A2AR Expression and Immunosuppressive Environment Independent of KRAS and GNAS Mutations in Pseudomyxoma Peritonei

In pseudomyxoma peritonei (PMP), KRAS and GNAS mutations are frequent. We hypothesized that these mutations may contribute to the suppression of antitumor immunity: KRAS may induce GMCSF expression, while GNAS may enhance the expression of cyclic adenosine monophosphate and A2AR signaling. This study aimed to explore possible mechanisms facilitated by KRAS and GNAS mutations for escaping immune surveillance. Additionally, we looked for new potential therapeutic and prognostic targets in this rare disease which is poorly characterized at the molecular level. GM-CSF, A2AR, CD73, CD39, and PD-L1 expression was investigated by immunohistochemistry in 40 PMPs characterized for GNAS and KRAS mutational status. Immune cell populations were studied by immunohistochemistry and nanostring nCounter®. Following the criteria of a prognostic nomogram reported for PMP, we stratified the patients into two different risk groups, with 28 “low-risk” and 12 “high-risk” patients. We observed the expression of GM-CSF (74%); CD39 (37%); CD73 (53%); A2AR (74%); and PD-L1 (16%) which was unrelated to GNAS or KRAS status. The tumor microenvironment showed the presence of CD4+ T cells (86%); CD8+ T cells (27%); CD20+ B (67%); CD15+ cells (86%); and CD163+ M2 macrophages (67%), while CD56+ NK cells were absent. CD163 expression (27%) in PMP tumor cells was associated with poor prognosis. GNAS mutation and A2AR expression were not associated with a specific immune transcriptional signature. However, the expression assay revealed 21 genes associated with prognosis. The “high-risk” patients exhibited worse progression-free survival (HR = 2.3, CI 95%: 1.1–5.1, p = 0.034) and significant downregulation of MET, IL8, PPARG, DTX4, HMGA1, ZIC2, WNT5B, and CCRL2. In conclusion, we documented the presence of immunosuppressive factors such as GM-CSF, A2AR, and PD-L1 in PMP. These factors were not associated with GNAS and KRAS status and could be explored as therapeutic molecular targets. Additionally, a set of potential prognostic biomarkers, including CD163 expression in tumor cells, deserve further investigation.


Introduction
Pseudomyxoma peritonei (PMP) is a rare clinical entity characterized by the presence of mucinous ascites, leading to significant abdominal distension and bowel obstruction [1]. PMP most commonly originates from a mucinous neoplasm in the appendix [2]. The combined treatment of cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) with cisplatin plus mitomycin has been established as the standard of care for PMP [3]. However, despite such aggressive intervention, recurrence rates after complete cytoreduction range from 24% to 30%. PMP relapsing or progressing after CRS/HIPEC were treated with curative-intent surgery or palliative options such as systemic chemotherapy [4].
For its part, GNAS encodes the α-subunit of a stimulatory G-protein (Gαs), an intracellular signal transducer that activates adenylate cyclase. GNAS mutation induces the oncogenic activation of Gαs, resulting in increased cyclic adenosine monophosphate (cAMP) levels. Subsequently, cAMP binds to protein kinase A (PKA), leading to the release of two catalytic subunits that can phosphorylate various transcription factors, including NF-kB, HIF-1α, and CREB. The PKA/CREB signaling promotes mucin expression, and the inhibition of the cAMP/PKA pathway in PMP in in vivo models has been shown to decrease mucinous tumor growth [10].
Furthermore, both mutations may modulate immune reactions, contributing to the suppression of antitumor immunity. KRAS mutation, for instance, in colorectal cancer induces the expression of immune mediators such as the cytokine granulocyte macrophage-colony stimulating factor (GM-CSF), which acts as a strong immunosuppressive factor [11,12]. Additionally, cAMP derived from GNAS mutation plays a physiological role as an immunosuppressive messenger of T-cell function [13].
It is noteworthy that, regardless of GNAS mutation, Gαs is involved in "adenosine signaling," through its coupling with the A2A-adenosine receptor (A2AR). Upon adenosine binding to A2AR, Gαs triggers the cAMP-mediated activation of PKA. Adenosine levels are regulated by the conversions of extracellular ATP into AMP and of AMP into adenosine, catalyzed by the ectoenzymes CD39 and CD73, respectively [14]. In contrast, the expression of CD39, CD73, and A2AR on tumor and stromal cells is promoted by the hypoxia-and c-AMP/PKA-mediated activation of HIF1α and NF-kB [15,16]. Owing to persistent inflammation and hypoxia in the microenvironment of solid tumors, ATP and adenosine levels may remain elevated, promoting the A2AR axis, which may further suppress antitumor immunity [17]. Specifically, A2AR signaling inhibits natural killer (NK) and CD8+ T cells and increases the generation of immunosuppressive cells, including myeloid-derived suppressor cells and tumor-associated macrophages [18][19][20].
Given the current understanding of the molecular landscape of PMP and the presence of an immunosuppressive tumor microenvironment associated with KRAS and GNAS mutations in other tumor types, this study aimed to investigate the potential role of immunosuppressive factors in PMP, allowing tumor cells to evade immune surveillance. Additionally, we aimed to explore new potential therapeutic and prognostic targets in this poorly characterized disease, which currently lacks effective treatment options.

