In solid tumors, the presence of lymph node–like structures called tertiary lymphoid structures (TLS) is associated with improved patient survival. However, little is known about how TLS develop in cancer, how their function affects survival, and whether they are affected by cancer therapy. In this study, we used multispectral microscopy, quantitative pathology, and gene expression profiling to analyze TLS formation in human lung squamous cell carcinoma (LSCC) and in an experimental model of lung TLS induction. We identified a niche of CXCL13+ perivascular and CXCL12+LTB+ and PD-L1+ epithelial cells supporting TLS formation. We also characterized sequential stages of TLS maturation in LSCC culminating in the formation of germinal centers (GC). In untreated patients, TLS density was the strongest independent prognostic marker. Furthermore, TLS density correlated with GC formation and expression of adaptive immune response–related genes. In patients treated with neoadjuvant chemotherapy, TLS density was similar, but GC formation was impaired and the prognostic value of TLS density was lost. Corticosteroids are coadministered with chemotherapy to manage side effects in LSCC patients, so we evaluated whether they impaired TLS development independently of chemotherapy. TLS density and GC formation were each reduced in chemotherapy-naïve LSCC patients treated with corticosteroids before surgery, compared with untreated patients, a finding that we confirmed in the experimental model of lung TLS induction. Overall, our results highlight the importance of GC formation in TLS during tumor development and treatment.

Significance: Corticosteroid treatment during chemotherapy negatively affects the development of tertiary lymphoid structures and abrogates their prognostic value in patients with lung cancer. Cancer Res; 78(5); 1308–20. ©2018 AACR.

Tertiary lymphoid structures (TLS) resemble follicles of secondary lymphoid organs (SLO) both by structure and function and develop in response to chronic inflammation (1) and cancer (2, 3). The presence of germinal centers (GC) within TLS correlates with exacerbated autoimmunity and transplant rejection due to the generation of autoreactive B cells and high endothelial venules (HEV; ref. 1). In models for infection, TLS serve as priming sites for T cells and contribute to pathogen clearance (4, 5). The first prognostic analysis of TLS in human cancer showed that TLS containing GC correlate with improved survival for patients with hepatocellular carcinoma (6). Further studies showed that either total TLS numbers or the numbers of TLS-associated immune cells including follicular helper T (TFH) cells, follicular B cells, mature dendritic cells (DC), and HEVs are associated with prolonged survival in many different tumor types (3, 7–10). Specifically in non–small cell lung cancer (NSCLC), the number of CD208+ (DC-LAMP+) antigen-presenting mature DCs, which reside in T-cell areas of TLS, significantly correlated with improved survival (11), especially in combination with the number of CD8+ T cells (12) or follicular B cells (13). Tumor-associated TLS provide the necessary specialized vasculature and chemoattractants that allows T-cell infiltration (14) and significantly correlate with tumor-infiltrating T cells and plasma cells (13, 15, 16). B-cell and T-cell receptor sequencing data show clonal expansion of lymphocytes in tumor-associated TLS, suggesting a direct role for TLS in the priming of antitumor immunity (17–19). Moreover, the application of CXCL13 or lymphotoxin (LT), two components of a positive feedback-loop orchestrating the organogenesis of SLO (20) and TLS development in adult mice (21), resulted in tumor control in models of colorectal cancer (22) or melanoma (23). To normalize tumor-associated blood vessels, Johansson–Percival and colleagues delivered LIGHT to such vessels, resulting in susceptibility of RIP-TAg pancreatic tumors to treatment with anticancer vaccines and immune checkpoint inhibition. Interestingly, targeting LIGHT to tumor-associated blood vessels induced formation of TLS, which can be explained by the fact that LIGHT is an essential ligand for SLO and TLS development (24). In human pancreatic ductal adenocarcinoma, TLS developed in response to GVAX immunotherapy in around 85% of patients and their presence is associated with an increased ratio of intratumoral Teffector:Treg cells, improved postvaccination responses and prolonged survival (25).

In light of these studies, deliberate induction of tumor-associated TLS has been proposed as a novel immunotherapeutic approach (2, 3). However, the mechanisms underlying TLS development in different human tissues are poorly understood and may be affected by the nature of the inflammatory stimulus and the organ-specific environments (7). In lungs for example, viral or bacterial stimuli activate the LT/CXCL13 or IL17/CXCL12 pathway, respectively, resulting in the formation of TLS, also known as inducible bronchus associated lymphoid tissue (iBALT; ref. 26). The LT/CXCL13 pathway is crucial for differentiation of follicular dendritic cells (FDC) and functional GC (26, 27). Both pathways participate in TLS development in response to cigarette smoke or LPS, which is a major component of cigarette smoke (28), in patients and animal models of chronic obstructive pulmonary disease (COPD; refs. 29, 30). Cigarette smoke and inhaled particulates such as alum drive the release of IL1α by alveolar macrophages and lead to iBALT formation (31–33). Additional factors like CCL19, CCL21, and IL22 also contribute to iBALT development (34, 35), but which stimuli are upstream of these pathways is not known. A crucial role for B cells, T cells, and DCs in the generation and maintenance of TLS has been demonstrated in various animal models (35). Finally, Krautler and colleagues showed that ubiquitous perivascular PDGFRβ+ mesenchymal cells are the precursors of FDCs, which orchestrate B-cell accumulation and TLS development in a LTβ receptor (LTBR)-dependent manner explaining why TLS can arise in any vascularized organ (36). The abovementioned chemokines and immune cells are linked to the presence of TLS in human cancers (3, 37). However, there is little information about the sequence of events that drive the development of tumor-associated TLS or about which features of TLS are relevant for patients' survival. To address these questions, we characterized the TLS-associated microenvironment and TLS composition in human LSCC tissues and an experimental model of lung TLS induction. We identified three distinct maturation stages of TLS culminating in GC formation with significant relevance for patient survival and expression of adaptive immunity-related genes. Corticosteroids are frequently used to manage side effects of chemo- or radiotherapy or to treat other comorbidities of cancer patients. Our data suggest that corticosteroids have a negative impact on TLS development and specifically on GC formation in LSCC patients and lungs of mice.

Patient material

Patient analyses were conducted according to the Declaration of Helsinki. Ethical approval for performing research on anonymized, archival patient material was obtained from the cantonal ethics commission Zurich (KEK-ZH-2013-0584 amendment 26.11.2015, KEK-ZH-29-2009 amendment 20.05.2016), and Ethics Committee of Biomedical Research of Riga East University Hospital with signed informed consents.

Samples from patients who underwent resection of LSCC regardless of tumor stage in University Hospital Zurich (USZ) were collected between 2003 and 2010, and were analyzed for their baseline and outcome clinical information retrospectively from the in-house diagnostic and treatment databases (n = 245) by a clinical pathologist (A. Soltermann) and a medical oncologist (A. Curioni-Fontecedro). Following criteria were used to exclude patients: Mixed tumor histology, prior diagnosis of a squamous cell carcinoma of head and neck, or incomplete diagnostic and clinical data. Patients who passed the exclusion criteria were sorted by tumor stage and similar proportions of early- and late-stage patients were selected for analysis (Zurich cohort, n = 138, Table 1). These patients were followed-up until December 2015 (median follow-up time from diagnosis 1,429 days).

Table 1.

Clinical and histologic characteristics of the LSCC patient cohort (n = 138) from the University Hospital Zurich

