Share this article on:

DNA methylation profiling can classify HIV-associated lymphomas

Matsunaga, Akihiroa; Hishima, Tsunekazub; Tanaka, Norikoc; Yamasaki, Mariac,d; Yoshida, Luia,e; Mochizuki, Makotof,i; Tanuma, Junkog; Oka, Shinichig; Ishizaka, Yukihitoa; Shimura, Maria; Hagiwara, Shotaroh

doi: 10.1097/QAD.0000000000000120
Clinical Science

Background: HIV-positive patients have a 60-fold to 200-fold increased incidence of non-Hodgkin lymphomas, including Burkitt lymphoma, diffuse large B-cell lymphoma, and primary central nervous system lymphoma. HIV-associated lymphomas frequently have features such as extranodal involvement, decreased responses to standard chemotherapy, and high relapse rates, which indicate a poor prognosis. General pathological features do not clearly differentiate HIV-associated lymphomas from non-HIV lymphomas.

Methods: To investigate the features of HIV-associated lymphomas, we performed genome-wide DNA methylation profiling of HIV and non-HIV lymphomas using Illumina GoldenGate Methylation Cancer Panel I and Illumina Infinium HumanMethylation450 BeadChip microarrays. DNA methylation profiles in HIV-associated and non-HIV lymphomas were characterized using unsupervised hierarchical clustering analyses.

Results: The analyses of promoter regions revealed unique DNA methylation profiles in HIV-associated lymphomas, suggesting profile differences compared with non-HIV lymphomas, which implies specific gene regulation in HIV-associated lymphoma involving DNA methylation. Based on HumanMethylation450 BeadChip data, 2541 target sites were selected as differing significantly in comparisons between HIV-associated and non-HIV-associated lymphomas using Wilcoxon's rank-sum test (P <0.05) and Δβ values more than 0.30. Recurrent cases of HIV-associated lymphoma had different profiles compared with nonrecurrent HIV lymphomas.

Conclusion: DNA methylation profiling indicated that 2541 target sites differed significantly in HIV-associated lymphoma, which may partly explain the poor prognosis. Our data indicate that the methylation profiles of target genes have potential in elucidating HIV-associated lymphomagenesis and can serve as new prognostic markers.

Supplemental Digital Content is available in the text

aDepartment of Intractable Diseases, Research Institute, National Center for Global Health and Medicine, Shinjuku

bDepartment of Pathology, Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital, Honkomagome, Bunkyo

cBiostatistics Section, Department of Clinical Research and Informatics, Clinical Research Center, National Center for Global Health and Medicine, Shinjuku

dDepartment of Human Genetic, School of Medicine, The University of Tokyo, Bunkyo, Tokyo

eGraduate School of Frontier Sciences, The University of Tokyo, Kashiwa, Chiba

fDepartment of Pathology

gAIDS Clinical Center

hDivision of Hematology, Internal Medicine, National Center for Global Health and Medicine Hospital, Shinjuku

iDepartment of Pathology, School of Medicine, Kyorin University, Mitaka, Tokyo, Japan.

Correspondence to Dr Mari Shimura, Department of Intractable Diseases, Research Institute, National Center for Global Health and Medicine, 1-21-1 Toyama, Shinjuku, Tokyo 162-8655, Japan. Tel: +81 3 32027181; fax: +81 3 32027364; e-mail:

