Nuclear factor kappa B pathway associated biomarkersin AIDS defining malignancies

Juan-Carlos Ramos1, Sang-Hoon Sin2, Michelle R. Staudt2, Debasmita Roy2, Wolfgang Vahrson2, Bruce J. Dezube3,

William Harrington, Jr.1 and Dirk P. Dittmer2

1 The Viral Oncology Program, Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine, Miami, FL2 Department of Microbiology and Immunology, Lineberger Comprehensive Cancer Center, Center for AIDS Research (CfAR),

The University of North Carolina at Chapel Hill, Chapel Hill, NC3 Division of Hematology/Oncology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA

The nuclear factor kappa B (NFjB) pathway is essential for many human cancers. Therapeutics such as bortezomib (VelcadeTM)

that interfere with NFjB signaling are of great clinical interest. NFjB signaling, however, is multifaceted and variable among

tissues, developmental and disease entities. Hence, targeted biomarkers of NFjB pathways are of prime importance for clinical

research. We developed a novel real-time qPCR-based NFjB array. Only mechanistically validated NFjB targets were included. We

then used random-forest classification to define individual genes and gene combinations within the NFjB pathways that define

viral lymphoma subclasses as well as Kaposi sarcoma (KS). Few NFjB targets emerged that were universally present in all tumor

types tested, underscoring the need for additional tumor-type specific biomarker discovery. (i) We uncovered tissue of origin-

specific tumor markers, specifically CD69, CSF-1 and complement factor B (C1QBP) for primary effusion lymphoma (PEL); IL1-beta,

cyclinD3 and CD48 for KS. We found that IL12, jun-B, msx-1 and thrombospondin 2 were associated with EBV co-infection in PEL.

(ii) We defined the NFjB signature of Epstein–Barr virus (EBV) positive AIDS-associated Burkitt lymphoma (BL). This signature

identified CCR5 as the key marker. (iii) This signature differed from EBV negative BL consistent with the idea that EBV not only

activates NFjB activity but that this virus also reprograms NFjB signaling toward different targets.

Nuclear factor kappa B (NFjB) is a sequence-specific tran-scription factor. NFjB is crucial for the survival of B-celllymphoma. Further studies established its importance as aprosurvival factor in many cancers, not just those of lymph-oid origin (reviewed in Ref. 1). NFjB is a mixed multimer ofNFjB 1 (p50), NFjB 2 (p52), Rel A (p65), Rel B and c-rel.Subunit composition, posttranscriptional modifications andcooperating transcription factors determine target gene speci-ficity beyond recognition of the NFjB consensus motif.

Hence, mRNA profiling of NFjB responsive genes can beused to stratify malignancies and to characterize cellularresponses to natural (viral infection, growth-factor signaling)or artificial (chemotherapy drugs) stimuli.

Chemotherapy induces NFjB as part of the prosurvivalresponse to DNA damage.2 Several therapeutics directly tar-get NFjB or NFjB regulating factors. VelcadeTM/bortezomibinhibit the proteasome and thereby prevent degradation ofIkB. This inhibits NFjB signaling, resulting in cell death(though bortezomib’s clinical efficacy may be due to othereffects as well). Bay 11-7082 has emerged as an experimentalNFjB inhibitor, which causes rapid cell death in NFjB-de-pendent cancer models, including primary effusion lym-phoma (PEL)3 and Epstein–Barr virus (EBV)-infected lym-phoblastoid cell lines.4

Kaposi sarcoma (KS) is the leading cancer in people livingwith HIV/AIDS followed by forms of non-Hodgkin lym-phoma (NHL), including Burkitt lymphoma (BL). Hodgkinlymphoma (HL) is an AIDS-associated cancer that hasincreased in the post-HAART era.5 Other so-called commoncancers have also become more prevalent. These AIDS-asso-ciated lymphomas and KS are associated with gamma-herpes-virus infection. Gamma-herpesvirus infection, reactivation(and subsequent replication) is regulated by NFjB.6–12 Thisrelationship seems almost self-evident, because gamma her-pesviruses with the exception of herpesvirus saimiri establishlifelong latency in B cells, for which NFjB is the central reg-ulator of survival and differentiation.