CharacteristicsNumber (%)Median PFSHRa (95% CI)
Gender 
 Male 111 (80.4) 827 1.1 
 Female 27 (19.6) 1,060 (0.6–2.1) 
Age (mean ± SEM-63 ± 0.8)   
 <Median 60 (43.5) 894 (0.97–1) 
 = Median (62) 12 (8.7) 1,236  
 >Median 66 (47.8) 796  
Smoking history   0.9 
 Current 79 (57.2) 1,172 (0.7–1.8) 
 Former 59 (42.8) 555  
Pack-years (mean ± SEM-52±2)   
 <Median 56 (40.6) 533 (0.99–1) 
 = Median (50) 22 (15.9) 1,539  
 >Median 56 (40.6) 916  
 ND 4 (2.9) 1,187  
pT/ypT 
 0 7 (5.1) 1,434 1.3g 
 1a 26 (18.8) 1,390 (1.1–1.5) 
 1b 13 (9.4) 2,167  
 2a 27 (19.6) 1,123  
 2b 11 (8.0) 1,055  
 3 31 (22.5) 652  
 4 23 (16.7) 360  
pN 
 0 62 (44.9) 1,394 1.7h 
 1 48 (34.8) 761 (1.2–2.3) 
 2 28 (20.3) 418  
pM 
 0 132 (95.7) 1,058 7.3g 
 1a 3 (2.2) 180 (2.9–18) 
 1b 3 (2.2)  
Stage 
 0 6 (4.3) 1,518 1.4g 
 1a 22 (15.9) 1,678 (1.2–1.7) 
 1b 10 (7.2) 1,073  
 2a 29 (21.0) 1,283  
 2b 16 (11.6) 797  
 3a 43 (31.2) 414  
 3b 8 (5.8) 512  
 4 4 (2.9) 91  
Neoadjuvant chemotherapyb   
 Yes 51 (37.0) 495 (0.62–1.7) 
 No 87 (63.0) 1,060  
Grade    
 2 61 (44.2) 1,060 1.1 
 3 77 (55.8) 725 (0.55–1.83) 
Tumor size (mean ± SEM = 4.3 ± 0.2)   1.2g 
 <Median 67 (48.6) 1,471 (1.1–1.3) 
 = Median (3.8) 4 (2.9) 415  
 >Median 67 (48.6) 451  
Pleural Invasion   1.5 
 Yes 41 (29.7) 509 (0.92–2.5) 
 No 97 (70.3) 1,123  
Vascular invasion   1.6 
 Yes 55 (39.9) 414 (0.97–2.6) 
 No 83 (60.1) 1,146  
Adjuvant therapyc   1.9h 
 Chemotherapy 23 (16.7) 509 (1.3–2.8) 
 Radiotherapy 16 (11.6) 339  
 Chemo+Radio 5 (3.6) 414  
 None 94 (68.1) 1,162  
Corticosteroid treatmentd   1.2 
 Local 12 (8.7) 627 (0.74–2) 
 Systemic 57 (41.3) 495  
 None 63 (45.7) 1,172  
 ND 6 (4.3) 720  
TLS densitye   0.4h 
 TLS-low 47 (34.1) 451 (0.26–0.71) 
 TLS-high 91 (65.9) 1,381  
GC formationf   0.6 
 No 53 (38.4) 495 (0.39–1) 
 Yes 85 (61.6) 1,123  
TIL score (mean ± SEM = 4.1 ± 0.1)   0.75h 
 <Median 47 (34.1) 461 (0.6–0.93) 
 = Median (4) 21 (15.2) 410  
 >Median 57 (41.3) 1,178  
 ND 13 (9.4) 1,511  
Plasma cell score (mean ± SEM = 3.2 ± 0.2)   0.81h 
 <Median 56 (40.6) 465 (0.7–0.94) 
 = Median (3.5) 9 (6.5) 1,055  
 >Median 60 (43.5) 1,426  
 ND 13 (9.4) 1,511  
CharacteristicsNumber (%)Median PFSHRa (95% CI)
Gender 
 Male 111 (80.4) 827 1.1 
 Female 27 (19.6) 1,060 (0.6–2.1) 
Age (mean ± SEM-63 ± 0.8)   
 <Median 60 (43.5) 894 (0.97–1) 
 = Median (62) 12 (8.7) 1,236  
 >Median 66 (47.8) 796  
Smoking history   0.9 
 Current 79 (57.2) 1,172 (0.7–1.8) 
 Former 59 (42.8) 555  
Pack-years (mean ± SEM-52±2)   
 <Median 56 (40.6) 533 (0.99–1) 
 = Median (50) 22 (15.9) 1,539  
 >Median 56 (40.6) 916  
 ND 4 (2.9) 1,187  
pT/ypT 
 0 7 (5.1) 1,434 1.3g 
 1a 26 (18.8) 1,390 (1.1–1.5) 
 1b 13 (9.4) 2,167  
 2a 27 (19.6) 1,123  
 2b 11 (8.0) 1,055  
 3 31 (22.5) 652  
 4 23 (16.7) 360  
pN 
 0 62 (44.9) 1,394 1.7h 
 1 48 (34.8) 761 (1.2–2.3) 
 2 28 (20.3) 418  
pM 
 0 132 (95.7) 1,058 7.3g 
 1a 3 (2.2) 180 (2.9–18) 
 1b 3 (2.2)  
Stage 
 0 6 (4.3) 1,518 1.4g 
 1a 22 (15.9) 1,678 (1.2–1.7) 
 1b 10 (7.2) 1,073  
 2a 29 (21.0) 1,283  
 2b 16 (11.6) 797  
 3a 43 (31.2) 414  
 3b 8 (5.8) 512  
 4 4 (2.9) 91  
Neoadjuvant chemotherapyb   
 Yes 51 (37.0) 495 (0.62–1.7) 
 No 87 (63.0) 1,060  
Grade    
 2 61 (44.2) 1,060 1.1 
 3 77 (55.8) 725 (0.55–1.83) 
Tumor size (mean ± SEM = 4.3 ± 0.2)   1.2g 
 <Median 67 (48.6) 1,471 (1.1–1.3) 
 = Median (3.8) 4 (2.9) 415  
 >Median 67 (48.6) 451  
Pleural Invasion   1.5 
 Yes 41 (29.7) 509 (0.92–2.5) 
 No 97 (70.3) 1,123  
Vascular invasion   1.6 
 Yes 55 (39.9) 414 (0.97–2.6) 
 No 83 (60.1) 1,146  
Adjuvant therapyc   1.9h 
 Chemotherapy 23 (16.7) 509 (1.3–2.8) 
 Radiotherapy 16 (11.6) 339  
 Chemo+Radio 5 (3.6) 414  
 None 94 (68.1) 1,162  
Corticosteroid treatmentd   1.2 
 Local 12 (8.7) 627 (0.74–2) 
 Systemic 57 (41.3) 495  
 None 63 (45.7) 1,172  
 ND 6 (4.3) 720  
TLS densitye   0.4h 
 TLS-low 47 (34.1) 451 (0.26–0.71) 
 TLS-high 91 (65.9) 1,381  
GC formationf   0.6 
 No 53 (38.4) 495 (0.39–1) 
 Yes 85 (61.6) 1,123  
TIL score (mean ± SEM = 4.1 ± 0.1)   0.75h 
 <Median 47 (34.1) 461 (0.6–0.93) 
 = Median (4) 21 (15.2) 410  
 >Median 57 (41.3) 1,178  
 ND 13 (9.4) 1,511  
Plasma cell score (mean ± SEM = 3.2 ± 0.2)   0.81h 
 <Median 56 (40.6) 465 (0.7–0.94) 
 = Median (3.5) 9 (6.5) 1,055  
 >Median 60 (43.5) 1,426  
 ND 13 (9.4) 1,511  

NOTE: TNM was assessed according to the 7th Edition of the TNM Staging System for non-small cell lung cancer.

Abbreviation: ND, not determined.

aHR was calculated by the univariate Cox regression analysis for PFS comparing no versus yes or low versus high groups in case of categorical variables (pleural invasion, vascular invasion, neoadjuvant and adjuvant chemotherapy, steroid treatment, GC formation, TLS density) or comparing low to increasing values for continuous variables (age, pack years, pT, pN, pM, stage, grade, size). Patients that had died sooner than 1 month after surgery were excluded from this analysis (n = 6).

bFor neoadjuvant chemotherapy, platinols were used in combination with either gemcitabine (n = 19) or taxols (n = 27) or others (n = 5).

cHR was calculated for all neoadjuvant chemotherapy-treated patients versus untreated patients.

dTreatment of corticosteroids was considered if applied up to one month before surgery. All neoadjuvant chemotherapy-treated patients received systemic corticosteroids. Eighteen chemotherapy-naïve patients received corticosteroids because of cancer nonrelated comorbidities either systemically or locally (inhalation). HR was calculated for all steroid-treated versus untreated patients.

eTLS density was measured as the number of TLS per mm2 in the tumor periphery and defined as high if >0.165 TLS/mm2.

fA tumor was considered GC-positive if at least one TLS showed the characteristic morphology of proliferating centroblasts.

gP < 0.0005, Wald test.

hP < 0.005, Wald test.

Information on patients who underwent resection of lung cancer in Riga Eastern Clinical University Hospital (Riga cohort, n = 78) is provided in Supplementary Methods and Supplementary Table S1.

Histologic evaluation

The number of dense lymphocytic aggregates was quantified per 10× high-power field (HPF) in all tumor-containing hematoxylin and eosin (H&E)-stained diagnostic sections of the Zurich cohort. TLS density was calculated as the number of TLS per mm2 in peritumoral and intratumoral regions. A patient was considered as GC-positive if at least one TLS showed the characteristic morphology of proliferating centroblasts. The sample analyst (K. Silina) was blinded for the patients' clinical data at the time of histologic evaluation. Further details are provided in the Supplementary Methods.

Gene expression analysis

A representative set of 27 chemotherapy-naïve patients was selected to match the whole cohort for the variability of TLS density, survival, and stage and included 10 TLS-low and 17 TLS-high tumors. Full details on tissue preparation and qPCR data analysis according to the MIQE guidelines (38) are provided in Supplementary Methods and Supplementary Fig. S1A and S1B. Primer sequences are shown in Supplementary Table S2.

Immunostaining and quantitative pathology

Antibodies and detection reagents used for immunostaining are listed in Supplementary Table S3. Full details on IHC and immunofluorescence (IF) staining protocols are provided in Supplementary Methods. To analyze TLS maturation in different TLS density groups, a set of 61 chemotherapy-naïve patients was selected to match the whole cohort for the variability of TLS density, survival, and stage and included 24 TLS-low and 37 TLS-high tumors (Supplementary Fig. S1C). To analyze the effects of neoadjuvant treatments, 28 TLS-high patients treated with neoadjuvant chemotherapy (Zurich cohort) and all TLS-high patients that received neoadjuvant radiotherapy (Riga cohort, n = 4) were selected. To analyze the effects of corticosteroids, chemotherapy-naïve patients of the Zurich cohort that had received corticosteroids before surgery to treat other comorbidities (n = 15; Supplementary Table S4) were compared with patients without any steroid or chemotherapy treatment (n = 43).

For TLS maturation analysis, slides were costained for CD21, CD23, and CXCL13. All dense lymphocytic aggregates irrespective of CD21/CD23 signal were imaged as multispectral HPF (200×) by Vectra 3.0 imaging system (PerkinElmer). Spectral unmixing is described in Supplementary Methods. Each HPF image was evaluated for the TLS maturation stage as follows: early TLS (E-TLS), dense lymphocytic aggregates lacking CD21 and CD23 signal; primary follicle-like TLS (PFL-TLS), dense lymphocytic aggregates with CD21 but no CD23 signal; secondary follicle-like TLS (SFL-TLS), dense lymphocytic aggregates with CD21 and CD23 signal. The proportion of TLS in each maturation stage was determined for each patient. To determine the size of GCs, tissue segmentation algorithm of Inform software (PerkinElmer) was trained to recognize areas that coexpressed CD21 and CD23. The GC size was calculated as the average pixel count per GC area for each patient. Segmentation quality was verified for all images by the sample analyst (K. Silina) who was blinded for the patients' clinical data at the time of image analysis. Inappropriately segmented images were excluded from calculations. For the analysis of DCs and HEVs in the context of TLS maturation, slides were costained for CD20, CD21, CD23, PNAD, and DC-LAMP. All dense lymphocytic aggregates and all areas with PNAD and DC-LAMP signal were imaged and analyzed as above.

