A novel crosstalk between immunogenic and oncometabolic pathways triggered by T cell-released interferon-gamma (IFN-ɣ) has been recently identified. This IFN-ɣ-pyruvate kinase M2-β-catenin axis relies on fibroblast growth factor 2 (FGF2) signaling in tumor cells and leads to hyperprogressive disease on immune checkpoint blockade (ICB) in preclinical models. This result underlines how IFN-ɣ signaling may have distinct effects on tumor cells depending on their oncogenic and metabolic features. On the basis of these data, this study aims to explore the relationship between genomic tumor FGF2 or FGF/FGF receptor (FGFR) amplification and immunotherapy response in patients with metastatic solid cancers. We used a large genomic data set of 545 ICB-treated patients and compared outcomes between those with and without FGF2 genomic amplification. Patients with no FGF2 genomic amplification had significantly longer progression-free survival (PFS) (HR=0.55 (95% CI 0.4, 0.8); p value=0.005) and overall survival (OS) (HR=0.56 (0.3, 0.9); p value=0.02) than patients harboring an FGF2 amplification. We next questioned whether such an observation may extend to genomic amplification of the FGF/FGFR pathway. Similarly, patients with no FGF/FGFR genomic amplification had longer PFS (HR=0.71 (0.8, 0.9), p value=0.004) and OS (HR=0.77 (0.6, 1); p value=0.06). RNA sequencing analysis of tumors between the amplified and non-amplified populations showed distinct expression profiles concerning oncogenic pathways. Importantly, using a cohort of patients untreated with ICB from the The Cancer Genome Atlas, we show that FGF2 and FGF/FGFR genomic amplification were not associated with prognosis, thus demonstrating that we identified a predictive biomarker of immunotherapy resistance.
- Immune Checkpoint Inhibitors
- Tumor Biomarkers
- Biomarkers, Tumor
- Metabolic Networks and Pathways
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Immune Checkpoint Inhibitors
- Tumor Biomarkers
- Biomarkers, Tumor
- Metabolic Networks and Pathways
Immune checkpoint blockade (ICB) given as a monotherapy or in combination, is the first-line standard of care for a growing number of metastatic solid cancers1 and its use now extends to earlier stage diseases.2 3 Immunotherapy reinvigorates T cells, thus eliciting a prolonged antitumor immune response.4 Nevertheless, less than one-fifth of patients are expected to respond to such therapy.5 When treated with ICB, some patients experience a response pattern with rapid progression that has been characterized as hyperprogressive disease (HPD),6 resulting in reduced overall survival (OS). Interferon-gamma (IFN-ɣ), a cytokine produced within the tumor microenvironment (TME) by numerous immune cells, seems to exert a dual role on the antitumor immune response in the context of cancer immunotherapy.7 Emerging data suggest that tumors harboring IFN-ɣ signaling gene mutations may be more prone to respond to ICB.8 Li et al recently reported that IFN-ɣ release by CD8+ T cells during ICB promotes fibroblast growth factor 2 (FGF2) production and signaling in tumor cells, as an oncometabolic reprogramming pathway through enhanced β-catenin acetylation, which drives HPD.9 FGF signaling could be induced by IFN-ɣ pathway signaling but also by oncogenic genomic variation like fusion, mutation or amplification.10 Because of the frequent amplification of FGF genes in solid tumors we sought to investigate whether there is a relationship between genomic tumor FGF2 or FGF gene amplification and survival outcomes under ICB in order to identify a potential predictive biomarker.
Material and methods
Genomic data set of ICB-treated patients
We combined two genomic data sets of patients treated with ICB. The first one comprised diagnosed with metastatic solid cancer included in the precision medicine trials EXOMA1 (NCT02840604) and EXOMA2 (NCT04614480) (n=190), and treated with immunotherapy (antiprogrammed death-1/ligand-1: anti-PD1/PD-L1; anticytotoxic T-lymphocyte-associated protein-4: anti-CTLA-4; anti-T cell immunoreceptor with Ig and ITIM domains: anti-TIGIT) in first or further lines.11 FGF1-23 and FGFR1-4 genomic amplifications were detected using the SuperFreq12 algorithm, and for each gene amplification was retained if the software detected more than two copies of the gene concerned. Patients with at least one amplification of FGF1-23 or FGFR1-4 genes were considered respectively FGF or FGF receptor (FGFR) amplified. As paired normal-tumor samples were not available for the whole cohort from trials EXOMA1 and EXOMA2, only tumor samples were considered in this analysis for copy number alterations estimation. SuperFreq algorithm does not require a matched normal and instead relies on unrelated controls. The second genomic data set contained data on melanoma-harboring patients treated with immunotherapy in first or further lines (n=355).13–16 For the different cohorts downloaded, copy number variations were obtained from Affymetrix SNP 6.0 platform.
