Interferon-� Resistance Associated Genes inRenal Cell Carcinoma Identified by Expression ProfilingJames E. Korkola, G. Varuni Kondagunta, Victor E. Reuter, Robert J. Motzer*and R. S. K. Chaganti†From the Departments of Cell Biology (JEK, RSKC), Medicine (GVK, RJM, RSKC) and Pathology (VER), Memorial Sloan-KetteringCancer Center, New York, New York

Purpose: We identified differentially expressed genes associated with response to pegylated interferon-� treatment inpatients with renal cell carcinoma.Materials and Methods: We performed expression profiling on renal cell carcinoma specimens isolated from 23 patientswith metastatic disease who were subsequently treated with interferon. Significance Analysis for Microarrays software wasused to identify genes that were differentially expressed between patients with partial response compared to those withdisease progression.Results: A candidate gene approach looking at VHL and known target genes did not identify any genes whose expressioncorrelated with patient response. A global analysis of approximately 54,000 probe sets identified 4 genes that had expressioncorrelated with response. Reverse transcriptase-polymerase chain reaction analysis of 2 of these genes confirmed that theywere more highly expressed in tumors from patients who responded to interferon-�. Interestingly, both of these genes mappedto 4q31-32, a region that has been implicated as the site of a potential tumor suppressor gene in renal cell carcinoma.Conclusions: We have identified 4 genes (3 uncharacterized and 1 known) that may prove useful in predicting response tointerferon-� treatment in patients with renal cell carcinoma.

Key Words: interferon-alpha; gene expression profiling; genes, tumor suppressor; carcinoma, renal cell

Renal cell carcinoma is the most common neoplasm ofthe kidney, with approximately 32,000 cases diag-nosed each year in the United States.1,2 RCC tumors

present as distinct histological subtypes, namely clear cell,papillary and chromophobe subtypes.1,2 Early stage diseaseis treatable with 5-year survival rates approaching 95%.2

However, the survival rates for patients with metastaticdisease are extremely low, with a 2-year survival rate of lessthan 20%.2 This poor survival rate reflects the refractorynature of metastatic RCC to chemotherapeutic agents. Al-though new classes of drugs such as sunitinib and sorafenibthat target receptor tyrosine kinases show great promise forthe treatment of metastatic kidney cancer,3,4 the currentstandard of care is treatment with cytokines such as highdose interleukin-2 or interferon-�. The response rates tothese therapies are low, with only 10% to 20% of patientswith RCC responding to IFN-� therapy.2,5,6

IFNs are glycoproteins that activate a complex series ofcellular responses. IFNs can elicit their effects through anti-

Submitted for publication July 7, 2006.Study received institutional review board approval.Supported by the Byrne Fund at Memorial Sloan-Kettering

Cancer Center and Schering-Plough Research Institute.Reviewed by Schering-Plough Research Institute before submis-

sion for publication.* Financial interest and/or other relationship with Schering

Plough.† Correspondence: Memorial Sloan-Kettering Cancer Center, 1275

York Ave., New York, New York 10021 (telephone: 212-639-8121;FAX: 212-717-3541; e-mail: chagantr@mskcc.org).

See Editorial on page 1224.

0022-5347/07/1774-1264/0THE JOURNAL OF UROLOGY®

Copyright © 2007 by AMERICAN UROLOGICAL ASSOCIATION

1264

proliferative, antiviral, antitumor and immunomodulatoryactivities.7 A number of studies have attempted to identifygenes and functional pathways associated with an IFN re-sistant phenotype in a variety of tumor types.8,9 Many ofthese studies have been performed in vitro and, thus, anyhost or microenvironmental effects would be missed. Further-more, these studies usually focus on changes in expressionfollowing IFN treatment rather than on preexisting geneexpression differences that could predict response. As a re-sult many of the genes involved in IFN resistance in RCCremain unknown.

We describe expression profiling on a panel of RCC tu-mors from patients with known response to IFN-� treat-ment. These patients were treated with a pegylated form ofIFN-�-2b (PEG-Intron®), which is reported to have betterstability and reduced toxicity compared to free IFN-�.6 Wereport the identification of a small set of genes whose ex-pression was associated with a clinical response to IFN-�.

MATERIALS AND METHODS

PatientsTumor material was collected from a panel of 23 patientswith metastatic RCC who had received PEG-Intron treat-ment at Memorial Sloan-Kettering Cancer Center between2002 and 2004. Patient material was obtained under aninstitutional review board approved protocol. Fresh tissuewas collected using standardized procedures with all speci-mens snap frozen in isopentane immediately after removal

and processing. Of the 23 tumors 15 were from the primary

Vol. 177, 1264-1268, April 2007Printed in U.S.A.

