Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Gene expression signatures predictive of bevacizumab/erlotinib therapeutic benefit in advanced non-squamous non-small cell lung cancer patients (SAKK 19/05 trial) Anca Franzini1, Florent Baty1, Ina I. Macovei2, Oliver Dürr3, Cornelia Droege4, Daniel Betticher5, Bogdan D. Griogriu2, Dirk Klingbiel6, Francesco Zappa7, and Martin H. Brutsche1 1

Department of Pulmonary Medicine, Cantonal Hospital St. Gallen, St. Gallen, Switzerland

2

Department of Pulmonary Diseases, University of Medicine and Pharmacy, Iasi, Romania

3

Institute of Data Analysis and Process Design, Zürich University of Applied Sciences, Winterthur, Switzerland

4

Männedorf Hospital, Männedorf, Switzerland

5

Cantonal Hospital Fribourg, Fribourg, Switzerland

6

Swiss Group for Clinical Cancer Research (SAKK) Coordinating Center, Bern, Switzerland

7

Department of Medical Oncology, Clinica Luganese, Lugano, Switzerland

Corresponding author: Martin H. Brustche, Cantonal Hospital St. Gallen, 95 Rorschacher Strasse, 9007 St. Gallen, Switzerland. Phone: 004171-494-1004; Fax: 004171-494-6118; E-mail: [email protected]

Grant Support Swiss Cancer League & Swiss Cancer Center (KLS-2880-02-2012): Martin H Brutsche, KSSG Medical Research Center (MFZF_2014_001): Anca Franzini, Romanian–Swiss Research Program (IZERZO_142235/1): Ina I. Macovei. Abstract Purpose: We aimed to identify gene expression signatures associated with angiogenesis and hypoxia pathways with predictive value for treatment response to bevacizumab/erlotinib (BE) of nonsquamous advanced NSCLC patients. Experimental design: Whole genome gene expression profiling was performed on 42 biopsy samples (from SAKK 19/05 trial) using Affymetrix exon arrays, and associations with the following endpoints: time-to-progression (TTP) under therapy, tumor-shrinkage (TS), and overall survival (OS) were investigated. Next, we performed gene set enrichment analyses using genes associated with the angiogenic process and hypoxia response to evaluate their predictive value for patients’ outcome. Results: Our analysis revealed that both the angiogenic and hypoxia response signatures were enriched within the genes predictive of BE response, TS and OS. Higher gene expression levels (GELs) of the 10-gene angiogenesis-associated signature and lower levels of the 10-gene hypoxia response signature predicted improved TTP under BE, 7.1 months vs. 2.1 months for low vs. high-risk patients (P = 0.005), and median TTP 6.9 months vs. 2.9 months (P = 0.016), respectively. The hypoxia

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

response signature associated with higher TS at 12 weeks and improved OS (17.8 months vs. 9.9 months for low vs. high risk patients, P = 0.001). Conclusions: We were able to identify gene expression signatures derived from the angiogenesis and hypoxia response pathways with predictive value for clinical outcome in advanced non-squamous NSCLC patients. This could lead to the identification of clinically relevant biomarkers, which will allow for selecting the subset of patients who benefit from the treatment and predict drug response.

STATEMENT OF TRANSLATIONAL RELEVANCE Clinical outcome of non-small cell lung cancer (NSCLC) could be improved by molecular stratification of patients. Antiangiogenic therapy is approved for treatment of advanced cancers, however because only a subset of patients treated with angiogenesis inhibitors show objective clinical response, there is an increased need for predictive biomarkers. We identified 10-gene signatures associated with angiogenesis and hypoxia-response pathways, which have predictive value for response to combined anti-VEGF/anti-EGFR in nonsquamous advanced NSCLC. Subclassification of the patients using these signatures indicated that patients most likely to respond (low-risk) showed higher levels of angiogenesis associated genes mainly involved in maintaining the vascular barrier integrity and lower levels of hypoxia-response genes. Moreover, they showed increased percentile tumor shrinkage and improved overall survival. The inverse trend was observed for the high-risk patients. These findings open the possibility for clinical use of these signatures as predictive biomarkers for identifying patients who would benefit from antiangiogenic therapies. Introduction Solid tumor cells in their primary goal to avoid senescence and achieve uncontrolled proliferation co-opt neighboring stromal cells to assist them with the expansion of the malignant tissue.1, 2 These tumor-associated stromal components have been demonstrated to have an important role in tumor growth, disease progression3 as well as in response and resistance to therapeutic agents.4,

5

Moreover, tumors are strongly dependent on

angiogenesis, the formation of neovessels from the preexisting vasculature, to obtain the nutrients and oxygen essential for their rapid growth.6 Consequently, antiangiogenic therapeutic strategies were extensively exploited in the last decade in search for more efficient cancer treatments.7

2

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Though multiple therapeutic strategies have been developed against vascular endothelial growth factor (VEGF),8 bevacizumab,9 a monoclonal antibody targeting VEGFA, is the first approved antiangiogenic drug for clinical use. Bevacizumab is mostly used in combination with standard chemotherapy for treatment of metastatic cancers,10-12 and as single agent in recurrent glioblastoma.13 However, survival benefits are observed only in a subset of patients, as a significant number of patients show only modest response presumably because of intrinsic or rapidly acquired resistance to antiangiogenic therapy.14 One proposed mechanism of resistance to antiangiogenic agents is the onset of hypoxia within the tumor as a result of vessel regression during the course of antiangiogenic therapy.15, 16 Moreover, effective inhibition of neovascularization using antiangiogenic therapy was shown in some cases to change the phenotype of tumors by increasing their invasion and metastatic potential.17 Additionally, significant rates of adverse effects were reported for patients receiving anti-VEGF therapy, as well as a mortality rate of 1%, which was a direct consequence of bevacizumab administration.18,

19

Thus, understanding all cascades of

vascular signaling involved in the response to antiangiogenic therapy and subsequent resistance is critical to achieve full potential of this therapeutic approach. Despite these obvious limitations of antiangiogenic therapy, because a subset of cancer patients treated with angiogenesis inhibitors show objective clinical response, there is an increased need to identify robust predictive biomarkers,20 which could allow for selecting the subgroup of patients who would benefit from the treatment. In a recent study, aimed at identifying novel biomarkers for response to combined antiVEGFA/anti-EGFR (epidermal growth factor receptor) therapy in non-squamous NSCLC by exploring gene expression at exon-level, we identified EGFR exon 18 as a predictive marker

3

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

for patients with metastatic non-squamous NSCLC who have received no previous therapy.21 The gene expression profiles obtained from this set of microarray data are derived, however, from highly heterogeneous clinical biopsies consisting of both tumor and activated stromal cells. In a previous study, Baty et al. revealed that prediction of survival was independent of tumor cell content present in each NSCLC biopsy.22 This suggests a strong predictive contribution from the tumor microenvironment compartments in NSCLC. Additionally, recent studies have demonstrated that tumor microenvironment can provide independent and reliable predictors of clinical outcome.23

