This study enrolled two independent patient cohorts, including 393 patients with bladder cancer who were treated with radical cystectomy (RC) at Zhongshan Hospital of Fudan University from 2008 to 2012 (ZSHS cohort, n=215) and Fudan University Shanghai Cancer Center from 2002 to 2014 (FUSCC cohort, n=178). A total of 132 patients were excluded: 95 patients without MIBC, 19 patients without urothelial carcinoma, and 18 patients with unavailabe clinical or follow-up data. Because of the immunohistochemistry (IHC) detachment, a specimen was lost on the TMA in each cohort. Therefore, 259 eligible patients with MIBC were included (ZSHS cohort, n=141; FUSCC cohort, n=118). There were 119 patients of the two cohorts who received adjuvant cisplatin-based chemotherapy and lasted at least one therapeutic cycle. Patients received follow-up every 3 months in the first year, every 6 months for 2 years and once per year afterwards, which included clinical history, physical examination and laboratory test. All follow-up data were collected until July 2016. The overall survival (OS) and the recurrence-free survival (RFS) were defined the time from the date of RC to the date of death and the first recurrence, or to the last follow-up.

IHC staining was performed on formalin-fixed, paraffin-embedded tissue microarray (TMA) as described previously.22 The IHC antibodies are listed in online supplementary table 1. In brief, the slides were baked at 60°C for 6 hours, deparaffinized in xylene (three times, 15 min each) and rehydrated in graded alcohol. Next, the slides were immersed in sodium citrate buffer (0.01 M sodium citrate buffer, pH=6) for antigen retrieval and then blocked with 3% H₂O₂ in methanol at 37°C for 30 min. For single IHC staining, the slides were incubated with the primary antibodies at 4°C overnight and visualized by 3,3′-diaminobenzidine (DAB) stain system. For double IHC staining, after being processed as the same of single IHC DAB staining, the slides were incubated with the second primary antibodies at 4°C for 2 hours, and then Vector Blue AP Substrate Kit (Vector Laboratories) was applied. All TMA slides were evaluated under Leica DM6000 B Microsystems by PZ and LC independently, who were blinded to clinical data. The positive cells were enumerated from the representative view of the three sections in high-power field (HPF, ×200 magnification), and the mean value was adopted. The cut-off value was determined by X-tile V.3.6.1 (Yale University). For CD8⁺ T-cells, the cut-off value was 34 cells/HPF. For TIGIT⁺ CD8⁺ cells, the cut-off value was 8 cells/HPF.

Fresh samples, including tumor tissues (n=26) and peritumor tissues (n=13), were collected from five different clinical centers (Zhongshan Hospital of Fudan University, Fudan University Shanghai Cancer Center, Ruijin Hospital, Shanghai General Hospital and Shanghai Ninth People’s Hospital). The peritumor tissues are obtained from an area of ≥2 cm from the tumor margin.

Single-cell suspension was performed as described.23 Then samples were stained with the indicated mAbs for 30 min at 4°C after lysing red blood cells. Cells were stimulated for 5 hours with phorbol myristate acetate (50 ng/mL) and ionomycin (1 µg/mL) in the presence of GolgiStop protein transport inhibitor (1:1000) for intracellular cytokine measurement. Cells were stained with interested surface markers, and Fixation/Permeabilization Solution Kit (BD Biosciences) was used for intracellular protein staining according to the manufacturer’s instructions. Stained cells were washed and resuspended in phosphate-buffered saline/0.1% bovine serum albumin coupled with azide. Flow cytometry data were analyzed by FlowJo software (Tree Star, San Carlos, California, USA). All flow cytometry antibodies are listed in online supplementary table 2.

Descriptive statistics was used to summarize patients’ baseline characteristics and disease factors. Results are shown as mean±SD, and Mann-Whiney U test, Wilcoxon signed-rank test, χ² test and Spearman correlation analysis were used in this study. OS and RFS were determined by the Kaplan-Meier method, which was evaluated by log-rank tests. Multivariate analyses of the Cox regression model were applied to estimate HRs and 95% CIs. A two-tailed p value of <0.05 was considered statistically significant in our study. All statistical analyses were conducted using IBM SPSS Statistics V.25.0, R V.3.5.1, GraphPad Prism Software V.8.0 and MedCalc V.15.

