Original research••

Enhanced immune responses are accompanied by increased MAGEA expression in osteosarcoma metastases

•,,,,,,,,,,,,,,,,,,.

...

Abstract

Objective Osteosarcoma is the most common primary bone sarcoma. About 50% of patients develop metastatic disease and their 5-year survival lingers at around 20%–30%. T cell checkpoint blockade immunotherapies have revolutionised cancer treatment in the last decade, but their impact remains limited in osteosarcoma.

Methods and analysis In order to reveal potentially novel immunotherapeutic strategies for advanced osteosarcoma, we conducted an immunogenomic characterisation of a unique sample set comprising 30 osteosarcoma samples from seven patients, collected throughout disease progression. We performed RNA-sequencing and imaging mass cytometry analysis on these samples to reveal the immunological landscape during osteosarcoma progression.

Results Transcriptional and phenotypical hallmarks of cytotoxic T cell-driven anticancer immunity were enriched in metastatic lesions as compared with primary tumours. Spatial analysis showed T cells infiltrating central regions of osteosarcoma metastases, indicating the absence of an immune excluded environment. In parallel, we found a pronounced increase in the expression of cancer testis antigens, particularly melanoma antigen family A (MAGEA)-related antigens, in osteosarcoma metastases, which was validated in an independent cohort (N=91). In addition, we demonstrated the presentation of MAGE-derived peptides in three out of four osteosarcoma cell lines.

Conclusion These findings indicate a concurrent augmentation of cytotoxic antitumour immune responses and expression of MAGEA antigens from primary to metastatic osteosarcoma. This observation suggests the exploration of MAGEA antigens as potential targets for immunotherapy in the treatment of advanced osteosarcoma.

What is already known on this topic

Osteosarcoma patients commonly develop metastases and show poor survival rates. New strategies are urgently needed to improve outcomes. While T cell checkpoint blockade immunotherapies have revolutionised cancer treatment, their impact remains limited in osteosarcoma.

What this study adds

In this study, we showed increased hallmarks of cytotoxic T cell-driven anti-cancer immunity and elevated expression of cancer testis antigens of the melanoma antigen family A (MAGEA) family in osteosarcoma metastases as compared with the primary tumours. Furthermore, presentation of MAGE-derived peptides was demonstrated on osteosarcoma cell lines.

How this study might affect research, practice or policy

The increased infiltration by cytotoxic T cells in osteosarcoma metastases renders them potentially responsive towards immunotherapy. Additionally, our findings suggest the exploration of cancer testis antigens of the MAGEA family as potential targets to be explored in the context of T cell-based immunotherapies in advanced osteosarcoma.

Introduction

Osteosarcoma is the most common primary bone sarcoma, predominantly occurring in children and young adults, with a second incidence peak in older adults.1 2 The 5-year survival rate for localised osteosarcoma is approximately 70% but drops to 20%–30% in the case of metastatic disease, which develops in approximately 50% of osteosarcoma patients.1 3 Unfortunately, the clinical prognosis of osteosarcoma patients has barely improved in the last few decades, underlining the need for the development of more effective therapies.4 5

T cell checkpoint blockade immunotherapies have revolutionised cancer treatment, but their impact has been limited in osteosarcoma.6 7 For instance, treatment with the programmed cell death 1 (PD-1) antibody pembrolizumab resulted in a clinical response in only 1 out of 22 osteosarcoma patients.6 Additionally, a phase 2 clinical trial with the anti-PD-L1 antibody durvalumab, in combination with the CTLA-4 blocker tremelimumab, yielded no responses in five patients with metastatic osteosarcoma.8 In another completed phase 2 clinical trial that assessed the efficacy of avelumab in 18 patients with recurrent or progressive osteosarcoma, 17 patients showed disease progression and 1 died off study (NCT03006848). The lack of sensitivity of osteosarcomas to these immunotherapies may be attributed to their low mutation burden, which affects the availability of neoantigens.9 Neoantigen load is closely linked to the therapeutic activity of checkpoint blockade antibodies.10 A paradigmatic example for this association is provided by the successful outcomes following treatment with PD-1 blockade in a small proportion of patients diagnosed with undifferentiated soft tissue sarcomas that presented with high mutation burden.11 Instead, most osteosarcomas present a highly complex genome where structural variants and copy number alterations can further contribute to immune evasion.12 13

Despite lacking immunogenic features, evidence of antitumour immunity has been encountered in osteosarcomas: infiltration by tumour-associated macrophages in primary osteosarcomas is associated with reduced metastasis and improved patient survival.14 Furthermore, us and others showed an increase in T cell infiltration and expression of immune checkpoints (including PD-1, PD-L1 and lymphocyte activation gene 3 (LAG3)) in osteosarcoma metastases compared with primary tumours.15–19 Indeed, neoantigen-specific T cell responses are known to occur in cancers with low mutation burden despite their lack of sensitivity to immune checkpoint blockade.20 21 On the other hand, ‘self’ antigens such as cancer-testis antigens (CTAs) have also been shown to elicit antitumour immune responses and may constitute attractive targets for immunotherapy, particularly in tumours with low mutation burden. For instance, targeting NY-ESO-1, a CTA, with vaccines or adoptive T cell therapy has proven successful in patients with synovial sarcoma.22–25

In this study, we aimed to identify novel immunotherapeutic targets for osteosarcoma patients with advanced disease by exploring the immunological changes that accompany osteosarcoma progression. A unique sample set composed of sequential lesions obtained from seven patients during the course of disease was analysed by RNA-sequencing and imaging mass cytometry (IMC). Transcriptomic and phenotypical hallmarks of cytotoxic T cell-driven anticancer immunity were enriched in metastatic lesions as compared with primary tumours. In parallel, we identified an increased expression of multiple CTAs in osteosarcoma metastases, in particular of melanoma antigen family A (MAGEA)-related antigens.

