Journal of Clinical Neuroscience 22 (2015) 326–330

Contents lists available at ScienceDirect

Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn

Clinical Study

Human cytomegalovirus viral load in tumor and peripheral blood samples of patients with malignant gliomas Eldar Priel a, Anton Wohl b, Michal Teperberg c, Dvora Nass d, Zvi R. Cohen b,⇑ a

Department of Internal Medicine E, Sheba Medical Center, Tel Hashomer, Israel Department of Neurosurgery, Sheba Medical Center, Tel Hashomer, 52621, Israel c National Department of Virology, Sheba Medical Center, Tel Hashomer, Israel d Department of Pathology, Sheba Medical Center, Tel Hashomer, Israel b

a r t i c l e

i n f o

Article history: Received 9 June 2014 Accepted 12 June 2014

Keywords: Glioma Human cytomegalovirus Infection

a b s t r a c t Malignant gliomas are the most common primary brain tumors in adults. The disease has no known etiology, progresses rapidly, and is fatal despite current therapies. Human cytomegalovirus (HCMV) is a beta herpes virus that is trophic for glial cells and infects 50% to 90% of the adult human population. HCMVmediated disease in immunosuppressed patients has highlighted the possible role of this virus in the development of other diseases, particularly inflammatory diseases such as vascular diseases, autoimmune diseases, and certain malignancies. Sensitive detection of viral DNA, mRNA, and antigens in tumor tissues, as well as seroepidemiologic evidence, suggest a link between HCMV and several human malignancies. HCMV gene products are proposed to dysregulate multiple cellular pathways involved in oncogenesis, such as cell cycle regulation, apoptosis, migration, and angiogenesis. These theories, currently being researched, suggest that HCMV acts as an oncomodulator in malignancies. We investigated the association between HCMV infection and reactivation, and malignant gliomas. An open, matched casecontrol, parallel group pilot study was performed in a tertiary referral center. The HCMV viral load in peripheral blood and tumor samples of 19 patients newly diagnosed with glioblastoma multiforme was compared with a matched control cohort comprising 19 patients newly diagnosed with non-malignant brain tumors. There was no significant correlation between peripheral blood and tumor tissue HCMV viral load in patients with glioblastoma multiforme compared to the control cohort. The findings of the present study did not support an oncomodulatory role for HCMV in malignant gliomas. Ó 2014 Elsevier Ltd. All rights reserved.

1. Background Malignant gliomas, such as glioblastoma multiforme (GBM), pose a significant challenge for treatment. Despite broad research in this area, increasing progress in oncologic knowledge, and integrated treatment consisting of aggressive surgical resections (while using novel neuroimaging modalities pre- and intra-operatively), radiotherapy, and chemotherapy as adjuvant therapies, the prognosis for GBM remains grave. Median patient survival is approximately 14 months after diagnosis [1] with 5% of patients surviving more than 2 years. To date, the etiology of this disease remains unknown. Numerous research studies are aimed at better understanding the positive prognostic factors, determining which patients survive longer, and finding novel methods to treat patients and extend their life expectancy with good quality of life. ⇑ Corresponding author. Tel.: +972 3 530 2650; fax: +972 3 535 4413. E-mail address: [email protected] (Z.R. Cohen). http://dx.doi.org/10.1016/j.jocn.2014.06.099 0967-5868/Ó 2014 Elsevier Ltd. All rights reserved.

Human cytomegalovirus (HCMV) is a beta herpes virus carried by most of the world’s population [2]. The prevalence of asymptomatic individuals carrying the virus has been extensively researched, including in remote aboriginal tribes, and indeed HCMV is carried in all populations studied. The prevalence of seropositivity for antigens specific for HCMV increases with age. One epidemiologic study reported a prevalence of 47% in those aged between 10– 12 years, 68% in those aged 15–35 years, and 81% in those aged 36–60 years [3]. For many years it was widely accepted that HCMV is not a clinically important pathogen in the healthy human, and reactivation of the virus, if manifested clinically, presents as an infectious mononucleosis-like disease; in contrast, in immunocompromised hosts, the virus can cause complicated and sometimes deadly diseases. Reactivation of HCMV in these patients may present with high fever, leukopenia, viral hepatitis, interstitial pneumonia, gastrointestinal disease, and acute retinitis. A large percentage (50–90%) of bone marrow transplant and solid organ transplant recipients experience an active HCMV infection after

