B7-H4 Further Stratifies Patients With Endometrial Cancer Exhibiting a Nonspecific Molecular Profile Open Access

This retrospective study, which was based on our previously reported research,^5,10  included diagnostic (curettage) and final hysterectomy specimens from patients diagnosed with grade 3 or 2–3 endometrioid EC (EEC) and non-EEC (including serous carcinoma, clear cell carcinoma, undifferentiated/dedifferentiated carcinoma, carcinosarcoma, and mixed carcinomas) between January 2009 and December 2019, or with grade 1 or 1–2 EEC between January 2014 and December 2015, at the Peking Union Medical College Hospital (Beijing, China). Representative areas with tumor tissues were marked on hematoxylin-eosin–stained slides and sampled for tissue microarray (TMA) blocks. TMA blocks with duplicate 2-mm cores per case were constructed by using a tissue arrayer (MiniCore, Mitogen, Hertford, United Kingdom). Patients who received neoadjuvant chemotherapy before surgery and those with inadequate formalin-fixed, paraffin-embedded tissue blocks were excluded from the investigation. This study was approved by the Institutional Review Board (SK-995); informed consent was waived owing to the study’s retrospective nature.

As described previously,^11,12  DNA extracted with the QiaAMP DNA micro kit (QIAGEN Ltd, Manchester, United Kingdom) was used as a polymerase chain reaction template to amplify POLE exons 9–14; next, 150- to 200-bp products were amplified by using 100 ng of DNA derived from formalin-fixed, paraffin-embedded samples. Sequencing was performed with BigDye v3.1 terminator cycle sequencing chemistry on an ABI 3730 DNA analyzer (Applied Biosystems Inc, Foster City, California). All validated POLE mutations (P286R, V411L, S297F, A456P, S459F) underwent bidirectional Sanger sequencing twice.

Immunohistochemistry was performed as previously described.^5,13–16  Briefly, 4-μm TMA serial sections were deparaffinized and subjected to heat-induced epitope retrieval with 10 mM sodium citrate (pH 6.0) at 95°C for 20 minutes. The endogenous peroxidase activity was quenched by using a 0.3% hydrogen peroxide solution. TMA sections were incubated with primary antibodies against PD-L1, B7-H4, B7-H3, VSIG-3, CD8, p53, and the MMR-related proteins MSH2, MSH6, MLH1, and PMS2. The antibodies are described in detail in Supplemental Table 1 (see the supplemental digital content at https://meridian.allenpress.com/aplm in the November 2023 table of contents, containing 4 tables and 1 figure). Human tonsil or placenta tissues treated with primary antibodies were used as positive controls for PD-L1, B7-H4, B7-H3, and CD8 straining; stromal and inflammatory cells served as internal controls for MMR and p53, while the same tissues with isotype-matched antibodies comprised the negative controls.

The assessment of PD-L1 and CD8 has been described previously⁵ ; a combined positive score of 1 or greater denoted positive PD-L1 expression. We noted that samples expressing B7-H3, B7-H4, and VSIG-3 in tumor cells showed positive staining in 100% of these cells in each core, and that the staining in 2 cores per case was consistent. Therefore, the expression of these 3 proteins in tumor cells was considered either negative (no staining) or positive (weak-to-strong staining). B7-H3 in tumor-infiltrating immune cells was defined as positive when 1 or more B7-H3–stained immune cells per high-power field (magnification ×400) was observed in tumor cores.

A tumor that exhibited a complete loss of nuclear expression of any of the MMR proteins (MLH1, PMS2, MSH2, and/or MSH6) was considered MMRd, whereas the detection of all 4 MMR proteins in tumor cells (in the presence of an intact internal control) was considered MMR-proficient. Mutation-type p53 staining was defined as intense and diffuse nuclear staining, the complete absence of nuclear staining, or significant cytoplasmic staining in the presence of variable nuclear staining; wild-type expression was defined as weak and heterogeneous nuclear staining. As described previously,^5,10  ECs are categorized into the following molecular subgroups: POLEmut (tumors with pathogenic variants in the exonuclease domain of POLE), MMRd (tumors with MMRd in the absence of POLE mutation), p53mut (tumors with mutation-type p53 staining in the absence of POLE mutations or MMRd), and NSMP (lacking any of POLE, MMR, or p53 alterations).