Materials and methods

Sample information

Osteosarcoma patients were selected based on the availability of frozen tissue specimens obtained from the primary tumour, local recurrence and/or distant metastases. In total, seven patients were included (P1–P7; M/F=4/3; age at diagnosis range=14–30 years), including primary biopsies (chemotherapy naïve) (N=5), primary resections (N=5), local recurrences (N=3) and metastases (N=10) (online supplemental tables S1,S2). Tumour percentage of the samples was at least 70% as estimated by an experienced pathologist (JVMGB). Approval of institutional review board (Medisch-Ethische Toetsingscommissie Leiden Den Haag Delft) was obtained for the use of these tissue samples from the LUMC bone and soft tissue tumour biobank (Leiden, the Netherlands) (B20.067). Written informed consent from patients was obtained. All specimens were pseudoanonymised and handled according to the ethical guidelines described in ‘Code for Proper Secondary Use of Human Tissue in The Netherlands’ of the Dutch Federation of Medical Scientific Societies. Patients or the public were not involved in the design, conduct, reporting or dissemination plans of our study.

RNA-sequencing and data analysis

RNA was isolated from tissue sections (thickness=20 µm) with TRIzol (15596018, Ambion) according to manufacturer’s instructions. RNA purification was performed using RNaesy mini kit (74104, Qiagen) and included DNAse treatment according to manufacturer’s instructions. RNA concentrations and quality were checked on NanoDrop (ThermoFisher). Total RNA-sequencing Paired-End was performed by GenomeScan (Leiden, the Netherlands) on the NovaSeq6000 platform for a total of 24 samples.26 Raw RNA-seq data were processed by the RNA-seq BioWDL pipeline using default settings, developed by Sequencing Analysis Support Core (https://biowdl.github.io/, LUMC, Leiden, the Netherlands). Reads were aligned against hg38 with STAR and expression quantification was performed using HTseq-count.27 One low-quality sample (RNA quality number <6 and mapped unique read pairs <1%) was excluded from data analysis. Gene counts were normalised with the ‘median of ratios’ method in the DESeq2 R package.28 Differential gene expression analysis was performed between primary biopsies (if available) and first occurring metastases, accounting for paired samples by including patient as covariate to the model. Differentially expressed genes were defined by Benjamini and Hochberg (BH)-corrected p values below 0.05 and log2FC above 1 or below −1. For principal component analysis (PCA), the patient effect was corrected in limma and PCA was performed using prcomp function in the stats R package. OptiType was used to determine human leukocute antigen (HLA) class I genotypes.29 To assess cytotoxic immune responses, the immunologic constant of rejection (ICR) score was calculated.30 31 For data visualisation, R packages ggplot2, EnhancedVolcano, survival and pheatmap were used.

TMA construction

Two osteosarcoma tissue micro arrays (TMA) containing formalin-fixed paraffin-embedded (FFPE) samples were produced. TMA-1 was used for IMC and internal validations, containing 28 samples of the 7 patients who were included in the transcriptome analysis (online supplemental table S1). Two cores in peripheral regions and two cores in central regions of the tumour were punched for each sample, whenever possible. TMA-2 was used as a validation set, and contained 150 samples of 91 patients, including primary biopsies (N=79), primary resections (N=47) and metastases (N=24). Each TMA contained three cores per sample. Clinical information is summarised in online supplemental table S1. Construction of the TMA for external validation has been described previously.14 32 For these, a waiver of consent was obtained from the institutional review board (protocol number: B17.036).

Imaging mass cytometry

IMC was performed on TMA-1 (online supplemental table S1). A tumour immune microenvironment panel of 40 markers was used as described previously.33 Anti-TIGIT (BLR047F, 1:50, incubation time: 5 hours, metal: 156 Gd, incubation temperature: RT, Abcam) and anti-TCRgd (H41, 1:50, incubation time: ON, metal: 148 Nd, incubation temperature: 4C, Santa Cruz) were used instead of anti-PD-L1 and anti-CD73, respectively. Furthermore, anti-histone (D1H2, 1:50, incubation time: ON, metal: 209 Bi, incubation temperature: 4C, CST) was added to the panel. Immunodetection by IMC was performed as described previously.33 Hyperion mass cytometry imaging system (Fluidigm) at the Flow Core Facility (LUMC, Leiden) was used to ablate regions of interest (1000×1000 µm). Processing of the images included normalisation of signal intensity at pixel level using Matlab (V.R2021a), background removal in Ilastik (V.1.3.3) and creation of cell segmentation masks in CellProfiler (V.2.2.0) as described previously.34 Due to decalcification of the FFPE material, DNA staining was suboptimal for cell segmentation in a number of images. Next, single-cell Flow Cytometry Standard (FCS) files with relative frequencies of positive pixels were acquired using ImaCyte35 and subsequently used for cell phenotyping. Markers with weak or undetectable expression, including LAG3, PD-1 and forkhead box P3 (FOXP3), were not included in the analysis. An overview of lineage markers used for cell phenotyping is listed in online supplemental table S3.

Spatial interaction analysis

Spatial interaction analysis was performed on all imaged samples using ImaCyte. Enriched immune cell interactions were compared between (1) primary biopsies (if available) and the first occurring metastasis and (2) central regions and peripheral regions within first occurring metastases. In short, the neighbours of each cell were determined at a 10 µm distance, after which a 1000-times iteration permutation test was performed for each image to calculate the Z-score for the interaction between adjacent cells. A positive Z-score indicates an enriched interaction between cell phenotypes as compared with random distribution of cell phenotypes. Given the range of all calculated Z-scores, a Z-score higher than 3 was considered a significant enriched interaction. Comparative interaction plots were generated with ggplot2 to visualise the significant interactions between all immune cell phenotypes.