E. Priel et al. / Journal of Clinical Neuroscience 22 (2015) 326–330

transplant. In acquired immunodeficiency syndrome patients, the prevalence of HCMV infection nears 100% [2]. These immunocompromised patients are the largest risk group for contacting a clinically significant infection with HCMV [4,5]. Herpes viruses have recently been linked with the development of certain malignancies. It is an accepted hypothesis that viruses are responsible for at least 15% of human malignancies. There is evidence for herpesviridae involvement in lymphoma, nasopharyngeal carcinoma, Kaposi sarcoma, and cervical carcinoma [6]. New studies link the presence of genetic products of HCMV with miscellaneous malignancies, such as colorectal carcinoma [7], GBM [8], non-Epstein–Barr virus-related Hodgkin’s lymphoma [9], and prostate carcinoma [10]. These studies and others reveal active HCMV in tumor cells, and in contrast its absence in nonmalignant cells in the tumor’s periphery, although it is currently unclear why the presence of lytic phase HCMV in tumor cells does not infect non-tumor cells. These findings are impossible to ignore, though the question of whether reactivation of the virus constitutes a part of the pathogenesis for tumor development or whether it is a result of tumor development does not have an unequivocal answer [2]. The role of HCMV as an oncogenic virus has been extensively studied in recent decades. It is currently accepted that cells infected with HCMV do not develop malignant transformations, because cells expressing HCMV viral proteins actively stop dividing and eventually die. Therefore, HCMV is not considered an oncogenic virus. A recent suggestion, based on the fact that HCMV infects cells and changes their characteristics in ways that do not consist of malignant transformations, is that HCMV is an oncomodulatory virus. Under this premise, a malignant transformation activates the immune system, which recognizes the tumor as anomalous and initiates a local inflammatory response; HCMV is carried by macrophages to the inflamed area and reactivates. HCMV infection of cells would then change certain characteristics of the cell, thus influencing the aggression and resistance of the tumor [11]. Reviews by Michaelis [12], Soderberg-Naucler [5], and Cinatl [13] present several mechanisms by which HCMV contributes to tumor development, as described briefly below. 1.1. Influence of HCMV on cancer cell apoptosis mechanisms Resistance to apoptosis is a common feature of cancer cells. Not only do apoptotic mechanisms become ineffective due to the pathogenesis of the malignancy, but also ineffective apoptosis mechanisms protect the tumor cells from chemotherapy, and contribute to treatment resistance. It has been proposed that HCMV activates anti-apoptotic mechanisms on fibroblasts in tumor cells by numerous mechanisms, such as by inhibiting caspases and protecting against p53-mediated apoptosis. 1.2. Influence of HCMV on tumor cell adhesion to the endothelium, tumor invasion, and migration These factors play a crucial role in the evolution of metastases. HCMV downregulates neural cell adhesion molecule receptors, thus increasing tumor cell adhesion to the endothelium, and invasion through the endothelium. This effect on neural cell adhesion molecules also weakens the connection between adjacent tumor cells, thus increasing the risk for developing metastasis. 1.3. Influence of HCMV on angiogenesis Recruiting blood vessels to redirect blood to tumor cells is a crucial stage of tumor genesis. HCMV overexpresses interleukin-8 and encodes the human cytomegalovirus US28 protein, thus increasing proangiogenic factors within the host. Moreover, HCMV suppresses