Key words: Kaposi sarcoma, EBV, KSHV, Burkitt lymphoma, random

forest, tree, real-time qPCR, ATL

Additional Supporting Information may be found in the online

version of this article.

Grant sponsors: AIDS Malignancy Clinical Trials Consortium

(AMC), The University Cancer Research Grant (UCRF), The Center

for AIDS Research (CfAR) and The Leukemia and Lymphoma

Society; Grant sponsor: NIH; Grant numbers: CA112217,

CA121935, DE018304, CA019014

DOI: 10.1002/ijc.26302

History: Received 22 Feb 2011; Accepted 31 May 2011; Online 25

Jul 2011

Correspondence to: Dirk P. Dittmer, Department of Microbiology

and Immunology, Lineberger Comprehensive Cancer Center, Center

for AIDS Research (CfAR), The University of North Carolina at

Chapel Hil, USA, Tel.: +919-966-7960; Fax: +919-962-8103,

E-mail: ddittmer@med.unc.edu

ShortReport

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC

International Journal of Cancer

IJC

ResultsNFjB signaling in KS

To investigate nuclear factor kappa B (NFjB) signaling, wedeveloped a targeted real-time qPCR array (Supporting Infor-mation Table 1). We obtained validated NFjB target genesthrough literature review. We designed primer pairs, suchthat these share a common melting temperature, similaramplicon length and amplification efficiency.

To test the hypothesis that there exist common NFjBresponsive genes among AIDS defining cancers, we per-formed unsupervised clustering (Fig. 1). We included twoAIDS-associated post germinal center (GC) NHL, namely BL(n ¼ 10) and PEL (n ¼ 4 in biological replicates). Weincluded primary KS biopsies (n ¼ 8) and KSHV-infectedcell lines (n ¼ 2) as a gamma herpesvirus-dependent nonhe-matological cancer. We included non-HIV associated adultT-cell leukemia (ATL) (n ¼ 17) and AIDS-associated, virus-

negative, NHL (n ¼ 4) as controls. In addition, we includedreplicates of nontemplate control (ntc, n ¼ 5). The differentAIDS-tumor types segregated according to tissue of origin.

This demonstrates that profiling for just NFjB responsivegenes sufficed to classify these tumor types. All KS biopsies,which are of endothelial lineage,13–15 clustered together. AllPEL, which are of post GC B-cell lineage,16,17 clustered to-gether. All ATL, which are of T-cell lineage, clustered together;although we identified two subtypes (labeled ATL and A’ inFig. 1). Two independently derived, KSHV-infected cell lines18

clustered together (labeled T in Fig. 1), but separate from KSbiopsies. One ATL sample (labeled ‘‘o’’ in Fig. 1) appeared asoutlier and was removed from subsequent analysis.

We identified two novel subtypes of BL: those thatco-clustered with other NHL as well as a HL case, i.e., othervirus negative B-cell lymphomas (middle of Fig. 1, labeledBL:NHL) and those that clustered separately (left cluster in

Figure 1. (a) Heat map representation of unsupervised clustering of all study data. Normalized (dCT) and median-centered relative mRNA

levels were clustered by gene and by sample. ‘‘Blue’’ indicates reduction and ‘‘Red’’ indicates induction relative to the median for each

sample. (a) and (b) indicate the two main clusters of mRNAs. The clusters of samples matched their presumptive origin: BL, Burkitt

lymphoma; ATL, Adult T cell leukemia; A0, a novel subtype of ATL; BL:NHL, a novel subtype of BL, which clustered with non-BL NHL

samples; KS, Kaposi sarcoma biopsies; T, KS-like cell lines; PEL, primary effusion lymphoma; ntc, nontemplate control; o, outlier ATL

sample. Relative scale is indicated on the top left (The data points for the ntc appear white in Fig. 1a, because the normalized data were

median-centered by sample).