In vivo experiments

All animal experiments were approved by the cantonal veterinary office Zurich (license number 127/2015) and conducted in accordance to the Swiss federal and cantonal law on animal protection. Female C57BL/6NRj mice (8–12 weeks old) were obtained from Janvier Labs (Saint Berthevin, France) and maintained in specific pathogen-free facilities at the University of Zurich (Zurich, Germany). Mice received intranasal (i.n.) administration of alum and ovalbumin (Ova) as described previously (33) and received daily oral dexamethasone (0.3 mg/kg) or PBS. Full details on treatment protocols and analysis are provided in Supplementary Methods.

Statistical analysis

Statistical analyses were performed using RStudio (V1.0.44) software. The threshold of TLS density was identified using the receiver operating characteristics (ROC) curve analysis. Three years of progression-free survival (PFS) was set as the discriminant for short- and long-term survival. Neoadjuvant chemotherapy-treated patients (n = 51) as well as chemotherapy-naïve patients whose follow-up was shorter than 3 years without relapse (n = 8) were excluded from this analysis. The prognostic significance of TLS density was assessed by the Kaplan–Meier curve, univariate, and multivariate Cox regression analyses. Patients who died up to one month after surgery (n = 6) were excluded from PFS and disease-free survival (DFS) analysis. PFS was calculated from date of surgery for patients with complete resection (R0; n = 121) or from the end date of adjuvant therapy for R>0 patients (n = 11) until diagnosis of relapse. DFS was analyzed for R0 patients without pleural infiltration (n = 118) from the date of surgery until the date of relapse. Disease-specific (DSS) and overall survival (OS) was defined as the number of days from surgery until death and was analyzed for all patients (n = 138). Diagnoses of relapse or cancer-related death were considered as events for the PFS/DFS and DSS/OS analysis, respectively. Patients were censored at the day of last follow-up if lost from follow-up, diagnosed with a new unrelated tumor, died of a cancer nonrelated cause (not for OS), were alive (for DSS/OS), and not relapsed (for DFS/PFS).

Quantification data are presented in dot plots with interquartile ranges and Tukey whiskers, or percent bar plots. Two-tailed Wilcoxon rank-sum test was used to compare all measured parameters between LSCC patient groups. For gene expression analysis, the Benjamini–Hochberg method was used to correct the P values for multiple testing. The correlation between gene expression, histologic, and clinical parameters was assessed by the Spearman correlation or χ2 test with Yates' correction.

Multiple group comparison was done as follows: Quantification of TLS maturation in chemotherapy-naïve patients versus neoadjuvant chemotherapy and radiotherapy-treated patients was compared by Kruskal–Wallis test with Dunn multiple comparison test. For the mouse experiment, groups were compared by the one-way ANOVA with Dunnett multiple comparison test. Alum-only mice (group 2) were compared with negative control mice (group 1) and alum + Ova mice (group 3). Alum + Ova mice (group 3) were compared to alum-only (group 2) and dexamethasone-treated mice (group 4).

The development of tumor-associated TLS follows sequential stages of maturation

We initially assessed the presence of TLS in H&E-stained primary LSCC tissues using full diagnostic cases of 138 patients from the University Hospital Zurich (Table 1). We defined TLS as dense lymphocytic clusters and found that >95% of tumors contain such structures predominantly in the tumor periphery (Fig. 1A–C), especially in the bronchoalveolar tissues (Fig. 1D). In about 50% of tumors, we found TLS with a characteristic morphology of GC (Fig. 1A, arrowheads). Because such GC-containing TLS were highly similar to the secondary follicles in SLO (Fig. 1E and F), we proposed that TLS without GC represents the analogues of the lymph node primary follicles, which contain mature FDC networks but lack GC. We used multiparameter immunofluorescence (IF) to test this in serial LSCC sections spanning a depth of 250 μm. Indeed, TLS containing mature CD21+ FDC and lacking GC (CD23) were present in the lungs of LSCC patients (Fig. 1G). Moreover, we observed a third TLS phenotype, namely dense lymphocytic clusters containing B and T cells without any apparent FDC networks or GC (Fig. 1H and I; Supplementary Fig. S2A–S2C). All three TLS phenotypes shared a common localization near CXCL13+ blood vessels (Fig. 1F–I, arrowheads, Supplementary Fig. S2A–S2C, arrowheads). Because CXCL13-expressing perivascular cells, B and T cells are essential for the development of TLS, we propose that clusters without FDC represent the first stage of TLS development and term these here as early TLS (E-TLS). In analogy to the stages of SLO follicles, we called FDC-containing TLS without GC as primary follicle-like TLS (PFL-TLS), and GC-containing TLS as secondary follicle-like TLS (SFL-TLS).

Figure 1.

Development of LSCC-associated TLS. A, A representative image of H&E-stained LSCC showing TLS in the adjacent normal tissue (N) in close proximity to the invasive front (white dashed line) of the tumor (T). GC in TLS were identified by a lighter central area (arrowheads and left zoomed-in segment). Scale bar, 500 μm. B and C, The number of dense lymphocytic aggregates was determined in peritumoral lung parenchyma (B) and intratumoral regions (C) in 138 LSCC patients. The number of TLS per mm2 defined the TLS density for each patient. D, TLS density was compared in bronchoalveolar tissue (n = 138) and pleura where possible (n = 14) by two-tailed Wilcoxon rank-sum test. E, The structure and phenotype of GC-containing TLS was analyzed in serial FFPE LSCC tissues from 5 patients with high TLS density by IHC. Scale bar, 100 μm. F–H, IF staining for FDCs (CD21), GC cells (CD23), and CXCL13 was performed on serial sections spanning a depth of 250 μm of two LSCC patients with high TLS density. Slides were imaged using the multispectral microscopy system Vectra 3.0 (PerkinElmer). Each TLS was evaluated through the different depths and the images showing the widest TLS diameter (the central part) are shown. SFL-TLS, secondary follicle-like TLS with a GC reaction (white circle); PFL-TLS, primary follicle-like TLS with differentiated FDCs (white circle); E-TLS, early TLS. Arrowheads, TLS-associated CXCL13+ blood vessel. See also Supplementary Fig. S2. I, IF staining for T cells (CD3), B cells (CD20), and CXCL13 was performed for 5 LSCC patients with high TLS density. Slides were imaged by AxioScan (Zeiss). A representative image showing dense lymphocytic cluster forming around a CXCL13+ blood vessel (arrowhead) is shown. Scale bars, 100 μm. J–K, IF staining for FDCs (CD21), GC cells (CD23), B cells (CD20), DCs (DC-LAMP), and HEVs (PNAD) was performed on sections from 5 LSCC patients with high TLS density. J, The number of TLS positive for HEVs and DCs was determined and compared in each TLS maturation stage. K, Representative images of TLS in each maturation stage and nonorganized lymphocytic infiltrate containing DCs (white arrowheads), HEVs (magenta arrowheads), and DC-LAMP–expressing lung epithelial cells (cyan arrowheads) are shown.

Figure 1.

Development of LSCC-associated TLS. A, A representative image of H&E-stained LSCC showing TLS in the adjacent normal tissue (N) in close proximity to the invasive front (white dashed line) of the tumor (T). GC in TLS were identified by a lighter central area (arrowheads and left zoomed-in segment). Scale bar, 500 μm. B and C, The number of dense lymphocytic aggregates was determined in peritumoral lung parenchyma (B) and intratumoral regions (C) in 138 LSCC patients. The number of TLS per mm2 defined the TLS density for each patient. D, TLS density was compared in bronchoalveolar tissue (n = 138) and pleura where possible (n = 14) by two-tailed Wilcoxon rank-sum test. E, The structure and phenotype of GC-containing TLS was analyzed in serial FFPE LSCC tissues from 5 patients with high TLS density by IHC. Scale bar, 100 μm. F–H, IF staining for FDCs (CD21), GC cells (CD23), and CXCL13 was performed on serial sections spanning a depth of 250 μm of two LSCC patients with high TLS density. Slides were imaged using the multispectral microscopy system Vectra 3.0 (PerkinElmer). Each TLS was evaluated through the different depths and the images showing the widest TLS diameter (the central part) are shown. SFL-TLS, secondary follicle-like TLS with a GC reaction (white circle); PFL-TLS, primary follicle-like TLS with differentiated FDCs (white circle); E-TLS, early TLS. Arrowheads, TLS-associated CXCL13+ blood vessel. See also Supplementary Fig. S2. I, IF staining for T cells (CD3), B cells (CD20), and CXCL13 was performed for 5 LSCC patients with high TLS density. Slides were imaged by AxioScan (Zeiss). A representative image showing dense lymphocytic cluster forming around a CXCL13+ blood vessel (arrowhead) is shown. Scale bars, 100 μm. J–K, IF staining for FDCs (CD21), GC cells (CD23), B cells (CD20), DCs (DC-LAMP), and HEVs (PNAD) was performed on sections from 5 LSCC patients with high TLS density. J, The number of TLS positive for HEVs and DCs was determined and compared in each TLS maturation stage. K, Representative images of TLS in each maturation stage and nonorganized lymphocytic infiltrate containing DCs (white arrowheads), HEVs (magenta arrowheads), and DC-LAMP–expressing lung epithelial cells (cyan arrowheads) are shown.

Close modal

Because HEVs and mature DCs are often located within TLS and their numbers are associated with improved outcome of cancer patients (12, 39), we analyzed whether these cell types were associated with any particular TLS maturation stage by using multiparameter IF. We detected DC-LAMP+ (DCs) and PNAD+ cells (HEVs) in TLS at all maturation stages with similar frequencies (Fig. 1J) as well as in lymphocyte-rich areas without apparent organization into TLS (Fig. 1K, right).

