameters was identified in the pure cases.
In the mixed cohort, higher numbers of B lymphocytes, irrespective of their location and topographic distribution, were associated significantly with larger (invasive and in situ) tumour size (P = 0.019), higher invasive tumour grade (P = 0.005), the presence of DCIS necrosis (P = 0.042), lymphovascular invasion (P = 0.022), lymph node metastases (P = 0.033), negative ER/PR status (P = 0.04) and positive HER2 status (P = 0.008). A higher number of plasma cells was associated significantly with ER/PR-negative tumours (P = 0.01) and HER2 positivity (P = 0.019). Outcome analysis (Figure 2) revealed that pure DCIS cases associated with low numbers of peri-and paratumoral B lymphocytes had a longer RFS (P = 0.008 and P = 0.04, respectively). Less dense peritumoral and low count of stromal B lymphocytes was associated with longer RFS (P = 0.04 and P = 0.01, respectively). There was a non-significant association between low plasma cell count around the invasive component and longer survival (P = 0.07). Intratumoral TIL-Bs did not show a significant association with patient outcome.

Discussion
The role of immune cells in breast carcinogenesis remains questionable. 33 It was thought initially that tumour-infiltrating immune cells play a protective role in tumorigenesis. 34 However, there is increasing evidence supporting the fact that the infiltrating immune cells play a role in carcinogenesis, 35 and there is a plethora of data to propose powerful links between infiltrating immune cells and carcinogenesis. 33 TILs are an important immune component of the response to cancer. 36 It is now well accepted that the immune system has a dual role in cancer development and progression. It can eradicate emerging malignant cells by an orchestrated action of innate and adaptive branches, thus preventing tumour  growth. Paradoxically, it can promote growth of malignant cells, their invasive capacity and their ability to metastasize. The presence of immune cells with tumour-suppressive and tumour-promoting activity in the cancer microenvironment and in peripheral blood is associated usually with good clinical outcome and poor clinical outcome, respectively. 36 Furthermore, the intensity of tumour immune response influences the effectiveness of cancer therapy, and affects the clinical outcome positively in several solid tumours. 4 TILs have been identified previously as prognostic and predictive biomarkers in several cancers, including breast cancer. However, the exact role of the different components of TILs remains unclear. 16 Although most of the TILs in breast cancer have focused on T lymphocytes 5-7 the role of TIL-Bs remains poorly defined, 37 with few studies having assessed the role of TIL-Bs in the breast. [38][39][40] In this study, we aimed to determine the potential role of TIL-Bs in DCIS and its role in DCIS behaviour and progression to invasive disease.  In the current study, the density of CD19 was much lower than that of CD20 and the number of follicles identified by the CD19 antibody was a subset of those identified by the CD20 antibody. This finding is in line with findings in other tumours, such as chronic lymphocytic leukaemia where, in 1998, Ginaldi et al. 41 showed that CD19 had low density and a lower number than CD20-positive cells. This can be explained as a reflection of an early stage of maturation of TIL-Bs in comparison to their counterparts in normal blood. 41 This may suggest that tumour immunity develops as a response to the early stage of tumour development, such as DCIS.
Some authors have reported that TILs are observed more commonly in DCIS than in invasive carcinoma, 5,42,43 and this was confirmed in the current study. One study reported that the proportion of luminal-like subtypes decreased, while HER2 + and basal-like subtypes increased with the development of invasion. 44 Conversely, other studies have reported that microinvasive carcinoma might be a distinct entity.
Martinet et al. 43 showed that tumour-associated lymphocytes and mature dendritic cell densities were significantly higher in DCIS than in invasive carcinoma. However, the relationship between the presence or abundance of TLSs and specific DCIS subtypes has been unclear, 45 although the presence of TLSs around HR-/HER2-positive tumours are also corroborated in the current study. It could be explained that once the carcinogenic events have settled (i.e. invasive cancers), a generalized increase in lymphocyte infiltration is observed that does not differ among various tumour subtypes. 33 Moreover, the DCIS tumour microenvironment overexpresses a variety of inflammatory mediators (probably released from infiltrating leucocytes), including interleukin signalling. 46 Taken together, these findings suggest a role of leucocytes in the early stages of breast cancer development. In support of this, there is evidence that lymphocytes play a key role in creating a tumour-promoting microenvironment. 34 However, this needs to be investigated further in breast carcinogenesis.