ShortReport

Ramos et al. 2729

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC

Fig. 1, labeled BL). The latter represent the EBV þ BL sam-ples. BL group membership in these subtypes correlated per-fectly with EBV status, even though EBV mRNAs levels werenot included in the unsupervised clustering. This demon-strates that the NFjB pathway in BL differs based upon thepresence of EBV.

Next, we inspected individual genes. One group of geneswas highly transcribed in all tumor classes, B-cell lymphoma,T-cell lymphoma and endothelial lineage KS alike (Fig. 1,cluster a.). The least variable were SelectinP/CD62, FasL,IL11, IL6, CSF3, prostacyclin synthase, MMP9, tenascin-C,MIP-2c, mad-3, PDGF-b and VCAM-1. Others in cluster ashowed variation within either the PEL (serpinA1 and Rel)or the ATL and KS (Lamb2, TGM1 and VEGF-C) classes.The remainder exhibited tumor type and sample specific var-iation (Fig. 1, cluster b). Several mRNA were highly tran-scribed in KS biopsies, but not exclusively so: CyclinD3,CD48, IGF-BP1, IL1-b, proenkephalin and TAP1. Conversely,several mRNAs were highly transcribed in PEL and others,but not KS: CD69-early activation antigen, CSF-1-colonystimulating factor 1, CFB-complement factor B, which wasexpected as these are lymphoid-associated NFB target genes.

We performed extensive quality control assessments. Sup-porting Information Figure S1 (panel A) shows the distribu-tion of CT levels, i.e., the distribution of mRNA levels on alog 2 scale across all samples and all genes. Undetectable orabsent mRNAs account for �50% of the data (single bar atCT ¼ 40). This is consistent with the idea that many NFjBtarget genes are tissue specific and tightly regulated. They aretranscriptionally silent in one tumor type and induced inanother. Our targeted array design enriched for this type ofmRNA, because many of the mRNAs in our array were ini-tially identified by less sensitive, semiquantitative methods,such as Northern-blot.

The remaining mRNAs varied broadly in their level oftranscription. On the basis of the density distribution, weestimate a range of �2(40–20) ¼ 210 or three orders of magni-tude. This large range suggests that these mRNAs are highlyvariable and thus may be developed into useful diagnosticmarkers. Supporting Information Figure S1 (panel B) showsthe distribution of CT levels for the n ¼ 470 individual non-template control reactions. With the exception of very fewoutliers, we did not detect any signal in the absence of tem-plate [mean ¼ 39.81, median ¼ 40.0, with a median absolutedeviation (m.a.d.) of 0.00 cycle units].

To investigate the levels of these candidate mRNAsbetween grossly different tumor classes, we conducted pair-wise comparisons for the differentially transcribed mRNAs asidentified by unsupervised clustering. We ranked the mRNAsby their differential expression (significance after adjustmentfor false discovery rate). For this and subsequent analyses, weremoved ntc, the outlier ‘‘o’’ and the single class samplesVG1, ATL#31, DLBCL and BL#3. KS, an endothelial tumor,was easily distinguished from the lymphomas by any one ofthree single mRNAs: IL1-beta, cyclinD3 and CD48 were pres-ent at higher levels in KS than in any of the other tumors(Figs. 2a–2c). These three markers were virtually absent inPEL (q < 0.001). PEL stood out on the basis of high-levelexpression for CD69, complement factor 3 and CSF-1(q < 0.001), which were absent from the other samples(Figs. 2d–2e).

Figure 2. Supervised class comparison of study data represented

as Box plots. The box covers the 25–75 percentile of the data,

that is, the interquartile range (IQR). The horizontal line indicates

the median. Whiskers demarcate the minimum and maximum of

the data. Open circles denote outliers (1.5 * IQR from the first and

third quartiles). (a–c) shown are the top most discriminating

mRNAs that distinguish KSHV-associated KS from AIDS-associated

lymphoma. (d–f) shown are the top most discriminating mRNAs

that distinguish PEL from AIDS-associated lymphoma and ATL. For

these markers p < 0.001 after adjustment for false-discovery rate.

