Females experience a greater benefit from surgical resection in an immune-dependent manner in a preclinical glioblastoma model

Abstract

Introduction

Glioblastoma (GBM) is 1.6 times more common in males, who also have a shorter survival. Recent evidence has demonstrated sex differences in the immune response to GBM. While preclinical models are often used to interrogate GBM, resection is often not performed, despite its role in standard of care. It is also unclear how biological sex impacts outcome after resection in preclinical models.

Methods

To address this question, we developed a preclinical resection model in which mice were implanted with syngeneic GBM cells and underwent tumor resection. RNA sequencing was used to characterize transcriptional changes, immune phenotype was investigated via flow cytometry, and patient outcomes were analyzed using the National Cancer Database.

Results

In an immunocompetent model, females survived longer than males post-resection. The survival bias favoring females was abrogated in immunocompromised mice, suggesting an immune-mediated mechanism for prolonged survival in female mice. After resection, immunocompetent female mice had higher immune cell infiltration in the resection cavity, and males had notably decreased regulatory T cells (Tregs). Correspondingly, when Tregs were depleted in both male and female mice, the sex bias was abrogated. Finally, GBM patients did not display a survival sex bias post-biopsy, but in patients who underwent resection, females survived longer.

Conclusions

These results demonstrate an immune-dependent sex bias in survival post-resection and suggest that Tregs may serve a protective role for survival post-operatively. These data underscore the complexity of sex bias and the immune response in GBM and the need to account for biological sex and resection in preclinical models to investigate next-generation therapies.

Key Points.

• A preclinical GBM resection model demonstrates an immune-dependent sex bias in survival.

• T regulatory cells may serve a protective role for survival post-operatively following GBM resection.

Importance of the Study

Glioblastoma (GBM) portends a dismal prognosis and is 1.6 times more common in biological males, who have a shorter overall survival than biological females. In a preclinical GBM resection model, we showed that female mice survive longer than males after resection. This sex bias was dependent on the immune system, and male mice had decreased Treg cells post-resection. When Tregs were depleted, the sex bias in survival was abrogated. Analysis of GBM patients in the National Cancer Database found that while female patients survive longer than male patients after resection, there is no sex bias in survival after biopsy alone. These results demonstrate an immune-dependent sex bias in survival after resection. Tregs may serve a protective role post-operatively and should be further studied, preclinically and in perioperative patients, to better understand how and when they become important for survival.

Glioblastoma (GBM) is the most common primary malignant brain tumor and has a median survival of 15-20 months despite standard of care treatment, which includes surgical resection, radiation, and chemotherapy.^1–4 GBM is 1.6 times more common in males, and overall survival can be up to 10 months shorter in males than in females.⁵ Females are more responsive to standard-of-care therapy, but the mechanism behind these differences has yet to be fully elucidated.⁶ While the immune system is known to be suppressed in GBM patients,⁷ recent studies from our group and others have also found sex differences in immune response.^8-11 For example, males have elevated levels of monocytic myeloid-derived suppressor cells (MDSCs) in their tumor microenvironment, while females have an increase in granulocytic MDSCs in their periphery, suggesting that MDSC-mediated tumor progression occurs in a sex-dependent manner.¹⁰ These initial observations necessitate the consideration of sex-specific treatment targets.¹¹ Recent findings in GBM preclinical models and humans demonstrate that females have higher T cell infiltration and more CD8+ effector cells, respectively,⁹ further underscoring the complexity of sex differences in GBM and the need to account for these parameters in preclinical models and next-generation therapeutic strategies.

While preclinical GBM models are used to interrogate pathophysiology and response to therapeutic interventions, few of these models include surgical resection, despite its role in the standard of care for patients and as a defining therapy in prolonging survival.¹² Using a preclinical GBM resection model, we previously discovered that male mice had decreased circulating CD8+ T cells in their blood following tumor resection, as compared to sham surgery, demonstrating that surgery impacts immune response and is essential to incorporate into preclinical models.^7,13 Many previously published preclinical GBM resection models utilize immunocompromised rodents (both mice and rats), which has limitations given the important role that the immune system plays in tumor progression and survival.^14-18 Moreover, sex differences have not been accounted for in these resection models.^{7,14-17,19-21}

To date, there have been no immunocompetent preclinical models of GBM used to study sex differences in survival and immune phenotype after surgical resection.¹³ Based on recent studies in other tumors and from our group in GBM, we hypothesized that females have a survival advantage due to a sex-biased immune response that is amplified by surgical resection.^8-11,22,23 To address this hypothesis, we created a preclinical model of GBM resection using male and female immunocompetent mice implanted with syngeneic GBM cells.

