Manipulation of tumour-infiltrating B cells and tertiary lymphoid structures: a novel anti-cancer treatment avenue?

Abstract

Combining different standard therapies with immunotherapy for the treatment of solid tumours has proven to yield a greater clinical benefit than when each is applied separately; however, the percentage of complete responses is still far from optimal, and there is an urgent need for improved treatment modalities. The latest literature data suggest that tertiary lymphoid structures (TLS), previously shown to correlate with the severity of autoimmune diseases or transplant rejection, are also formed in tumours, have a significant beneficial effect on survival and might reflect the generation of an effective immune response in close proximity to the tumour. Thus, the facilitation of TLS formation in tumour stroma could provide novel means to improve the efficiency of immunotherapy and other standard therapies. However, little is known about the mechanisms regulating the formation of tumour-associated TLS. Studies of chronic inflammatory diseases and transplant rejection have demonstrated that TLS formation and/or function requires the presence of B cells. Additionally, the infiltration of B cells into the tumour stroma has been demonstrated to be a significant prognostic factor for improved survival in different human tumours. This suggests that B cells could play a beneficial role in anti-tumour immune response not only in the context of antibody production, antigen presentation and Th1-promoting cytokine production, but also TLS formation. This review focuses on the latest discoveries in tumour-infiltrating B cell functions, their role in TLS formation and relevance in human tumour control, revealing novel opportunities to improve cancer therapies.

Introduction

The combination of immunotherapy with standard therapies has recently proven to yield a much more prominent beneficial effect than each of the treatments applied separately [1]. This implies that (i) chemo- and radiotherapies, as well as small molecule inhibitors [2] can be effective immune stimulants via different mechanisms such as the facilitation of tumour cell death, the release of cancer antigens and danger signals [3, 4], and that (ii) the tumour-specific immunity induced as a result of these standard treatments and strengthened by the addition of immunotherapies is capable of generating complete responses. Nevertheless, also in combination therapy trials, the percentage of complete responses is still far from 100 % and emphasises the very complex nature of tumour and immune system interactions as a priority research field.

Depending on the cellular composition of the tumour microenvironment, including the variety of effector and regulatory immune cells and stromal cells, the elicited immune response can be either tumour promoting or host protective, thereby greatly influencing the course of disease [5]. Good prognosis is often associated with the so-called type 1-immune-response-dominating microenvironments involving mature dendritic cells (DC), M1 macrophages, CD4⁺ type 1 T helper (Th1) cells, activated cytotoxic CD8⁺ T cells (CTLs), γδT cells and natural killer (NK) cells, while chronic inflammation involving M2 macrophages, myeloid-derived suppressor cells, neutrophils, Th2, as well as Th17 and regulatory T (Treg) cells has often been correlated with the promotion of cancer growth (recently reviewed in [5, 6]). The role of human tumour-infiltrating B (TIB) cells, however, has not been a focus of attention for several reasons. The infiltration of B cells is observed more rarely than T cells [7], and their absolute numbers are often much lower than T cells among the tumour-infiltrating immune cells [8–11]. Early studies on immune factors affecting transplanted [12] or chemically induced carcinogenesis [13] showed a tumour-promoting impact of B cells—an observation that has been convincingly reproduced by many studies in various mouse models [14, 15]. This has established an almost dogmatic notion that B cells are potent tumour- and metastasis-promoting effectors facilitating chronic inflammation [16] and an immunosuppressive microenvironment. Nevertheless, not all mouse studies demonstrate the same effect (see recent review [17]), nor do most human studies conform to this notion. Studies performed in the 1990s demonstrated that TIB cells can exert anti-tumour effects in humans [18, 19] as well as mice [20]. Even a direct antibody-dependant cellular cytotoxicity (ADCC) for splenic B cells from tumour-bearing, but not healthy mice was demonstrated [21]. Furthermore, systemic functional impairment of B cells in mice [22] and cancer patients [23] has been reported. The latest results from work with human tumours show that the infiltration of B cells or the presence of B-cell-associated gene signatures are significant and often independent favourable prognostic factors in a growing list of different human tumours including colorectal [10, 24, 25], breast [24, 26–31] and other cancers. A possible explanation for this inconsistency between the mouse and human studies could be the differential B cell activation status in the particular experimental settings: the studies in mouse models often address the whole B cell pool, including naïve and regulatory B cells, which have been shown to be immunosuppressive or acquire such capacity [17], while B cells that mostly infiltrate human tumours are activated and possess protective effector activities [32, 33].

Besides antibody production, recent studies have demonstrated many other functional activities of B cells. Upon activation, they can differentiate into distinct effector subsets, namely Be1 and Be2, that produce similar cytokine profiles as the Th1 and Th2 cells, respectively, thus regulating the type1/type2 polarity of the immune response [34]. B cells can also (i) affect the inflammatory microenvironment and induce the activation/differentiation of cellular immune responses [35, 36], (ii) facilitate the formation of CD4⁺ T cell memory [37], (iii) enhance the proliferation and survival of activated CTLs [38] or (iv) become immunosuppressive [39]. This implies that B cells should no longer be viewed as the “B-oring” cousins of the “T-errific” T cells, and solicits an important paradigm shift when considering novel ways to improve cancer immunotherapy.

Studies of transplant rejection and autoimmunity have provided yet another novel function of B cells: their capacity to induce and/or sustain the formation and function of tertiary lymphoid structures (TLS). These are lymphoid aggregates that resemble lymph node (LN) architecture and function, evolve in response to continued stimulus for leucocyte extravasation and persistent source of antigen, and exacerbate inflammatory reactions [40]. A number of studies have recently indicated that TLS might be important in anti-tumour immune responses, indicating a better prognosis. Thus, the presence of such structures in the vicinity of cancer might reflect the formation of a functional cancer-specific immune response. Here, we review recent data about the role of B cells in the human tumour microenvironment with a focus on their role in the formation of TLS, the importance of TLS for disease outcome and outline novel potential avenues for therapeutic manipulation.