A key constituent of the tumor

microenvironment is the blood vasculature, which undergoes angiogenesis to sustain the high proliferative rate of the tumor, and is the direct target of antiangiogenic therapies. Because there is a high degree of cross-talk between epidermal growth factor receptor EGF(R) and VEGF(R) pathways, they have been identified as potentially synergistic for dual targeting.24

EGFR

transcriptionally

pathway

up-regulating

is

involved VEGF

in

growth

expression.25

factor–induced Additionally,

angiogenesis,

multiple

studies

demonstrated that hypoxia can trigger the angiogenic switch in solid tumors.26 Therefore, we aimed to investigate whether angiogenesis and hypoxia-associated gene expression signatures could predict the combined anti-VEGF/anti-EGFR treatment response in advanced non-squamous therapy naїve NSCLC patients unselected for EGFR mutation status. Our analysis identified 10-gene angiogenesis-associated and hypoxia-response signatures predictive of therapeutic response to bevacizumab/erlotinib (BE) having time-to-progression (TTP) and tumor shrinkage (TS) as endpoints. We also identified for 10-gene signatures with prognostic value. These signatures hold great potential for clinical application allowing for identification of biomarkers, which can identify the patients most likely/ less likely to respond to targeted therapy. 4

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Methods Study design. Our study is based on clinical bronchoscopic biopsies available from 42 patients for which genome-wide gene expression was studied in a microarray platform. These patients (88% adenocarcinoma, 57% female, 31% never smoker) were enrolled in the Swiss Group for Clinical Cancer Research (SAKK) 19/05 phase II trial.27 Identification of predictive gene-expression signatures from RNA gene expression analysis was a predefined goal of this trial. The detailed clinical information of these patients was published previously.21 For these patients with stage IIIB or IV (93%) non-squamous NSCLC, BE was used as first-line therapy (independent of the EGFR mutation status) followed by standard platinum-based/gemcitabine chemotherapy (CT) after disease progression, (Figure 1). Time to disease progression and percentage tumor shrinkage at 12 weeks (assessed by CT scans) were defined according to RECIST criteria. To confirm that there was no sample selection bias for the 42 patients included in our study, we performed statistical analysis and found no significant differences between the study group and the patients with no available biopsies. The results are summarized in Supplementary Table 1.

Gene expression analysis Total RNA from 42 bronchoscopic biopsy samples were extracted using miRNeasy Mini Kit (Qiagen) according to the manufacturer's recommendations. Affymetrix Human Exon 1.0 ST arrays (Affymetrix, Santa Clara, CA, USA) were used for mRNA hybridization. The gene level probesets were preprocessed, quality-checked and normalized using the Robust Multi-array Average (RMA) procedure.28 Data were expressed as log2 ratio of fluorescence intensities of

5

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

the sample and the reference, for each element of the array. Additional experimental details for this section were previously reported by us.21

Statistical analysis of gene expression. Survival and time-to-event analysis were performed by applying univariate Cox proportional hazards regression and principal component analysis (metagene approach) according to a previously described method.29 We built a binary score (low/high risk) using the median of the metagene scores. The classification accuracy of our algorithm was assessed by leaveone-out cross-validation (LOOCV). Time-to-event and survival results were displayed using Kaplan-Meier curves, and log-rank tests are reported. Hierarchical cluster analysis was carried out using the Euclidean distance together with the complete linkage agglomerative method. The median follow up time was estimated using the reverse Kaplan-Meier method. All statistical tests were performed using the R statistical software version 3.1.0 (http://www.R-project.org). A P value of 0.05 was set as threshold for significance for all study outcomes. Gene set enrichment analysis (GSEA). We performed enrichment analysis using GSEA v2.1.0 software (http://www.broad.mit.edu/gsea) and GSEA-preranked function.30 The gene lists were ranked by using as metric –log10(P) resulted from the Cox-regression analysis for all endpoints. Quantitative real-time PCR analysis. Forty samples of RNA isolated from non-squamous NSCLC biopsies and previously analyzed by microarrays were available for reverse transcription (RT). RT was performed using QuanTiect® Reverse Transcription Kit from

6

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Qiagen starting from 10 ng RNA for each qPCR reaction. The quantity and integrity of RNAs were assessed using an UVS-99 micro-volume spectrophotometer (ACTGene). For this study we investigated the expression levels of nine angiogenesis-associated predictive genes. For RT-qPCR experiments we used predesigned optimized primer sets (annealing temperature 55 °C) for GPR116, EMCN, ITGA9, GNG11, KDR, PECAM1, S1PR1, JAM2 and RHOJ (QuantiTect Primer Assay, Qiagen). For sequences please refer to the Supplementary Table 2. All samples were processed in duplicate for the qPCR with the LightCycler®480 SYBR Green I Master in a 20 µL reaction volume containing 4 µL water, 1 µL PCR primer, 10 µL Master mix and 5 µL cDNA (10 ng). Quantitative real-time PCR experiments were performed on a LightCycler® 480 II instrument using the initial denaturation at 95 °C for 10 minutes followed by 45 cycles: 95 °C for 10 sec, 55 °C for 20 sec, 72 °C for 20 sec. Controls containing no reverse transcriptase were included for each sample. The mRNA expression levels were calculated relative to HPRT1 house-keeping gene and the relative quantification of the gene expression was performed using 2─ΔCp method.31 The correlation between the RT-qPCR gene expression levels (GELs) and microarray GELs was measured by means of Spearman's correlation coefficient. Results 10-gene angiogenesis-associated and hypoxia-response signatures predict response to BE therapy. We hypothesized that the genes which could associate with the response to BE therapy are genes involved in angiogenic signaling pathways. To investigate this, we performed GSEA30 using a core gene signature (43 genes) specific for angiogenesis transcriptional program predefined by integrative meta-analysis of the expression profiles of over 1,000 primary human cancers.32