Methods

Cell Culture

SB28 cells were a kind gift from Dr. Hideho Okada, University of California San Francisco. The SB28 cell line was created via the intraventricular injection of pT2/C-Luc/PGK-SB100 and sleeping beauty transposon (SB)-flanked proto-oncogenes (pT2/CAG-NRasV12 and pT2/shp53/mPDGF) into a neonatal C57BL/6 mouse followed by harvesting of the tumor cells.²⁴ SB28 cells have been noted to demonstrate some invasion into surrounding brain tissue.²⁵ SB28 cells were sent to KaryoLogic Inc. for karyotyping, and no Y chromosome was observed (Supplementary Figure S1). Mouse GBM cells were grown in incubators at 37°C and 5% CO₂. The cells grew adherently on tissue-culture plates in RPMI-1640 medium (Media Preparation Core at Cleveland Clinic Research) with 10% fetal bovine serum (Thermo Fisher Scientific) and 1% penicillin/streptomycin (Media Preparation Core at Cleveland Clinic Research).

NPD cells were a kind gift from Dr. Maria G. Castro, University of Michigan. The NPD cell line was derived from genetically engineered tumors generated by the SB transposase system as well, and their genotype is shp53, NRAS, and PDGFβ.^26,27 These cells were karyotyped by KaryoLogic Inc., and no Y chromosome was found (Supplementary Figure S1). Cells were grown in F12 media with L-glutamine (prepared at Media Preparation Core at Cleveland Clinic Research), supplemented with 1% penicillin/streptomycin (Media Preparation Core at Cleveland Clinic Research), 1% B27 (ThermoFisher Scientific), 0.5% N2 (ThermoFisher Scientific), 20 ng/mL EGF (R&D Systems) and 20 ng/mL FGF (R&D Systems). Cells were grown in adherent conditions by pre-incubating tissue culture-treated surfaces with Geltrex (ThermoFisher Scientific) diluted 1:500 in null F12 media.

Mouse Models

Male and female C57BL/6J mice from The Jackson Laboratory (000664) were purchased at the age of 6 weeks and implanted with syngeneic cells between the ages of 6 and 12 weeks. Male and female NOD scid gamma (NSG) mice from The Jackson Laboratory (005557) were purchased at the age of 6 weeks and implanted with cells between the ages of 6 and 8 weeks. Foxp3^DTR mice that express a diphtheria toxin receptor following the Foxp3 promoter were obtained from The Jackson Laboratory (#016958) and bred in the Biological Research Unit at Cleveland Clinic Research. Age- and sex-matched Foxp3^DTR mice between the ages of 7 and 12 weeks were used. All animal experiments were performed under approved Institutional Animal Care and Use Committee protocols.

Resection Model

Age-matched male and female mice were each implanted with 15 000 SB28 cells in 5 µL of null RPMI-1640 media or 5000 NPD cells in 5 µL of null F12 media using a stereotactic frame. The mice were anesthetized with 1%-3% isoflurane and 2 L/min of 100% oxygen. Once adequate anesthesia was confirmed, the mice were immobilized on a stereotactic frame. The tumor cells were implanted in the right hemisphere of each mouse’s brain approximately 0.5 mm rostral and 2 mm lateral to the bregma at a 1 mm depth from the scalp, using an insulin syringe.

On day 7 post-implantation, tumors were resected in half of the mice using methodology previously described in the Lathia laboratory.⁷ Sham surgery was performed on the remaining implanted mice. All surgeries were conducted with the use of an operative microscope (Zeiss). Briefly, the resection consisted of anesthetizing the mouse with 1%-3% isoflurane and 2 L/min of 100% oxygen. After corneal and pedal reflexes were lost, the mouse was immobilized in a stereotactic frame while in a nose cone providing continuous anesthesia with 2% isoflurane and 2 L/min of 100% oxygen. After applying ointment to the eyes, clipping fur on the head, and sterilizing with alcohol and iodine, bupivacaine was injected, and a small linear incision was made on the right side of the scalp. The periosteum was cleaned off, and a burr hole was created using a 1 mm round burr and a micro-drill (Braintree Scientific). This burr hole was centered around the prior implantation hole from the insulin syringe. After the bone was removed, a corticectomy was created using a micro-spatula (Fine Science Tools). For the sham surgeries, at this point, hemostasis was obtained with irrigation, monopolar cautery, and placement of Surgicel over the craniectomy site. For the resections, at this point, the micro-spatula was used to create planes around the tumor, which had some invasion into the surrounding parenchyma, and a glass Pasteur pipet attached to gentle wall suction was used to aspirate the tumor. After satisfactory tumor resection was achieved using gross inspection under the operative microscope, irrigation and monopolar cautery were used to achieve hemostasis, followed by placement of a piece of Surgicel over the craniectomy site. The scalp was closed in a simple, running fashion with 5-0 PROLENE suture. Next, the mice received subcutaneous normal saline for hydration and extended-release buprenorphine for pain control. Mice were closely monitored in their cage over a warming blanket while they awoke from anesthesia. Post-operatively, all mice were monitored for signs of neurologic endpoint, which included poor grooming, weight loss, lethargy, hunched posture, loss of coordination, poor grasp, etc. Once mice demonstrated signs of endpoint, they were euthanized with CO₂ asphyxiation and cervical dislocation. All survival experiments were analyzed using the Kaplan-Meier survival analysis and log-rank test.