Immunohistochemistry

Immunohistochemistry was performed on TMA-1 and TMA-2 sections using the anti-MAGEA3 antibody (ab223162, clone EPR19065, Abcam). Tissues from testis and cervix were used for positive and negative control, respectively. Antigen retrieval was performed using 10 mM citrate buffer (pH 6) and a blocking step with 5% non-fat dry milk was performed for MAGEA3 staining. TMA sections were incubated overnight with the primary antibody (MAGEA3, 1:4000) at 4°C and subsequently washed with PBS. Next, sections were incubated with poly-HRP anti-mouse/rabbit secondary antibody (DPVO110HRP, BrightVision, Immunologic), washed with PBS and developed with DAB+Chromogen (K3468, Dako, Agilent Technologies, California). Sections were counterstained with haematoxylin, rinsed with distilled water, dehydrated and mounted. Loss of cores during experimental processes reduced the number of samples available for scoring.

TMA scoring

TMAs were manually and semiquantitatively scored for MAGEA3 expression by two independent observers, including one experienced pathologist (SWL). Scoring was based on staining intensity (0=negative, 1=weak, 2=moderate, 3=strong) and percentage of positive tumour cells (0=0%, 1=1%–24%, 2=25%–49%, 3=50%–74%, 4=75%–100%). The sum of the staining intensity score and percentage of positive tumour cells score is regarded as the TMA score. TMA score discrepancies between the two observers were discussed to reach consensus. Cumulative scores for each core were calculated. The average of regions for each resulted in the final score.

Cell culture

Human osteosarcoma cell lines U2OS, SAOS2, MNNG and OHS were cultured in RPMI 1640 (Gibco, Invitrogen Life-Technologies, Scotland, UK), supplemented with 10% Fetal Calf Serum (F7524, Sigma-Aldrich, Saint Louis, MO, USA) in a humidified incubator (37°C, 5%CO₂). STR profiling (GenePrint V.10 System, Promega, Madison, Wisconsin) and mycoplasma tests were performed on a regular basis.

Reverse transcriptase quantitative PCR

RNA was isolated from osteosarcoma cell lines using TRIzol (15596018, Ambion). RNA purification was performed using RNaesy kit (74104, Qaigen) according to manufacturer’s instructions. cDNA was synthesised using AMV Reverse Transcriptase (Roche, 109118). Primers used to assess MAGEA3, GAPDH and HPRT expression are listed in online supplemental table S4. Reverse transcriptase quantitative PCR was performed using iQ SYBR Green Supermix (1708886, Bio-rad) and a Thermal Cycler (Bio-rad). Samples were run in technical duplicates. MAGEA3 gene expression was normalised using house-keeping genes GAPHD and HPRT.

Statistical analyses

All statistical analyses were performed in R V.4.0.2. When data of primary biopsy and resection, or multiple metastases were available from a single patient, the primary biopsy and the first occurring metastasis were considered to calculate significant differences between primary osteosarcomas and osteosarcoma metastases with Wilcoxon signed rank tests. Differences in immunohistochemistry scores between unmatched primary and metastatic tumours from an extended validation cohort were analysed using Mann-Whitney U test. Kruskall-Wallis with post hoc Dunn tests were performed to identify differences in immune phenotypes among ICR groups. Correlations between ICR scores and MAGEA gene expression were analysed using Spearman rank correlation coefficient. P values below 0.05 were considered significant.

Methods employed for HLA genotyping and HLA ligandome analysis are provided in supplementary methods.

Results

Hallmarks of T cell-driven antitumour immunity are enhanced in osteosarcoma metastases

To explore the transcriptomic changes that accompany osteosarcoma progression, RNA-sequencing was performed on a discovery cohort of osteosarcoma samples that included primary biopsies (N=5) and resections (N=5), with matching local recurrences (N=3) and metastatic lesions (N=10), from a total of seven patients (online supplemental table S1). A principal components analysis with paired samples including the primary biopsy (when available) (primary biopsy N=4, primary resection N=2) and the first occurring metastasis of every patient (N=6) showed a clear separation between primary and metastatic lesions on prinicpal component (PC) 2 (figure 1A). Paired differential gene expression analysis between the primary tumours (biopsies when available) and the first occurring metastasis revealed 141 differentially expressed genes, of which 74 and 67 exhibited higher expression in primary tumours and metastases, respectively (figure 1B, online supplemental table S5). In primary tumours, the increased expression of genes such as S100A8, FCGR3B and CXCR2 could represent increased myeloid cell infiltration, in particular of granulocytes.36–38 Interestingly, CXCL9, a Th1-associated T cell modulator and chemoattractant, and FOXP3, a regulatory T cell (T_reg) marker were expressed at higher levels in osteosarcoma metastases, suggesting enhanced T cell infiltration in these lesions. To further assess how cytotoxic immune responses vary during osteosarcoma progression, the ICR score was calculated and compared between primary tumours and metastases. A high expression of ICR genes typifies immune active tumours characterised by the presence of a T helper 1 (Th-1)/cytotoxic immune response.30 31 On all RNAseq osteosarcoma samples (N=23), osteosarcoma metastases showed a relatively high expression of ICR genes compared with the primary tumours (figure 1C). Additionally, when performing a paired analysis between primary biopsies with the first occurring metastases, ICR scores, as calculated by the average expression of ICR genes per sample, were higher in the metastases (figure 1D).

Request permission

Figure 1

Differential gene expression analysis and ICR profiles between osteosarcoma primary tumours and metastases. (A) PCA plot of osteosarcoma primary tumours and first occurring metastases, corrected for patient effect. (B) Volcano plot highlighting significantly differentially expressed genes between primary tumours and first occurring metastases (DESeq2, accounting for paired samples by adding patient as co-variate). Yellow and purple regions represent genes overexpressed in primary tumours and metastases, respectively. Most significant genes (FDR<1e-7 and log2FC>1 or log2FC < −1) and immune-related genes are highlighted. (C) Heatmap depicting ICR genes, clustering samples into ICR low (left), moderate (middle) and high (right) groups. (D) Scatterplot depicting ICR scores (average expression of the 20 genes of the ICR gene signature) of osteosarcoma primary tumours compared with first occurring osteosarcoma metastases (Wilcoxon signed rank exact test, p<0.05). Primary and metastasis pairs are indicated with dashed lines. FDR, false discovery rate; ICR, immunologic constant of rejection; PCA, principal component analysis.