7

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

GSEA revealed a significant enrichment of the angiogenesis-associated genes within the genes that associate with TTP under BE therapy endpoint (mean rank = 4246, P=0.004, Figure 2A). As VEGF expression is directly activated under hypoxic conditions by the transcription factor hypoxia inducible factor 1 alpha (HIF1α), we decided to additionally investigate hypoxia-response genes by GSEA. For this analysis we used a previously derived common hypoxia metagene (51 genes) across cancer types.33 Within the genes which associate with TTP under BE therapy, we identified by GSEA a significant enrichment of the hypoxia-response genes (mean rank = 4798, P = 0.001, Figure 2B). The top 12-ranked angiogenesis-associated and hypoxia-response genes, which significantly correlate with TTP under BE therapy are given in Figure 2C and 2D. We

then

applied

unsupervised

hierarchical

clustering

to

the

top

10-ranked

angiogenesis-associated genes to investigate whether there are gene expression patterns correlating with the BE treatment response. The clustering output for the angiogenesisassociated signature is displayed in Figure 3A. We identified three distinct gene clusters: Cluster A1 (low risk, 9 patients, 21 %) was characterized by an increased expression of the angiogenesis-associated signature and associated with improved TTP under BE 7.1 months, 95% confidence interval (95% CI): 4.0 ─ ∞; Cluster A2 (high risk, 12 patients, 29 %) was characterized by a decreased expression of the angiogenesis-associated gene signature and associated with reduced response to BE treatment with a mean TTP under BE of 2.1 months (95% CI: 1.4 ─ ∞). The patients within the third cluster A3 (medium risk, 21 patients, 50 %) showed intermediate angiogenesis-associated GELs and a median TTP under BE of 4.1 months (95%CI: 3.1 ─ 7). The Kaplan-Meier TTP curves are shown in Figure 3C. GELs for the top 10-ranked angiogenesis-associated genes for patients included in each cluster are given in Supplementary Figure 1. 8

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Hierarchical clustering showing the variability of gene expression of hypoxia-response top 10-ranked genes is given in Figure 3B. For this signature, we identified two significant clusters: Cluster H1 (18 patients, 43 %) with lower hypoxia-response GELs and Cluster H2 (24 patients, 57 %) with hypoxia-response higher GELs (Supplementary Figure 2). Dichotomization of the patients into low-risk (Cluster H1) and high-risk (Cluster H2), subgroups based on the gene expression levels of the hypoxia-response signature revealed a marked difference in TTP under BE between the two groups (Figure 3D). We obtained a median TTP for the high-risk patients of 2.9 months (95%CI: 1.8 ─ 4.1) and of 6.9 months (95% CI: 4.0 ─ 9.7) for the low-risk patients.

Hypoxia-response gene signature predictive of tumor shrinkage after BE treatment in non-squamous NSCLC patients. An additional secondary endpoint of interest was tumor shrinkage (TS) measured at 12 weeks after BE treatment, which indicates clinically relevant direct anti-tumor activity. Because of lack of measurements at 12 weeks, 14 patients had to be excluded from this analysis. We analyzed the genes correlating with TS in our patients using GSEA using both the angiogenesis-associated and hypoxia-response gene signatures (Figure 4A and B). Both gene signatures where significantly enriched in the gene set correlating with TS (mean rank = 5104, P = 0.002 and mean rank = 7795, P = 0.038, respectively). The resulting top 12-ranked genes for both signatures are given in Figure 4C and D. Further, we classified the patients based on the metagene score calculated for the top 10-ranked genes for each gene signature. For the angiogenesis-associated signature, the resulting two patients groups (lowrisk and high-risk) showed a non-significantly different median TS, 13.7 % interquartile range (IRQ): -0.8 ─ 26.2 vs. 0 %, IQR: -3.2 ─ 16.4, P = 0.755. In contrast, when dichotomization was 9

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

performed using the 10-gene hypoxia-response signature, the low-risk patients had a significantly higher median TS than the high-risk patients (16.1 %, IRQ: 0 ─ 26.2 vs. -0.4 %, IRQ: -2.5 ─ 2.6, P = 0.013), indicating a higher potential for assessing treatment response using this signature. A heat map showing the gene expression levels of the top 10-ranked hypoxia response genes predictive of tumor shrinkage is given in Supplementary Figure 3.

Angiogenesis-associated and hypoxia-response gene signatures have prognostic value for non-squamous NSCLC patients. Lastly, we analyzed the prognostic value of both angiogenesis and hypoxia gene signatures by investigating the genes correlating with the overall survival (OS). The median follow-up time was 24.9 months (95% CI, 23.9 - ∞). The results of the GSEA for OS are given in Figure 5A and 5B, and the derived top 12-ranked genes are given in Figure 5C and 5D, respectively. Both gene signatures where significantly enriched in the gene set correlating with OS (mean rank = 5064, P = 0.031 and mean rank = 4555, P = 0.001, respectively). Using the top 10-ranked genes for both signatures to dichotomize the patients in low-risk (longer OS) and high-risk (shorter OS) revealed marked differences in OS. Figures 5E and 5F show the KaplanMeier OS curves for both gene signatures used. Using the angiogenesis-associated gene signature led to a median OS for the high-risk patients of 10.6 months [95%CI, 5.2 ─ 19.4] and for the low-risk patients of 14.1 months [95% CI, 10.5 ─ ∞], P = 0.035. The hypoxiaresponse gene signature showed a higher prognostic value (P = 0.001) resulting in a median OS for the high-risk patients of 9.9 months [95% CI 4.8 ─ 13.4] and 17.8 months [95% CI 16.6 ─ ∞] for the low-risk patients. A heat map displaying the gene expression levels of the top

10

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

10-ranked hypoxia response genes significantly associated with OS is given in Supplementary Figure 4. Validation of microarray gene expression levels (GELs) by RT-qPCR. RT-qPCR was used to assess the GELs that were established by microarray analysis. RT-qPCR was performed on 40 out of 42 RNA samples and for nine genes, plus three control genes. The Spearman’s rank correlation coefficients (r) were between 0.51─0.71. The correlation was significant for each gene. We obtained P values for these associations of < 0.0008, demonstrating good agreement between the two complementary methods in quantifying the gene expression levels (Figure 6). This outcome indicates a statistically significant correlation between the microarray GELs and GELs assessed by RT-qPCR, which is a more convenient and less expensive technique for routine application in a clinical setting. These results are in agreement with the expected level of correlation considering the fact that there is no designed sequence overlap between the qPCR primers and the microarray probes.