Functional properties of TIB cells

Prognostic significance

The summarised literature data in Table 1 document that TIB cells can be associated with different outcomes in various cancer types and histological subtypes. The infiltration of B cells into breast, lung and colorectal cancers has been convincingly associated with improved survival as demonstrated in the large number of patients in the analysed cohorts and significance in multivariate analysis. In contrast, no significant association of B cell infiltration with survival has been reported for gastric cancer, while studies of ovarian, prostate, hepatocellular, head and neck cancers and melanoma show somewhat conflicting results. Inconsistent results have also been reported regarding specific cancer histotypes, especially for the squamous cell carcinoma and adenocarcinoma subtypes of non-small-cell lung cancer (NSCLC). The cut-offs used to discriminate patient groups by the intensity of B cell infiltration and applied for survival analyses vary greatly among the published studies and might partially explain the observed inconsistencies; however, further research is warranted to identify the mechanisms underlying the differential prognostic impact of TIB cells in various cancer histotypes.

Table 1.

Prognostic significance of elevated B cell infiltration in human solid tumours

Patient information, reference	Marker (method)	Location (mean extent) of infiltration	Mode of distribution (% frequency)	Prognostic significancea
				Threshold for group definition	Histotype (♯ of patients, frequency of positive cases)	HR	p value
Breast cancer
Node-negative [29]	IGKC (IHC)	Tumour stroma	Diffuse	<1 cell≤	All (335, 76 %) ER− (64) Luminal B (55) ER + (271), Luminal A (224)	DFS 0.5 DFS < 1 DFS < 1 NA	0.004 0.044b 0.009b NS
[30]	CD20 (IHC)	Tumour stroma, neoplastic epithelium (1 cell/mm2), and peritumoural (12 cells/mm2)	Diffuse (>80 %), aggregates (>20 %)	<5 cells/mm2	All (1470, 55 %) Grade 3 (736) ER− (321) Basal (249) HER2+ (153) Grade 1 + 2 (726), ER+	DSS 0.75 DSS 0.61 DSS 0.61 DSS 0.46 DSS 0.56 DSS NA	0.025 <0.001 0.008 <0.001 0.023 NS
[31]	CD19 (IHC)	Tumour stroma	Diffuse	Mean infiltration level in <5 years versus >7 years relapse-free survival groups	All (17)	DFS < 1	0.046c
	IGKC (GE)	NA	NA	MEL in groups of <5 years versus >7 years DFS	All (12, 75 %)	DFS < 1	<0.001c
Node-negative, 3 independent cohorts [27]	B cell metagene (GE)	NA	NA	<2 (relative metagene expression level) ≤	All (200), low proliferative (105) Highly proliferative (95)	DFS 0.79 DFS 0.66	NS 0.034
					All (286) Highly proliferative (184) Low proliferative (102)	DFS 0.78 DFS 0.74 DFS 0.91	0.034 0.03 NS
					All (302) Highly proliferative (151) Low proliferative (151)	DFS 0.83 DFS 0.82 DFS 1	0.021 0.035 NS
Node-negative [24]	IGKC (IHC)	Tumour stroma	Diffuse	<1 cell≤	All (410, ~75 %)	MFS < 1	<0.001b
	IGKC (GE)	NA	NA	<MEL<	All (965) ER + HER2− (610) ER-HER2- (220) HER2 + (135)	MFS 0.81 MFS 0.82 MFS 0.81 MFS 0.62	<0.0001d 0.0002d 0.017d 0.0001d
[26]	B/plasma cell metagene (GE)	NA	NA	<MEL<	Low + medium proliferative (1297) Highly proliferative (657) Highly proliferative (421)	NA MFS < 1 MFS 0.59	NSb <0.001b <0.0001
Colorectal cancer
Resectable, no NAT [25]	Plasma cells (HE staining)	Invasive margin stroma (11 % of NCs)	NA	<Median infiltration<	All (130, >50 %)	DSS 2.99f DSS NA	0.002e NS
[24]	IGKC (GE)	NA	NA	<MEL<	All (513)	MFS 0.88	0.043e
No post-operative therapy [161]	CD20 (IHC)	Neoplastic epithelium (~0.12 cells/mm2), tumour stroma and peritumoural	Diffuse, aggregates in peritumour (35 %)	(NA) Low versus high infiltration in neoplastic epithelium	All (117, ~100 %)	NA	NS
Two independent cohorts [10]	CD20 (IHC)	Invasive margin stroma (350 cells/mm2), intratumour (30 cells/mm2)	NA	(NA) High versus low infiltration	All (107)	DFS 3.7f	0.002e
					All (415)	DFS 2.2f	<0.05e
Gastric cancer
[162]	CD20 (IHC)	Neoplastic epithelium, tumour stroma and invasive margin	Diffuse	<Median infiltration<	All (100, >43 %)	OS 0.555 DFS 0.632	0.045 NS
[163]	CD20 (IHC)	Tumour stroma (438.7 cells/mm2), neoplastic epithelium (18.9 cells/mm2)	Diffuse	<Median infiltration<	All (52)	NA	NS
Resectable, no NAT [164]	CD20 (IHC)	Tumour stroma (~200 cells/mm2)	Diffuse	<Mean infiltration<	All (220)	NA	NS
Hepatocellular cancer
Resectable, no NAT [135]	CD20 (IHC)	Neoplastic epithelium (5.5 cells/mm2), tumour margin	Diffuse, aggregates (~6 %)	<400 lymphocytes/mm2≤	All (163)	DFS < 1	0.01g