Relative levels on a log 10 scale for the indicated mRNAs are

shown on the vertical and classes on the horizontal axis.

ShortReport

2730 NFjB Pathway-Associated Biomarkers

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC

NFjB pathway profiling distinguishes AIDS-associated

lymphoma

The result of our unsupervised analyses was that we coulddistinguish PEL for BL and that we could establish subsets ofBL and subsets of ATL on the basis of target nuclear factorkappa B (NFjB) responsive gene profiling. We used super-vised, pair-wise comparison to establish significance levels foreach gene in the array. Figure 3a plots the p value distribu-tion for each of the three comparisons: PEL to BL (purpleline), the two novel BL subsets to each other (blue line) andthe two novel ATL subsets to each other (green line). Thedistribution for each comparison set was bi-modal, i.e., someof the genes showed a significant difference in gene expres-sion [lower log(p-value) or left peak], others did not [log(p-value) centered around 0 or right peak]. As expected, the dif-ferences between the two histological variant B-cell lympho-mas PEL and BL were much greater [log(p-value) for the leftpeak at (�5 . . . �10] compared to differences within the BLor the ATL subsets. It is important to remember that eachtumor within the BL and ATL subsets carried the same histo-logical diagnosis, and these subsets could only be identifiedafter profiling. The exception is BL. BL can also be subdi-vided based on EBV status. This analysis thus identified a setof novel biomarkers, which differed between subsets of BLand a set of biomarkers, which differed between two novelsubsets of ATL. Conversely, this supervised analysis demon-strates that there exist two different subtypes of BL and twodifferent subtypes of ATL.

Next, we identified individual biomarkers that can differ-entiate between BL and ATL subtypes. Again, we startedwith the easy comparison that between the histologically dif-ferent PEL and BL. After adjustment for multiple compari-sons,19 �40 genes were differentially transcribed (Fig. 3,panel PEL). Their unadjusted p-values were <0.00001. Wecalculated a false positive rate of <1. This suggests that anyone of these mRNAs can be used to molecularly differenti-ate PEL from BL. As an alternative method, we used ran-dom forest classification (reviewed in Ref. 20). This identi-fied BcL-xl, Tap1, CSF-1, CSF-2, FasL and complementfactor B as the most robust classifiers to distinguish BLfrom PEL

Approximately 20 genes in our array were differentiallytranscribed between the two subsets of BL as identified byunsupervised clustering (Fig. 3, panel BL). Their unadjustedp-values were < 0.001. As expected, subsets of BL were moreclosely related to each other than BL to PEL. Cox-2, TNFaand CCR5 were the most robustly differentially transcribedmRNAs between the two BL subsets. High level of thesethree mRNAs correlated with the presence of EBV. This sug-gests on one hand that these three mRNAs can be used tomolecularly differentiate EBVþ BL from EBV� BL. On theother hand, if one wanted to test the efficacy of an NFjB in-hibitor for BL, these would be good biomarkers only forEBVþ BL, not for EBV� BL.

Figure 3. Tree-based classification of PEL, BL subsets and ATL

subsets. (a) Density plot of p values for each individual mRNA in the

array that result from two-sample nonparametric comparison of PEL

to all other lymphomas (PEL, purple), the two subsets of BL (BL,

blue) and the two subsets of ATL (ATL, green). (b) Corresponding all

other NHL diagnostic plots for q-value calculation. For each

comparison, the left panel plots the number of significant tests, that

is, genes at a given q-value cut-off level. Note that for PEL � 40

genes were significant to q � 0.02, for BL and ATL subsets only �20.

For each comparison, the right panel plots the expected number of

false-positive markers among the top most significant tests on the

vertical and the number of significant tests on the horizontal axis.

Note that for PEL < 1 false positive is expected within the top 40

most differentially regulated genes, for BL subsets three to four false

positives are expected and for ATL subsets also < 1.

ShortReport

Ramos et al. 2731

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC

Approximately 20 genes in our array were differentiallytranscribed between the two subsets of ATL as identified byunsupervised clustering (Fig. 3, panel ATL). Their unadjustedp-values were not significant, suggesting that the heterogene-ity among ATL cases was not very well captured in a targetedarray based on NFjB signaling.

