CMV infection and glioma, a highly controversial concept struggling in the clinical arena

Abstract

Human cytomegalovirus (CMV), a member of the herpes virus family, persists after a usually unrecognized infection lifelong in the body.1 The sero-prevalence ranges from 50% up to more than 90% of the population.2 CMV is the most frequent opportunistic CNS infection in severely immunocompromised patients. Postmortem CMV nucleic acid can be detected in the CNS of more than 20% of individuals without CNS diseases,3 confirming tropism of CMV to CNS cells and indicating that CMV may persist in the CNS lifelong without causing (obvious) harm. In a consensus manuscript, a group of neurooncologists and virologists have agreed on the following positions: (i) There is sufficient evidence that CMV sequences and viral gene expression exist in most gliomas. (ii) There is no evidence for lytic infection of glioblastoma tissue. (iii) CMV in glioblastoma does not fit classic definitions of latent infection. (iv) There is sufficient evidence that CMV could modulate the malignant phenotype in glioblastoma. (v) A specific role for CMV in gliomas remains to be defined. (vi) CMV could serve as a potential novel target for a variety of therapeutic strategies including valganciclovir or vaccination.4 Of note, at the time of the consensus statement, data from the negative randomized clinical trial were still pending. New, in part highly controversial data has led to a critical reevaluation of current data on the key questions: (i) Can CMV transcripts or proteins be detected in glioma tissue? (ii) If so, does this mean that glioma cells are infected with CMV? (iii) If so, does CMV play a role in gliomagenesis and/or progression of gliomas? (iv) If so, does an anti-CMV therapy alter the course of disease? Without a definitive answer on questions 1–3, scientists have now jumped directly to addressing question number 4 by conducting a randomized clinical trial, which is the proper way to collect evidence for efficacy of any novel therapeutic strategy. A randomized trial conducted and published previously by Stragliotto and colleagues did not show antitumor activity of an anti-CMV-directed valganciclovir therapy. This trial included patients with newly diagnosed glioblastoma who were randomized to receive valganciclovir or placebo in addition to standard radiochemotherapy with temozolomide. Progression-free survival estimates in the intention-to-treat (ITT) population were 5.6 months in the valganciclovir and 5.5 months in the placebo group. Median overall survival was 17.9 months in the valganciclovir group and 17.4 months in the placebo group. Two-year survival rates were 27.3% in the valganciclovir group and 25% in the placebo group, respectively.5 In a reassembled series from that study and some individual treatments, the authors claimed a therapeutic effect of valganciclovir treatment despite the negative controlled trial.6,7 Setting aside the obvious conceptual flaws of this retrospective analysis,8–10 it would be of interest to know whether the CMV status correlates with whether patients benefited from therapy with valganciclovir, it would be interesting to know whether the CMV status of the benefitting or nonbenefitting patients is relevant. Similarly, it would be interesting and a relevant proof-of-concept to see an anti-CMV therapy effect in the tumor tissue after therapy at recurrence. This is particularly important as potential off-target effects may be an alternative explanation for any effects of valganciclovir.9 So, with the current data at hand we conclude that there is no proof of biological efficacy of valganciclovir as an anti-glioma agent and it should thus not be used outside clinical trials, which is an important message to patients and caregivers. Does this mean that questions 1–3 can be answered with no? Not necessarily, but if CMV were to play a role in gliomagenesis and/or progression one would expect valganciclovir to work. With a negative trial at hand, one could argue that it is not necessary to answer questions 1 and 2, but a no to these questions would prevent further (unnecessary) clinical trials with anti-CMV strategies such as vaccination. This justifies a critical analysis of the current data on the presence of CMV in gliomas. Cobbs et al first reported the expression of CMV proteins and oligonucleotides in a high percentage of gliomas in 2002.11 However, similar to the remarkable discrepancies of CMV RNA/DNA levels obtained by sequencing methods, immunohistochemistry (protein) as well as in-situ hybridization techniques (nucleic acid) in glioblastoma samples also remained highly controversial, and a French series concluded that CMV is unlikely to be implicated in the development of human malignant gliomas.12 Of note, a combination of morphology-based and molecular-biological detection methods seemed to reduce the likelihood of detecting relevant CMV nucleic acids and proteins in both normal CNS tissues and glial tumors.13 In addition, evaluations of genome-wide transcriptome analyses failed to detect CMV transcripts in glioblastoma tissues.14 Further, new analyses applying more stringent multimodal methods have been unable to recapitulating the presence of CMV in glioblastoma. While certain CMV epitopes may be present in some glioblastoma tissues, there is certainly no proof as of now that CMV is active in glioblastoma patients. In fact, large transcriptome analyses were unable to provide evidence for the presence of relevant amounts of CMV transcripts (<.05 p.p.m.) in glioblastoma tissue. Out of 22.8 billion sequencing reads from 167 tumors, only 1 sequence corresponded to CMV RNA.15 To say the least, these results imply that CMV does not replicate in gliomas. Despite the absence of relevant amounts of DNA in a relevant number of cells, it is proposed that many cells in glioblastomas are positive for CMV proteins. Here, there are methodological problems that are beyond the scope of this opinion paper but it is fair to state that there are both groups that are able to detect relevant amounts of CMV protein in glioma cells of patient tissues and other that are not. What's clear also in the consensus statement is that a positive immunohistochemistry of the CMV immediate early antigen in the tumor tissue, as detected by many authors, does not constitute an infection per se. This needs to be separated from the preclinical cell line work, where several groups see an altered glioma phenotype with CMV infection, despite CMV not being a tumor virus with proven transformation potential, and some efficacy of anti-CMV therapy.16–18 Thus, despite some data and early anti-CMV therapy trials, from our view, unequivocal evidence even for the existence of relevant CMV in ex vivo glioma samples is missing and consensus on the detection methods as well as details on the potential complex biology of CMV in glioblastoma are yet to be provided and understood. Since current protocols for in vivo CMV detection in tumor-associated CMV infection are considerably different from high-throughput diagnostic methods, there is an urgent need for new diagnostic tools that reliably allow for the detection of CMV in tumors. For the potential role of CMV in glioblastoma patients, we would like to propose that evidence of an infection should represent the most reliable evidence for CMV to be relevant. CMV serology of glioblastoma patients rather than CMV, RNA or protein in tumor tissue would provide a convincing piece of information to base preclinical or even clinical research on. So far, these data are missing and it remains to be determined whether patients with glioblastoma are infected with CMV at all. Despite the consensus paper, in our view there is conflicting information on the existence of CMV nucleic acids or protein in glioma tissue and an absence of evidence for the role of CMV in glioma patients. In addition, a controlled intervention trial that did not provide a signal for CMV as a therapeutic target with the currently available measures.10 As gliomas remain a huge challenge and there is a potentially growing role for viral infections in other solid cancers, the CMV efforts may be better devoted otherwise.