- Email: [email protected]
Differential gene expression profiling of matched primary renal cell carcinoma and metastases reveals upregulation of extracellular matrix genes
Annals of Oncology Advance Access published December 19, 2016
Differential gene expression profiling of matched primary renal cell carcinoma and metastases reveals upregulation of extracellular matrix genes T. H. Ho1, D. J. Serie2, M. Parasramka3, J. C. Cheville4, B. M. Bot5, W. Tan6, L. Wang7, R. W. Joseph6, T. Hilton2, B. C. Leibovich8, A. S. Parker2, J. E. Eckel-Passow7# 1
Division of Hematology and Medical Oncology, Mayo Clinic, Scottsdale, USA Department of Health Sciences Research, Mayo Clinic, Jacksonville, USA 3 Department of Cancer Biology, Mayo Clinic, Jacksonville, USA 4 Laboratory Medicine and Pathology, Mayo Clinic, Rochester, USA 5 Computational Oncology, Sage Bionetworks, Seattle, USA 6 Division of Hematology/Oncology, Mayo Clinic, Jacksonville, USA 7 Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, USA 8 Department of Urology, Mayo Clinic, Rochester, USA 2
Corresponding Author: Prof. Jeanette E. Eckel-Passow Division of Biomedical Statistics and Informatics Mayo Clinic 200 First Street SW Rochester, MN 55905 USA Phone: 001-507-538-6512 E-mail: [email protected]
Word count of abstract: 273 Word count of text: 3007 Number of figures: 1 Number of tables: 4 Key message: We validated gene expression profiles using a large set of patient-matched primary and metastatic renal cell carcinoma tumors and identified up regulation of extracellular matrix genes in metastases. Our study implicates up regulation of extracellular matrix genes as a critical molecular event leading to visceral, bone and soft tissue metastases in renal cell carcinoma. Supplementary Data are available online.
© The Author 2016. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
#
ABSTRACT Background: The majority of renal cell carcinoma (RCC) studies analyze primary tumors, and the corresponding results are extrapolated to metastatic RCC tumors. However, it is unknown if gene expression profiles from primary RCC tumors differs from patient-matched metastatic tumors. Thus, we sought to identify differentially expressed genes between patient-matched primary and metastatic RCC tumors in order to understand the molecular mechanisms
Patients and Methods: We compared gene expression profiles between patient-matched primary and metastatic RCC tumors using a two-stage design. First, we used Affymetrix microarrays on 15 pairs of primary RCC (14 clear cell RCC, 1 papillary) tumors and patientmatched pulmonary metastases. Second, we used a custom Nanostring panel to validate seven candidate genes in an independent cohort of 114 clear cell RCC (ccRCC) patients. Differential gene expression was evaluated using a mixed effect linear model; a random effect denoting patient was included to account for the paired data. Third, The Cancer Genome Atlas (TCGA) data were used to evaluate associations with metastasis-free and overall survival in primary ccRCC tumors. Results: We identified and validated up regulation of seven genes functionally involved in the formation of the extracellular matrix (ECM): DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. In primary ccRCC, CEACAM6 and LUM were significantly associated with metastasis-free and overall survival (P<0.01). Conclusions: We evaluated gene expression profiles using the largest set to date, to our knowledge, of patient-matched primary and metastatic ccRCC tumors and identified up regulation of ECM genes in metastases. Our study implicates up regulation of ECM genes as a critical molecular event leading to visceral, bone and soft tissue metastases in ccRCC.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
underlying the development of RCC metastases.
Keywords: Renal cell carcinoma, extracellular matrix, metastases
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
INTRODUCTION Complete surgical excision, in the form of either partial or radical nephrectomy, for localized renal cell carcinoma (RCC) offers an opportunity for cure in RCC patients. Even with strict adherence to the 2014 National Comprehensive Cancer Network or American Urological Association surveillance guidelines, one third of RCC recurrences will be missed [1]. Metastases of RCC tumors have been reported to virtually all organs; however, the most
variable and impact prognosis and response to therapy.
Although treatment options for metastatic RCC have increased over the past decade, mortality rates and five-year survival are unacceptably poor [3,4]. Thus, there is a clear clinical need to develop novel management and treatment strategies for metastatic RCC that will lower mortality rates and extend survival [5]. Due to tissue availability, genome-wide interrogation of biomarkers associated with metastatic progression and cancer-specific death has primarily been based on observations made in the primary tumor [6-8] and not in the more lethal, and more therapeutically-relevant metastatic lesion. And, genome-wide studies that have preliminarily interrogated metastatic specimens have done so using small sample sizes [9,10]. Thus, the molecular mechanisms that lead to RCC metastases remain largely unknown and require further study.
To address these gaps in knowledge, we identified genes that support RCC metastases by comparing gene expression profiles between metastatic tumors and their patient-matched primary tumor. We primarily focused on clear cell RCC (ccRCC), which accounts for more than 85% of RCC. Particularly, we employed a two-stage design where we first utilized gene
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
common sites are pulmonary and bone [2]. The location and number of metastases are highly
expression microarray technology to identify candidate genes that are associated with RCC metastases, and subsequently, validated the candidate genes in a large cohort of patientmatched primary and metastatic ccRCC tumors using a custom Nanostring assay.
METHODS
Discovery Cohort We identified 15 patients at Mayo Clinic Rochester who were treated surgically for RCC (14 ccRCC, 1 papillary), underwent metastasectomy for a pulmonary metastatic lesion, and had fresh-frozen tissue available from their primary tumor in addition to their pulmonary metastasis. One of the 15 patients had two separate pulmonary metastases; both were included in the study. A single pathologist (J.C.C.) performed a blinded comprehensive review of all tumors (primary and metastatic) to confirm histological subtype (1997 AJCC/UICC classification), tumor stage, 2012 International Society of Urological Pathology (ISUP) tumor grade, tumor size, and coagulative tumor necrosis.
Validation Cohort We identified 114 additional patients at Mayo Clinic Rochester who were treated surgically for ccRCC between 1990 and 2005, had synchronous or metachronous metastases, and underwent metastasectomy for at least one of their metastatic lesions and therefore had formalin-fixed, paraffin-embedded tissue available from their primary tumor and at least one patient-matched metastatic tumor. Unilateral multifocal disease and bilateral disease were not
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Patient Selection
considered metastatic ccRCC. A single pathologist (J.C.C.) performed a blinded comprehensive review of all tumors (primary and metastatic) to confirm histological subtype (1997 AJCC/UICC classification), tumor stage, 2012 International Society of Urological Pathology (ISUP) tumor grade, tumor size, and coagulative tumor necrosis. A representative paraffin-embedded tissue block with the highest grade and presence of necrosis was chosen from each resected tumor. If serial metastatic tumors were available, then all available metastatic tumors were analyzed in order to evaluate inter-tumor heterogeneity across serial metastatic tumors. Metastases were
nephrectomy (< 30 days) or metachronous if the diagnosis and resection of metastases occurred after (>30 days) nephrectomy. This study was approved by the Mayo Clinic IRB.
Affymetrix Gene Expression Microarrays (Discovery Stage) Laser capture microdissection of tumor cells was performed and samples were hybridized to the Affymetrix U133 Plus2 microarray chip, a genome-wide gene expression assay. Microarray analysis was conducted according to manufacturer’s instructions for the Affymetrix One Cycle Target Labeling and Control Reagents kit (Santa Clara, CA). Briefly, cDNA was generated from 5 µg of total RNA using SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA) and T7 Oligo(dT) primer. Subsequently, the products were column-purified (Affymetrix) and then in vitro transcribed to generate biotin-labeled cRNA. The transcribed products were then columnpurified, fragmented, and hybridized onto Affymetrix U133 Plus 2.0 GeneChips® at 45º C for 16 h. Subsequent to hybridization, the arrays were washed and stained with streptavidinphycoerythrin, then scanned in an Affymetrix GeneChip® Scanner 3000 (Santa Clara, CA). The data have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE85258.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
classified as synchronous if the metastatic tumor was resected at the same time as
Nanostring (Validation Stage) Total RNA was isolated from three slides per tumor at 10-µm sections of tumor-rich areas of FFPE tissue blocks using the AllPrep DNA/RNA FFPE kit reagents (Qiagen) following vendor's standard protocols. Isolated FFPE RNAwas treated with 20 units DNase I. The NanoString platform was used to quantify gene expression of seven candidate genes: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. 200 ng of each total RNA sample was prepared as per the manufacturer’s instructions. Gene expression was quantified on the NanoString
extracted using the nSolver software (Nanostring).
