Overexpression of B7H5/CD28H is associated with worse survival in human gastric cancer

Abstract Gastric cancer (GC) is a common malignancy with low 5‐year overall survival (OS). Recently, immune therapy has been used to treat cancer. B7H5 and CD28H are novel immune checkpoint molecules. However, the prognostic value of B7H5/CD28H expression in patients with GC remains unclear. In this study, seventy‐one patients diagnosed with GC were included in this study. Patients' GC tissues and matched adjacent tissue constructed a tissue microarray. The expression levels of B7H5 and CD28H were examined using immunohistochemistry. Correlations between the expression of B7H5 and CD28H and the clinical data were evaluated. We found that the expression of B7H5 and CD28H (both P = .001) were higher in GC tumour tissues than in adjacent noncancerous tissues. B7H5/CD28H expression acted as an independent predictive factor in the OS of patients with GC. High expression of B7H5 and CD28H predicted poor outcome. Patients in the B7H5+CD28H+ group had a lower 5‐year OS compared with patients in the B7H5−CD28− group (4.5% vs 55.6%, P = .001). A significant difference was found in the 5‐year OS between patients in the B7H5+CD28H− and B7H5+CD28H+ groups (33.5% vs 4.5%, P = .006). However, there was no correlation between B7H5 and CD28H expression (P = .844). Therefore, B7H5 and CD28H expression are up‐regulated in GC and are independent prognostic factors for overall survival in patients with GC. Although there was no correlation between B7H5 and CD28H expression, high expression of B7H5 and CD28H predicts poor prognosis, especially when both are highly expressed.


| INTRODUC TI ON
Gastric cancer (GC) is a common malignancy and has been estimated to account for one-third of cancer deaths. 1 Surgery is the optimal treatment for patients with GC, which provides the best chance of long-term survival. However, the early stage of the disease is often asymptomatic, and GC is frequently diagnosed in the later stages, when it is characterized by invasion or metastasis. The therapeutic options of neoadjuvant therapy, targeted drugs and immunotherapy have opened a new field of cancer treatment in recent years; however, the 5-year overall survival (OS) rate of GC is unsatisfactory. 2 While cancers suppress the immune response by establishing a microenvironment, the immune cells decrease cancer immunogenicity in the process of tumorigenesis and cancer progression. 3 Therefore, the interaction between the microenvironment and the immune system is a new direction in GC research. 4 However, patients with GC have not benefited from immunotherapy.
Immune checkpoint therapy is a novel treatment that targets regulatory pathways in T cells to enhance antitumour immune responses.
The B7/CD28 family of ligands and receptors play an important role in cancer pathogenesis. Ten members are included in the B7/CD28 family, including B7-1/CD28, B7-2/cytotoxic T lymphocyte antigen 4 (CTLA-4), ICOS-L/ICOS and programmed cell death 1 (PD-1)/PD-1 ligand (PD-L1). 5 PD-L1/PD-1 has become a hot area of research in recently years. 6,7 The B7/CD28 family regulates the proliferation and function of T cells as antigen-presenting cells (APCs). In addition, some tumours can evade immune elimination, because of overexpression of PD-L1/PD-1. 8,9 Inhibitors of PD-L1/PD-1 have been approved by Food and Drug Administration (FDA) as second-line treatments for lung cancer, because of their therapeutic benefit in clinical trials.
The B7H5/CD28H pathway is a novel receptor-ligand interaction in the B7/CD28 family that can reduce the expression of interleukin-5 (IL-5), IL-13, IL-10 and tumour necrosis factor gamma (TNF-γ) and suppress the activity of CD4+ and CD8+ T cells. 10 Some studies showed low expression of B7H5 in normal tissue, 11 while B7H5 is overexpressed in some cancers. 12,13 However, the prognostic value of B7H5/CD28H in patients with GC is controversial. In the present study, we aimed to evaluate the expression levels of B7H5 and CD28H in patients with GC.

| Tissue microarray (TMA) construction and immunohistochemistry (IHC) analysis
Seventy-one formalin-fixed, paraffin-embedded (FFPE) GC tissues and corresponding adjacent noncancerous tissues were collected from the Department of Abdominal Surgery, Zhejiang Cancer Hospital. All FFPE GC tissues were screened by two pathologists independently to confirm the diagnosis of GC. The most representative tumour and noncancerous tissues were selected to construct the TMA slide. Seventy-one paired GC tissues and matched noncancerous tissues were included in the TMA.
The IHC technique has been described in detail previously. 14 TMA sections were deparaffinized, and hydrogen peroxide (3%) was applied to repair the antigen. Nonspecific staining was blocked using 10% goat serum at room temperature for 30 minutes.
Immunostaining of histological sections was performed using mono- Cytoplasmic expression was assessed using the H-score sys-

| Post-operative evaluation and follow-up
Post-operative chemotherapy was performed according to each patient's individual pathological diagnosis. If an infiltrating margin was found, adjuvant irradiation was necessary. The patients were examined regularly every 3 months during the first year and every 6 months thereafter.

| Statistical analysis
Statistical analysis was performed using IBM SPSS statistics for

| B7H5 expression in cytoplasm in GC and noncancerous tissues
B7H5 protein levels in GC tissues are currently unknown. As shown in Figure 1, B7H5 was expressed in the cytoplasm in the noncancerous and GC tissues. However, the staining intensity of B7H5 in the cytoplasm in GC tissues was obviously stronger than that in noncancerous tissues ( Figure 2A). The median H-score of B7H5 was 8 (range: 1-12).
The median score was used to determine the cut-off value of low-or high-level B7H5 expression. An H-score <8.0 was defined as low B7H5

| B7H5 expression in the nucleus in GC and noncancerous tissues
B7H5 was expressed in the nucleus in the noncancerous tissues and GC tissues. However, the staining intensity of B7H5 in the nuclei in GC tissues was obviously stronger than that in noncancerous tissues ( Figure 2B).