Foxp3^DTR Model

For the Treg depletion animal studies, age- and sex-matched Foxp3^DTR mice between the ages of 7 and 12 weeks were used. For mice who underwent depletion, 1 µg diphtheria toxin was administered per mouse via intraperitoneal (IP) injection. The diphtheria toxin (Sigma) was suspended as a 1 mg/mL solution. A total of 1 µL of this solution was added to 99 µL 1X D-PBS, and the 100 µL solution was administered to each mouse undergoing depletion. For the placebo-controlled mice, 100 µL of 1X D-PBS was administered via IP injection every time diphtheria toxin was administered to the depleted mice. All mice underwent an IP injection 2 days prior to implantation of the syngeneic tumor cells, on the day of implantation of the tumor cells, and then every 7 days after tumor cell implantation. Treg depletion was confirmed via flow cytometry on blood of a mouse who received the diphtheria toxin injections. These mice were followed for neurological endpoint, similarly to how the wildtype and NSG mice were followed post-operatively.

Imaging

Seven days post-implantation with syngeneic tumor cells, age- and sex-matched mice underwent MRI for confirmation of tumor presence in the brains. The images were taken on a 7-Tesla MRI (Bruker BioSpec 70/20 USR) using a 23-mm volume coil setup. The Small Animal Imaging Core at the Cleveland Clinic was utilized. Briefly, the mice were anesthetized with 1-3% isoflurane and 2 mL/min 100% oxygen, which was maintained throughout the acquisition of the scan. Respirations and body temperature were monitored throughout the scan acquisition. T1 sequences pre- and post-administration of gadolinium via tail vein injection were obtained. The images were reviewed using 3D Slicer (v 5.4.0).

Bulk RNA Sequencing

For bulk RNA sequencing, 8-week-old C57BL/6J male and female mice were implanted with 15 000 SB28 cells in 5 µL of null RPMI in the right hemisphere, as described earlier in the methods. Seven days post-implantation, male and female mice underwent either tumor resection surgery or sham surgery. Another set of male and female mice did not undergo any surgery at this time, and they served as the non-operative arm of this study. Seven days post-surgery, a total of 14 days post-tumor cell implantation, the tumor and resection bed tissues were collected from the right hemisphere of the age- and sex-matched mice. This tissue was immediately rinsed with pre-chilled RNAase-free water. Next, the tissue was sliced into small pieces on dishes that were pre-chilled on dry ice. These small pieces (50-100 mg) were placed in pre-chilled RNase-free Eppendorf tubes with a screw cap and flash frozen in liquid nitrogen. Then, the tubes were stored in a −80 °C freezer until they were shipped on dry ice to Novogene for RNA extraction and bulk RNA sequencing using the NovaSeq X Plus from Illumina to a depth of approximately 20 million reads (Figure 2A). The genomes were aligned to mm39, and downstream analyses were performed. Differential expression analysis was performed using DESeq2 (v1.30.1) with the default parameters. Gene ontology analyses were performed using the default parameters. All plots were generated using R v4.0.5.

Figure 2.

Surgical resection alters gene expression in the tumor microenvironment. (A) Schematic detailing the experimental groups analyzed by bulk RNA-sequencing. (B) PCA plot of 6 experimental groups followed by Venn diagram showing the number of genes uniquely expressed in each experimental group, as well as those shared between groups. (C) Volcano plots demonstrating differentially expressed genes (DEGs) between male resection and female resection groups with X and Y chromosome genes removed. (D) Volcano plots demonstrating DEGs between male resection and male sham surgery groups. (E) Volcano plots demonstrating DEGs between female resection and female sham surgery groups.

Flow Cytometry for Immunophenotyping

The following steps were followed at either 3 days, 7 days, or 10 days post-operatively. At the indicated time post-operatively and post-implantation, mice were euthanized via CO₂ asphyxiation as approved by the Cleveland Clinic IACUC. The mice were pinned, the thorax was opened, and a small cut was made in the right atrium. Blood was collected with RAM Scientific Safe-T-Fill Capillary Blood collection tubes (Fisher Scientific). After incubation with red blood cell lysis buffer (BioLegend), the cells were transferred to a 96-well plate for staining.

After their blood was collected, the mice underwent perfusion with 10 mL of pre-cooled 1X D-PBS slowly injected into the apex of the left ventricle via a 20 G needle. After successful perfusion, the spleen was harvested from the left upper quadrant of the mouse and placed into 1 mL of pre-cooled Hank’s balanced salt solution (HBSS) in a single well of a 24-well plate. Next, the tibia was isolated from the right limb, and this was also placed in an HBSS-containing well in a 24-well plate on ice. Finally, the tumor with surrounding resection cavity in the tumor-bearing right hemisphere was carefully dissected and placed in an HBSS-containing well in a 24-well plate on ice.

Following the harvest of the aforementioned organs, processing ensued as noted. For the tumors, a single-cell suspension of cells from the tumor-bearing right hemisphere was prepared via enzymatic digestion with collagenase IV (Sigma) and DNase I (Sigma). These were digested in gentleMACS C tubes (Miltenyi) on a Miltenyi Biotec gentleMACS Octo Dissociator using the program “37C_abdk_1.” Following digestion, the tissue was filtered through a 70 µm strainer, and the cells were enriched via gradient centrifugation using a 30% Percoll solution (Sigma). After the enrichment, cells were incubated with red blood cell lysis buffer (Biolegend) and then washed and transferred to a 96-well plate for staining.