TCGA control data set of patients untreated with ICB
We used a set of patients untreated with ICB from The Cancer Genome Atlas (TCGA) data set as control. We included in this cohort patients with lung, breast, colorectal or melanoma tumors. Genomic data were downloaded using the R package TCGAbiolinks.17 As for the ICB-treated cohort, FGF1-23 and FGFR1-4 amplifications were retained if the patient had more than two copies of the relevant genes. Patients with at least one FGF1-23 or FGF1-4 amplification were considered respectively FGF or FGFR amplified.
For each patient from the TCGA cohort, single-sample Gene Set Enrichment Analysis (ss-GSEA) was used to estimate expression of cancer-associated fibroblast (CAF) signaling pathways and Hallmark gene set signatures (https://www.gsea-msigdb.org/gsea/msigdb). Expanded immune gene (EIG),18 T-score19 and CXC chemokine ligand (CXCL) signatures20 were estimated by averaging expression of relevant genes (online supplemental table 1).
Patient characteristics are described as median and IQR for continuous variables and as number and percentage (%) for qualitative variables.
Progression-free survival (PFS) and OS were censured at 12 months for patients treated with ICB and disease-free survival (DFS) and OS were censured at 60 months for patients untreated with ICB.
Survival analysis was performed using the survival R library. The prognostic value of the different variables were tested using univariate and multivariate Cox models for PFS and OS. Survival probabilities were estimated using the Kaplan-Meier method, and survival curves were compared using the log-rank test. P values less than 0.05 were considered statistically significant. The association between FGF/FGFR amplification status and Hallmark oncogenic pathways or immunologic signatures was determined using median comparison and Wilcoxon tests. P values less than 0.05 were considered significant.
Patient baseline characteristics
Overall, the ICB-treated cohort included 545 patients with mainly melanoma (n=364, 67%), lung (n=67, 12%), colorectal (n=20, 3.7%), or breast cancer (n=15, 2.8%); the majority were treated with ICB as monotherapy (n=456; 84%) and in the second line of metastatic disease (n=232; 43%) (table 1). This cohort included 79 (14.5%) patients with other tumor types (online supplemental table 2). FGF2 amplification was reported for 32 (6%) patients, any FGF genes amplification alone for 86 (16%) patients, FGFR amplification alone for 36 (6.6%) patients, and amplification in any FGF and FGFR genes 115 (21%) patients. Among 545 patients 308 (57%) patients did not have any amplification of the FGF or FGFR gene. The control TCGA cohort included 2466 patients with similar tumor type, that is, breast (n=1,065, 43%), lung (n=498, 20%), melanoma (n=452, 18%), and colorectal cancer (n=451, 18%). In this data set, FGF2 amplification was reported for 864 (35%) patients, any FGF gene amplification alone for 459 (19%) patients, any FGFR gene amplification alone for 53 (2%) patients, and amplification of any FGF and FGFR gene for 1487 (60%) patients; while 467 (59%) patients did not have any amplification of the FGF or FGFR gene. The distribution of FGF2 amplifications according to FGF/FGFR amplifications is provided in online supplemental table 3.
Survival on ICB among patients harboring FGF2 and FGF/FGFR amplification
In order to assess the relationship between genomic tumor FGF2 amplification and immune checkpoint blockade response, we compared outcomes between patients with and without FGF2 amplification. In our ICB-treated cohort, patients with no FGF2 amplification had significantly better PFS (HR=0.55 (95% CI 0.4, 0.8); p value=0.005, median PFS was 3.5 (95% CI 3.1, 4.4) vs 2.6 months (95% CI 2.2, 3.7) in FGF2Amp group) and OS (HR=0.56 (95% CI 0.3, 0.9); p value=0.02, median OS was Not Reached (NR) (95% CI NR, NR) vs 8.3 months (95% CI 5.1, NR) in FGF2Amp group) than patients with FGF2 amplification (figure 1A,B). We next investigated whether this observation could be extended to the FGF/FGFR pathway. Similarly to what was observed with FGF2, patients with no FGF or FGFR amplification had better PFS than patients with at least one FGF or FGFR amplification (HR=0.71 (95% CI 0.8, 0.9), p value=0.004, median PFS was 4.2 (95% CI 3.2, 5.6) in FGFNo Amp/FGFRNo Amp group vs 2.8 months (95% CI 2.6, 3.3) in Other group). A similar trend was observed for OS (HR=0.77 (95% CI 0.6, 1); p value=0.06, median OS was not reached in both groups) (figure 1C,D). Subgroups analyses showed that patients classified