Statistical Methods Affymetrix Gene Expression Microarray Data (Discovery Stage) Data were normalized using fastlo [11]. The primary objective was to identify genes that were differentially expressed between patient-matched primary and metastatic tumors. Differential expression was tested using a linear mixed-effect model; models were fit per probe set using a random effect for patient to account for the paired data. Probe sets with p-value (P) <0.00002 and a fold change > 2 were considered candidate differentially expressed genes. We were particularly interested in genes that were up regulated in metastatic tumors as these genes have the largest potential to be therapeutically targetable.
Nanostring Data (Validation Stage) Bland-Altman plots were used to visualize agreement of gene expression between primarymetastatic tumor pairs as well as between serial metastatic tumors from the same individual. Linear mixed-effect models were used to determine if each of the seven genes was differentially expressed between the patient-matched primary and metastatic tumors; models were fit per gene using a random effect for patient to account for the paired data. Linear mixed-effects
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
nCounterTM and raw counts were generated with nSolverTM. Raw counts per gene were
models were also used to determine if the difference in gene expression between primarymetastatic tumor pairs was associated with metastatic site, metastatic timing (M0/M1), metastatic tumor grade, metastatic tumor necrosis (yes/no), and metastatic tumor sarcomatoid status (yes/no). To account for multiple testing, P < 0.01 was used for statistical testing.
The Cancer Genome Atlas (TCGA) ccRCC Data
Data Portal for 428 ccRCC subjects. Cox proportional hazards regression was used to evaluate the association of each gene with metastasis-free survival and overall survival, adjusting for age at diagnosis and tumor stage. Gene expression was modeled as a continuous variable and a linear relationship was verified. Multivariable Cox regression was used to evaluate independent gene predictors of outcome. To account for testing associations of seven genes, P < 0.01 was used.
RESULTS Discovery Stage: Differential Gene Expression Profiling of Patient-Matched Primary and Metastatic RCC Tumors The discovery cohort consisted of 15 patients that developed pulmonary metastatic tumors: 6 had a synchronous (M1) metastasis at the time of diagnosis and 9 had a metachronous (M0) metastasis (Table 1). The majority of patients were male (87%) with Fuhrman grade 3 primary RCC tumors (60%). Affymetrix gene expression profiling identified 38 probesets that were more than two-fold upregulated in metastatic tumors relative to patient-matched primary RCC tumors (P<2x10-5; Supplementary Table S1). Many of these probesets mapped to genes that were associated with lung surfactants and thus were excluded from further consideration, as were
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Level 3 RSEM normalized RNASeqV2 files were downloaded on January 8, 2013, from TCGA
probesets that mapped to open reading frames. The remaining probesets were observed to be enriched for extracellular matrix (ECM) genes. Thus, we chose to validate differential expression of the seven ECM genes that had probesets that met our significance criteria: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3 (Table 2).
Notably, one of the 15 patients in the discovery cohort had a papillary primary RCC tumor whereas the remaining 14 patients had a clear cell primary RCC tumor. Thus, as a sensitivity
seven candidate ECM genes remain highly significant.
Validation Stage: Independent Validation of Differentially Expressed ECM Genes Of the 114 patients identified, 97 (85%) had a primary tumor or a metastatic tumor that was successfully profiled by Nanostring (Figure 1; Table 1); 88 patients had a primary tumor (43 synchronous and 45 metachronous) successfully profiled and 87 patients had a metastatic tumor that was successfully profiled. A total of 118 metastases were available from the 87 patients who had at least one metastatic tumor successfully profiled (Table 3); 59 patients had Nanostring data for only a single distant metastasis and 28 patients had Nanostring data on two or more metastases. The median time from nephrectomy to the first metachronous metastasis was 1.85 years (37 days, maximum 10.82 years). Only one patient received systemic treatment between the initial nephrectomy and metastectomy. Metastases to pulmonary (37%) were the most common (Table 3). Comparing patient-matched primary and metastatic tumor pathologic characteristics, 59% had concordant grade, 60% had concordant necrosis status and 87% had concordant sarcomatoid dedifferentiation status (Supplementary Table S3).
In the validation cohort, DCN, SLIT2, LUM, LAMA2, ADAMTS12, and LMO3 were all significantly upregulated in metastatic tumors relative to patient-matched primary ccRCC tumors
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
analysis, we analyzed only the 14 ccRCC patients. As shown in Supplementary Table S2, the
(P<0.005; Table 2, Supplementary Figure S1). Supplementary Table S4 shows the results for testing whether differential expression was associated with metastatic site. There was a significant association between differential expression of LMO3 (P=0.0001) and CEACAM6 (P=0.0001) with metastatic site. Specifically, pulmonary metastases had larger differences of LMO3 gene expression between primary and metastatic tumors than metastases to bone, brain, liver, skin and nodes (P<0.01). Similarly, pulmonary metastases had larger differences of CEACAM6 gene expression than metastases to most other sites (P<0.01). In fact, CEACAM6
(M1) metastases had larger differences of DCN gene expression between primary and patientmatched metastatic ccRCC tumors in comparison to metachronous (M0) metastases (P<0.01; Supplementary Table S4). Non-necrotic metastases had larger differences of SLIT2 gene expression between primary and patient-matched metastatic ccRCC tumors in comparison to necrotic metastases (P<0.01; Supplementary Table S4). We did not observe a significant association between metastatic tumor sarcomatoid differentiation and difference in gene expression in any of the seven genes (P>0.01; Supplementary Table S4).
Association of ECM Genes with Survival in TCGA Primary ccRCC Tumors After adjusting for age at diagnosis, CEACAM6 and LUM were significantly associated with metastasis-free and overall survival (P<0.01; Table 4). When including both CEACAM and LUM in a multivariable Cox model (Supplemental Table S5), only LUM remained significantly (P<0.05) associated with overall survival whereas both LUM and CEACAM6 were significantly associated with metastasis-free survival (P<0.05). After adjustment for both age at diagnosis and tumor stage (Supplemental Table S6); none of the seven genes were significantly associated with either overall survival or metastasis-free survival (P>0.01).
DISCUSSION
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
was primarily expressed in pulmonary metastases (Supplementary Figure S1). Synchronous
Although treatment options for metastatic RCC have increased over the past decade, mortality and five-year survival remain unacceptably poor [3,4]. TCGA identified somatic mutations, cytogenetic and epigenetic alterations in primary ccRCC tumors [8]; however, less is known regarding common molecular alterations in the more lethal and therapeutically-relevant distant metastatic tumor. In fact, it has been suggested in other cancers that the primary tumor may not be representative of the metastatic tumor and more effective molecular targets could be discovered by interrogating the metastatic tumor [12,13]. Thus, studies that interrogate
cause of patient mortality.
We evaluated gene expression profiles using the largest set, to our knowledge, of patientmatched primary and metastatic ccRCC tumors and identified up regulation of seven ECM genes: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. We utilized a two-stage design where we first identified the seven candidate ECM genes using a whole-genome screen, and subsequently, validated the upregulation of these seven genes in a large independent validation cohort that included a range of metastatic sites. We additionally observed that two of these genes (CEACAM6 and LUM) were significantly associated with metastatic-free and overall survival in primary ccRCC tumors.
The ECM pathway has been suggested to have a significant role in tumor progression and a therapeutic target in cancer [14]. However, the mechanism associated with the role of ECM with tumor progression differs across tumor sites and remains unknown for kidney cancer. Specifically, although known to be a tumor suppressor, DCN has been shown to be up regulated in oral cancer [15] and has been reported to be associated with poor outcome in RCC [16]. LUM encodes a member of the small leucine-rich proteoglycan family that includes DCN
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
metastatic tumors directly are critical to understanding the biology of metastases, the major
and also has been shown to be associated with poor prognosis in RCC [16] as well as in other cancers [17]. ADAMTS12 has been shown to be associated with outcome in colorectal cancer [18]; however, to our knowledge this is the first time that ADAMTS12 has been linked to RCC. SLIT2 and LMO3 have been reported to be hyper-methylated in pulmonary tumors (another smoking-related cancer) relative to patient-matched normal pulmonary tissue [19] and LMO3 has been shown to have oncogenic potential in neuroblastoma [20]. The SLIT/ROBO pathway is reported to be a therapeutic target for cancer [21]. And, although SLIT2 has been shown to be
has not been systematically studied. However, SLIT2 has been shown to be associated with metastases in other cancers [23,24]. CEACAM6 is an oncogene that has been found to be associated with tumor progression and metastasis in pancreatic, breast, colon, lung and gastric cancer [25,26] and to be associated with poor outcome in RCC [16]. Similar to our observations in ccRCC, CEACAM6 has been shown to be upregulated in colon metastasis in comparison to primary tumors [25]. Overall, these genes represent possible targets for metastatic ccRCC and deserve further attention.