Distinct immune microenvironments in osteosarcoma primary tumours and metastases

The differences observed in ICR scores between primary osteosarcoma tumours and metastases suggest the presence of a distinct tumour immune microenvironment between lesions. Therefore, a more in-depth exploration of the immune contexture was investigated by IMC in 28 osteosarcoma cases (primary biopsies (N=6), primary resections (N=6), local recurrences (N=4) and metastases (N=12)), including 21 samples that were profiled by RNA-seq (online supplemental table S1, online supplemental table S2). In total, 37 markers, mostly corresponding to immune cell markers, were investigated at single cell level by IMC. Lineage markers use for immune cell phenotyping are listed in online supplemental table S3. Unsupervised clustering analysis of immune cell populations revealed that the cluster with the highest levels of immune cell infiltration, in particular T cells and myeloid cells, predominantly contained ICR moderate/high samples and osteosarcoma metastases (figure 2A, online supplemental figure S1A). Metastatic lesions presented on average 3.6 times more T cells per mm² than primary tumours (figure 2B–D, online supplemental figure S1B). Only a single patient (P5) had more T cells in the primary resection compared with the metastatic lesion, although this patient had little T cell infiltrate in both lesions (online supplemental figure S1B). Interestingly, the primary resection of P3 clustered with samples with high immune cell infiltrate and showed increased T cell infiltrate compared with the treatment naïve tumour (figure 2A and online supplemental figure S1B). This increase might be explained by a good response towards chemotherapy, which was only observed in P3 and P6. Unfortunately, no IMC data were available for the primary resection of P6. Further analysis of the T cell compartment showed that a large number of T cells infiltrating osteosarcoma metastases corresponded to CD8⁺, CD4⁺ and proliferating T cells (online supplemental figure S1C).

Request permission

Figure 2

Abundance of immune cell phenotypes in osteosarcoma samples by IMC. (A) Unsupervised clustering of immune phenotypes in Z-scores. Boxplots above the heatmap indicate absolute values of the number of cells per mm² per sample for each phenotype. Samples are annotated based on Primary/recurrence/metastasis (PRM), patient ID (PAT_ID) and ICR group. (B) H&E staining of primary and metastatic samples corresponding to IMC images (C) showing low and high T cell infiltrate, respectively. (D) Boxplot depicting the number of T cells/mm2 in primary tumours and metastases as calculated from IMC data. DCs, dendritic cells; ICR, immunologic constant of rejection; ILCs, innate lymphoid cell.

Interactions between T cell populations and macrophages are enriched in osteosarcoma metastases

To explore the spatial localisation of T cells in relation to tumour cells, the number of T cells in central and peripheral regions of the tumours was compared (figure 3A). There were no discernable patterns across all samples regarding T cell enrichment in either the central or peripheral regions. In P2Pr, P2M1 and P3Pr, the majority of T cells were localised in central regions. Conversely, only P4M4 showed most T cell infiltration in peripheral regions of the tumour, suggesting an immune excluded environment in this sample. Direct cell–cell interactions between T cell phenotypes and other immune cells were subsequently investigated with a spatial interaction analysis. Comparing primary tumours with the first occurring metastasis revealed that interactions between different T cell phenotypes were enriched in osteosarcoma metastases (figure 3B). Interestingly, interactions between CD8⁺ T cell, macrophages (including CD11+populations) and monocytes were also specifically enriched in metastatic lesions. This observation could suggest the activation of CD8⁺ T cells through antigen cross-presentation by M1-like macrophages. Notably, approximately half of the CD11c⁺ macrophage population in osteosarcoma metastases expressed HLA-DR (online supplemental figure S2A,B). The interactions between different T cell populations were enriched in central regions of the tumours, while interactions between T cells and myeloid cells were more prevalent at peripheral regions (figure 3C). Interestingly, neighbourhoods comprising both CD8⁺ T cells and CD11c⁺ macrophages were present at both central and peripheral regions. In primary tumours, there was an enrichment of interactions between myeloid cells, in particular, ones involving CD163+ macrophages.

Request permission

Figure 3

Localisation of immune cells in primary tumours and metastases. (A) Barplot visualising the number of T cells/mm² in central and peripheral regions. (B) Neighbourhood analyses showing enriched interactions in primary tumours and metastases and (C) central and peripheral regions in osteosarcoma metastases. Highlighted interactions are indicated with red boxes or red arrows. DCs, dendritic cells; ILCs, innate lymphoid cell; Mϕ, macrophages.

Increased expression of MAGEA cancer testis antigens in osteosarcoma metastases

The enrichment of immune cells and, more specifically, T cells in osteosarcoma metastases could indicate an increased immunogenicity of those lesions as compared with primary tumours. We postulated that such increase could be due to the increased availability of CTAs as these were previously shown to be overexpressed in osteosarcomas.39 Therefore, the expression of all genes coding for CTAs annotated in the CTA database was evaluated (http://www.cta.lncc.br/). Unsupervised clustering on all RNAseq osteosarcoma samples (N=23) revealed that genes belonging to the MAGEA and chondrosarcoma-associated genes (CSAG) family were overexpressed in the majority of osteosarcoma metastases (figure 4A,B, online supplemental figure S3). In particular, CSAG1, MAGEA3, MAGEA6 and MAGEA12 (MAGEA3/6/12) were highly expressed in osteosarcoma metastases as compared with primary tumours (figure 4A,B). Additionally, the gene expression of these CTAs was positively correlated with the ICR score of the tumours (Spearman correlation, R>0.58, p<0.01) (online supplemental figure S4).

