All procedures performed in this study involving human participants were in accordance with the ethical standards of the Institutional Review Board of Beijing Shijitan Hospital, Capital Medical University (sjtkyll-1x-2021(108)), as well as the 1964 Helsinki Declaration and its later amendments or comparable ethical standards (16). Given the retrospective and de-identified nature of the study, the aforementioned Institutional Review Board waived the requirement for written informed consent (16).

This study was a retrospective cohort study. 227 female patients with a pathological diagnosis of primary BC were recruited from the Department of Breast Surgery, Beijing Shijitan Hospital, Capital Medical University from 2010 to 2018. Inclusion criteria: ① Patients had no diagnosis of pregnancy, lactation, or other malignancies; ②Patients had no experience of neoadjuvant chemotherapy or target therapy; ③Patients under 75 years of age. Exclusion criteria: ①Patients previously received any form of immunotherapy; ②Patients had a diagnosis of autoimmune disease; ③Patients had an Eastern Cooperative Oncology Group (ECOG) score >2; ④Patients had dysfunction of the heart, brain, kidneys, and other vital organs.

The clinical factors included age, pathological diagnosis, estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER-2) status, tumor histological grade, tumor stage, Ki-67 status, PD-1 and PD-L1 expression status.

The positive expression threshold of ER and PR in immunohistochemistry (IHC) detection was set to 1% TC staining. IHC tests with 3+ staining or positive results in fluorescence in situ hybridization (FISH) indicated positive HER-2 expression. By contrast, IHC tests with less than 2+ staining or negative FISH results showed negative expression. Patients with 2+ staining in IHC tests were required to undergo FISH testing. Ki-67 expression was defined as a brown nucleus in BC cells by IHC on 4 μm-thick formalin fixed paraffin-embedded sections. Meanwhile, the Ki-67 index was calculated as the proportion of BC cells expressing Ki-67 within the hot-spot area. The hot-spot area was determined under a low-power field, and an index≥14% was defined as a high expression of Ki-67. Molecular subtypes were defined as Luminal A (HER-2 negative, PR/ER positive, Ki-67 low expression), Luminal B (HER-2 negative, PR/ER positive, Ki-67 high expression), TNBC (HER-2 negative, ER negative, PR-negative, Ki-67 arbitrary), and HER-2 overexpression (HER-2 positive, Ki-67 arbitrary). Tumor histological grading was performed using the Nottingham grading system, integrating the proportion of gland formation, nuclear pleomorphism, and mitotic count to determine the overall grade of the tumor. The score ranges for grades I, II, and III were 3-5, 6-7, and 8-9, respectively. Tumor staging was performed using the tumor node metastasis (TNM) classification system following the guidelines of the 8th edition of the American Joint Committee on Cancer. TNM stage was classified as I (T₁N₀M₀, T_0~1N_1miM₀), II (T_0~1N₁M₀, T₂N_0~1M₀, T₃N₀M₀), III (T_0~2N₂M₀, T₃N_1~2M₀, T₄N_0~2M₀, T_0~4N₃M₀), and IV(T_0~4N_0~3M₁).

Cancer recurrence and death were study outcomes. The follow-up interval was set as six months and data was collected from clinic visits, telephonic interviews, as well as hospital records. The diagnosis of BC recurrence relied on biopsy, bone scanning, as well as CT/MRI. Information about all-cause deaths was gathered from both patients and their caregivers. Disease-free survival (DFS) was defined as the period from surgery to cancer recurrence or death. Overall survival (OS) was defined as the interval from surgery to death.

The expression of CD155, TIGIT, CD96, and CD226 on the membrane of TC and stromal TILs was detected by IHC using the EnVision two-step method. Stromal area was demarcated as the region falling within the boundaries of the invasive tumor. Areas featuring crush artifacts, necrosis, regressive hyalinization, and the biopsy site were excluded from this definition. Scored cells comprised mononuclear cells, specifically lymphocytes and plasma cells, while polymorphonuclear leukocytes were excluded. TILs were measured as the average counts in 10 random high-power fields (HPF, ×400) on IHC sections.