Received 9 July, 2013

Revised 18 October, 2013

Accepted 18 October, 2013

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website (

Back to Top | Article Outline


The incidence of non-Hodgkin's lymphoma is 60-fold to 200-fold higher in patients with HIV infection [1,2]. Most HIV-associated lymphomas are high-grade B-cell lymphomas such as diffuse large B-cell lymphoma, Burkitt lymphoma, and primary central nervous system lymphoma. The clinical course is often aggressive, with a poor prognosis [2]. Since the introduction of highly active antiretroviral therapy, the risk for opportunistic infections and the incidence of AIDS-defining malignancies, including HIV-associated lymphomas, have declined, and prognoses have improved. Nevertheless, lymphomas remain a major cause of death for HIV-infected patients [3]. It is important to identify differences between HIV-associated lymphomas and non-HIV lymphomas, as their clinical and general pathological features do not clearly distinguish them [2]. Recent studies have revealed that the DNA methylation patterns can differentiate among disease subtypes, suggesting that epigenetic DNA alterations are related to carcinogenesis [4,5]. Epigenetic silencing of functionally important genes may contribute to the development of lymphomas [5,6], and promoter hypermethylation of CpG islands (CGIs) in some genes has been reported in aggressive-phenotype lymphoma with a poor prognosis [7]. In this study, we examined DNA methylation of CGIs in a promoter region clustered with HIV-associated lymphomas and non-HIV lymphomas, and investigated the prognostic significance of DNA methylation. Our findings contribute to an understanding of the lymphomagenesis of HIV-associated lymphomas and suggest specific DNA methylation as a useful prognostic biomarker.

Back to Top | Article Outline



HIV-associated lymphoma is a pathologically diagnosed malignant lymphoma in HIV patients. Two cohorts were studied. Cohort I consisted of 11 HIV-associated and 18 non-HIV lymphoma patients who visited Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital (CICK), and two non-HIV lymphoma patients who visited the National Center for Global Health and Medicine Hospital (NCGM). Cohort II included nine HIV-associated and 12 non-HIV lymphoma patients who visited NCGM. Formalin-fixed, paraffin-embedded tissues and fresh-frozen tissues were collected from NCGM and CICK, following approval by the ethics committees of both hospitals and in accordance with the Declaration of Helsinki. All patients gave written informed consent for their tissue to be used and for review of their clinical records. Diagnosis was made using the 2008 WHO classification [2]. Hematologists reviewed the tumor specimens and classified them histologically as diffuse large B-cell lymphoma, Burkitt lymphoma, primary central nervous system lymphoma, follicular lymphoma, or Hodgkin's lymphoma. Non-HIV lymphoma samples were randomly selected from among the Burkitt lymphomas, diffuse large B-cell lymphomas, follicular lymphoma, and Hodgkin's lymphoma. Epstein–Barr virus (EBV) status was determined by Epstein–Barr encoded RNA (EBER) in situ hybridization and Southern blotting. BCL2 expression was examined by immunostaining.

Back to Top | Article Outline

HumanMethylation450 microarray analysis

Cohort I was analyzed using an Infinium HumanMethylation450 BeadChip microarray [8], which covered 485 577 methylation sites. Genomic DNA was isolated using a DNeasy mini kit (QIAGEN, Valencia, California, USA) according to the manufacturer's protocol. After 1 μg of DNA was ligated at 24°C for 30 min, the reaction was stopped by 5 min at 95°C (REPLI-g FFPE kit; QIAGEN) [9]. The DNA was subjected to genome-wide DNA methylation profiling using an Infinium HumanMethylation450 BeadChip (Illumina, San Diego, California, USA) [8], according to the manufacturer's instructions. The methylation status of specific cytosines is indicated by the β value, with 1 indicating complete methylation and 0 indicating no methylation. We first filtered the probes and samples using the Bioconductor IMA package to load files created by Illumina GenomeStudio software, using the IMA.methy450R function. With this package, we performed filtering steps using the IMA.methy450PP function. The inclusion criteria were as follows: sample call rate, more than 99.5%; detection P value, <0.05; site call rate, more than 90%; probes with no SNPs based on snpsite.txt provided in the IMA package [10]; and probes outside the XY chromosomes. We converted the initial file created by Illumina GenomeStudio to a new file to reflect the filtering results. The data were normalized by entering the filtered data into the Bioconductor lumi package [11]. Using the lumi package, methylation data were first analyzed by the color balance check and then scaled based on the mean of all probes, using methylation simple scaling normalization (SSN) implemented in the lumi package. The Infinium array methylation data are available in the Gene Expression Omnibus database under the accession number GSE42372.

