In the present study, CMV seropositivity was associated with indicators of glucose regulation in terms of more frequent diagnoses of type 2 diabetes, elevated levels of HbA1c and non-fasting glucose in very elderly people infected with this virus. The titer of CMV IgG antibody, however, was not significantly associated with these indicators. Our findings suggest a role for CMV infection in the pathogenesis of type 2 diabetes in the elderly.
CMV might be involved in accelerating pancreatic failure to compensate for insulin resistance via at least two possible mechanisms. First, it could influence the pancreatic cells directly; secondly, it might act indirectly by influencing the immune system which in turn affects the pancreas. Consistent with the first possibility is the report that CMV may infect and reside in pancreatic cells without causing cytopathic effects but nonetheless influencing insulin production directly after repeated reactivations . Additionally, infection of human pancreatic β-cells with CMV induced the release of pro-inflammatory cytokines and increased cellular immunogenicity . The indirect effects of CMV could be exerted via infected monocyte production of IL-1β which induces TNF-α production in human pancreatic duct cells, driving cells into apoptosis and thus compromising β-cell function [24, 31]. Other components of the immune system, influenced by prolonged CMV infection, could hypothetically also contribute to a more pro-inflammatory environment, which is an important feature of type 2 diabetes . CMV seropositivity is associated with accumulations of potentially senescent late- differentiated T-cells and elevated numbers of CD4+ and CD8+ effector cells  which are more likely to produce pro-inflammatory cytokines .
Thus far, no relationships between glucose regulation and CMV seropositivity had been confirmed at the population level [27–30]. Three earlier studies investigated notably younger participants with comorbidities in relatively small cohorts [27–29]. Age, ethnicity, income and education (socioeconomic factors) have not always been considered important confounders for CMV and diseases, although this is very likely to be the case especially for the latter [33, 34]. In 1000 participants aged 45–84 years in the Multiethnic study of atherosclerosis (MESA), associations between CMV infection, pathogen burden > 3 and diagnoses of type 2 diabetes were indeed observed. After adjustment for demographic covariates (race/ethnicity and education), however, all associations became non-significant, suggesting no aetiological role for pathogens/CMV in the occurrence of type 2 diabetes .
Three possible explanations for why positive relationship of CMV seropositivity with glucose regulation emerges in the oldest old can be given. First, the direct deleterious effects of CMV infection on pancreatic cells might only become significant after a long period of activation and reactivation of CMV and therefore only detectable in the oldest old. Second, systemic inflammation may interfere with the action of insulin on cells by suppressing intracellular insulin signal transduction . This possible indirect systemic effect of CMV infection on the immune system would also take time to develop. Third, our study population of 85-year-olds is of course highly selected for longer-than-average survival, and has thus already lost individuals with other known risk factors for cardiovascular diseases and type 2 diabetes. These risk factors could have overshadowed the impact of CMV infection in glucose regulation in younger people.
An alternative explanation posits that hyperglycaemia may impair host defenses, predispose to infection and therefore leads to a higher seroprevalence of CMV in diabetic patients . Thus, higher prevalence of CMV would be a result, not a cause, of disease. However, CMV infection is often acquired during childhood. Once the host is infected, the virus persists in various tissues. The chance of becoming infected with CMV is highly influenced by socioeconomic conditions and close contacts allowing the virus to spread [33, 34]. It is therefore unlikely that diabetes precedes CMV infection, but it still cannot be ruled out that there are undetected predisposing factors to both diabetes and CMV infection.
In the present study, CMV IgG antibody titer did not associate significantly with diagnoses of type 2 diabetes in the oldest old. Although CMV antibody level has been reported to associate with mortality among community-dwelling older adults [37, 38], it is still unclear what the clinical implications of CMV antibody level on the onset of diseases may be and what the effect of chronological age on CMV antibody level is [39, 40]. Recently, CMV antibody level has been reported not to be related to any major genetic determinants either .
The cross-sectional nature of the relationship between CMV seropositivity and indicators of glucose regulation is one limitation of our study. One cross-sectional study is not sufficient to prove whether the contribution of CMV infection is causative. This relationship should be confirmed in studies with a prospective design in younger participants with long-term follow-up and regular monitoring of incidence of CMV infection. A second limitation is that the blood samples were collected under non-fasting conditions. It is therefore likely that we have underestimated the effect of CMV seropositivity on glucose level, which would increase the robustness of our findings. Still, the old age of the participants could limit the applicability of these findings to the general population, but the fact that our study population is relatively large and all participants were 85 years of age could also be seen as a strength. It might be feasible to study the effect of CMV seropositivity only in such special participants as the impact of CMV may become pronounced after years of latent infection plus periodic reactivations. The large size of our study and the homogeneity of the age could have made the impact of CMV easier to determine.
Our finding that CMV seropositivity is associated with more diagnoses of type 2 diabetes in the oldest olds leads to the hypothesis that CMV seropositivity may facilitate the onset of type 2 diabetes in the long term. It is crucial to demonstrate the association using a prospective study design, starting with a younger study population, as CMV infection may be a “hidden” cause of morbidity .