Heterogeneity of tumor-infiltrating T cells and its impact on anti-PD-1/PD-L1 immunotherapy: insights from integrated multicancer single-cell and bulk transcriptomic analysis

Highlights

• Single-cell transcriptomic analysis of multicancer cohort reveals significant heterogeneity among tumor-infiltrating T cells.

• Heterogeneity in T-cell subtypes suggests potential biomarkers for efficacy of anti-PD-1/PD-L1 immunotherapy.

• Bulk transcriptomic data further supports the role of T-cell subtypes as potential biomarkers for immunotherapy response.

Immunotherapy targeting the programmed cell death 1/programmed cell death ligand 1 (PD-1/PD-L1) pathway has shown remarkable efficacy in treating diverse cancers. A successful antitumor immune response crucially depends on the presence and activation of tumor-infiltrating T cells¹. Tumor-infiltrating T cells exhibit significant phenotypic and functional diversity, which directly impacts the efficacy of anticancer therapies. The prognostic significance of tumor-infiltrating T-cell heterogeneity remains to be fully understood. It is, therefore, imperative to study the biological relevance and clinical significance of different subtypes, especially in the context of immunotherapy, where T cells play a major role². Single-cell transcriptomic analysis precisely offers a key platform for unraveling the heterogeneity of T cells, enhancing understanding of their diverse roles in immunotherapy.

To investigate the impact of heterogeneous tumor-infiltrating T cells on immunotherapy, we fully retrieved and integrated single-cell transcriptomic data of pretreatment tumor samples from 61 patients across 7 cancer types who received anti-PD-1/PD-L1 immunotherapy and had available response status (Fig. 1A). Details of data integration and participant characteristics can be found in the Supplemental Files, Table S1 (Supplemental Digital Content 1, http://links.lww.com/JS9/D568) and Table S2 (Supplemental Digital Content 1, http://links.lww.com/JS9/D568). Preprocessing ensured the integration of data from different cancer types without batch effects (Fig. S1, Supplemental Digital Content 1, http://links.lww.com/JS9/D568). For subsequent analysis, 50 156 CD8⁺ T cells and 42 974 CD4⁺ T cells were divided into 10 and 7 subtypes, respectively, mainly including exhausted T cells (T_EX), naive T cells (T_N), effector memory T cells (T_EM), stress response T cells (T_STR), terminally differentiated effector memory T cells (T_EMRA), regulatory T cells (T_REG), and follicular helper T cells (T_FH) (Fig. 1B, C). Additionally, 28 717 CD8⁺ T cells and 28 534 CD4⁺ T cells from 31 patients had paired T-cell receptor (TCR) sequences available. According to the expression of marker genes and the Shannon equitability index of TCR sequences, obvious heterogeneity can be found among different subtypes (Fig. 1D, E). For example, T_EX subtypes (CD8 C1 and CD4 C3) exhibit higher immune checkpoint expression and lower Shannon equitability index, indicating greater clonal expansion, while the opposite is true for T_N subtypes (CD8 C2 and CD4 C1). Signature scoring of functional gene sets further demonstrated functional heterogeneity among subtypes (Fig. 1F, G). Additionally, single-cell gene regulatory network analysis revealed that different subtypes exhibited specific activated transcription factors, underscoring distinct gene expression regulation patterns (Fig. S2, Supplemental Digital Content 1, http://links.lww.com/JS9/D568).

Figure 1.

Multicancer tumor-infiltrating T cell profile at single-cell resolution from patients undergoing anti-PD-1/PD-L1 immunotherapy. (A) Schematics of the single-cell transcriptome of tumor-infiltrating T cells (created with BioRender.com). (B) UMAP plot of 10 CD8⁺ T cell subtypes (left) and bar graphs showing summary statistics for the number of CD8⁺ T cells by cancer type, response to immunotherapy, and target of immunotherapy (right). (C) The same plot as in (B) is applied to CD4⁺ T cells. (D) Heatmap showing the expression of marker genes for each CD8⁺ T cell subtype (left) and box plots showing the Shannon equitability index of TCR sequences for each CD8⁺ T cell subtype (right). (E) The same plot as in (D) is applied to CD4⁺ T cells. BCC, basal cell carcinoma; ESCA, esophageal cancer; HCC, hepatocellular carcinoma; LUAD, lung adenocarcinoma; MELA, melanoma; SCC, squamous cell carcinoma; TNBC, triple-negative breast cancer.