Although we observed that seven ECM genes were upregulated in RCC metastases, currently no drugs exist to specifically target these seven genes. However, other ECM components (e.g., integrins, heparanases and matrix metalloproteinases) have inhibitors that are currently under study in phase I/II/III clinical trials [27-33]. Kinase genes have also recently been reported to be up-regulated in metastatic tumors relative to primary ccRCC tumors [34]. Thus, further study is required to determine if ECM or kinase genes represent therapeutic targets in metastatic RCC.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
methylated in approximately 25% of primary RCC tumors [22], its expression in metastatic RCC
There are some noteworthy limitations to our study. Primary and metastatic tumors were microdissected in the discovery cohort (analyzed fresh frozen tissue) but not the validation cohort (analyzed FFPE tissue). The process of microdissection minimizes stromal contamination; however, it may also decrease the ability to detect large global changes in gene expression in the tumor microenvironment. We acknowledge that intra-tumor and microenvironment heterogeneity can impact RNA profiles; thus, a single pathologist centrally reviewed all primary and metastatic tumors blinded to the patient identifiers and selected a
discovery cohort consisted solely of pulmonary metastases, we validated that the seven candidate genes were up regulated across metastatic locations in an independent validation cohort, suggesting that the presence of adjacent stroma from the affected site did not significantly influence gene expression. We also acknowledge that there are biological differences between subjects with distant metastases who undergo metastasectomy and subjects ineligible for metastasectomy. Patients with high-volume disease, rapid progression or poor performance status are less likely to undergo metastasectomy [35,36]. Conversely, the presence of pancreatic metastases in RCC is associated with a more indolent clinical course [37]. Thus, our cohort of patient-matched primary and metastatic tumors may not be representative of all subjects with metastatic disease. It is also important to note that the metastases were treated with metastasectomy rather than systemic therapy with the exception of one patient. Lastly, our study is hypothesis generating and thus requires further functional validation to demonstrate that therapeutic targeting of ECM genes inhibits tumor growth and improves clinical outcomes.
In summary, the results described herein provide important molecular evidence that metastatic ccRCC tumors are different than primary tumors and thus the metastatic tumor should be
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
representative tissue block (highest grade and presence of necrosis) for analysis. Although the
interrogated to define the key mechanisms underlying the development of metastatic disease. Specifically, our study indicates that ECM genes are upregulated in metastases in comparison to patient-matched primary RCC tumors and thus potentially offer insight into novel therapeutic targets. Overall, these ECM genes may be part of an ECM ‘remodeling’ program to facilitate metastatic colonization. Future preclinical studies are warranted to selectively therapeutically target upregulated ECM genes in metastases while maintaining proper ECM integrity in normal tissue.
The authors acknowledge the Mayo Clinic Cancer Center Biospecimens Accessioning and Processing Core, Pathology Research Core and the Gene Analysis Shared Resource. We thank all patients and their families for their contributions to this study. The authors would also like to acknowledge the support provided by Gloria A. and Thomas J. Dutson and the Jr. Kidney Research Endowment.
FUNDING This work was supported by the Mayo Clinic Center for Individualized Medicine Clinomics Program (T.H.H.); Gerstner Family Career Development Award (T.H.H.), the National Cancer Institute at the National Institutes of Health [K12CA90628 to T.H.H., R01CA134466 to A.S.P., R21CA176422 to J.E.E.].
DISCLOSURE The authors have declared no conflicts of interest.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
ACKNOWLEDGEMENTS
REFERENCES 1. 2. 3. 4.
7. 8. 9.
10.
11. 12.
13. 14. 15. 16.
17. 18.
19. 20.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
5. 6.
Gerlinger M, Horswell S, Larkin J, et al. Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing. Nat Genet 2014; 46: 225-233. Park WH, Eisen T. Prognostic factors in renal cell cancer. BJU Int 2007; 99: 1277-1281. Courtney KD, Choueiri TK. Updates on novel therapies for metastatic renal cell carcinoma. Ther Adv Med Oncol 2010; 2: 209-219. Pal SK, Williams S, Josephson DY, et al. Novel therapies for metastatic renal cell carcinoma: efforts to expand beyond the VEGF/mTOR signaling paradigm. Mol Cancer Ther 2012; 11: 526537. Milella M, Felici A. Biology of metastatic renal cell carcinoma. J Cancer 2011; 2: 369-373. Takahashi M, Rhodes DR, Furge KA, et al. Gene expression profiling of clear cell renal cell carcinoma: gene identification and prognostic classification. Proc Natl Acad Sci U S A 2001; 98: 9754-9759. Kosari F, Parker AS, Kube DM, et al. Clear cell renal cell carcinoma: gene expression analyses identify a potential signature for tumor aggressiveness. Clin Cancer Res 2005; 11: 5128-5139. TCGA TCGA. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 2013; 499: 43-49. Tan X, Zhai Y, Chang W, et al. Global analysis of metastasis-associated gene expression in primary cultures from clinical specimens of clear-cell renal-cell carcinoma. Int J Cancer 2008; 123: 1080-1088. Wuttig D, Baier B, Fuessel S, et al. Gene signatures of pulmonary metastases of renal cell carcinoma reflect the disease-free interval and the number of metastases per patient. Int J Cancer 2009; 125: 474-482. Ballman KV, Grill DE, Oberg AL, et al. Faster cyclic loess: normalizing RNA arrays via linear models. Bioinformatics 2004; 20: 2778-2786. Poplawski AB, Jankowski M, Erickson SW, et al. Frequent genetic differences between matched primary and metastatic breast cancer provide an approach to identification of biomarkers for disease progression. Eur J Hum Genet 2010; 18: 560-568. Malek JA, Mery E, Mahmoud YA, et al. Copy number variation analysis of matched ovarian primary tumors and peritoneal metastasis. PLoS One 2011; 6: e28561. Jarvelainen H, Sainio A, Koulu M, et al. Extracellular matrix molecules: potential targets in pharmacotherapy. Pharmacol Rev 2009; 61: 198-223. Banerjee AG, Bhattacharyya I, Lydiatt WM, et al. Aberrant expression and localization of decorin in human oral dysplasia and squamous cell carcinoma. Cancer Res 2003; 63: 7769-7776. Brannon AR, Reddy A, Seiler M, et al. Molecular Stratification of Clear Cell Renal Cell Carcinoma by Consensus Clustering Reveals Distinct Subtypes and Survival Patterns. Genes Cancer 2010; 1: 152-163. Matsuda Y, Yamamoto T, Kudo M, et al. Expression and roles of lumican in lung adenocarcinoma and squamous cell carcinoma. Int J Oncol 2008; 33: 1177-1185. Wang D, Zhu T, Zhang FB, et al. Expression of ADAMTS12 in colorectal cancer-associated stroma prevents cancer development and is a good prognostic indicator of colorectal cancer. Dig Dis Sci 2011; 56: 3281-3287. Kwon YJ, Lee SJ, Koh JS, et al. Genome-wide analysis of DNA methylation and the gene expression change in lung cancer. J Thorac Oncol 2012; 7: 20-33. Isogai E, Ohira M, Ozaki T, et al. Oncogenic LMO3 collaborates with HEN2 to enhance neuroblastoma cell growth through transactivation of Mash1. PLoS One 2011; 6: e19297.
21. 22. 23. 24. 25. 26.
28.
29.
30. 31.
32.
33.
34.
35. 36. 37.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
27.
