Identification and Validation of a Prognostic Model Based on Three TLS- Related Genes in Oral Squamous Cell Carcinoma

doi:10.21203/rs.3.rs-4358767/v1

Download PDF

Research Article

Identification and Validation of a Prognostic Model Based on Three TLS- Related Genes in Oral Squamous Cell Carcinoma

https://doi.org/10.21203/rs.3.rs-4358767/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 26 Oct, 2024

Read the published version in Cancer Cell International →

You are reading this latest preprint version

Background: The tertiary lymphoid structures (TLSs) have an immunomodulatory function and have a positive impact on the survival outcomes of patients with oral squamous cell carcinoma (OSCC). However, there is a lack of standard approaches for quantifying TLSs and prognostic models using TLS-related genes (TLSRGs). These limitations limit the widespread use of TLSs in clinical practice.

Methods: A convolutional neural network was used to automatically detect and quantify TLSs in HE-stained whole slide images. By employing bioinformatics and diverse statistical methods, this research created a prognostic model using TCGA cohorts, and explored the connection between this model and immune infiltration. The expression levels of three TLSRGs in clinical specimens were detected by immunohistochemistry.

Results: TLSs were found to be an independent predictor of both overall survival (OS) and disease-free survival in OSCC patients. A larger proportion of the TLSs area represented a better prognosis. After analysis, we identified 69 differentially expressed TLSRGs, and selected three pivotal TLSRGs to construct the risk score model. This model emerged as a standalone predictor for OS and exhibited close associations with CD4+ T cells, CD8+ T cells, and macrophages. Immunohistochemistry revealed high expression levels of CCR7 and CXCR5 in TLS+OSCC samples, while CD86 was highly expressed in TLS-OSCC samples.

Conclusions: This is the first prognostic model based on TLSRGs, that can effectively predict survival outcomes and contribute to individual treatment strategies for OSCC patients.

Oral squamous cell carcinoma

Tertiary lymphoid structure

Prognostic model

Immune infiltration

Convolutional neural network.

Oral squamous cell carcinoma (OSCC) is the most common type of head and neck squamous cell carcinoma (HNSCC), and is characterized by a notable tendency for metastasis to cervical lymph nodes at a high rate (1, 2). According to the Global Cancer Observatory (GCO), there were 389,846 diagnosed cases in 2022, including 188,438 deaths. There will be a growth in incidence and mortality until 2040, as predicted by the GCO. Currently, prognosis prediction for patients primarily involves assessing tumor dimensions, lymph nodes and distant organ metastases (3). However, the prognoses of patients with identical TNM stages are different, which shows that the TNM stage may fail to capture the immune heterogeneity of OSCC (4, 5).

The tertiary lymphoid structures (TLSs), an ectopic lymphoid organ that forms in nonlymphoid tissues (6), are correlated with favorable clinical outcomes and an improved response to immunological therapy (7, 8) across various cancers (9–11). In the absence of external fibrous capsules, TLSs are directly exposed to the tumor microenvironment and have a stronger specific immune response than other noninfiltrating lymphocytes (7, 12, 13). A recent study revealed that TLSs serve as an innovative prognostic indicator in the field of HNSCC (14). The presence of TLSs, which is frequently observed in oral tongue squamous cell carcinoma (OTSCC), is utilized for predicting immunotherapy sensitivity (15). Furthermore, the presence of TLSs independently correlates with the overall survival (OS) rate and disease-free survival (DFS) rate in patients of OSCC (16). Thus, TLSs have the potential to be a complementary indicator for predicting clinical outcomes after OSCC surgery. Although TLSs have received increasing attention in prognostic evaluation, there is still no recommended TLS-related molecular marker for building a prognostic model for the accurate stratification of OSCC patients, which limits the ability of TLSs to help clinicians make personalized clinical decisions.

In this study, we used a convolutional neural network to automatically identify and quantify TLSs in routine HE-stained whole slide images (WSIs) of OSCC (17), aiming to make the assessment of TLSs more efficient and accurate. We explored TLS-related genes (TLSRGs) related to OSCC through bioinformatics analysis. Based on the TLSRGs, we established a risk score model capable of predicting the prognosis of OSCC patients and evaluated the tumor immune microenvironment. By introducing clinical variables into the model, we constructed a nomogram model (5) aimed at providing individualized evaluations for OSCC patients.

Data Acquisition

High-throughput sequence data and corresponding clinicopathological information for 503 HNSCC (Additional file 1) and 54 adjacent normal mucosa tissues were downloaded from TCGA. According to the tumor location, the clinical information and data of 336 OSCC tissues were extracted for subsequent analysis. Patients lacking survival information were omitted from the prognostic analysis.

TLSs Quantification

Using a convolutional neural network, the presence of TLSs was identified in HE-stained WSIs of OSCC patients. This approach can be used to determine the area occupied by TLSs and their density, and establish a heatmap of lymphocytes (17), which allows us to define TLSs in tissues. TLSs were categorized into three types based on a previously published scale (18, 19): (1) lymphoid aggregates (Agg), characterized by indistinct masses of lymphocytes; (2) primary follicle-like TLSs (FL1), comprising immature TLSs composed of round lymphocytes lacking germinal centers; and (3) secondary follicle-like TLSs (FL2), representing mature TLSs as round clusters of lymphocytes with the formation of germinal centers. Patients were categorized as TLS + if they had at least one TLS and as TLS- if they had no TLSs. Among TLS + tumors, patients with an area proportion above the series median were categorized into the high-TLS group, while patients with an area proportion below the series median were classified into the low-TLS group. The TLS + tumors could also be classified as follows: (1) Agg + FL1 group: tumors comprising at least one FL1, without FL2, with or without Agg; and (2) Agg + FL1 + FL2 group: tumors containing at least one FL2, irrespective of the presence of Agg and FL1.