The residency of TIGIT⁺ CD8⁺ T-cells in MIBC was identified through double IHC staining and flow cytometry (figure 1A,B and online supplementary figure 1). The patient characteristics are listed in online supplementary table 3. We found the proportion of patients with high TIGIT⁺ CD8⁺ T-cell infiltration expanded with progression of pathology Tumor (pT) or pathology Node (pN) stage (figure 1C). Moreover, the infiltration of TIGIT⁺ CD8⁺ T-cells, as well as the proportion of TIGIT⁺ CD8⁺ T-cell among CD8⁺ T-cells, was positively associated with the increased tumor stages (figure 1D and online supplementary figure 2). Furthermore, compared with peritumor tissues, we found that TIGIT⁺ CD8⁺ T-cells were dramatically more infiltrated in matched tumor tissues (figure 1E). In conclusion, these results confirmed the existence of TIGIT⁺ CD8⁺ T-cells in MIBC and indicated that intratumoral TIGIT⁺ CD8⁺ T-cells were correlated with MIBC progression.

The prognostic ability of TIGIT⁺ CD8⁺ T-cells was next explored. We found that CD8⁺ T-cell infiltration could prolong the OS of patients with MIBC both in the ZSHS cohort and the FUSCC cohort (p=0.023 and p=0.007, figure 2A). However, the RFS showed no difference with CD8⁺ T-cell strata (p=0.859 and p=0.142, figure 2B). Interestingly, the TIGIT⁺ CD8⁺ cells high infiltration group possessed inferior OS and RFS than the TIGIT⁺ CD8⁺ cells low infiltration group in both cohorts (OS: p=0.010 and p=0.013, figure 2C; RFS: p=0.009 and p=0.047, figure 2D), which was entirely contrary to the prognosis of CD8⁺ T-cell infiltration. Additionally, multivariate Cox regression analysis showed that TIGIT⁺ CD8⁺ T-cell infiltration was an independent prognostic factor for patients with MIBC after adjustment for age, gender, grade, lymphovascular invasion, adjuvant chemotherapy (ACT), pathological T/N stage and CD8⁺ T-cell infiltration as confounders (online supplementary table 4). Therefore, TIGIT⁺ CD8⁺ T-cells proved to be an independent unfavorable factor to predict survival and recurrence in patients with MIBC.

ACT has been widely used in the treatment for patients with MIBC with pT3/4 or pN+ disease.24 Thus, we evaluated whether intratumoral TIGIT⁺ CD8⁺ T-cell abundance was associated with ACT responsiveness. Both cohorts were combined for further investigation. In all patients or patients with stage III MIBC, intratumoral TIGIT⁺ CD8⁺ T-cell infiltration failed to predict ACT responsiveness (online supplementary figure 3). No evidence suggested that ACT could improve OS or RFS in patients with stage II MIBC (OS: p=0.087, RFS: p=0.128; figure 3A,D). CD8⁺ T-cells still failed to predict ACT effectiveness in patients with stage II MIBC (online supplementary figure 4). However, after dividing patients into TIGIT⁺ CD8⁺ T-cells high and low infiltration subgroups, we found that ACT successfully prolonged both OS and RFS in patients with low TIGIT⁺ CD8⁺ T-cell infiltration (OS: p=0.043, RFS: p=0.013; figure 3C,F), while no survival benefit was observed in patients with high TIGIT⁺ CD8⁺ T-cell infiltration (OS: p=0.261, RFS: p=0.769; figure 3B,F). In addition, univariate Cox regression analysis was performed to assess the relationship between TIGIT⁺ CD8⁺ T-cell infiltration and ACT benefit, which suggested that patients with stage II MIBC with TIGIT⁺ CD8⁺ T-cells low infiltration could receive more benefit from ACT (OS: HR, 0.340, 95% CI, 0.114 to 0.950, p=0.049; RFS: HR, 0.239, 95% CI, 0.070 to 0.817, p=0.022; online supplementary table 5). Herein, these results suggested that TIGIT⁺ CD8⁺ T-cells high abundance predicted suboptimal ACT responsiveness within patients with stage II MIBC. Patients with low TIGIT⁺ CD8⁺ T-cell infiltration could benefit more from ACT.