Monoclonal antibody against CD155 (rabbit anti-human, #81254S) were purchased from Cell Signaling Technology Co. Ltd. Monoclonal antibodies against CD226 (rabbit anti-human, #ab2120772) and TIGIT (rabbit anti-human, #ab243903) were purchased from Abcam Co. Ltd. A polyclonal antibody against CD96 (rabbit anti-human, #PA5-97568) was purchased from Invitrogen Co. Ltd. Monoclonal antibodies against PD-L1 (rabbit anti-human, #SP142) were purchased from Roche Shanghai Co. Ltd. Monoclonal antibody against PD-1 (mouse anti-human, # UMAB199) and secondary antibodies were purchased from Beijing Zhongshanjinqiao Biotechnology Co. Ltd.

Positive expression was recorded by brown staining of the cells. PD-L1 positive expression was denoted by the appearance of brown staining in the cytoplasm and/or cell membrane of both immune cells and TC. PD-1 positive expression was manifested as brown-stained cytoplasm within immune cells. Positive expression of CD155/TIGIT/CD96/CD226 on TC in the tumor microenvironment (TME) was defined as brown staining on the cytomembrane. Positive expression of TIGIT/CD96/CD226 in stromal TILs was defined as brown cytoplasmic staining. The expression of CD155, TIGIT, CD96, and CD226 on TC was evaluated by integrating staining intensity and the proportion of positive cells: the proportion of positive TC was categorized into 4 grade based on percentage (grade 0 = 0% positive cells, grade 1 = <1/3 positive cells, grade 2 = 1/3-2/3 positive cells, grade 3 = >2/3 positive cells), and staining intensity was scored as 0 (negative), 1 (weak and incomplete cytomembrane staining), 2 (weak and complete or strong and incomplete cytomembrane staining), or 3 (strong and complete cytomembrane staining); a composite score, with a total range of 0-9, was calculated by multiplying the proportion grade by the corresponding intensity score for all positive TC (i.e., composite score = proportion grade × intensity score).The percentage grade of positive TC indicated the proportion category of TC with molecular expression in the whole section, and the percentage of stromal TILs indicated the proportion of stromal TILs with positive molecule expression in the whole section. High expression of CD155, TIGIT, CD96, and CD226 was defined as a composite score of more than 3 for TC. Because the staining intensity on TILs could not be determined, high stromal expression was defined as more than 2/3 positive cells. Two pathologists estimated the IHC scoring, and a third, higher-level pathologist re-evaluated inconsistent estimations between the two pathologists.

R version 4.3.1 was used to conduct all statistical analyses. Differences between groups were analyzed using Pearson’s chi-square or Fisher’s exact test. The influence of missing data was eliminated during the analysis. Survival curves were plotted using the Kaplan-Meier method, and differences between groups were evaluated using the log-rank test. Cox proportional hazards regression models were established to control for confounding factors. The analysis steps were as follows: ①clinical factors related to prognosis and molecular expression were screened in univariate analysis; ②the selected relevant factors (p < 0.10) were adjusted in combination with CD155-TIGIT/CD96/CD226 molecular expression levels, and the relationship between each molecular expression and prognosis was estimated using the Cox multivariate model, with the estimation of hazard ratio (HR) and 95% Confidence interval (CI). All analyses were two-sided and the significance level was set at 0.05.

Both stromal TILs and TC expressed CD226, CD96, and TIGIT ( Figures 1A–C ), but only TC expressed CD155 ( Figure 1D ). Among them, 37.2%, 25.6%, and 17.1% of the patients had more than 2/3 of stromal TILs expressing CD226, CD96, and TIGIT, respectively, and 24.9%,26.9%,84.8% and 56.7% of the patients had high expression levels of CD155, CD226, CD96, and TIGIT in TC, respectively ( Table 1 ).

High CD226 expression in stromal TILs was associated with ER(p = 0.006), PR(p = 0.041), and PD-1(p = 0.012, Table 2 ). High CD226 expression in TC correlated with low HER-2 expression (p = 0.050, Table 2 ). More TNBC patients had high expression of CD96 in stromal TILs (p = 0.044, Table 2 ), which was significantly associated with PD-L1 expression in stromal TILs (p = 0.012, Table 2 ). High CD96 expression on TC correlated with increased TILs in the TME (p = 0.002, Table 2 ). High expression of TIGIT in stromal TILs was correlated with low PR (p = 0.025, Table 2 ) but high PD-L1 expression in stromal TILs (p = 0.014, Table 2 ). High TIGIT expression in TC was correlated with lower histological grade (p = 0.004, Table 2 ). High expression of CD155 was associated with higher TILs levels (p = 0.033, Table 2 ), higher proportion of histological grade (p = 0.023, Table 2 ), higher Ki-67 index (p < 0.001, Table 2 ), higher proportion of TNBC (p = 0.020, Table 2 ), and PD-L1 expression on stromal TILs (p = 0.040, Table 2 ).