Gara RK, Kumari S, Ganju A, et al. Slit/Robo pathway: a promising therapeutic target for cancer. Drug Discov Today 2015; 20: 156-164. Astuti D, Da Silva NF, Dallol A, et al. SLIT2 promoter methylation analysis in neuroblastoma, Wilms' tumour and renal cell carcinoma. Br J Cancer 2004; 90: 515-521. Yang XM, Han HX, Sui F, et al. Slit-Robo signaling mediates lymphangiogenesis and promotes tumor lymphatic metastasis. Biochem Biophys Res Commun 2010; 396: 571-577. Bartholow TL, Becich MJ, Chandran UR, et al. Immunohistochemical staining of slit2 in primary and metastatic prostatic adenocarcinoma. Transl Oncol 2011; 4: 314-320. Blumenthal RD, Leon E, Hansen HJ, et al. Expression patterns of CEACAM5 and CEACAM6 in primary and metastatic cancers. BMC Cancer 2007; 7: 2. Zang M, Zhang B, Zhang Y, et al. CEACAM6 promotes gastric cancer invasion and metastasis by inducing epithelial-mesenchymal transition via PI3K/AKT signaling pathway. PLoS One 2014; 9: e112908. Vansteenkiste J, Barlesi F, Waller CF, et al. Cilengitide combined with cetuximab and platinumbased chemotherapy as first-line treatment in advanced non-small-cell lung cancer (NSCLC) patients: results of an open-label, randomized, controlled phase II study (CERTO). Ann Oncol 2015; 26: 1734-1740. Stupp R, Hegi ME, Gorlia T, et al. Cilengitide combined with standard treatment for patients with newly diagnosed glioblastoma with methylated MGMT promoter (CENTRIC EORTC 26071-22072 study): a multicentre, randomised, open-label, phase 3 trial. Lancet Oncol 2014; 15: 1100-1108. Elez E, Kocakova I, Hohler T, et al. Abituzumab combined with cetuximab plus irinotecan versus cetuximab plus irinotecan alone for patients with KRAS wild-type metastatic colorectal cancer: the randomised phase I/II POSEIDON trial. Ann Oncol 2015; 26: 132-140. Feagan BG, Rutgeerts P, Sands BE, et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis. N Engl J Med 2013; 369: 699-710. Khasraw M, Pavlakis N, McCowatt S, et al. Multicentre phase I/II study of PI-88, a heparanase inhibitor in combination with docetaxel in patients with metastatic castrate-resistant prostate cancer. Ann Oncol 2010; 21: 1302-1307. Liu CJ, Lee PH, Lin DY, et al. Heparanase inhibitor PI-88 as adjuvant therapy for hepatocellular carcinoma after curative resection: a randomized phase II trial for safety and optimal dosage. J Hepatol 2009; 50: 958-968. Ritchie JP, Ramani VC, Ren Y, et al. SST0001, a chemically modified heparin, inhibits myeloma growth and angiogenesis via disruption of the heparanase/syndecan-1 axis. Clin Cancer Res 2011; 17: 1382-1393. Ghatalia P, Yang ES, Lasseigne BN, et al. Kinase Gene Expression Profiling of Metastatic Clear Cell Renal Cell Carcinoma Tissue Identifies Potential New Therapeutic Targets. PLoS One 2016; 11: e0160924. Alt AL, Boorjian SA, Lohse CM, et al. Survival after complete surgical resection of multiple metastases from renal cell carcinoma. Cancer 2011; 117: 2873-2882. Dabestani S, Marconi L, Hofmann F, et al. Local treatments for metastases of renal cell carcinoma: a systematic review. Lancet Oncol 2014; 15: e549-561. Kalra S, Atkinson BJ, Matrana MR, et al. Prognosis of patients with metastatic renal cell carcinoma and pancreatic metastases. BJU Int 2016; 117: 761-765.
FIGURE LEGEND Figure 1: Consort diagram for the validation stage. Differential expression analyses utilized data from the 97 patients who had Nanostring data available on their primary tumor or their metastatic tumor. That is, patients who had Nanostring data available on only their primary tumor (or similarly, on only their metastatic tumor) provide information regarding within group variance in the differential expression analyses.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
254x190mm (96 x 96 DPI)
Table 1: Clinical and pathological characteristics of the discovery and validation cohorts.
M0 (N=9) Subtype Papillary RCC Clear Cell RCC
Discovery Cohort M1 Total (N=6) (N=15)
M0 (N=47) 0 (0.0%) 47 (100.0%)
Validation Cohort M1 Total (N=50) (N=97)
1 (11.1%) 8 (88.9%)
0 (0.0%) 6 (100.0%)
1 (6.7%) 14 (93.3%)
0 (0.0%) 50 (100.0%)
0 (0.0%) 97 (100.0%)
1 (11.1%)
1 (16.7%)
2 (13.3%)
13 (27.7%)
13 (26.0%)
26 (26.8%)
37 (74.0%)
71 (73.2%)
Gender Female
5 (83.3%)
13 (86.7%)
Mean
61.4
54.9
58.8
62.9
58.0
60.4
Median
62.2
56.1
57.8
65.3
58.8
61.4
Range
(45.4-73.4)
(46.2-62.6)
(45.4-73.4)
(34.9-78.8)
(38.2-73.7)
(34.9-78.8)
Mean
11
10.7
10.9
9.7
10.9
10.3
Median
10
10.9
10.7
9.0
10.0
9.5
Range
(9.0-14.4)
(4.5-20.0)
(4.5-20.0)
(2.5-18.0)
(2.1-23.0)
(2.1-23.0)
9 (19.1%)
0 (0.0%)
9 (9.3%)
Male Age at Surgery
Tumor Size
TNM Stage
1 2
3 (33.3%)
0 (0.0%)
3 (20.0%)
16 (34.0%)
0 (0.0%)
16 (16.5%)
3
6 (66.7%)
0 (0.0%)
6 (40.0%)
21 (44.7%)
0 (0.0%)
21 (21.6%)
4
0 (0.0%)
6 (100.0%)
6 (40.0%)
1 (2.1%)
50 (100.0%)
51 (52.6%)
0 (0.0%)
1 (2.0%)
1 (1.0%)
Grade 1 2
1 (11.1%)
0 (0.0%)
1 (6.7%)
10 (21.3%)
4 (8.0%)
14 (14.4%)
3
5 (55.6%)
4 (66.7%)
9 (60.0%)
28 (59.6%)
29 (58.0%)
57 (58.8%)
4
3 (33.3%)
2 (33.3%)
5 (33.3%)
9 (19.1%)
16 (32.0%)
25 (25.8%)
0
3 (33.3%)
0 (0.0%)
3 (20.0%)
24 (51.1%)
15 (30.0%)
39 (40.2%)
1
6 (66.7%)
6 (100.0%)
12 (80.0%)
23 (48.9%)
35 (70.0%)
58 (59.8%)
Necrosis
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
8 (88.9%)
34 (72.3%)
Table 2: Discovery (Affymetrix U133 Plus2) and validation (Nanostring) of seven genes that were differentially expressed between patient-matched primary and metastatic RCC tumors. Discovery Cohort
Chro m 4p
Affymetrix Probeset 228850_s_at
Fold Change (=Met/Primary) 2.37
P 1.42E-06
Fold Change (=Met/Primary) 1.68
P 0.0048
5p
226997_at
2.41
5.57E-06
1.84
0.00039
6q 12p 12q 12q 19q
213519_s_at 204424_s_at 211813_x_at 201744_s_at 203757_s_at
2.33 6.86 4.17 11.53 11.12
5.00E-06 1.44E-05 4.56E-07 1.51E-06 9.90E-06
2.48 4.4 6.32 5.9 2.26
2.11E-05 8.76E-10 4.82E-13 2.31E-12 1.06E-05
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Gene SLIT2 ADAMTS1 2 LAMA2 LMO3 DCN LUM CEACAM6
Validation Cohort
Table 3: Clinical and pathological features associated with the 118 metastatic tumors representing 87 patients. Total (N=118)* 17 (14.4%) 7 (5.9%) 5 (4.2%) 9 (7.6%) 6 (5.1%) 6 (5.1%) 6 (5.1%) 44 (37.3%) 4 (3.4%) 14 (11.9%)
Metastatic Grade 2 3 4
24 (20.3%) 71 (60.2%) 23 (19.5%)
Metastatic Necrosis No Yes
71 (60.2%) 47 (39.8%)
Metastatic Sarcomatoid No 110 (93.2%) Yes 8 (6.8%) *Some patients had more than one serial metastatic tumor analyzed.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Metastatic Site Bone Brain Contralateral Adrenal Ipsilateral Adrenal Liver Non-regional Nodes Pancreas Pulmonary Skin Other
Table 4: Association of gene expression with overall survival and metastasis-free survival in primary ccRCC tumors using TCGA data.