Our study showed that higher numbers of B lymphocytes and TLSs were associated with higher tumour grade, presence of necrosis, vascular invasion, negative hormone status and HER2 positivity. These findings are supported by Schalper et al., 47 who suggested that the tumour biology itself may play a possible role in lymphocytes induction. Higher CD20-positive TILs have been observed similarly in high-grade DCIS by Campbell et al. and, as indicated, an orchestrated increase of FoxP3 + , CD68 + , HLA-DR + and CD4 + cells are observed in higher grades of DCIS. 48 The relationship with vascular invasion is interesting because, in studies as early as 1997 by Lee et al., it was apparent from morphology correlates that clusters of B and T cells may be recruited in DCIS by high endothelial venules and the authors speculate that cytokines released by the DCIS, along with its immune cells, may stimulate new vessel formation and create a prometastatic milieu. 48 DCIS cells should adapt to the hypoxic and nutrientdeprived ductal microenvironment. We assume that the presence of necrosis in DCIS might be associated with increased release of damage-associated molecular substances such as adenosine triphosphate that could result in the subsequent recruitment of immune cells into the tumour microenvironment. 49 Ma et al. 50 reported a strong immune response signature, resulting in activation of other leucocytes and interferon signalling present, particularly around highgrade DCIS. They speculate that the presence of an immune response signature around high-grade DCIS may represent a phase where the cancer cells resist immune attack and instead are able to utilize the abundant cytokines produced by immune cells to facilitate invasion. 50 Unlike invasive cancer, where the presence of abundant TILs has been linked to better prognosis, this does not appear to be the case for DCIS. 29 In the current study, we found decreased RFS in cases associated with more TIL-Bs. This agrees with Knopfelmacher et al., 51 who found that the presence of dense chronic inflammation surrounding DCIS was associated significantly with a high oncotype Dx DCIS score, and hence a high recurrence risk. Although there are no genes related directly to the immune system in the DCIS score, research has elucidated that there are genetic changes in the microenvironment, including stromal fibroblasts, myoepithelium and inflammatory cells, which are associated with progression from in-situ to invasive disease. 52,53 In cases where DCIS is associated with early invasion, a dense chronic inflammatory infiltrate often surrounds these microinvasive foci. We could speculate that dense periductal chronic inflammation around DCIS suggests a role for the immune response in DCIS progression and deserves further investigation. Despite the prognostic role of TIL-Bs, it was not an independent prognostic factor in the Cox regression model, and this might be explained by the small number of cases included in the study. In this study, it is noteworthy that assessment of TILs in DCIS was different from that of invasive cancer. Evaluation of stromal-associated lymphocytes, within the confinement of an invasive tumour, was challenging because of the presence of scattered ducts in DCIS. However, the scoring was performed as objectively as possible. For better reproducibility, standardization of the TILs scoring method in DCIS is needed.
The use of CD138 as a clinical marker remains controversial. In our study, CD138 was expressed in the entire epithelium as well as plasma cells but it did not stain lymphocytes, and this finding is consistent with the finding of Barbareschi et al. 54 We also found that more plasma cells were observed in pure DCIS cases when compared to DCIS cases associated with invasion. A high plasma cell count represented by CD138 positivity is correlated with ER/PR negativity. In fact, it could be hypothesized that in ER-negative tumours that have lost the ability to respond to the oestrogen-dependent proliferative pathway, high CD138 expression may confer a particularly important growth advantage by enhancing the response to other growth factors. 54 This correlates with poor prognosis, findings consistent with those reported by Barbareschi et al. 54 CD138 is, in fact, implicated in several essential physiological cell functions, such as control of cell proliferation, differentiation, adhesion and migration. 55 One of the best-known biological functions of CD138 is related to its interaction with fibroblast growth factors (FGFs), which are known angiogenic and mitogenic growth factors for breast carcinoma cells, 56 binding to FGFs and to their receptors in a ternary signalling complex. 57 CD138 can also function as a potent FGF-2 activator through physiological shedding and degradation of its extracellular domain by enzymes, such as heparanase. 58 The limitations of this study are that a limited subset of cases were studied, as well as the limited availability of data for hormone receptor status and HER2 expression, especially for the DCIS component. This might have affected the statistical associations with some parameters. Expansion of this study to include a larger patient cohort is therefore warranted.
In conclusion, this study suggests that B cells, perhaps as part of the adaptive humoral immune response, may have a role in breast cancer. Expansion of this work on a larger series of patients is warranted, as well as the development of a standardized scoring approach. Studies of a holistic nature exploring the cross-talk of both B and T cell pathways may reveal the immune switch, enabling tumour progression from the in-situ stage to invasive disease. As B cell activation may be both T cell-dependent and -independent, studying them side by side alongside released mediators will aid clearer understanding of the correlations between biology and morphology. As the role of B cells in the preinvasive to invasive stages becomes clearer, novel options for immune modulation to prevent breast cancer progression may become evident.