The χ² test was used to determine the association between categorical variables. Relapse-free survival (RFS) was defined as the interval between the date of surgery and that of the detection of the first local, regional, and/or distant relapse. Disease-specific survival (DSS) was defined as the interval between the date of surgery and that of death caused by EC. Survival curves were plotted by using the Kaplan-Meier method and compared by using the log-rank test. To identify prognostic predictors, univariate and multivariate survival analyses were performed by using the Cox proportional hazards regression model, and hazard ratios (HRs) with 95% CIs for recurrence and death were calculated. The predictive capacity (discrimination) of molecular classification was evaluated by the Harrell concordance index. The integrated discrimination improvement (IDI) for molecular classification was calculated by using the R package. All statistical analyses were conducted by using the Statistical Package for the Social Sciences (version 20.0; IBM Corp, Armonk, New York). A 2-sided P value <.05 was considered statistically significant.

A total of 833 patients with EC were included in this study; their median age was 58 years (range, 25–88 years). The clinicopathologic characteristics of these patients are summarized in Supplemental Table 2. Molecular subgroup analysis data were available for 590 patients, including 452 with EEC and 138 with non-EEC. B7-H3 was expressed in both tumor cells and immune cells (predominantly in the former); B7-H4 and VSIG-3 were detected exclusively in tumor cells and exhibited a cytoplasmic/membranous staining pattern (Figure 1, A through D). B7-H3 expression in immune and tumor cells was observed in 352 (42.3%) and 741 (89.0%) of the 833 patients, respectively; B7-H4 expression was noted in 596 (71.5%) and VSIG-3 in 831 (99.8%).

VSIG-3 was widely expressed in 831 of the 833 EC samples (99.8%), although no associations between its expression and pathologic characteristics were observed. The associations between B7-H3 expression in tumor cells/immune cells, B7-H4 expression, and the patients’ clinicopathologic characteristics are presented in Table 1. B7-H4 positivity in tumor cells was observed significantly more frequently in patients 58 years of age or older as well as in samples with favorable pathologic features (early stage, low-grade tumors, absence of LVSI, endometrioid histology, and <50% myometrial invasion). Among the 4 molecular subgroups, tumors that were NSMP (194 of 246 [78.9%]) and p53mut (101 of 131 [77.1%]) exhibited higher relative frequencies of positive B7-H4 expression than did those that were POLEmut (27 of 50 [54.0%]) and MMRd (109 of 163 [66.9%]). B7-H4 expression was not associated with PD-L1 expression or CD8⁺ T-cell density.

Similar to B7-H4, positive B7-H3 expression in tumor cells was more frequently observed in samples with low-grade tumors, early FIGO stage, and endometrioid histology. In contrast, B7-H3 positivity in immune cells was associated with unfavorable pathologic features (advanced stage, high-grade tumors, presence of LVSI, non-EEC histology, and deep myometrial invasion). Intriguingly, tumors with the p53mut subtype showed the highest and lowest rates of B7-H3 positivity in immune and tumor cells, respectively. B7-H3 expression in immune cells was positively correlated with PD-L1 expression and a high density of CD8⁺ T cells, but was negatively correlated with B7-H4 expression in tumor cells (Supplemental Table 3). B7-H3 in tumor cells was not associated with PD-L1, CD8⁺ T cells, or B7-H4.