Request permission

Figure 4

Increased expression of cancer testis antigens in osteosarcoma metastases. (A) Heatmap and (B) boxplots of normalised gene expression of genes from the MAGEA and CSAG family in osteosarcoma primary tumours, local recurrences and metastases. (C) IHC images of a negative, weak, moderate and strong staining of MAGEA3. (D) Boxplot depicting TMA scores for primary tumours and metastases of external validation series (Mann-Whitney U test, p<0.05). Dashed lines connect primary tumours and metastases from the same patient. CSAG, chondrosarcoma-associated genes; IHC, immunohistochemistry; MAGEA, melanoma antigen family A.

Subsequently, MAGEA protein expression was detected by immunohistochemistry with an anti-MAGEA3 antibody (also cross-reacting with MAGEA6), in a tissue microarray containing 14 primary osteosarcomas and 12 osteosarcoma metastases that included the samples subjected to RNA sequencing (figure 4C, online supplemental figure S5). All primary tumours were negative or had low MAGEA3 expression, while metastases that displayed MAGEA3 RNA expression were also positive at the protein level. Next, we expanded these observations by assessing MAGEA3 expression in an additional tissue microarray (online supplemental table S1). Immunohistochemical detection of MAGEA3 was evaluated in 66 primary tumours and 14 osteosarcoma metastases. Overall, MAGEA3 expression was significantly higher in metastatic lesions (p<0.05, unpaired Mann-Whitney U test) (figure 4D).

MAGEA peptide presentation on the cell surface of osteosarcoma cell lines

To investigate whether overexpressed CTAs could be potential targets for T cell recognition, we determined whether CTA-derived epitopes were presented on HLA class I molecules in osteosarcoma cell lines. MAGEA3 expression was observed in cell lines U2OS, OHS and SAOS2, by qPCR and immunohistochemistry (online supplemental figure S6). No MAGEA3 expression was detected in cell line MNNG. Next, it was confirmed that membranous expression of HLA class I/β2-microglobulin was detectable in all MAGEA3-expressing cell lines (online supplemental figure S7). HLA genotyping was performed on all cell lines and patients included in the discovery cohort. Common overlapping HLA class I alleles between cell lines and patients included HLA-A 02:01 and HLA-A 24:01 (online supplemental table S6). The HLA class I ligandome of the MAGEA3-expressing cell lines was determined by mass spectrometry analysis following immunoprecipitation of HLA class I complexes. Various MAGEA peptides were found in cell lines U2OS (N=18), OHS (N=9) and SAOS2 (N=15) (online supplemental table S7). Two peptides, one from MAGEA1 (KVLEYVIKV) and one from MAGEA3/6 (KIWEELSVLEV), were identified in the ligandome of the three cell lines and were strong binders towards HLA-A02:01 (table 1). A Blast search (https://blast.ncbi.nlm.nih.gov/) on the MAGEA1 peptide sequence showed no sequence homology with other proteins. The MAGEA3/6 peptide had 91% overlap with MAGEA12. These findings suggest that the MAGEA1 and MAGEA3/6 peptides may represent recurrent targets for T cell-based therapy in osteosarcoma.

Table 1

Common peptides presented on HLA-A 02:01 between three osteosarcoma cell lines

Discussion

The limited treatment options available for metastasised osteosarcoma contribute to the poor clinical prognosis of patients with this condition. In our study, we conducted an immunogenomic analysis of paired primary and metastatic osteosarcoma samples and found an enrichment of T cells and other immune cell subsets in osteosarcoma metastases. By integrating these findings with transcriptome analysis, we aimed to identify targets that could account for the increased T cell infiltration in metastases and potentially serve as effective targets for immunotherapy. This led to the identification of CTAs, particularly MAGEA-related antigens, as potential targets for immunotherapy.

Analysis of gene expression in osteosarcoma samples revealed increased expression of FOXP3 and CXCL9 as well as enrichment of the ICR signature in osteosarcoma metastases, indicating an ongoing T cell-driven antitumour immune response in these lesions. Our multidimensional immunophenotyping analysis confirmed the increased infiltration of T cells in metastatic samples compared with primary osteosarcomas, consistent with a previous study showing a positive correlation between ICR scores and infiltration of various immune cell populations in soft tissue sarcomas.40 Moreover, T cell cytotoxicity transcriptional signatures, including ICR, have been associated with improved overall survival in primary soft tissue sarcomas, and both overall and event-free survival in osteosarcomas.18 30 41 Next to T cell abundances, our spatial analysis showed that T cells infiltrated peripheral regions as well as central regions of the tumour. This suggests that most of the metastatic lesions investigated in this study did not display an immune excluded phenotype. This is in contrast to reported observations where T cells were mainly detected at peripheral regions of osteosarcoma metastases.18 However, these observations were focused on lung metastases, while the current study included other locations such as colon, pancreas and skin. Next to increased infiltration of T cells in osteosarcoma samples with high ICR scores, the myeloid compartment, including macrophages and monocytes, was also more abundant in these samples and could potentially contribute to the antitumour immune response. More specifically, a neighborhood analysis showed that cytotoxic T cells and CD11c+ macrophages, which also express HLA-DR, colocalise with each other, suggesting the occurrence of antigen presentation processes in those tumours.

The increased ICR score, immune infiltrate and cytotoxic T cell responses in osteosarcoma metastases suggest a potential for positive outcomes with immunotherapy, as cancers infiltrated by (cytotoxic) T cells are typically more responsive towards immunotherapy,42 although it should be noted that responses towards immune checkpoint inhibitors in osteosarcoma remain limited.6 7 To understand the basis of the increased immune response in osteosarcoma metastases and identify potential targets for immunotherapy, we profiled the expression of CTAs during osteosarcoma progression. We demonstrate that several MAGEA CTAs exhibit increased expression during osteosarcoma progression in matched samples. Furthermore, we observed a significant association between ICR scores and increased MAGEA expression in metastatic osteosarcoma samples, suggesting a potential causal effect. Previous studies have suggested increased infiltration by T cells and enhanced MAGEA expression in metastatic osteosarcoma, but the association between these factors had not been investigated.15 17 43 This finding is particularly important because a link between cytotoxic T cell signatures and positive responses to MAGEA3 antigen-based immunotherapy has been established in a therapeutic setting for metastatic melanoma.44 Together, these results reinforce the potential of MAGEA antigens as targets for immunotherapy in advanced osteosarcoma.

