Serological identification of the cytomegalovirus (CMV) status in children less than 18 months of age is complicated by the variable persistence of maternal antibodies. As T cells are not passively transferred, we attempted to assess whether CMV-specific cellular immunity may be superior to determine the actual CMV status; we also performed a functional characterization of T-cell immunity in childhood. Antibodies from 59 mothers and 168 children were determined, and specific CD4+ T cells were identified by induction of IFN-γ, IL-2, TNF-α, IL-4, and IL-17 after CMV-specific and polyclonal stimulation. Agreement between both tests was perfect for mothers and children more than 18 months. Among infants less than 18 months, 17/30 were concordantly negative. Interestingly, 8/13 seropositive children had detectable CMV-specific T cells, whereas only 5/13 were T-cell negative, indicating passive immunity. CMV-specific T cells from young infants differed in cytokine profiles from that of older age groups, and polyclonal effector T-cell frequencies were higher in young infants with detectable CMV-specific T cells compared with those without. In conclusion, the majority of young infants with CMV-specific antibodies show evidence of true infection, which indicates that passive immunity is overestimated. Our data may have important implications for improved risk stratification and CMV management in infants in the setting of transplantation.
Complications related to cytomegalovirus (CMV) are among the most serious infectious complications after organ transplantation in both adults and children [1-3]. The risk for CMV infection and disease largely depends on the CMV status of the donor and the recipient, which is determined prior to transplantation based on CMV serology. While the determination of CMV-specific humoral immunity is a reliable and widely used method to determine evidence of prior infection, it has limitations in clinical situations where passive antibodies may exist due to transfusion of plasma products or due to natural existence of maternal antibodies in young infants. This raises uncertainty as to the correct assignment of the CMV status in recipients or donors after plasma transfusion or in children below the age of 18 months.
Based on this uncertainty, current guidelines recommend that the highest risk should be assumed for any transplant arrangement where passive antibody titers may exist in either the donor or the recipient, i.e. the recipient should always be assumed as CMV-negative and the donor should be considered as CMV-positive [1, 2]. Based on this clinical dilemma, the development of alternative approaches to assess the actual infection status was recently defined as an important research priority to improve posttransplant management in children .
We have previously shown in adults that the determination of CMV-specific CD4+ T cells is equally effective in assigning the CMV status as determination of CMV-specific antibody titers . This study even provided evidence that the presence of CMV-specific T cells may be superior in situations of ambiguous antibody titers close to the detection limit . CMV-specific T cells may be rapidly quantified directly from whole blood using flow-cytometry based on the induction of intracellular cytokines after antigen-specific stimulation . Conceptually, the limitations of antibody testing in situations of potential passive immunity may be resolved by the determination of CMV-specific T cells, as cellular immunity is not passively transferred. Indeed, in a proof of principle study, we recently showed that this approach allowed for a more accurate assignment of the CMV status in an adult renal transplant candidate who was CMV-seronegative and received plasma products prior to AB0-incompatible transplantation of a graft from his CMV-seropositive mother . Up to now, however, the suitability of this approach for application in children has not been formally assessed and studies on CMV-specific cellular immunity in young infants are limited. We therefore performed a comparative analysis of CMV-specific cellular and humoral immune responses in infants younger than 18 months. Samples from cord blood, from children above 18 months, and from adult mothers served as controls. As T-cell immunity in young infants may have cytokine response profiles different from adults , this study was accompanied by a detailed analysis of Th1, Th2, and Th17 cytokine patterns from CMV specific and polyclonally stimulated T cells across a wide age range.