Gene SLIT2 ADAMTS12 LAMA2 LMO3 DCN LUM CEACAM6
Chrom* 4p 5p 6q 12p 12q 12q 19q
Overall Survival (N=428, 139 events) P HR (95% CI) 0.928 (0.841 - 1.025) 0.14 1.003 (0.905 - 1.113) 0.95 0.994 (0.911 - 1.085) 0.9 1.026 (0.952 - 1.106) 0.5 1.057 (0.995 - 1.123) 0.07 1.093 (1.022 - 1.17) 0.0099 1.246 (1.081 - 1.437) 0.0024
Metastasis-free Survival (N=357, 60 events) P HR (95% CI) 1.111 (0.956 - 1.291) 0.17 1.095 (0.926 - 1.295) 0.29 1.113 (0.971 - 1.276) 0.12 1.113 (0.989 - 1.251) 0.075 1.122 (1.018 - 1.237) 0.02 1.166 (1.045 - 1.301) 0.0059 1.329 (1.073 - 1.648) 0.009324
*Chromosome arm Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Differential gene expression profiling of matched primary renal cell carcinoma and metastases reveals upregulation of extracellular matrix genes T. H. Ho1, D. J. Serie2, M. Parasramka3, J. C. Cheville4, B. M. Bot5, W. Tan6, L. Wang7, R. W. Joseph6, T. Hilton2, B. C. Leibovich8, A. S. Parker2, J. E. Eckel-Passow7# 1
Division of Hematology and Medical Oncology, Mayo Clinic, Scottsdale, USA Department of Health Sciences Research, Mayo Clinic, Jacksonville, USA 3 Department of Cancer Biology, Mayo Clinic, Jacksonville, USA 4 Laboratory Medicine and Pathology, Mayo Clinic, Rochester, USA 5 Computational Oncology, Sage Bionetworks, Seattle, USA 6 Division of Hematology/Oncology, Mayo Clinic, Jacksonville, USA 7 Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, USA 8 Department of Urology, Mayo Clinic, Rochester, USA 2
Corresponding Author: Prof. Jeanette E. Eckel-Passow Division of Biomedical Statistics and Informatics Mayo Clinic 200 First Street SW Rochester, MN 55905 USA Phone: 001-507-538-6512 E-mail: [email protected]
Word count of abstract: 273 Word count of text: 3007 Number of figures: 1 Number of tables: 4 Key message: We validated gene expression profiles using a large set of patient-matched primary and metastatic renal cell carcinoma tumors and identified up regulation of extracellular matrix genes in metastases. Our study implicates up regulation of extracellular matrix genes as a critical molecular event leading to visceral, bone and soft tissue metastases in renal cell carcinoma. Supplementary Data are available online.
© The Author 2016. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
#
ABSTRACT Background: The majority of renal cell carcinoma (RCC) studies analyze primary tumors, and the corresponding results are extrapolated to metastatic RCC tumors. However, it is unknown if gene expression profiles from primary RCC tumors differs from patient-matched metastatic tumors. Thus, we sought to identify differentially expressed genes between patient-matched primary and metastatic RCC tumors in order to understand the molecular mechanisms
Patients and Methods: We compared gene expression profiles between patient-matched primary and metastatic RCC tumors using a two-stage design. First, we used Affymetrix microarrays on 15 pairs of primary RCC (14 clear cell RCC, 1 papillary) tumors and patientmatched pulmonary metastases. Second, we used a custom Nanostring panel to validate seven candidate genes in an independent cohort of 114 clear cell RCC (ccRCC) patients. Differential gene expression was evaluated using a mixed effect linear model; a random effect denoting patient was included to account for the paired data. Third, The Cancer Genome Atlas (TCGA) data were used to evaluate associations with metastasis-free and overall survival in primary ccRCC tumors. Results: We identified and validated up regulation of seven genes functionally involved in the formation of the extracellular matrix (ECM): DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. In primary ccRCC, CEACAM6 and LUM were significantly associated with metastasis-free and overall survival (P<0.01). Conclusions: We evaluated gene expression profiles using the largest set to date, to our knowledge, of patient-matched primary and metastatic ccRCC tumors and identified up regulation of ECM genes in metastases. Our study implicates up regulation of ECM genes as a critical molecular event leading to visceral, bone and soft tissue metastases in ccRCC.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
underlying the development of RCC metastases.
Keywords: Renal cell carcinoma, extracellular matrix, metastases
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
INTRODUCTION Complete surgical excision, in the form of either partial or radical nephrectomy, for localized renal cell carcinoma (RCC) offers an opportunity for cure in RCC patients. Even with strict adherence to the 2014 National Comprehensive Cancer Network or American Urological Association surveillance guidelines, one third of RCC recurrences will be missed [1]. Metastases of RCC tumors have been reported to virtually all organs; however, the most
variable and impact prognosis and response to therapy.
Although treatment options for metastatic RCC have increased over the past decade, mortality rates and five-year survival are unacceptably poor [3,4]. Thus, there is a clear clinical need to develop novel management and treatment strategies for metastatic RCC that will lower mortality rates and extend survival [5]. Due to tissue availability, genome-wide interrogation of biomarkers associated with metastatic progression and cancer-specific death has primarily been based on observations made in the primary tumor [6-8] and not in the more lethal, and more therapeutically-relevant metastatic lesion. And, genome-wide studies that have preliminarily interrogated metastatic specimens have done so using small sample sizes [9,10]. Thus, the molecular mechanisms that lead to RCC metastases remain largely unknown and require further study.
To address these gaps in knowledge, we identified genes that support RCC metastases by comparing gene expression profiles between metastatic tumors and their patient-matched primary tumor. We primarily focused on clear cell RCC (ccRCC), which accounts for more than 85% of RCC. Particularly, we employed a two-stage design where we first utilized gene
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
common sites are pulmonary and bone [2]. The location and number of metastases are highly
expression microarray technology to identify candidate genes that are associated with RCC metastases, and subsequently, validated the candidate genes in a large cohort of patientmatched primary and metastatic ccRCC tumors using a custom Nanostring assay.
METHODS
Discovery Cohort We identified 15 patients at Mayo Clinic Rochester who were treated surgically for RCC (14 ccRCC, 1 papillary), underwent metastasectomy for a pulmonary metastatic lesion, and had fresh-frozen tissue available from their primary tumor in addition to their pulmonary metastasis. One of the 15 patients had two separate pulmonary metastases; both were included in the study. A single pathologist (J.C.C.) performed a blinded comprehensive review of all tumors (primary and metastatic) to confirm histological subtype (1997 AJCC/UICC classification), tumor stage, 2012 International Society of Urological Pathology (ISUP) tumor grade, tumor size, and coagulative tumor necrosis.
Validation Cohort We identified 114 additional patients at Mayo Clinic Rochester who were treated surgically for ccRCC between 1990 and 2005, had synchronous or metachronous metastases, and underwent metastasectomy for at least one of their metastatic lesions and therefore had formalin-fixed, paraffin-embedded tissue available from their primary tumor and at least one patient-matched metastatic tumor. Unilateral multifocal disease and bilateral disease were not
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Patient Selection
considered metastatic ccRCC. A single pathologist (J.C.C.) performed a blinded comprehensive review of all tumors (primary and metastatic) to confirm histological subtype (1997 AJCC/UICC classification), tumor stage, 2012 International Society of Urological Pathology (ISUP) tumor grade, tumor size, and coagulative tumor necrosis. A representative paraffin-embedded tissue block with the highest grade and presence of necrosis was chosen from each resected tumor. If serial metastatic tumors were available, then all available metastatic tumors were analyzed in order to evaluate inter-tumor heterogeneity across serial metastatic tumors. Metastases were
nephrectomy (< 30 days) or metachronous if the diagnosis and resection of metastases occurred after (>30 days) nephrectomy. This study was approved by the Mayo Clinic IRB.
Affymetrix Gene Expression Microarrays (Discovery Stage) Laser capture microdissection of tumor cells was performed and samples were hybridized to the Affymetrix U133 Plus2 microarray chip, a genome-wide gene expression assay. Microarray analysis was conducted according to manufacturer’s instructions for the Affymetrix One Cycle Target Labeling and Control Reagents kit (Santa Clara, CA). Briefly, cDNA was generated from 5 µg of total RNA using SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA) and T7 Oligo(dT) primer. Subsequently, the products were column-purified (Affymetrix) and then in vitro transcribed to generate biotin-labeled cRNA. The transcribed products were then columnpurified, fragmented, and hybridized onto Affymetrix U133 Plus 2.0 GeneChips® at 45º C for 16 h. Subsequent to hybridization, the arrays were washed and stained with streptavidinphycoerythrin, then scanned in an Affymetrix GeneChip® Scanner 3000 (Santa Clara, CA). The data have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE85258.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
classified as synchronous if the metastatic tumor was resected at the same time as
Nanostring (Validation Stage) Total RNA was isolated from three slides per tumor at 10-µm sections of tumor-rich areas of FFPE tissue blocks using the AllPrep DNA/RNA FFPE kit reagents (Qiagen) following vendor's standard protocols. Isolated FFPE RNAwas treated with 20 units DNase I. The NanoString platform was used to quantify gene expression of seven candidate genes: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. 200 ng of each total RNA sample was prepared as per the manufacturer’s instructions. Gene expression was quantified on the NanoString
extracted using the nSolver software (Nanostring).
Statistical Methods Affymetrix Gene Expression Microarray Data (Discovery Stage) Data were normalized using fastlo [11]. The primary objective was to identify genes that were differentially expressed between patient-matched primary and metastatic tumors. Differential expression was tested using a linear mixed-effect model; models were fit per probe set using a random effect for patient to account for the paired data. Probe sets with p-value (P) <0.00002 and a fold change > 2 were considered candidate differentially expressed genes. We were particularly interested in genes that were up regulated in metastatic tumors as these genes have the largest potential to be therapeutically targetable.