DEGs, GO, KEGG analyses

Both TLS + OSCC tissues and TLS- OSCC tissues were compared with normal tissues for differential analysis. Analysis of differentially expressed genes (DEGs) was performed using the R package in combination with the Wilcoxon test. Visualization of the data was accomplished through the creation of heatmaps using the pheatmap package, complemented by the generation of volcano plots using R software. Based on the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases, we performed functional enrichment analysis of TLSRGs using the clusterProfiler package. Those functional categories with an adjusted p-value < 0.05 were regarded as significantly enriched.

Construction of the risk score model and prognostic nomogram

Univariate Cox regression analysis was utilized to assess the relationship between TLSRGs and OS in the TCGA cohort. To prevent overfitting and streamline the gene selection process, stepwise Cox regression analysis was utilized to identify the optimal genes based on genes detected via the Lasso algorithm. The risk score was computed by multiplying gene expression with a linear combination of regression coefficients derived from the multivariate Cox regression analysis. Patients were stratified into a high-risk group and a low-risk group based on the median risk score. Kaplan-Meier (K-M) survival curves were generated to compare OS between the low- and high-risk groups. The precision of the risk score model was evaluated through receiver operating characteristic (ROC) curve analysis via the survival ROC package. Utilizing multivariate Cox regression analysis, we constructed a nomogram integrating risk scores with additional clinical parameters. The nomogram's predictive precision was evaluated through calibration curves, ensuring its reliability and effectiveness in forecasting outcomes.

Immune cell infiltration estimation

To investigated the relationship between the presence of TLSs and immune cell infiltration, we used the Tumor Immune Estimation Resource (TIMER) (20), which can estimate the levels of six tumor-infiltrating immune subsets. Relationships between the risk score model and six types of immune cells were analyzed with Pearson’s correlation coefficient. The same method was utilized to evaluate the correlations between the proportion of TLSs area and six immune cell types.

Tissue samples

Primary OSCC tissues and adjacent normal mucosa tissues, surgically resected at the Department of Oral and Maxillofacial Surgery of the Second Xiangya Hospital between April 2010 and July 2013, were utilized in this study. Participants who received chemotherapy, radiation, or preoperative interventions were ineligible for participation in this study.

Immunohistochemistry

The formalin-fixed paraffin-embedded specimens were sectioned into 4 µm slices for immunohistochemical staining. The sections were baked at 60°C for 2 hours and subsequently dewaxed with xylene. Different concentrations of alcohol were utilized for hydration treatment, and antigen extraction was performed by citrate solution combined with high-pressure repair. Upon cooling, endogenous peroxidase was removed from tissue sections using 3% hydrogen peroxide, followed by triple rinsing with freshly prepared PBS. To mitigate nonspecific binding, the sections were incubated for 60 minutes in goat serum. The sections were placed in a humidified box at 4°C, where they were incubated with primary antibodies (anti-CCR7, Proteintech, 25898-1-AP, 1:200; anti-CXCR5, Abcam, ab254415, 1:400; anti-CD86, Abcam, ab269587, 1:100) overnight. Following washing, the sections were treated with secondary antibodies for 60 minutes at 37°C. After washing, the sections were stained with DAB and hematoxylin.

Statistical analyses

The data from the experiments were rigorously analyzed with GraphPad Prism 8.0, R software v4.0.1, and ImageJ to ensure precision. For comparisons between two groups, we applied the Student’s t test and Wilcoxon test. K-M analysis was used to construct survival curves, and the log-rank test was used to compare survival variances across groups. Univariate and multivariate Cox regression analyses were employed to determine the OS-related prognostic factors. P < 0.05 was considered to indicate statistical significance. *, **, and *** represent p < 0.05, p < 0.01, and p < 0.001, respectively.

Clinicopathological characteristics and TLSs expression of OSCC patients

In our analysis of 336 OSCC patients from the TCGA database, we collected data on various clinicopathological characteristics: age (<or ≥ 60 years), gender, smoking, drinking, pT classification, pN classification, pTNM classification, grade, perineural invasion (PNI), and nodal extracapsular spread (NES). These features are summarized in Table 1. Then the convolutional neural network was used to determine the three types of TLSs and their respective area proportions in routine HE-stained WSIs, which were obtained from the TCGA database (Fig. 1). We identified 265 TLS + samples among the 336 OSCC samples, and the TLSs area accounted for 0.1–6.2% of the total tumor area (mean 2.8%). Among the TLS + OSCC samples, ninety cases (34.0%) were Agg-type TLSs, one hundred cases (37.7%) were FL1-type TLSs, and seventy-five cases (28.3%) were FL2-type TLSs (Additional file 2). Our research investigated the correlations between TLSs expression and clinicopathological parameters in 336 OSCC patients. The results revealed that the occurrence of TLSs was negatively correlated with the pT classification (p = 0.003), pN classification (p = 0.003), and pTNM classification (p = 0.032). In addition to the PNI (p = 0.013), TLS + OSCC was also negatively correlated with the NES (p = 0.022, Table 1).

