The prognostic significance of tumor-immune microenvironment in ascites of patients with high-grade serous carcinoma

Ovarian carcinoma is a gynecological malignancy with the highest mortality rate in Western countries and the sixth leading cause of cancer-related deaths among women.¹ High-grade serous carcinoma (HGSC) is the most common and aggressive histological type. Lack of symptoms and adequate screening methods usually result in delayed diagnosis and advanced stage with less than 40% of a 5-year survival rate for HGSC patients.² Cytoreductive surgery combined with carboplatin/paclitaxel chemotherapy (with or without bevacizumab) is still the standard treatment approach.³ Despite the good overall response, 70% of the patients experience relapse or develop metastatic disease and resistance. Unfortunately, no significant improvement has been achieved in the last three decades, except for BRCA1/2 mutated tumors where poly-ADP-ribose polymerase (PARP) inhibitors slightly shifted the care paradigm for ovarian carcinoma.⁴ However, accumulating evidence is showing that tumor-immune microenvironment (TME) in ovarian carcinoma can open the door for the discovery of new prognostic markers and the development of immunotherapeutic treatment approaches. For instance, the presence of tumor-infiltrating CD3⁺ T cells in the primary tumor positively correlates with progression-free survival (PFS) and overall survival (OS) of ovarian carcinoma patients, as does high CD8/CD4 ratio.⁵ Strong association with better OS was also seen on CD3⁺ T cells expressing CD103 tissue resident marker,⁶ while infiltration of regulatory T cells has an opposite impact, and the contribution of B cells remains undefined.⁷ Cells from innate immunity, such as natural killer (NK) cells, dendritic cells (DCs), and macrophages contribute to improved outcomes, except for macrophage subsets that polarize from tumor-inhibiting (M1) to tumor-promoting (M2) phenotype and are associated with disease progression.⁸ Moreover, increased expression of PD-1 and PD-L1 is one of the inhibition mechanisms of anti-tumor response by induction of peripheral tolerance, and TME has a significant role in its activation.⁹ Several studies have examined the feasibility of using PD-1 and PD-L1 to serve as prognostic biomarkers for ovarian carcinoma, although their role is still controversial.

Ascites is the most common sign of advanced ovarian carcinoma. Over the last years, studies have demonstrated that ascites contains almost the same immune cells (ICs) and extracellular components as the primary tumor.^2,10,11 However, there is a lack of quantitative data about the percentages of ICs in ascites and data on their clinical importance. Moreover, the role of immune checkpoints is also poorly described. Our objective was to quantitate ICs in HGSC ascites at disease presentation, assess the expression of PD-1 and PD-L1 on ICs, and investigate their prognostic significance for PFS and OS.

HGSC is the most aggressive gynecological malignancy which is usually diagnosed at advanced stages when the disease has already spread in the peritoneum.¹⁴ Ascites is therefore often the first sign of the disease.¹⁵ We hypothesized that ICs in ascites might be a promising source of novel prognostic markers for HGSC. We assessed the presence of different ICs together with CD103, PD-1, and PD-L1 expression levels and showed that percentages of cytotoxic ICs (CD8⁺, CD56^brightCD16⁻ NK cells) as well as macrophages, might affect patient's survival. We also showed that DCs are independent prognostic marker for PFS of HGSC patients.

As expected, our results on clinical and treatment characteristics of HGSC patients included in the study aligned with the already published data on the impact of age at diagnosis, FIGO stage, surgery, residual disease, and chemotherapy and maintenance therapy.^2,3,4 This data confirms the adequacy of our analyzed patient cohort.

According to the evaluation of ICs in ascites, our findings demonstrated that CD3⁺ T cells (median 51%) are the predominant population in the ascites of HGSC patients (FIGO stage ≥ III, ascites collected before initiation of treatment), with CD4⁺ significantly more abundant than CD8⁺, while the amount of Tregs was low as well as the other ICs investigated. Auer et al. reported higher percentages of CD3⁺ (median 80%) compared to our results, with an equal ratio of CD4⁺ and CD8⁺ subsets, and higher percentages of NKT cells (median 15%).¹ They did not specify patient's FIGO stage and when ascites was collected, which could explain the difference in the results of our and their study. However, the percentages of Tregs, NK cells, and B cells in their study were in concordance with our findings.¹ Similar percentages of CD4⁺ and Tregs, T cells, and DCs as in our study were reported in other studies.^2,16 We detected low percentages of macrophages (median 5%) in our series of ascites samples; the majority of them were M1-like. On the contrary, Steitz et al. described roughly 70% of macrophages, with equal ratios of both M1-like and M2-like subsets. Unfortunately, they did not describe disease progression by FIGO stage and ascites collection time either.¹⁷ Therefore, we speculate that lower percentages of macrophages in our study might be related to the inclusion of ascites samples from HGSC patients at the time of diagnosis, and not later at disease progression.