In addition to calculating the overall Shannon equitability index of each subtype, we also compared differences in TCR clone sizes (Fig. 2A, B). Although the clonal frequency distribution among different subtypes is uneven for both CD8⁺ and CD4⁺ T cells, we can also find that the overall clonal frequency of CD8⁺ T cells is higher, which means that the expansion of CD8⁺ T cells is more common in the tumor microenvironment and may be related to their specific recognition and elimination of tumor cells (Fig. 2C, D). According to the overlap coefficient of TCR sequences, it can be found that subtypes with more clone expansion, such as CD8 C1 and CD4 C3, often have higher TCR overlap with other subtypes, suggesting that the expansion of T cells in the tumor microenvironment is typically global rather than subtype-specific (Fig. 2E, F). T cells with the same TCR sequence but different subtypes gradually develop functional heterogeneity during the process of proliferation and differentiation.

Figure 2.

Distribution and similarity of TCR sequences among different T cell subtypes. (A) UMAP plot of CD8⁺ T cell with clone size = 1 (top) and clone size ≥2 (bottom). (B) The same plot as in (A) is applied to CD4⁺ T cells. (C) UMAP plot (left) and bar graph (right) showing the distribution of TCR clonal frequency among CD8⁺ T cell subtypes. (D) The same plot as in (C) is applied to CD4⁺ T cells. (E) Bubble plot showing the similarity of TCR sequences among CD8⁺ T cell subtypes. (F) The same plot as in (E) is applied to CD4⁺ T cells.

Furthermore, we evaluated the association of the frequencies of various tumor-infiltrating T-cell subtypes with the efficacy of anti-PD-1/PD-L1 immunotherapy. By analyzing the compositions of these subtypes in pretreatment tumors, we found that the frequency of CD8⁺ T_N (CD8 C2) was significantly higher in responsive tumors, while the frequency of CD4⁺ T_EX (CD4 C3) was significantly higher in nonresponsive tumors (Fig. 3A, B). Interestingly, while not statistically significant, the relationship between the frequency of CD8⁺ T_EX and tumor response exhibited a trend similar to that of CD4⁺ T_EX, as well as that between CD4⁺ T_N and CD8⁺ T_N (Fig. 3A, B). We also explored the influence of prior treatments on T-cell profiles and found higher CD4⁺ T_EX (CD4 C3) frequency in pretreated patients (Fig. S3, Supplemental Digital Content 1, http://links.lww.com/JS9/D568). This observation may be attributed to the fact that T_N cells, as precursors, undergo activation and differentiation upon encountering and recognizing tumor antigens, thus forming antigen-experienced T cells and responding to immunotherapy³. Conversely, unlike T_N, tumor-infiltrated T_EX are in an inflexible state and resistant to immunotherapy, which is consistent with previous reports^4,5.

Figure 3.

Association of the frequencies of various T cell subtypes with the efficacy of anti-PD-1/PD-L1 immunotherapy. (A) Comparisons of frequency of each CD8⁺ T cells subtype between nonresponsive and responsive tumors. (B) The same plot as in (A) is applied to CD4⁺ T cells. (C) Radar chart showing enrichment score of CD8 C2 signature for responsive and nonresponsive samples in each bulk transcriptomic cohort undergoing anti-PD-1/PD-L1 immunotherapy. The distance from the dots to the center of the circle represents the average enrichment score of each cohort. (D) The same plot as in (C) is applied to the CD4 C3 signature. In box plots, the centerline indicates the median value, lower and upper hinges represent the 25th and 75th percentiles, respectively, and whiskers denote the 1.5× interquartile range. The two-sided Wilcoxon rank-sum test was applied to calculate the P-values, followed by false discovery rate (FDR) correction for multiple comparisons. GSVA, gene set variation analysis; MELA, melanoma; RCC, renal cell carcinoma; UC, urothelial cancer.