Patients
This study had the approval of the International Review Board (IRB) of the Fondazione IRCCS Istituto Nazionale Tumori of Milan and was conducted on a monoinstitutional series of 40 patients with appendiceal PMP treated with complete cytoreduction (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC). This series included 31 patients from a previous study and for which the tissue material was still sufficient for new analyses [8]. The median age was 57 years (range: 27-71); the male/female distribution was 17/22; 89% had low-grade (PSOGI classification) disease [21]; and the median peritoneal cancer index was 24 (range: 3-39).

NGS, Immunohistochemistry, and Transcriptional Analysis
We assessed KRAS and GNAS mutational status using targeted next-generation sequencing (T-NGS) on genomic DNA extracted from formalin-fixed paraffin-embedded (FFPE) PMP samples following standard procedures, as previously described [8].
We extracted RNA from FFPE tumor tissue using the miRNeasy FFPE Kit (Qiagen, Germantown, Maryland MD, USA) according to the manufacturer's instructions. We performed transcriptional analysis by nCounter ® (Nanostring Technologies, Seattle, WA, USA). Specifically, we used 150 ng of mRNA and the PanCancer Pathways and the PanCancer Immune Profiling panels. The first panel quantifies the expression of 770 human genes from cancer-associated canonical pathways, including MAPK, JAK-STAT, PI3K, RAS, Cell Cycle-Apoptosis, Hedgehog, Wnt, DNA Damage-Repair, Transcriptional Misregulation, Chromatin Modification, TGF-SS, NOTCH, and Driver Genes. The PanCancer Immune Profiling panel quantifies the immune counterpart. We used the NanoStringNorm package for R software 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria) to assess quality; the geometric mean of the counts relative to each sample, the mean plus two standard deviations and the total sum of counts options were used to correct the data for technical, background, and batch-effect issues, respectively. We used the expression counts of housekeeping genes and quantile normalization to account for intersample variations with the PanCancer Pathways Panel. We used Canberra distance and Ward as the distance metric and linkage method, respectively, to assess the natural grouping of samples. We used the DESeq2 package for R for differential expression analyses, under standard parameters.
Gene set enrichment analysis (GSEA) and single-sample(ss)-GSEA were, finally, used for functional annotation analyses under default conditions (with weighted statistic score and 10,000 permutations of gene sets).

Prognosis and Statistical Analysis
Overall survival (OS) was defined as the time interval from the surgery to death from any cause, and progression-free survival (PFS) as the period from the surgery to the disease progression. We stratified patients into two different risk groups (28 "low-risk" and 12 "high-risk" patients) according to a prognostic nomogram for PMP patients based on large-scale population data [23]. Briefly, the nomogram provided an individual estimate of patient survival by incorporating demographic and disease-related parameters with critical prognostic significance.
The Cox proportional-hazards model in the globaltest package for R was used to test the association between gene expression levels, assumed as continuous variables, and PFS or OS as clinical outcomes. Globaltest was run with 100,000 permutations on genes that varied mostly over the dataset, namely, those whose expression exceeded two variation coefficients (i.e., the standard deviation on the mean ratio). The Cox proportional-hazards model was also used to assess the correlation between PFS or OS and risk or the expression in TC or TME of the following markers: GM-CSF, CD39, CD73, A2AR, PD-L1, CD4, CD8, CD20, CD15, and CD163.