Two independent cohorts [80]	CD20 (IHC)	Tumour margin (192.7 cells/mm2), intratumour (12 cells/mm2), peritumour (50 cells/mm2)	Diffuse	<Median infiltration<	All (120, >50 %)	OS 0.44 DFS 0.49	0.003 0.012
					All (200)	OS 0.53 DFS 0.6	0.001 0.015
Resectable, no NAT [165]	CD20 (IHC)	Neoplastic epithelium	Diffuse	<Median infiltration<	All (362)	NA	NS
Biliary tract [166]	CD20 (IHC)	Neoplastic epithelium (rarely), tumour stroma (< 25 cells/mm2)	Diffuse, aggregates (7 %)	<1 cell≤ <1 aggregate≤	Adenocarcinoma (323, ~46 %)	OS < 1 OS < 1	0.032b NS
Head and neck cancer
Resectable, no NAT [167]	CD20 (IHC)	Intratumour (0.23 cells/100TCs), intratumour of LNM (1.8 cells/100TCs), peritumour of LNM (~ 2200 cells/mm2)	NA	<Median infiltration in peritumour of LNM<	SCC (33, >50 %)	DFS < 1	0.039b
[23]	B cell (IHC)	Tumour stroma and neoplastic epithelium	NA		SCC (40)	NA	NS
[168]	CD20 (IHC)	Intratumour	NA	<0.02 cells/100 TCs≤	SCC early stage (62)	DFS < 1	0.021b
				<0.08 cells/100 TCs≤	SCC late stage (53)	DFS > 1	0.03b
Resectable, no NAT [169]	CD20 (IHC)	Intratumour (~3 cells/HPF)	NA	NA	SCC of the larynx (50)	NA	NS
Lung cancer
Resectable [170]	CD20 (IHC)	Peritumoural	NA	<1 cell≤	All (113, 57.5 %) Non- SCC (66) SCC (47)	OS < 1 OS < 1 NA	0.04e <0.001e NSd
Resectable, no NAT [171]	CD20 (IHC)	Tumour stroma and neoplastic epithelium	Diffuse	≤1 % of NCs<	All (335, ~ 99 %) SCC (191) Non-SCC (144)	DSS < 1 DSS < 1 NA	<0.001b 0.03b NSb
Resectable, no NAT [172]	CD138 (IHC)	Tumour stroma (>25 %), neoplastic epithelium (> 5 %)	Diffuse	Stromal: ≤25 % of NCs< Intraepithelial: ≤ 5 % of NCs<	All (335, ~99 %) T3 (27) N2 (27) Stage IIIA (32)	NA DSS < 1 DSS < 1 DSS < 1	NSb 0.046b 0.029b 0.034b
[24]	IGKC (GE)	NA	NA	Continuous variable	All (196)	MFS 0.81	0.032
				<MEL<	SCC (66) Adenocarcinoma (106)	MFS ~ 1 MFS < 1	NS 0.002e
				<MEL<	All (1,056)	MFS 0.91	0.011e
Resectable [173]	CD138 (IHC)	Tumour stroma	Diffuse	≤20 % of NCs<	All (350) Adenocarcinoma (191)	OS 0.74 OS 0.54	0.041 0.004
	IGKC (IHC)	Tumour stroma	Diffuse	≤20 % of NCs<	All (350) Adenocarcinoma (193) Adenocarcinoma (81)	OS 0.72 OS 0.57 DFS < 1	0.035 0.013 0.044b
	CD20 (IHC)	Tumour stroma	Aggregates	≤20 % of NCs<	All (350)	NA	NS
	IGKC (GE)	NA	NA	<MEL<	All (196)	OS 0.63	0.007
Melanoma
Resectable, no NAT [76]	CD20 (IHC)	Peritumour (78.7 cells/mm2), neoplastic epithelium, (4.9 cells/mm2), tumour stroma (10 cells/mm2)	Diffuse, aggregates (26 %)	<mean peritumour infiltration (170 cells/mm2)<	All (106, 86 %)	OS < 1 OS NA	0.0136b NS
				<Mean intraepithelial infiltration (3 cells/mm2)<	All (106)	OS < 1	0.0391b
[174]	CD20 (IHC)	Peritumoural, intratumoural	Diffuse, aggregates	NA	All (364, ~5 %)	OS 1.73	0.004e
[145]	CD20 (IHC)	Tumour stroma (65 cells/mm2)	Diffuse	<Median infiltration (4.8 cells/mm2)<	Metastasis (147, >50 %)	DFS 1.88f	0.04
	CD138 (IHC)	Tumour stroma (26 cells/mm2)	Diffuse	<Median infiltration (2.3 cells/mm2)<	Metastasis (147)	DFS 1.47f	0.01
Ovarian cancer
FIGO stage III [175]	CD20 (IHC)	Tumour stroma and neoplastic epithelium (1-5 cells/400 × HPF)	NA	<1 cell/400 × HPF<	Serous (97, ~58 %)	NA	NS
[78, 79]	CD20 (IHC)	Neoplastic epithelium, tumour stroma	Diffuse	Intraepithelial: <1 cell/mm2≤	Optimally Db HGS (198, 42 %) Suboptimally Db HGS (220, 16 %) Endometroid (125, 21 %) Clear cell carcin. (132, 18 %) Mucinous (31, 10 %)	DSS < 1 NA NA NA NA	0.0033b NSb NSb NSb NA
[24]	IGKC (GE)	NA	NA	<MEL<	All (426)	OS 0.98	NSe
[109]	CD19 (IHC)	Tumour stroma (> 50 %)	Aggregates	<50 % of positive cells<	All (49)	OS 3.93 OS 2.03	0.003e NS
Other cancer
Cervical cancer [176]	CD20 (IHC)	Tumour stroma	Diffuse, aggregates	(NA) Low versus high infiltration	All (61, ~ 18 %)	DFS < 1	<0.05b
Cervical cancer [177]	CD20 (IHC)	Intratumoural (~50 cells/mm2), peritumoural (~400 cell/mm2)	NA	Mean peritumour infiltration level in < 5 year versus > 5 year relapse-free survival groups	Stage IB SCC (40)	DFS < 1	0.002c
Prostate cancer [112]	CD20 (IHC)	Tumour stroma and neoplastic epithelium	Diffuse	(NA) Low versuss high infiltration	All (75)	PSA RFS NA	NSe
Prostate cancer, resectable [178]	CD20 (IHC)	Tumour stroma	NA	NA	All (188, 15 %)	PSA RFS NA	NSb
Soft tissue sarcoma [179]	CD20 (IHC)	Peritumoural, intratumoural	Diffuse	<1 cell/mm2 in peritumour≤	All (77, 15 %)	DSS 2.7	0.03
Thymic carcinoma [81]	CD20 (IHC)	Tumour stroma and neoplastic epithelium	Diffuse	<1 % of NCs in tumour stroma<	All (32, stromal 41 %, epithelial 16 %)	OS < 1	0.045b

