Prognostic value of HHLA2 expression in solid tumors

Abstract Background: Human endogenous retrovirus-H long terminal repeat-associating protein 2 (HHLA2), a newly discovered member of the B7 family, is overexpressed in numerous tumors. However, the prognostic impact of HHLA2 in human cancers remains controversial. Thus, we performed this meta-analysis to explore the prognostic value of HHLA2 in Chinese patients with solid tumors. Methods: PubMed, Embase, Web of Science, Chinese National Knowledge Infrastructure, and WanFang databases were systematically searched for eligible studies that evaluated the impact of HHLA2 on overall survival (OS) in patients with cancer. Hazard ratios (HRs) and 95% confidence intervals (CIs) were combined to evaluate the association between HHLA2 expression and OS in solid tumors. Odds ratios (ORs) and 95% CIs were pooled to assess the correlation between HHLA2 expression and clinicopathological characteristics in solid tumors. Results: A total of 12 studies, including 15 cohorts and 1747 patients, were included in this meta-analysis. We found that high HHLA2 expression was significantly associated with shorter OS (HR = 1.65, 95% CI: 1.12–2.43). Subgroup analysis by cancer type demonstrated that high HHLA2 expression was associated with poor OS in patients with clear cell renal cell carcinoma (HR = 3.42, 95% CI: 2.39–4.91), gastric cancer (HR = 2.03, 95% CI: 1.31–3.16), intrahepatic cholangiocarcinoma (HR = 1.77, 95% CI: 1.24–2.53), lung cancer (HR = 2.14, 95% CI: 1.33–3.44) and other cancer types (HR = 2.08, 95% CI: 1.34–3.24), but not in patients with epithelial ovarian cancer (HR = 0.52, 95% CI: 0.08–3.56). Nevertheless, high HHLA2 expression was associated with better OS in patients with pancreatic ductal adenocarcinoma (HR = 0.45, 95% CI: 0.32–0.64). Furthermore, high HHLA2 expression was associated with old age (OR = 1.30, 95% CI: 1.03–1.63), lymph node metastasis (OR = 1.99, 95% CI: 1.41–2.81), and vascular invasion (OR = 1.69, 95% CI: 1.18–2.42). Conclusions: HHLA2 may serve as a potential prognostic biomarker for solid tumors in Chinese population, by predict the prognosis of cancer patients based on their tumor types.


Introduction
Malignant tumors have a high incidence and mortality rate, and are the major public health problem worldwide, leading to heavy medical and socioeconomic burdens. [1] Despite substantial advances in surgical, radio-and chemo-therapies in recent decades, poor tumor prognosis remains a tremendous challenge for researchers and clinicians. [2] As a Consequence, the development of novel therapeutics for solid tumors is needed.
Cancer immunotherapy is a rapidly developing cancer treatment designed to stimulate the immune system to fight and eliminate tumors. [3,4] Among various immunotherapy strategies, immune checkpoint blockade seems to be an effective approach. [5] Immune checkpoint proteins are surface molecules on certain immune cell populations that activate or inhibit immune function when engaged to their ligands. [5] Several checkpoint inhibitors have been extensively studied for cancer treatment, and some have been used in clinical trials in cancer patients, such as antibodies against cytotoxic T lymphocyteassociated antigen-4 (CTLA-4), programmed cell death protein 1 (PD-1), or its ligand PD-L1. [6][7][8] However, the response rate of anti-PD-1 and anti-CTLA4 antibody therapy is rather low in some types of cancer. [9][10][11][12] Hence, the development of other immune checkpoint molecules for immunotherapy to improve the survival rate of cancer patients is urgently required.

Materials and methods
This meta-analysis was performed in accordance with the preferred reporting items for systematic reviews and metaanalyses guidelines. [27] Ethical approval was not required, as the study was conducted based on existing literature and did not involve direct human or animal participation.

Search strategy
A comprehensive electronic search of PubMed, Embase, Web of Science, Chinese National Knowledge Infrastructure, and WanFang databases was performed to identify relevant studies that focused on the prognostic value of HHLA2 in cancer patients prior to November 2020. The following search terms were included "human endogenous retrovirus-H long terminal repeatassociating protein 2" or "HHLA2" or "B7H7" or "B7H5" and "tumor" or "cancer" or "neoplasm" "carcinoma" and "prognosis" or "survival" or "outcome". The citation lists of the identified studies were also screened for other pertinent studies.