After excluding patients who only had curettage specimens available and whose follow-up times were less than 3 months, 664 subjects who underwent final hysterectomy with complete adjuvant systemic therapy (when necessary) were subjected to survival analysis. There were no significant differences between the entire cohort of 833 patients and the survival analysis–only subgroup in terms of clinicopathologic parameters (data not shown). After a median follow-up of 31 months (range, 4–121 months), 87 patients (13.1%) had a relapse and 63 (9.5%) had died of EC as of July 2020.

Kaplan-Meier analyses showed that B7-H3 positivity in tumor cells was significantly associated with favorable prognosis, with HRs of 0.32 (95% CI, 0.20–0.52) for RFS and 0.24 (95% CI, 0.14–0.41) for DSS (Figure 2, A and B, and Supplemental Table 4). In contrast, B7-H3 positivity in immune cells was associated with poor outcomes (Figure 2, C and D); the HRs were 2.89 (95% CI, 1.86–4.47) for RFS and 2.49 (95% CI, 1.49–4.14) for DSS. B7-H4 expression tended to be associated with improved RFS and DSS (Supplemental Figure 1). On multivariate analysis that incorporated age, FIGO stage, tumor grade, histology, depth of myometrial invasion, LVSI status, and molecular subgroup, B7-H3 expression in tumor cells remained significantly associated with longer RFS and DSS, and B7-H4 expression status in tumor cells was an independent predictor of longer DSS (Table 2).

To identify biomarkers that could refine the molecular subtyping of NSMP, we analyzed the association between B7-H3/B7-H4 and survival in patients stratified by molecular subtype. Kaplan-Meier analyses showed that B7-H3 expression in tumor cells as well as B7-H4 expression had significant prognostic impacts within the NSMP subgroup, but was not associated with survival within the POLEmut, MMRd, or p53mut subgroups. Considering the association between these proteins and the patients’ pathologic features, we analyzed their associations within the NSMP subgroup (Table 3). Consistent with data from the entire cohort, B7-H3 and B7-H4 expression in tumor cells was observed more frequently in samples with early FIGO stage, low-grade tumor, and endometrioid histology; B7-H3 in immune cells was associated with advanced FIGO stage and high-grade tumors. Accordingly, B7-H3 and B7-H4 positivity in tumor cells was associated with favorable prognoses among the 194 patients with the NSMP subtype, while B7-H3 positivity in immune cells was associated with unfavorable prognoses (Table 4). Univariate survival analyses showed that FIGO stage, LVSI, tumor grade, and histology were associated with survival in patients with the NSMP subtype (Table 4). On multivariate analysis, positive B7-H4 expression was a predictor of improved RFS (HR, 0.27; 95% CI, 0.09–0.78; P = .02) and DSS (HR, 0.16; 95% CI, 0.05–0.58; P = .005) independent of FIGO stage, LVSI status, or histology (Table 4).

To further explore the prognoses of patients with NSMP tumors, we compared the survival curves and IDI according to molecular subtype, and also incorporated B7-H4 status within NSMP in our analyses. Patients categorized in the NSMP and MMRd subgroups had different survival outcomes when B7-H4 status was incorporated into the NSMP molecular subtype (Figure 3, A through D). The HRs for B7-H4⁻ NSMP tumors were 5.76 (95% CI, 2.10–15.8; P = .001) for RFS and 6.67 (95% CI, 2.06–21.7; P = .002) for DSS (Table 5). The IDI for RFS at 24 months was 1.2 (95% CI, 0.3–2.4; P = .001) and that for DSS at 24 months was 1.1 (95% CI, −0.7 to 3.3; P = .24). These data showed that molecular classification that incorporates B7-H4 within NSMP could improve the ability to predict disease relapse.