The characteristics of B cell infiltration were depicted from the published results, provided images, discussion and supplementary data

DSS disease specific survival, DFS disease free survival (also includes NED no evidence of disease, and recurrence free survival), MFS metastasis-free survival, OS overall survival, PSA RFS prostate specific antigen recurrence-free survival, IHC immunohistochemistry, GE gene expression, LNM lymph node metastasis, NSCLC non-small cell lung cancer, HR hazard ratio, NAT neoadjuvant therapy, NA not available, NS not significant, NCs nucleated cells, TIL tumour-infiltrating lymphocytes, MEL median expression level, SCC squamous cell carcinoma, TCs tumour cells, Db HGS debulked high-grade serous, HPF high power field

^aThe HR is depicted here for the high infiltration of B cells or elevated expression of B-cell-related markers and p values are calculated by multivariate Cox regression analysis, if not indicated otherwise

^bHR < 1—improved survival in Kaplan–Meier curve without specified HR value, logrank test p value

^cStudent’s t test p value

^dFixed effect model p value

^eUnivariate Cox regression analysis p value

^fThe indicated HR shows the association to decreased B cell infiltration

^gWilcoxon test p value

Autoantibodies

It has been shown that B cells infiltrating breast cancer [41–48], colorectal cancer [49–51], lung [52, 53], cervical cancer [54], germ cell tumours [55], melanoma [56, 57], nasopharyngeal carcinoma [58], malignant mesothelioma [59] and other cancers [60, 61] produce class-switched affinity-matured anti-tumour antibodies in situ, and anti-tumour effects of these antibodies have been demonstrated [62–64]. Autoantibodies against cancer-cell-derived molecules have also been observed systemically—in sera of patients with most, if not all, cancer types [65], and their anti-tumour effects have been demonstrated in melanoma [66]. However, there is currently no clear consensus on what is the role of the systemic autoantibody response in cancer immunity, as it has been correlated with diverse clinical outcomes (reviewed in [67]). For example, autoantibodies against one of the most studied cancer-testis antigens NY-ESO-1 (CTAG1B) have been correlated with decreased overall survival of prostate cancer [68], but not oesophageal cancer patients [69]. The data on antibodies against p53 are also controversial—in NSCLC, breast, gastric, and oral cancer patients they have been shown as markers of worse prognosis, but were associated with improved survival of hepatocellular carcinoma patients, while both associations have been published for colon cancer. However, it is not clear whether the observed correlation is associated with the TP53 mutation status or the production of autoantibodies [67, 70]. Our own studies of autoantibody signatures have demonstrated that there are different sets of autoantibodies that show correlations with longer or shorter time to disease progression in melanoma patients, as well as shorter or longer overall survival of gastric cancer patients (unpublished results and [71]). An explanation of such discrepancies might be the different affinities of various IgG subclasses with the activating and inhibitory Fc gamma receptors (FcγR) [72]. Indeed, a differential role for immunoglobulin G (IgG) subclasses has been demonstrated in melanoma, where specific IgG4 antibodies failed to induce tumouricidal effects as compared to their IgG1 counterparts, decreased the IgG1-mediated protective responses and inversely correlated with patient survival [73]. In contrast, in NSCLC the presence of IgG4-producing plasma cells was associated with prolonged overall survival [74]. The net outcome of autoantibodies in the tumour microenvironment is likely dependent on the combination of several factors such as the respective antigen, the Ig class and IgG subclass produced, and the characteristics of the prevailing infiltrating immune cells including the effector versus regulatory phenotypes and the expressed FcR subtypes [75].

Support for T cell response

Plasma cells have often been reported as rare and scarce tumour infiltrates, while CD20⁺ TIB cells have been shown to constitute even >50 % of all nucleated cells, especially in the stromal areas and often correlate with improved survival (Table 1), suggesting that cancer-specific Ig production is not the main or the only activity of TIB cells. In patients with cutaneous melanoma, the presence of both CD20⁺ and OX40^high cells had the strongest correlation with longer survival as compared to either population alone [76]. High (CD20 + CD3)/CD68 ratio in invasive ductal breast carcinomas [77] and high co-expression of B cell, T cell and DC metagenes in highly proliferative breast tumours [26] were superior to either parameter separately for predicting prolonged metastasis-free survival. Similarly, the co-localisation of CD20⁺ and CD8⁺ cells in colorectal [10], serous ovarian [78, 79], hepatocellular [80] and thymic [81] cancers shows the most marked survival benefit as compared to either population alone. This suggests that TIB cells could act as enhancers of T cell responses possibly through co-stimulatory surface molecules, cytokine production and/or antigen presentation. Figure 1 summarises the diverse functional properties of B cells that have been demonstrated in mouse and human studies.

Fig. 1.

Functional properties of B cells in solid tumours. 1. B cells infiltrating various human cancers [41–48, 59] produce class-switched affinity-matured anti-tumour antibodies with protective effects [62–64]. 2. Human TIB cells produce cytokines enhancing type 1 cellular immunity [80]. 3. B cells concentrate low-dose antigens by membrane IgG-mediated antigen capture [83], and induce CD4⁺ and CD8⁺ T cell priming via antigen presentation [49, 78, 80] and cross-presentation [84–86], respectively. 4. Human B cells have been shown to secrete TRAIL, inducing in vitro tumour cell killing [96], as well as to produce Granzyme B after BCR and IL-21 receptor stimulation [97], but whether this activity results in anti-tumour effects or immune suppression is not known. 5. Human peripheral blood B cells promote the survival and proliferation of CTLs via CD70/CD27 interaction [38], but whether such mechanism is active also in tumour microenvironment is not known. 6. TIB cells acquire a regulatory phenotype and produce immunosuppressive cytokines and have been shown to suppress tumour immunity in mice [103–106]. 7. B cells produce CXCL13 and Lymphotoxin (LT), which induce the formation of tertiary lymphoid structures in mouse tumour models [118, 120], and under inflammatory conditions in humans [121]. Whether this also takes place in the human tumour microenvironment is not known. 8. TIB cells promote angiogenesis in a STAT3-dependent manner [108] or promote lymphangiogenesis [107] that facilitates tumour growth and dissemination, respectively, in mice. TIB-cell-derived LTα stimulates the recurrence of castration-resistant prostate tumours in mice [110, 111]. TIB-cell-derived BAFF might be involved in the promotion of human pancreatic cancer [113]

The data from studies on the phenotypic characterisation of meningioma [82], ovarian [78], breast [42], hepatocellular [80] and colorectal cancer [49] TIB cells provide support for these ideas. In hepatocellular cancer, TIB cells were shown to produce the Be1 cytokines IFNγ and IL-12 [80], suggesting their potential involvement in the facilitation of type 1 cellular immunity. The majority of CD20⁺ TIB cells were shown to be antigen-experienced memory cells of an atypical phenotype. These cells express high levels of molecules necessary for antigen presentation, such as MHC Class I and II, CD80, CD86 and CD40, but not the canonical memory marker CD27, which might indicate the uncoupling of antibody production and skewing towards antigen presenting cell (APC) functions [49, 78, 80]. B cells as antigen presenters might even be superior to DCs in the tumour microenvironment due to the unique potential of B cells to concentrate low-dose antigen by membrane IgG-mediated antigen capture [83] and the ability to cross-present [84–86]. Thus, TIB cells could represent the necessary APCs to stimulate tumour-specific T cells over a long period of time that is crucial for the control of human tumours [32, 87].

