Impact of Cytomegalovirus infection on immune signatures in cancer patients

Human Cytomegalovirus (CMV) is a wide-spread persistent beta-herpes virus, infection with which has a substantial effect on the distribution of different T-cell subsets, especially CD8+ T-cells, driving them towards a late-differentiated, more “elderly” phenotype [1–3]. In an infected host, a significant fraction of the immune system (even up to 30% of CD4+ and CD8+ memory responses in some individuals) is committed to controlling the virus [4]. Our unpublished results in hormone-resistant non-metastatic prostate cancer show that CMV-seropositive patients have significantly lower levels of naïve CD8+ T-cells and a higher frequency of late-differentiated CD27− and TEMRA (CCR7−, CD45RA+) CD8+ T-cells compared to CMV-seronegative patients. Such an “old” CD8 compartment has also been observed in healthy CMV-seropositive individuals [3, 5] and is rapidly induced after primary infection with the virus [6–8], suggesting CMV (and not only chronological age) as a driving force for human immunosenescence.

In an article entitled “Immune impairment in patients with terminal cancers: influence of cancer treatments and cytomegalovirus infection” [9], Chen et al. report decreased frequencies of naïve and central memory (CM) lymphocytes, accompanied by an increase in effector memory (EM) lymphocytes, in heavily-treated stage IV cancer patients (HT) compared to healthy age-matched controls. Furthermore, they show reduced expression levels of costimulatory receptors CD27 and CD28 on lymphocytes in HT patients compared to newly diagnosed (treatment naïve, TN) stage III/IV patients and healthy controls. However, I understand from the manuscript that they did not determine the CMV-serostatus of the patients and controls. The authors have looked at IE-1 copy numbers in plasma, but usually IE-1 DNA is only detected in plasma during active CMV infection and does not reflect latent infection (personal communication, Dr. R. Beck, Department of Medical Virology, Tübingen). The late-differentiated phenotype observed in HT patients may simply be due to a higher CMV infection rate in this group and not necessarily due to tumor stage and/or previous treatment as the authors suggest. Indeed, the authors show a significantly higher frequency of CD8+ T-cells carrying the receptor for an HLA-A2-restricted epitope from CMV in HT patients. However, it is not clear from the manuscript if data from only CMV-seropositive individuals or all participants are presented. As it is not specified otherwise, I assume that the frequency of tetramer-positive T-cells is compared in all HLA-A2+ donors tested. Thus, the higher frequency of these cells in HT patients might only be due to a higher seropositivity rate in this group, which would correlate with the observed more “late-differentiated” T-cell phenotype. Additionally, the authors show higher IE-1 DNA copy numbers in HT patients compared to other groups. This in itself is an interesting finding, suggesting a higher reactivation rate of this virus in heavily-treated cancer patients, as already published by the authors themselves [10]. Thus, the changes observed in the constellation of different T-cell subsets in HT patients can be attributed to this parameter and not to tumor stage and/or previous treatment directly.

Another important point discussed in the manuscript by Chen et al. is the question of immunological fitness of cancer patients enrolled in immunotherapy trials. Their data clearly show a significant difference between cancer patients in the composition of different lymphocyte populations. Such inter-individual differences might well influence the outcome of cancer vaccination in these patients, whatever the reason for their existence. Whether a reduced fraction of naïve cells and accumulation of late-differentiated T-cells (widely presumed to be dysfunctional) actually does influence responses to immunotherapy in cancer patients is a crucial question that remains to be explored. A limited body of evidence on the outcome of influenza vaccination, however, suggests that this might indeed be the case. Whereas ‘flu vaccination is effective in a majority of young adults (70–90%), only 17–53% of the elderly respond to the vaccine by mounting a protective anti-viral antibody response [11, 12]. Interestingly, an increased frequency of CD28−CD8+ T-cells and seropositivity for CMV have been reported to correlate with poor responsiveness to influenza vaccination in the elderly [13, 14]. Considering the enormous impact of CMV on immune signatures, especially within the CD8+ T-cells—which are the target cells in most immunotherapy trials in cancer patients—a pre-existing infection with CMV and the resulting more “exhausted” status of the immune system might well have an impact on their responsiveness to active immunotherapy procedures. The search for the “holy grail” of cancer vaccination, namely the ability to predict and select patients who would respond to a given vaccination strategy, is still ongoing. We believe that to identify reliable immune biomarkers in any clinical setting, the CMV-serostatus of the individual must be taken into account [2].