Tumor-associated TLS develop within a specialized microanatomic niche

To identify factors involved in the development of LSCC-associated TLSs, we selected a set of patients with high and low TLS densities and investigated the expression of chemokine transcripts associated with lymphoid neogenesis in lung inflammation (3, 37). We found a significantly higher expression of CXCL13, LTB, and CCL21 in TLS-high LSCCs (Fig. 2A). Because of high intergroup variability, we did not detect a significant difference in the expression of CXCL12 between TLS-high and TLS-low patients. Nevertheless, we found a direct correlation between CXCL12 expression and TLS density by Spearman correlation analysis (R = 0.44, P = 0.019), while IL17A, CCL19, and other previously published TLS-associated factors did not correlate (Supplementary Table S5).

Figure 2.

A specialized niche supports TLS development. A, qRT-PCR analysis for transcripts of genes involved in lymphoid neogenesis was performed for chemotherapy-naïve patients with TLS-low (n = 10) and TLS-high tumors (n = 17). Relative expression (2Ct(cohort mean)−Ct(target gene)) was normalized against the normalization factor (NF) calculated using three most stable reference genes in the selected cohort. Groups were statistically compared using the two-tailed Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. B–D, Multiparameter-IF was performed in 5 LSCC patients with high TLS density for CXCL12, LTB, PD-L1, and cell-type markers CDH1 (E-cadherin), CD45, CD21, and CD23. Representative images of hyperplastic TLS-associated epithelium are shown. Slides were imaged using a laser scanning confocal microscope (Leica; B and C) or the Vectra imaging system (D). Scale bar, 50 μm. See also Supplementary Fig. S3.

Figure 2.

A specialized niche supports TLS development. A, qRT-PCR analysis for transcripts of genes involved in lymphoid neogenesis was performed for chemotherapy-naïve patients with TLS-low (n = 10) and TLS-high tumors (n = 17). Relative expression (2Ct(cohort mean)−Ct(target gene)) was normalized against the normalization factor (NF) calculated using three most stable reference genes in the selected cohort. Groups were statistically compared using the two-tailed Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. B–D, Multiparameter-IF was performed in 5 LSCC patients with high TLS density for CXCL12, LTB, PD-L1, and cell-type markers CDH1 (E-cadherin), CD45, CD21, and CD23. Representative images of hyperplastic TLS-associated epithelium are shown. Slides were imaged using a laser scanning confocal microscope (Leica; B and C) or the Vectra imaging system (D). Scale bar, 50 μm. See also Supplementary Fig. S3.

Close modal

We analyzed the expression of the relevant chemokines in LSCC tissues and found that CCL21 was expressed by HEVs within TLS, and CXCL13 by perivascular and TLS-associated stromal cells as well as by 75% of tumors (Supplementary Fig. S3A). CXCL12, which is produced by alveolar and bronchial epithelium in steady state, was highly expressed by 90% of the tumors and by focal clusters of hyperplastic epithelial cells nearby TLS (Supplementary Fig. S3A). Cells in such clusters also expressed variable amounts of PD-L1 and LTB (Fig. 2B–D; Supplementary Fig. S3B). We observed such clusters in all patients with TLS in the alveolar compartment and independently of the TLS maturation stage (Fig. 2D). Taken together, lung parenchyma provides a favorable environment for TLS development, which involves CXCL13 production by perivascular and stromal cells, CCL21 production by HEVs as well as LTB, PD-L1 and CXCL12 production by hyperplastic alveolar epithelial cells.

The density of LSCC-associated TLS predicts survival and is associated with an adaptive immune response–related signature

To test the prognostic significance of TLS density in our cohort, we defined a threshold for separating patients with high and low TLS densities by ROC curve analysis (Fig. 3A). In Kaplan–Meier analysis high TLS density (>0.165 TLS/mm2) significantly correlated with improved progression-free (PFS), disease-free and overall survival (Fig. 3B). Among the significant covariates (Table 1) TLS density and tumor stage were the only two independent predictors of PFS in a multivariate Cox regression analysis (Supplementary Table S6). We confirmed the positive correlation between high TLS density and improved survival in another, geographically distinct NSCLC patient cohort (Supplementary Table S1) by scoring TLS density in a single H&E image per patient (Supplementary Fig. S4A).

Figure 3.

TLS density correlates with improved survival and adaptive immune response. A and B, TLS were quantified in H&E sections as described in Fig. 1. A, ROC curve analysis was performed on chemotherapy-naïve patients with complete 3 year follow-up data (n = 79) and identified 0.165 TLS/mm2 as a threshold for separating TLS-high and TLS-low tumors with prognostic relevance. AUC, area under the curve. B, Survival was compared for patients with high (>0.165 TLS/mm2) and low TLS density by Kaplan–Meier curves. Confidence intervals are indicated as shaded areas surrounding survival curves. Significance was calculated by log-rank test. Patient numbers per group are indicated in brackets and numbers at risk are displayed for each 1,000 days of follow-up. C, qRT-PCR analysis for immune response-related transcripts was performed as described in Fig. 2. Groups were compared using the two-tailed Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. See also Supplementary Fig. S4.

Figure 3.

TLS density correlates with improved survival and adaptive immune response. A and B, TLS were quantified in H&E sections as described in Fig. 1. A, ROC curve analysis was performed on chemotherapy-naïve patients with complete 3 year follow-up data (n = 79) and identified 0.165 TLS/mm2 as a threshold for separating TLS-high and TLS-low tumors with prognostic relevance. AUC, area under the curve. B, Survival was compared for patients with high (>0.165 TLS/mm2) and low TLS density by Kaplan–Meier curves. Confidence intervals are indicated as shaded areas surrounding survival curves. Significance was calculated by log-rank test. Patient numbers per group are indicated in brackets and numbers at risk are displayed for each 1,000 days of follow-up. C, qRT-PCR analysis for immune response-related transcripts was performed as described in Fig. 2. Groups were compared using the two-tailed Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple testing. See also Supplementary Fig. S4.

Close modal

To understand the possible reasons of the prognostic benefit, we investigated whether TLS are associated with the immunologic makeup of a tumor. We compared the expression of various immune response–related genes (Supplementary Table S5) in a set of TLS-high and TLS-low patients. The expression of transcripts related to adaptive immunity including CD27, CD8A, IL21, IGKC, and others (Fig. 3C) was significantly upregulated in TLS-high tumors or directly correlated with TLS density (Supplementary Fig. S4B), whereas the expression of innate response-related genes, such as NKP46 and ITGAM (encoding CD11b), was similar in both patient groups (Supplementary Table S5). The expression of most TLS-associated genes was prognostic for improved PFS in an independent LSCC cohort (n = 141) available from the online repository, Kaplan–Meier plotter (Supplementary Fig. S4C; ref. 40). Taken together, high TLS density is an independent positive prognostic marker for LSCC patients and suggests an increased intratumoral adaptive immune response.

Neoadjuvant chemotherapy impairs TLS maturation and abrogates the prognostic power of TLS

Histology analysis revealed a strong positive correlation between TLS density and the presence of GC (Fig. 4A), and a negative correlation between GC and the use of neoadjuvant chemotherapy (Supplementary Table S7). A significantly lower proportion of neoadjuvant chemotherapy-treated patients showed GC+ TLS (Fig. 4B) compared with chemotherapy-naïve patients, while TLS density was similar (Fig. 4C). We performed an in-depth analysis of TLS maturation in different patients groups using multiparameter IF and quantitative pathology (Fig. 4D). We determined the number of TLS in each maturation stage and compared the proportion of each stage in patients with high and low TLS densities and in patients after neoadjuvant chemotherapy. TLS maturation significantly correlated with TLS density in chemotherapy-naïve patients (Spearman R = 0.32; P = 0.01), but not in chemotherapy-treated patients (Spearman R = −0.09; P = 0.6) Specifically, TLS maturation was arrested at the E-TLS stage in TLS-low patients (Fig. 4E, left), resulting in a significantly reduced SFL-TLS proportion (Fig. 4E, right) and decreased GC size (Fig. 4F) when compared with TLS-high patients.

Figure 4.

The prognostic value of TLS density is abrogated after neoadjuvant chemotherapy due to impaired GC formation. A, TLS density was analyzed in H&E sections as described in Fig. 1B. A tumor was considered GC-positive if at least one TLS showed the characteristic morphology of proliferating centroblasts. TLS density was compared in patients with apparent GC (n = 85, GC+) versus no GC (n = 53, GC-) by two-tailed Wilcoxon rank-sum test. B, The proportion of patients with GC+ TLS in TLS-high patients with (n = 37) or without (n = 55) neoadjuvant chemotherapy (−/+ NeoCh), and TLS-low patients with (n = 14) and without (n = 34) neoadjuvant chemotherapy. χ2 test with Yates' correction was used to compare the proportion of patients with GC+ TLS in +NeoCh and −NeoCh groups (P = 0.012). C, TLS density was compared in chemotherapy-naïve patients (−NeoCh, n = 87) and -treated patients (+NeoCh, n = 51) by two-tailed Wilcoxon rank-sum test. D, Quantitative pathology approach: IF was performed for B cells (CD20), T cells (CD3) and epithelial cells (panCK; top row) or FDCs (CD21), GCs (CD23), and CXCL13 (middle row). To analyze TLS maturation stages and GC size, tissue segmentation algorithm was trained to recognize dense DAPI cell clusters as lymphocytic aggregates, CD21+ areas as FDC, and CD21+CD23+ areas as GC (bottom row). The number of TLS in each maturation stage was determined and expressed as the proportion of all TLS for each patient. GC size was measured as the average number of pixels per GC area in each patient. E and F, TLS maturation and GC size compared with chemotherapy-naïve patients with high (n = 37) or low TLS density (n = 24) by two-tailed Wilcoxon rank-sum test. G and H, TLS maturation and GC size were compared with TLS-high chemotherapy-naïve (-, n = 37), neoadjuvant chemotherapy-treated (+Ch, n = 28), or radiotherapy-treated (+RT, n = 4) patients by two-tailed Wilcoxon rank-sum test. The same significant differences were detected by the Kruskal–Wallis test with Dunn multiple comparison test comparing naïve versus +Ch and naïve versus +RT groups. I, Patients were stratified by neoadjuvant chemotherapy treatment and PFS was compared in patients with high and low TLS density as described in Fig. 3. Patient numbers per group are indicated in brackets and numbers at risk are displayed for each 1,000 days of follow-up. n.s., nonsignificant.