One challenge in targeting CTAs through immunotherapy is their potential lack of tumour specificity. Neurological toxicity has been observed as a result of anti-MAGEA3 TCR gene therapy, resulting in fatalities in some fatalities.45 However, this was proposed to result from an off-target activity of the anti-MAGEA3 TCR that recognised a highly homologous peptide sequence derived from EPS8L2, a protein expressed in the brain.46 Several clinical trials are currently underway, including a MAGEA3-based vaccine immunotherapy for advanced cancers (NCT02285816, NCT04908111). In osteosarcoma, one clinical trial demonstrated the safety and efficacy of T cell receptor immunotherapy targeting MAGEA3 in a patient with metastasised osteosarcoma, resulting in a partial response.47 Furthermore, a T cell clone was previously established that recognises the peptide MAGEA1: KVLEYVIKV and demonstrated a safe reactivity profile when cocultured with a large panel of MAGE-negative tumour cell lines.48 This specific peptide was also found through HLA peptidomics in all investigated osteosarcoma cell lines in our study, making it a potential immunotherapy target.

Absence of immunogenic traits in primary samples could indicate poor immune surveillance contributing to the high likelihood of osteosarcoma progression to the metastatic setting. Following the metastatic process, increased infiltration of lesions by immune cells might be explained by exposure of immune cells to circulating tumour cells or accessibility of immune cells to the organs of metastasis. The infiltration of cytotoxic T cells in osteosarcoma could promote the adoption of immune evasive mechanisms by tumour cells. One such mechanism could be the loss of HLA class I molecules, which has been reported to occur in 6%–20% of osteosarcomas.17 49 However, the frequency of HLA class I loss was shown to be similar between primary tumours and metastases.17 Interestingly, expression of PD-L1 has been shown to be more frequently expressed in osteosarcoma metastases compared with primary tumours.17

Our study design, which involved primary osteosarcomas with corresponding metastases, was crucial in uncovering important transcriptomic differences, despite the limitation of a small sample size. This matched design helped to highlight differences in gene expression that could have otherwise been overlooked, considering the known high heterogeneity between osteosarcoma patients.50 It is important to note that the requirement for fresh-frozen samples from sequential osteosarcomas significantly limited our ability to include more patients in this study. As a result, additional biological processes associated with osteosarcoma progression may not have been captured and warrant further investigation.

In summary, osteosarcoma metastases exhibit increased infiltration by cytotoxic T cells compared with the matched primary tumours, rendering them potentially responsive to immunotherapy. Furthermore, we identified CTAs of the MAGEA family as potentially interesting targets to be explored in the context of T cell-based immunotherapies.

NdM and JVMGB contributed equally.
Contributors: DMM: conceptualisation, methodology, project administration, data curation, software, formal analysis, validation, investigation, visualisation, writing-original draft, writing—review and editing. SvO: conceptualisation, investigation, writing—review and editing. JPR, DR, IHB-dB, BEvdA, ABK, PMW-K, MvdP, MI, SWL, AHdR, RTNT: investigation, writing—review and editing. MAJvdS, HG: resources, writing—review and editing. PAVV: methodology, data curation, investigation, writing—review and editing. MLK: conceptualisation, methodology, supervision, writing—review and editing. NdM, JVMGB: conceptualisation, methodology, supervision, writing—original draft, writing—review and editing. Guarantor: JVMGB.
Funding: This work was financially supported by the intramural Leiden Center for Computational Oncology strategic fund. NdM. is funded by the European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Program (grant agreement number 852832) and by the VIDI ZonMW (project number: 09150172110092).
Competing interests: None declared.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review: Not commissioned; externally peer-reviewed.
Supplemental material: This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Data availability statement

Data are available in a public, open access repository. The RNA-seq data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE237033. The IMC images generated in this study are publicly available in BioImage Archive at S-BIAD833. The code used to process data is publicly available on the LUMC Gitlab.

Ethics statements

Patient consent for publication:

Ethics approval:

This study involves human participants and was approved by Medisch-Ethische Toetsingscommissie Leiden Den Haag Delft) B20.067. Participants gave informed consent to participate in the study before taking part.

Acknowledgements

The authors thank Dr. A.M. Cleton-Jansen for helpful discussions; Dr. M.H.M. Heemskerk and her research group for great support and discussions; The Sequencing Analysis Support Core (LUMC, Leiden) and the Flow Core Facility (LUMC, Leiden) for their support.