To the best of our knowledge, ours is the first study to investigate the expression of B7-H3, B7-H4, and VSIG-3 (a ligand of VISTA) across molecular subtypes in a large cohort of patients with EC. Our findings demonstrated that the expression patterns and prognostic significance of these proteins are different in various EC tumor types. B7-H4 was exclusively expressed in tumor cells and was detected significantly more frequently in patients with NSMP and p53mut tumors, and in those with favorable pathologic features and survival outcomes. B7-H3 was detected in both immune cells and tumor cells; its positivity in tumor cells was associated with favorable survival, while its presence in immune cells was associated with poor outcomes. VSIG-3 was widely expressed in almost all tumor samples but was not associated with pathologic features or prognosis. Importantly, B7-H4 expression was significantly associated with longer survival in patients with NSMP, and was a predictor of improved RFS and DSS independent of FIGO stage or pathologic features.

NSMP is the largest molecular subgroup of EC, comprising approximately half of all cases. It is characterized by the absence of defining molecular features and an intermediate prognosis,¹⁷  and is a heterogeneous grouping with diverse clinical and histologic presentations as well as varying clinical outcomes.¹⁸  Although most patients with NSMP have endometrioid histology, this subtype has also been reported in endometrial clear cell carcinoma¹⁹  and undifferentiated carcinoma.²⁰  Some patients with the NSMP subtype have an excellent prognosis, whereas others have poor outcomes. The lack of predictive biomarkers makes NSMP the most challenging molecular subtype to treat, and current adjuvant therapy is mainly determined by FIGO stage, grade, LVSI status, and histologic type; all these factors are subject to interobserver variability.¹⁸  Recently, several biomarkers were reported to be possible prognostic factors for risk stratification of NSMP tumors. Results from 2 large, randomized trials of early-stage EEC (PORTEC-1 and -2) demonstrated that, within the NSMP subgroup, patients with CTNNB1 (β-catenin) mutations had worse prognoses than did those without.²¹  Similarly, a previous study found that CTNNB1 mutation was associated with significantly worse RFS in patients with low-grade early-stage EEC.²²  In a study of 452 patients with EC, Kommoss et al²³  demonstrated that L1 cell adhesion molecule (L1CAM) overexpression was predictive of worse outcomes among those with NSMP tumors, and that L1CAM was an independent predictor of DSS on multivariate analysis. Recently, a study of 465 patients with EC by Sengal et al²⁴  found that FGFR2c mesenchymal isoform expression was associated with poor prognosis within the NSMP and MMRd subgroups. In our study of 833 patients with EC wherein the molecular subtypes of 590 were known, we found that B7-H4 expression was a predictor of improved RFS and DSS independent of FIGO stage or pathologic features, and that incorporating B7-H4 in the molecular classification of NSMP significantly improved the ability to predict disease relapse. Unlike previous studies, a high proportion of high-grade tumors (429 of 833 [51.5%]) and non-EECs (238 of 833 [28.6%]) were included in this study. Incorporation of new biomarkers that can further stratify the prognoses of patients with NSMP and MMRd within the current molecular classification framework could improve the management of patients with EC in clinical practice. Further studies from independent groups are warranted to validate the prognostic significance of the abovementioned markers.

B7-H4, a member of the B7 family of immunomodulatory molecules, has been identified as a coinhibitory molecule that negatively regulates the T-cell immune response and promotes immune escape.²⁵  Previous studies found that B7-H4 was overexpressed in many tumors and was associated with poor survival and impaired antitumor immune responses.^26–29  An increasing number of studies have investigated the potential of B7-H4 as a therapeutic immune target^30–32 ; however, the evidence continues to suggest that B7-H4 in tumor cells is not associated with a T-cell-inhibitory microenvironment but rather plays a positive role in limiting tumor growth in animal models.^33–35  Our analysis showed that B7-H4 expression correlated with improved survival in our cohort of patients with EC, including those with NSMP subtype—a PD-1–resistant subgroup that lacks effective immunotherapy targets. Rahbar et al³³  found that high B7-H4 expression correlated with improved RFS in 2681 patients with breast cancer, which is consistent with our findings in patients with EC. Together with the recently discovered positive role for B7-H4 in antitumor immunity,³³  our data suggest that targeting B7-H4 through monoclonal antibodies may not be beneficial for patients with EC, especially those with NSMP.