FGFNo Amp/FGFRNo Amp had significantly better PFS and OS than patients classified FGFAmp/FGFRAmp (HR=0.65 (95% CI 0.5, 0.9), p value=0.003; and HR=0.7 (95% CI 0.5, 1), p value=0.03, respectively) (figure 1E,F). PFS were not significantly different between FGFAmp/FGFRAmp and FGFNo Amp/FGFRAmp (HR=0.93 (95% CI 0.6, 1.5), p value=0.76) as between FGFAmp/FGFRAmp and FGFAmp/FGFRNo Amp patients (HR=0.78 (95% CI 0.5, 1.2), p value=0.21). Similar results were found for OS with no significant difference between FGFAmp/FGFRAmp and FGFNo Amp/FGFRAmp (HR=1 (95% CI 0.6, 1.8), p value=0.91) as between FGFAmp/FGFRAmp and FGFAmp/FGFRNo Amp patients (HR=0.72 (95% CI 0.5, 1.1), p value=0.13). We performed univariate Cox analysis for OS and PFS according to sex, number of lines at ICB initiation, treatment regimen, and cancer type (melanoma vs other). Only significant variables were selected to perform a multivariate Cox analysis with the FGF/FGFR status. Patients who received ICB at the second or further line and patients with melanoma cancer had a worse PFS than other patients (HR=1.84 (95% CI 1.44, 2.33); p<0.001 and HR=1.27 (95% CI 1.03, 1.58); p=0.03 respectively, online supplemental figure 1A). Using multivariate Cox analysis with these variables and FGF/FGFR status, all variables remained significantly associated with PFS. Moreover, patients who received ICB at second or further lines had a worse OS than others (HR=2.23 (95% CI 1.63, 3.06); p<0.001, online supplemental figure 1B). A multivariate Cox analysis was performed, including FGF/FGFR status and line of therapy, and only the latter remained significantly associated with OS. These results underline that the FGF/FGFR status was an independent factor for PFS but not for OS. Altogether, these findings demonstrate that genomic tumor FGF2 amplification is associated with poorer outcomes on ICB. Additionally, we extended such an observation on ICB resistance to FGF and FGR genomic amplification.
FGF2 and FGF/FGFR amplification are predictive biomarkers of ICB resistance
To determine the predictive versus prognostic value of genomic tumor FGF2 and FGF/FGFR amplification, we used our TCGA control data set of patients untreated with ICB. Importantly, FGF2 amplification was not associated in this cohort with DFS or OS (HR=0.9 (95% CI 0.7, 1.2), p value=0.29 and HR=1.04 (95% CI 0.8, 1.2), p value=0.96), respectively (figure 2A,B). Similar results were observed with FGF/FGFR amplification status. FGFNo Amp/FGFRNo Amp patients had similar survival compared with patients with amplification of at least one of FGF or FGFR amplification (DFS: HR=0.77 (95% CI 0.5, 1.1), p value=0.2, OS: HR=0.91 (95% CI 0.7, 1.1), p value=0.29) (figure 2C,D). Collectively, these data demonstrate that genomic tumor FGF2 and FGF/FGFR amplification are predictive biomarkers of response to ICB but are not prognostic.
Transcriptomic characterization of FGF/FGFR amplified tumors
Next, to test the association between FGF/FGFR amplification and ICB biomarkers, we used RNA sequencing (RNAseq) data from the TCGA cohort. FGFAmp/FGFRAmp tumors significantly upregulated the MYC and MTOR oncogenic pathways compared with other tumors (Wilcoxon p value<0.05) using GSEA Hallmark analysis. Regarding immunologic signatures, FGFAmp/FGFRAmp was associated with downregulation of T-score and EIG signature (figure 3). No differences were found neither on chemokines (CXCL9, CXCL10, and CXCL11) nor on immune checkpoint expression (PDCD1, CTLA4). These data highlight the transcriptomic discrepancies existing between FGF/FGFR amplified and non-amplified tumors.
In our study, we show in a large genomic data set of ICB-treated patients that genomic tumor FGF2 amplification is associated with a shorter survival on immunotherapy and confirm its role as a predictive biomarker of ICB resistance. FGF2 has recently been identified as a key component of the IFN-ɣ-pyruvate kinase M2 (PKM2)-β-catenin axis, which promotes HPD on immunotherapy.9 Reinvigorated T cells release IFN-ɣ thus achieving immunologic signaling, which promotes FGF2 activation. Consequently, it triggers the FGF2 pathway, which results through PKM2 phosphorylation, in the activation of β-catenin signaling genes that drive cancer hyperprogression. β-catenin pathway activation is a classical mode of resistance to immunotherapy, notably by its capacity to limit CCL4 production and dendritic cells recruitment.21 Thus, these preclinical findings may biologically explain our observations of poorer outcomes in ICB-treated patients harboring an FGF2 genomic amplification.