Nanostring Data (Validation Stage) Bland-Altman plots were used to visualize agreement of gene expression between primarymetastatic tumor pairs as well as between serial metastatic tumors from the same individual. Linear mixed-effect models were used to determine if each of the seven genes was differentially expressed between the patient-matched primary and metastatic tumors; models were fit per gene using a random effect for patient to account for the paired data. Linear mixed-effects
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
nCounterTM and raw counts were generated with nSolverTM. Raw counts per gene were
models were also used to determine if the difference in gene expression between primarymetastatic tumor pairs was associated with metastatic site, metastatic timing (M0/M1), metastatic tumor grade, metastatic tumor necrosis (yes/no), and metastatic tumor sarcomatoid status (yes/no). To account for multiple testing, P < 0.01 was used for statistical testing.
The Cancer Genome Atlas (TCGA) ccRCC Data
Data Portal for 428 ccRCC subjects. Cox proportional hazards regression was used to evaluate the association of each gene with metastasis-free survival and overall survival, adjusting for age at diagnosis and tumor stage. Gene expression was modeled as a continuous variable and a linear relationship was verified. Multivariable Cox regression was used to evaluate independent gene predictors of outcome. To account for testing associations of seven genes, P < 0.01 was used.
RESULTS Discovery Stage: Differential Gene Expression Profiling of Patient-Matched Primary and Metastatic RCC Tumors The discovery cohort consisted of 15 patients that developed pulmonary metastatic tumors: 6 had a synchronous (M1) metastasis at the time of diagnosis and 9 had a metachronous (M0) metastasis (Table 1). The majority of patients were male (87%) with Fuhrman grade 3 primary RCC tumors (60%). Affymetrix gene expression profiling identified 38 probesets that were more than two-fold upregulated in metastatic tumors relative to patient-matched primary RCC tumors (P<2x10-5; Supplementary Table S1). Many of these probesets mapped to genes that were associated with lung surfactants and thus were excluded from further consideration, as were
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Level 3 RSEM normalized RNASeqV2 files were downloaded on January 8, 2013, from TCGA
probesets that mapped to open reading frames. The remaining probesets were observed to be enriched for extracellular matrix (ECM) genes. Thus, we chose to validate differential expression of the seven ECM genes that had probesets that met our significance criteria: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3 (Table 2).
Notably, one of the 15 patients in the discovery cohort had a papillary primary RCC tumor whereas the remaining 14 patients had a clear cell primary RCC tumor. Thus, as a sensitivity
seven candidate ECM genes remain highly significant.
Validation Stage: Independent Validation of Differentially Expressed ECM Genes Of the 114 patients identified, 97 (85%) had a primary tumor or a metastatic tumor that was successfully profiled by Nanostring (Figure 1; Table 1); 88 patients had a primary tumor (43 synchronous and 45 metachronous) successfully profiled and 87 patients had a metastatic tumor that was successfully profiled. A total of 118 metastases were available from the 87 patients who had at least one metastatic tumor successfully profiled (Table 3); 59 patients had Nanostring data for only a single distant metastasis and 28 patients had Nanostring data on two or more metastases. The median time from nephrectomy to the first metachronous metastasis was 1.85 years (37 days, maximum 10.82 years). Only one patient received systemic treatment between the initial nephrectomy and metastectomy. Metastases to pulmonary (37%) were the most common (Table 3). Comparing patient-matched primary and metastatic tumor pathologic characteristics, 59% had concordant grade, 60% had concordant necrosis status and 87% had concordant sarcomatoid dedifferentiation status (Supplementary Table S3).
In the validation cohort, DCN, SLIT2, LUM, LAMA2, ADAMTS12, and LMO3 were all significantly upregulated in metastatic tumors relative to patient-matched primary ccRCC tumors
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
analysis, we analyzed only the 14 ccRCC patients. As shown in Supplementary Table S2, the
(P<0.005; Table 2, Supplementary Figure S1). Supplementary Table S4 shows the results for testing whether differential expression was associated with metastatic site. There was a significant association between differential expression of LMO3 (P=0.0001) and CEACAM6 (P=0.0001) with metastatic site. Specifically, pulmonary metastases had larger differences of LMO3 gene expression between primary and metastatic tumors than metastases to bone, brain, liver, skin and nodes (P<0.01). Similarly, pulmonary metastases had larger differences of CEACAM6 gene expression than metastases to most other sites (P<0.01). In fact, CEACAM6
(M1) metastases had larger differences of DCN gene expression between primary and patientmatched metastatic ccRCC tumors in comparison to metachronous (M0) metastases (P<0.01; Supplementary Table S4). Non-necrotic metastases had larger differences of SLIT2 gene expression between primary and patient-matched metastatic ccRCC tumors in comparison to necrotic metastases (P<0.01; Supplementary Table S4). We did not observe a significant association between metastatic tumor sarcomatoid differentiation and difference in gene expression in any of the seven genes (P>0.01; Supplementary Table S4).
Association of ECM Genes with Survival in TCGA Primary ccRCC Tumors After adjusting for age at diagnosis, CEACAM6 and LUM were significantly associated with metastasis-free and overall survival (P<0.01; Table 4). When including both CEACAM and LUM in a multivariable Cox model (Supplemental Table S5), only LUM remained significantly (P<0.05) associated with overall survival whereas both LUM and CEACAM6 were significantly associated with metastasis-free survival (P<0.05). After adjustment for both age at diagnosis and tumor stage (Supplemental Table S6); none of the seven genes were significantly associated with either overall survival or metastasis-free survival (P>0.01).
DISCUSSION
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
was primarily expressed in pulmonary metastases (Supplementary Figure S1). Synchronous
Although treatment options for metastatic RCC have increased over the past decade, mortality and five-year survival remain unacceptably poor [3,4]. TCGA identified somatic mutations, cytogenetic and epigenetic alterations in primary ccRCC tumors [8]; however, less is known regarding common molecular alterations in the more lethal and therapeutically-relevant distant metastatic tumor. In fact, it has been suggested in other cancers that the primary tumor may not be representative of the metastatic tumor and more effective molecular targets could be discovered by interrogating the metastatic tumor [12,13]. Thus, studies that interrogate
cause of patient mortality.
We evaluated gene expression profiles using the largest set, to our knowledge, of patientmatched primary and metastatic ccRCC tumors and identified up regulation of seven ECM genes: DCN, SLIT2, LUM, LAMA2, ADAMTS12, CEACAM6 and LMO3. We utilized a two-stage design where we first identified the seven candidate ECM genes using a whole-genome screen, and subsequently, validated the upregulation of these seven genes in a large independent validation cohort that included a range of metastatic sites. We additionally observed that two of these genes (CEACAM6 and LUM) were significantly associated with metastatic-free and overall survival in primary ccRCC tumors.
The ECM pathway has been suggested to have a significant role in tumor progression and a therapeutic target in cancer [14]. However, the mechanism associated with the role of ECM with tumor progression differs across tumor sites and remains unknown for kidney cancer. Specifically, although known to be a tumor suppressor, DCN has been shown to be up regulated in oral cancer [15] and has been reported to be associated with poor outcome in RCC [16]. LUM encodes a member of the small leucine-rich proteoglycan family that includes DCN
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
metastatic tumors directly are critical to understanding the biology of metastases, the major
and also has been shown to be associated with poor prognosis in RCC [16] as well as in other cancers [17]. ADAMTS12 has been shown to be associated with outcome in colorectal cancer [18]; however, to our knowledge this is the first time that ADAMTS12 has been linked to RCC. SLIT2 and LMO3 have been reported to be hyper-methylated in pulmonary tumors (another smoking-related cancer) relative to patient-matched normal pulmonary tissue [19] and LMO3 has been shown to have oncogenic potential in neuroblastoma [20]. The SLIT/ROBO pathway is reported to be a therapeutic target for cancer [21]. And, although SLIT2 has been shown to be
has not been systematically studied. However, SLIT2 has been shown to be associated with metastases in other cancers [23,24]. CEACAM6 is an oncogene that has been found to be associated with tumor progression and metastasis in pancreatic, breast, colon, lung and gastric cancer [25,26] and to be associated with poor outcome in RCC [16]. Similar to our observations in ccRCC, CEACAM6 has been shown to be upregulated in colon metastasis in comparison to primary tumors [25]. Overall, these genes represent possible targets for metastatic ccRCC and deserve further attention.