Discussion There are several gene expression signatures identified as prognostic biomarkers for lung cancer, mostly for lung adenocarcinoma.34 However, gene expression signatures with predictive value for treatment response to antiangiogenic therapy are still lacking. Therefore, the aim of this study was to evaluate the predictive potential of angiogenesisassociated and hypoxia-response gene signatures for the benefit of BE treatment. We performed GSEA for each endpoint, which led to the identification of specific angiogenesis and hypoxia–derived 10-gene signatures. These signatures were then evaluated for their predictive (TTP und TS endpoints) and prognostic value (OS) and tested in an independent 11

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

data set. Our findings might shed light on the mechanism of antiangiogenic treatment response and resistance. Moreover, they may lead to the identification of the causal mechanism behind the high proportion of patients treated with angiogenesis inhibitors showing partial response, as demonstrated by increased progression free survival (PFS), however, with no improvement in their OS. We took a closer look at the angiogenesis-associated genes with predictive value for TTP under BE therapy. The expression of the first ranked gene, adhesion G-protein-coupled receptor 116 (GPR116), was shown to be significantly correlated with tumor progression, recurrence, and poor prognosis in human breast cancer. However, here we identify this gene to play a protective role and its expression to be associated with lower risk of disease recurrence in NSCLC. The biological role of GPR116 in angiogenesis and endothelial proliferation remains to be assessed, however this protein is highly expressed in normal human lung tissue and it was recently demonstrated to regulate lung surfactant homeostasis.35 The second ranked gene, endomucin (EMCN), is an endothelial sialomucin and an endothelial-specific marker, involved in cell-cell and cell-extracellular matrix interactions.36 Integrin α9 (ITGA9), which forms a heterodimeric receptor with activated β1 integrin, has been demonstrated to bind directly to VEGFA, and to contribute to angiogenesis.37 α9 β1 integrin ligand, tenascin-C, enhances secretion of S1P (sphingosine 1phosphate) in endothelial cells. In turn, S1P promotes endothelial cell barrier integrity,38 acting as an anti-permeability agent39 and modulating vessel integrity through its cognate receptor S1PR1. S1PR1 inhibits VEGFR2 signaling and suppresses endothelial hypersprouting via stabilization of junctional VE-cadherin, which leads to enhanced cell-cell adhesion.40 PECAM-1 and JAM2 are adhesive proteins that accumulate in adherens junctions and maintain the restrictiveness of the endothelial barrier. RhoJ, an endothelial-enriched Rho 12

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

GTPase, regulates angiogenesis and vessel integrity.41 GNG11 is a member of the γ subunit family of heteromeric G-protein, which regulates cellular senescence in response to environmental stimuli.42 To obtain the ranking of the discriminating power for each gene, we performed optimized between-group classification (OBC)43 sensitivity analysis for this signature. OBC suggests that for angiogenesis-associated signature S1PR1, GPR116, and PECAM1 have the highest discriminating power for TTP under BE (Supplementary Figure 5a ). The hypoxia-response genes with predictive value for TTP under BE were mostly genes involved in the glycolytic and other metabolic pathways44: PFKP (phosphofructokinase, platelet), LDHA (lactate dehydrogenase A), GPI (glucose-6-phosphate isomerase), ALDOA (aldolase A), ACOT7 (acyl-CoA thioesterase 7), PGK1 (phosphoglycerate kinase), SLC25A32 (solute carrier family 25, mitochondrial folate carrier, member 32) and SLC2A1 (solute carrier family 2, glucose transporter, member 1). The first-ranked hypoxia-response gene DDIT4 (DNA-damage-inducible transcript 4 protein, also known as REDD1) has been shown to be implicated in inhibition of the mTORC1 (mammalian target of rapamycin kinase) signaling pathway, which is relevant in tumor suppression;45 MIF (macrophage migration inhibitory factor) has been revealed to act within the tumor microenvironment to stimulate angiogenesis and promote immune evasion;46 ADM (adrenomedullin) is involved in promoting tumor progression by sustaining proliferation and angiogenesis.47 For this signature, OBC indicated that DDIT4 and MIF have the highest discriminatory power (Supplementary Figure 5b). Importantly, the hypoxia-response signature shows both high predictive and prognostic value and could potentially guide future clinical decisions. Our gene expression analyses suggest that the low-risk patients (most likely to respond to antiangiogenic combined BE therapy and have better outcome) have a tumor

13

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

microenvironment that sustains controlled angiogenesis, developing blood vessels with increased levels of integrity and reduced permeability. The controlled angiogenesis is associated with lower risk of hypoxia within the tumors (lower levels of hypoxia response genes). On the other hand, the high-risk patients (less likely to respond to BE treatment and have worse outcome) have a tumor microenvironment that sustains aberrant angiogenesis with reduced vascular stability and increased vascular permeability. This phenotype is further associated with increased tumor hypoxia and an earlier onset of disease progression. Moreover, we observed that the 10-gene hypoxia-response signature has a higher prognostic power than the angiogenesis-associated signature. This most likely indicates that CT has a relatively high contribution to OS, which is expected taking into account that hypoxic tumors are less responsive to CT than normoxic tumors.48 Additionally, we performed LOOCV from the original dataset in order to test the robustness of the gene signatures associated with TTP under BE and OS endpoints. The perturbations resulting from LOOCV had an insignificant impact on our findings demonstrating the robustness of the discriminatory power of our gene signatures (Supplementary Table 3). Unfortunately, we could not investigate the performance of our gene expression signatures using an identical independent data set, as an additional gene expression data set from pretreatment biopsies of treatment-naïve NSCLC patients receiving the same therapeutic scheme is not available for validation. However, we tested our 10-gene signatures correlating with TTP under BE using a data set comprising GELs of biopsies from a pretreated NSCLC population (BATTLE-1 study,49 trial registration ID: NCT00409968, raw data GEO series accession number: GSE33072). We analyzed the association between our 10-gene expression signatures and PFS for two treatment arms: erlotinib (25 patients, all EGFR-WT) and sorafenib (31 patients, all EGFR-WT; the patients with squamous cell or adenosquamous 14

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

carcinoma were excluded). This analysis revealed that our 10 gene angiogenesis-associated signature had no predictive value for either erlotinib or sorafenib response as second line therapy. Likewise, the 10-gene hypoxia-response signature had no predictive value for erlotinib response. However, for the second line sorafenib-treated patient group (majority erlotinib resistant), we found that the 10-gene hypoxia-response signature discriminated between low-risk (responders) and high-risk patients (non-responders): PFS 3.65 months (95%CI: 1.87 ─ 8.74) and 2.61 months (95%CI: 1.81 ─ 3.61), respec vely (P = 0.0186, Supplementary Figure 6). Sorafenib, a multikinase inhibitor, achieves antiangiogenic effects by blocking VEGFR and PDGFR. In addition, several studies suggest that sorafenib exerts a negative regulatory effect on angiogenesis by suppressing expression of VEGF via inhibition of HIF-1α accumulation and activation.50 Although caution in extrapolating data from one clinical trial to the other is required, our findings suggest that the hypoxia-response signature may be a predictive biomarker of anti-VEGF(R) treatment response (such as bevacizumab and sorafenib) and has no predictive power for erlotinib activity. The lack of predictive value of the 10-gene angiogenesis in the context of second line sorafenib treatment could indicate that the expression of the angiogenesis-associated genes significantly changes from treatment naïve tumors to CT resistant tumors, whereas hypoxiaresponse GELs are affected to a lesser extent. Importantly, only 18 patients in sorafenibtreated group had biopsies originating from the lung; the gene expression signatures of tumor-associated vascular endothelial cells originating from other organs could vary greatly. Nonetheless, our findings suggest that the 10-gene hypoxia-response has a high predictive value of treatment response even in the context of second line antiangiogenic therapy and a great potential for future clinical use.

