Detection of spontaneous anti-neoepitope T-cell responses in non-metastatic bladder cancer patients

Abstract

Background: Bladder carcinomas are immunogenic, and patients with bladder cancer benefit from immune checkpoint therapy. This is correlated to a high tumor mutation burden, which provides a higher number of neoepitopes that can be recognized by tumor-specific CD8⁺ T cells. Intravesical Bacillus Calmette-Guérin (BCG) is used to treat non-muscle invasive bladder cancer (NMIBC), but its mechanism of action remains elusive. Most lymphocytes appearing in the urine of BCG-treated patients are CD4⁺ T cells though preclinical studies showed that CD8⁺ T cells are also necessary for BCG treatment efficacy. It is currently unknown which proportion of patients with non-metastatic bladder cancer develop a spontaneous antitumor CD8⁺ response, and if BCG treatment influences this response.

Methods: In a first cohort of 15 NMIBC and 9 muscle invasive bladder cancer patients, we used IFN-y ELISPOT assays to screen for the presence of anti-neoepitope CD8⁺ T cells in the blood, tumor and urine. In a second cohort of 4 NMIBC patients, we analyzed the features and specificity of CD8⁺ T cells infiltrating the tumoral or bladder tissues before and after BCG using single cell transcriptomic analyses. A total of 31 tumor-infiltrating CD8⁺ clonotypes were screened against neoepitopes and tumor cDNA libraries.

Results: 9 out of 24 patients from the first cohort mounted a spontaneous and functional anti-neoepitope T-cell response in blood and/or tumor. In 5 patients from this cohort who were treated with BCG, no neoantigen-specific T cells were detected in urine during treatment. In the second cohort, 6 out of 6 TCRs from exhausted CD8⁺ TILs from one patient recognized 5 different neoepitopes. T-cell receptor (TCR) repertoire analyses indicated that the frequencies of these tumor-specific T cells did not increase after BCG instillations, neither in the bladder nor in the blood. None of the 25 other TCRs of CD8⁺ T cells recognized tumor-specific antigens.

Conclusions: We show that one third of patients with non-metastatic bladder cancer mount a spontaneous and functional anti-neoepitope CD8⁺ T-cell response detectable in blood or tumor. In 4 patients with NMIBC, BCG treatment did not boost or induce the anti-neoepitope response, suggesting alternative mechanisms of action for its efficacy.

Keywords: BCG treatment; T lymphocyte; bladder cancer; immunotherapy; neoepitope.

Figure 1

Identification of neoepitope-specific T cells in TILs and PBMCs from bladder cancer patients. (A) Representative examples of deconvolution of peptide pools in TILs or PBMCs from two patients (URO483 and URO291). (B) Neoepitope repertoire from the 9 patients with neoantigen reactivities. Each colored square represents one neoepitope for which CD8⁺ T-cell reactivity was observed either exclusively in PBMCs (blue), exclusively in TILs (red) or in both compartments (orange). (C) CD8⁺ T-cell reactivity from TILs or PBMCs measured by IFN-y ELISpot against neoepitope (red dot) and wild-type (WT) non-mutated peptide (blue dot). (D) CD107a, IFN-y, TNF-α and IL-2 production measured by intracellular labelling upon neoepitope stimulation by TILs from 6 patients (one color per patient). (E) Total single-nucleotide variant (SNV)-mutational burden in patients with or without detectable neoepitope-specific CD8⁺ T-cell responses in TILs or PBMCs. Indicated P values were determined by Wilcoxon t-test (C, E) and one-way ANOVA (D). Significant differences between paired groups are indicated by *p < 0.05 and **p < 0.01. PP, peptide pool.

Figure 2

Transcriptomic profiles of CD8⁺ T cells extracted from the bladder of patients UC1 to UC4. (A) T-distributed stochastic neighbor embedding (t-SNE) projection of Smart-Seq2-based CD8⁺ single-cell data from 2040 CD8⁺ T cells extracted from the tumors and adjacent bladder tissues collected during TURBT, or from the same bladder regions 6 weeks after the last BCG instillation. (B) Heatmap of CD8⁺ T-cell clusters from (A) showing representative genes for each cluster. (C) Representations of (A) restricted to the indicated cell populations.

Figure 3

Identification of mutant peptides recognized by TCRs of CD8⁺ TILs from patient UC1. (A) On the left, TCR repertoire (48 TCR for 222 single T cells) of the ‘exhausted’ cluster shown on (Figure 2), ordered according to TCR frequencies. Six TCRs (#19, #49, #31, #10, #24 and #8), which were increased in tumor tissues compared to non-tumor tissues and blood, were screened for recognition of tumor antigens. (B) Example of peptide screening. Reporter Jurkat-2D3 expressing TCR #24 (15,000 cells/well) were incubated overnight with 30,000 autologous EBV-B cells pulsed with one of the indicated peptides. NFAT-GFP reporter gene activation was assessed with flow cytometry. (C) Example of cDNA library screening.

Figure 4

Frequencies of CD8⁺ T cell clonotypes in bladder and blood, pre- and post-BCG (A) Frequencies of TCR clonotypes were established on the basis of the TRA and TRB sequences identified in the single cell RNA-Seq data. Only clonotypes detected in at least two cells are shown. Those with statistically significant differences between pre-BCG and post-BCG frequencies are indicated in red (Fisher’s exact test, P < 0.05). Clonotypes identified with a number were screened for recognition of mutant peptides and of tumor cDNA library. Lowest frequency thresholds are indicated by dotted lines. For patient UC1 the TCR clonotypes with a demonstrated tumor-specificity are boxed. (B) Frequencies of the same TCR clonotypes in blood before and after therapy, established on the basis of TRB repertoires on blood DNA (Adaptive Biotechnologies).