For the spleen, tissue was placed on top of a 40 µm strainer and gently mashed down with the wide end of a syringe plunger. The cells were washed through the strainer with HBSS. After centrifugation, cells were incubated with red blood cell lysis buffer (Biolegend) and then washed and transferred to a 96-well plate for staining.

For the tibia marrow, marrow was washed out of the harvested tibia and through a 40 µm strainer with HBSS and a 27 G needle. The cells were washed through the strainer with HBSS. After centrifugation, cells were incubated with red blood cell lysis buffer (Biolegend) and then washed and transferred to a 96-well plate for staining.

After all cells were transferred to the 96-well plate, they underwent live/dead staining with Live Dead blue fluorescent dye (Thermo Fisher, 1:1000 in 1x D-PBS) and were incubated with FCR blocking buffer (Miltenyi Biotec, 1:50 in 1X D-PBS) in the dark for 15 minutes on ice. After washing, the cells were incubated with surface antibodies conjugated to fluorochromes, diluted in Brilliant Stain Buffer (BD Biosciences) at 1:250. They were incubated on ice, in the dark, for 25 minutes. After washing, the cells were stored overnight at 4 °C in the dark in fixation/permeabilization concentrate and diluent (Thermo Fisher). The next morning, the cells were washed in 1X permeabilization buffer and incubated with fluorochrome-conjugated antibodies against intracellular markers, at concentrations varying from 1:100 to 1:250 in permeabilization buffer, for 45 minutes in the dark at room temperature. Stained cells were measured using the Aurora (Cytek) and analyzed using FlowJo software (v10.8.0, BD Biosciences). Cells were gated based on the strategy shown in Supplementary Figure S2.

Reagents

For immunophenotyping in this mouse model, the following fluorophore-conjugated antibodies at concentrations of 1:100—1:250 were used: CD11b (BD Biosciences, Cat# 563553), CD69 (BD Biosciences, Cat# 741234), CD11c (BD Biosciences, Cat# 612796), CTLA4 (BioLegend, Cat# 106312), Ly6G (BD Biosciences, Cat# 560603), PD1 (BioLegend, Cat# 135241), B220 (BioLegend, Cat# 103237), Ki67 (BioLegend, Cat# 652413), TIM3 (BioLegend, Cat# 119727), CD3 (BD Biosciences, Cat# 564379), CD45 (BioLegend, Cat# 103132), Foxp3 (Thermo Fisher 15-5773-82), LAG3 (BioLegend, Cat# 125224), MHCII (BioLegend, Cat# 107606), NK1.1 (BioLegend, Cat# 108716), CD4 (BioLegend, Cat# 100422), CD8 (BioLegend, Cat# 100712), CD206 (BioLegend, Cat# 141712), Lyg6C (BioLegend, Cat# 128024), CD68 (BioLegend, Cat# 137024), and F4/80 (BioLegend, Cat# 123118).

Retrospective Data Analysis

Data were retrieved from the National Cancer Database (NCDB), which reports data for >85% of new primary brain tumor diagnoses in the United States.²⁸ Because NCDB data are de-identified and publicly available, this study was conducted under a protocol determined to be exempt by the Duke University School of Medicine Institutional Review Board. Individuals 18 years old or older, diagnosed with a histologically confirmed IDH-wildtype GBM (identified using International Classification of Disease for Oncology, third edition histology code 9440/3 with Brain Molecular Markers Site-Specific Data Item indicating IDH-wildtype) between 2018 and 2020 who received chemotherapy and had radiation treatment lasting for at least 14 days were included. Patients were categorized into 2 groups, those who underwent neurosurgical resection and those who had biopsy (as indicated by histologic confirmation) but no resection. Survival was defined as the number of months between the month of diagnosis and date of death or date of last follow-up. Individuals were stratified by sex and age (grouped as 18-40 years, 41-50 years, 51-60 years, 61-70 years, 71-80 years, and 81-90 years). Retrospective survival analyses were conducted using Cox proportional hazard models to estimate hazard ratios (HR), 95% confidence intervals, and P value. All models were adjusted for age (unless stratified by age), race, Charlson comorbidity score, and Hispanic ethnicity. Data processing and analyses were conducted in R 4.3.

Statistical Analysis

GraphPad Prism (Version 10.1.2, GraphPad Software Inc.) was used for statistical analysis and creation of figures unless otherwise noted. All survival experiments were analyzed using the Kaplan-Meier survival analysis and log-rank test. One-way/two-way analysis of variance (ANOVA) testing was used with multiple comparisons to compare groups for the immune profiling experiments. P < .05 were considered statistically significant.

