Epstein–Barr virus- and cytomegalovirus-specific immune response in patients with brain cancer

Human malignant glioma (brain tumor) and pancreatic cancer are characterized by poor prognosis following diagnosis. Patients with pancreatic cancer have a 5-year survival rate of 5% [1], while the situation is worse for patients diagnosed with glioblastoma multiforme (GBM), the most malignant form of glioma, whose 5-year survival rate is frequently below 3% despite treatment [1‐3]. Studies have revealed that the tumor microenvironment in patients with either cancer diagnosis is dysfunctional and immunosuppressive in nature, characterized by the low number of tumor infiltrating lymphocytes (TILs), dysfunctional T-cell receptors (TCR), T-helper 2 (Th2) skewed cytokine and generally low tumor immunogenicity [2, 4, 5]. It is therefore plausible to assume that these underlying biological factors may inevitably have an observable effect on immune responses to common infectious agents, including Epstein–Barr virus (EBV) and cytomegalovirus (CMV).

CMV- or EBV-associated disease mostly occurs in individuals with a compromised immune system arising from transplantation, human immunodeficiency virus co-infection (HIV), autoimmunity or cancer to list a few [6]. EBV and to a lesser extent CMV are implicated in cancer owing to their onco-modulatory properties. Nasopharyngeal carcinoma (NPC) and post-transplant lymphoproliferative disorder (PTLD) are attributed to EBV-driven pathogenesis. Furthermore, active EBV can perpetrate acute clinical pancreatitis, where reduced numbers of CD57+ CD8 T cells have been observed, indicating hampered cell-mediated immune responses in conjunction with chronic immune activation and reduced responsiveness to cognate epitopes [7, 8]. Interferon gamma (IFNγ) and tumor necrosis factor alpha (TNFα) production by CD4+ T helper 1 (Th1) targeting EBV nuclear antigen 1 (EBNA-1) in patients with breast cancer harboring the latent form of EBV correlated with better clinical outcome compared to those infected with the replicative form of the virus [9]. No direct association of EBV infection in brain tumor or pancreatic cancer has been shown to date although its implication in glioma has drawn recent attention [10]. Although initial studies link anticancer responses potentially orchestrated by EBV-specific T cells, this hypothesis requires formal testing.

CMV-derived components such as viral DNA and protein antigens in GBM tumors have been long debated and remain controversial [11]. CMV-positive patients with GBM receiving valganciclovir adjunctively to standard chemotherapy showed overall improved survival after antiviral therapy [12], yet other groups have not observed any evidence of increased CMV DNA load in the tumor [13]. Interestingly, human cytotoxic T cells specific for CMV structural protein pp65 (CMV-pp65) can recognize and kill human primary GBM as well as leukemic cells [14, 15], while a CMV-based vaccine candidate facilitated enhanced survival among patients with GBM in a recent clinical trial [16].

Several reports have described the humoral response to CMV and EBV in association with brain tumor [17‐19]. However, to the best of our knowledge, specific IFNγ production to EBV and CMV in patients with pancreatic cancer or brain tumor remains unexplored. CMV-pp65 is a strong TCR agonist that is widely used to gauge the immune competence of individuals, while EBNA-1 is associated with latent EBV infection [20], and contains epitopes capable of inducing strong CD4+ T-cell responses and antibody production.

While the specific IFNγ production to tumor associated antigen (TAA) would highlight the exclusive response to cancer antigen, IFNγ response to common viral antigens in Sweden such as CMV-pp65 and EBNA-1 (prevalence above 80%) may reflect general immunological balance versus immune exhaustion [21, 22]. We reported here the anti-CMV/EBV humoral and cellular immune response profile of patients with brain tumor in comparison to those with pancreatic cancer and healthy individuals (HD). The results of this study are expected to contribute to developing CMV/EBV-directed immune reactivity as valuable biomarkers of immune fitness in patients with advanced cancer.