Figure 4.

The prognostic value of TLS density is abrogated after neoadjuvant chemotherapy due to impaired GC formation. A, TLS density was analyzed in H&E sections as described in Fig. 1B. A tumor was considered GC-positive if at least one TLS showed the characteristic morphology of proliferating centroblasts. TLS density was compared in patients with apparent GC (n = 85, GC+) versus no GC (n = 53, GC-) by two-tailed Wilcoxon rank-sum test. B, The proportion of patients with GC+ TLS in TLS-high patients with (n = 37) or without (n = 55) neoadjuvant chemotherapy (−/+ NeoCh), and TLS-low patients with (n = 14) and without (n = 34) neoadjuvant chemotherapy. χ2 test with Yates' correction was used to compare the proportion of patients with GC+ TLS in +NeoCh and −NeoCh groups (P = 0.012). C, TLS density was compared in chemotherapy-naïve patients (−NeoCh, n = 87) and -treated patients (+NeoCh, n = 51) by two-tailed Wilcoxon rank-sum test. D, Quantitative pathology approach: IF was performed for B cells (CD20), T cells (CD3) and epithelial cells (panCK; top row) or FDCs (CD21), GCs (CD23), and CXCL13 (middle row). To analyze TLS maturation stages and GC size, tissue segmentation algorithm was trained to recognize dense DAPI cell clusters as lymphocytic aggregates, CD21+ areas as FDC, and CD21+CD23+ areas as GC (bottom row). The number of TLS in each maturation stage was determined and expressed as the proportion of all TLS for each patient. GC size was measured as the average number of pixels per GC area in each patient. E and F, TLS maturation and GC size compared with chemotherapy-naïve patients with high (n = 37) or low TLS density (n = 24) by two-tailed Wilcoxon rank-sum test. G and H, TLS maturation and GC size were compared with TLS-high chemotherapy-naïve (-, n = 37), neoadjuvant chemotherapy-treated (+Ch, n = 28), or radiotherapy-treated (+RT, n = 4) patients by two-tailed Wilcoxon rank-sum test. The same significant differences were detected by the Kruskal–Wallis test with Dunn multiple comparison test comparing naïve versus +Ch and naïve versus +RT groups. I, Patients were stratified by neoadjuvant chemotherapy treatment and PFS was compared in patients with high and low TLS density as described in Fig. 3. Patient numbers per group are indicated in brackets and numbers at risk are displayed for each 1,000 days of follow-up. n.s., nonsignificant.

Close modal

To determine the impact of neoadjuvant therapy on TLS maturation, we analyzed only TLS-high patients and observed that the proportion of SFL-TLS (Fig. 4G) as well as GC size (Fig. 4H) were significantly reduced in patients after neoadjuvant radio- or chemotherapy. Because neoadjuvant chemotherapy is used for patients with advanced disease, we separately compared TLS maturation in chemotherapy-naïve patients with early- and late-stage disease. We observed that only early-stage patients had a significantly higher proportion of SFL-TLS than chemotherapy-treated patients, while GC size was significantly higher in both early- and late-stage chemotherapy-naïve patients (Supplementary Fig. S5A and S5B).

Because GC results from cognate interactions between T and B cells, their presence in TLS may point toward an ongoing immune response, which may be tumor-specific. We thus hypothesized that the reduction of GC in neoadjuvant chemotherapy-treated patients indicates dysfunctional TLS and abrogates the prognostic relevance of TLS density. Indeed, in neoadjuvant chemotherapy-treated patients TLS density had no prognostic power (Fig. 4I). In contrast, TLS density was an independent prognostic marker, stronger than tumor stage in chemotherapy-naïve patients (Table 2). To verify the prognostic relevance of TLS maturation in the absence of potential negative impact of chemotherapy we determined the association between the number of each TLS maturation stage determined in the IF images and the PFS in chemotherapy-naïve patients. Only the number or the proportion of SFL-TLS significantly correlated with improved survival (Supplementary Fig. S5C–S5E). Taken together, these data show that neoadjuvant therapy impairs TLS maturation and reveal the essential role of GC for the prognostic power of TLS density in patients with LSCC.

Table 2.

Multivariate Cox regression analysis of PFS of neoadjuvant chemotherapy-naïve patients (n = 85)

Variables in the equationPHR (95% CI)
Model 1a 
 Stageb 0.007 1.38 (1.1–1.73) 
 Adjuvant therapyc 0.95 1 (0.61–1.69) 
 TLS densityc 0.0003 0.28 (0.14–0.56) 
 TIL scoreb 0.5 0.88 (0.61–1.28) 
 Plasma cell scoreb 0.84 0.97 (0.75–1.27) 
Model 2a 
 Stagec 0.006 2.7 (1.33–5.43) 
 Adjuvant therapyc 0.94 1 (0.61–1.72) 
 TLS densityc 0.0002 0.26 (0.13–0.52) 
 TIL scoreb 0.35 0.84 (0.59–1.21) 
 Plasma cell scoreb 0.96 1 (0.77–1.32) 
Model 3a 
 pT/ypTb 0.011 1.32 (1.1–1.63) 
 pNb 0.59 1.13 (0.72–1.8) 
 pMb 0.11 4.2 (0.73–24) 
 Adjuvant therapyc 0.66 1.13 (0.65–1.95) 
 TLS densityc 0.002 0.34 (0.17–0.67) 
 TIL scoreb 0.29 0.82 (0.56–1.19) 
 Plasma cell scoreb 0.87 0.98 (0.74–1.29) 
Variables in the equationPHR (95% CI)
Model 1a 
 Stageb 0.007 1.38 (1.1–1.73) 
 Adjuvant therapyc 0.95 1 (0.61–1.69) 
 TLS densityc 0.0003 0.28 (0.14–0.56) 
 TIL scoreb 0.5 0.88 (0.61–1.28) 
 Plasma cell scoreb 0.84 0.97 (0.75–1.27) 
Model 2a 
 Stagec 0.006 2.7 (1.33–5.43) 
 Adjuvant therapyc 0.94 1 (0.61–1.72) 
 TLS densityc 0.0002 0.26 (0.13–0.52) 
 TIL scoreb 0.35 0.84 (0.59–1.21) 
 Plasma cell scoreb 0.96 1 (0.77–1.32) 
Model 3a 
 pT/ypTb 0.011 1.32 (1.1–1.63) 
 pNb 0.59 1.13 (0.72–1.8) 
 pMb 0.11 4.2 (0.73–24) 
 Adjuvant therapyc 0.66 1.13 (0.65–1.95) 
 TLS densityc 0.002 0.34 (0.17–0.67) 
 TIL scoreb 0.29 0.82 (0.56–1.19) 
 Plasma cell scoreb 0.87 0.98 (0.74–1.29) 

aStage was analyzed either as a continuous variable (lower to higher, Model 1) or as a categorical variable [early stage (0–2a) vs. late stage (3a–4; Model 2)] in separate models. Because stage is dependent on TNM parameters, we analyzed these in a separate model (Model 3).

bContinuous variables (smaller to larger).

cCategorical variables (no vs. yes or low vs. high)

Corticosteroids impair TLS development in the lungs

The cytotoxic effects of chemo- and radiotherapy may have a negative impact on the rapidly proliferating GC B cells, thus leading to GC shrinkage; however, all patients receiving neoadjuvant therapy were also concomitantly treated with corticosteroids to manage side effects. Thus, we hypothesized that the immunosuppressive properties of corticosteroids negatively influence TLS development. To investigate this possibility we compared TLS density and maturation in neoadjuvant therapy-naïve LSCC patients, who were treated with corticosteroids before resection because of noncancer-related comorbidities (Supplementary Table S4) or who had not received corticosteroids. Tumors of corticosteroid-treated patients showed a significantly lower TLS density, reduced proportion of SFL-TLS as well as GC size (Fig. 5A–C), independently of whether steroids were given locally or systemically.

Figure 5.

TLS maturation is hampered by corticosteroids. A, TLS density was analyzed in H&E-stained sections as described in Fig. 1. TLS density was compared in chemotherapy-naïve patients with no history of steroid treatment (n = 63) and patients who received steroids until at least one month before surgery (n = 18) by two-tailed Wilcoxon rank-sum test. B and C, TLS maturation and GC size were compared in a selection of chemotherapy-naïve patients as described in Fig. 4. Steroid-treated patients (n = 15) were compared to untreated patients (n = 43) by two-tailed Wilcoxon rank-sum test. D, Experimental design for TLS induction by intranasal administration of alum (200 μg/mouse) and ovalbumin (Ova, 50 μg/mouse; adapted from ref. 33) and dexamethasone treatment (0.3 mg/kg). E, Representative images of mouse lungs stained by IF to visualize different TLS maturation stages. Slides were imaged by Vectra 3.0. Scale bars, 100 μm. F, TLS density was determined as the number of dense B-cell aggregates per lung lobe and compared in mice after TLS induction and dexamethasone treatment by one-way ANOVA and Dunnett multiple comparison test. G, TLS maturation stages were analyzed as described in Fig. 4, with CD21/35+ areas recognized as FDCs and CD21/35+PNA+ areas recognized as GCs; PNA also stained alveolar epithelium. H, Spleens were collected at the time of sacrifice (Fig. 5D) and were analyzed by flow cytometry. The following gates were applied prior to gating on the target populations: singlets/live cells/CD45+. The proportions of B cells (B220+), T cells (TCRβ+), and myeloid cells (CD11b+) from live cells is indicated. F–H, Different treatment groups were compared by one-way ANOVA and Dunnett multiple comparison test.