DOI:10.1016/j.juro.2006.11.087

INTERFERON-� RESISTANCE ASSOCIATED GENES IN RENAL CELL CARCINOMA 1265

site while the remaining 8 were from metastases. The clin-ical features of these patients and their clinical response areshown in the Appendix. Further description of these pa-tients is being prepared for a separate report.

RNA IsolationRNA was isolated from frozen tumor blocks as previouslydescribed.10 Briefly, tumors were grossly dissected to ex-clude normal tissue, then minced and homogenized. RNAwas purified through RNeasy columns (Qiagen, Valencia,California) according to manufacturer protocols. RNA quan-tity was assessed by ultraviolet absorbance, and RNA qual-ity was verified on denaturing agarose gels by the presenceof distinct 28S and 18S ribosomal bands. RNA from a patientwith no evidence of kidney cancer was also extracted as acontrol.

cRNA Labeling and HybridizationProbes for hybridization to Human Genome U133 Plus 2.0microarrays (Affymetrix, Santa Clara, California) were pre-pared as described previously.10 Briefly, 12 �g of total RNAwas converted into double stranded cDNA using a T7 mod-ified oligo dT primer and a ds cDNA synthesis kit (Invitro-gen, Valencia, California). Biotinylated cRNA probe wasproduced from the ds cDNA template using a HighYield™RNA transcript labeling kit (Enzo Biochem, Farmingdale,New York). The cRNA probe was quantitated by ultravioletabsorbance, and 15 �g were fragmented according to man-ufacturer protocols for hybridization to the U133 Plus 2.0microarrays. Hybridization, washing and imaging were per-formed according to manufacturer protocols.

RT-PCR AnalysisRT-PCR was done as described previously.10 Briefly, 1 �gof total RNA was reverse transcribed using random hex-amers (Invitrogen) and superscript II reverse transcrip-tase (Invitrogen) in 50 �l total volume. cDNA products(1 �l) were then amplified using 35 cycles of 95C melting(45 seconds), annealing (30 seconds) and 72C extension(45 seconds), with gene specific primers: Hs.94122, 52Cannealing, FWD: 5=-GGGCAATTTGTTGCTTTACAA-3=,REV: 5=-TCAACAGCAACGATATGACATT-3=; Hs.11325,56C annealing, FWD: 5=-CCCTGAATTTGGTTTGCAGT-3=, REV: 5=-AGCCTCGTAATGCAAAAAGC-3=; ACTB, 56Cannealing, FWD: 5=-ATCTGGCACCACACCTTCTAGAAT-GAGCTGCG-3, REV: 5=-CGTCATACTCCTGCTTGCTGATCCACATCTGC-3=. PCR products were visualized on 1%agarose gels.

Data Normalization and Statistical AnalysisRaw.cel files were normalized, background subtracted andlog transformed as described previously.10 Significantly dif-ferentially expressed genes were identified using SAM.11

Candidate genes were tested for significance using the t testwith Benjamini-Hochberg correction for multiple compari-sons within Bioconductor for R.12

RESULTS

Expression profiling using Affymetrix U133 Plus 2.0 micro-arrays, which consist of 54,627 probe sets, was performed on

23 renal cell carcinoma specimens from patients with met-

astatic disease who were subsequently treated with IFN.Eight patients showed partial response, 9 had stable diseaseas the best response and 6 had overt disease progressionwith treatment. The characteristics of the tumors that wereprofiled are shown in the Appendix.

Expression of Genes Known to be Involved in RCCA number of proteins have been previously implicated asbeing important in RCC, most notably VHL and its down-stream targets HIF-1� and HIF-2�.1 A number of geneshave been identified through expression profiling and otherstudies that are regulated in response to changes in VHL orHIF levels, including CA9, VEGF and CCND1.13,14 We ex-amined the expression levels of 32 genes (90 probe sets) thathave been implicated as target genes for correlation withresponse (progression of disease, stable disease and partialor moderate response). Unsupervised clustering of the tu-mor samples based on expression of these genes did notshow any separation on the basis of response to IFN (fig. 1).Statistical analysis for differential expression comparing pa-tients with partial response to those with stable disease andprogression of disease showed 2 significant probe sets beforecorrection for multiple comparisons (both GLS), with nosignificant probe sets following this correction. Similarly,comparison of patients with partial response or stable dis-ease to those with progression of disease yielded 1 signifi-cant probe set (GLS) before the multiple comparison correc-tion, but none following this correction.