Results

Female Mice Demonstrate Longer Survival Post-Resection Compared to Male Mice in a Preclinical Model of GBM

To assess whether there is a sex bias in survival in this preclinical resection model of GBM, male and female immunocompetent C57BL/6 mice were implanted with a syngeneic GBM model, SB28, using a stereotactic frame (Figure 1A). The mice developed tumors in the right hemisphere of their brains by day 7 post-implantation, as verified by MRI (Figure 1B). Mice either underwent tumor resection or sham surgery and were then followed for signs of neurologic endpoint, at which time they were euthanized (Figure 1A). Kaplan-Meier survival analysis did not identify any statistically significant differences in survival between male and female mice who underwent sham surgery only (Figure 1C). In contrast, female mice who underwent tumor resection survived significantly longer post-implantation than male mice (32 days vs 24 days, P = .0424, Figure 1C). Similar results were obtained when we performed the experiment in a second GBM model, the NPD model (Supplementary Figure S3), further highlighting the sex bias in survival after tumor resection.

Figure 1.

There is a sex bias in survival after GBM resection. (A) Experimental schematic of how C57BL/6 mice were implanted with syngeneic tumor cells and underwent surgery 7 days post-implantation. Mice were then monitored for progression to neurologic endpoint. (B) Representative MRI images (T1 with gadolinium) 7 days post-implantation. The arrows designate tumor. (C) Kaplan-Meier survival curves demonstrating survival differences between male and female mice who underwent either sham surgery or tumor resection surgery after implantation with 15 000 SB28 cells. Significance calculated by log-rank test.

Surgical Resection Alters Gene Expression in the Tumor Microenvironment

To further explore how surgical resection of GBM tumor or sham surgery differentially impacted males and females, we performed RNA sequencing of ex vivo tumor specimens. This approach allowed for testing of whether there are changes on a transcriptional level in the GBM cells that could be driving the survival differences we observed. Male and female immunocompetent C57BL/6 mice were implanted with SB28 cells using a stereotactic frame. There were 6 groups investigated: male mice whose GBM tumors were resected (male resection), female mice whose GBM tumors were resected (female resection), male mice who underwent craniectomy and corticectomy without tumor removal (male sham), female mice who underwent craniectomy and corticectomy without tumor removal (female sham), male mice with a tumor who underwent no further surgery post-implantation (male non-operative), and female mice with a tumor who underwent no further surgery post-implantation (female non-operative). Seven days post-operatively (or 14 days post-implantation in the case of the non-operative groups), the tumors and their surrounding resection cavities were harvested from the right hemisphere. RNA was extracted and subjected to RNA sequencing (Figure 2A). Principal component analysis (PCA) demonstrated that the mice that underwent resection of their tumors (Rsxn) clustered together, away from the mice that underwent either sham surgery (Sham) or no surgery (NonOp) (Figure 2B). The mice that did not undergo resection of their tumors tended to cluster by sex (Figure 2B). Each experimental group had its own set of uniquely expressed genes, as well as genes uniquely expressed by different combinations of groups (Figure 2B and Supplementary Figure S4A and B).

Of the differentially expressed genes (DEGs) among experimental groups, there were fewer differentially expressed between male resection and female resection compared to the other pairs (Figure 2C-E and Supplementary Figure S4B). With genes found on the X and Y chromosome removed, the top upregulated DEGs in the male resection cohort compared to the female resection cohort were growth hormone (Gh) and Purkinje cell protein 2 (Pcp2) (Figure 2C). The top downregulated genes in the male resection cohort were C-X-C motif chemokine ligand 13 (Cxcl13), MyoD family inhibitor domain containing (Mdfic2), and special AT-rich sequence binding protein 2 (Satb2). No gene ontology terms were significantly different between the male and female resection groups based on the upregulated DEGs (Supplementary Figure S4C). For male resection vs male sham surgery groups, the top upregulated DEGs were lipoprotein lipase (Lpl), CD5 antigen-like (Cd5l), and 3-hydroxy-3-methylglutaryl-coenzyme A synthase 2 (Hmgcs2) (Figure 2D). The top downregulated DEGs were dentin matrix protein 1 (Dmp1), prostaglandin F receptor (Ptgfr), and notum palmitoleoyl-protein carboxylesterase (Notum) (Figure 2D). The top gene ontology terms for upregulated DEGs in male resection vs male sham surgery included terms involved in the immune response: leukocyte chemotaxis, myeloid leukocyte activation, positive regulation of cytokine production, and myeloid cell activation involved in the immune response (Supplementary Figure S4D). For female resection vs female sham surgery groups, the top upregulated DEGs were 5-hydroxytryptamine receptor 2B (Htr2b), Lpl, and matrix metallopeptidase 8 (Mmp8). The top downregulated DEGs were leucine rich repeat and Ig domain containing 3 (Lingo3), corticotropin releasing hormone binding protein (Crhbp), and ATP synthase membrane subunit 6, pseudogene (Gm10925) (Figure 2E). The top gene ontology terms for upregulated DEGs in female resection vs female sham surgery groups were all related to immune response and included: positive regulation of cytokine production, response to bacterium, leukocyte-mediated immunity, adaptive immune response, leukocyte activation involved in immune response, cell activation involved in immune response, leukocyte migration, regulation of leukocyte activation, positive regulation of immune response, and myeloid leukocyte activation (Supplementary Figure S4E). The lack of significant gene ontology terms identified by RNA-sequencing suggested that cell-intrinsic programs were not likely to be the dominant determinant of sex differences in the preclinical GBM resection model.