The loss of CD27 on Ig-switched (IgD⁻) B cells has also been shown at the systemic level in the PBMCs of melanoma, glioma, breast and pancreatic cancer patients as compared to healthy donors [88]. This observation mimics the alterations of peripheral B cell pool in the elderly [89], as well as patients suffering from SLE [90, 91]. The IgD⁻CD27⁻ or the so-called double-negative (DN) memory B cell population has been suggested to represent either cells derived via the extrafollicular pathway, incomplete germinal centre reaction or senescent, exhausted memory cells [92–94]. Whether the observed DN TIB cells originate via a particular pathway in the draining LN (or in situ) or represent the infiltrates of systemically deregulated B cells subsets characteristic of the elderly, is an open question.

A mechanism of B cell help to CD8⁺ cells, however, requires the expression of CD27. The engagement of activated CD8⁺ CTLs with B cells via CD70/CD27 interactions, but independent of antigen presentation, has been demonstrated to promote the survival and proliferation of CTLs [38]. Whether TIB cells that were observed to co-localise with CD8⁺ T cells in various human cancers [10, 78–80] also express CD27 and whether this is the pathway involved in tumour-infiltrating CTL stimulation by TIB cells, remains to be determined.

Cytotoxicity

A direct tumour-cell-killing capacity of B cells has also been shown recently. Adoptive transfer experiments in mice demonstrated that activated B cells were able to lyse autologous tumour cells in vitro through an as yet unknown mechanism [95], and human B cells mediated in vitro tumour cell killing via the TRAIL/Apo1 pathway [96]. Additionally, in acute viral infection the DN B cells produce Granzyme B in a BCR and IL-21 stimulation-dependant manner [97] when CD40 ligation is absent [98]. These cells contribute to early antiviral immunity and are speculated to participate also in the early control of neoplasia before specific T cells have emerged [99]. Such cells were further demonstrated to infiltrate human breast, cervical, prostate, colorectal and ovarian cancers and co-localise with IL-21-producing cells; however, the authors suggest that these cells are immunosuppressive because B cells isolated from healthy PBMCs and subjected to the above stimulation, upregulated markers of regulatory cells such as CD25 and IDO and produced Granzyme B that suppressed T cells via degradation of the TCRζ-chain [100]. The ability of B cells to engage tumour cells directly through protein, carbohydrate and lipid antigens together with their cytotoxicity is an attractive feature in the context of tumour immunotherapy, but very little is known about the pathways inducing the killer phenotype in either effector or regulatory context. This represents an exciting research area.

Human tumour promotion

B cells have the capacity to acquire a regulatory phenotype. These regulatory B cells function via the expression of suppressive cytokines (IL-10, TGFβ) and death ligands (FasL, TRAIL) [101]. They have mostly been studied in the context of controlling autoimmunity and transplant rejection (for recent reviews see [39, 102]). Recently, B-cell-derived IL-10 and TGFβ have been demonstrated to suppress cancer immunity in mice [103–106], but whether that applies also to human tumours in vivo is not known.

B cells might contribute to oncogenesis also by other means, such as the facilitation of lymphangiogenesis [107] or angiogenesis [108] as demonstrated in mouse melanoma models and suggested for human prostate [108] and ovarian cancers from indirect evidence [109]. Cytokines, namely LTα [110], produced by B cells infiltrating murine prostates after androgen deprivation therapy were shown to activate the IKKa–E2F1–BMI1 cascade that is necessary for prostate regeneration, but also stimulates recurrence of castration-resistant tumours [111]. However, no significant correlation between B cell infiltration and survival of prostate cancer patients undergoing androgen deprivation therapy was demonstrated [112].

A study on human pancreatic ductal adenocarcinoma (PDAC) demonstrated that infiltrating B cells might promote cancer progression via the production of BAFF (TNFSF13B) that was shown to induce the expression of epithelial-mesenchymal transition gene signature and increase the invasion and motility of a BAFF-R-positive PDAC cell line. In addition, PDAC patients had significantly increased serum levels of BAFF that correlated with advanced stage and tumour size [113]. Together, this suggests that B-cell-derived molecules can have a tumour-promoting effect in a specific physiological context defined by both tumour-intrinsic as well as B cell functional phenotype-related properties.

Insights from colorectal cancers

The study from Bindea et al. [10] on the spatiotemporal dynamics of various tumour-infiltrating immune cells in colorectal cancer used an integrated systems approach to tackle the local coordination of the various immune compartments across the phases of cancer progression and the genome plasticity of cancer cells in the same patients. By analysing over one hundred colorectal cancer patients, this pivotal study showed that B cells along with cytotoxic and memory T cells fell within the central network of adaptive immune cells in the tumour microenvironment and were significant positive prognostic factors (Table 1), especially when combined with CD8 T cells, as well as T follicular helper (T_FH) cells in the invasive margin [10]. By characterising the chromosomal alterations in tumour cells, the authors showed that the genetic loss of CXCL13 (the B cell and T_FH cell attractant) was significantly associated with shorter relapse-free survival and with lower numbers of B and T_FH cells in the invasive margin. Gene expression data further revealed that CXCL13 together with IL-21 (the major T_FH cytokine) formed a close network with B, T_FH, Th1 and cytotoxic T cells and that high abundance of each of these parameters positively affected survival. Many studies have demonstrated that TIB cells are observed within lymphoid aggregates rather than dispersed in the tumour stroma (Table 1). B cells together with T_FH cells are necessary to form the germinal centre [114] that itself is a lymphoid aggregate. Recently, the formation of such aggregates or TLS in cancer has been shown to be associated with better prognosis. Besides, as discussed below, both CXCL13 and CXCR5 (the receptor of CXCL13 on B and T_FH cells) are necessary for the development of TLS [40], and a question raises whether this might represent the physiological structure through which B and T_FH cells exert their beneficial effect on tumour immunity and, if so, how it can be induced or manipulated in tumours that lack or have “inefficient” or, possibly, tolerogenic TLS.