Back to Top | Article Outline

Cancer Panel I microarray analysis

Cohort II was analyzed using the Illumina GoldenGate Methylation Cancer Panel I microarray, a cancer-focused methylation analysis covering 1505 CpG loci from 807 genes (Illumina) [12]. Genomic DNA was isolated (Agencourt FormaPure kit; Beckman Coulter, Brea, California, USA), subjected to sodium bisulfite conversion, labeled with fluorescent dyes, and hybridized to the microarrays according to the manufacturer's protocol. The methylation status of specific cytosines was indicated by the β value (1, complete methylation; 0, no methylation). Only probes with detection P value at <0.01 were used for the analyses. The X chromosome loci were removed from the analysis, leaving 1421 CpG loci. Raw average β values were not normalized and were used for analyses as per the manufacturer's recommendations. The GoldenGate array methylation data are available in the Gene Expression Omnibus database under the accession number GSE42626.

For the statistical analysis, enrichment analysis of target genes, validation by combined bisulfite restriction analysis (COBRA), and bisulfite DNA sequences, see the Supplementary Methods,

Back to Top | Article Outline


To identify differences between HIV-associated and non-HIV lymphomas, genome-wide DNA methylation array analyses were performed using Infinium HumanMethylation450 BeadChip technology. DNA from formalin-fixed and paraffin-embedded or fresh-frozen lymphoma tissues collected from the 11 HIV-positive and 20 HIV-negative Asian patients in Cohort I was analyzed (Table 1). DNA methylation throughout the genome was examined using probes targeting six gene regions (Fig. 1a): within 1500 bps of a transcription start site (TSS1500), within 200 bps of a transcription start site (TSS200), and the 5′ untranslated region (5′UTR), first exon (1stExon), body, and 3′ untranslated region (3′UTR) and intergenic regions. Three HIV-negative lymphomas were excluded from the analyses in the filtering steps (see Methods for details). The differences in methylation status between HIV-associated and non-HIV lymphomas were significantly greater for CGIs in the various target regions, compared with non-CGI methylation (Supplementary Fig. 1, Hierarchical clustering analysis of CGI methylation markers of TSS1500, TSS200, 5′UTR, and 1stExon (Fig. 1b) produced roughly two groups that distinguished HIV-associated lymphomas from non-HIV lymphomas (Groups 1 and 2; Fig. 1b, upper left), with a few exceptions. By contrast, the analysis of non-CGI methylation and CGI methylation in the body and 3′UTR and intergenic gene targets did not give clear groupings (Fig. 1b, upper right and lower images, Supplementary Fig. 2, As all HIV patients in this study were men (Table 1), we next analyzed male patients only. The CGI results for TSS1500, TSS200, 5′UTR, and 1stExon again clustered into two groups (Supplementary Fig. 3,, suggesting that gender does not affect the results. Generally, patients with HIV-associated lymphomas were younger than patients with non-HIV lymphomas (Table 1) [13]. When we excluded age-related target sites, as previously suggested [14], the analysis of CGI methylation in TSS1500, TSS200, 5′UTR, and 1stExon again produced two groups that distinguished between HIV-associated and non-HIV lymphomas (Supplementary Fig. 4, These results suggest that DNA methylation of CGIs in promoter regions (TSS1500, TSS200, 5′UTR, and 1stExon) probably distinguishes HIV-associated from non-HIV lymphomas. Among the targets measured, those with a significant absolute difference between HIV-associated and non-HIV lymphomas were used for further analyses (Supplementary Methods, Compared with non-HIV lymphoma DNA, HIV-associated lymphoma DNA tended to be hypomethylated (Fig. 1c). Representative genes were used to validate the array analyses. Using COBRA, three of the five non-HIV lymphomas cases were methylated as positive controls, whereas none of the HIV-associated lymphomas was detected as methylated at either RARRES1 or FGF5 (Fig. 1d, upper). Bisulfite DNA sequencing gave consistent results (Fig. 1d, lower), confirming this tendency toward hypomethylation in Group 1 (Fig. 1d). These findings encouraged us to examine previously analyzed cases in Cohort II.