Coinhibitory receptors, also known as immune checkpoints, are often coexpressed on dysfunctional CD8⁺ T-cells.25 Compared with TIGIT^- CD8⁺ T-cells, TIGIT⁺ CD8⁺ T-cells expressed higher levels of immune checkpoints, including PD-1, CTLA-4, Lag-3 and Tim-3 (online supplementary figure 5A). Meanwhile, TIGIT⁺ CD8⁺ T-cells had a higher level of proliferative marker (Ki-67), effector markers (interferon (IFN)-γ, tumor necrosis factor (TNF)-α and interleukin (IL)-2) and cytolytic marker (CD107a) compared with their TIGIT⁻ counterparts (online supplementary figure 5B–D), which inferred that TIGIT⁺ CD8⁺ T-cells could be closely related to a terminally exhausted CD8⁺ T-cell phenotypes as previously reported.26

The global characterization of CD8⁺ T-cells was subsequently investigated according to TIGIT⁺ CD8⁺ T-cell abundance. We found that CD8⁺ T-cells in TIGIT⁺ CD8⁺ T-cells high infiltration tumors expressed increased immune checkpoints, including PD-1, CTLA-4 and Lag-3 (figure 4A) while exhibiting more proliferative ability (Ki-67) than their counterparts in TIGIT⁺ CD8⁺ T-cells low infiltration tumors (figure 4B). Additionally, the effector cytokines (IFN-γ and TNF-α) and cytotoxicity activation molecules (granzyme B (GZMB)) expressed by CD8⁺ T-cells were decreased in TIGIT⁺ CD8⁺ T-cells high infiltration tumors (figure 4C,D). These results indicated that high TIGIT⁺ CD8⁺ T-cells were associated with impaired CD8⁺ T-cell antitumor immunity. In addition to CD8⁺ T-cells, CD45⁺ T-cells in TIGIT⁺ CD8⁺ T-cells high infiltration tumors expressed increased immune checkpoints, including PD-1, CTLA-4 and Tim-3, while the percentage of CD45⁺ T-cells expressing effector cytokines (IFN-γ and TNF-α) decreased in TIGIT⁺ CD8⁺ T-cells high infiltration tumors. (online supplementary figure 6). Interestingly, we found that TIGIT⁺ CD8⁺ T-cells could produce immunosuppressive cytokine IL-10 (online supplementary figure 5E). TIGIT⁺ CD8⁺ T-cell infiltration was also found positively associated with intratumoral IL-10⁺ CD8⁺ T-cells and IL-10 expression in the tumor microenvironment (figure 4E,F). These results preliminarily verified our conjecture that intratumoral TIGIT⁺ CD8⁺ T-cell abundance may contribute to immune suppression and dampen CD8⁺ T-cell immune response in MIBC.

Next, the association between immune contexture and TIGIT⁺ CD8⁺ T-cell abundance was explored in MIBC. Eleven types of immune cells were evaluated inthe ZSHS cohort (figure 5A, n=141). The representative images of immune cells are illustrated in online supplementary figure 7. TIGIT⁺ CD8⁺ T-cell infiltration was positively correlated with protumor Th2 cells (Spearman r=0.317, p<0.001), Tregs (Spearman r=0.309, p<0.001), mast cells (Spearman r=0.334, p<0.001), neutrophils (Spearman r=0.178, p=0.035) infiltration and antitumor NK cells (Spearman r=0.183, p=0.029), and M1 macrophages (Spearman r=0.251, p=0.003) infiltration (figure 5B,C). These data indicated that intratumoral TIGIT⁺ CD8⁺ T-cell abundance indicated tumor-promoting immune microenvironment in MIBC.