Although we observed that seven ECM genes were upregulated in RCC metastases, currently no drugs exist to specifically target these seven genes. However, other ECM components (e.g., integrins, heparanases and matrix metalloproteinases) have inhibitors that are currently under study in phase I/II/III clinical trials [27-33]. Kinase genes have also recently been reported to be up-regulated in metastatic tumors relative to primary ccRCC tumors [34]. Thus, further study is required to determine if ECM or kinase genes represent therapeutic targets in metastatic RCC.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
methylated in approximately 25% of primary RCC tumors [22], its expression in metastatic RCC
There are some noteworthy limitations to our study. Primary and metastatic tumors were microdissected in the discovery cohort (analyzed fresh frozen tissue) but not the validation cohort (analyzed FFPE tissue). The process of microdissection minimizes stromal contamination; however, it may also decrease the ability to detect large global changes in gene expression in the tumor microenvironment. We acknowledge that intra-tumor and microenvironment heterogeneity can impact RNA profiles; thus, a single pathologist centrally reviewed all primary and metastatic tumors blinded to the patient identifiers and selected a
discovery cohort consisted solely of pulmonary metastases, we validated that the seven candidate genes were up regulated across metastatic locations in an independent validation cohort, suggesting that the presence of adjacent stroma from the affected site did not significantly influence gene expression. We also acknowledge that there are biological differences between subjects with distant metastases who undergo metastasectomy and subjects ineligible for metastasectomy. Patients with high-volume disease, rapid progression or poor performance status are less likely to undergo metastasectomy [35,36]. Conversely, the presence of pancreatic metastases in RCC is associated with a more indolent clinical course [37]. Thus, our cohort of patient-matched primary and metastatic tumors may not be representative of all subjects with metastatic disease. It is also important to note that the metastases were treated with metastasectomy rather than systemic therapy with the exception of one patient. Lastly, our study is hypothesis generating and thus requires further functional validation to demonstrate that therapeutic targeting of ECM genes inhibits tumor growth and improves clinical outcomes.
In summary, the results described herein provide important molecular evidence that metastatic ccRCC tumors are different than primary tumors and thus the metastatic tumor should be
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
representative tissue block (highest grade and presence of necrosis) for analysis. Although the
interrogated to define the key mechanisms underlying the development of metastatic disease. Specifically, our study indicates that ECM genes are upregulated in metastases in comparison to patient-matched primary RCC tumors and thus potentially offer insight into novel therapeutic targets. Overall, these ECM genes may be part of an ECM ‘remodeling’ program to facilitate metastatic colonization. Future preclinical studies are warranted to selectively therapeutically target upregulated ECM genes in metastases while maintaining proper ECM integrity in normal tissue.
The authors acknowledge the Mayo Clinic Cancer Center Biospecimens Accessioning and Processing Core, Pathology Research Core and the Gene Analysis Shared Resource. We thank all patients and their families for their contributions to this study. The authors would also like to acknowledge the support provided by Gloria A. and Thomas J. Dutson and the Jr. Kidney Research Endowment.
FUNDING This work was supported by the Mayo Clinic Center for Individualized Medicine Clinomics Program (T.H.H.); Gerstner Family Career Development Award (T.H.H.), the National Cancer Institute at the National Institutes of Health [K12CA90628 to T.H.H., R01CA134466 to A.S.P., R21CA176422 to J.E.E.].
DISCLOSURE The authors have declared no conflicts of interest.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
ACKNOWLEDGEMENTS
REFERENCES 1. 2. 3. 4.
7. 8. 9.
10.
11. 12.
13. 14. 15. 16.
17. 18.
19. 20.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
5. 6.
Gerlinger M, Horswell S, Larkin J, et al. Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing. Nat Genet 2014; 46: 225-233. Park WH, Eisen T. Prognostic factors in renal cell cancer. BJU Int 2007; 99: 1277-1281. Courtney KD, Choueiri TK. Updates on novel therapies for metastatic renal cell carcinoma. Ther Adv Med Oncol 2010; 2: 209-219. Pal SK, Williams S, Josephson DY, et al. Novel therapies for metastatic renal cell carcinoma: efforts to expand beyond the VEGF/mTOR signaling paradigm. Mol Cancer Ther 2012; 11: 526537. Milella M, Felici A. Biology of metastatic renal cell carcinoma. J Cancer 2011; 2: 369-373. Takahashi M, Rhodes DR, Furge KA, et al. Gene expression profiling of clear cell renal cell carcinoma: gene identification and prognostic classification. Proc Natl Acad Sci U S A 2001; 98: 9754-9759. Kosari F, Parker AS, Kube DM, et al. Clear cell renal cell carcinoma: gene expression analyses identify a potential signature for tumor aggressiveness. Clin Cancer Res 2005; 11: 5128-5139. TCGA TCGA. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 2013; 499: 43-49. Tan X, Zhai Y, Chang W, et al. Global analysis of metastasis-associated gene expression in primary cultures from clinical specimens of clear-cell renal-cell carcinoma. Int J Cancer 2008; 123: 1080-1088. Wuttig D, Baier B, Fuessel S, et al. Gene signatures of pulmonary metastases of renal cell carcinoma reflect the disease-free interval and the number of metastases per patient. Int J Cancer 2009; 125: 474-482. Ballman KV, Grill DE, Oberg AL, et al. Faster cyclic loess: normalizing RNA arrays via linear models. Bioinformatics 2004; 20: 2778-2786. Poplawski AB, Jankowski M, Erickson SW, et al. Frequent genetic differences between matched primary and metastatic breast cancer provide an approach to identification of biomarkers for disease progression. Eur J Hum Genet 2010; 18: 560-568. Malek JA, Mery E, Mahmoud YA, et al. Copy number variation analysis of matched ovarian primary tumors and peritoneal metastasis. PLoS One 2011; 6: e28561. Jarvelainen H, Sainio A, Koulu M, et al. Extracellular matrix molecules: potential targets in pharmacotherapy. Pharmacol Rev 2009; 61: 198-223. Banerjee AG, Bhattacharyya I, Lydiatt WM, et al. Aberrant expression and localization of decorin in human oral dysplasia and squamous cell carcinoma. Cancer Res 2003; 63: 7769-7776. Brannon AR, Reddy A, Seiler M, et al. Molecular Stratification of Clear Cell Renal Cell Carcinoma by Consensus Clustering Reveals Distinct Subtypes and Survival Patterns. Genes Cancer 2010; 1: 152-163. Matsuda Y, Yamamoto T, Kudo M, et al. Expression and roles of lumican in lung adenocarcinoma and squamous cell carcinoma. Int J Oncol 2008; 33: 1177-1185. Wang D, Zhu T, Zhang FB, et al. Expression of ADAMTS12 in colorectal cancer-associated stroma prevents cancer development and is a good prognostic indicator of colorectal cancer. Dig Dis Sci 2011; 56: 3281-3287. Kwon YJ, Lee SJ, Koh JS, et al. Genome-wide analysis of DNA methylation and the gene expression change in lung cancer. J Thorac Oncol 2012; 7: 20-33. Isogai E, Ohira M, Ozaki T, et al. Oncogenic LMO3 collaborates with HEN2 to enhance neuroblastoma cell growth through transactivation of Mash1. PLoS One 2011; 6: e19297.
21. 22. 23. 24. 25. 26.
28.
29.
30. 31.
32.
33.
34.
35. 36. 37.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
27.