The Sex Bias in Survival After Resection Is Dependent on the Immune System

Our RNAseq analysis revealed that transcriptional changes in brain tissue after resection were primarily enriched in immune responses (Supplementary Figure S4D and E). To evaluate the role of the immune system on the sex bias in survival after surgical resection, male and female immunocompromised NSG mice were implanted with the syngeneic GBM model SB28 and underwent either tumor resection or sham surgery. Mice were then followed for signs of neurologic endpoint, at which time they were euthanized. No significant difference in survival was observed between male and female mice who underwent resection surgery or sham surgery only (Figure 3A). These survival studies demonstrate that the sex bias in survival is abrogated in an immunocompromised preclinical model of GBM resection.

Figure 3.

The sex bias in survival after resection is dependent on the immune system. (A) Kaplan—Meier survival curve demonstrating survival curves for male (blue) and female (red) NSG mice who underwent either sham surgery or tumor resection surgery 7 days after implantation with 15 000 SB28 cells. (B) Schematic detailing how C57BL/6 mice were implanted with syngeneic tumor cells, underwent surgery 7 days post-implantation, and then underwent organ harvest either 3, 7, or 10 days post-operatively for flow cytometry. (C) Flow cytometry results showing CD45^hi cell infiltration into the tumor cavity either 3, 7, or 10 days post-operatively in male sham surgery, male resection, female sham surgery, and female resection experimental groups, as a percentage of total live cells. ANOVA statistical testing did not demonstrate significantly different levels between the groups at these time points (P = .23, P = .57, and P = .38 at post-operative days 3, 7, and 10, respectively). (D) Flow cytometry results showing regulatory T cell infiltration into the tumor cavity as a percentage of CD4+ cells in male sham surgery, male resection, female sham surgery, and female resection experimental groups. For male sham vs male resection, Treg infiltration was higher and significant by unpaired T test, P = .0008. (E) Kaplan-Meier survival curve demonstrating survival for male (blue) and female (red) Foxp3^DTR mice that all underwent tumor resection and received intraperitoneal injections of either PBS (control dotted lines) or 1 µg diphtheria toxin to deplete regulatory T cells (solid lines).

Sex-Biased Immune Changes Were Identified following Tumor Resection

Next, flow cytometry was used to characterize the immune phenotype in immunocompetent male and female mice undergoing GBM resection over time. Tumor, blood, spleen, and tibia bone marrow were harvested from male and female C57BL/6 mice that underwent either surgical resection of their GBM tumor or sham surgery. Those tissues were then processed for flow cytometry analysis and stained with both surface and intracellular markers for myeloid and lymphoid cell populations (Figure 3B). Three days post-operatively, there were relatively low levels of CD45^hi cells as a percentage of total live cells infiltrating the tumor bed (Figure 3C). At 7 days post-operatively, male sham tumor-bed levels of CD45^hi cells were increased, while the male resection tumor-bed levels remained low. The opposite phenotype was observed in the female mice, with the female resection cohort demonstrating higher levels of CD45^hi cell infiltration compared to the female sham surgery group levels. By day 10 post-operatively, the differential phenotypes were no longer visible. Both male and female mice had lower levels of CD45^hi cells in the resection groups and higher levels CD45^hi cells in the same surgery groups (Figure 3C). Of note, CD45^hi levels were not statistically different between the groups at these time points (P  = .23, P  = .57, and P  = .38 by ANOVA at post-operative days 3, 7, and 10, respectively). Seven days post-operatively, there was also a significant reduction in T regulatory cells (Tregs) as a percentage of total CD4+ cells in the tumor bed of the male resection group compared to the male sham group (Figure 3D). No significant differences were found in blood, tibia bone marrow, or spleen (Supplementary Figure S5).

To further explore the role of Tregs in a preclinical resection model of GBM, Foxp3^DTR mice were implanted with 15 000 SB28 cells in their right hemisphere and received serial intraperitoneal injections of either 1 µg of diphtheria toxin to deplete their Tregs or D-PBS as a control. All mice underwent resection of their tumors 7 days after implantation of the syngeneic GBM cells. The D-PBS control group demonstrated a trend toward females surviving longer than males (24.5 days vs 20.5 days, P  = .0781), but the diphtheria toxin Treg-depleted group demonstrated similar survival times of 14.5 days for females and 16 days for males (P  = .2166) (Figure 3E). The male mice who underwent resection with depleted Tregs survived for a significantly shorter time than their D-PBS control counterparts (16 days vs 20.5 days, P  = .0208) (Figure 3E). Similarly, the female mice with depleted Tregs that underwent resection survived for significantly shorter time than their D-PBS control counterparts (14.5 days vs 24.5 days, P  = .0067, Figure 3E). These results demonstrate that Tregs play a role in the sex difference observed in survival in GBM preclinical resection models.