Tertiary lymphoid structures

Lymphoid organogenesis and neogenesis

TLS can arise in places of chronic inflammation such as persistent infection, transplant rejection, autoimmunity and cancer via the so-called process of lymphoid neogenesis, which largely resembles the development of secondary lymphoid organs (SLO) during embryogenesis (reviewed in [115, 116]). For SLO formation, the interaction between two cell types, namely lymphoid tissue inducer (LTi) cells of haematopoietic origin and organiser (LTo) cells of mesenchymal origin, is crucial. Prenatally the signal for the initial clustering of these cells is thought to be neuronal cell-derived retinoic acid that induces the expression of CXCL13 in LTo cells and attracts the first CXCR5-expressing LTi cells. Additionally, CCL21/CCR7 signalling might be involved in the initial attraction of LTi cells to the site of LN development. The contact between the clustered LTi cells induces the expression of surface lymphotoxin-α1β2 (LTα₁β₂) either through IL-7R or TRANCER (also known as RANK) signalling, enabling them to bind to the lymphotoxin-β receptor (LTβR) on LTo cells. This further induces the expression of adhesion molecules (VCAM1, ICAM1, MADCAM1) and homeostatic chemokines (CCL19, CCL21 and CXCL13) that attract additional haematopoietic cells to the site of developing LNs and creates a positive feedback loop [115]. LTo cells then differentiate into various subsets of lymphoid organ stromal cells including follicular dendritic cells (FDCs) and fibroblast reticular cells (FRC) [116]. A recent report has also demonstrated an important role of LTβR signalling in endothelial cells that facilitate the differentiation of high endothelial venules (HEVs) and the extravasation of lymphocytes, thus further supporting the development of LNs [117].

It has been shown that the same principal chemokines and their receptors, namely CXCL13/CXCR5, CCL21/CCR7, IL-7/IL-7R and LTα₁β₂/LTβR, are necessary and sufficient to induce the formation of TLS in various transgenic mouse models, as well as models of autoimmunity (see recent review in [40]). The specialised LTi cells have not been identified in adult humans, but DC, B and Th17 cells are capable of producing these cytokines and have been proposed to play the role of LTi cells during TLS formation, while smooth muscle cells and tissue-resident fibroblasts have been proposed to acquire the functions of LTo cells. The homeostatic formation of TLS in the gut, such as isolated lymphoid follicles (ILF), is induced by commensal microbiota, while various inflammatory stimuli and loss of tolerance to autoantigens can induce TLS formation in other mucosal and non-mucosal tissues [40].

B cells as LTi-like cells

The role of B cells in TLS formation has been demonstrated by several studies on ILFs. While the initial formation of immature ILFs could be induced by T-cell- or NK-cell-derived lymphotoxin (LT), naïve LT-producing B cells were crucial for the maturation of ILFs into functional lymphoid nodules [118]. Following the lead that CCR6-deficient mice show impaired mucosal immunity, the authors later reported that it was due to arrest in ILF development and that ILF B cells were the major population expressing CCR6 in wild-type animals [119]. This indicates that B cells can respond to other chemokines besides the abovementioned, namely CCL20 (the ligand of CCR6) to support the formation of TLS. In another ILF study, gut epithelial cells were engineered to overexpress CXCL13, resulting in a marked increase of infiltrating B cells and cells likely representing the adult counterparts of the embryonic LTi cells, as well as lymphoid follicles in the small intestine [120]. No antigenic stimulation was performed in this study and the majority of infiltrating B cells in response to CXCL13 overexpression in the gut epithelium were IgM-positive, indicating their naïve phenotype [120]. In human chronic obstructive pulmonary disease, B cells were shown to be important producers of CXCL13 and LT in response to TLR4 signalling that was activated by cigarette smoke extract, H₂O₂ and lipopolysacharide (LPS) in the absence of any specific antigen stimulation; furthermore, the level of CXCL13 correlated with lymphoid follicle density in the inflamed lungs [121]. These studies collectively suggest that B cells do not need to be activated by antigen in order to aggregate into TLS and that the homeostatic cytokines CXCL13 and LT (often produced by infiltrating B cells) and others (like CCL20) play a definitive part in the induction of TLS formation similar to SLO formation during embryogenesis. What are the signals and responsible cell types that produce the initial “clustering” cytokine CXCL13 in inflammatory microenvironments that further attracts B cells (or other cells with LTi capacities), as well as what is the role of antigen availability or is there a time point when it becomes crucial for this process, are all important questions when considering novel ways to control human inflammatory diseases, autoimmunity, transplant rejection as well as cancer.

TLS formation in the cancer microenvironment

The importance of LT for TLS formation has also been demonstrated in the tumour setting. First, a recombinant tumour-specific antibody–LTα fusion protein was used in a murine melanoma model and was shown to induce B cell aggregation in the tumour periphery of athymic nu/nu mice [122]. In a later study, the application of this antibody in wild-type mice with xenografted human melanomas resulted in the formation of lymphoid-like tissues in the tumour microenvironment, containing T cells, antigen-presenting cells, B cell aggregates and PNAd⁺/CCL21⁺ HEVs [123]. By using an LTα-expressing cell line as a transplanted tumour model in nude mice, it was demonstrated that the local expression of LTα, but not TNF-α, IFN-γ or IL-4, attracted naïve B cells that formed intra-tumoural follicles together with differentiated FDCs and VCAM1⁺PNAd⁺MADCAM1⁺ HEVs. This cell line in SCID mice induced the formation of lymphoid tissue stroma, containing HEVs that lacked the expression of MADCAM1, and immature FDC [124]. Although the respective receptor of LTα in this setting was not clear, these data suggest that the formation of TLS in tumours might be driven by the same homeostatic cytokines as in the abovementioned lymphoid organogenesis and neogenesis. This notion is also supported by several findings in human cancer. Studies in NSCLC showed the frequent presence of TLS containing T cells in relation to the expression of CCL19, CCL21, CXCL13, CCL17, CCL22, IL16, and adhesion molecules ICAM2, ICAM3, VCAM1, MAdCAM1 and PNAd [125]. Gene expression profiling studies have shown that high expression of 12 chemokine signature (CCL2, CCL3, CCL4, CCL5, CCL8, CCL18, CCL19, CCL21, CXCL9, CXCL10, CXCL11 and CXCL13) significantly correlated with the presence of intra- or peritumoural TLS in colorectal cancer [126] and melanoma metastases [127]. These are chemokines defining the migration of all the major cell types in lymphoid organs. Furthermore, the TLS in human colorectal cancer have been shown to possess a high degree of compartmentalisation, containing B and T cells, mature DCs and a network of CD21⁺ FDCs [128]. As mentioned earlier, colorectal cancer cells can produce CXCL13, and CXCL13-positive tumours had an increased infiltration by B and T_FH cells [10]. It would be of great interest to determine whether this results also in a TLS-enriched tumour microenvironment and suggests that some tumours themselves can generate the TLS formation-initiating signals.