Expression DifferencesCorrelated With Clinical ResponseWe next examined expression differences in the entire gene setfor correlations with response. An analysis of the full expres-sion profiles of patients for therapeutic benefit (patients withPR/SD vs those with POD) identified 3 significantly differen-tially expressed genes (SAM estimated False Discovery Rate0), consisting of Hs.76704 (Affymetrix probe set 226192_at),Hs.11325 (228046_at), and ABCD3 (202850_at). Furtheranalysis of just PR cases vs POD identified an additionalexpressed sequence tag [Hs.94122 (228423_at)] that wassignificantly differentially expressed (SAM estimated FalseDiscovery Rate 0.33). The average expression of each ofthese genes within PR, SD and POD groups relative tonormal kidney is shown in figure 2. All 4 shared a similarexpression pattern with higher levels in the PR group andlower levels in the POD group. To ensure that these differ-ences were not due to differences in primary compared tometastatic lesions, we compared expression of these 4 genesbetween the primary tumors and metastases. None of thegenes showed significant differences in expression betweenthese 2 groups.

Validation of Expression by RT-PCRTo validate the array results the 2 expressed sequence tagswere chosen for RT-PCR analysis. Hs.11325 and Hs.94122were analyzed within a subset of the tumor samples. Ofthese tumors 4 showed PR while the remaining 4 showedPOD. As shown in figure 3 RT-PCR analysis confirmed thedifferential expression observed with the microarrays, asboth genes were more highly expressed in patients with PR

compared to those with POD.

in black indicate patients with disease progression. Normal kidneysample is indicated in gray.

INTERFERON-� RESISTANCE ASSOCIATED GENES IN RENAL CELL CARCINOMA1266

DISCUSSION

The current options for treatment of metastatic RCC arelimited. Although new drugs such as sunitinib, sorafeniband bevacizumab offer great promise,3 the current standardtherapies such as IFN-� have responses in the 10% to 20%range.1,2 The use of genomic and expression profiling datapromises to improve patient diagnosis and treatment as wegain a better understanding of genetic alterations in cancer.Therapies to subsets of patients who aberrantly expresstarget genes have already improved patient outcome such aswith Herceptin® in ERBB2 positive breast cancers or ima-tinib in chronic myelogenous leukemia.15 Recent evidencesuggests that different chemotherapeutic regimens may alsobenefit patients who are stratified based on genetic alter-ations.16 Currently there is little known about the underly-ing genetic features that affect patient response to IFN-� inRCC.

We performed expression profiling on a panel of RCCswith known response to IFN with the goal of identifyinggenes that are associated with response. Using a candidategene approach we examined known VHL target genes todetermine if expression levels of these genes were associatedwith patient response. Clustering based on this set of 32genes did not result in separation of patients on the basis ofIFN response. Similarly a statistical analysis of these genesdid not identify any genes that had significant correlationsbetween expression and patient response.

FIG. 2. Average fold change in expression levels relative to normalkidney for genes significantly associated with progression after IFNtreatment. Error bars are SEM.

FIG. 3. RT-PCR analysis of Hs.94122 (228423_at) and Hs.11325

FIG. 1. Heat map showing clustering of tumor samples based onreported VHL responsive genes (green indicates low relative geneexpression, red indicates high relative gene expression). Tumorsamples in light blue indicate patients who had partial response,samples in orange indicate patients with stable disease and samples

(228046_at) validates differential gene expression in IFN respond-ers and nonresponders in patients with renal cell carcinoma.

INTERFERON-� RESISTANCE ASSOCIATED GENES IN RENAL CELL CARCINOMA 1267

From our global analysis of the approximately 54,000probe sets on the Affymetrix U133 Plus microarrays, wewere only able to identify 1 known and 3 uncharacterizedgenes that were associated with partial response status,none of which have been previously implicated in RCC. All 4genes, ABCD3, Hs.76704, Hs.11325 and Hs.94122, weremore highly expressed in patients who had a response afterIFN treatment compared to those in whom disease pro-gressed. We chose to validate the last 2 for several reasons.They showed the largest fold change in expression betweenthe POD and PR groups and progressive decrease from PR toSD to POD patients. Additionally, both mapped to 4q31-32.The entire 4q arm has previously been implicated as the siteof several putative tumor suppressor genes in multiple tu-mor types, including renal.17,18 Comparative genomic hy-bridization analysis of RCC has identified loss of the entire4q arm as a frequent event,18,19 occurring in up to 50% ofcases in some studies. Furthermore, oncogenetic tree anal-ysis has indicated that loss of 4q is an important early eventin RCC etiology.17 These 2 genes may represent putativecandidate genes for suppressor genes active in RCC andother cancers that frequently exhibit loss of 4q. Additionalwork will be required to determine if these genes are directlyinvolved in IFN-� resistance or are simply markers of tumoraggressiveness or tumors more likely to progress early.