Table 1

Table 1

Fig. 1

Fig. 1

Data from nine HIV-associated lymphoma samples derived from the first visit of Cohort II, which had been previously analyzed using Illumina GoldenGate Methylation Cancer Panel I (see Methods), were used for hierarchical clustering analyses. The results showed two apparent methylation profiles for HIV-associated lymphomas (Groups 3 and 4, Fig. 2a). The genes with a significant absolute difference between two clusters were used for further analyses (Supplementary Method, Group 3 tended to be hypermethylated compared with Group 4 (Fig. 2b). COBRA indicated that all of the Group 3 cases were methylated, whereas fewer in Group 4 were methylated among those tested (Fig. 2c, upper). Bisulfite DNA sequencing clearly showed that Group 3 was highly methylated (Fig. 2c, lower), confirming the tendency toward hypermethylation in Group 3. Two cases in Group 3 subsequently showed recurrence, representing a significant patient characteristic (P = 0.083), if 0.1 was considered a significant level (Table 2). In another case in Group 3, a tumor mass appeared in the cervical spinal cord about 17 months later, although recurrence was not confirmed pathologically. Notably, the methylation profile of nonrecurrent HIV-associated lymphomas (Group 4) did not differ significantly from that of non-HIV lymphomas (non-Group 3, Supplementary Fig. 5 and Supplementary Table 1, These data suggest that recurrent HIV-associated lymphomas have a specific methylation profile.

Fig. 2

Fig. 2

Table 2

Table 2

Back to Top | Article Outline


The prognosis of HIV-associated lymphoma has improved with the development of HIV and cancer therapies [15]. Nevertheless, it is important to identify the mechanism responsible for the aggressiveness of HIV-associated lymphomas. Our data suggest that the DNA methylation profile is a molecular indicator of prognosis.

In the methylation analyses, we examined nine or 11 HIV-associated lymphomas. This number was relatively small because of the small HIV-positive population in Japan [13]. Even so, our data clearly suggest that DNA methylation profiles, especially CGI methylation in promoter regions, differ between HIV-associated and non-HIV lymphomas. As the tumor location varies in HIV-associated lymphoma [2], it is essential to know whether tumor location influenced our analyses. Lymph nodes were the most frequent tumor location and were broadly similar in Groups 1 and 2 (P = 0.45; Supplementary Fig. 6a,, although Group 1 had more extra-node variation, probably due to the high proportion of HIV-associated lymphoma. It is noteworthy that Group 1 had narrower correlation distances than Group 2, indicating that the DNA methylation profiles in Group 1 were quite similar, although Group 1 included various tumor locations (Supplementary Fig. 6b, Additionally, the lymph node cases in Group 1 were very dissimilar from the lymph node cases in Group 2. The data suggested that the clustered results were not due to tumor location. The differences between the profiles may not be related to antiretroviral therapy either, as only two HIV-positive lymphomas in Cohort I were treated with antiretroviral therapy. Coinfections such as EBV with HIV may influence DNA methylation profiles, but we found no significant difference between HIV-associated and non-HIV lymphomas in terms of EBV infection status in our study. However, we cannot exclude the influence of HIV infection on methylation profiles. One of our validation genes, RARRES1, is a cancer methylation target [16] that is differentially expressed in various tumors [17,18], although its clinical relevance to lymphomas remains unknown. FGF5 is reported to be a bone metastasis-related gene related to angiogenesis [19]. As angiogenic growth factors have been implicated in a poor prognosis in non-Hodgkin lymphomas [20], hypomethylated FGF5 may similarly influence the prognosis in HIV-associated lymphomas. Note that several significant pathways related to cell adhesion were found (Supplementary Table 2, Of these, those involving laminins, collagens, N-cadherin, and caveolin2 were significantly hypomethylated in HIV-associated lymphomas, suggesting that their increased expression initiates and promotes tumors and results in a poor prognosis [21–23]. These data partly support the poor prognosis seen in HIV-associated lymphomas.

Clustering analysis of the Cohort II data obtained using Cancer Panel I placed recurrent or suspicious and nonrecurrent HIV-associated lymphomas into separate groups, suggesting that recurrence of HIV-associated lymphomas is attributable to specific gene regulation involving DNA methylation. PTCH2, which was used for validation, was a significant component of the Hedgehog signaling pathway (Supplementary Table 3,, which is related to relapse rate in carcinomas [24]. The data imply that the DNA methylation profile is a good indicator of prognosis. Recently, specific methylation targets have been reported as candidates for new biomarkers of prognosis or metastasis [25,26]. Careful determinations in more cases will identify biomarkers for recurrence in HIV-associated lymphomas.