15

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

The positive correlation between pretreatment angiogenesis-associated and hypoxiaresponse GELs from tumor biopsies and clinical outcomes following BE treatment derived from our analyses supports further evaluation of these candidate gene signatures as potential biomarkers for the selection of the patient subpopulation most likely to obtain benefit from antiangiogenic therapy. There are, however, several limitations that accompany our study. Our study comprises a relatively low number of patients, and control groups (no treatment, and bevacizumab-only and erlotinib-only treatment) are absent. Further validation with larger number of patients and adequate control arms is needed. Nevertheless, we found highly statistically significant differences in the hypoxia-response GELs of responders vs. non-responders to antiangiogenic therapy in both our data set and an additional independent data set. This is very promising and suggests that the identified signatures may be clinically useful for further stratifying non-squamous NSCLC patients and allow for personalized treatment to avoid unnecessary costs and patient exposure to toxicity. Conclusions We identified 10-gene angiogenesis and hypoxia signatures, which can predict the subgroup of patients with higher likelihood of responding to angiogenic therapy. These patients had higher GELs of the genes mainly involved in maintaining the vascular barrier integrity and lower levels of hypoxia-response genes. Moreover, these patients showed improved OS and 75 % of them experienced a high tumor shrinkage level (between 16 - 76 % TS) at 12 weeks after the beginning of treatment. Although there is a very important implication to patient selection for antiangiogenic therapy, the results of this study are preliminary and need to be further validated.

16

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed.

Authors' Contributions Conception and design: A. Franzini, M. H. Brutsche Development of methodology: A. Franzini, M. H. Brutsche Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Franzini, I. I. Macovei, C. Droege, D. Betticher, F. Zappa Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Franzini, F. Baty, I. I. Macovei, O. Dürr, D. Klingbiel, M. H. Brutsche Writing, review, and/or revision of the manuscript: A. Franzini, F. Baty, B. D. Grigoriu, D. Klingbiel, M. H. Brutsche Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Franzini, B. D. Grigoriu, C. Droege, D. Betticher, F. Zappa, M. H. Brutsche Study supervision: A. Franzini, M. H. Brutsche

Supplementary Information Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

Grant Support This work was supported by the Swiss Cancer League & Swiss Cancer Center (KLS-2880-02-2012), KSSG Medical Research Center (MFZF_2014_001) and Romanian–Swiss Research Program (IZERZO_142235/1).

Acknowledgements We thank all the investigators involved in the SAKK 19/05 trial for collecting the samples and SAKK for funding the Affymetrix arrays.

References 1. Horimoto Y, Polanska UM, Takahashi Y, Orimo A. Emerging roles of the tumor-associated stroma in promoting tumor metastasis. Cell Adh Migr 2012;6:193-202. 2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646-74. 3. Brabek J, Mierke CT, Rosel D, Vesely P, Fabry B. The role of the tissue microenvironment in the regulation of cancer cell motility and invasion. Cell Commun Signal 2010;8:22. 4. McMillin DW, Negri JM, Mitsiades CS. The role of tumour-stromal interactions in modifying drug response: challenges and opportunities. Nat Rev Drug Discov 2013;12:217-28. 5. Meads MB, Gatenby RA, Dalton WS. Environment-mediated drug resistance: a major contributor to minimal residual disease. Nat Rev Cancer 2009;9:665-74. 6. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996;86:353-64.

17

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

7. Weis SM, Cheresh DA. Tumor angiogenesis: molecular pathways and therapeutic targets. Nat Med 2011;17:1359-70. 8. Ellis LM, Hicklin DJ. VEGF-targeted therapy: mechanisms of anti-tumour activity. Nat Rev Cancer 2008;8:579-91. 9. Ferrara N, Hillan KJ, Gerber HP, Novotny W. Discovery and development of bevacizumab, an antiVEGF antibody for treating cancer. Nat Rev Drug Discov 2004;3:391-400. 10. Hurwitz H, Fehrenbacher L, Novotny W, Cartwright T, Hainsworth J, Heim W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 2004;350:2335-42. 11. Escudier B, Bellmunt J, Negrier S, Bajetta E, Melichar B, Bracarda S, et al. Phase III trial of bevacizumab plus interferon alfa-2a in patients with metastatic renal cell carcinoma (AVOREN): final analysis of overall survival. J Clin Oncol 2010;28:2144-50. 12. Perren TJ, Swart AM, Pfisterer J, Ledermann JA, Pujade-Lauraine E, Kristensen G, et al. A phase 3 trial of bevacizumab in ovarian cancer. N Engl J Med 2011;365:2484-96. 13. Kreisl TN, Kim L, Moore K, Duic P, Royce C, Stroud I, et al. Phase II trial of single-agent bevacizumab followed by bevacizumab plus irinotecan at tumor progression in recurrent glioblastoma. J Clin Oncol 2009;27:740-5. 14. Bergers G, Hanahan D. Modes of resistance to anti-angiogenic therapy. Nat Rev Cancer 2008;8:592-603. 15. Dang DT, Chun SY, Burkitt K, Abe M, Chen S, Havre P, et al. Hypoxia-inducible factor-1 target genes as indicators of tumor vessel response to vascular endothelial growth factor inhibition. Cancer Res 2008;68:1872-80. 16. Rapisarda A, Melillo G. Overcoming disappointing results with antiangiogenic therapy by targeting hypoxia. Nat Rev Clin Oncol 2012;9:378-90. 17. Ebos JM, Kerbel RS. Antiangiogenic therapy: impact on invasion, disease progression, and metastasis. Nat Rev Clin Oncol 2011;8:210-21. 18. Miles DW, Chan A, Dirix LY, Cortes J, Pivot X, Tomczak P, et al. Phase III study of bevacizumab plus docetaxel compared with placebo plus docetaxel for the first-line treatment of human epidermal growth factor receptor 2-negative metastatic breast cancer. J Clin Oncol 2010;28:3239-47. 19. Robert NJ, Dieras V, Glaspy J, Brufsky AM, Bondarenko I, Lipatov ON, et al. RIBBON-1: randomized, double-blind, placebo-controlled, phase III trial of chemotherapy with or without bevacizumab for first-line treatment of human epidermal growth factor receptor 2-negative, locally recurrent or metastatic breast cancer. J Clin Oncol 2011;29:1252-60. 20. Gyanchandani R, Kim S. Predictive biomarkers to anti-VEGF therapy: progress toward an elusive goal. Clin Cancer Res 2013;19:755-7. 21. Baty F, Rothschild S, Fruh M, Betticher D, Droge C, Cathomas R, et al. EGFR exon-level biomarkers of the response to bevacizumab/erlotinib in non-small cell lung cancer. PLoS One 2013;8:e72966. 22. Baty F, Facompre M, Kaiser S, Schumacher M, Pless M, Bubendorf L, et al. Gene profiling of clinical routine biopsies and prediction of survival in non-small cell lung cancer. Am J Respir Crit Care Med 2010;181:181-8. 23. Edlund K, Lindskog C, Saito A, Berglund A, Ponten F, Goransson-Kultima H, et al. CD99 is a novel prognostic stromal marker in non-small cell lung cancer. Int J Cancer 2012;131:2264-73. 24. Larsen AK, Ouaret D, El Ouadrani K, Petitprez A. Targeting EGFR and VEGF(R) pathway cross-talk in tumor survival and angiogenesis. Pharmacol Ther 2011;131:80-90. 25. Maity A, Pore N, Lee J, Solomon D, O'Rourke DM. Epidermal growth factor receptor transcriptionally up-regulates vascular endothelial growth factor expression in human glioblastoma cells via a pathway involving phosphatidylinositol 3'-kinase and distinct from that induced by hypoxia. Cancer Res 2000;60:5879-86. 26. Fong GH. Mechanisms of adaptive angiogenesis to tissue hypoxia. Angiogenesis 2008;11:121-40. 27. Zappa F, Droege C, Betticher D, von Moos R, Bubendorf L, Ochsenbein A, et al. Bevacizumab and erlotinib (BE) first-line therapy in advanced non-squamous non-small-cell lung cancer (NSCLC) (stage 18