There Is No Sex Bias in Survival in Patients Who Undergo Biopsy Alone Without Surgical Resection

To investigate whether surgery plays a role in the sex bias in survival observed in epidemiologic studies, clinical patient data was reviewed. The NCDB was queried for patients diagnosed with IDH-wildtype GBM.²⁹ Of the patients who only had a biopsy and did not undergo surgical resection but did receive chemotherapy and at least 14 days of radiotherapy, 1465 were male and 997 were female. For all ages, the median survival for male patients was 8.8 months, while the median survival for female patients was 9.13 months (P  = .1) (Figure 4A). There was no sex bias in survival when the patients were stratified by age group (Supplementary Table S1). In the same database, patients who underwent resection with chemotherapy and at least 14 days of radiotherapy were analyzed; of these patients, 9203 were male and 6249 were female. Within this cohort of treated individuals, a sex bias in survival was observed, with female patients surviving significantly longer than male patients (17.1 months vs 15.7 months, P < .001) (Figure 4B). Subset analyses were performed for those who received subtotal resection (STR) vs gross total resection (GTR), and a statistically significant female survival advantage was found in both groups (GTR: P = .001; STR: P < .001) (Supplementary Figure S6A and B). These results demonstrate that the sex difference in resection observed in the preclinical model is present in human GBM patients.

Figure 4.

There is no sex bias in survival in patients who underwent biopsy alone, without surgical resection. Analyses were conducted with data from the National Cancer Database on patients with IDH-wildtype GBM. (A) Survival curve of male (blue) and female (red) patients who did not have a surgical resection but did receive chemotherapy and at least 14 days of radiation. (B) Survival curve of male (blue) and female (red) patients who had a surgical resection, received chemotherapy, and underwent at least 14 days of radiation.

Discussion

In this study, we present the first preclinical, immunocompetent GBM resection model that allows for the study of sex differences and immune response, with findings supported by a robust cohort of human survival data. It is essential to use a GBM preclinical model that includes surgical resection given that resection is a component of the current standard of care for patients. Understanding the role of the immune phenotype in GBM progression and how it differs in a sex-specific manner in post-surgical resection will allow the field to identify which aspects of the immune system confer survival benefits at given points in the treatment timeline and provide information on valuable targets for novel therapeutic interventions.

The preclinical finding that female immunocompetent mice survive longer than male immunocompetent mice supports results from larger epidemiologic studies showing that female patients have a longer overall survival than male patients after undergoing standard-of-care treatment (involving maximal safe surgical resection).³⁰ Interestingly, this sex bias was not noted in the mice that underwent sham surgery. The lack of sex bias in survival observed with sham surgery compared with prior reports of female mice surviving longer than males with no surgery could be explained by surgery itself impacting the immune system, even in conditions without a complete resection.^7,31,32 The lack of sex bias observed with sham surgery supports the idea that removal of the tumor may be needed to appreciate a female survival advantage, compared to the alterations induced by a surgery alone. Alternatively, there may be a characteristic shared by males with GBM that negatively affects their ability to mount an appropriate immune response against GBM. A study examining only male mice with different syngeneic models of GBM found that resection led to decreased circulating levels of cytotoxic T cells.⁷ There could be cell-intrinsic changes that occur, either in cancer cells themselves or in the immediate tumor microenvironment, that drive this survival phenotype. To better understand why males and females display a sex bias in their survival after surgical resection of tumor but not with sham surgery, the tumors and their resection beds were sent for RNA sequencing.

This analysis revealed that there were larger transcriptional differences between resection and sham groups within the same sex compared to the transcriptional differences observed between the male and female resection groups. This demonstrates that cell-intrinsic changes are not directly responsible for the female survival advantage after surgical resection. The gene ontology terms associated with the genes that were upregulated in the male and female groups when the resection cohort was compared to the sham surgery cohort were related to the immune system, especially to immune response, activation, and migration. This is not unexpected, as other injuries to the brain, such as ischemic stroke, have been shown to induce changes in the immune system.^33-36

When the role of the immune system was examined, it was found that both males and females had a shorter overall survival without an intact immune system, with the females reaching neurologic endpoint as quickly as the males. The fact that sex bias relies on the immune system has previously been shown in preclinical models of GBM that do not incorporate surgical resection.⁹ Given the known sex differences observed in the immune system and immune responses in cancer, including GBM, this is not an unexpected finding.^10,37,38

The observation that male mice who undergo resection may have diminished Tregs in their resection cavity was surprising, as Tregs are often thought to be an immunosuppressive cell population. The Foxp3^DTR survival experiment demonstrated that both males and females have significantly lower survival when their Tregs are depleted, indicating that Tregs play a protective role in this post-operative population. The observation that both males and females display similar survival upon Treg depletion demonstrates that Tregs are not only important for survival post-operatively but that they may also play a role in the sex bias seen in non-depleted mice. Tregs play a role not only in the immune response but also in wound healing and recovery. They are reportedly important for neurological recovery after stroke by producing amphiregulin, which helps to decrease levels of astrogliosis and prevent neurotoxicity.^39,40 This work demonstrates that Tregs may have an essential role in survival for this preclinical resection model, and that role could be in recovery after the neurologic injury incurred by a tumor resection surgery. Additional work is needed to examine whether Tregs function in a wound-healing capacity during a specific time period post-operatively. The effect of the other components of standard of care on Tregs will also need to be characterized to determine the timing of administration of surgery, radiotherapy, temozolomide, and potential new therapies.