Functional role of tumour-associated TLSs

In autoimmunity and transplantation immunity, TLSs have often been associated with exacerbated inflammatory reactions and graft rejection [40], and a diminished peripheral tolerance in these structures has been proposed [129, 130]. In contrast, also a regulatory capacity of TLS has been demonstrated in both autoimmunity and transplantation in a mouse model of autoimmune gastritis [131] and tolerance of renal allografts [132]. In chronic infections, the presence of TLS has been correlated to improved pathogen clearance [115], while the physiological role of TLS formation in cancer is a matter of intense debate. Studies in murine melanoma models yielded contradictory results; for example, an immunosuppressive effect of TLS was shown by Treg cell stimulation [133]. Considering the B cell role, early studies on transplanted LTα-expressing tumours demonstrated that tumour growth was much more inhibited in nude versus SCID mice [20]. Studies using recombinant antibody–LTα fusion protein demonstrated a protective effect of TLS in nu/nu mice via NK cell and B cell infiltration and stimulation [122]. In wild-type mice the infiltration by L-selectin⁺ naïve T cells through the de novo acquired HEVs and a marked clonal expansion of TILs displaying reactivity against melanoma cells was achieved in TLS, but was not present in the draining LNs, and resulted in melanoma reduction and prevention of pulmonary metastasis [123]. Thus, B-cell-/LTα-induced TLS in tumours might serve as entry portals and activation sites for tumour-specific lymphocytes, but the signals that define TLS-dependent protective versus immunosuppressive outcomes in the tumour microenvironment are unknown.

The formation and prognostic value of TLS in human tumours has been addressed by several recent studies. A study of NSCLC demonstrated that intra-tumoural germinal centres were detected significantly more often in early- than in late-stage tumours, while no difference was observed for peritumoural germinal centres [134]. TLS formation in tumour tissues of around 6 % of hepatocellular carcinoma patients was observed and found to correlate with improved survival (Wilcoxon test p = 0.01) [135]. In breast cancer, the presence of TLS was detected in about 37 % of patients and was associated with prolonged metastasis-free survival of hormone-receptor negative tumour patients (HR = 0.25, p = 0.033) [136]. Also for patients with colorectal cancer [126, 137, 138] and melanoma [127], the presence of TLS in intra-tumoural or peritumoural areas was a strong indicator of better prognosis. Furthermore, such cell types as HEVs, FDCs, and mature DC-LAMP⁺ DCs in tumours are seen exclusively in or highly enriched in TLS, thus indirectly indicating the presence of TLS, and also correlate with improved survival [125, 139–143].

The functional characteristics of TLS have been assessed so far in few human cancer studies. TLSs in breast cancer were shown to possess B cells, T cells and CD21⁺ FDCs and yield affinity-matured and clonally expanded B cells that produced breast cancer-reactive IgG antibodies [42, 45]. In contrast, the draining LNs were shown to possess a markedly decreased number of clonal groups compared with the TIB cell pool, suggesting that the intra-tumoural TLS hosted the activation of naïve B cells [42]. In lung cancer, the same cellular composition was demonstrated as in the breast cancer TLS [140]. The expression of a variety of chemokines involved in naïve and memory T cell, B cell and mature DC attraction, as well as the presence of these cell types in the lymphoid structures was also later demonstrated; the authors suggest that the TLS in NSCLC establish a favourable microenvironment for T cell priming by tumour antigen-loaded DC [125, 142]. The characterisation of human primary melanoma and metastatic lesions has revealed that TLS containing B cells, T cells, FDCs and HEVs form in about 1/7th of cutaneous metastasis and host the affinity maturation of TIB cells, but primary melanomas merely host the formation of B cell aggregates surrounding HEVs, but lacking FDCs [139]. In breast cancer, high densities of HEVs were shown to correlate with elevated numbers of infiltrating B cells, T cells and DC-LAMP⁺ LTβ-producing mature DCs [143]. Together, these data support the notion that tumour-associated TLS allow for the generation of antigen-specific immune responses in situ that are involved in the control of human cancers. The induction of antigen-specific regulatory cells in these structures is another plausible outcome, and further research is warranted to determine the signals that shape the generated immune cell functionality in tumour-associated TLS.

B cells and TLS in cancer therapy

Predictive value

Combination immunotherapy with different standard treatments may be superior to either separately, owing to the immunostimulatory features of low-dose chemo- and radiotherapies [1]. Hence, the underlying mechanisms and effects of standard cancer therapies on various immune cells have become a very intense field of research. The assessment of the infiltration of effector and regulatory T cells has proven to be among the most precise prognostic biomarkers in various cancers, and several studies have analysed also their predictive value (recently reviewed in [144]). To the best of our knowledge, there are currently no prospective clinical trials that included the assessment of pre-existing or therapy-induced TIB cells or tumour-associated TLS for the prognosis or the prediction of therapy outcome. Some studies have analysed these factors in prospective and retrospective cohorts and addressed the changes in the pre- and post-treatment tumour samples. The resected tumours from melanoma patients that remained disease-free more than 5 years after cancer vaccine therapy (n = 9) were characterised by prominent B cell infiltration when compared to the rest of the patients (n = 40) [145]. Intra-tumoural or peritumoural infiltration of CD20⁺ B cells has been associated with improved clinical outcome of BCG-based immunotherapy for melanoma [146] and bladder carcinoma patients [147], as well as IRX-2 immunotherapy for head and neck cancer patients [148]. A study of ovarian cancer showed that in recurrent lesions all analysed immune cell types, including B cells, were decreased [149]. In contrast, another study demonstrated that B cells in the ovarian cancer metastatic effusion samples after platinum-based chemotherapy correlated with worse overall survival (OS) in univariate Cox regression analysis. However, the authors conclude that the analysed immune response parameters (including NK cells, B cells, T cells and expression of various chemokine receptors) were weak predictors of outcome in this setting [150]. Neoadjuvant chemotherapy of breast cancer was also shown to decrease the number of TIB cells and CD4⁺ T cells in residual tumours, while the percentage of CD8⁺ T cells was not altered; the authors propose that this represents a shift from Th2 immunity towards a more beneficial cytotoxic response. However, no survival analysis was performed [151], while other studies convincingly demonstrate an association of TIB cells with improved outcome after various breast cancer treatments. CD20 expression was associated with good prognosis in patients who received chemotherapy (p = 0.012, log-rank = 6.30) in univariate Cox regression analysis, and in patients who received endocrine therapy also in multivariate analysis (HR = 0.61, p = 0.031) [30]. Gene expression profiling in LN-negative highly proliferative breast cancer patients showed that the high expression of all three metagenes, namely B cell/plasma cell, T cell/NK cell and monocyte/DC metagenes, significantly improved metastasis-free survival as compared to any other marker combination irrespective of the ER status and post-operative treatment (none, tamoxifen monotherapy or chemotherapy) [26]. Another study on node-negative breast cancer showed the importance of Ig kappa chain (IGKC) as a positive predictor of anthracycline-based neoadjuvant chemotherapy efficacy (p < 0.001) [24]. These data provide the first evidence for a link between TIB cells and improved response to immunotherapies and standard cancer treatments.

