Patients, Study Design, and Tissue Microarray (TMA) Construction
This study comprises two cohorts. In the training cohort, LSCC cases were collected from the Otolaryngology Head and Neck Surgery Department of the Eye, Ear, Nose, and Throat Hospital between 2014 and 2018. The LSCC cohort included TMA cores from 80 patients, with 5 patients excluded due to loss during follow-up. Tumor recurrence was defined as either local recurrence or distant metastases. The study was ethically approved by the Ethics Boards of the Eye and ENT Hospital of Fudan University, and it was conducted in adherence to the principles of the Declaration of Helsinki (No. KJ2008-01). Informed consent was obtained from all participants. Supplementary Table S1 provides baseline clinical characteristics of the LSCC patients included in the TMA analysis.
In the validation cohort, RNA-seq data and clinical information of 116 LSCC sample and 12 noncancerous adjacent tissue samples were obtained from the TCGA database (https://portal.gdc.cancer.gov), which includes transcriptome information and clinical data. The clinical information for these 116 LSCC patients comprises survival time, TNM classification, stage, age, and gender.
Immunohistochemistry (IHC) and IS Construction
Sections of 4-5μm thickness were cut from each selected block of formalin-fixed, paraffin-embedded TMA tissue (Wuhan Biosci Biotechnology Co.; Ltd, Wuhan, China). The staining protocol was established as previously reported (4, 14). CD3+, CD8+, CD4+, and Foxp3+ staining were imaged using an Aperio digital slide scanner (Leica Biosystems) and analyzed with QuPath v. 0.2.3 (Queen's University, Belfast, Northern Ireland) to determine the number of positive cells on the TMA. Two board-certified pathologists confirmed the definitions of tumor inferior (TI) and tumor stroma (TS) areas through performing Hematoxylin and eosin (HE) staining. Tumor-infiltrating lymphocytes (TILs) were calculated by summing individual cells from both the tumor interior (TI) and the tumor stroma (TS). The cutoff value for each TIL type, which separated patients into two groups (low or high infiltration), was determined using the Youden index. Based on the infiltration of CD3+ and CD8+ T cells, the mean of the four percentiles (two markers, two regions) was calculated and converted into an Immunoscoring (IS) system. Specifically, the IS categories are as follows: 1) IS=0 indicates low infiltration of both CD3+ and CD8+ T cells; 2) IS=1 indicates high density of at least one marker in one region (TI or TS); 3) IS=2-4 indicates a total score for each region with high density(15).
Immunofluorescence Staining and CIL Construction
For TMA analysis, anti-CD68 primary antibody (1:200, 76437, clone D4B9C, CST) or goat monoclonal anti-CD206 (1:100, AF2534, R&D Systems) were incubated overnight at 4°C. Following three 10-minute washes with PBS, the sections were incubated in the dark for 60 minutes with different fluorochrome-conjugated secondary antibodies. After washing, the slides were mounted using Vectashield Hardset mounting medium, with 4′,6-diamidino-2-phenylindole (1:200, DAPI; Solarbio, C0060). Co-expression of both CD206+ and CD68+ was considered M2 macrophages, while co-expression of both CD66b+ (1:200, 305102, clone G10F5, BioLegend) and CXCR4+ (ab124824, UMB2, Abcam) was considered N2 neutrophils. QuPath was used to evaluate cell detection through nuclei (DAPI), considering cytoplasmic fluorescence of each cell with a mean intensity greater than 100 as positive.
CILM2N2 was computed as the ratio of CD8+ cell density to the sum of CD8+, M2 (CD206+CD68+), and N2 (CD66b+CXCR4+). CILM2 was computed as the ratio of CD8+ cell density to the sum of CD8+ and M2 (CD206+CD68+). Similarly, CILN2 was computed as the ratio of CD8+ cell density to the sum of CD8+ and N2 (CD66b+CXCR4+).
Analysis of TCGA RNA Data: Enhancing Immunologiclal Profiling in Laryngeal Squamous Cell Carcinoma
The IS-TCGA was derived by calculating the average of expression CD8A and CD3E, following a previously reported methodology(16). To assess the landscape of tumor-infiltrating lymphocytes (TILs) in TCGA-LSCC, the CIBERSORT algorithm, utilizing R software (Version 458, https://www.r-project.org/), was employed. Additionally, CILM2-TCGA was determined as the ratio of CD8+ cell density to the combined CD8+, M0 and M2 cell population.
Statistical Analysis: Evaluating Predictive Accuracy and Survival Outcomes
To assess the predictive accuracy of CIL, IS, and clinic-pathological parameters, integrated area under the ROC curve (iAUC) was computed with 500× bootstrap resampling through R Studio (Version 2022.02.2). Overall Survival (OS) and Disease-Free Survival (DFS) curves were generated using the Kaplan-Meier method with SPSS (version 25.0) and GraphPad Prism (version 8.0), respectively. The risk of OS and DFS was analyzed through univariate and multivariate analyses using the log-rank (Mantel-Cox) test and Cox proportional hazard models, respectively. Additionally, the hazard ratio (HR) with 95% confidence intervals (CI) was determined through univariate Cox regression.
We determined the relative importance of parameters in estimating survival risk using a multi-variable Cox proportional hazards model. In this analysis, we considered clinical factors, CIL, and IS as co-variables for the 'cph' function from the 'rms' R package. By applying the 'anova' function to the 'chp' object, we obtained a matrix of predictors that reflects the variables' importance in the model, assessed by the Wald chi-square (χ2) statistic, as previous reported(16). The specific code is provided in the supplementary materials.
Associations between TILs density, SIRI, and recorded clinicopathological characteristics were examined using χ2 tests for categorical variables. A P-value of 0.05 (two-tailed) was considered statistically significant.