The median follow-up was 10 years (95%CI:8.8-11.0). BC patients with low expression of CD155 had 10-year DFS and OS rates of 70.37% and 80.64%, respectively, which were significantly higher than those with high expression of CD155 (58.81% for DFS, p = 0.033; 58.24% for OS, p = 0.002, Figure 2 ). Among patients with low TME infiltration of stromal TILs, the 10-year DFS and OS rates differed among different biomarker expression groups. For CD155, the rates were 56.49% and 62.70% in the low-expression group and 25.00% (p = 0.019) and 25.00% (p = 0.013) in the high-expression group, respectively ( Figure 3 ). Regarding TC CD96, the low expression group of patients had rates of 78.57% and 78.57%, while the high expression group of patients had rates of 42.21% (p = 0.029) and 50.73% (p = 0.098), respectively ( Figure 3 ). Patients with low TIGIT expression on TC had 10-year DFS and OS rates of 38.89% and 42.11%, respectively, and the high expression group had 62.75% (p = 0.042) and 69.96% (p = 0.020), respectively ( Figure 3 ). In patients with high TME infiltration of stromal TILs, patients with low expression of CD155 had 10-year DFS and OS rates of 75.19% and 86.89%, respectively, compared with 61.12% (p = 0.091) and 62.26% (p = 0.008) among those with high expression of CD155, respectively ( Figure 4 ). Additionally, patients with low expression of CD226 on stromal TILs had 10-year DFS and OS rates of 59.23% and 74.38%, respectively, whereas the 10-year rates of DFS and OS were 83.04% (p = 0.057) and 89.58% (p = 0.120) among patients with high expression, respectively ( Figure 4 ).

In all BC patients, high expression of CD155 increased 2.21-fold the risk of relapse (95%CI:1.18-4.13, Figure 5A ) and a 2.57-fold higher risk of death (95%CI:1.29-5.10, Figure 6A ). High expression of TC CD226 (HR = 1.79, 95%CI: 1.03-3.11), TC CD96 (HR = 2.65, 95%CI:1.09-6.44), and TIGIT on stromal TILs (HR = 2.06, 95%CI:1.02-4.14) were associated with an increased risk of relapse ( Figure 5A ). However, high expression of TC TIGIT was associated with a 55% reduction in relapse risk (HR = 0.45, 95%CI:0.24-0.83, Figure 5A ) and a 68% reduction in death risk (HR = 0.32, 95%CI:0.16-0.66, Figure 6A ). Among patients with low TME infiltration of stromal TILs, high expression of CD155 was associated with poor DFS (HR = 4.39, 95%CI:1.17-16.51, Figure 5B ) and OS (HR = 5.18, 95%CI:1.32-20.28, Figure 6B ). High TC CD96 expression (HR = 3.50, 95%CI: 1.01-12.21, Figure 5B ) was associated with unfavorable DFS, while high TC TIGIT expression was associated with favorable DFS (HR = 0.31, 95%CI:0.12-0.80, Figure 5B ) and OS (HR = 0.36, 95%CI: 0.14-0.94, Figure 6B ). In patients with high TME infiltration of stromal TILs, high expression of CD155 was associated with a 2.86-fold higher risk of death (HR = 2.86, 95%CI:1.10-7.42, Figure 6C ) and high expression of TC CD226 was associated with a 2.29-fold high risk of relapse (HR = 2.29, 95%CI:1.10-4.76, Figure 5C ) and a 2.56-fold higher risk of death (HR = 2.56, 95%CI: 1.07-6.11, Figure 6C ), while high expression of CD226 on stromal TILs was associated with a favorable DFS (HR = 0.38, 95%CI:0.16-0.89, Figure 5C ) and OS (HR = 0.27, 95%CI:0.08-0.84, Figure 6C ).