Table 1

Associations between clinicopathological characteristics and TLSs expression in 336 OSCC patients in the TCGA cohort
variable(No.)	Negative (173)	Positive (163)	p
Age (Years)			0.442
< 60	71(41%)	74(45.4%)
≥ 60	102 (59.0%)	89 (54.6%)
Gender			0.707
Male	131 (75.7%)	120 (73.6%)
Female	42 (24.3%)	43 (26.4%)
Smoking			1
Yes	80 (46.2%)	76 (46.6%)
No	93 (53.8%)	87 (53.4%)
Drinking			0.639
Yes	117 (67.6%)	108 (66.3%)
No	56 (32.4%)	55 (33.7%)
pT classification			0.003
T1 ~ T2	51 (29.5%)	74 (45.4%)
T3 ~ T4	122 (70.5%)	89 (54.6%)
pN classification			0.003
N0	55 (31.7%)	79 (48.5%)
N1 ~ N3	118 (68.2%)	84 (51.5%)
pTNM classification			0.032
Stage I ~ Stage II	28(16.2%)	42(25.8%)
Stage III ~ Stage IV	145 (83.8%)	121 (74.2%)
Grade			0.383
G1 ~ G2	125 (72.2%)	125 (76.7%)
G3 ~ G4	48 (27.7%)	38 (23.3%)
Perineural invasion			0.013
Positive	83 (47.9%)	58 (35.6%)
Negative	90 (52.1%)	105 (64.4%)
Nodal extracapsular spread			0.022
Gross Extension	37 (21.4%)	25 (15.3%)
Microscopic Extension	46 (26.6%)	29 (17.8%)
No Extranodal Extension	90 (52.0%)	109 (66.9%)

TLSs correlated with a positive prognosis in OSCC patients

Then the prognostic significance of TLSs in OSCC patients was assessed. According to the K-M survival analyses, TLS + OSCC was correlated with good OS and DFS (Fig. 2A, D). Based on the TLSs classification in the Methods section, we conducted subgroup analyses of TLS + OSCC patients. Compared with those in the low-TLS group, OS and DFS in the high-TLS group improved significantly (Fig. 2B, E). The K-M survival assessments indicated enhanced OS for patients in the Agg + FL1 + FL2 group compared with those in the Agg + FL1 group. Conversely, disease-free survival (DFS) rates between these groups showed no notable disparity (Fig. 2C, F). These results revealed that the presence of TLSs was a good prognostic indicator for OSCC patients, and patients with a greater proportion of TLSs had better OS and DFS.

To evaluate the potential of TLSs as a standalone predictor of outcomes in OSCC patients, we performed both univariate and multivariate Cox regression analyses. According to the univariate analysis, older age (> 60 years, hazard ratio (HR) = 1.022; 95% confidence interval (CI), 1.009–1.036; p < 0.001), pT3-4 stage (HR = 1.868; 95% CI, 1.375–2.539; p < 0.001), and lymph node metastasis (HR = 1.523; 95% CI, 1.148–2.021; p = 0.004) were linked to an elevated risk of OS, while the gender (male, HR = 0.737; 95% CI, 0.551–0.987; p = 0.040) and the presence of TLSs (HR = 0.430; 95% CI, 0.324–0.572; p < 0.001) were correlated with a reduced risk (Fig. 2G). According to the multivariate analysis, age, pT stage, lymph node metastasis status, and TLSs were found to be independent prognostic factors for the OS rate of OSCC patients. Specifically, only the presence of TLSs (HR = 0.484; 95% CI, 0.361–0.649; p < 0.001) was linked to a decreased risk of mortality (Fig. 2H).

Differential expression of genes in TLS + and TLS- OSCC tissues

TLS + and TLS- OSCC tissues were compared with normal mucosa tissues to identify DEGs. The heatmaps clearly distinguished between OSCC samples and normal samples based on the DEGs (Fig. 3A, 3D). The volcano plots revealed that 13 DEGs were downregulated and 203 DEGs were upregulated in the TLS + OSCC group (Fig. 3B), while 32 DEGs were downregulated and 207 DEGs were upregulated in the TLS-OSCC group (Fig. 3E). By comparing the upregulated DEGs in the TLS + and TLS − groups, we obtained 64 DEGs via a Venn diagram. Similarly, we obtained 5 DEGs that were downregulated in the TLS + OSCC tissues but not in the TLS- OSCC tissues. These 69 (64 + 5) DEGs, defined as TLSRGs, were used for further modeling (Fig. 3C).

To further explore the biological functions of the TLSRGs, we conducted GO and KEGG analyses on the 69 TLSRGs. GO analysis identified the primary functional categories in biological processes as the chemokine-mediated signaling pathway, regulation of dendritic cell antigen processing and presentation, and dendritic cell apoptotic process. Regarding molecular function, these TLSRGs were mainly involved in receptor-ligand activity and cytokine activity. KEGG analysis revealed that the TLSRGs were associated with cytokine-cytokine receptor interaction and chemokine signaling pathway (Fig. 3F).

Construction and identification of a prognostic model for OSCC

To determine whether the TLSRGs were associated with patient clinical outcomes, the 69 TLSRGs were evaluated by univariate Cox regression analysis (Additional Table 1). A total of 11 TLSRGs significantly influenced the OS of OSCC patients (Fig. 4A). These 11 TLS-related genes (TLSRGs) were subjected to Lasso Cox analysis, resulting in the filtration of 9 genes. (Fig. 4B). Multivariate Cox analysis was subsequently performed, and three key TLSRGs (CD86, CXCR5, and CCR7) were chosen to construct a model for prediction (Table 2). The model formula was as follows: risk score = (0.3169 × CD86 expression) + (-1.3345× CXCR5 expression) + (-0.2974 × CCR7 expression). With respect to the median risk score, the OSCC patients were then divided into two groups: one at high risk and the other at low risk (Fig. 4C). The patients’ survival time and the heatmaps of the three TLSRGs are presented in Fig. 4D and 4E. These findings indicated a direct correlation between heightened risk scores and elevated CD86 expression, concurrently revealing a decrease in CXCR5 and CCR7 expression. An increased risk score was also accompanied by a decrease in survival time. The K-M curve demonstrated that the survival rate of the low-risk group (n = 165) was markedly greater than that of the high-risk group (n = 165) (Fig. 4F). Additionally, the ROC curves revealed AUC values of 0.674, 0.652, and 0.549 for 1, 3, and 5 years respectively, suggesting that the prognostic model exhibited good sensitivity and specificity (Fig. 4G).