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

IIIB/IV) followed by platinum-based chemotherapy (CT) at disease progression: a multicenter phase II trial (SAKK 19/05). Lung Cancer 2012;78:239-44. 28. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003;4:249-64. 29. Bair E, Tibshirani R. Semi-supervised methods to predict patient survival from gene expression data. PLoS Biol 2004;2:E108. 30. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 2005;102:15545-50. 31. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001;25:402-8. 32. Masiero M, Simoes FC, Han HD, Snell C, Peterkin T, Bridges E, et al. A core human primary tumor angiogenesis signature identifies the endothelial orphan receptor ELTD1 as a key regulator of angiogenesis. Cancer Cell 2013;24:229-41. 33. Buffa FM, Harris AL, West CM, Miller CJ. Large meta-analysis of multiple cancers reveals a common, compact and highly prognostic hypoxia metagene. Br J Cancer 2010;102:428-35. 34. Zhu C-Q, Tsao M-S. Prognostic markers in lung cancer: is it ready for prime time? Transl Lung Cancer Res 2014;3:149-58. 35. Yang MY, Hilton MB, Seaman S, Haines DC, Nagashima K, Burks CM, et al. Essential regulation of lung surfactant homeostasis by the orphan G protein-coupled receptor GPR116. Cell Rep 2013;3:1457-64. 36. Ueno M, Igarashi K, Kimura N, Okita K, Takizawa M, Nobuhisa I, et al. Endomucin is expressed in embryonic dorsal aorta and is able to inhibit cell adhesion. Biochem Biophys Res Commun 2001;287:501-6. 37. Vlahakis NE, Young BA, Atakilit A, Hawkridge AE, Issaka RB, Boudreau N, et al. Integrin alpha9beta1 directly binds to vascular endothelial growth factor (VEGF)-A and contributes to VEGF-Ainduced angiogenesis. J Biol Chem 2007;282:15187-96. 38. Garcia JG, Liu F, Verin AD, Birukova A, Dechert MA, Gerthoffer WT, et al. Sphingosine 1phosphate promotes endothelial cell barrier integrity by Edg-dependent cytoskeletal rearrangement. J Clin Invest 2001;108:689-701. 39. Argraves KM, Gazzolo PJ, Groh EM, Wilkerson BA, Matsuura BS, Twal WO, et al. High density lipoprotein-associated sphingosine 1-phosphate promotes endothelial barrier function. J Biol Chem 2008;283:25074-81. 40. Gaengel K, Niaudet C, Hagikura K, Lavina B, Muhl L, Hofmann JJ, et al. The sphingosine-1phosphate receptor S1PR1 restricts sprouting angiogenesis by regulating the interplay between VEcadherin and VEGFR2. Dev Cell 2012;23:587-99. 41. Kim C, Yang H, Fukushima Y, Saw PE, Lee J, Park JS, et al. Vascular RhoJ is an effective and selective target for tumor angiogenesis and vascular disruption. Cancer Cell 2014;25:102-17. 42. Hossain MN, Sakemura R, Fujii M, Ayusawa D. G-protein gamma subunit GNG11 strongly regulates cellular senescence. Biochem Biophys Res Commun 2006;351:645-50. 43. Baty F, Bihl MP, Perriere G, Culhane AC, Brutsche MH. Optimized between-group classification: a new jackknife-based gene selection procedure for genome-wide expression data. BMC Bioinformatics 2005;6:239. 44. Doherty JR, Cleveland JL. Targeting lactate metabolism for cancer therapeutics. J Clin Invest 2013;123:3685-92. 45. Ben Sahra I, Regazzetti C, Robert G, Laurent K, Le Marchand-Brustel Y, Auberger P, et al. Metformin, independent of AMPK, induces mTOR inhibition and cell-cycle arrest through REDD1. Cancer Res 2011;71:4366-72. 46. Girard E, Strathdee C, Trueblood E, Queva C. Macrophage migration inhibitory factor produced by the tumour stroma but not by tumour cells regulates angiogenesis in the B16-F10 melanoma model. Br J Cancer 2012;107:1498-505. 19

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

47. Nikitenko LL, Fox SB, Kehoe S, Rees MC, Bicknell R. Adrenomedullin and tumour angiogenesis. Br J Cancer 2006;94:1-7. 48. Harrison L, Blackwell K. Hypoxia and anemia: factors in decreased sensitivity to radiation therapy and chemotherapy? Oncologist 2004;9 Suppl 5:31-40. 49. Kim ES, Herbst RS, Wistuba, II, Lee JJ, Blumenschein GR, Jr., Tsao A, et al. The BATTLE trial: personalizing therapy for lung cancer. Cancer Discov 2011;1:44-53. 50. Liu LP, Ho RL, Chen GG, Lai PB. Sorafenib inhibits hypoxia-inducible factor-1alpha synthesis: implications for antiangiogenic activity in hepatocellular carcinoma. Clin Cancer Res 2012;18:566271.