Immunohistochemical Analysis
The detailed data obtained from 38 PMPs (2 cases were not evaluable) are presented in Table 1, providing information on the combined scores assessed in the tumor cells (TCs) and tumor microenvironment (TME), as well as KRAS and GNAS status.
Overall, combined TC and TME analysis revealed GM-CSF expression in 28 (74%) cases. There was no significant correlation between GM-CSF expression and KRAS or GNAS status (Table 2).
CD39 expression in the TME was found in only three (8%) cases ( Figure 1D). Notably, these three CD39+ cases were KRAS mutated, with two of them being GNAS wild type (WT) ( Table 2).

T Cells
We successfully characterized the TME by investigating the distribution of the immune cell types in 37 cases (Tables 1 and 3). Overall, the expression of the immune cells was not significantly associated with KRAS or GNAS status. Table 3. Distribution of CD4, CD8, CD56, CD20, CD15, and CD163 immunohistochemical expression according to KRAS and GNAS mutational status and A2AR expression. Markers were evaluated in the tumor microenvironment (TME) only, except for CD163, which was evaluated both in tumor cells (TC) and tumor microenvironment (TME). CD4+ T cells were observed in all but 5 (86%) cases ( Figure 2B), with 14 of 32 (44%) positive cases displaying strong CD4 expression. CD8+ T cells were detected in 10 (27%) cases ( Figure 2C). Among the CD8+ T cases, all but one (90%) showed low expression (combined score 2, 3, or 4). A higher fraction of CD8+ T cells were observed in A2AR positive cases compared to negative cases, although the difference was not statistically significant (37% vs. 8%).

Natural Killer (NK) Cells
CD56+ NK cells were absent in the TME of all cases ( Figure 2D).

B Cells
CD20+ B cells were observed in 25 (67%) cases ( Figure 2E and Table 3) with the majority of cases (76%) showing low CD20 expression.

Transcriptional Analysis
We conducted an analysis of the gene expression profile associated with GNAS mutation by using the PanCancer Pathways panel. Out of 36 PMPs, 20 cases passed the quality check and were suitable for analysis. We found no statistically significant difference in expression levels between the GNAS-mutated and WT cases (10 vs. 10), except for the downregulated BID gene and the upregulated CDKN1C, MGMT, and WNT10B genes in the GNAS-mutated cases ( Figure S1A). Next, we investigated the possible gene expression profile associated with A2AR expression in 16 A2AR positive and 6 A2AR negative PMPs. These two groups exhibited no statistically significant differences in gene expression levels, except for the downregulation of IRS1 and FOXL2 in the A2AR positive cases ( Figure S1B).
We further studied the immune cell populations in PMP using the PanCancer Immune Profiling panel. The data were acquired for 36 PMPs, and after the quality check, 33 cases were suitable for analysis. We did not observe any expression differences between the CD163 TC positive (7 cases) and negative (26 cases), GNAS-mutated (18 cases) and WT (15 cases), and A2AR positive (19 cases) and negative (14 cases) PMPs. GSEA analysis revealed a group of 13 patients characterized by the enrichment of genes involved in immunological and inflammation pathways. However, no significant correlation was found with GNAS mutation, A2AR, or CD163 expression ( Figure S2).

Prognostic Analysis
In order to identify potential prognostic markers, we investigated the correlation between the expression of the biomarkers with overall survival (OS) or progression-free survival (PFS). The Cox proportional-hazards model showed a significant association between CD163 expression in TCs with poor PFS (HR = 2.3, 95% CI: 1-5.4, p = 0.047) ( Figure 3A). However, in multivariate analysis, CD163 expression did not retain prognostic significance. Although detected in a limited number of cases and not statistically significant (HR = 2.4, CI 95%: 0.92-6.3, p = 0.073), PD-L1 expression was predominantly observed in patients with poor PFS.