In the literature, we have found some data about percentages of ICs in HGSC ascites, but to our knowledge, there was no data providing information about the association of ICs at disease presentation and treatment assigned to the patients. Interestingly, in our study, we observed higher percentages of CD103⁺, CD8⁺, Tregs, NK cells, and DCs, and lower percentages of CD4⁺ cells in the ascites of patients with less tumor burden that underwent primary surgery compared to inoperable patients, which due to the size of the tumor were no eligible for surgery. These results might indicate an association of the amount and cell type of ICs in ascites at disease presentation with the extent of the tumor burden. Furthermore, we observed higher percentages of CD103⁺ T cells and DCs in patients who underwent interval surgery compared to the inoperable group. This data suggests the possibility of using these ICs to help us predict which patients, after receiving neoadjuvant chemotherapy, are likely to be eligible for surgery later on and have a lower amount of residual tumor. However, a much larger patient cohort is needed to confirm these findings. And as mentioned above, we have not found any similar studies to compare our results with.

Furthermore, most of the research on the influence of ICs on HGSC patient's survival is carried out on primary tumor tissues, and very little is known about the role of ICs in ascites. For instance, it has been reported that T cells in primary tumors improve the survival of HGSC patients. In fact, CD8⁺ cells correlated with improved survival, and Tregs, as well as CD4⁺ cells were seen as an indicator of poor prognosis.^18,19 On the other hand, studies on T cells in ascites have failed to confirm this correlation, even though a trend towards improved survival in patients with low CD4⁺ T cells was reported.² However, the ratio between CD8⁺ and CD4⁺ T cells or even Tregs has been reported as a more appropriate indicator of better OS.^20,21,22,23 In our study, similarly, we observed an association with significantly longer OS for patients stratified by having low CD4⁺ and high CD8⁺ T cells, and high CD8/CD4 index compared with the patients stratified by having high CD4⁺ and low CD8⁺ T cells, and also a low CD8/CD4 index. Regarding NKT cells, data on their role in ovarian carcinoma survival is generally limited. According to our results, there was not a significant correlation between NKT cells and survival rate for patients stratified as having high percentages of NKT cells. We also investigated the role of CD103 on T cells in HGSC ascites. CD103 is a subunit of the αE/β7 integrin that helps to retain expressing cells on the epithelium.²⁴ CD103 has been proposed as a marker of activated and tumor-reactive CD8⁺ T cells in ascites HGSC^25,26 but no data correlated with survival was given. We showed that CD103 was mostly expressed on CD8⁺, and not on CD4⁺ subsets, which was shown by us and by two other studies.^26,27 Furthermore, patients stratified by having high percentages of CD103⁺CD3⁺ T cells in HGSC ascites were associated with better PFS and OS. CD103⁺CD3⁺ T cells in ascites seem to have the same potential of prognostic information as reported for the CD103⁺ tumor-infiltrated T cells in the primary tumor²⁷ and we speculate that these cells in the ascites might be involved in the improvement of the antitumor response in the peritoneum. Of course, a larger patient cohort and additional tests are needed to gain a more comprehensive understanding of the significance and role of CD103⁺CD3⁺ T cells in HGSC ascites.

The role of B cells regarding their contribution to impaired antitumor immunity in HGSC has not been investigated as much as the role of T cells. However, there are few reports showing a trend towards worse OS in patients with high infiltration of B cells in ascites. These findings are consistent with ours.^28,29 Interestingly, opposite findings have been reported in primary tumors where a high percentage of B cells correlated with favorable survival, indicating that more studies are needed to estimate the role of B cells in ovarian tumors.^30,31

DCs in ascites have been poorly investigated. Only one study described a trend of high percentages of DCs in HGSC ascites toward a better survival outcome.² Similarly, we confirmed a significant association between DCs and patient's survival. Consistently, patients stratified by having high percentages of DCs were associated with improved prognosis in the primary tumors as well.^31,32

NK cells have attracted attention due to their ability to kill tumor cells without prior sensitization. There is limited data on the contribution of NK cell immunity to the clinical outcome of ovarian carcinoma. Infiltration of NK cells in primary tumors has shown a contradictory impact on survival outcomes in HGSC.³³ However, recently, one study showed an association of high percentages of CD56⁺ NK cells in ascites with better PFS and OS.¹⁴ Similarly, we showed the same association of both CD56^brightCD16⁻ and CD56^dimCD16⁺ NK cells with the survival outcome in our patient cohort. It is generally thought that CD56^brightCD16⁻ NK cells have a higher capacity for cytokine production and have mainly proliferative potential, and on the contrary, CD56^dimCD16⁺ NK cells have weak cytotoxic activity,^34,35 which explains why we observed significant results for CD56^brightCD16⁻ and only a trend towards CD56^dimCD16⁺.