Table 2

Three key TLSRGs were selected after multivariate Cox analysis
TLSRGs	Coefficient
CD86	0.3169
CXCR5	-1.3345
CCR7	-0.2974

Correlations between the prognostic model and clinical parameters

To further substantiate the reliability of the model, we investigated the correlations between the expression levels of the three key TLSRGs and clinical data in OSCC patients. First, our research focused on the correlations between survival outcomes and the expression of the three TLSRGs in OSCC patients. The K-M survival analyses revealed that the elevated expressions of CCR7 and CXCR5 were directly correlated with improved OS (Fig. 5A, B), while the high expression of CD86 indicated a poor prognosis (Fig. 5C). Second, we investigated the relationships between the clinical parameters and the expression levels of the three TLSRGs. Our findings revealed a significant correlation between the downregulation of CCR7 (p = 0.036) and CXCR5 (p = 0.0051), alongside the upregulation of CD86 (p = 0.018), with the incidence of positive lymph node metastasis (Fig. 5D-F). In addition, CCR7 expression was closely related to the pT classification (p = 0.00059), pTNM classification (p = 0.002), NES (p = 0.041), and PNI (p = 0.025) (Fig. 5D; Additional Fig. 1A). Additionally, observable disparities in CD86 expression were noted across varying grade (p = 0.043) and PNI groups (p = 0.0026) (Fig. 5F; Additional Fig. 1C). We employed a heatmap to delineate the relationships between the risk score and various clinical parameters. This visual representation revealed a significant association between the high-risk group and pT classification, pTNM stage, PNI, NES, and lymph node metastasis (Fig. 5G). The above results indicated that the three TLSRGs were closely related to lymph node metastasis and the prognostic model exhibited favorable predictive ability.

Correlations of TLSs, risk score, and tumor-infiltrating immune cells

Given the crucial role of TLSs in antitumor immunity, TIMER2.0 was utilized to investigate the characteristics of immune cell infiltration in the TLS + OSCC group. As shown in Fig. 6A, strong correlations are noted between TLSs and B cells, CD4 + T cells, CD8 + T cells, and macrophages in OSCC. A high ratio of TLSs area suggested abundant B cell, CD4 + T cell, and macrophage infiltration in the microenvironment (Fig. 6B). The infiltration of CD4 + T cells, CD8 + T cells, and macrophages exhibited a negative correlation with the risk score (Fig. 6C), indicating a potential inverse relation between immune cell presence and risk assessment metrics. These results suggested that these three immune cell types played important roles in the formation of TLSs and bolstered anti-tumor defenses.

Establishment of the prognostic nomogram for individualized evaluation

Immunohistochemistry was used to investigate the expression of the characteristic proteins corresponding to the three TLSRGs in clinical samples (Fig. 7A). The result showed that the percentage of CD86-positive cells was the lowest in the TLS + OSCC group (Fig. 7D). Compared with those in the normal group and TLS- OSCC group, the percentages of cells positive for CCR7 and CXCR5 were greater in TLS + OSCC group (Fig. 7B, C).

To further provide reliable prognostic information tailored to individual patients, a predictive nomogram was developed to estimate the survival rates at one, three, and five years for OSCC patients (Fig. 8A). Additionally, a calibration plot was generated to illustrate the model's good predictive value, depicting the predicted probabilities at 1, 3, and 5 years relative to the actual observations (Fig. 8B).

In this study, we analyzed HE-stained WSIs for the detection of TLSs by employing convolutional neural networks. Our findings validated TLSs as the robust, independent prognostic marker for OSCC patients. To make TLSs broadly applicable in clinical settings, we developed a risk score model including three TLSRGs and verified its independent prognostic value. We also investigated the relationships between TLSs, risk score, and immune cell infiltration. After verifying the protein expression of the three TLSRGs in clinical samples, a nomogram model was constructed for the individualized evaluation of OSCC patients by incorporating routine clinical metrics.

We found that TLSs were unevenly distributed in OSCC, with its area less than 2.8% of the tumor area. However, in a few cases, the proportion could reach 4–6%. This suggests that the patch- or tile-based approaches for image analysis used in many studies are inappropriate, which may lead to misestimation of the presence of TLSs. Zeng et al. (21) reported that the percentage of TLS-positive breast ductal carcinomas was 24.7%, which is significantly different from the 60.3% reported in a Korean study (22). Wirsing's study revealed that analyzing a single level in OSCC tissue blocks failed to detect approximately one-third of TLS + patients (23). Most previous studies have used multiplex immunohistochemistry (mIHC) or immunofluorescence (mIF) approaches to detect TLSs (24, 25). Despite the potential of multiplex immunohistochemistry (mIHC) and multiplex immunofluorescence (mIF) for providing detailed cellular insights, their adoption in clinical settings is restricted by their prohibitive costs and operational complexity. Conversely, hematoxylin and eosin (HE) staining stands as the cornerstone of histopathological analysis due to its affordability and widespread availability, offering a practical alternative for routine examinations. TLSs vary greatly in size, density, maturity, and distribution, and the evaluation of TLSs is affected by pathologists’ experience (26, 27). Manual assessment is labor-intensive, relies on expertise, and often yields poor reproducibility. Therefore, we believe that it is more reliable and standardized to use a convolutional neural network to identify TLSs on HE-stained WSIs at multiple levels.