Risk Classification
To further evaluate the prognostic significance, we classified the patients into two risk groups, namely "low-risk" and "high-risk", using the criteria of a prognostic nomogram based on extensive population data for PMP patients [23]. In our series, we identified 28 patients in the "low-risk" group and 12 patients in the "high-risk" group, each exhibiting distinct prognostic and molecular features. In detail, in the univariate analysis, we found a statistically significant correlation between the "high-risk" group and poor PFS (HR = 2.3 C 95% CI: 1.1-5.1, p = 0.034) ( Figure 3B). Moreover, the "high-risk" group displayed a significant downregulation of PPARG, DTX4, WINT5B, RET, MAP3K12, MET, IL8, CSF1R, HMGA1, ETS2, DUSP5, FGF9, and JUN ( Figure S1C). These findings highlight the molecular differences between the two risk groups and suggest their potential role as prognostic markers in PMP.

Discussion
This study aimed to investigate the presence of an immunosuppressive microenvironment potentially associated with KRAS and GNAS mutations in PMP.
Our findings revealed GM-CSF overexpression in 74% of cases with strong expression observed in TCs in 68% of positive cases. Previous research has reported the transcriptional upregulation of GM-CSF mediated by KRAS attributed to Ras-regulated transcription factor-binding sites in the GM-CSF promoter region [12]. Consistently, inhibition of the PI3K or MAPK pathway has been shown to abolish GM-CSF expression in KRAS-mutated mouse pancreatic ductal epithelial cells [12]. In our study, 69% of KRAS-mutated PMP showed GM-CSF positive TCs. However, it is worth noting that four KRAS WT cases were also GM-CSF positive, indicating that other molecular mechanisms independent of KRAS may contribute to GM-CSF upregulation. GM-CSF is a potent cytokine that plays a complex role in the interplay between myeloid cells and antitumor immunity or immunosuppression [24]. Specifically, the antitumor effect depends on the ability of this cytokine to promote the differentiation and maturation of antigen-presenting dendritic cells (DCs) and their migration to lymph nodes, thereby inducing T-cell activation. Conversely, high levels of GM-CSF can lead to tumor immune tolerance by impeding DC maturation or activation and promoting the generation of immunosuppressive cells such as Tregs, myeloid-derived suppressor cells (MDSCs), and granulocytes, which restrain cytotoxic T lymphocyte-mediated antitumor responses [24]. In the context of PMP, the presence of an immunosuppressive effect is suggested by the elevated levels of GM-CSF detected. This finding aligns with evidence reported in other tumors, such as mesothelioma and pancreatic ductal adenocarcinoma, where the blockade of GM-CSF with a neutralizing antibody significantly attenuated the immunosuppressive potential of myeloid cells, leading to restored T-cell proliferation and activity [12,25].
Regarding the relation between GNAS and adenosine signaling, our study focused on the A2AR axis, which is typically promoted by the c-AMP/PKA-and hypoxia-mediated activation of NF-kB and HIF-1α [15,17]. Our data revealed frequent and strong A2AR (80%) expression in TCs, with higher levels observed in GNAS WT (82%) cases compared to GNAS-mutated (58%) cases. Additionally, CD73 (53%) expression, predominantly with a high score, and CD39 (37%) expression were also detected. A2AR and/or CD39 and/or CD73 expression was found in 36 (95%) cases. Overall, GNAS activation either through mutation or A2AR expression was present in the majority (87%) of cases. Considering the role of adenosine signaling in suppressing antitumor immunity, A2AR expression may contribute to critical tumor-mediated immune escape in PMP. Adenosine, through the A2AR-mediated enhancement of c-AMP/PKA signaling, inhibits the activity of CD8+ T and NK cells, while stimulating MDSC and type 2 (M2) macrophages to suppress T cells and release protumorigenic cytokines [18][19][20]. Targeting A2AR or CD73 shows promise as it enhances antitumor responses in vivo through T-and NK-cell activation and it may also synergistically potentiate the effects of chemotherapy and immunotherapy in vitro [26]. Various inhibitors of A2AR are currently being investigated for cancer therapy and their clinical efficacy has been observed both as monotherapy and in combination with other agents in patients with advanced solid tumors. Importantly, the effect of A2AR inhibitors may be further enhanced by oxygenation, a novel class of antitumor drugs. These molecules can attenuate HIF-1α-mediated immunosuppression (known as oxygen immunotherapy) by reducing both extracellular adenosine accumulation and the levels of CD39/CD73 adenosine generating enzymes [27,28].