327

the expression of angiogenesis inhibitors such as thrombospondins 1 and 2. Cyclooxygenase-2 (COX-2) contributes to angiogenesis in tumors to such an extent that treatment with COX inhibitors (nonsteroidal anti-inflammatory drugs [NSAID]) is a proposed strategy for tumor prevention and even as an adjuvant therapy in some malignancies [14]. Infection with HCMV leads to overexpression of COX-2 in the human body, and thus contributes to angiogenesis. To summarize, reactivation of HCMV promotes angiogenesis by more than one mechanism, thus contributing to tumor growth and survival. 1.4. Influence of HCMV on tumor immunogenicity When reactivated, HCMV encodes a number of proteins (US2, US3, US6, US11) which decreases the cell surface expression of major histocompatibility complex I and major histocompatibility complex II on infected cells, thus interfering with the recognition of these tumor cells by the adaptive immune system. Moreover, HCMV stimulates the production of transforming growth factor beta 1, a cytokine that suppresses the cellular immune response, in some tumors. Michaelis proposed that transforming growth factor beta 1 is the most prominent glioblastoma immunosuppressant [12]. If the virus assists in reducing tumor immunogenicity, it is a critical factor for tumor survival inside the human body, and may be a therapeutic challenge and target. 1.5. Influence of HCMV on chromosomal stability HCMV encourages chromosomal damage and genetic instability, which might contribute to the development of malignancies. These effects become even more pronounced when cytotoxic treatment is initiated. Under the influence of HCMV, specific chromosomal strand breaks develop at positions 1q21 and 1q42. In these areas, deletion mutations are correlated with the development of malignant gliomas. 1.6. Influence of HCMV on telomerase activation Telomeres are nucleotide sequences at the ends of chromosomes, which become shorter with every cell cycle. When telomeres become too short, the cell cycle stops in an irreversible way, a process known as senescence. The physiologic mechanism by which telomere shortening is prevented and senescence is delayed is the action of telomerase-dependent and RNA-dependent DNA polymerase, which lengthens telomere DNA. When telomerases are redundant, they bestow ‘‘eternal life’’ upon the cell, and thus malignant properties. In 2009, Straat found that infecting cells with HCMV activates telomerases, thereby conferring malignant properties to the cell [15]. This is yet another mechanism by which HCMV changes the characteristics of cells and improves their survival. These findings represent advances in determining the cause versus the effect of virus infection and malignancy development. 1.7. Recent discoveries Although it was reported approximately 30 years ago that HCMV infects malignant cells in patients with carcinomas of the colon and prostate [16,17], later pathologic studies contradict these findings [18]. The development of new, highly sensitive laboratory techniques enables the identification of minute amounts of viral nucleic acids, proteins, and antigens inside different tumors. These advances have reignited the discussion of the role of HCMV in the development of malignancies. Indeed, there have been some impressive findings in this area, as discussed below. In 2002, Cobbs et al. [9] used immunohistochemistry and in situ hybridization to identify HCMV nucleic acids and viral proteins in gliomas, and

328

E. Priel et al. / Journal of Clinical Neuroscience 22 (2015) 326–330

found that HCMV viral proteins were present in a higher percent of specimens of malignant gliomas than in other central nervous system samples (normal brain, meningioma, stroke, Alzheimer’s disease, paraneoplastic encephalitis, cryptococcal cerebritis): 27/27 in glioma samples versus 0/23 in other samples. In 2007, Mitchell et al. [19] used samples from a tumor bank to perform immunohistochemistry for viral proteins and in situ hybridization for HCMV nucleic acids (IE1). They also sampled peripheral blood from patients recently diagnosed with malignant gliomas, and processed these samples by polymerase chain reaction (PCR) for HCMV glycoprotein B (UL55). Immunohistochemistry showed positive results in 42 of 45 samples (93%). Using a more specific method, positive results were obtained for 30 of 33 samples (91%). Specific hybridization for HCMV was positive in 16 of 16 samples. PCR for HCMV DNA was positive in 21 of 34 tumor samples (61.7%) and in 16 of 20 peripheral blood samples (80%). PCR for HCMV DNA was negative in 17 of 17 control group samples taken from healthy volunteers. Eleven of the samples were seropositive for HCMV [19]. Thus, a correlation can be demonstrated between HCMV and malignant gliomas using several different research methods. As stated previously, however, cause and effect have not been clarified. 1.8. Study goals and objectives HCMV is present in malignant gliomas, though it is not known if its presence is part of the pathogenesis, thus constituting a cause of tumor development, or if its presence arises from local inflammatory processes, thus constituting a consequence of tumor development. Our goal was to enrich medical knowledge about GBM and its connection with HCMV. A positive correlation might stimulate widespread studies designed to clarify whether antiviral treatment could change the disease course and prognosis.