WHO Classification of Tumours Editorial Board. WHO classification of tumours of soft tissue and bone. Lyon, France, IARC Press 2020;
Google Scholar
Sadykova LR, Ntekim AI, Muyangwa-Semenova M, et al. Epidemiology and Risk Factors of Osteosarcoma. Cancer Invest 2020; 38:259–69.
doi:10.1080/07357907.2020.1768401•Google Scholar
Lindsey BA, Markel JE, Overview K, et al. Rheumatol Ther. 2017; 4:25–43.
doi:10.1007/s40744-016-0050-2•Google Scholar
Mirabello L, Troisi RJ, Savage SA, et al. Osteosarcoma incidence and survival rates from 1973 to 2004: data from the Surveillance, Epidemiology, and End Results Program. Cancer 2009; 115:1531–43.
doi:10.1002/cncr.24121•Google Scholar
Allison DC, Carney SC, Ahlmann ER, et al. A meta-analysis of osteosarcoma outcomes in the modern medical era. Sarcoma 2012; 2012.
doi:10.1155/2012/704872•Google Scholar
Tawbi HA, Burgess M, Bolejack V, et al. Pembrolizumab in advanced soft-tissue sarcoma and bone sarcoma (SARC028): a multicentre, two-cohort, single-arm, open-label, phase 2 trial. Lancet Oncol 2017; 18:1493–501.
doi:10.1016/S1470-2045(17)30624-1•Google Scholar
Le Cesne A, Marec-Berard P, Blay J-Y, et al. Programmed cell death 1 (PD-1) targeting in patients with advanced osteosarcomas: results from the PEMBROSARC study. Eur J Cancer 2019; 119:151–7.
doi:10.1016/j.ejca.2019.07.018•Google Scholar
Somaiah N, Conley AP, Parra ER, et al. Durvalumab plus tremelimumab in advanced or metastatic soft tissue and bone sarcomas: a single-centre phase 2 trial. Lancet Oncol 2022; 23:1156–66.
doi:10.1016/S1470-2045(22)00392-8•Google Scholar
Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome Med 2017; 9.
doi:10.1186/s13073-017-0424-2•Google Scholar
Cristescu R, Mogg R, Ayers M, et al. Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy. Science 2018; 362.
doi:10.1126/science.aar3593•Google Scholar
Cheung LS, Chen L, Oke TF, et al. Anti-PD-1 elicits regression of undifferentiated pleomorphic sarcomas with UV-mutation signatures. J Immunother Cancer 2021; 9.
doi:10.1136/jitc-2021-002345•Google Scholar
Davoli T, Uno H, Wooten EC, et al. Tumor aneuploidy correlates with markers of immune evasion and with reduced response to immunotherapy. Science 2017; 355.
doi:10.1126/science.aaf8399•Google Scholar
Chen X, Bahrami A, Pappo A, et al. Recurrent somatic structural variations contribute to tumorigenesis in pediatric osteosarcoma. Cell Rep 2014; 7:104–12.
doi:10.1016/j.celrep.2014.03.003•Google Scholar
Buddingh EP, Kuijjer ML, Duim RAJ, et al. Tumor-infiltrating macrophages are associated with metastasis suppression in high-grade osteosarcoma: a rationale for treatment with macrophage activating agents. Clin Cancer Res 2011; 17:2110–9.
doi:10.1158/1078-0432.CCR-10-2047•Google Scholar
Toda Y, Kohashi K, Yamada Y, et al. PD-L1 and IDO1 expression and tumor-infiltrating lymphocytes in osteosarcoma patients: comparative study of primary and metastatic lesions. J Cancer Res Clin Oncol 2020; 146:2607–20.
doi:10.1007/s00432-020-03242-6•Google Scholar
Wang D, Niu X, Wang Z, et al. Multiregion Sequencing Reveals the Genetic Heterogeneity and Evolutionary History of Osteosarcoma and Matched Pulmonary Metastases. Cancer Res 2019; 79:7–20.
doi:10.1158/0008-5472.CAN-18-1086•Google Scholar
Sundara YT, Kostine M, Cleven AHG, et al. Increased PD-L1 and T-cell infiltration in the presence of HLA class I expression in metastatic high-grade osteosarcoma: a rationale for T-cell-based immunotherapy. Cancer Immunol Immunother 2017; 66:119–28.
doi:10.1007/s00262-016-1925-3•Google Scholar
Ligon JA, Choi W, Cojocaru G, et al. Pathways of immune exclusion in metastatic osteosarcoma are associated with inferior patient outcomes. J Immunother Cancer 2021; 9.
doi:10.1136/jitc-2020-001772•Google Scholar
Wu C-C, Beird HC, Andrew Livingston J, et al. Immuno-genomic landscape of osteosarcoma. Nat Commun 2020; 11.
doi:10.1038/s41467-020-14646-w•Google Scholar
van den Bulk J, Verdegaal EME, Ruano D, et al. Neoantigen-specific immunity in low mutation burden colorectal cancers of the consensus molecular subtype 4. Genome Med 2019; 11.
doi:10.1186/s13073-019-0697-8•Google Scholar
Parkhurst MR, Robbins PF, Tran E, et al. Unique Neoantigens Arise from Somatic Mutations in Patients with Gastrointestinal Cancers. Cancer Discov 2019; 9:1022–35.
doi:10.1158/2159-8290.CD-18-1494•Google Scholar
D’Angelo SP, Melchiori L, Merchant MS, et al. Antitumor Activity Associated with Prolonged Persistence of Adoptively Transferred NY-ESO-1 c259T Cells in Synovial Sarcoma. Cancer Discov 2018; 8:944–57.
doi:10.1158/2159-8290.CD-17-1417•Google Scholar
Robbins PF, Kassim SH, Tran TLN, et al. A pilot trial using lymphocytes genetically engineered with an NY-ESO-1-reactive T-cell receptor: long-term follow-up and correlates with response. Clin Cancer Res 2015; 21:1019–27.
doi:10.1158/1078-0432.CCR-14-2708•Google Scholar
Somaiah N, Block MS, Kim JW, et al. First-in-Class, First-in-Human Study Evaluating LV305, a Dendritic-Cell Tropic Lentiviral Vector, in Sarcoma and Other Solid Tumors Expressing NY-ESO-1. Clin Cancer Res 2019; 25:5808–17.
doi:10.1158/1078-0432.CCR-19-1025•Google Scholar
Ishihara M, Tono Y, Miyahara Y, et al. First-in-human phase I clinical trial of the NY-ESO-1 protein cancer vaccine with NOD2 and TLR9 stimulants in patients with NY-ESO-1-expressing refractory solid tumors. Cancer Immunol Immunother 2020; 69:663–75.