Because PEL can either be singly infected with KSHValone or dually infected with KSHV and EBV, we were inter-ested whether any NFjB-dependent genes were associatedwith EBV infection.

DiscussionAIDS-associated malignancies are in dire need of newtherapies. Because their frequency in the general popula-tion is comparably low and because patients who developAIDS-associated malignancies are often too sick to partici-pate in experimental treatments, clinical trials in AIDS-associated malignancies are faced with low enrollment.How then, can we nevertheless obtain clinical data thatsupport the efficacy of novel therapeutics? This is a generalobstacle faced by all who study disease that affect marginalor especially vulnerable populations. One solution is theuse of biomarkers to support studies of drug action. Here,we used a targeted approach to identify biomarkers thatare relevant for AIDS-associated malignancies and that canbe developed into biomarkers for the evaluation of drugsthat affect the NFjB pathway. We focused on the NFjBpathway, because it is central to gamma-herpesvirus-associ-ated cancers, which in turn are overrepresented amongAIDS patients.21

We identified CD69, CSF-1 and complement factor B(C1QBP) as PEL specific biomarkers that are known down-stream targets of NFjB. We identified IL1-b, cyclin D3 andCD48 as NFjB-regulated KS specific biomarkers. TheseNFjB targets were not expressed in BL or ATL.

We obtained evidence in support of two molecular sub-types of BL, which can be differentiated on the basis of theirEBV status, and we determined that Cox-2, TNFa and CCR5were robustly expressed only in EBV positive BL. This cor-roborates a recent study by Piccaluga et al.,22 who were ableto distinguish BL from other NHL, but more importantlywho also distinguished three types of BL—endemic, sponta-neous and HIV-associated—on the basis of mRNA profiling.Furthermore, this study found that genes in the TNF/NFjBpathway were overrepresented among the mRNAs that differ-entiated among BL subsets. This underscores the utility ofpathway-targeted profiling if biological insights or larger scaledata suggest particular biological pathways.

We obtained evidence in support of molecular subtypes ofATL that were not linked to histological stage. However, wedid not have the statistical power to identify any one specificbiomarker in our targeted arrays, which could differentiatebetween these two subtypes.

Lymphomas have been the subject of innumerous profil-ing studies. Studies on AIDS-associated lymphomas or KS

are rarer.13,14,16,17,23 Some of the new biomarkers that we dis-covered here were previously investigated: C1QBP had beendetected in PEL, but the other two markers (CD69 and CSF-1) were not previously associated with PEL. CD69 was origi-nally identified as a T-cell activation marker, but is alsohighly transcribed in follicular lymphoma.24 This also holdstrue for CSF-1 and C1QBP. In regard to KS, profiling dataare even more limited. Cyclin D3 is a bonafide endothelialcell marker, expression of which correlates with CD31 levels.CD48 is expressed in endothelial cells as well as leukocytesand can be upregulated by LPS (via NFjB) and VEGF-1.

Cox-2, TNFa and CCR5 were the most robustly differen-tially transcribed NFjB targets in our two BL subsets. HighermRNA levels correlated with the presence of EBV. In PEL,high levels of IL12, jun-B, msx-1 and thrombospondin 2were best correlated with the presence of EBV. This suggestsan important generalization, whereas EBV modulates NFjBsignaling, it is the tumor tissue of origin that restricts whichNFjB target gene repertoire that is open to regulation by thevirus.

Some NFjB target genes emerged that were universallypresent and that could be used to follow drug effects in alarge variety of different cancers. This is not entirely unex-pected and underscores the importance of multianalyte assaysin cancer research. Unfortunately, this would also suggestthat single biomarker responses cannot be compared betweentumor types or even between lymphoma subtypes.

These biomarkers are clinically relevant, because many agentsthat target NFjB in cancer are in advanced clinical testing. Thesebiomarkers are robust, because they can be measured individu-ally by real-time qPCR. We hope that they facilitate the testingof novel NFjB inhibitors for AIDS-associated malignancies.