2. Methods 2.1. Study design We designed a matched, case-controlled pilot study in a tertiary medical center, namely the Chaim Sheba Medical Center, Tel Hashomer, Israel. We compared HCMV viral loads in peripheral blood and tumor cells in 19 patients recently diagnosed with malignant gliomas to a matched control group comprising 19 patients recently diagnosed with non-malignant brain tumors. Each patient was tested for seropositivity or negativity for HCMV immunoglobulin (Ig) G/IgM. Subjects were matched according to the following parameters: age (±3 years), sex, and smoking habit. The study was approved by Sheba Medical Center’s Institutional Review Board. 2.2. Study population The study group comprised 19 patients from the neurosurgical ward at Chaim Sheba Medical Center who were recently diagnosed with malignant gliomas. Tissue samples were obtained by biopsy or surgical resection, and were reviewed by a board certified pathologist. Each patient signed an informed consent form to participate in this study. The control group comprised 19 patients from the neurosurgical ward at Chaim Sheba Medical Center who were recently diagnosed with non-malignant brain tumors. Each patient signed an informed consent form to participate in this study. Exclusion criteria were lymphopenia, prior medical treatment with steroids, prior medical history consistent with malignancies, and chemotherapy or radiotherapy treatment at enlistment or within 24 months prior to enlistment.

2.3. Sample collection Peripheral blood samples were obtained with disposable needles. Peripheral blood (5 ml) was drawn into an ethylenediaminetetraacetic acid-coated test tube and immediately sent to the national virologic laboratory at Sheba Medical Center. Every sample was tested for HCMV IgG and IgM, and real-time PCR for HCMV DNA was performed for every sample. Tumor cell samples were obtained during surgical resection or with a needle biopsy. For PCR, samples were kept cool (4°C) and sterile, and were sent immediately to the virology laboratory at Sheba Medical Center. All samples were processed within 24 hours of procurement. Every sample was tested for HCMV DNA using real-time PCR. Real-time PCR was performed using the TaqMan method in the National Virology Lab at Sheba Medical Center. To produce nucleic acids from the different samples we used the MagNa pure system (Roche Applied Science, Indianapolis, IN, USA). Approximately 200 lL of sample produced 100 lL DNA, from which 10 lL was used for every reaction and 5 lL was used as an internal control. To minimize the possibility of contamination, different phases of the reaction were performed in separate and distant areas: nucleic acids were extracted in one room; reagents and primers for the PCR were prepared in a second, clean room; and the reaction was performed in a third room. DNA amplification was performed by real-time PCR with specific primers for HCMV glycoprotein B and IE. An internal control was obtained using RNASE-P. We aimed at amplifying cytomegalovirus glycoprotein B (UL55) and immediate early proteins. Table 1 lists the primers and probes that we used. Positive results were sequenced for HCMV glycoprotein B.

2.4. Statistical analysis This was a case-controlled pilot study, aimed at performing a quantitative comparison between HCMV viral loads in patients newly diagnosed with malignant gliomas to those in patients with newly diagnosed non-malignant brain tumors. This pilot study was designed to test the following two hypotheses: (1) HCMV viral loads in patients with malignant glioma are greater than zero; and (2) HCMV viral loads in patients with non-malignant brain tumors equal zero. An additional goal was to present preliminary data on the quantitative characteristics of viral loads in patients. A sample size calculation was carried out before performing the study, using Student’s t-test to reject the assumption that there is no correlation between HCMV viral loads in the different mediums tested. An r value greater than 0.55 was considered statistically significant with a power of 80%. A type I error was set to 5%. The sample size was necessarily small because GBM is a rare disease. Because no published data to our knowledge provide evidence for the quantification of HCMV viral loads, we could not predict the size of the difference between the groups. We planned to perform a statistical analysis after obtaining the preliminary results. As the data accumulated, however, no higher prevalence of HCMV viral loads was demonstrated compared with the control group; therefore, we could not determine the sample size needed to demonstrate statistically significant differences between the groups. Means and standard deviation were calculated for the continuous variables (e.g., age). The distribution of continuous variables was tested with a one-sample Kolmogorov–Smirnov test. We used the Mann–Whitney test to determine whether there was a significant difference in the mean age between study and control groups. We used the chi-squared test to assess whether there was a statistically significant difference between the groups in the incidence of discrete variables.