Research indicates that the presence of mature TLSs is positively correlated with enhanced OS and DFS among patients with early-stage OTSCC (15). OSCC patients without TLSs have a poorer prognosis than those with TLSs (16). Similarly, we found that the presence of TLSs was a positive factor for OS and DFS and was an independent prognostic factor for OSCC. OSCC patients with FL2 generally had longer overall survival than those without FL2, but there was no significant difference in DFS between the two groups. In OSCC, the predictive value of the TLSs area ratio is better than that of maturity. Patients with a high TLSs area have a favorable prognosis.

Despite these advancements, validated molecular markers linked to TLSs that can consistently predict the prognosis of OSCC patients are lacking. Through a comparison of sequencing data from TLS + and TLS- OSCC tissues, we identified 69 TLSRGs, which were mainly enriched in the chemokine-mediated signaling pathway, regulation of dendritic cell antigen processing and presentation, and cytokine-cytokine receptor interaction. Based on the abundance of chemokines and cytokines expressed in TLSs, some studies have established the disease-specific gene signatures to assess the presence of TLSs in tumors (18, 28). Liu et al. (14) propose a 13-gene signature to assess the level of TLSs in HNSCC and these selected genes partially overlap with our screening results. In our research, CCR7, CXCR5, and CD86 were identified by LASSO regression and stepwise Cox regression analysis, and a prognostic model was constructed.

CXCR5 and CCR7 play pivotal roles in lymphoid organogenesis and the preservation of lymphoid tissue architecture, orchestrating the migration of lymphocytes and dendritic cells (DCs) to secondary lymphoid structures (29). CCR7 is highly expressed on some CD4 + T cells, B cells, and DCs, which migrate to the T-cell zone via CCL19 and CCL21. DCs introduce antigens to uninitiated T cells, facilitating the transformation of naive CD4 + T cells into follicular helper T cells (Tfh cells), which are essential for adaptive immunity (30). The expression of the chemokine receptor CXCR5 on Tfh cells gradually increases, and the expression of CCR7 decreases (31).

CXCR5, which is predominantly expressed on B cells, Tfh cells, and mature DCs, plays a crucial role in cell migration (32–34). Its likely sole ligand, CXCL13, is expressed by follicular dendritic cells (FDCs) and various stromal cells situated in the B-cell regions of secondary lymphoid organs (35, 36). Tfh cells and B cells with high CXCR5 expression migrate to the B-cell zone through CXCL13 (31). B cells and Tfh cells engage with FDCs to foster the germinal center reaction, which leads to the evolution of B cells into memory B cells and enduring plasma cells (37). TLSs are structurally and functionally similar to secondary lymphoid organs (6) and are the site of effector T cell, memory T cell as well as B cell differentiation (38). Therefore, we hypothesize that CXCR5 and CCR7 play similar roles in TLSs, such as recruiting immune cells and promoting TLSs formation.

Primarily located on antigen-presenting cells, CD86 serves as a crucial ligand for CD28 and CTLA-4, which are found on the surface of T cells (39). The interaction of CD86 with CD28 stimulates T cell activation, while the interaction of CD86 with CTLA-4 suppresses T cell activation and decreases the immune response (40, 41). According to previous studies, CTLA-4 has a greater binding affinity for CD86 than for CD28, and the CTLA-4-CD86 interaction counteracts the CD86-CD28 interaction, leading to immune suppression (42, 43).

Our research revealed a significant correlation between TLSs and the presence of B cells, CD4 + T cells, CD8 + T cells, and macrophages in OSCC, highlighting their potential role in the tumor microenvironment. A rise in the percentage of TLSs area was favorably connected with immune cell infiltration, although it was adversely correlated with the risk score. The upregulation of CCR7 and CXCR5 and the downregulation of CD86 and the risk score were negatively correlated with lymph node metastasis, indicating a better prognosis. These findings indicate that the presence of TLSs is negatively correlated with lymph node metastasis and predicts a better prognostic outcome. Our findings are supported by other literature. According to the Human Protein Atlas (44), high CCR7 expression is associated with a better prognosis in OSCC patients (45). In melanoma, CCR7 + DCs play a key role in trafficking tumor antigens to lymphoid tissues and activating T cells (46). Zhang et al. reported that CXCR5 + CD4 + Tfh cells play a pivotal role in developing and sustaining TLSs and are associated with a good prognosis in HNSCC patients (47). Wang et al. confirmed that high levels of CXCR5 + CD8 + T cells correlate with improved OS in gastric cancer patients (48). Upregulation of CTLA4 is an important immunosuppressive mechanism in HNSCC (49). Wakasu and Zhang et al. reported that patients with mature TLSs have a significantly lower incidence of lymph node metastasis (10, 50).

We propose that TLSs distributed in tumors are on the front line of antitumor immunity. A high TLSs area ratio and maturity indicate strong antitumor immunity. When the strength of antitumor immunity around highly malignant tumors is weak, the incidence of lymph node metastasis is greatly increased. However, some studies suggest that cancer cells can metastasize to lymph nodes by upregulating CCR7 expression (51), leading to a worse prognosis (52–54). Two opposite effects of CCR7 have been reported in different cancer studies, and one possible explanation is that the changes in the tumor microenvironment alter the effect of CCR7. In the early stage of cancer, a small minority of CCR7 + tumor cells migrate to lymph nodes, where they may play a role in presenting tumor antigens and activating immunity. With tumor progression, alterations in the microenvironment may facilitate the colonization of CCR7 + tumor cells in lymph nodes. Second, the heterogeneities of tumors at various sites are large, and the interactions between immune cells and tumors are complex. A single indicator cannot accurately predict the prognosis of all cancers, so it is necessary to combine multiple indicators to make a more accurate prediction. To a certain extent, these findings validate the appropriateness of selecting these three TLSRGs to construct a prognostic model. We further confirmed the prognostic model's effectiveness in clinical applications at the tissue level by performing immunohistochemical staining of three TLSRGs. Finally, routine clinical parameters were integrated into the risk score model, and the nomogram model was constructed to enhance the specificity of individual prognosis prediction.