Considering TLS, a comprehensive study addressing the importance of various CD4 helper T cell subsets in breast cancer patients with a 10-year follow-up demonstrated a significant association of T_FH gene signature or CXCL13 alone to pathological complete response after neoadjuvant chemotherapy for all histological cancer subtypes [152]. In that study, the authors demonstrated that the observed elevated expression of a T_FH gene signature reflects the formation of tumour-associated TLS. Another study on hormone receptor-negative breast cancer patients showed that the presence of TLS correlated with longer distant disease-free survival (DFS), while for patients with hormone receptor-positive tumours, the adjuvant endocrine therapy improved survival in the absence of TLS [136]. These data suggest that TLS can affect various treatments differently and needs to be explored in detail in the context of response to each specific therapy type.

Therapeutic applications

Considering the cancer-promoting capacities of B cells reported by many mouse studies, the inhibition of B cells for example by using the anti-CD20 antibody Rituximab^® would seem an idea worth trying in order to improve cancer therapy [153, 154]. A few studies report such attempts in colorectal cancer [155], renal cell carcinoma and melanoma [156], but no marked clinical benefit was achieved, which might be explained by the notion that Rituximab^® does not inhibit the CD20^low B regulatory cells [153]. A study in mice using resveratrol showed that this treatment effectively inhibited lung metastasis by preventing the generation and function of tumour-evoked B regulatory cells [105]. Nevertheless, other studies in mice have provided data for the usefulness of B cells as therapeutic effectors exhibiting both T cell stimulatory and direct tumouricidal activities (recently reviewed in [33]). Recent reports further support this notion. The immunogenicity of DC-derived exosomes loaded with cancer antigens as a cancer immunotherapy tool was shown to depend on both intact T and B cells [157, 158]. Also the role of CD40 on DCs and B cells has gained appreciation as a potential anti-tumour immunity stimulator via the CD40-CD40L activating pathway [159]. In a mesothelioma model, low-dose stimulatory anti-CD40 antibody administered directly into the tumour bed avoided toxic side effects (reported for high-dose intra-venous application) and prolonged survival in 60 % of the mice, with most actually cured. CD4⁺ or CD8⁺ depletion only marginally decreased the efficacy of anti-CD40 antibody treatment, the presence of follicular B cells was required [160]. After CD40 stimulation, the potential of B cells as APCs and T cell activators is comparable to DCs; in addition, B cells can proliferate in vitro more easily than DCs, thus offering an attractive tool for adoptive cellular immunotherapy [159].

The possible significance of TLS induction for cancer therapy can be appreciated indirectly in a study using a unique orthotopic colon cancer model in wild-type, RAG^−/− and CXCR5^−/− mice [10]. It was demonstrated that (i) tumours grew in CXCR5^−/− mice with similar accelerated kinetics as in RAG^−/− mice when compared to wild-type mice, and that (ii) the injection of recombinant CXCL13 endoscopically within the colonic submucosa of wild-type mice resulted in tumour rejection in 80 % (4/5) of cases, but not in CXCL13-untreated mice (0/5) [10]. The authors did not analyse the formation of TLS, but taking into account the pivotal role of CXCL13 in the formation of lymphoid structures, it is tempting to speculate that its local delivery in cancer might result in the formation of TLS, contributing to the beneficial outcome and represents a possible new option to improve cancer therapy.

Conclusions

Overall, the literature provides substantial data that B cells are important contributors to tumour control and warrants updating the current paradigm of B cells from sole tumour-promoting immune cells to multifaceted effectors capable of different functions including the formation of TLS. It is likely that the effect of the TIB cells on anti-tumour immune response depends on the combination of various factors, including the tissue of origin, the phenotype of tumour cells, the TIB cell functional subtype, as well as the applied therapy. TIB cells frequently form aggregates or TLS together with T cells and FDCs in the tumour microenvironment and the presence of such aggregates often correlates with improved survival in human cancer. Thus, the stimulation of B cells and TLS formation using homeostatic cytokines such as CXCL13 and lymphotoxin in the tumour microenvironment might facilitate the development of physiological structures assisting immune cell entry into the tumour, as well as generating protective immune responses, and offers novel possibilities to improve the efficiency of cancer vaccines. Further studies are necessary to characterise the mechanisms and the relevance of TIB cells in TLS formation in cancer, as well as to define the effects of standard therapies on the TIB cell and TLS functionality. This approach might yield novel beneficial treatments or treatment combinations.

Abbreviations

APC

Antigen presenting cell

CTL

Cytotoxic T lymphocyte

DC

Dendritic cell

DN

Double negative

ER

Estrogen receptor

FDC

Follicular dendritic cell

HEV

High endothelial venule

HR

Hazard ratio

Ig

Immunoglobulin

IGKC

Immunoglobulin kappa chain

ILF

Isolated lymphoid follicle

LN

Lymph node

LT

Lymphotoxin

LTi cell

Lymphoid tissue inducer cell

LTo cell

Lymphoid tissue organiser cell

NK cell

Natural killer cell

NSCLC

Non-small-cell lung cancer

PDAC

Pancreatic ductal adenocarcinoma

SLO

Secondary lymphoid organ

T_FH cell

T follicular helper cell

Th cell

T helper cell

TIB cell

Tumour-infiltrating B cell

TLS

Tertiary lymphoid structure

Treg cell

T regulatory cell