329

E. Priel et al. / Journal of Clinical Neuroscience 22 (2015) 326–330 Table 1 Primers and probes Primer set

CMV region

Sequence Forward primer

Reverse primer

Probe

1

Gb (UL55)

TGG GCG AGG ACA ACG AA

TGG GCA ACC ACC GCA CTG AGG

2

IE

GAC TAG TGT GAT GCT GGC CAAG

TGA GGC TGG GAA GCT GAC AT GCT ACA ATA GCC TCT TCC TCA TCTG

AGC CTG AGG TTA TCA GTG TAA TGA AGC GCC

hand, local characteristics of the tumor enabled viral reactivation, the expected dynamics would be inverted; that is, local reactivation of the virus, leading to brain and cerebral spinal fluid infection, and not necessarily to peripheral viremia. We predicted a correlation between peripheral viremia and tissue prevalence of HCMV copies. Our hypothesis was not supported by the data. In light of the results of Cobbs et al. and Mitchell et al. [9,19], we predicted the opposite effect. Below we speculate on the possible reasons for this contradiction. 4.1. Study methods

CMV = cytomegalovirus, Gb = glycoprotein, IE = immediate early.

All data analyses were performed using the Statistical Package for the Social Sciences version 17.0 (SPSS, Chicago, IL, USA). 3. Results We recruited 25 patients to the study group, six of whom were excluded. The most common cause of exclusion was a prior medical history of malignancy. Of the 19 patients who were included, all peripheral blood samples were negative for HCMV DNA, and all tumor samples were negative for HCMV DNA. HCMV copies were detected in the peripheral blood sample of a patient who was excluded due to a prior malignancy. No HCMV DNA, however, was detected in this patient’s tumor cells. In all but four patients in the study group, there was serologic evidence of prior HCMV exposure. We recruited 30 patients for the control group, all of whom were recently diagnosed with non-malignant brain tumors. Of these, three were excluded; one patient due to chronic corticosteroid therapy, one patient due to prior malignancy, and one patient due to nosocomial meningitis after a neurosurgical procedure. After matching cases and controls (using age, sex, and smoking status), data were processed for the 19 matched controls. All control samples were negative for HCMV viral copies. Of this group, tumor samples were obtained from seven patients. All samples were negative for HCMV copies. All control patients except one had serologic evidence for prior exposure to HCMV. One patient in this group had serologic evidence for both recent and past exposure to HCMV (both IgG and IgM). The results are presented in Table 2. 4. Discussion We predicted that we would find evidence for HCMV in the serology and tissue samples of the study group. To our knowledge, no data have been published that quantify the dynamics of viral loads in serology or tissue samples, or about the average viral copies demonstrated. If HCMV were involved in the pathogenesis of malignant gliomas, the expected dynamics would include peripheral viremia leading to viral infection of the cerebral spinal fluid, and, as a consequence, parenchymal brain infection. If, on the other