Among the 23 members of the FGF family, there are 18 FGFR ligands that are divided into six families. These proteins bind to the extracellular domain of four tyrosine kinase receptors22 (FGFR1, FGFR2, FGFR3, and FGFR4) or FGFRL1, which does not have an intracellular domain.23 FGF2 is a secreted FGFR ligand that belongs to the FGF1 subfamily and binds preferentially to FGFR1 and FGFR3. On activation, phosphorylated FGFRs promote downstream signaling cascades such as the Ras/Raf-MEK-mitogen-activated protein kinases (MAPKs), the signal transducer and activator of transcription 1, 3, and 5 (STAT1, STAT3, STAT5), the phophastidylinositol-3 kinase/protein kinase B (PI3K/AKT), the phospholipase Cɣ (PLCɣ) and the wingless-related integration site (WNT)/β-catenin pathway.24 Thus, FGFs and their associated receptors (FGFRs) control a number of cytosolic signaling pathways involved in cell migration, angiogenesis, survival, and proliferation that are of particular importance in the context of cancer.
Aberrant FGF signaling in cancer cell can occur through several mechanisms. On one hand, FGFR activating mutations, chromosomal translocations or, gene amplification may lead to ligand-independent signaling. On the other hand, activation of the FGF/FGFR pathway in cancer cells may be ligand-dependent in the case of FGF genomic amplification, overexpression or, mutations that lead to the establishment of an autocrine or a paracrine loop if FGFs are produced by cancer or stromal cells, respectively.10
FGF signaling also affects the constitution and immune polarization of the TME by reducing antitumor immunity and promoting immunosuppressive features.25 For example, FGFR1 activation in lung squamous cell carcinoma is associated with tumor immune evasion, and FGFR1 knockdown allowed the reversion of such a phenotype with the recruitment of CD8, CD4 T cells, and M1 macrophages in vivo and the reduction of PD-1 expression on CD8 T cells.26 Testifying of the role of FGF signaling in the TME, FGF2 produced by CAFs promotes the FGF2-FGFR1 paracrine loop in breast tumors that could favor tumor growth.27 In our RNAseq analysis, it is therefore unsurprising that FGF/FGFR amplification was associated with a downregulation of immunological signatures. Of note, in line with previously published data, FGF/FGFR amplification was associated with upregulation of two major oncogenic pathways, MYC28 and MTOR.29 Nevertheless, these oncogenic and immunologic features observed with FGF/FGFR amplification remain unassociated with prognosis (figure 2). FGFR signaling regulates PD-1/PD-L1, inhibits MHC (major histocompatibility complex) I and II expression, and may promote a non-T cell infiltrative phenotype.25 Thus, impairing the FGFR pathway could be synergic with immune checkpoint blockade. Tyrosine kinase inhibitors that target FGFRs are now validated options for patients harboring FGFR alterations since erdafitinib, futibatinib, infigratinib, and pemigatinib are Food and Drug Administration-approved drugs. While the first is available for patients with metastatic urothelial carcinoma that have FGFR3 or FGFR2 genetic alterations, the last three drugs are an option for cholangiocarcinoma with FGFR2 fusion or other rearrangement. Clinical trials that assess FGFR inhibitors in combination with immunotherapy are recruiting, notably for urothelial carcinoma (NCT05173142, NCT05775874), liver (NCT04828486), gastric/gastroesophageal cancer and intrahepatic cholangiocarcinoma (ICC) (NCT05173142).
A limitation of our study is the heterogeneity of the cohort in terms of tumor and treatment types. Because of the retrospective design, these results should be considered exploratory. In line with a preclinical study that have identified an oncometabolic reprogramming pathway in FGF2+ tumor cells with ICB, we found that patients harboring genomic tumor FGF2 amplification have worse outcomes with immunotherapy, and we have extended this observation to amplification of the FGF/FGFR signaling pathway. A strength of our study is the use of a control group treated without ICB, which strongly supports a predictive effect of FGF/FGFR amplification rather than a prognostic effect. Our study highlights that genomic tumor FGF/FGFR amplification could be added as an additional biomarker of ICB resistance.
Data availability statement
Data are available upon reasonable request.
Patient consent for publication
The study was conducted according to the guidelines of the Declaration of Helsinki and European legislation and approved by the CNIL (French national commission for data privacy) and the Georges François Leclerc Cancer Center (Dijon, France) local ethics committee (13.085). Participants gave informed consent to participate in the study before taking part.
We wish to thank Fiona Ecarnot (EA3920, University of Franche-Comté, Besançon, France) for English correction and helpful comments.
Contributors Conceptualisation: NR, FG. Methodology: JL, CT. Validation: CT, FG. Formal analysis: JL. Data acquisition: NR, LD, JL, CT, LF. Writing—original draft preparation: NR, JL, CT, FG. Visualisation: JL, CT. Supervision: CT, FG. All authors have read and agreed to the published version of the manuscript. Guarantor: FG.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
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.