Figure 5.

TLS maturation is hampered by corticosteroids. A, TLS density was analyzed in H&E-stained sections as described in Fig. 1. TLS density was compared in chemotherapy-naïve patients with no history of steroid treatment (n = 63) and patients who received steroids until at least one month before surgery (n = 18) by two-tailed Wilcoxon rank-sum test. B and C, TLS maturation and GC size were compared in a selection of chemotherapy-naïve patients as described in Fig. 4. Steroid-treated patients (n = 15) were compared to untreated patients (n = 43) by two-tailed Wilcoxon rank-sum test. D, Experimental design for TLS induction by intranasal administration of alum (200 μg/mouse) and ovalbumin (Ova, 50 μg/mouse; adapted from ref. 33) and dexamethasone treatment (0.3 mg/kg). E, Representative images of mouse lungs stained by IF to visualize different TLS maturation stages. Slides were imaged by Vectra 3.0. Scale bars, 100 μm. F, TLS density was determined as the number of dense B-cell aggregates per lung lobe and compared in mice after TLS induction and dexamethasone treatment by one-way ANOVA and Dunnett multiple comparison test. G, TLS maturation stages were analyzed as described in Fig. 4, with CD21/35+ areas recognized as FDCs and CD21/35+PNA+ areas recognized as GCs; PNA also stained alveolar epithelium. H, Spleens were collected at the time of sacrifice (Fig. 5D) and were analyzed by flow cytometry. The following gates were applied prior to gating on the target populations: singlets/live cells/CD45+. The proportions of B cells (B220+), T cells (TCRβ+), and myeloid cells (CD11b+) from live cells is indicated. F–H, Different treatment groups were compared by one-way ANOVA and Dunnett multiple comparison test.

Close modal

To provide direct experimental evidence for our observation that corticosteroids interfere with lung TLS development and/or maturation, we induced TLS in the lungs of mice by intranasal (i.n.) administration of alum with or without Ova as an antigen (adapted from ref. 33) and investigated the impact of systemic low-dose dexamethasone treatment (Fig. 5D). We characterized TLS maturation by multiparameter IF using B220, CD21/35 and PNA to stain B cells, FDC and GC, respectively. We observed that TLS development in response to intranasal alum followed the same maturation stages as in LSCC patients (Fig. 5E). However, the addition of Ova was necessary to induce high TLS density and GC formation (Fig. 5F and G). Systemic dexamethasone treatment did not significantly affect B cell, T cell, and myeloid cell proportions in the spleen (Fig. 5H) and did not affect the density of TLS in the lungs (Fig. 5G), but significantly reduced the proportion of PFL- and SFL-TLS (Fig. 5H). Collectively, our data suggest that corticosteroids have a negative impact on TLS development in the lungs independently of cytotoxic therapies.

We performed a comprehensive analysis of LSCC tissues to study TLSs as whole microanatomic structures. Healthy human lungs are devoid of TLS (34), while most LSCC patients had peritumoral TLS in close vicinity to hyperplastic alveolar epithelial cells, which express CXCL12 and LTB. A fraction of these cells expressed considerable amounts of PD-L1 rather than chemokines. Cells with similar morphology are present next to TLS in infected human lungs (41) and in advanced-stage COPD patients where they generate an alveolar–lymphoid interface containing antigen-capturing DC (42). We propose that these cells support TLS development as we observed them close to TLS in all patients and at all TLS maturation stages. Our observation that the pleural wall, which lacks such cells, contained significantly less TLS than the alveolar compartment (Fig. 1D) supports this idea. However, direct experimental evidence is needed to verify this assumption. The morphology of this cell cluster resembles the dome-shaped lymphoepithelium (M cells) of the gut-associated TLS, which has a unique ability to transport antigens from the gut lumen to the underlying DC (43). M cell–like cells are present also in the respiratory tract of various animal species and deliver material from the airway lumen to DCs in the underlying lymphoid follicles (44, 45). Whether the hyperplastic epithelial cells surrounding TLS in human lungs likewise participate in antigen delivery to TLS is currently unknown.

We showed that expression of CXCL13, LTB, CCL21, and CXCL12 correlates with TLS density in LSCC. In contrast, IL17A and other chemokines relevant for iBALT formation in different lung inflammatory conditions (34) do not seem to be involved in the generation of LSCC-associated TLS. In the lungs, B cells are the main source of CXCL13 in response to cigarette smoke or LPS and are plausible drivers of iBALT development in COPD patients via Toll-like receptor and LTBR signaling (29). In LSCC, however, CXCL13 was mainly produced by TLS-associated perivascular and stromal cells and may serve as entry sites for lymphocytes from the bloodstream. We thus propose that B-cell and T-cell clusters without differentiated FDCs near CXCL13+ blood vessels represent the initial stage of lung TLS development. Various stimuli including IL1 (46, 47) and cigarette smoke (48) induce the production of CXCL13 in lung mesenchymal cells. However, the mechanisms driving the expression of CXCL13 by perivascular cells or the formation of the hyperplastic TLS-associated epithelium in LSCC are currently unknown.

In mice, naïve B cells respond to CXCL13 by producing LTB (49), which is crucial for the differentiation of FDCs from perivascular mesenchymal precursors and TLS development (36). We observed that hyperplastic epithelial cells and lymphocytes within early TLS produce LTB (Fig. 2C), which might drive the differentiation of FDC and generate the PFL-TLS in human lungs. The subsequent establishment of SFL-TLSs presumably requires antigen-dependent interactions between DCs, T cells, and B cells. CCL21 is mainly expressed by HEVs and lymphatic vessels, and attracts DCs and T cells to the developing TLS in different tissues (7). Our data suggest an upstream function of DCs and HEVs in the development of TLS because they are already present in E-TLS as well as in lymphocyte-rich areas without apparent organization. However, it is currently unclear whether such areas are the initiation sites of TLS or how these cell types dictate the generation of a mature TLS. We observed that TLS development follows the same steps in colorectal cancer (50) and in the lungs of mice after intranasal administration of alum + Ova. Thus, the described TLS maturation sequence may represent a general mechanism of TLS development. A clear understanding of TLS dynamics in the tumor microenvironment is currently lacking. It has been shown in the context of viral lung infection that TLS persist in lungs for at least 4 weeks after the clearance of the virus (4); however, it is unclear whether each single TLS was maintained for at least 4 weeks or whether TLS developed and disappeared at a certain rate during these 4 weeks. Tumor models with synchronized TLS development could be used to address this question; targeting of LIGHT to tumor vasculature (24) may represent such a model, although it is currently unknown whether GC develop in this setting.

We identified TLS density as the most significant and independent prognostic marker in untreated LSCC patients, which even outperformed tumor stage. Because TLS often correlates with increased lymphocytic infiltrate that has a pronounced prognostic benefit on its own (8), the causal relationship between TLS and lymphocytic infiltrate is unclear. Here we demonstrate that TLS and not the diffuse infiltration of lymphocytes or plasma cells is an independent prognostic factor for PFS (Table 2; Supplementary Table S6), suggesting that at least in LSCC, the structural organization of lymphocytes is the confounding factor for prognosis.

TLS density can be assessed in diagnostic H&E sections and can thus be easily introduced in routine pathology to serve as a relevant prognostic parameter. Neoadjuvant chemotherapy-treated patients, however, showed similar TLS density but significantly less and smaller GC when compared with untreated patients. Because of the lack of paired pre- and post-therapy samples, we cannot exclude that neoadjuvant chemotherapy-treated patients had a reduced TLS maturation already before the therapy. Nevertheless, we did observe a significant reduction in GC size when compared with stage-matched chemotherapy-naïve patients (Supplementary Fig. S5B). In mice, de novo formation of B-cell follicles and the growth of GC size is crucial for an effective immune response (51). We thus propose that the low frequency and reduced size of GC in tumor-associated TLS after neoadjuvant chemotherapy reflects impaired TLS function. This is in line with the observation that the prognostic power of TLS density was lost in these patients. Furthermore, we demonstrate that only SFL-TLS, but not E-TLS or PFL-TLS correlated with improved survival in chemotherapy-naïve patients. The notion that the presence of GC reflects the function of TLS in antitumor immunity is also supported by Posch and colleagues showing that the integration of TLS maturation and TLS density into a joint TLS immunoscore had a superior predictive power for the recurrence risk in untreated nonmetastatic colorectal cancer than TLS density alone (50). However, direct evidence for the presence and activation of tumor-specific B and/or T cells in cancer-associated TLS is still lacking.

We propose that corticosteroids are at least partly responsible for the impaired TLS development in the neoadjuvant chemotherapy-treated LSCC patients for the following reasons. First, we observed a significantly lower TLS density and reduced GC formation in LSCC patients who were treated with corticosteroids but did not receive neoadjuvant chemotherapy. Second, we observed that dexamethasone significantly decreased the development of alum + Ova-induced mature TLS in the lungs of mice. Third, corticosteroid-treated COPD patients have a significantly reduced TLS density in comparison with COPD patients who did not receive corticosteroids (52). Finally, corticosteroid treatment abolishes the formation of GC in BALT in experimental rat pulmonary granuloma (53). A possible explanation is offered by a study demonstrating that prednisolone induces apoptosis of human GC B cells by blocking both surface immunoglobulin- and CD40-mediated survival pathways (54). Despite the concomitant use of corticosteroids, the TLS density in chemotherapy-treated patients was not diminished. We propose that chemotherapy-induced inflammation in the tumor microenvironment (55) overrides the negative impact of steroids on TLS initiation (reflected by total TLS density), but is not sufficient to override the negative impact on TLS maturation (reflected by proportion of SFL-TLS). Similarly, in the lungs of mice receiving proinflammatory alum intranasally, steroid treatment did not decrease TLS density (Fig. 5F), whereas it reduced TLS maturation (Fig. 5G). Our findings indicate that corticosteroid treatment of cancer patients may not only decrease their spontaneous immune defense to cancer but also suppress the response to immunotherapy as well as to treatments that induce immunogenic cell death such as radiotherapy (56) and chemotherapy (57). Thus, the use of alternative nonsteroid drugs for managing adverse effects of chemo-, radio-, or immune therapy should be considered.