Our study has several limitations. First, it is important to note that the results primarily rely on the TCGA dataset and require further validation in additional databases. Some individuals had received immune or targeted treatments, which influenced the prognosis analysis. Although the three TLSRGs were validated in clinical samples, the biological functions of these genes in OSCC necessitate further verification through experiments, and expanding the sample size would be beneficial.

At present, only a few studies have demonstrated that TLSs can act as the prognostic biomarker and the predictor of immunotherapy efficacy in OSCC patients. Many different markers, which cannot be widely used in clinical practice, have been used to characterize TLSs. Compared to previous studies, this study offers the following advantages: (1) it used a convolutional neural network to accurately identify TLSs and underscores the potential of integrating advanced image analysis techniques in oncological prognostication; (2) it explored the relationships between TLSs, lymph node metastasis, and immune infiltration; and (3) it is the first OSCC nomogram prediction model based on TLSRGs, which can be used as a supplementary indicator of the prognosis in OSCC patients. Its use may be related to further refinement of the current staging system and improvement of risk stratification.

Agg

Lymphoid aggregates

AUC

Area under the curve

Confidence interval

DCs

Dendritic cells

DEGs

Differentially expressed genes

DFS

Disease-free survival

FDCs

Follicular dendritic cells

FDR

False discovery rate

FL1

Primary follicle-like TLSs

FL2

Secondary follicle-like TLSs

GCO

Global Cancer Observatory

Gene Ontology

Hematoxylin and eosin

HNSCC

Head neck squamous cell carcinoma

Hazard ratio

KEGG

Kyoto Encyclopedia of Genes and Genomes

mIHC

Multiplex immunohistochemistry

mIF

Multiplex immunofluorescence

NES

Nodal extracapsular spread

OTSCC

Oral tongue squamous cell carcinoma

Overall survival

OSCC

Oral squamous cell carcinoma

PNI

Perineural invasion

ROC

Receiver operating characteristic

TCGA

The Cancer Genome Atlas

Tfh cells

Follicular helper T cells

TIMER

Tumor Immune Estimation Resource

TLSs

Tertiary lymphoid structures

TLSRGs

TLS-related genes

WSIs

Whole slide images.

Ethics approval and consent to participate

This study received approval from the Ethics Committee of the Second Xiangya Hospital of Central South University (No.2022111194), and all participants provided informed consent prior to their inclusion.

Consent for publication

Not applicable.

Availability of data and materials

The datasets generated and analyzed during the current study are available in the TCGA repository, https://www.cancer.gov/ccg/research/genome-sequencing/tcga.

Competing interests

The authors declare that they have no competing interests.

Funding

This study was supported by the National Engineering Research Center of Oral Biomaterials and Digital Medical Devices of Peking University and Hospital of Stomatology (PKUSS20230202), the Department of Science and Technology of Hunan Province (2023JJ41028, 2024JJ6352), and the Natural Science Foundation of Hainan Province (821MS312).

Authors' contributions

BS and FZ designed the study. CG and FZ performed the bioinformatics analyses. YT and QX contributed to the clinical sample collection and experiments. BS wrote the manuscript, and KW helped to revise the manuscript. All the authors have read and approved the final manuscript.

Acknowledgements

Not applicable.

