Research Article

The Ratio between Monocytic Myeloid Derived Suppressor Cells/Monocytes and Dendritic Cells is A Prognostic Intratumoral and Blood Biomarker for Survival of Epithelial Ovarian Cancer Patients

by Anne F. de Groot1, Saskia J. Santegoets1, Ziena Abdulrahman1, Eveline M. Dijkgraaf2, Judith R. Kroep2, Sjoerd H. van der Burg1*

1Department of Medical Oncology, Oncode Institute, Leiden University Medical Center, P.O. Box 9600, 2300 RC Leiden, the Netherlands

2Department of Medical Oncology, P.O. Box 9600, 2300 RC Leiden, Leiden University Medical Center, the Netherlands

*Corresponding author : Sjoerd .H. van der Burg, Department of Medical Oncology, Leiden University Medical Center, P.O. Box 9600, 2300 RC Leiden, the Netherlands.

Received Date: 5 June 2024

Accepted Date: 10 June 2024

Published Date: 13 June 2024

Citation: de Groot AF, Santegoets SJ, Abdulrahman Z, Dijkgraaf EM, Kroep JR, et al. (2024) The ratio between monocytic myeloid derived suppressor cells/monocytes and dendritic cells is a prognostic intratumoral and blood biomarker for survival of epithelial ovarian cancer patients . Gynecol Obstet Open Acc 8: 202. https://doi.org/10.29011/2577-2236.100202.

Abstract

Predictive biomarkers are essential for tailoring therapy of patients with epithelial ovarian cancer (EOC). In this study, various immune populations in both tumor tissue and blood samples from EOC patients were analyzed to identify an easy accessible biomarker.

The frequencies of circulating immune cells were determined by flow cytometry and subsequent analysis by opt SNE and FLOWSOM using OMIQ in 36 EOC patients. Primary tumor material (n = 21) was analyzed by multiplex immunofluorescence using Vectra imaging and InForm software. Data were related to overall survival using Kaplan–Meier curves and log-rank testing.

A low ratio between mMDSC to DC (p = 0.0004), total monocyte to DC (p = 0.0026), mMDSC to CD19hi B cells (p = 0.0013) and monocyte to CD19hi B cells (p = 0.003) in blood was associated with prolonged survival in EOC patients. Importantly, low mMDSC to DC (p = 0.006) and CD14+HLA-DR+ monocyte to DC ratios (p = 0.006) in tumor tissue were also strongly prognostic for better survival. 

This is the first study showing that both the intratumoral and systemic balance between two cell types functions as prognostic biomarkers for EOC survival. Validation of their predictive power is ongoing in a prospective study and clinical immunotherapy trial.

Keywords: epithelial ovarian cancer, mMDSC to DC ratio, monocyte to DC ratio, prognostic, biomarker, overall survival 

Introduction

The majority of patients with epithelial ovarian cancer, fallopian tube cancer, and primary peritoneal cancer (all abbreviated as EOC) present with advanced disease. Treatment is mainly based on chemotherapy in combination with surgery and maintenance PARP inhibition [1]. Although most patients have an initial good response, around 70-80% of the patients will relapse with subsequent short survival [2,3]. Therefore, there is a need for clinical biomarkers to identify the poor responding patients as they may prefer treatment by experimental strategies.

The tumor microenvironment (TME) is composed of a wide variety of immune cells, and is well studied since subsets of immune cells have been shown to be prognostic and/or predictive for clinical outcome in the response to (chemo)therapy [4,5]. Generally, EOC hosts an abundancy of immunosuppressive cell subsets in the TME, such as tumor-associated macrophages (TAMs), myeloidderived suppressor cells (MDSCs) and regulatory T cells (Tregs). In addition, the TME can contain CD8+ cytotoxic T cells and dendritic cells (DCs) which are associated with better clinical outcome of patients after standard of care therapy [5-9]. 

To identify and monitor biomarkers, either as a prognostic tool or for prediction of therapy response, tumor sampling is often used. However, liquid biopsies have proven to be an attractive alternative since it is less invasive, inexpensive and easy to repeat [10,11]. Yet, it is essential to use blood biomarkers that are a good reflection of the TME [12-14]. Earlier we showed that the presence of high numbers of monocytes/macrophages and monocytic MDSCs (mMDSCs) in baseline blood samples was associated with reduced OS, whereas high levels of DCs were associated with an OS benefit. Importantly, the ratio of monocyte/ macrophage to DC, and in particular of mMDSC to DC, formed an independent prognostic factor for OS [15]. In the current study, we analyzed these myeloid immune populations in tumor tissue to investigate if they have the same prognostic capacity and thus, if the blood can be considered a reflection of the TME. Furthermore, we aimed to identify additional blood biomarkers. Here, we report that relatively low numbers of monocytes and mMDSCs as well as higher numbers of DCs were predictive for longer survival either when measured in the tumor or in the blood. Since these cells are also key to suppress or activate, respectively, tumor specific T-cell reactivity [15,16], they may also predict the response to immunotherapy.