In summary, the work presented here extends the current knowledge by (1) characterizing the niche that fosters TLS development in the lungs, (2) determining stages of TLS maturation that affect the prognostic relevance of TLS density, and (3) discovering the negative impact of corticosteroids on TLS development in LSCC that is associated with poor outcome in chemotherapy-naïve patients.

No potential conflicts of interest were disclosed.

Conception and design: K. Siliņa, A. Soltermann, H. Moch, A. Linē, M. van den Broek

Development of methodology: K. Silina, A. Soltermann, A.C. Fontecedro, A. Linē

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Siliņa, A. Soltermann, F.M. Attar, R. Casanova, Z.M. Uckeley, H. Thut, S. Isajevs, A.C. Fontecedro, P. Foukas, M.P. Levesque, H. Moch, A. Linē

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Siliņa, P.F. Cheng, M. van den Broek

Writing, review, and/or revision of the manuscript: K. Siliņa, A. Soltermann, M.P. Levesque, H. Moch, A. Linē, M. van den Broek

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Soltermann, R. Casanova, H. Moch, M. van den Broek

Study supervision: A. Soltermann, H. Moch, M. van den Broek

Other (collection and assembly of data): M. Wandres

We thank Stefanie Hiltbrunner and Paulino Tallon de Lara (Institute of Experimental Immunology, University of Zurich, Zurich, Switzerland) for critical reading of the manuscript and helpful discussions, Mathias Heikenwaelder (Helmholtz Zentrum Munich, Munich, Germany) for providing the anti-LTB antibody, and the personnel of the Laboratory Animal Services Center (University of Zurich) for expert animal care. This work was financially supported by the Swiss National Science Foundation (SNSF, 31003A_152851, MvdB), the Sciex Foundation (13.046 to M. van den Broek, K. Siliņa, A. Linē), the Science Foundation for Oncology (to M. van den Broek), the Cancer League Zurich (KLZ_2015_vandenBroek to M. van den Broek, K. Siliņa), the Novartis Research Foundation (15B096 to K. Siliņa), the University Research Priority Program “Translational Cancer Research” (to M. van den Broek and K. Siliņa), Swiss Cancer League (to A. Soltermann), SNSF SystemsX (to A. Soltermann), and the Latvian National Research Program “Biomedicine” (to A. Linē).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Neyt
K
,
Perros
F
,
GeurtsvanKessel
CH
,
Hammad
H
,
Lambrecht
BN
. 
Tertiary lymphoid organs in infection and autoimmunity
.
Trends Immunol
2012
;
33
:
297
305
.
2.
Dieu-Nosjean
MC
,
Giraldo
NA
,
Kaplon
H
,
Germain
C
,
Fridman
WH
,
Sautes-Fridman
C
. 
Tertiary lymphoid structures, drivers of the anti-tumor responses in human cancers
.
Immunol Rev
2016
;
271
:
260
75
.
3.
Silina
K
,
Rulle
U
,
Kalnina
Z
,
Line
A
. 
Manipulation of tumour-infiltrating B cells and tertiary lymphoid structures: a novel anti-cancer treatment avenue?
Cancer Immunol Immunother
2014
;
63
:
643
62
.
4.
Halle
S
,
Dujardin
HC
,
Bakocevic
N
,
Fleige
H
,
Danzer
H
,
Willenzon
S
, et al
Induced bronchus-associated lymphoid tissue serves as a general priming site for T cells and is maintained by dendritic cells
.
J Exp Med
2009
;
206
:
2593
601
.
5.
Moyron-Quiroz
JE
,
Rangel-Moreno
J
,
Kusser
K
,
Hartson
L
,
Sprague
F
,
Goodrich
S
, et al
Role of inducible bronchus associated lymphoid tissue (iBALT) in respiratory immunity
.
Nat Med
2004
;
10
:
927
34
.
6.
Wada
Y
,
Nakashima
O
,
Kutami
R
,
Yamamoto
O
,
Kojiro
M
. 
Clinicopathological study on hepatocellular carcinoma with lymphocytic infiltration
.
Hepatology
1998
;
27
:
407
14
.
7.
Hiraoka
N
,
Ino
Y
,
Yamazaki-Itoh
R
. 
Tertiary lymphoid organs in cancer tissues
.
Front Immunol
2016
;
7
:
244
.
8.
Fridman
WH
,
Zitvogel
L
,
Sautes-Fridman
C
,
Kroemer
G
. 
The immune contexture in cancer prognosis and treatment
.
Nat Rev Clin Oncol
2017
;
12
:
717
34
.
9.
Garnelo
M
,
Tan
A
,
Her
Z
,
Yeong
J
,
Lim
CJ
,
Chen
J
, et al
Interaction between tumour-infiltrating B cells and T cells controls the progression of hepatocellular carcinoma
.
Gut
2015
;
66
:
342
51
.
10.
Knief
J
,
Reddemann
K
,
Petrova
E
,
Herhahn
T
,
Wellner
U
,
Thorns
C
. 
High density of tumor-infiltrating B-Lymphocytes and plasma cells signifies prolonged overall survival in adenocarcinoma of the esophagogastric junction
.
Anticancer Res
2016
;
36
:
5339
45
.
11.
Dieu-Nosjean
MC
,
Antoine
M
,
Danel
C
,
Heudes
D
,
Wislez
M
,
Poulot
V
, et al
Long-term survival for patients with non-small-cell lung cancer with intratumoral lymphoid structures
.
J Clin Oncol
2008
;
26
:
4410
7
.
12.
Goc
J
,
Germain
C
,
Vo-Bourgais
TK
,
Lupo
A
,
Klein
C
,
Knockaert
S
, et al
Dendritic cells in tumor-associated tertiary lymphoid structures signal a Th1 cytotoxic immune contexture and license the positive prognostic value of infiltrating CD8+ T cells
.
Cancer Res
2014
;
74
:
705
15
.
13.
Germain
C
,
Gnjatic
S
,
Tamzalit
F
,
Knockaert
S
,
Remark
R
,
Goc
J
, et al
Presence of B cells in tertiary lymphoid structures is associated with a protective immunity in patients with lung cancer
.
Am J Respir Crit Care Med
2014
;
189
:
832
44
.
14.
de Chaisemartin
L
,
Goc
J
,
Damotte
D
,
Validire
P
,
Magdeleinat
P
,
Alifano
M
, et al
Characterization of chemokines and adhesion molecules associated with T cell presence in tertiary lymphoid structures in human lung cancer
.
Cancer Res
2011
;
71
:
6391
9
.
15.
Kroeger
DR
,
Milne
K
,
Nelson
BH
. 
Tumor-Infiltrating plasma cells are associated with tertiary lymphoid structures, cytolytic T-cell responses, and superior prognosis in ovarian cancer
.
Clin Cancer Res
2016
;
22
:
3005
15
.
16.
Goc
J
,
Germain
C
,
Vo-Bourgais
TK
,
Lupo
A
,
Klein
C
,
Knockaert
S
, et al
Dendritic cells in tumor-associated tertiary lymphoid structures license the positive prognostic value of tumor-infiltrating CD8+ T cells
.
Cancer Res
2014
;
74
:
705
15
.
17.
Nzula
S
,
Going
JJ
,
Stott
DI
. 
Antigen-driven clonal proliferation, somatic hypermutation, and selection of B lymphocytes infiltrating human ductal breast carcinomas
.
Cancer Res
2003
;
63
:
3275
80
.
18.
Coronella
JA
,
Spier
C
,
Welch
M
,
Trevor
KT
,
Stopeck
AT
,
Villar
H
, et al
Antigen-driven oligoclonal expansion of tumor-infiltrating B cells in infiltrating ductal carcinoma of the breast
.
J Immunol
2002
;
169
:
1829
36
.
19.
Zhu
W
,
Germain
C
,
Liu
Z
,
Sebastian
Y
,
Devi
P
,
Knockaert
S
, et al
A high density of tertiary lymphoid structure B cells in lung tumors is associated with increased CD4+ T cell receptor repertoire clonality
.
Oncoimmunology
2015
;
4
:
e1051922
.
20.
van de Pavert
SA
,
Mebius
RE
. 
New insights into the development of lymphoid tissues
.
Nat Rev Immunol
2010
;
10
:
664
74
.
21.
Luther
SA
,
Lopez
T
,
Bai
W
,
Hanahan
D
,
Cyster
JG
. 
BLC expression in pancreatic islets causes B cell recruitment and lymphotoxin-dependent lymphoid neogenesis
.
Immunity
2000
;
12
:
471
81
.
22.
Bindea
G
,
Mlecnik
B
,
Tosolini
M
,
Kirilovsky
A
,
Waldner
M
,
Obenauf
AC
, et al
Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer
.
Immunity
2013
;
39
:
782
95
.
23.
Schrama
D
,
Voigt
H
,
Eggert
AO
,
Xiang
R
,
Zhou
H
,
Schumacher
TN
, et al
Immunological tumor destruction in a murine melanoma model by targeted LTalpha independent of secondary lymphoid tissue
.
Cancer Immunol Immunother
2008
;
57
:
85
95
.
24.
Johansson-Percival
A
,
He
B
,
Li
ZJ
,
Kjellen
A
,
Russell
K
,
Li
J
, et al