Material and MethodsSample processing

Frozen biopsy samples were obtained under informed con-sent at the participating institutions. Diagnosis was confirmedby pathology review and de-identified samples were used foranalysis. RNA was isolated following our published meth-ods25 and quality established using an Agilent Bioanalyzer(Supporting Information Fig. S2).

Analysis

Multiple housekeeping genes were included in the qPCRarray. If >50% of housekeeping genes failed to yield a signal,the entire sample was excluded from further analysis. If<50% of housekeeping genes failed, the individual datapoints were omitted from further analysis. Next, the medianCT of all housekeeping genes was calculated for each sampleand subtracted from the CT for all other genes (dCT ¼CTgene � medianCThouse). The dCT data were standardized bysample and subjected to unsupervised clustering. We dividedthe ATL and BL each into two subsets according to Figure 1and rounded our real-time qPCR data (dCT) to the nearestinteger. We next calculated individual Wilcox-rank tests with

ShortReport

2732 NFjB Pathway-Associated Biomarkers

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC

continuity correction as implemented in R version 2.8.0 foreach gene in the array between KS and all non-KS samples.We adjusted for multiple comparison using q-value.19 This

yielded 10 genes with q � 0.001. We used random forestclassifiers to identify the most robust markers for eachsubset.20

References

1. Ghosh S. Handbook of transcription factorNF-jB. Boca Raton: CRC Press, 2007.

2. Elkon R, Rashi-Elkeles S, Lerenthal Y,Linhart C, Tenne T, Amariglio N, RechaviG, Shamir R, Shiloh Y. Dissection of aDNA-damage-induced transcriptionalnetwork using a combination ofmicroarrays, RNA interference andcomputational promoter analysis. GenomeBiol 2005;6:R43.

3. Keller SA, Schattner EJ, Cesarman E.Inhibition of NF-jB induces apoptosis ofKSHV-infected primary effusionlymphoma cells. Blood 2000;96:2537–42.

4. Cahir-McFarland ED, Davidson DM,Schauer SL, Duong J, Kieff E. NF-jBinhibition causes spontaneous apoptosis inEpstein-Barr virus-transformedlymphoblastoid cells. Proc Natl Acad SciUSA 2000;97:6055–60.

5. Engels EA, Biggar RJ, Hall HI, Cross H,Crutchfield A, Finch JL, Grigg R, HyltonT, Pawlish KS, McNeel TS, Goedert JJ.Cancer risk in people infected withhuman immunodeficiency virus in theUnited States. Int J Cancer 2008;123:187–94.

6. Krug LT, Moser JM, Dickerson SM, SpeckSH. Inhibition of NF-jB activation in vivoimpairs establishment ofgammaherpesvirus latency. PLoS Pathog2007;3:e11.

7. Grossmann C, Podgrabinska S, Skobe M,Ganem D. Activation of NF-jB by thelatent vFLIP gene of Kaposi’s sarcoma-associated herpesvirus is required for thespindle shape of virus-infected endothelialcells and contributes to theirproinflammatory phenotype. J Virol 2006;80:7179–85.

8. Chugh P, Matta H, Schamus S, ZachariahS, Kumar A, Richardson JA, Smith AL,Chaudhary PM. Constitutive NF-jBactivation, normal Fas-induced apoptosisand increased incidence of lymphoma inhuman herpes virus 8 K13 transgenic mice.Proc Natl Acad Sci USA 2005;102:12885–90.

9. Guasparri I, Keller SA, Cesarman E. KSHVvFLIP is essential for the survival ofinfected lymphoma cells. J Exp Med 2004;199:993–1003.

10. Brown HJ, Song MJ, Deng H, Wu TT,Cheng G, Sun R. NF-jB inhibitsgammaherpesvirus lytic replication. J Virol2003;77:8532–40.

11. Cahir-McFarland ED, Carter K, RosenwaldA, Giltnane JM, Henrickson SE, StaudtLM, Kieff E. Role of NF-jB in cell survivaland transcription of latent membraneprotein 1-expressing or Epstein-Barr viruslatency III-infected cells. J Virol 2004;78:4108–19.