Macrophages in ovarian ascites are gaining a lot of attention in recent years, due to their plasticity to switch from antitumor M1 to protumor M2 phenotypes.⁸ M2-type macrophages have been characterized by the expression of markers such as the scavenger receptors CD206 or CD163.³⁶ Published data is speculating that M2-type macrophages are taking the main role in immune suppression and angiogenesis promotion to sustain tumor growth.^37,38 Even though we identified lower percentages of macrophages than reported, we showed that patients stratified by having higher percentages of M1-like macrophages than M2-like macrophages were associated with better survival. We speculate that when the diagnosis is given, even though the total macrophage count is low, M1-like macrophages are predominant and are probably the ones contributing to a better outcome. However, during disease progression, M2-like macrophages outnumbered the M1 subset and most probably contributed to tumor progression and poor outcome of the disease.³⁹

Immune tolerance is defined by the inability of ICs to express immune checkpoints such as PD-1 and PD-L1. PD-1 receptor is an inhibitor of both adaptive and innate immune responses and can be expressed on CD8⁺ T cells, CD4⁺ T cells, and Tregs in ovarian tumors, whereas PD-L1 is expressed on activated T cells, tumor-infiltrating macrophages or fibroblasts, contributing to tumor immune escape.⁴⁰ However, the expression of PD-1 and PD-L1 in HGSC ascites and its correlation with survival has not yet been fully investigated. In the present study, we showed that PD-1 expression is present in almost all ICs, except on NK cells and B cells (less than 1%). We detected roughly 20% PD-1 expression on CD4⁺, CD8⁺ T cells and Tregs. However, Imai et al. reported 2x higher level of PD-1⁺CD4⁺ and PD-1⁺CD8⁺ cells in the ascites.41 Possible reasons for this discrepancy might be the different PD-1 clone selections for the analysis, as well as the diversity of the patient cohort. Imai et al. performed the analysis on different types of malignant epithelial ovarian carcinomas, and a few cases on borderline and benign tumors. Similar to us, they did not find a correlation between survival and PD-1 expression on CD4⁺ and CD8⁺ T cells. However, Sato et al. found an association between CD8⁺ cells and PD-1 expression in advanced epithelial ovarian carcinoma.²⁰ On the contrary, Pawłowska et al. demonstrated an association of high percentages of both PD-1⁺CD4⁺ and PD-1⁺CD8⁺ cells in ascites with worse outcomes, indicating a negative regulation of the anticancer immune response and exhaustion of T cells in the ascites.⁴² Results on primary ovarian tumors have also reported a correlation of higher PD-1 expression on T cells with shorter survival and worse prognosis.^43,44 For the other ICs in ovarian carcinoma ascites, data on PD-1 expression and survival correlation is also missing. We showed that high expression of PD-1 on Tregs and CD56^brightCD16⁻ NK cells is associated with better survival. We do not know how to interpret the correlation of high PD-1 expression with better instead of worse survival as expected. Additional studies are necessary to clarify if PD-1 expression on Tregs and CD56^brightCD16⁻ NK cells could be a positive prognostic marker for patient survival. We are also the first to confirm that PD-L1 is not expressed on ICs in HGSC ascites, since we did not find data on PD-L1 expression on ICs in ascites. One study on primary tumors has found that almost 2/3 of the tumors had a low level of PD-L1 expression, mainly on ICs rather than tumor cells, and the expression of PD-L1 was associated with significantly worse prognosis,⁴⁵ indicating location-dependent loss of expression of PD-L1 on ICs in ascites.

Nevertheless, it is worth mentioning that all published data on ovarian carcinoma also reports results on low and high percentages of ICs in ascites in correlation with patient survival without stratifying patients in subgroups according to their treatment characteristics. This is due to the low number of HGSC patients^{2,16,26,27,28,32} each research group confronts, and also the reason why we did not conduct that kind of analysis within the treatment subgroups either. Yet, the multivariate analysis indicated that residual tumor is the only independent prognostic marker for PFS and OS, and DCs are an independent prognostic marker for PFS only. We believe that multicentric studies on large patient cohorts could give more accurate information on the prognostic meaning of DCs and other ICs in the ascites.

In conclusion, we found that CD3⁺ were the predominant cells in HGSC ascites at disease presentation and showed that high levels of CD103⁺CD3⁺ T cells, CD56^brightCD16⁻ NK cells and DCs improve both PFS and OS, whereas high levels of CD8⁺, CD8/CD4 index, macrophages, PD-1⁺ Tregs and PD-1⁺CD56^brightCD16⁻ NK cells, and low levels of CD4⁺ improve OS only. We also confirmed that the residual disease is the only clinical independent prognostic marker for PFS and OS, and we showed that DCs are the only ICs that might become an independent prognostic marker for PFS. Data obtained highlight the potential of ascites as a source to provide additional prognostic information for HGSC patients. However, a larger patient cohort and longer follow-up are necessary to assess the independent prognostic significance of ICs together within different treatment characteristics.