Figure legends: Figure 1: Treatment course for patients with no prior therapy included in the SAKK 19/05 phase II trial, study endpoints and microenvironment compartments analyzed by GSEA; TTP ─ me to progression, TS ─ tumor shrinkage, OS ─ overall survival. Figure 2: Enrichment plot from GSEA shows statistically significant enrichment of the angiogenesis-associated genes (P=0.004) (A) and in the hypoxia-response genes (P=0.001) (B) in the gene set predictive of TTP under BE. Top 12-ranked angiogenesis-associated genes (C) and hypoxia-response genes (D) derived from GSEA. Figure 3: Hierarchical clustering of top 10-ranked angiogenesis-associated genes (A) and hypoxia-response genes (B) significantly associated with TTP under BE showing gene expression variation among patients. The heat maps represent relative intensity values of gene expression levels. Kaplan–Meier TTP under BE curves of the three patients groups (lowrisk, medium-risk and high-risk) defined by the 10-gene angiogenesis-associated signature, P = 0.013 for Cluster A1 vs. Cluster A2, (C) and two patient groups (low-risk and high-risk) defined by the 10-gene hypoxia-response signature, P = 0.016 (D). The P values of these associations were determined by log-rank test. Figure 4: Angiogenesis and hypoxia-associated gene signatures associate with tumor shrinkage in NSCLC. Enrichment plot from GSEA shows statistically significant enrichment of the angiogenesis-associated genes (P=0.002) (A) and in the hypoxia-response genes (P=0.038) (B) in the gene signature correlating with TS. Top 12-ranked angiogenesisassociated genes (C) and hypoxia-response (D) derived from GSEA, which correlate with TS. Correlation between the two patients groups defined by the 10-gene angiogenesisassociated signature (E) and hypoxia-response signature (F) (high-risk and low-risk; n = 28) and tumor shrinkage at 12 weeks after BE treatment (box plots). The boxes represent the median ± interquartile range (IQR). Whiskers delimit the highest and lowest non-outlier data points (defined as greater/less than 1.5 × IQR). Figure 5: Angiogenesis and hypoxia-associated gene signatures predicting OS in NSCLC. Enrichment plot from GSEA shows statistically significant enrichment of the angiogenesisassociated genes (P = 0.031), (A) and hypoxia-response genes (P = 0.001) (B) in the gene signature predictive of OS. Top 12-ranked angiogenesis-associated genes (C) and top 1220

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. Gene signatures predictive of response to therapy in NSCLC

ranked hypoxia-response genes (D) derived from GSEA, which significantly correlate with OS. Kaplan–Meier OS curves of the two patients groups (low-risk and high-risk) defined by the 10-gene angiogenesis-associated signature (P = 0.035) (E) and 10-gene hypoxia signature (P = 0.001) (F). The P values of the associations were determined by log-rank test. Figure 6: Comparison between microarray and RT-qPCR GELs. Spearman's correlations were performed between relative gene expression levels (GELs) determined by RT-qPCR (x-axis) and RNA microarray (y-axis) for nine angiogenesis-associated genes for 40 out of the 42 patients. The mRNA levels for each gene of interest were determined by RT qPCR and correlated with microarray expression scores determined after data processing, both calculated relative to HPRT1 gene. r, Spearman’s rank correlation coefficient.

21

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

Figure 1

Biopsies from patients with stage IIIB or IV non-squamous NSCLC with no prior therapy

Bevacizumab/Erlotinib (BE) therapy n × 3 weeks until progression

Standard chemotherapy (CT) 6 × 6 weeks or until progression

12 weeks Endpoints: Compartments:

TS

TTP under BE Angiogenesis

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

OS Hypoxia

B 0.5 0.3 0.1 4 3 2 0

1

Ranked list metric

4 3 2 1 0

Ranked list metric

Hypoxia

5

0.1

0.3

0.5

Enrichment score (ES)

Angiogenesis

5

Enrichment score (ES)

A

Figure 2

0

5000

10000

15000

0

Rank in Ordered Dataset

C Gene symbol GPR116 EMCN ITGA9 LDB2 MEF2C GNG11 CALCRL KDR PECAM1 S1PR1 JAM2 RHOJ

5000

10000

15000

Rank in Ordered Dataset

D Gene symbol Rank in gene list Rank metric score Enrichment score (ES) DDIT4 403 1.924 0.027 MIF 611 1.700 0.060 PFKP 648 1.667 0.103 ADM 707 1.629 0.143 805 1.561 0.179 LDHA GPI 829 1.546 0.219 865 1.522 0.257 ALDOA ACOT7 987 1.455 0.289 SLC25A32 1019 1.435 0.326 PGK1 1117 1.391 0.357 TUBA1B 1288 1.313 0.382 Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer SLC2A1 1364 1.289 Research. 0.412

Rank in gene list 18 33 53 69 132 137 238 373 976 1110 1238 1420

Rank metric score 3.539 3.276 2.991 2.835 2.513 2.494 2.211 1.975 1.460 1.394 1.331 1.268

Enrichment score (ES) 0.082 0.157 0.226 0.291 0.346 0.404 0.450 0.488 0.487 0.512 0.535 0.554

A

Figure 3

B

Angiogenesis

Hypoxia Color Key

Color Key

−2

Row

−2

0

2 Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Z−Score Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Cluster A1

Cluster A2

0

2

4

Row Z−Score

Cluster H1

Cluster A3

Cluster H2 GPI

KDR

MIF

LDB2

ALDOA

S1PR1

LDHA

MEF2C

PGK1

ITGA9

ADM

PECAM1

DDIT4

GPR116

ACOT7

GNG11

PFKP

EMCN

SLC25A32

002 078 102 055 103 056 094 060 070 091 090 058 077 051 080 065 038 099 057 063 088 064 082 083 097 081 096 098 074 095 049 093 069 068 067 076 023 101 061 084 075 087

103 102 094 093 002 078 058 083 082 077 023 060 074 056 099 055 084 098 080 091 038 096 057 063 097 061 051 049 075 095 081 067 101 070 076 087 068 065 090 088 064 069

CALCRL

D Cluster A1

90

Cluster A2

Proportion of patients without progresion

100

Cluster A3

80 70 60 50 40 30 20 10 0

0

2

4

6 8 Time (months)

10

12

14

Patients at risk: Cluster A1 9 9 7 5 4 2 1 0 Cluster A2 12 6 clincancerres.aacrjournals.org 3 2 1 18, 2015. © 0 2015 American 0 Association 0 for Cancer Downloaded from on May Research. Cluster A3 21 17 12 6 4 2 1 0

Proportion of patients without progresion

C

100

Cluster H1

90

Cluster H2

80 70 60 50 40 30 20 10 0

0 Patients at risk: Cluster H2 18 Cluster H1 24

2

4

6 8 Time (months)

17 15

13 9

10 3

7 2

10

12

14

2 2

1 1

0 0

Figure 4

B

Hypoxia 0.1 0.2 0.3 0.4 0.5

0.3

Enrichment score (ES)

0.5

Angiogenesis

0.1

Enrichment score (ES)