Data extraction and quality assessment
Data were extracted and examined from the included manuscripts by two independent reviews (ZCM and XJ). Any disagreement in the literature assessment was resolved through consensus with a third author (YJ). The following data were extracted: the clinicopathological characteristics, first author's name, publication year, country, cancer type, clinical stage, follow-up time, sample size, cut-off value, number and proportion of patients with high HHLA2 expression, analysis method, language, HR and 95% CI. If univariate and multivariate HRs existed, the latter was selected to minimize bias.
The methodological quality of each recruited study was evaluated independently by two investigators (ZCM and YJ), according to the Newcastle-Ottawa Scale (NOS). The scale included three dimensions as follows: selection (0-4), comparability (0-2), and outcome assessment (0-3), with a total score of 0-9. [28] Articles with a NOS score of ≥ 6 were considered to be of high quality.

Statistical analysis
The Stata version 12.0 software (StataCorp LLC, College Station, Texas, USA) was used to analyze the correlation between HHLA2 expression and OS, and to evaluate the clinicopathological significance of HHLA2 expression in human cancers. Heterogeneity among the included studies was measured by the I 2 and Chi-square Q tests. When significant heterogeneity (I 2 > 50% or P < .05) existed, the random effects model was used for the meta-analysis. Otherwise, a fixed effects model was adopted. Subgroup analysis was performed to examine the potential sources of statistical heterogeneity. Sensitivity analysis was conducted by sequentially omitting each individual cohort to evaluate the stability of the results. The potential publication bias was investigated using the funnel plot and Begg/Egger test. A P-value of < .05 was considered statistically significant.

Study selection and characteristics
A flow diagram showing our literature search and screening strategy is presented in Figure 1. A total of 121 articles were initially obtained according to the search strategies. After removing 37 duplicate studies and 31 apparently irrelevant articles, the remaining 53 records were screened by reading the titles and abstracts. Further, 32 studies were excluded (8 for reviews, 4 for meeting abstract, 5 for academic dissertations, 15 for animal studies or basic research). Twenty-one full-text studies were evaluated for eligibility, and finally 12 articles with 15 cohorts were included in the meta-analysis.
All of these studies were published between 2018 and 2020, and all the patients included in these studies were from China. They consisted of the following cancer types: clear cell renal cell carcinoma (ccRCC), [13,19] gastric cancer (GC), [17,18] epithelial ovarian cancer (EOC), [9,26] pancreatic ductal adenocarcinoma (PDAC), [23,24] oral squamous cell carcinoma (OSCC), [25] intrahepatic cholangiocarcinoma (ICC), [20] lung adenocarcinoma, [21] lung squamous cell carcinoma, [21] and colorectal cancer (CRC). [22] The sample sizes ranged from 63 to 403, with a total of 1747 patients. HHLA2 expression was measured by IHC in all cohorts. HRs and the corresponding 95% CIs of OS were obtained by the multivariate analysis in 13 cohorts and univariate analysis or Kaplan-Meier curves in 2 cohorts. The NOS scores of all these studies were between 6 and 8 points, indicating that each article was of high quality. Further detailed descriptions of these eligible articles are listed in Table 1.

Association of HHLA2 expression with OS
The random effect model was used to calculate the pooled HR and 95% CI because of apparent statistical heterogeneity (I 2 = 83.0%, P < .001). The results indicated that high expression of HHLA2 in human tumor tissue was associated with poor OS compared to low expression of HHLA2 (HR = 1.65, 95% CI: 1.12-2.43, P = .011) (Fig. 2, Table 2).
Considering the significant heterogeneity, we performed subgroup analysis to explore the potential factors that may cause heterogeneity. We classified the included cohorts and conducted subgroup analysis based on cancer type, clinical stage, sample size, proportion of patients with high HHLA2expression, and analysis method ( Table 2). Subgroup analysis by cancer type revealed that HHLA2 overexpression was correlated with poor

Sensitivity analysis and publication bias
We performed sensitivity analysis by successively ignoring individual cohort studies, and the results showed that no individual cohort affected the relationship between HHLA2 expression and OS in patients with solid tumors (Fig. 3). Moreover, no publication bias was detected, which was verified by the Begg test (P = .075), Egger's test (P = .507), and funnel plot (Fig. 4). Therefore, our results were robust and reliable.