Compared to Mitchell et al.’s 2008 study, our analysis was aimed at the same expression of lytic HCMV infection: glycoprotein B (UL55) and IE1. The fact that we selected patients recently diagnosed with the disease is also similar to the PCR analysis in this study. There was no difference in the study methods that can explain the discrepancy. 4.2. Study weaknesses 4.2.1. Selection bias First, our patients were all newly diagnosed with their diseases, and we have not tested our hypothesis using patients at other disease stages. Second, this small-scale pilot study was performed in the neurosurgery department of Sheba Medical Center, a tertiary hospital, and one of only a few centers in Israel focusing on the diagnosis and treatment of GBM. Even if the power of our study allowed us to generalize our conclusions, they are, at most, valid only for the Israeli population. When taking into account that our study deals with phenomena related to infectious diseases in which epidemiology varies, this is an important limitation of our results. This limitation has been used as a possible explanation for discrepancies between different studies in this area, and for why, in studies with positive results, HCMV presence in gliomas is not expressed at the same levels [20]. Third, we instituted strict exclusion criteria, which might have caused us to select patients with a more durable immune system. 4.2.2. Small groups As a pilot study, our study and control groups were small, comprising 19 patients per group. We addressed this problem by matching study patients with control patients, thus eliminating confounding factors and contributing to the power of the study. The use of COX inhibitors (NSAID) may suppress reactivation of HCMV [14]. In some cases, tumor development is accompanied by pain. In Israel, where NSAID are sold over the counter for analgesic treatment, some of our patients may have used COX inhibitors prior to the study, thus influencing the results. HCMV is a special virus, in that, contrary to other oncogenetic viridae, it does not classically express at a lytic or latent phase [21]. As a virus that can express over 166 genes [22], it might be that studies finding HCMV viral proteins and genes present a picture that is merely the ‘‘tip of the iceberg’’. There is yet much to be investigated regarding the importance of HCMV in malignant gliomas. A study

Table 2 Study results Units Age Female Serum IgM Serum IgG Serum CMV DNA Tumor CMV DNA

Mean and SD Number (%) Number (%) Number (%) Number (%) Number (%)

Study group

Control group

RR (CI)

p value

57.9 ± 13.2 12/19 (63.1%) 0/18 (0.0%) 15/18 (83.3%) 0/19 (0.0%) 0/17 (0.0%)

58.3 ± 11.5 12/19 (63.1%) 1/18 (5.5%) 17/18 (94.4%) 0/19 (0.0%) 0/7 (0.0%)

N/A 1.000 (0.268–3.737) N/A 0.294 (0.028–3.138) N/A N/A

0.781 0.999 0.999 0.603 N/A N/A

CI = confidence interval, CMV = cytomegalovirus, Ig = immunoglobulin, N/A = not applicable, RR = relative risk, SD = standard deviation.

330

E. Priel et al. / Journal of Clinical Neuroscience 22 (2015) 326–330

of cytomegaloviruses by Bhattacharjee et al. [23] sheds some light on this area. 5. Conclusions It is not clear whether HCMV comprises a part of the pathogenesis of brain tumors or whether the tumor itself creates a convenient environment for the virus to reactivate. The presence of the virus inside or near tumor cells, whether a cause or effect of tumor growth, may serve as a basis for developing future adjuvant therapies for malignant gliomas. These treatment options may include antiviral or biologic therapies, targeting viral characteristics to attack tumor cells. In contrast to a recent consensus [24] that HCMV sequences and viral gene expression exist in most, if not all, malignant gliomas, we found no difference between the HCMV viral load in patients with malignant gliomas compared to patients with non-malignant brain tumors. Our negative results reveal the need for further research to determine if patients with malignant gliomas comprise two separate patient groups: one with evidence of HCMV reactivation, and the other without such evidence. If these two patient groups could be identified in other studies, it would be interesting to investigate whether there is a difference in survival rates. If HCMV modulates the malignant phenotype in glioblastomas by interacting with key signaling pathways, then HCMV could serve as a novel target for a variety of therapeutic strategies. The findings of the present study, however, did not support an oncomodulatory role for HCMV in malignant gliomas. Conflicts of Interest/Disclosures The authors declare that they have no financial or other conflicts of interest in relation to this research and its publication. This study was conducted as a part of the academic requirements to obtain an M.D. degree, at Sackler School of Medicine, Tel Aviv University, Israel. References [1] Adamson C, Kanu OO, Mehta AI, et al. Glioblastoma multiforme: a review of where we have been and where we are going. Expert Opin Investig Drugs 2009;18:1061–83.