Of the 2 remaining genes, ABCD3 (also known as PMP70)is the better characterized. This gene encodes an adenosinetriphosphate binding cassette transporter protein that isfound in peroxisomal membranes. Amplification of this genehas been implicated in drug resistance in cancer cell lines,20

although we observed lower levels of expression in patientswho had POD compared to those with PR or SD. Hs.76704 isuncharacterized and maps to the X chromosome.

CONCLUSIONS

We have identified 4 genes that show an association withpoor response to IFN treatment in patients with RCC. Thesegenes may prove useful in identifying patients with RCCwho would benefit from IFN-� based therapy. Finally, 2 ofthese genes may represent putative novel candidate tumorsuppressor genes on 4q.

ACKNOWLEDGMENTS

PEG-Intron was provided by Schering-Plough ResearchInstitute, Kenilworth, New Jersey. Hybridization, washingand imaging of the Affymetrix microarrays were performedat the Memorial Sloan-Kettering Cancer Center GenomicsCore facility.

APPENDIX

Clinical characteristics of RCC panel

Sample Disease Site Histology IFN Response

KC10 Abdominal wall Clear Cell PRKC11 Primary Clear Cell PRKC12 Primary Clear Cell SDKC13 Primary Clear Cell SDKC14 Primary Clear Cell PODKC15 Primary Clear Cell SDKC16 Parotid (Rt) Clear Cell SDKC17 Primary Clear Cell* PODKC18 Primary Clear Cell PRKC19 Lymph Node Unclassified SD

(appendix continued)

APPENDIX continued

Clinical characteristics of RCC panel

Sample Disease Site Histology IFN Response

KC2 Lung Metastasis Clear Cell PRKC20 Lung Metastasis Clear Cell PODKC21 Primary Clear Cell SDKC25 Primary Clear Cell SDKC26 Primary Clear Cell PODKC27 Primary Clear Cell PODKC29 Bone Metastasis Clear Cell PRKC3 Primary Clear Cell SDKC4 Primary Clear Cell PRKC6 Bone Metastasis Clear Cell SDKC7 Primary Clear Cell PODKC8 Primary Clear Cell PRKC9 Lung Metastasis Clear Cell* PR

* Tumors were clear cell with sarcomatoid features.

Abbreviations and Acronyms

IFN � interferonPOD � progression of disease

PR � partial responseRCC � renal cell carcinoma

RT-PCR � reverse transcriptase-polymerasechain reaction

SAM � Significance Analysis for MicroarraysSD � stable disease

REFERENCES

1. Linehan WM and Zbar B: Focus on kidney cancer. Cancer Cell2004; 6: 223.

2. Motzer RJ, Bander NH and Nanus DM: Renal-cell carcinoma.N Engl J Med 1996; 335: 865.

3. Patel PH, Chaganti RS and Motzer RJ: Targeted therapy formetastatic renal cell carcinoma. Br J Cancer 2006; 94: 614.

4. Motzer RJ, Rini BI, Bukowski RM, Curti BD, George DJ,Hudes GR et al: Sunitinib in patients with metastatic renalcell carcinoma. JAMA 2006; 295: 2516.

5. Motzer RJ, Rakhit A, Ginsberg M, Rittweger K, Vuky J, Yu Ret al: Phase I trial of 40-kd branched pegylated interferonalfa-2a for patients with advanced renal cell carcinoma.J Clin Oncol 2001; 19: 1312.

6. Bukowski RM, Tendler C, Cutler D, Rose E, Laughlin MM andStatkevich P: Treating cancer with PEG Intron: pharmaco-kinetic profile and dosing guidelines for an improved inter-feron-alpha-2b formulation. Cancer 2002; 95: 389.

7. Caraglia M, Marra M, Pelaia G, Maselli R, Caputi M, MarsicoSA et al: Alpha-interferon and its effects on signal trans-duction pathways. J Cell Physiol 2005; 202: 323.

8. Leaman DW, Chawla-Sarkar M, Jacobs B, Vyas K, Sun Y,Ozdemir A et al: Novel growth and death related interferon-stimulated genes (ISGs) in melanoma: greater potency ofIFN-beta compared with IFN-alpha2. J Interferon CytokineRes 2003; 23: 745.