De novo induction of intratumoral lymphoid structures and vessel normalization enhances immunotherapy in resistant tumors
.
Nat Immunol
2017
.
25.
Lutz
ER
,
Wu
AA
,
Bigelow
E
,
Sharma
R
,
Mo
G
,
Soares
K
, et al
Immunotherapy converts nonimmunogenic pancreatic tumors into immunogenic foci of immune regulation
.
Cancer Immunol Res
2014
;
2
:
616
31
.
26.
Fleige
H
,
Ravens
S
,
Moschovakis
GL
,
Bolter
J
,
Willenzon
S
,
Sutter
G
, et al
IL-17-induced CXCL12 recruits B cells and induces follicle formation in BALT in the absence of differentiated FDCs
.
J Exp Med
2014
;
211
:
643
51
.
27.
Luther
SA
,
Bidgol
A
,
Hargreaves
DC
,
Schmidt
A
,
Xu
Y
,
Paniyadi
J
, et al
Differing activities of homeostatic chemokines CCL19, CCL21, and CXCL12 in lymphocyte and dendritic cell recruitment and lymphoid neogenesis
.
J Immunol
2002
;
169
:
424
33
.
28.
Hasday
JD
,
Bascom
R
,
Costa
JJ
,
Fitzgerald
T
,
Dubin
W
. 
Bacterial endotoxin is an active component of cigarette smoke
.
Chest
1999
;
115
:
829
35
.
29.
Litsiou
E
,
Semitekolou
M
,
Galani
IE
,
Morianos
I
,
Tsoutsa
A
,
Kara
P
, et al
CXCL13 production in B cells via Toll-like receptor/lymphotoxin receptor signaling is involved in lymphoid neogenesis in chronic obstructive pulmonary disease
.
Am J Res Crit Care Med
2013
;
187
:
1194
202
.
30.
Roos
AB
,
Sanden
C
,
Mori
M
,
Bjermer
L
,
Stampfli
MR
,
Erjefalt
JS
. 
IL-17A is elevated in end-stage chronic obstructive pulmonary disease and contributes to cigarette smoke-induced lymphoid neogenesis
.
Am J Res Crit Care Med
2015
;
191
:
1232
41
.
31.
Nikota
JK
,
Shen
P
,
Morissette
MC
,
Fernandes
K
,
Roos
A
,
Chu
DK
, et al
Cigarette smoke primes the pulmonary environment to IL-1alpha/CXCR-2-dependent nontypeable Haemophilus influenzae-exacerbated neutrophilia in mice
.
J Immunol
2014
;
193
:
3134
45
.
32.
Morissette
MC
,
Jobse
BN
,
Thayaparan
D
,
Nikota
JK
,
Shen
P
,
Labiris
NR
, et al
Persistence of pulmonary tertiary lymphoid tissues and anti-nuclear antibodies following cessation of cigarette smoke exposure
.
Respir Res
2014
;
15
:
49
.
33.
Kuroda
E
,
Ozasa
K
,
Temizoz
B
,
Ohata
K
,
Koo
CX
,
Kanuma
T
, et al
Inhaled fine particles induce alveolar macrophage death and interleukin-1alpha release to promote inducible bronchus-associated lymphoid tissue formation
.
Immunity
2016
;
45
:
1299
310
.
34.
Hwang
JY
,
Randall
TD
,
Silva-Sanchez
A
. 
Inducible bronchus-associated lymphoid tissue: taming inflammation in the lung
.
Front Immunol
2016
;
7
:
258
.
35.
Jones
GW
,
Hill
DG
,
Jones
SA
. 
Understanding immune cells in tertiary lymphoid organ development: it is all starting to come together
.
Front Immunol
2016
;
7
:
401
.
36.
Krautler
NJ
,
Kana
V
,
Kranich
J
,
Tian
Y
,
Perera
D
,
Lemm
D
, et al
Follicular dendritic cells emerge from ubiquitous perivascular precursors
.
Cell
2012
;
150
:
194
206
.
37.
Goc
J
,
Fridman
WH
,
Sautes-Fridman
C
,
Dieu-Nosjean
MC
. 
Characteristics of tertiary lymphoid structures in primary cancers
.
Oncoimmunology
2013
;
2
:
e26836
.
38.
Bustin
SA
,
Benes
V
,
Garson
JA
,
Hellemans
J
,
Huggett
J
,
Kubista
M
, et al
The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments
.
Clin Chem
2009
;
55
:
611
22
.
39.
Martinet
L
,
Filleron
T
,
Le Guellec
S
,
Rochaix
P
,
Garrido
I
,
Girard
JP
. 
High endothelial venule blood vessels for tumor-infiltrating lymphocytes are associated with lymphotoxin beta-producing dendritic cells in human breast cancer
.
J Immunol
2013
;
191
:
2001
8
.
40.
Szasz
AM
,
Lanczky
A
,
Nagy
A
,
Forster
S
,
Hark
K
,
Green
JE
, et al
Cross-validation of survival associated biomarkers in gastric cancer using transcriptomic data of 1,065 patients
.
Oncotarget
2016
;
7
:
49322
33
.
41.
Tschernig
T
,
Pabst
R
. 
Bronchus-associated lymphoid tissue (BALT) is not present in the normal adult lung but in different diseases
.
Pathobiology
2000
;
68
:
1
8
.
42.
Mori
M
,
Andersson
CK
,
Svedberg
KA
,
Glader
P
,
Bergqvist
A
,
Shikhagaie
M
, et al
Appearance of remodelled and dendritic cell-rich alveolar-lymphoid interfaces provides a structural basis for increased alveolar antigen uptake in chronic obstructive pulmonary disease
.
Thorax
2013
;
68
:
521
31
.
43.
Randall
TD
. 
Bronchus-associated lymphoid tissue (BALT) structure and function
.
Adv Immunol
2010
;
107
:
187
241
.
44.
Wang
M
,
Gao
Z
,
Zhang
Z
,
Pan
L
,
Zhang
Y
. 
Roles of M cells in infection and mucosal vaccines
.
Hum Vaccin Immunother
2014
;
10
:
3544
51
.
45.
Pabst
R
,
Tschernig
T
. 
Bronchus-associated lymphoid tissue: an entry site for antigens for successful mucosal vaccinations?
Am J Respir Cell Mol Biol
2010
;
43
:
137
41
.
46.
Lisignoli
G
,
Cristino
S
,
Toneguzzi
S
,
Grassi
F
,
Piacentini
A
,
Cavallo
C
, et al
IL1beta and TNFalpha differently modulate CXCL13 chemokine in stromal cells and osteoblasts isolated from osteoarthritis patients: evidence of changes associated to cell maturation
.
Exp Gerontol
2004
;
39
:
659
65
.
47.
Neyt
K
,
GeurtsvanKessel
CH
,
Deswarte
K
,
Hammad
H
,
Lambrecht
BN
. 
Early IL-1 signaling promotes iBALT induction after influenza virus infection
.
Front Immunol
2016
;
7
:
312
.
48.
Bracke
KR
,
Verhamme
FM
,
Seys
LJ
,
Bantsimba-Malanda
C
,
Cunoosamy
DM
,
Herbst
R
, et al
Role of CXCL13 in cigarette smoke-induced lymphoid follicle formation and chronic obstructive pulmonary disease
.
Am J Res Crit Care Med
2013
;
188
:
343
55
.
49.
Ansel
KM
,
Ngo
VN
,
Hyman
PL
,
Luther
SA
,
Forster
R
,
Sedgwick
JD
, et al
A chemokine-driven positive feedback loop organizes lymphoid follicles
.
Nature
2000
;
406
:
309
14
.
50.
Posch
F
,
Silina
K
,
Leibl
S
,
Mündlein
A
,
Moch
H
,
Siebenhüner
A
, et al
Maturation of tertiary lymphoid structures predicts the risk for recurrence in stage II and III colorectal cancer
.
Oncoimmunology
2018
;
7
:
e1378844
.
doi.org/10.1080/2162402X.2017.1378844
.
51.
Dubey
LK
,
Lebon
L
,
Mosconi
I
,
Yang
CY
,
Scandella
E
,
Ludewig
B
, et al
Lymphotoxin-dependent B cell-FRC crosstalk promotes de novo follicle formation and antibody production following intestinal helminth infection
.
Cell Rep
2016
;
15
:
1527
41
.
52.
Hogg
JC
,
Chu
FS
,
Tan
WC
,
Sin
DD
,
Patel
SA
,
Pare
PD
, et al
Survival after lung volume reduction in chronic obstructive pulmonary disease: insights from small airway pathology
.
Am J Res Crit Care Med
2007
;
176
:
454
9
.
53.
Gemma
H
,
Sato
A
. 
[Effect of glucocorticoid on lung tissue and bronchus-associated lymphoid tissue of experimental granulomatous lung]
.
Kekkaku
1989
;
64
:
387
99
.
54.
Holder
MJ
,
Knox
K
,
Gordon
J
. 
Factors modifying survival pathways of germinal center B cells. Glucocorticoids and transforming growth factor-beta, but not cyclosporin A or anti-CD19, block surface immunoglobulin-mediated rescue from apoptosis
.
Eur J Immunol
1992
;
22
:
2725
8
.
55.
Apetoh
L
,
Ghiringhelli
F
,
Tesniere
A
,
Obeid
M
,
Ortiz
C
,
Criollo
A
, et al
Toll-like receptor 4-dependent contribution of the immune system to anticancer chemotherapy and radiotherapy
.
Nat Med
2007
;
13
:
1050
9
.
56.
Surace
L
,
Lysenko
V
,
Fontana
AO
,
Cecconi
V
,
Janssen
H
,
Bicvic
A
, et al
Complement is a central mediator of radiotherapy-induced tumor-specific immunity and clinical response
.
Immunity
2015
;
42
:
767
77
.
57.
Galluzzi
L
,
Buque
A
,
Kepp
O
,
Zitvogel
L
,
Kroemer
G
. 
Immunological effects of conventional chemotherapy and targeted anticancer agents
.
Cancer Cell
2015
;
28
:
690
714
.

Supplementary data