12. Uchida J, Yasui T, Takaoka-Shichijo Y,Muraoka M, Kulwichit W, Raab-Traub N,Kikutani H. Mimicry of CD40 signals byEpstein-Barr virus LMP1 in B lymphocyteresponses. Science 1999;286:300–3.

13. Hong YK, Foreman K, Shin JW, HirakawaS, Curry CL, Sage DR, Libermann T,Dezube BJ, Fingeroth JD, Detmar M.Lymphatic reprogramming of bloodvascular endothelium by Kaposi sarcoma-associated herpesvirus. Nat Genet 2004;36:683–5.

14. Wang HW, Trotter MW, Lagos D,Bourboulia D, Henderson S, Makinen T,Elliman S, Flanagan AM, Alitalo K,Boshoff C. Kaposi sarcoma herpesvirus-induced cellular reprogrammingcontributes to the lymphatic endothelialgene expression in Kaposi sarcoma. NatGenet 2004;36:687–93.

15. Sturzl M, Brandstetter H, Roth WK.Kaposi’s sarcoma: a review of geneexpression and ultrasructure of KS spindlecells in vivo. AIDS Res Hum Retroviruses1992;8:1753–62.

16. Klein U, Gloghini A, Gaidano G,Chadburn A, Cesarman E, Dalla-Favera R,Carbone A. Gene expression profileanalysis of AIDS-related primaryeffusion lymphoma (PEL) suggests aplasmablastic derivation and identifiesPEL-specific transcripts. Blood 2003;101:4115–21.

17. Jenner RG, Maillard K, Cattini N, WeissRA, Boshoff C, Wooster R, Kellam P.Kaposi’s sarcoma-associated herpesvirus-infected primary effusion lymphoma has aplasma cell gene expression profile. ProcNatl Acad Sci USA 2003;100:10399–404.

18. An FQ, Folarin HM, Compitello N, Roth J,Gerson SL, McCrae KR, Fakhari FD,Dittmer DP, Renne R. Long-term-infectedtelomerase-immortalized endothelial cells: amodel for Kaposi’s sarcoma-associatedherpesvirus latency in vitro and in vivo.J Virol 2006;80:4833–46.

19. Storey JD, Tibshirani R. Statisticalsignificance for genomewide studies. ProcNatl Acad Sci USA 2003;100:9440–5.

20. Fielding A. Cluster and classificationtechniques for the biosciences. Cambridge:Cambridge University Press, 2007.

21. Bouvard V, Baan R, Straif K, Grosse Y,Secretan B, El Ghissassi F, Benbrahim-Tallaa L, Guha N, Freeman C, Galichet L,Cogliano V. A review of humancarcinogens, Part B: biological agents.Lancet Oncol 2009;10:321–2.

22. Piccaluga PP, De Falco G, Kustagi M,Gazzola A, Agostinelli C, Tripodo C,Leucci E, Onnis A, Astolfi A, Sapienza MR,Bellan C, Lazzi S, et al. Gene expressionanalysis uncovers similarity and differencesamong Burkitt lymphoma subtypes. Blood2011;117:3596–608.

23. Fan W, Bubman D, Chadburn A,Harrington WJ, Jr, Cesarman E, KnowlesDM. Distinct subsets of primary effusionlymphoma can be identified based on theircellular gene expression profile and viralassociation. J Virol 2005;79:1244–51.

24. Bohen SP, Troyanskaya OG, Alter O,Warnke R, Botstein D, Brown PO, Levy R.Variation in gene expression patterns infollicular lymphoma and the response torituximab. Proc Natl Acad Sci USA 2003;100:1926–30.

25. Papin J, Vahrson W, Hines-Boykin R,Dittmer DP. Real-time quantitative PCRanalysis of viral transcription. Methods MolBiol 2004;292:449–80. Sh

ortReport

Ramos et al. 2733

Int. J. Cancer: 130, 2728–2733 (2012) VC 2011 UICC