To further validate the association between T cell subtype frequencies and immunotherapy efficacy, we collected bulk transcriptomic data from pretreatment tumor samples of 819 patients across six cohorts. These patients received anti-PD-1/PD-L1 immunotherapy and had available response statuses. We calculated scores for the CD8 C2 and CD4 C3 signatures in each tumor sample. While not entirely consistent with the results of single-cell transcriptomic data, responsive samples generally exhibited higher CD8 C2 scores and lower CD4 C3 scores across most cohorts (Fig. 3C, D).

In conclusion, this study underscores the importance of heterogeneity in tumor-infiltrating T cells as a key factor influencing immunotherapy outcomes. We integrated single-cell transcriptomic data from a multicancer cohort of 61 patients undergoing anti-PD-1/PD-L1 immunotherapy, revealing significant correlations between specific T-cell subtypes and therapeutic outcomes. These findings have important clinical implications, suggesting that T cell subtypes such as CD8⁺ T_N and CD4⁺ T_EX could serve as biomarkers for patient stratification and personalized treatment approaches. Clinically, this could be achieved through techniques like flow cytometry or immunohistochemistry to identify and quantify these subtypes in tumor biopsies. Additionally, therapeutic strategies could involve enhancing CD8⁺ T_N expansion or inhibiting CD4⁺ T_EX function through combination therapies, potentially improving immunotherapy efficacy in a more targeted manner. However, the study also faces limitations, primarily due to the small number of patients and cancer types included, which made subgroup analysis unfeasible. This limits the generalizability of the results. Future studies should validate these findings in larger, pan-cancer cohorts, and explore the use of additional datasets, such as spatial transcriptomics or multiomics data, to further explore the interplay between T-cell subtypes and other components of the tumor microenvironment. Expanding the analysis to include additional cancer types and integrating real-world clinical outcomes would offer clearer directions for future research to build upon and validate these results. Continued research in this area is essential to refine our understanding of T cell subtypes and their clinical applications, ultimately improving immunotherapy efficacy by targeting specific subtypes.

Ethical approval

As all data is publicly accessible, institutional review board approval was not required.

Consent

Consent is not applicable in this study.

Source of funding

This study was supported by National Natural Science Foundation of China (No. 82303772, No. 82303773, and No. 82204490), Postdoctor Research Fund of West China Hospital, Sichuan University (No. 2024HXBH006, No. 2024HXBH149), Natural Science Foundation of Sichuan Province (No. 2024NSFSC1908), Key Research and Development Program of Sichuan Province (No. 23ZDYF2836).

Author contribution

K.K.: writing – original draft, writing – review and editing, conceptualization, visualization, investigation, and funding acquisition; C.H. and L.W.: writing – original draft, writing – review and editing, visualization, and investigation; A.Z. and Y.W.: writing – review and editing, conceptualization, funding acquisition, and supervision.

Conflicts of interest disclosures

The authors declare no conflicts of interest.

Research registration unique identifying number (UIN)

1. Name of the registry: not applicable.

2. Unique identifying number or registration ID: not applicable.

3. Hyperlink to your specific registration (must be publicly accessible and will be checked): not applicable.

Guarantor

Yijun Wu.

Data availability statement

All data used in this study are publicly available as described in the Materials and Methods section in the Supplemental Files. Processed data were deposited on figshare (https://doi.org/10.6084/m9.figshare.25751100). Other further inquiries can be directed to the corresponding author.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Supplementary Material

js9-111-1642-s001.pdf ^{(858.9KB, pdf)}