9. Zimmer R and Thomas P: Expression profiling and interferon-beta regulation of liver metastases in colorectal cancer cells.Clin Exp Metastasis 2002; 19: 541.

10. Korkola JE, Houldsworth J, Dobrzynski D, Olshen AB, ReuterVE, Bosl GJ et al: Gene expression-based classification ofnonseminomatous male germ cell tumors. Oncogene 2005;24: 5101.

11. Tusher VG, Tibshirani R and Chu G: Significance analysis ofmicroarrays applied to the ionizing radiation response.Proc Natl Acad Sci U S A 2001; 98: 5116.

12. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M,

Dudoit S et al: Bioconductor: open software development for

INTERFERON-� RESISTANCE ASSOCIATED GENES IN RENAL CELL CARCINOMA1268

computational biology and bioinformatics. Genome Biol2004; 5: R80.

13. Maina EN, Morris MR, Zatyka M, Raval RR, Banks RE,Richards FM et al: Identification of novel VHL target genesand relationship to hypoxic response pathways. Oncogene2005; 24: 4549.

14. Zatyka M, da Silva NF, Clifford SC, Morris MR, Wiesener MS,Eckardt KU et al: Identification of cyclin D1 and othernovel targets for the von Hippel-Lindau tumor suppressorgene by expression array analysis and investigation of cy-clin D1 genotype as a modifier in von Hippel-Lindau dis-ease. Cancer Res 2002; 62: 3803.

15. Houshmand P and Zlotnik A: Targeting tumor cells. Curr OpinCell Biol 2003; 15: 640.

16. Pritchard KI, Shepherd LE, O’Malley FP, Andrulis IL, Tu D,Bramwell VH et al: HER2 and responsiveness of breast can-cer to adjuvant chemotherapy. N Engl J Med 2006; 354: 2103.

17. Jiang F, Desper R, Papadimitriou CH, Schaffer AA, KallioniemiOP, Richter J et al: Construction of evolutionary tree modelsfor renal cell carcinoma from comparative genomic hybrid-ization data. Cancer Res 2000; 60: 6503.

18. Sanjmyatav J, Schubert J and Junker K: Comparative study ofrenal cell carcinoma by CGH, multicolor-FISH and conven-tional cytogenic banding analysis. Oncol Rep 2005; 14:1183.

19. Moch H, Presti JC Jr, Sauter G, Buchholz N, Jordan P,Mihatsch MJ et al: Genetic aberrations detected by com-parative genomic hybridization are associated with clin-ical outcome in renal cell carcinoma. Cancer Res 1996;56: 27.

20. Yasui K, Mihara S, Zhao C, Okamoto H, Saito-Ohara F,Tomida A et al: Alteration in copy numbers of genes as amechanism for acquired drug resistance. Cancer Res 2004;64: 1403.

EDITORIAL COMMENT

With the excitement about new targeted agents in renal cellcarcinoma, it is important not to forget the small subset ofpatients who achieve significant tumor responses following

treatment with interferon-�. Prospective identification of

nonresponders would spare unnecessary toxicity and delayin receiving other potentially more effective agents. Thisstudy from Memorial Sloan-Kettering uses cDNA gene ex-pression profiling of 23 metastatic renal cell cancers to iden-tify a set of genes whose expression correlates to response tointerferon-�.

The fact that no differential expression of VHL-HIF path-way genes was found may reflect the technical limitations oftumor procurement. Many of the genes over expressed withVHL loss are artifactually induced within minutes of expo-sure to decreased oxygen tension, as is present after renalartery clamping.1 Thus, differential expression present un-der physiological conditions may have been masked.

Nonetheless, studies such as this are valuable not only todirect therapy but also to elucidate the mechanisms of anti-tumor effect and resistance. They suggest how tumor genet-ics can successfully influence tailored therapeutic regimensof the future.

Jodi K. MaranchieDepartment of Urology

University of Pittsburgh Medical CenterPittsburgh, Pennsylvania

1. Spruessel A, Steimann G, Jung M, Lee SA, Carr T, Fentz AK et al:Tissue ischemia time affects gene and protein expressionpatterns within minutes following surgical tumor excision.Biotechniques 2004; 36: 1030.

REPLY BY AUTHORS

Maranchie suggests that the reason we did not find anydifferential expression of VHL-HIF genes may be due tochanges in gene expression caused by renal artery clampingduring tumor procurement. We believe this is unlikely and,in fact, we did observe differential gene expression of VHL-HIF genes between tumors. However, there was no associa-tion between expression of VHL-HIF genes and patient re-

sponse.