Association of expression with clinicopathological parameters
Meta-analysis of the association between HHLA2 expression and clinicopathological parameters showed a significant relationship between high HHLA2 expression and old age (OR =

Discussion
In the past several decades, many studies have focused on identifying novel diagnostic and prognostic markers of tumors in order to advance treatment and outcomes by providing information for clinical decision-making. [29][30][31][32][33] The prognostic significance of HHLA2 expression has been evaluated in a variety of solid tumors. Here, we aimed to summarize and evaluate the results of published studies and extract valuable information that can be used in clinical decision-making for human solid tumors. A total of 12 studies, including 15 cohorts and 1747 patients, were included. Our results demonstrated that high HHLA2 expression predicted poor OS in Chinese patients with cancers. Furthermore, sensitivity analysis and publication bias proved that the results were robust and reliable. However, a significant heterogeneity existed among these studies. Considering the apparent heterogeneity, subgroup analysis was performed. Subgroup analysis showed that clinical stage, sample size, proportion of patients with high HHLA2 expression and analysis method did not change heterogeneity significantly. Subgroup analysis by cancer type showed that HHLA2 overexpression was correlated with poor OS in patients with ccRCC, GC, ICC, LC, and other cancer types, but not in patients with EOC. Nevertheless, high HHLA2 expression was related to better OS in patients with PDAC. In addition, the cancer type subgroup significantly reduced the heterogeneity within each subgroup. Thus, cancer type was the main source of heterogeneity and HHLA2 expression may play a different role in different types of cancer, especially in patients with PDAC. The mechanism should be further studied to reveal the influence of HHLA2 expression on the types of cancer. HHLA2 is involved in tumor immunosuppressive mechanisms and is a checkpoint for inhibition. [9] HHLA2 inhibits CD4 + and CD8 + T cell proliferation and cytokine production by binding to putative receptors in various immune cells. [34] In addition, HHLA2 binds to receptors on T cells and other immune cells to evade immune attacks. [35] Furthermore, HHLA2 binds to CD28H, which is expressed in naive T cells, natural killer cells,  endothelial cells and epithelial cells, but not in activated T cells, and plays an important a role in angiogenesis, cell-cell interaction, and cell migration. [36,37] However, the HHLA2/ CD28H interaction can also co-stimulate T cell proliferation and cytokine production via the AKT pathway. [18,38] Therefore, HHLA2 may have dual immune functions depending on the immune environment, receptor engagement or blockade, or interaction with different receptors, which is similar to the other members of the B7 family, such as B7-H3 and B7x. [24,39] In view of the conflicting results shown in HHLA2 so far, continued research and further understanding of these pathways is urgently needed to understand the causes of these differences among different tumors, to provide new strategies for tumor immunotherapy. Several limitations should be noted in the present study. First, all the patients included in this study were from China, which affected the applicability of the results to a certain extent. Second, the cut-off values of high HHLA2 expression were inconsistent, which may affect the effectiveness of HHLA2 as a predictor of cancer prognosis. Third, HR and 95% CI in some studies were calculated by extracting data from Kaplan-Meier curves rather than directly from the original literature, which inevitably led to small statistical deviations. Fourth, HHLA2 may act differently in patients with PDAC than in other cancer types, and PDAC patients were also included in the study. Given the limitations of this study, further well-designed studies that include evaluation of more tumor types with a larger sample size are needed.
In summary, high HHLA2 expression was associated with poor OS in Chinese patients with solid tumors, and might be used as a potential prognostic marker and tumor treatment target. However, high HHLA2 expression was related to better OS in patients with PDAC. Therefore, more mechanistic studies are needed for further analysis. Table 3 Meta-analysis of HHLA2 and clinicopathological features in cancer patients.

Categories
Trials (