Furthermore, the PD-L1 immune checkpoint may be a potential target in PMP, as we observed PD-L1 expression in TM in a subset of cases (16%). Gleeson et al. also reported PD-L1 and PD-1 expression in 18% and 36% of PMPs, respectively [5]. This finding is significant as several clinical trials are currently evaluating the efficacy of adenosine and PD-1 coinhibition options [29].
We conducted an analysis of immune cell populations that can influence the response to immune checkpoint blockade and we found a high proportion of CD4+T (86%) and CD20+B cells (68%). The fraction of CD8+T cells (27%), mostly showing a low score, was more prevalent in A2AR positive than negative cases, although the difference was not statistically significant. Overall, these immune cell populations are strongly associated with tumor surveillance and can impact the efficacy of checkpoint inhibitors [30]. However, the absence of CD56+NK and the enrichment of CD163+M2 macrophages (67%) and CD15+ (83%) cells may sustain an immunosuppressive TME in PMP. Moreover, CD20+B cells, which are typically associated with antitumor immune responses, can also exhibit immunosuppressive and tumor-promoting functions [31][32][33]. Interestingly, a strong expression of CD163 was also observed in TCs in 10 (26%) cases, with a trend for a higher frequency of A2AR positive cases (37% vs. 8%).
Moreover, in a subset of PMPs, we evaluated specific transcriptional signatures using nanostring panels. Our findings suggest that neither GNAS nor A2AR status markedly affect gene expression in PMPs. However, we speculate that the absence of a transcriptional difference suggests that GNAS mutation and A2AR expression represent two mechanisms leading to cAMP-PKA pathway activation. Since at least one of these two alterations was present in the majority (18 of 20) of analyzed PMPs, it is not surprising that a functional annotation analysis did not reveal any specific correlations with GNAS mutation or A2AR expression. However, our analysis segregated a group of patients characterized by the enrichment of genes whose expression is enhanced in immunological (e.g., T, NK, and B functions) and inflammation (chemokines and cytokines) pathways, which may represent a potential advantage in response to checkpoint inhibitors.
We also looked for prognostic biomarkers and found that patients with CD163 positive TCs exhibited worse PFS (HR = 2.3, 95% CI: 1-5.4, p = 0.047), consistent with the previous literature [34]. On the other hand, nanostring experiments identified a set of 21 genes whose expression was significantly associated with OS or PFS, including COL4A6, FGF14, LEP, TPO, and PTPN5 which are involved in the PI3K, MAPK, and JAK/STAT pathways, as well as regulators of T-and B-cell stimulation/development (BLK, BLNK, SH2D1A, FCER2). These findings warrant further investigation in larger cohorts of PMP cases.
Finally, the "high-risk" group of PMPs, defined by the nomogram, exhibited poor PFS and a significant downregulation of genes mainly involved in the Transcriptional Misregulation and Hedgehog pathways. Beyond the biological implications of these genes, our data suggest that "high-risk" PMPs may have unique molecular features, supporting the potential use of transcriptional features for improving the stratification of this rare disease.

Conclusions
In conclusion, our data, although derived from a limited series of PMP, provide preliminary evidence supporting the presence of immunosuppressive factors such as GM-CSF, the A2AR axis, and PD-L1 expression in PMP. While these factors were not significantly associated with GNAS and KRAS status, they hold potential as therapeutic targets. Considering the efficacy of checkpoint inhibitor-based therapies, it is important to further investigate the potential negative predictive role played by the immunosuppressive cell fraction (CD163+M2 macrophages, CD15+cells). Concurrently, the prognostic significance of a set of potential biomarkers, especially the expression of CD163 in TCs, deserves further investigation.