Detection of CMV-specific CD4 T cells in CMV-seropositive children and mothers
To evaluate detection of CMV-specific CD4 T cells in children with known CMV-IgG serological status that is unlikely to be confounded by passive antibody titers, children above the age of 18 months were recruited. In addition, mothers were analyzed as adult controls. Heparinized whole blood was stimulated with a CMV-antigen lysate for a total of 6 h and CMV-specific cells were identified after intracellular cytokine accumulation as CD69/IFN-γ double-positive CD4 T cells. Stimulations with a control lysate served as negative controls. The frequency of CMV-specific CD4 T cells after control stimulation was always subtracted from that obtained after CMV-specific stimulation. One sample of a CMV-seronegative mother was indeterminate due to excess reactivity in the control stimulation (0.27%), and was excluded from further analysis. Stimulation with Staphylococcus aureus enterotoxin B (SEB) served as a positive control that allowed for an additional assessment of general immune reactivity. Representative examples of a 13-years old CMV-seropositive child with 0.43% and a CMV-seropositive mother with 1.05% CMV-specific CD4 T cells are shown in Figure 1A. Figure 1B shows respective examples of a CMV-seronegative 8-year-old child and mother, where no specific immunity toward CMV was detectable. Clearly detectable immunity toward SEB indicated that both individuals had T cells that were generally able to produce IFN-γ (Fig. 1B).
Strong agreement between CMV-specific T-cell immu-nity and IgG in subjects without passive immunity
To assess the potential of CMV-specific T-cell immunity for identification of a positive CMV-status, the results for all 58 mothers and 109 children older than 18 months were analyzed and compared with those obtained by CMV-IgG serology. When qualitatively comparing the results of the two assays, there was a strong agreement in both groups (Fig. 2A and B). Among children, 24 (22.0%) were positive for both CMV-specific T cells and IgG, and 85 (78.0%) were negative in both assays. Discordant results were not found (Fig. 2A, K = 1.0, 100% sensitivity, and 100% specificity). Among mothers, a higher percentage of individuals (51.7%) was seropositive. As with children, agreement was almost perfect (Fig. 2B, K = 0.93). Two mothers even had clearly detectable T cells toward CMV, although respective IgG titers were below detection limit (Fig. 2B). Although serology may have limitations as a gold standard, this corresponds to a sensitivity of 100% and a specificity of 92.9%. When comparing CMV-specific T cells on a quantitative basis, their frequencies were above detection limit in all seropositive children, although median frequencies were lower as compared with those found in seropositive mothers (Fig. 2C, 0.38%, interquartile range 1.50% in children versus 0.95%, interquartile range 1.51%, p = 0.004). In CMV-IgG negative individuals, CMV-specific T-cell frequencies in all 85 children and in 26/28 mothers were below detection limit. The T-cell frequencies in the two mothers with discordant serology were lower as compared with those with positive serology (Fig. 2C, 0.10 and 0.23%, respectively). Taken together, CMV-specific T cells have similar potential to determine the CMV status as CMV-IgG in both children and adults. In line with previous findings , evidence of CMV-specific T-cell immunity in 2/28 seronegative mothers may even indicate a higher sensitivity of T-cell testing.
CMV-specific T cells are not detectable in cord blood samples
As CMV-specific T-cell immunity in term neonates is generally induced only after birth, cord blood samples should contain CMV-specific IgG depending on the serostatus of the mother, but should be negative for CMV-specific T cells. To assess this hypothesis, 29 cord blood samples from mature newborns were analyzed as controls for infants where the presence of specific T-cell immunity was unlikely. These samples were also compared with results from their mothers (n = 28, of whom one was a mother of twins). As expected, none of the 29 samples showed evidence of CMV-specific T-cell immunity, although 11 (37.0%) were CMV-IgG positive (Fig. 3A). CMV-specific IgG were exclusively found in children from CMV-seropositive mothers (Fig. 3B). In addition, there was no seropositive mother, where the titer of passive immunity in the respective child was below detection limit (Fig. 3B). These results clearly demonstrate that maternal CMV–IgG are invariably transferred to the child, indicating that all children of CMV-seropositive mothers should have passive humoral immunity at birth.