A

4 0

1

2

3

Ranked list metric

3 2 1 0

Ranked list metric

4

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

0

5000

10000

0

15000

5000

10000

15000

Rank in Ordered Dataset

Rank in Ordered Dataset

C

D

Gene symbol LDB2 VWF MYCT1 TEK EMCN CDH5 TIE1 JAM2 ZNF423 ROBO4 ADCY4 PTPRB

Rank in gene list 622 664 669 700 734 749 788 1072 1139 1197 1398 1690

Rank metric score 1.818 1.785 1.784 1.763 1.745 1.735 1.716 1.559 1.533 1.506 1.425 1.328

Enrichment score (ES) 0.010 0.053 0.098 0.141 0.183 0.227 0.268 0.291 0.326 0.361 0.386 0.402

E

Gene symbol MRPL15 PSMA7 ANKRD37 MIF LDHA GPI TUBA1B YKT6 PFKP ACOT7 ALDOA PGK1

Rank in gene list 61 82 170 228 473 605 1212 1522 2118 3154 3247 3600

Rank metric score 2.960 2.836 2.465 2.285 1.951 1.829 1.501 1.382 1.206 0.991 0.975 0.922

Enrichment score (ES) 0.090 0.179 0.252 0.321 0.368 0.418 0.430 0.456 0.459 0.429 0.455 0.463

80

80

60

60

TS(%)

TS(%)

F

40

40

20

20

0

0

−20

18

10

FALSE.

TRUE.

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Patients at risk (metagene score) Research.

−20

18

10

FALSE.

TRUE.

Patients at risk (metagene score)

A

Figure 5

B 0.6 0.4 0.2 0.0 3 2 1 0

Ranked list metric

4 3 2 1 0

Ranked list metric

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

4

5

0.1

0.3

Enrichment score (ES)

0.5

Hypoxia

5

Enrichment score (ES)

Angiogenesis

0

5000

10000

0

15000

5000

10000

15000

Rank in Ordered Dataset

Rank in Ordered Dataset

C

D

Gene symbol GPR116 MYCT1 ITGA9 RHOJ KDR CDH5 GNG11 CALCRL LDB2 ESAM JAM2 ROBO4

Rank in gene list 504 811 1228 1369 2137 2260 2354 2379 2485 2601 3091 3220

Rank metric score 1.798 1.563 1.346 1.297 1.059 1.031 1.009 1.004 0.987 0.962 0.879 0.859

Enrichment score (ES) 0.030 0.064 0.084 0.118 0.108 0.135 0.163 0.195 0.222 0.247 0.247 0.268

Rank in gene list 100 265 290 333 387 388 442 466 757 931 977 1307

Rank metric score 2.699 2.177 2.124 2.059 1.957 1.955 1.880 1.844 1.598 1.489 1.460 1.320

Enrichment score (ES) 0.061 0.105 0.156 0.204 0.249 0.297 0.341 0.385 0.407 0.434 0.467 0.480

F 100

Low risk

100

Low risk

90

High risk

90

High risk

80

Cumulative survival (%)

Cumulative survival (%)

E

Gene symbol MIF ALDOA MRPL15 LDHA ADM GPI DDIT4 ACOT7 SHCBP1 PFKP PSMA7 NDRG1

70 60 50 40 30

80 70 60 50 40 30

20

20

10

10 0

0

0 Patients at risk:

Low risk High risk

6

12

18 Time (months)

24

0

30

21 18from clincancerres.aacrjournals.org 13 8 4 18, 2015.3© 2015 American 2 Downloaded on May Association for Cancer 21 17 12 6 4 2 1 Research.

6

12

Patients at risk:

Low risk High risk

21 21

19 14

16 10

18 Time (months) 12 3

5 2

24

30 2 2

1 2

1.0

0.0 0.0

r = 0.6912 P < 0.0001 0.5

1.0

1.5

2.0

1.0

0.5

0.0 0.0

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

GNG11

2.0

1.5

1.0

0.5

r = 0.6437 P < 0.0001

0.0

0

5

10

GEL, microarray/HPRT1

GEL, microarray/HPRT1

GELs, RT-qPCR/HPRT1

15

1.0

0.5

r = 0.5102 P = 0.0008 1

2

3

GELs, RT-qPCR/HPRT1

4

GEL, microarray/HPRT1

GEL, microarray/HPRT1

1.5

0

0.4

0.6

0.8

ITGA9

2.0 1.8 1.6 1.4 1.2 1.0

GELs, RT-qPCR/HPRT1

2.0

1.5

1.0

r = 0.7131 P < 0.0001

0.0 0.0

0.1

0.2

1.5

1.0

0.0 0.0

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

r = 0.7030 P < 0.0001 0.5

1.0

1.5

GELs, RT-qPCR/HPRT1

1.5

2.0

1.5

1.0

0.5

r = 0.6534 P < 0.0001

0.0

0

2

4

6

8

10

GELs, RT-qPCR/HPRT1

2.0

0.5

1.0

PECAM1

2.5

0.3

JAM2

2.5

0.5

GELs, RT-qPCR/HPRT1

KDR

0.5

r = 0.6283 P < 0.0001

0.8 0.0

GELs, RT-qPCR/HPRT1

S1PR1

0.0

0.2

2.5

GELs, RT-qPCR/HPRT1 2.0

r = 0.7032 P < 0.0001

2.0

GEL, microarray/HPRT1

0.5

1.5

GEL, microarray/HPRT1

1.5

EMCN

2.0

GEL, microarray/HPRT1

2.0

GEL, microarray/HPRT1

GEL, microarray/HPRT1

GPR116

2.5

Figure 6

RHOJ

2.5

2.0

1.5

1.0

0.5

0.0 0.00

r = 0.6588 P < 0.0001 0.02

0.04

0.06

0.08

GELs, RT-qPCR/HPRT1

0.10

Author Manuscript Published OnlineFirst on April 28, 2015; DOI: 10.1158/1078-0432.CCR-14-3135 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Gene expression signatures predictive of bevacizumab/erlotinib therapeutic benefit in advanced non-squamous non-small cell lung cancer patients (SAKK 19/05 trial) Anca Franzini, Florent Baty, Ina I Macovei, et al. Clin Cancer Res Published OnlineFirst April 28, 2015.

Updated version Supplementary Material Author Manuscript

E-mail alerts Reprints and Subscriptions Permissions

Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-14-3135 Access the most recent supplemental material at: http://clincancerres.aacrjournals.org/content/suppl/2015/04/29/1078-0432.CCR-14-3135.DC1.html Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected] To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]

Downloaded from clincancerres.aacrjournals.org on May 18, 2015. © 2015 American Association for Cancer Research.

05 trial).

We aimed to identify gene expression signatures associated with angiogenesis and hypoxia pathways with predictive value for treatment response to beva...
1MB Sizes 0 Downloads 9 Views