| CD28H + T cells exist in GC and noncancerous tissues
CD28H protein levels in GC tissues are currently unknown. As shown in Figure 3, the CD28H protein was expressed in both GC and noncancerous tissues. In the GC tissues and noncancerous tissues, CD28H was expressed in immune cells. However, the staining F I G U R E 2 B7H5 protein was expressed in both gastric carcinoma and noncancerous tissues. A, B7H5 expression was higher in cytoplasm in GC tissues than in gastric tissue (P = .001); B, B7H5 expression was higher in the nucleus in GC tissues than in gastric tissue (P < .001); C, CD28h expression was higher in immune cells in GC tissues than in gastric tissue (P = .001); D, There was no correlation between B7H5 and CD28H expression (P = .844) intensity of CD28H in GC tissues was obviously stronger than that in noncancerous tissues. The median of H-score of CD28H was 8 expression on CD3 + T cells in GC tissues and gastric tissues using flow cytometry. As shown in Figure 4, the level of CD3 + T cells in GC tissues was higher than that in gastric tissues (P < .01), while the level of CD3 + CD28H + T cells was higher in GC than in gastric tissues (P < .01).

| B7H5 and CD28H expression in GC and clinicopathological variables
The association between B7H5 and CD28H expression and clinicopathological parameters in GC was investigated using the chi-square test. As listed in Table 1

| Prognostic value of B7H5 and CD28H expression in patients with GC
As shown in Figure 5, the prognostic value of B7H5 expression in There was a significant difference between B7H5 + CD28H − and B7H5 + CD28H + patients (P = .006), while there is no difference between B7H5 + CD28H − and B7H5 − CD28H + patients (P = .777).
In addition, there was no difference between B7H5 − CD28H + and B7H5 + CD28H + patients (P = .066). However, the results of the chisquare test confirmed that there was no correlation between B7H5 expression and CD28H expression in patients with GC (P = .844).

F I G U R E 4 CD28H is expression on T cells in GC tissues and gastric tissues. A-B, Representative flow cytometry plots
showing the count of CD3 + T cells in GC tissues and gastric tissues. C, The mean percentages showed that the count of CD3 + T cells in GC tissues was higher than that than gastric tissues, *, P < .05. D-E, Representative flow cytometry plots showing the count of CD3 + CD28H + T cells in CD3 + T cells in GC tissues and gastric tissues. F, The mean percentages showed that the count of CD3 + CD28H + T cells in CD3 + T cells in GC tissues was higher than that in gastric tissues, *, P < .05 blood between 111 patients with untreated GC and 20 healthy volunteers. Their study showed that patients with GC had significantly higher B7H5 mRNA levels and higher B7H5 expression was associated with a better 5-year OS. This result revealed a different B7H5 expression pattern to that shown in the present study. We are unable to explain this difference because of the different methods and antibodies used to detect B7H5 expression between the two studies.
However, we also showed that the expression of B7H5 was almost absent in B7H5 KO -BGC803 group in vivo. And it can also indicate the specificity of B7H5 (as showed in Figure S1). Other studies confirmed that high B7H5 expression was associated with poor prognosis in cer- B7H5 has two receptors on T cells, including CD28H and another, as yet unknown, receptor. B7H5 has co-stimulatory and co-inhibitory effects against the immune response of T cells by CD28H and the unknown receptor. 10 Therefore, we also examined the expression of CD28H. We found that the level of CD28H + T cells in the tumour tissues in patients with GC was higher than that in the adjacent noncancerous tissues.
Furthermore, patients in the B7H5 + CD28H + group had a lower 5-year OS compared with patients in the B7H5 − CD28H − group (P = .001). However,  10 Therefore, the interaction of B7H5 and CD28H may inhibit the immune response as a co-inhibitor in GC.
In conclusion, we confirmed that B7H5 and CD28H expression levels are up-regulated and predict low survival in patients with GC, and are independent prognostic factors of overall survival. Although there is no correlation between B7H5 and CD28H expression, high expression of B7H5 and CD28H predicts poor prognosis, especially F I G U R E 5 Correlation of B7H5/CD28H expression level and overall survival (OS) in GC patients. A, There is no difference in 5-year OS between high B7H5 expression and low expression in the nuclei in patients with GC (28.0% vs 21.6%, P = .254); B, High B7H5 expression predicted poorer survival than low B7H5 expression in patients with GC (19.6% vs 37.5%, P = .035); C, High CD28H expression predicted poorer survival than low CD28H expression in patients with GC (39.4% vs 6.9%, P = .002); D, Patients in the B7H5 + CD28H + group have a lower 5-year OS compared with patients in the B7H5 − CD28 − group (4.5% vs 55.6%, P = .001), a significant difference was found in the 5-year OS between patients in B7H5 + CD28H − and B7H5 + CD28H + groups (33.5% vs 4.5%, P = .006), and there was no significant difference between the B7H5 − CD28 − group and the B7H5 − CD28H + group (55.6% vs 33.3%, P = .111) when both are highly expressed, via inhibition of the immune response of T cells. Therefore, the B7H5/CD28H axis could be an attractive target for GC immunotherapy.

ACK N OWLED G EM ENTS
We would like to thank Yuwen Zhu in Department of Immunobiology, Yale University school of Medicine for donating of antibodies.

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.

E TH I C A L A PPROVA L
The study was approved by the ethics committee of Zhejiang Cancer Hospital. The study conformed to the tenets of the Declaration of Helsinki. All patients provided written informed consent before taking part in the study.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author upon request. Abbreviations: CI, confidence interval; HR, hazard ratio. a Statistically significant (P < .05).