Gara RK, Kumari S, Ganju A, et al. Slit/Robo pathway: a promising therapeutic target for cancer. Drug Discov Today 2015; 20: 156-164. Astuti D, Da Silva NF, Dallol A, et al. SLIT2 promoter methylation analysis in neuroblastoma, Wilms' tumour and renal cell carcinoma. Br J Cancer 2004; 90: 515-521. Yang XM, Han HX, Sui F, et al. Slit-Robo signaling mediates lymphangiogenesis and promotes tumor lymphatic metastasis. Biochem Biophys Res Commun 2010; 396: 571-577. Bartholow TL, Becich MJ, Chandran UR, et al. Immunohistochemical staining of slit2 in primary and metastatic prostatic adenocarcinoma. Transl Oncol 2011; 4: 314-320. Blumenthal RD, Leon E, Hansen HJ, et al. Expression patterns of CEACAM5 and CEACAM6 in primary and metastatic cancers. BMC Cancer 2007; 7: 2. Zang M, Zhang B, Zhang Y, et al. CEACAM6 promotes gastric cancer invasion and metastasis by inducing epithelial-mesenchymal transition via PI3K/AKT signaling pathway. PLoS One 2014; 9: e112908. Vansteenkiste J, Barlesi F, Waller CF, et al. Cilengitide combined with cetuximab and platinumbased chemotherapy as first-line treatment in advanced non-small-cell lung cancer (NSCLC) patients: results of an open-label, randomized, controlled phase II study (CERTO). Ann Oncol 2015; 26: 1734-1740. Stupp R, Hegi ME, Gorlia T, et al. Cilengitide combined with standard treatment for patients with newly diagnosed glioblastoma with methylated MGMT promoter (CENTRIC EORTC 26071-22072 study): a multicentre, randomised, open-label, phase 3 trial. Lancet Oncol 2014; 15: 1100-1108. Elez E, Kocakova I, Hohler T, et al. Abituzumab combined with cetuximab plus irinotecan versus cetuximab plus irinotecan alone for patients with KRAS wild-type metastatic colorectal cancer: the randomised phase I/II POSEIDON trial. Ann Oncol 2015; 26: 132-140. Feagan BG, Rutgeerts P, Sands BE, et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis. N Engl J Med 2013; 369: 699-710. Khasraw M, Pavlakis N, McCowatt S, et al. Multicentre phase I/II study of PI-88, a heparanase inhibitor in combination with docetaxel in patients with metastatic castrate-resistant prostate cancer. Ann Oncol 2010; 21: 1302-1307. Liu CJ, Lee PH, Lin DY, et al. Heparanase inhibitor PI-88 as adjuvant therapy for hepatocellular carcinoma after curative resection: a randomized phase II trial for safety and optimal dosage. J Hepatol 2009; 50: 958-968. Ritchie JP, Ramani VC, Ren Y, et al. SST0001, a chemically modified heparin, inhibits myeloma growth and angiogenesis via disruption of the heparanase/syndecan-1 axis. Clin Cancer Res 2011; 17: 1382-1393. Ghatalia P, Yang ES, Lasseigne BN, et al. Kinase Gene Expression Profiling of Metastatic Clear Cell Renal Cell Carcinoma Tissue Identifies Potential New Therapeutic Targets. PLoS One 2016; 11: e0160924. Alt AL, Boorjian SA, Lohse CM, et al. Survival after complete surgical resection of multiple metastases from renal cell carcinoma. Cancer 2011; 117: 2873-2882. Dabestani S, Marconi L, Hofmann F, et al. Local treatments for metastases of renal cell carcinoma: a systematic review. Lancet Oncol 2014; 15: e549-561. Kalra S, Atkinson BJ, Matrana MR, et al. Prognosis of patients with metastatic renal cell carcinoma and pancreatic metastases. BJU Int 2016; 117: 761-765.
FIGURE LEGEND Figure 1: Consort diagram for the validation stage. Differential expression analyses utilized data from the 97 patients who had Nanostring data available on their primary tumor or their metastatic tumor. That is, patients who had Nanostring data available on only their primary tumor (or similarly, on only their metastatic tumor) provide information regarding within group variance in the differential expression analyses.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
254x190mm (96 x 96 DPI)
Table 1: Clinical and pathological characteristics of the discovery and validation cohorts.
M0 (N=9) Subtype Papillary RCC Clear Cell RCC
Discovery Cohort M1 Total (N=6) (N=15)
M0 (N=47) 0 (0.0%) 47 (100.0%)
Validation Cohort M1 Total (N=50) (N=97)
1 (11.1%) 8 (88.9%)
0 (0.0%) 6 (100.0%)
1 (6.7%) 14 (93.3%)
0 (0.0%) 50 (100.0%)
0 (0.0%) 97 (100.0%)
1 (11.1%)
1 (16.7%)
2 (13.3%)
13 (27.7%)
13 (26.0%)
26 (26.8%)
37 (74.0%)
71 (73.2%)
Gender Female
5 (83.3%)
13 (86.7%)
Mean
61.4
54.9
58.8
62.9
58.0
60.4
Median
62.2
56.1
57.8
65.3
58.8
61.4
Range
(45.4-73.4)
(46.2-62.6)
(45.4-73.4)
(34.9-78.8)
(38.2-73.7)
(34.9-78.8)
Mean
11
10.7
10.9
9.7
10.9
10.3
Median
10
10.9
10.7
9.0
10.0
9.5
Range
(9.0-14.4)
(4.5-20.0)
(4.5-20.0)
(2.5-18.0)
(2.1-23.0)
(2.1-23.0)
9 (19.1%)
0 (0.0%)
9 (9.3%)
Male Age at Surgery
Tumor Size
TNM Stage
1 2
3 (33.3%)
0 (0.0%)
3 (20.0%)
16 (34.0%)
0 (0.0%)
16 (16.5%)
3
6 (66.7%)
0 (0.0%)
6 (40.0%)
21 (44.7%)
0 (0.0%)
21 (21.6%)
4
0 (0.0%)
6 (100.0%)
6 (40.0%)
1 (2.1%)
50 (100.0%)
51 (52.6%)
0 (0.0%)
1 (2.0%)
1 (1.0%)
Grade 1 2
1 (11.1%)
0 (0.0%)
1 (6.7%)
10 (21.3%)
4 (8.0%)
14 (14.4%)
3
5 (55.6%)
4 (66.7%)
9 (60.0%)
28 (59.6%)
29 (58.0%)
57 (58.8%)
4
3 (33.3%)
2 (33.3%)
5 (33.3%)
9 (19.1%)
16 (32.0%)
25 (25.8%)
0
3 (33.3%)
0 (0.0%)
3 (20.0%)
24 (51.1%)
15 (30.0%)
39 (40.2%)
1
6 (66.7%)
6 (100.0%)
12 (80.0%)
23 (48.9%)
35 (70.0%)
58 (59.8%)
Necrosis
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
8 (88.9%)
34 (72.3%)
Table 2: Discovery (Affymetrix U133 Plus2) and validation (Nanostring) of seven genes that were differentially expressed between patient-matched primary and metastatic RCC tumors. Discovery Cohort
Chro m 4p
Affymetrix Probeset 228850_s_at
Fold Change (=Met/Primary) 2.37
P 1.42E-06
Fold Change (=Met/Primary) 1.68
P 0.0048
5p
226997_at
2.41
5.57E-06
1.84
0.00039
6q 12p 12q 12q 19q
213519_s_at 204424_s_at 211813_x_at 201744_s_at 203757_s_at
2.33 6.86 4.17 11.53 11.12
5.00E-06 1.44E-05 4.56E-07 1.51E-06 9.90E-06
2.48 4.4 6.32 5.9 2.26
2.11E-05 8.76E-10 4.82E-13 2.31E-12 1.06E-05
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Gene SLIT2 ADAMTS1 2 LAMA2 LMO3 DCN LUM CEACAM6
Validation Cohort
Table 3: Clinical and pathological features associated with the 118 metastatic tumors representing 87 patients. Total (N=118)* 17 (14.4%) 7 (5.9%) 5 (4.2%) 9 (7.6%) 6 (5.1%) 6 (5.1%) 6 (5.1%) 44 (37.3%) 4 (3.4%) 14 (11.9%)
Metastatic Grade 2 3 4
24 (20.3%) 71 (60.2%) 23 (19.5%)
Metastatic Necrosis No Yes
71 (60.2%) 47 (39.8%)
Metastatic Sarcomatoid No 110 (93.2%) Yes 8 (6.8%) *Some patients had more than one serial metastatic tumor analyzed.
Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016
Metastatic Site Bone Brain Contralateral Adrenal Ipsilateral Adrenal Liver Non-regional Nodes Pancreas Pulmonary Skin Other
Table 4: Association of gene expression with overall survival and metastasis-free survival in primary ccRCC tumors using TCGA data.
Gene SLIT2 ADAMTS12 LAMA2 LMO3 DCN LUM CEACAM6
Chrom* 4p 5p 6q 12p 12q 12q 19q
Overall Survival (N=428, 139 events) P HR (95% CI) 0.928 (0.841 - 1.025) 0.14 1.003 (0.905 - 1.113) 0.95 0.994 (0.911 - 1.085) 0.9 1.026 (0.952 - 1.106) 0.5 1.057 (0.995 - 1.123) 0.07 1.093 (1.022 - 1.17) 0.0099 1.246 (1.081 - 1.437) 0.0024
Metastasis-free Survival (N=357, 60 events) P HR (95% CI) 1.111 (0.956 - 1.291) 0.17 1.095 (0.926 - 1.295) 0.29 1.113 (0.971 - 1.276) 0.12 1.113 (0.989 - 1.251) 0.075 1.122 (1.018 - 1.237) 0.02 1.166 (1.045 - 1.301) 0.0059 1.329 (1.073 - 1.648) 0.009324
*Chromosome arm Downloaded from http://annonc.oxfordjournals.org/ at Stanford University on December 21, 2016