Tan Y, Wang Z, Xu M, Li B, Huang Z, Qin S, et al. Oral squamous cell carcinomas: state of the field and emerging directions. Int J Oral Sci. 2023;15(1):44.
Mody MD, Rocco JW, Yom SS, Haddad RI, Saba NF. Head and neck cancer. Lancet. 2021;398(10318):2289-99.
Amin MB, Greene FL, Edge SB, Compton CC, Gershenwald JE, Brookland RK, et al. The Eighth Edition AJCC Cancer Staging Manual: Continuing to build a bridge from a population-based to a more "personalized" approach to cancer staging. CA Cancer J Clin. 2017;67(2):93-9.
Zhang SY, Ren XY, Wang CY, Chen XJ, Cao RY, Liu Q, et al. Comprehensive Characterization of Immune Landscape Based on Epithelial-Mesenchymal Transition Signature in OSCC: Implication for Prognosis and Immunotherapy. Front Oncol. 2021;11:587862.
Balachandran VP, Gonen M, Smith JJ, DeMatteo RP. Nomograms in oncology: more than meets the eye. Lancet Oncol. 2015;16(4):e173-80.
Pitzalis C, Jones GW, Bombardieri M, Jones SA. Ectopic lymphoid-like structures in infection, cancer and autoimmunity. Nat Rev Immunol. 2014;14(7):447-62.
Schumacher TN, Thommen DS. Tertiary lymphoid structures in cancer. Science. 2022;375(6576):eabf9419.
Zhao Z, Ding H, Lin ZB, Qiu SH, Zhang YR, Guo YG, et al. Relationship between Tertiary Lymphoid Structure and the Prognosis and Clinicopathologic Characteristics in Solid Tumors. Int J Med Sci. 2021;18(11):2327-38.
Hayashi Y, Makino T, Sato E, Ohshima K, Nogi Y, Kanemura T, et al. Density and maturity of peritumoral tertiary lymphoid structures in oesophageal squamous cell carcinoma predicts patient survival and response to immune checkpoint inhibitors. Br J Cancer. 2023;128(12):2175-85.
Zhang C, Wang XY, Zuo JL, Wang XF, Feng XW, Zhang B, et al. Localization and density of tertiary lymphoid structures associate with molecular subtype and clinical outcome in colorectal cancer liver metastases. J Immunother Cancer. 2023;11(2).
Wang Q, Sun K, Liu R, Song Y, Lv Y, Bi P, et al. Single-cell transcriptome sequencing of B-cell heterogeneity and tertiary lymphoid structure predicts breast cancer prognosis and neoadjuvant therapy efficacy. Clin Transl Med. 2023;13(8):e1346.
Badalamenti G, Fanale D, Incorvaia L, Barraco N, Listì A, Maragliano R, et al. Role of tumor-infiltrating lymphocytes in patients with solid tumors: Can a drop dig a stone? Cell Immunol. 2019;343:103753.
Sautès-Fridman C, Petitprez F, Calderaro J, Fridman WH. Tertiary lymphoid structures in the era of cancer immunotherapy. Nat Rev Cancer. 2019;19(6):307-25.
Liu Z, Meng X, Tang X, Zou W, He Y. Intratumoral tertiary lymphoid structures promote patient survival and immunotherapy response in head neck squamous cell carcinoma. Cancer Immunol Immunother. 2022.
Wang C, Huang Z, Zhang M, Xiong G, Chen X, Xie N. Prognostic value of tertiary lymphoid structures in early clinical stage oral tongue squamous cell carcinoma. J Oral Pathol Med. 2021;50(8):776-84.
Li Q, Liu X, Wang D, Wang Y, Lu H, Wen S, et al. Prognostic value of tertiary lymphoid structure and tumour infiltrating lymphocytes in oral squamous cell carcinoma. Int J Oral Sci. 2020;12(1):24.
Barmpoutis P, Di Capite M, Kayhanian H, Waddingham W, Alexander DC, Jansen M, Kwong FNK. Tertiary lymphoid structures (TLS) identification and density assessment on H&E-stained digital slides of lung cancer. PLoS One. 2021;16(9):e0256907.
Finkin S, Yuan D, Stein I, Taniguchi K, Weber A, Unger K, et al. Ectopic lymphoid structures function as microniches for tumor progenitor cells in hepatocellular carcinoma. Nat Immunol. 2015;16(12):1235-44.
Murakami J, Shimizu Y, Kashii Y, Kato T, Minemura M, Okada K, et al. Functional B-cell response in intrahepatic lymphoid follicles in chronic hepatitis C. Hepatology. 1999;30(1):143-50.
Li T, Fan J, Wang B, Traugh N, Chen Q, Liu JS, et al. TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Res. 2017;77(21):e108-e10.
Zeng L, Koh VCY, Chen XY, Tan PH. Tertiary lymphoid structures in breast ductal carcinoma in situ correlate with adverse pathological parameters. Histopathology. 2023;82(5):779-88.
Kim A, Heo SH, Kim YA, Gong G, Jin Lee H. An Examination of the Local Cellular Immune Response to Examples of Both Ductal Carcinoma In Situ (DCIS) of the Breast and DCIS With Microinvasion, With Emphasis on Tertiary Lymphoid Structures and Tumor Infiltrating Lymphoctytes. Am J Clin Pathol. 2016;146(1):137-44.
Wirsing AM, Rikardsen OG, Steigen SE, Uhlin-Hansen L, Hadler-Olsen E. Characterisation and prognostic value of tertiary lymphoid structures in oral squamous cell carcinoma. BMC Clin Pathol. 2014;14:38.
Ruffin AT, Cillo AR, Tabib T, Liu A, Onkar S, Kunning SR, et al. B cell signatures and tertiary lymphoid structures contribute to outcome in head and neck squamous cell carcinoma. Nat Commun. 2021;12(1):3349.
Rakaee M, Kilvaer TK, Jamaly S, Berg T, Paulsen EE, Berglund M, et al. Tertiary lymphoid structure score: a promising approach to refine the TNM staging in resected non-small cell lung cancer. Br J Cancer. 2021;124(10):1680-9.
Díaz A, Forner A. Prognosis assessment by pathologist: Is the detection of intratumoural tertiary lymphoid structures a reliable tool? J Hepatol. 2019;70(1):11-2.
Munoz-Erazo L, Rhodes JL, Marion VC, Kemp RA. Tertiary lymphoid structures in cancer - considerations for patient prognosis. Cell Mol Immunol. 2020;17(6):570-5.
Xu W, Ma C, Liu W, Anwaier A, Tian X, Shi G, et al. Prognostic value, DNA variation and immunologic features of a tertiary lymphoid structure-related chemokine signature in clear cell renal cell carcinoma. Cancer Immunol Immunother. 2022;71(8):1923-35.
Müller G, Höpken UE, Lipp M. The impact of CCR7 and CXCR5 on lymphoid organ development and systemic immunity. Immunol Rev. 2003;195:117-35.
Barnett LG, Simkins HM, Barnett BE, Korn LL, Johnson AL, Wherry EJ, et al. B cell antigen presentation in the initiation of follicular helper T cell and germinal center differentiation. J Immunol. 2014;192(8):3607-17.
Allen CD, Cyster JG. Follicular dendritic cell networks of primary follicles and germinal centers: phenotype and function. Semin Immunol. 2008;20(1):14-25.
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(4):782-95.
Garg R, Blando JM, Perez CJ, Abba MC, Benavides F, Kazanietz MG. Protein Kinase C Epsilon Cooperates with PTEN Loss for Prostate Tumorigenesis through the CXCL13-CXCR5 Pathway. Cell Rep. 2017;19(2):375-88.
Gu-Trantien C, Migliori E, Buisseret L, de Wind A, Brohée S, Garaud S, et al. CXCL13-producing TFH cells link immune suppression and adaptive memory in human breast cancer. JCI Insight. 2017;2(11).
Gunn MD, Ngo VN, Ansel KM, Ekland EH, Cyster JG, Williams LT. A B-cell-homing chemokine made in lymphoid follicles activates Burkitt's lymphoma receptor-1. Nature. 1998;391(6669):799-803.
Ansel KM, Ngo VN, Hyman PL, Luther SA, Förster R, Sedgwick JD, et al. A chemokine-driven positive feedback loop organizes lymphoid follicles. Nature. 2000;406(6793):309-14.
Tokunaga R, Naseem M, Lo JH, Battaglin F, Soni S, Puccini A, et al. B cell and B cell-related pathways for novel cancer treatments. Cancer Treat Rev. 2019;73:10-9.
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(7):832-44.
Bugeon L, Dallman MJ. Costimulation of T cells. Am J Respir Crit Care Med. 2000;162(4 Pt 2):S164-8.
Bolandi N, Derakhshani A, Hemmat N, Baghbanzadeh A, Asadzadeh Z, Afrashteh Nour M, et al. The Positive and Negative Immunoregulatory Role of B7 Family: Promising Novel Targets in Gastric Cancer Treatment. Int J Mol Sci. 2021;22(19).
Romo-Tena J, Gómez-Martín D, Alcocer-Varela J. CTLA-4 and autoimmunity: new insights into the dual regulator of tolerance. Autoimmun Rev. 2013;12(12):1171-6.
Krummel MF, Allison JP. CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation. J Exp Med. 1995;182(2):459-65.
Chikuma S. CTLA-4, an Essential Immune-Checkpoint for T-Cell Activation. Curr Top Microbiol Immunol. 2017;410:99-126.
Uhlen M, Zhang C, Lee S, Sjöstedt E, Fagerberg L, Bidkhori G, et al. A pathology atlas of the human cancer transcriptome. Science. 2017;357(6352).
Korbecki J, Grochans S, Gutowska I, Barczak K, Baranowska-Bosiacka I. CC Chemokines in a Tumor: A Review of Pro-Cancer and Anti-Cancer Properties of Receptors CCR5, CCR6, CCR7, CCR8, CCR9, and CCR10 Ligands. Int J Mol Sci. 2020;21(20).
Roberts EW, Broz ML, Binnewies M, Headley MB, Nelson AE, Wolf DM, et al. Critical Role for CD103(+)/CD141(+) Dendritic Cells Bearing CCR7 for Tumor Antigen Trafficking and Priming of T Cell Immunity in Melanoma. Cancer Cell. 2016;30(2):324-36.
Zhang B, Li H, Liu YT, Xiong D, Zhang L, Sun ZJ. Single-cell chemokine receptor profiles delineate the immune contexture of tertiary lymphoid structures in head and neck squamous cell carcinoma. Cancer Lett. 2023;558:216105.
Wang J, Li R, Cao Y, Gu Y, Fang H, Fei Y, et al. Intratumoral CXCR5(+)CD8(+)T associates with favorable clinical outcomes and immunogenic contexture in gastric cancer. Nat Commun. 2021;12(1):3080.
Yu GT, Bu LL, Zhao YY, Mao L, Deng WW, Wu TF, et al. CTLA4 blockade reduces immature myeloid cells in head and neck squamous cell carcinoma. Oncoimmunology. 2016;5(6):e1151594.
Wakasu S, Tagawa T, Haratake N, Kinoshita F, Oku Y, Ono Y, et al. Preventive effect of tertiary lymphoid structures on lymph node metastasis of lung adenocarcinoma. Cancer Immunol Immunother. 2023;72(6):1823-34.
Morein D, Erlichman N, Ben-Baruch A. Beyond Cell Motility: The Expanding Roles of Chemokines and Their Receptors in Malignancy. Front Immunol. 2020;11:952.
Sperveslage J, Frank S, Heneweer C, Egberts J, Schniewind B, Buchholz M, et al. Lack of CCR7 expression is rate limiting for lymphatic spread of pancreatic ductal adenocarcinoma. Int J Cancer. 2012;131(4):E371-81.
Ding Y, Shimada Y, Maeda M, Kawabe A, Kaganoi J, Komoto I, et al. Association of CC chemokine receptor 7 with lymph node metastasis of esophageal squamous cell carcinoma. Clin Cancer Res. 2003;9(9):3406-12.
Shang ZJ, Liu K, Shao Z. Expression of chemokine receptor CCR7 is associated with cervical lymph node metastasis of oral squamous cell carcinoma. Oral Oncol. 2009;45(6):480-5.