To our knowledge, this is the first report using molecular technology to distinguish HIV-associated lymphomas from non-HIV lymphomas. Our findings contribute to the understanding of HIV-associated lymphomagenesis and suggest new prognostic biomarkers.

Back to Top | Article Outline


This study was supported by a Grant-in-Aid for Research on Advanced Medical Technology from the Ministry of Health, Labor, and Welfare of Japan (Grant number: H22-AIDS-I-002, and H25-AIDS-I-002), a Grant for International Health Research (104-A), Center of Excellence Research from the Ministry of Education, Sports, Culture, Science, and Technology of Japan, and Grants-in-Aid from the Japan Science and Technology Agency.

Author contributions: S.H. is responsible for the study design and the clinical data. S.H., M.S. and Y.I. planned the experiments; J.T. and S.O. analyzed the clinical information; T.H. and M.M. prepared the pathology samples; A.M. prepared gDNA from clinical samples; A.M. and M.S. analyzed the microarray data; N.T., M.Y., and L.Y. were in charge of the statistical analyses and bioinformatics; and M.S. and S.H. wrote the article. All authors discussed the results of the article.

Back to Top | Article Outline

Conflicts of interest

There are no conflicts of interest.

Back to Top | Article Outline


1. Cote TR, Biggar RJ, Rosenberg PS, Devesa SS, Percy C, Yellin FJ, et al. Non-Hodgkin's lymphoma among people with AIDS: incidence, presentation and public health burden. Int J Cancer 1997; 73:645–650.
2. Raphael M, Said J, Borisch B, Cesarman E, Harris NL. Swerdlow SH, Campo E, Harris NL, Jaffe E, Pillar SA, Stein H, et al. Lymphomas associated with HIV infection. WHO Classification of tumours of haematopoietic and lymphoid tissues 4th edLyon:IARC; 2008. 340–342.
3. Simard EP, Engels EA. Cancer as a cause of death among people with AIDS in the United States. Clin Infect Dis 2010; 51:957–962.
4. Borinstein SC, Conerly M, Dzieciatkowski S, Biswas S, Washington MK, Trobridge P, et al. Aberrant DNA methylation occurs in colon neoplasms arising in the azoxymethane colon cancer model. Mol Carcinog 2010; 49:94–103.
5. Martin-Subero JI, Kreuz M, Bibikova M, Bentink S, Ammerpohl O, Wickham-Garcia E, et al. New insights into the biology and origin of mature aggressive B-cell lymphomas by combined epigenomic, genomic, and transcriptional profiling. Blood 2009; 113:2488–2497.
6. Richter J, Ammerpohl O, Martin-Subero JI, Montesinos-Rongen M, Bibikova M, Wickham-Garcia E, et al. Array-based DNA methylation profiling of primary lymphomas of the central nervous system. BMC Cancer 2009; 9:455.
7. Amara K, Trimeche M, Ziadi S, Laatiri A, Hachana M, Korbi S. Prognostic significance of aberrant promoter hypermethylation of CpG islands in patients with diffuse large B-cell lymphomas. Ann Oncol 2008; 19:1774–1786.
8. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, et al. High density DNA methylation array with single CpG site resolution. Genomics 2011; 98:288–295.
9. Thirlwell C, Eymard M, Feber A, Teschendorff A, Pearce K, Lechner M, et al. Genome-wide DNA methylation analysis of archival formalin-fixed paraffin-embedded tissue using the Illumina Infinium HumanMethylation27 BeadChip. Methods 2010; 52:248–254.
10. Wang D, Yan L, Hu Q, Sucheston LE, Higgins MJ, Ambrosone CB, et al. IMA: an R package for high-throughput analysis of Illumina's 450K Infinium methylation data. Bioinformatics 2012; 28:729–730.