The presence of CMV-specific T cells in seropositive children <18 months indicates a true CMV status
We next recruited 30 infants below the age of 18 months, where the true CMV status could not be unequivocally determined due to the potential existence of passively transferred maternal IgG. All children were comparatively analyzed for cellular and humoral immunity (Fig. 4). A total of 17 infants (56.7%) were seronegative, and none of those had evidence for CMV-specific T-cell immunity (Fig. 4A). Conversely, 13 infants were CMV-IgG positive. Among these, eight (61.5%) had detectable CMV-specific Tcells, which is indicative of a truly positive CMV status. In contrast, 5/13 showed no evidence of specific T-cell immunity that may be considered as a passively positive serostatus without evidence of true CMV infection. Dot plots of a 1.1-month-old infant and its 24-year-old mother are shown as examples of the five infants with passive immunity (Fig. 4B). These examples show no CMV-specific Tcells in the infant whereas such T cells were clearly detectable in the mother. This infant also reacted toward the polyclonal stimulus SEB, although this was only observed in 18/30 cases in this age group. In addition, dotplots of a CMV-seropositive infant (9.7 months of age) and its mother (26 years) are shown as examples of the eight seropositive infants with detectable CMV-specific T-cell responses (Fig. 4C). Although seropositive infants with detectable T-cell immunity appeared to be older than those who were T-cell negative (0.81 ± 0.60 versus 0.56 ± 0.68 years), this difference did not reach statistical significance (p = 0.50). Nevertheless, this is consistent with the assumption that CMV infection and hence the likelihood of a truly positive CMV status increases with age.
IFN-γ is the best readout cytokine for CMV-specific T-cell immunity in all age groups
Polyclonal stimulations with SEB revealed that most cord blood samples and some samples from infants under 18 months did not produce any IFN-γ. To assess whether the failure to detect CMV-specific T cells in those samples was due to a general inability to produce cytokines, we analyzed whether other Th1, Th2, or Th17 cytokines such as IL-2, TNF-α, IL-4, or IL-17 may be induced in a CMV-specific manner and thereby increase the sensitivity of T-cell testing particularly in the young age groups. Results were separately analyzed for individuals that did not or did react by producing IFN-γ after stimulation with the CMV lysate (Fig. 5A and B). Among individuals who produced IFN-γ, CMV-specific IL-4 induction was only observed in 15/32 mothers, 9/21 children, and in none of the infants, and IL-17 was not induced in a CMV-specific manner (data not shown). When analyzing Th1 cytokines, the majority of samples from individuals who produced IFN-γ were also positive for TNF-α and IL-2 across all age groups (Fig. 5B). Conversely, cells from most seronegative mothers and children who failed to produce IFN-γ did not show any pronounced production of TNF-α and IL-2 (Fig. 5A, 25/27 mothers and 83/85 children). This also held true for the IFN-γ negative samples from infants and for all cord blood samples (Fig. 5A). Thus, although CMV-specific T cells produce TNF-α and to a lesser extent IL-2 (Fig. 5B), these two cytokines are slightly less sensitive and do not provide additional independent evidence of CMV-specific T-cell reactivity in samples that do not react with IFN-γ. Therefore, IFN-γ appears to be a reliable readout cytokine to identify CMV-specific T-cell immunity across all age groups.
As shown in Fig. 5B, frequencies of cytokine producing CMV-specific T cells were lowest in infants and showed a significant increase with increasing age (i.e. p = 0.0008 for IFN-γ positive T cells). However, when qualitatively analyzing profiles of triple, dual, and single cytokine producing cells after CMV-specific stimulation, the largest fraction of reactive T cells was positive for IFN-γ and similar in magnitude in all age groups (Fig. 5C, black bars), demonstrating that the CMV-specific T-cell response is dominated by IFN-γ. Interestingly, however, in line with a more recent CMV contact, CMV-reactive T cells from infants exhibit a lower percentage of multifunctional triple positive and a concomitantly higher percentage of IFN-γ single-positive T cells (Fig. 5C, lower panel).