No competing interests reported.

Download PDF

Journal Publication

published 26 Oct, 2024

Read the published version in Cancer Cell International →

Editorial decision: Revision requested
03 Oct, 2024
Reviews received at journal
30 Sep, 2024
Reviewers agreed at journal
08 Sep, 2024
Reviews received at journal
21 Aug, 2024
Reviewers agreed at journal
10 Aug, 2024
Reviewers invited by journal
03 Jul, 2024
Submission checks completed at journal
03 May, 2024
Editor assigned by journal
03 May, 2024
First submitted to journal
02 May, 2024

You are reading this latest preprint version

Identification and Validation of a Prognostic Model Based on Three TLS- Related Genes in Oral Squamous Cell Carcinoma

Status:

Journal Publication

Version 1

Abstract

Figures

Background

Methods

Data Acquisition

TLSs Quantification

DEGs, GO, KEGG analyses

Construction of the risk score model and prognostic nomogram

Immune cell infiltration estimation

Tissue samples

Immunohistochemistry

Statistical analyses

Results

Clinicopathological characteristics and TLSs expression of OSCC patients

TLSs correlated with a positive prognosis in OSCC patients

Differential expression of genes in TLS + and TLS- OSCC tissues

Construction and identification of a prognostic model for OSCC

Correlations between the prognostic model and clinical parameters

Correlations of TLSs, risk score, and tumor-infiltrating immune cells

Establishment of the prognostic nomogram for individualized evaluation

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1