[2] Britt WJ, Mach M. Human cytomegalovirus glycoproteins. Intervirology 1996;39:401–12. [3] Klemola E, Kaariainen L. Cytomegalovirus as a possible cause of a disease resembling infectious mononucleosis. Br Med J 1965;2:1099–102. [4] Michelson-Fiske S. Human cytomegalovirus: a review of developments between 1970 and 1976.Part II. Experimental developments. Biomedicine 1977;26:86–97. [5] Soderberg-Naucler C. Does cytomegalovirus play a causative role in the development of various inflammatory diseases and cancer? J Intern Med 2006; 259:219–46. [6] Sissons JG, Carmichael AJ. Clinical aspects and management of cytomegalovirus infection. J Infect 2002;44:78–83. [7] Kadow JF, Regueiro-Ren A, Weinheimer SP. The role of viruses in human cancer development and antiviral approaches for intervention. Curr Opin Investig Drugs 2002;3:1574–9. [8] Harkins L, Volk AL, Samanta M, et al. Specific localisation of human cytomegalovirus nucleic acids and proteins in human colorectal cancer. Lancet 2002;360:1557–63. [9] Cobbs CS, Harkins L, Samanta M, et al. Human cytomegalovirus infection and expression in human malignant glioma. Cancer Res 2002;62:3347–50. [10] Huang G, Yan Q, Wang Z, et al. Human cytomegalovirus in neoplastic cells of Epstein-Barr virus negative Hodgkin’s disease. Int J Oncol 2002;21:31–6. [11] Samanta M, Harkins L, Klemm K, et al. High prevalence of human cytomegalovirus in prostatic intraepithelial neoplasia and prostatic carcinoma. J Urol 2003;170:998–1002. [12] Michaelis M, Doerr HW, Cinatl J. The story of human cytomegalovirus and cancer: increasing evidence and open questions. Neoplasia 2009;11:1–9. [13] Cinatl J, Scholz M, Kotchetkov R, et al. Molecular mechanisms of the modulatory effects of HCMV infection in tumor cell biology. Trends Mol Med 2004;10:19–23. [14] Pereg D, Lishner M. Non-steroidal anti-inflammatory drugs for the prevention and treatment of cancer. J Intern Med 2005;258:115–23. [15] Straat K, Liu C, Rahbar A, et al. Activation of telomerase by human cytomegalovirus. J Natl Cancer Inst 2009;101:488–97. [16] Huang ES, Roche JK. Cytomegalovirus D.N.A. and adenocarcinoma of the colon: evidence for latent viral infection. Lancet 1978;1:957–60. [17] Sanford EJ, Geder L, Laychock A, et al. Evidence for the association of cytomegalovirus with carcinoma of the prostate. J Urol 1977;118:789–92. [18] Hart H, Neill WA, Norval M. Lack of association of cytomegalovirus with adenocarcinoma of the colon. Gut 1982;23:21–30. [19] Mitchell DA, Xie W, Schmittling R, et al. Sensitive detection of human cytomegalovirus in tumors and peripheral blood of patients diagnosed with glioblastoma. Neuro Oncol 2008;10:10–8. [20] Fonseca RF, Kawamura MT, Oliveira JA, et al. The prevalence of human cytomegalovirus DNA in gliomas of Brazilian patients. Mem Inst Oswaldo Cruz 2012;107:953–4. [21] Moore PS, Chang Y. Why do viruses cause cancer? Highlights of the first century of human tumour virology. Nat Rev Cancer 2010;10:878–89. [22] Mocarski E, Shenk T, Pass R. Cytomegaloviruses. In: Knipe DM, Howley PM, editors. Fields Virology. 5th ed., Philadelphia: Lippincott, Williams & Wilkins; 2007. [23] Bhattacharjee B, Renzette N, Kowalik TF. Genetic analysis of cytomegalovirus in malignant gliomas. J Virol 2012;86:6815–24. [24] Dziurzynski K, Chang SM, Heimberger AB, et al. Consensus on the role of human cytomegalovirus in glioblastoma. Neuro Oncol 2012;14:246–55.