11. Davis S, Bilke S. methylumi: Handle Illumina methylation data. Bioconductor R package version 132 2010. [Accessed 24 September 2012].
12. Bibikova M, Lin Z, Zhou L, Chudin E, Garcia EW, Wu B, et al. High-throughput DNA methylation profiling using universal bead arrays. Genome Res 2006; 16:383–393.
13. Hagiwara S, Yotsumoto M, Odawara T, Ajisawa A, Uehira T, Nagai H, et al. Non-AIDS-defining hematological malignancies in HIV-infected patients: an epidemiological study in Japan. AIDS 2013; 27:279–283.
14. Teschendorff AE, Menon U, Gentry-Maharaj A, Ramus SJ, Weisenberger DJ, Shen H, et al. Age-dependent DNA methylation of genes that are suppressed in stem cells is a hallmark of cancer. Genome Res 2010; 20:440–446.
15. Shiels MS, Pfeiffer RM, Gail MH, Hall HI, Li J, Chaturvedi AK, et al. Cancer burden in the HIV-infected population in the United States. J Natl Cancer Inst 2011; 103:753–762.
16. Youssef EM, Chen XQ, Higuchi E, Kondo Y, Garcia-Manero G, Lotan R, et al. Hypermethylation and silencing of the putative tumor suppressor Tazarotene-induced gene 1 in human cancers. Cancer Res 2004; 64:2411–2417.
17. Davidson B, Stavnes HT, Risberg B, Nesland JM, Wohlschlaeger J, Yang Y, et al. Gene expression signatures differentiate adenocarcinoma of lung and breast origin in effusions. Hum Pathol 2012; 43:684–694.
18. Kloth M, Goering W, Ribarska T, Arsov C, Sorensen KD, Schulz WA. The SNP rs6441224 influences transcriptional activity and prognostically relevant hypermethylation of RARRES1 in prostate cancer. Int J Cancer 2012; 131:E897–E904.
19. Casimiro S, Luis I, Fernandes A, Pires R, Pinto A, Gouveia AG, et al. Analysis of a bone metastasis gene expression signature in patients with bone metastasis from solid tumors. Clin Exp Metastasis 2012; 29:155–164.
20. Salven P, Orpana A, Teerenhovi L, Joensuu H. Simultaneous elevation in the serum concentrations of the angiogenic growth factors VEGF and bFGF is an independent predictor of poor prognosis in non-Hodgkin lymphoma: a single-institution study of 200 patients. Blood 2000; 96:3712–3718.
21. Saha S, Lo PK, Duan X, Chen H, Wang Q. Breast tumour initiating cell fate is regulated by microenvironmental cues from an extracellular matrix. Integr Biol (Camb) 2012; 4:897–904.
22. Montenegro RC, de Vasconcellos MC, Barbosa GD, Burbano RM, Souza LG, Lemos TL, et al. A novel o-naphtoquinone inhibits N-cadherin expression and blocks melanoma cell invasion via AKT signaling. Toxicol In Vitro 2013; 27:2076–2083.
23. Ando T, Ishiguro H, Kimura M, Mitsui A, Mori Y, Sugito N, et al. The overexpression of caveolin-1 and caveolin-2 correlates with a poor prognosis and tumor progression in esophageal squamous cell carcinoma. Oncol Rep 2007; 18:601–609.
24. Chaudary N, Pintilie M, Hedley D, Fyles AW, Milosevic M, Clarke B, et al. Hedgehog pathway signaling in cervical carcinoma and outcome after chemoradiation. Cancer 2012; 118:3105–3115.
25. Bougel S, Lhermitte B, Gallagher G, de Flaugergues JC, Janzer RC, Benhattar J. Methylation of the hTERT promoter: a novel cancer biomarker for leptomeningeal metastasis detection in cerebrospinal fluids. Clin Cancer Res 2013; 19:2216–2223.
26. Huang RL, Gu F, Kirma NB, Ruan J, Chen CL, Wang HC, et al. Comprehensive methylome analysis of ovarian tumors reveals hedgehog signaling pathway regulators as prognostic DNA methylation biomarkers. Epigenetics 2013; 8:624–634.

CpG islands; DNA methylation microarray; HIV; HIV-associated lymphomas; poor prognosticators

Copyright © 2014 Wolters Kluwer Health, Inc.