Materials and methods

Phenotyping of PBMC

Immunostaining of PBMC and acquisition by multiparameter flow cytometry was previously performed [17, 18]. Data obtained with the monocyte/macrophage panel consisting of the live/dead marker yellow amine reactive dye, CD3, CD1a, CD11b, CD11c, CD16, CD19, CD45, HLA-DR, CD163 and CD206 antibodies, was used to re-analyze the monocyte/macrophage, mMDSC and other immune cell subsets by high-dimensional single cell data analysis using the OMIQ data science platform (n = 36). To this end, CD45+ cells were manually gated using FlowJo software version 10.8, after which newly generated FCS files were used for analysis by automated optimized parameters for T-distributed stochastic neighbor embedding (optSNE) followed by clustering in a self-organizing map (FlowSOM) using OMIQ data analysis software (www.omiq.ai).

Patients and tumor material

Formalin-fixed paraffin-embedded (FFPE) samples from primary tumors from EOC patients treated within the PITCH and/or CHIP study were collected (Table S1). From 21 out of 36 patients of whom PBMC phenotyping was performed, tumor material was available for analyses.

Eligibility criteria for these two trials have been described previously

[17,18]. In summary, the PITCH study was a phase I doseescalation study where patients with recurrent platinum sensitive EOC received standard chemotherapy (carboplatin/(pegylated liposomal)doxorubicin) with or without tocilizumab. The highest dose of tocilizumab was combined with interferon alpha. The CHIP study was a phase I/II study in patients with recurrent platinum resistant EOC in which patients received standard chemotherapy (gemcitabine), whether or not in combination with interferon alpha with or without a vaccine against p53. Both studies were conducted in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee Leiden, in agreement with the Dutch law for medical research involving humans, and registered in the clinical trial register (PITCH study: NCT01637532 and CHIP study: NTC01639885).

Multiplex immunofluorescence imaging of tumor tissue

A seven-color multispectral immunofluorescence panel was used for identification of mMDSCs,

DCs, macrophages and monocytes in tissue, consisting of antibodies against CD1c, CD163, CD11c, CD14, CD68, HLA-DR and DAPI. The panel is a combination of indirect (CD14, CD163) and direct (CD11c, CD68, HLA-DR) detection, and tyramide signal amplification with Opal (CD1c) (PerkinElmer, Waltham, Massachusetts). An overview of the panel can be found in Table S2.

Immunofluorescent stains were performed on 4µm tissue sections from FFPE blocks with tumor material from debulking procedures. Tissue sections were deparaffinized and endogenous peroxidase was blocked with hydrogen peroxide, after which epitope retrieval was performed with tris-EDTA (10 mM/1 mM, pH 9.0). Nonspecific binding sites were blocked with SuperBlock (ThermoFisher Scientific). Antibodies were applied following the recommended protocols in the following order: CD1c, unconjugated antibodies (CD14 and CD163 with overnight incubation) and conjugated antibodies (CD11c, HLA-DR, CD68 with 5h incubation). Finally, DAPI was used for nuclear staining. Slides that underwent the complete immunofluorescent staining procedure without primary antibodies served as negative controls, human tonsil slides served as positive control. 

Enumeration of immune cells in tumor sections

The numbers of mMDSCs, DCs, macrophages and monocytes were counted using the Vectra 3.0.5, an automated quantitative pathology imaging system, with InForm 2.4 software (PerkinElmer, Waltham, Massachusetts). mMDSCs were characterized as CD68CD14+HLA-DRCD11c-CD1c-, DCs as CD68-CD14-HLADR+CD11c+, M1-like macrophages as CD68+CD163-, M2like macrophages as CD68+CD163+, inflammatory macrophages as CD68+CD14+HLA-DR+, total macrophages as CD68+ and monocytes as CD68-CD14+HLA-DR+. In brief, slides were scanned for image acquisition (×4 magnification) after which five multispectral images per slide on average were selected manually (×20 magnification, one image representing roughly 0.33 mm2). Using the semi-supervised deep learning based InForm software, the software was trained to segment stroma, tumor epithelium, and empty tissue regions, and to distinguish between mMDSCs, DCs, macrophages, monocytes and other cell phenotypes. To improve the accuracy of the training process, the training was divided in three sub-analyses (HLA-DR and CD14, CD68 and CD163, CD68 and CD163) after which the phenotypes of the individual sub-analyses were merged per cell based on their X,Y‐positions with an R script to obtain the complete expression profile of each individual cell. Finally, the number of immune cells were calculated as number per mm2 stroma, mm2 tumor and mm2 stroma plus tumor.