Chronic local inflammation in response to pathogens and/or autoantigens, orchestrated by a combination of innate and adaptive immune cell activity can lead to the oncogenesis of solid tumors [31]. Such perturbation of the overall immunological equilibrium may compromise the immune system’s ability to deter productive infection by latent pathogens and associated clinical disease [32, 33]. Therefore, well-preserved pathogen-directed immune responses reflect a certain level of immunological fitness in patients with cancer. IFNγ produced by T cells has, among its various biological functions, an essential role in the anti-viral and anti-tumor immune defense [34]. Earlier clinical studies involving patients with advanced cancer indicate that reduced or impaired IFNγ signaling may negatively affect patient survival [35‐38]. The present report describes for the first time CMV- and EBV-specific immune responses—characterized by antigen-specific IFNγ production and humoral immune responses—in chemotherapy-naïve patients with pancreatic cancer or brain tumor.

First, we observed in the present study cohort that CMV-pp65-specific but not EBNA-1-specific plasma IgG levels were different among patients with cancer and healthy donors. Plasma IgG level is usually indicative of a past infection, and although not assayed in conjunction with other Ig subclasses recognizing EBV early antigen or viral capsid antigen, our data suggests that the patients in the cohort described here have well-preserved humoral immunity to EBV infection. On the contrary, patients with brain tumor showed a lower IgG response to CMV as opposed to patients with pancreatic cancer and healthy individuals. This observation agrees with a clinical study in which 8 out of 15 patients with high-grade glioma who presented with low-to-no CMV DNA present despite testing positive for CMV-specific IgG [13]. Earlier studies by Wrensch et al. [39, 40] described lower IgG response to varicella-zoster virus (VZV) but did not report any direct association with other herpesviruses i.e. CMV and EBV, while Amirian et al. [18] showed that among CMV IgG-positive patients, those with the lowest level of anti-CMV IgG presented the highest glioma risk. Of note, an exploratory study described the poor reliability of serology tests to define CMV infection in patients with GBM as the authors observed discrepancy between cellular and humoral immune reactivity to CMV peptides [19]. Additionally, CMV-specific T cells have been found at a low frequency yet protective against CMV infection in CMV-seronegative kidney transplant recipients and after a short CMV-pp65 peptide stimulation, 75% of healthy CMV-seronegative donors exhibited a CMV-specific T cell response [41, 42]. This discordant response between cellular and humoral immunity toward CMV antigens have been earlier described showing proliferative response to CMV antigens by PBMCs of CMV-seronegative donors [43]. Therefore EBV or CMV serology of healthy individuals or patients with cancer may not necessarily reflect the status of their cell-mediated immunity to these viruses.

Numerous studies and clinical observations have shown that immunological control of latent infection with CMV and/or EBV in humans requires adequately intact T-cell responses [9, 44, 45]. Thus, upon antigen encounter, these CMV-directed lymphocytes should be able to expand rapidly and respond with cytokine production (i.e. IFNγ, TNFα, IL-2) and/or cytotoxicity [46]. Interestingly, anti-CMV T-cell responses detected in human GBM tumor tissue have been suggested to play a role in driving anti-tumor immune responses in some patients [14, 47]. We however observed that exposure of peripheral blood of patients with malignant glioma to CMV-pp65 antigen in the absence of cytokine conditioning did not enhance significantly the IFNγ production. The same appeared to be true for exposure to EBNA-1 protein as well as PHA. It is noteworthy to point out here that patients with GBM or brain metastases displayed particularly meagre IFNγ responses to the viral proteins or mitogen. This is a stark representation of the general immune impairment in these individuals, which has been in part attributed to the reduced responsiveness of peripheral blood T cells from brain tumor patients to IL-2 further to the use of corticosteroids to reduce brain inflammation [28, 48]. Moreover, potentially elevated expression of immune checkpoint molecules i.e. programmed cell death 1 (PD-1), cytotoxic lymphocyte-associated antigen 4 (CTLA-4) could also drive general exhaustion of T cells in patients with GBM as previously reported [49, 50]. GBM is the most aggressive form of malignant glioma, while the occurrence of brain metastases is also suggested to impair cell-mediated immunity owing to increased transforming growth factor beta (TGF-β) levels, elevated numbers of circulating and tumor-infiltrating regulatory T cells (Treg) and non-productive inflammatory processes [3, 28, 51]. Also, CMV- and EBV-directed IFNγ production in peripheral blood has been linked to better survival among patients with tuberculosis (characterized by aberrant immunopathology leading to systemic immune dysregulation that also manifests in T-cell exhaustion) after successful completion of standard antibiotic therapy [52]. Thus, albeit immune checkpoint expression on T cells, their ability to react to stimulation by CMV/EBV is not abrogated and thus hints at a more intense nature of immune suppression experienced by patients with glioma compared to those with pancreatic cancer.