B7-H3, a member of the B7 family immune checkpoint proteins, reportedly acts as a T-cell costimulatory and coinhibitory factor in different tumor microenvironments.³⁶  It is expressed in tumor cells, tumor-associated vascular endothelial cells, and immune cells.^37,38  Most data demonstrate that B7-H3 inhibits T-cell activation and that it is associated with tumor aggressiveness, unfavorable clinicopathologic features, and poor outcomes.³⁶  In a study of 99 EC samples, Brunner et al⁸  found that B7-H3 expression in tumor cells was present in 75.7% of samples, and that overexpression of B7-H3 (ie, an immunohistochemical score ≥9) was associated with T-cell infiltration and shortened overall survival. In our present study, B7-H3 was detected in both tumor cells (741 of 833 samples [89.0%]) and immune cells (352 of 833 samples [42.3%]); however, Brunner et al⁸  did not investigate B7-H3 in immune cells. Our findings showed that B7-H3 in tumor cells was associated with favorable pathologic features, low-risk molecular subtypes, and improved survival outcomes; this contrasted with its expression in immune cells correlating with the high-risk p53mut subtype, unfavorable pathologic features, and poorer prognoses. Additionally, B7-H3 in immune cells, but not in tumor cells, correlated positively with PD-L1 and CD8⁺ T cells but negatively with B7-H4, in tumor cells. The inconsistency with the previous study may be attributed to the different cohorts and methods of B7-H3 assessment. The sample size of 833 in this present study was relatively large; thus, the results ought to be more generalizable. The different distributions and prognostic implications of B7-H3 in tumor and immune cells across pathologic and molecular subgroups indicate that the regulatory mechanisms of its expression in these 2 cell types are distinct. However, the mechanism of B7-H3 regulation in EC remains unknown; as such, future studies are warranted to identify the regulatory mechanisms associated with B7-H3 expression and affirm its prognostic significance in EC.

VSIG-3/IGSF11, which belongs to the immunoglobulin superfamily, was first identified in 2002 and is abundantly expressed in the testes, ovaries, and brain.³⁹  VSIG-3 functions as a cell adhesion molecule,⁴⁰  and was recently found to be an inhibitory ligand for VISTA.^6,41,42  Previous studies showed that VSIG-3 is highly expressed in colorectal cancer, hepatocellular carcinoma, intestinal-type gastric cancer, and advanced gliomas.^39,43  Considering the wide expression of VSIG-3 in human cancers and the inhibitory function of the VSIG-3/VISTA axis on T cells, VSIG-3 is a promising therapeutic target. Our study, which was the first to evaluate VSIG-3 expression in EC, showed that this protein was highly and abundantly expressed in the tumor cells of both EECs and non-EECs. However, the mechanisms of its upregulation remain unknown, and additional studies are needed to explore the functions and interactions of VSIG-3 and VISTA in EC, as well as their potentials for being immunotherapeutic targets.

We acknowledge some limitations in our study. First, it was a retrospective investigation, which can produce inherent, unavoidable biases. Second, the use of TMA blocks may not have accurately represented the entirety of each sample given intratumoral heterogeneity, although duplicate 2-mm TMA cores were sampled per case. Lastly, our study was limited by its single-center nature and lack of an independent validation cohort. Further studies from multicenter, independent cohorts are needed to validate our findings.

In conclusion, we found that B3-H3 and B7-H4 expression in tumor cells was more frequent in tumors with favorable pathologic features and prognoses, while B7-H3 in immune cells was associated with unfavorable pathologic features and poor prognoses. Additionally, our data suggest that B7-H4 is an independent predictor of improved survival in patients with NSMP molecular subtype tumors. The NSMP subgroup of EC can be further stratified by B7-H4 status, and molecular classification of NSMP that incorporates B7-H4 can improve the prediction of disease relapse.

We thank the medical record room staff for their assistance in retrieving the patients’ medical records.