Statistics

Statistical analyses were performed using SPSS (v.25.0 for Windows, IBM SPSS statistics). Patients from whom tumor material appeared non-evaluable after staining were excluded from analyses. Survival analyses were performed using Kaplan–Meier curves and log-rank testing. A variable for each individual immune cell population (low/high) was calculated using the median as cutoff value, since immune cell counts were not normally distributed. For the ratios, first an individual ratio per patient was calculated, after which the median ratio was used to differentiate between a low or high ratio. The clinical endpoint examined was OS, defined as time from surgery to death due to any cause. Correlations between different populations were tested using non-parametric Spearman r correlation analysis.

Results

Low blood mMDSC to DC, total monocyte to DC, mMDSC to CD19hi B cell and monocyte to CD19hi B cell ratios are associated with prolonged survival

In the past, we analyzed pre-defined cell populations on manually gated PBMC samples from EOC patients [15]. Newly developed state-of-the-art techniques such as opt SNE [19] and FLOWSOM [20] using OMIQ allow true objective and automatic analyses of the immune cell composition of PBMC, and these were utilized to re-analyze the data sets from 36 EOC patients of two phase I trials [17,18]. In addition, survival data were updated (cutoff date 2022-05-19). A total of thirteen individual immune cell clusters were identified, with CD14+HLA-DR+CD16- monocytes (population 2) and T-cells (population 7) being the two most substantial populations. CD14+HLA-DR- mMDSCs (population 1), CD14+CD16+ monocytes (population 3), B cells (populations 8 and 9), NK and NK/T cells (populations 11 and 12) were also prominently present (Figure 1A-C, Figure S1A). When patients were grouped into low or high according to the median frequency of each population, high frequencies of circulating CD14+HLA-DR- mMDSCs (population 1), CD14+HLA-DR+CD16- monocytes (population 2) and CD14+HLA-DR+ total monocytes (populations 2, 3 and 5) were associated with worse overall survival (OS), while high frequencies of circulating DCs (population 4) and strongly CD19 expressing B cells (population 9) were associated with OS benefit (Figure 1D, Figure S1B).

Figure 1: Circulating immune cell subsets associated with short or prolonged survival in EOC. Baseline frequencies of immune cells were determined in platinum-sensitive and –resistant. Epithelial Ovarian Cancer (EOC) patients using flow cytometry data acquired earlier [15]. Data was analyzed by subsequent optSNE and FLOWSOM using OMIQ. A) optSNE plot visualizing the FLOWSOM cluster for all EOC patients. B) Graph displaying the frequencies of the identified FLOWSOM clusters as percentage of viable cells for each patient. C) Phenotypic plots visualizing the expression of the indicated markers for the different immune cell clusters. KaplanMeier survival plots of 36 EOC patients for D) mMDSC (left), total CD14+ monocytes (center left), DC (center right) and CD19hi B cell (right) and E) mMDSC to DC ratio (left), total monocytes to DC ratio (center left), mMDSC to CD19hi B cell ratio (center right) and total monocytes to CD19hi B cell ratio (right). Patients were grouped into high or low groups according to the median frequency of the indicated myeloid cell subpopulations. The solid line depicts patients with frequencies above the median and the dotted line depicts patients with frequencies below the median. Statistical significance of the survival distribution was analyzed by log-rank testing.

Since these immune cell populations were associated with opposite clinical outcome, the ratios between mMDSCs and DCs or CD19hi B cells (population 9) as well as total monocytes (having a better separation of the survival curves than CD14+CD16- monocytes) and DCs or CD19hi B cells (population 9) were calculated. The differences in median OS were even more prominent when a low mMDSC to DC (p = 0.0004), a low total monocyte to DC (p = 0.0026), a low mMDSC to CD19hi B cell (p = 0.0013) or a low monocyte to CD19hi B cell ratio (p = 0.003) was used as biomarker (Figure 1D). Thus, the mMDSC to DC, monocyte to DC, mMDSC to CD19hi B cell and monocyte to CD19hi B cell ratio in the blood displayed the best prognostic capacity for survival in EOC patients.

A variety of myeloid immune cell populations infiltrate EOC tumors

To investigate whether the prognostic immune subsets in PBMCs have the same prognostic impact when measured in tumor tissue and thus may reflect what is locally going on, primary tumor material from 21 of the 36 patients from which PBMCs were analyzed could be collected, stained with a multispectral myeloid cell panel consisting of antibodies against CD1c, CD163, CD11c, CD14, CD68, HLADR and DAPI and analyzed. Material from the other patients was not available or not enough tumor was left for analysis. Patient and tumor characteristics are described in Table S1. Representative images of immunofluorescent stains are shown in Figure 2 and Figure S2.