Impairment of T-cell activity due to cellular exhaustion (where inflammation plays an important role) can generally predict disease progression, thus patient survival. Since immune exhaustion is characterized by reduced ability to produce cytokines and proliferate, the IFNγ production profile in whole blood presented represents an important characteristic of immune fitness—with regard to the cellular immune compartment [53, 54]. EBV and CMV specific T cells can have high PD-1 expression, resulting from persistent TCR stimulation during chronic infection [55], which is also governed by T-cell memory and activation status thus exerting a profound effect on their functionality [56, 57]. Interestingly, CMV-specific T cells have also been implicated in clearance of chronic hepatitis C virus infection [58], while being functionally intact in patients with chronic lymphocytic leukemia despite general T-cell impairment in addition to immune exhaustion [59]. Also, CMV- and EBV-directed IFNγ production in peripheral blood has been linked to better survival among patients with tuberculosis (characterized by chronic inflammation) after successful completion of standard antibiotic therapy [52].

Cytokine conditioning of peripheral blood in the present study had a pronounced effect on IFNγ production to viral antigen exposure. Furthermore, the influence of the cytokine conditioning-induced increase of IFNγ production in whole blood appeared to vary between patient groups and the viral antigens themselves (CMV-pp65 vs EBNA-1). T cell populations may display different degrees of sensitivity to cytokine exposure largely due to immune dysfunction-driven dynamics of the surface cytokine receptor expression, as reported in cancer [48]. Furthermore, we previously reported that TILs from pancreatic tumor as well as GBM tissue can be successfully expanded in culture medium containing IL-2, IL-15 and IL-21, leading to proliferation of central memory T cells with strong effector functions and a rich TCRVβ repertoire, hinting at a wide epitope recognition potential [29, 30]. In addition, cytokine effect in reducing the numbers of pre-apoptotic cells and restoring TCR function in patients with HIV infection or acute myeloid leukemia has been previously reported [60, 61]. In agreement with the existing evidence, treating whole blood of patients with cancer with IL-2, IL-15 and IL-21 markedly improved the antigen-specific IFNγ response regardless of existing immunosuppression, thereby suggesting clinical applicability.

Recognition of CMV antigens by plasma IgG in patients with GBM is generally lower as compared to healthy individuals and patients with pancreatic cancer, although CMV IgG-positive patients with GBM with a higher antibody titer do significantly better from a clinical perspective (pertaining to survival after surgery). We have shown that compared to patients with pancreatic cancer and healthy individuals, immune function represented by CMV- and EBV-specific cellular response of patients with brain tumor is impaired. This immune reactivity may be improved by cytokine conditioning of lymphocytes. Also, IFNγ production to EBV was found to be of interest in patients with GBM. This correlation was only observed with IL-2/IL-15/IL-21 conditioning of the peripheral blood affirming that our finding are not only related to the specific immune response to EBV an antigen but more likely also reflecting the immune fitness of the patient with GBM as the capability to respond to cytokines stimulation.

We acknowledge that this study is limited the investigation to the anti-viral immune response characterized by the CMV-specific IgG level and the IFNγ production by circulating immune cells in blood in patients with brain cancer in comparison with those with pancreatic cancer. While our results show that the competence of the general and CMV/EBV specific immune response impact on clinical outcome, we acknowledge that a myriad of other clinical parameters govern patient survival as were shown in a previous publication from our laboratory concerning the predictability of the survival of patients with brain metastases based on anti-mesothelin immune responses in blood shown by multivariate analysis [62]. Nevertheless, this is the first report to show link between immune response patterns to CMV and EBV in relation to survival of patients with brain tumor, and be developed for future studies with the inclusion of necessary clinical criteria [63, 64].

Antiviral immune responses, especially to common pathogens such as CMV and EBV are an integral component of immune competence in humans. These findings shed new light on our understanding of antiviral immune responses in patients with brain tumor while reflecting the general immunological status of patients with cancer in response to T cell-oriented cytokines and/or common viral antigens. Thus, efforts to better understand and characterize the cellular exhaustion/impairment caused by persistent chronic infection as well as the tumor microenvironment may lead to novel strategies for exploiting functional and clinically relevant immune responses for therapy.