Review and analysis of human data from a national registry support the preclinical model findings that there is a sex bias in survival following GBM resection. This bias was not seen for patients who underwent biopsy alone, without resection of tumor. This reinforces the findings of our preclinical model and helps to validate the findings observed in mice. Surgical resection itself may exacerbate a sex bias in the pathophysiology of GBM, leading to a greater difference in survival between the sexes. Alternatively, there could be no significant sex bias in survival seen in patients who did not undergo surgical resection because their survival was too short to display a meaningful difference in survival between male and female patients. In the biopsy cohorts, median survival for both male and female patients was unfortunately less than a year, highlighting the aggressiveness of this disease. With surgical resection, patients survived longer, and this may have unmasked a sex bias in survival. These results corroborate findings using the preclinical model and reinforce the need to include surgical resection in preclinical investigations.

Despite the promising results of these studies, it is important to note their limitations. Although the syngeneic GBM cells were created with mutations that mirror those found in GBM, this is not the same disease that grows in humans and therefore may not faithfully recapitulate how the human disease would progress or react to potential interventions. RAS mutations are not frequently noted in human GBM but are found in the syngeneic line used in these experiments. An immediate priority is to elucidate how consistent these results are with other syngeneic murine models with different driver mutations. Additionally, while this approach allows for the study of the immune system, both systemically and in the microenvironment, it precludes the use of human GBM. It is also important to consider the immunogenicity of the syngeneic lines. GBM is an immunologically cold tumor, known to have a microenvironment lacking in CD3+ T cells.^41-44 Clinical trials studying immune checkpoint inhibitors in GBM have failed, and it is vital that preclinical models used to identify and screen potential therapies recapitulate the immunogenicity of GBM as accurately as possible.^45,46 Response to immunotherapy is linked to mutational burden, with higher levels correlating with immunogenicity of the cancer.⁴⁷ Whole-exome sequencing has revealed that the conventional mouse syngeneic GBM model GL261 has a higher tumor mutational burden than SB28, which is not unexpected given that the line was generated through carcinogen exposure.⁴⁸ Moreover, flow cytometry revealed that GL261 expresses MHC-I constitutively and MHC-II when induced with interferon-γ; however, SB28 only expressed MHC-I when induced with interferon-γ and never expressed MHC-II, a characteristic that is commonly observed in GBM.^48,49 For these reasons, GL261 was not used for any studies here. Regarding immunogenicity, the cells used for this study were karyotyped as XX (Supplementary Figure S1); therefore, male mice are not expected to mount baseline immune responses due to a sex mismatch. Both male and female mice were both able to grow SB28 and NPD tumors, and this has been a consistent observation in previous studies from the group.9,¹⁰ An additional limitation of this work is the inherent biological variability observed when working with an in vivo model. There was variability noted among the animals during the flow cytometry experiments. Total live cell counts per tissue for each animal were recorded, and counts of immune cell populations were normalized to total CD45^hi cells, but this did not necessarily acknowledge any differences between mice in terms of their tumor size, resection cavity, and size/quantity of other organs. To better understand variation in tumor size between the biological sexes and different surgical interventions, a histologic study should be conducted with hematoxylin and eosin staining to characterize tumor quantity at specific time points post-operatively. Future directions include exploring the role of Tregs in the perioperative setting, especially as they influence wound healing and survival. Finally, it is important to add radiation and temozolomide, as well as dexamethasone, to the preclinical model to fully recapitulate the GBM standard of care.

Supplementary Material

Supplementary material is available online at Neuro-Oncology Advances (https://academic.oup.com/noa).

Author Contributions

J.V.: design, implementation, analysis, interpretation, writing. A.A.V.: implementation. J.L.: design, interpretation. E.H.: analysis, interpretation. Q.O.: analysis. C.B.: analysis. S.A.: implementation. A.K.: implementation. S.W.: implementation. K.K.: implementation. G.B.: implementation. E.M.H.: editing. M.G.: design. A.L.: design. A.D.: interpretation. J.D.L.: design, interpretation, and writing.

Conflict of Interest Statement

J.D.L. is listed as an inventor on intellectual property involving cancer therapeutics held by Cleveland Clinic that is not directly relevant to this work. No other conflicts are declared for any other author.

Funding

Funding for this work was provided by the National Institutes of Health grants P01 CA245705 (J.D.L.) and R35 NS127035 (J.D.L.), the Cleveland Clinic (J.D.L. and J.L.), and Case Comprehensive Cancer Center (J.D.L.), as well as by the Crile Fellowship (J.V.), American Brain Tumor Association (J.L.), a VeloSano Postdoctoral Fellowship (A.A.-V.) and a Ramon Areces Foundation Postdoctoral Fellowship (A.A.-V.).

Ethics Statement

Animal protocols and procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the Cleveland Clinic. Experiments were conducted under the IACUC protocol 2849. Human data were accessed retrospectively using a de-identified public database under an IRB protocol determined to be exempt.

Data Availability

Data will be made available upon reasonable request.