doi:10.1007/s00262-020-02483-1•Google Scholar
Meijer DM, Miranda N, Bovée J, et al. MAGEA antigens as immunotherapeutic targets for advanced osteosarcoma. Gene Expression Omnibus 2024;
Available: here
Google Scholar
Anders S, Pyl PT, Huber W, et al. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 2015; 31:166–9.
doi:10.1093/bioinformatics/btu638•Google Scholar
Love MI, Huber W, Anders S, et al. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014; 15:550.
doi:10.1186/s13059-014-0550-8•Google Scholar
Szolek A, Schubert B, Mohr C, et al. OptiType: precision HLA typing from next-generation sequencing data. Bioinformatics 2014; 30:3310–6.
doi:10.1093/bioinformatics/btu548•Google Scholar
Roelands J, Hendrickx W, Zoppoli G, et al. Oncogenic states dictate the prognostic and predictive connotations of intratumoral immune response. J Immunother Cancer 2020; 8.
doi:10.1136/jitc-2020-000617•Google Scholar
Wang E, Worschech A, Marincola FM, et al. The immunologic constant of rejection. Trends Immunol 2008; 29:256–62.
doi:10.1016/j.it.2008.03.002•Google Scholar
Mohseny AB, Tieken C, van der Velden PA, et al. Small deletions but not methylation underlie CDKN2A/p16 loss of expression in conventional osteosarcoma. Genes Chromosomes Cancer 2010; 49:1095–103.
doi:10.1002/gcc.20817•Google Scholar
Ijsselsteijn ME, van der Breggen R, Farina Sarasqueta A, et al. A 40-Marker Panel for High Dimensional Characterization of Cancer Immune Microenvironments by Imaging Mass Cytometry. Front Immunol 2019; 10.
doi:10.3389/fimmu.2019.02534•Google Scholar
Ijsselsteijn ME, Somarakis A, Lelieveldt BPF, et al. Semi-automated background removal limits data loss and normalizes imaging mass cytometry data. Cytometry A 2021; 99:1187–97.
doi:10.1002/cyto.a.24480•Google Scholar
Somarakis A, Van Unen V, Koning F, et al. ImaCytE: Visual Exploration of Cellular Micro-Environments for Imaging Mass Cytometry Data. IEEE Trans Vis Comput Graph 2021; 27:98–110.
doi:10.1109/TVCG.2019.2931299•Google Scholar
Zhang W, Wang H, Sun M, et al. CXCL5/CXCR2 axis in tumor microenvironment as potential diagnostic biomarker and therapeutic target. Cancer Commun (Lond) 2020; 40:69–80.
doi:10.1002/cac2.12010•Google Scholar
Sprenkeler EGG, Zandstra J, van Kleef ND, et al. S100A8/A9 Is a Marker for the Release of Neutrophil Extracellular Traps and Induces Neutrophil Activation. Cells 2022; 11.
doi:10.3390/cells11020236•Google Scholar
Browne T, Dearman RJ, Poles A, et al. Human neutrophil antigens: Nature, clinical significance and detection. Int J Immunogenet 2021; 48:145–56.
doi:10.1111/iji.12514•Google Scholar
Wei R, Dean DC, Thanindratarn P, et al. Cancer testis antigens in sarcoma: expression, function and immunotherapeutic application. Cancer Lett 2020; 479:54–60.
doi:10.1016/j.canlet.2019.10.024•Google Scholar
Bertucci F, Niziers V, de Nonneville A, et al. Immunologic constant of rejection signature is prognostic in soft-tissue sarcoma and refines the CINSARC signature. J Immunother Cancer 2022; 10.
doi:10.1136/jitc-2021-003687•Google Scholar
Sherif S, Roelands J, Mifsud W, et al. The immune landscape of solid pediatric tumors. J Exp Clin Cancer Res 2022; 41.
doi:10.1186/s13046-022-02397-z•Google Scholar
Galon J, Bruni D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat Rev Drug Discov 2019; 18:197–218.
doi:10.1038/s41573-018-0007-y•Google Scholar
Zou C, Shen J, Tang Q, et al. Cancer-testis antigens expressed in osteosarcoma identified by gene microarray correlate with a poor patient prognosis. Cancer 2012; 118:1845–55.
doi:10.1002/cncr.26486•Google Scholar
Ulloa-Montoya F, Louahed J, Dizier B, et al. Predictive gene signature in MAGE-A3 antigen-specific cancer immunotherapy. J Clin Oncol 2013; 31:2388–95.
doi:10.1200/JCO.2012.44.3762•Google Scholar
Morgan RA, Chinnasamy N, Abate-Daga D, et al. Cancer Regression and Neurological Toxicity Following Anti-MAGE-A3 TCR Gene Therapy. J Immunother 2013; 36:133–51.
doi:10.1097/CJI.0b013e3182829903•Google Scholar
Martin AD, Wang X, Sandberg ML, et al. Re-examination of MAGE-A3 as a T-cell Therapeutic Target. J Immunother 2021; 44:95–105.
doi:10.1097/CJI.0000000000000348•Google Scholar
Lu Y-C, Parker LL, Lu T, et al. Treatment of Patients With Metastatic Cancer Using a Major Histocompatibility Complex Class II-Restricted T-Cell Receptor Targeting the Cancer Germline Antigen MAGE-A3. J Clin Oncol 2017; 35:3322–9.
doi:10.1200/JCO.2017.74.5463•Google Scholar
de Rooij MAJ, Remst DFG, van der Steen DM, et al. A library of cancer testis specific T cell receptors for T cell receptor gene therapy. Mol Ther Oncolytics 2023; 28:1–14.
doi:10.1016/j.omto.2022.11.007•Google Scholar
Tsukahara T, Kawaguchi S, Torigoe T, et al. Prognostic significance of HLA class I expression in osteosarcoma defined by anti-pan HLA class I monoclonal antibody, EMR8-5. Cancer Sci 2006; 97:1374–80.
doi:10.1111/j.1349-7006.2006.00317.x•Google Scholar
Martin JW, Squire JA, Zielenska M, et al. The genetics of osteosarcoma. Sarcoma 2012; 2012.
doi:10.1155/2012/627254•Google Scholar

Overview

Abstract
Introduction
Materials and methods
Results
Discussion
References
Supplementary files
Footnotes
Publication history
Rapid Responses

Article metrics

Altmetric

Dimensions