Polyclonal neonatal Th1 and Th17 responses are higher in subjects with CMV-specific T-cell immunity
To further assess whether cord blood samples or samples from some infants were generally unreactive toward other T-cell stimuli, the induction of the cytokines IFN-γ, TNF-α, IL-2, IL-17, and IL-4 was analyzed after polyclonal stimulation with SEB (Fig. 6). Again, results were stratified for individuals who did or did not react by producing IFN-γ after stimulation with the CMV lysate (black and white circles, respectively). This analysis demonstrates that SEB reactive T cells from infants who were unresponsive toward CMV showed the ability to produce TNF-α and IL-2, whereas the number of IFN-γ reactive samples in this group was considerably low. Interestingly, samples from infants with detectable immunity toward CMV had significantly higher frequencies of SEB-reactive T cells, a finding that applied not only for IFN-γ, but also for TNF-α, IL-2, and IL-17. Of note, no such difference was found for IL-4 (Fig. 6). Moreover, no difference was observed in the levels of any of the cytokines between CMV-positive and CMV-negative older children and mothers. Together this suggests that CMV infection in early childhood contributes to the induction of cytokine-producing effector Th1 and Th17 cells that may already be revealed at a polyclonal level.
Adult and pediatric transplant patients share similar risk factors for the development of CMV disease after transplantation, which is largely determined by the CMV infection status of the donor and the recipient [1, 2]. Yet, the determination of the CMV status by serological analysis in young infants is hampered by the potential existence of passively transferred maternal antibodies. In this study, we provide evidence that the determination of CMV-specific CD4 T cells may represent a promising alternative to serology in this clinical situation. Among infants with potential passive immunity, more than 60% of infants less than 18 months had detectable CMV-specific T cells as evidence of a true CMV status, which indicates that the extent of passive immunity is overestimated. Our data may even indicate that the age threshold to recommend assumption of passive immunity may be reduced to 12 months, as among the 13 seropositive individuals below 18 months, four infants were older than 12 months and all had detectable CMV-specific cellular immunity. This is in line with decay kinetics of maternal antibodies toward other viruses that largely seem to decrease within 9–11 months . Thus, although serology is generally a reliable diagnostic assay to assess the CMV status in most clinical situations in adults, this study highlights a clinically relevant area where the determination of CMV-specific T cells may be superior.
The use of CMV-specific T cells to assign the actual infection status in infants with potential passive immunity may be of particular value in the setting of transplantation of newborns. Due to the uncertainty of serological analysis in infants under 18 months, current guidelines recommend assuming the highest possible risk level for donors or recipients in this age group [1, 2]. Consequently, donors younger than 18 months should be assumed to be CMV positive if the CMV serological test is positive. Likewise, all CMV-seropositive recipients under 18 months should be regarded as CMV-naïve. Given that these classifications have direct implications for the management of treatment strategies, T-cell analyses may reduce the need for universal prophylaxis in arrangements where the recipient is clearly positive or the donor is clearly negative. Together this would facilitate more specific targeting of prophylaxis in constellations involving young infants, and overall may aid in reducing the burden of antiviral drugs with attendant side effects. On the other hand, as children are more often CMV-naïve at the time of transplant as compared with adults, the absence of cellular immunity may be used to more specifically identify high risk groups likely benefiting from antiviral prophylaxis. In more general terms, other areas where T-cell analysis may be advantageous include constellations of any age, where pretransplant serology is equivocal. In this situation, guidelines recommend that equivocal serology in the donor should be assumed positive and equivocal results in the recipient should be assumed negative . Finally, as shown in a proof-of-principle study, the presence or absence of CMV-specific CD4 T cells may serve as a more accurate surrogate for the actual CMV status in transplant recipients with passively positive CMV-serology due to infusion of plasma preparations .
Up to now, antigen-specific analyses of cellular immune responses toward CMV have largely been restricted to adults with and without immunodeficiency [5, 9-14], and studies in children are limited [15-18]. CMV-specific T-cell levels in asymptomatic individuals show an age-dependent increase with median frequencies of approximately 1–2% [11, 14, 15]. In transplant recipients, the failure to maintain or develop CMV-specific CD4 T cells has been associated with poor viral control and shown to be predictive of CMV disease [5, 11-13]. When directly assessing CMV-specific T cells in children, we show that both their frequency and cytokine profile are consistent with a more recent infection, in that overall frequencies are lower in young infants and increase with age, and the predominance of IFN-γ single-positive T cells as opposed to multifunctional T cells is similar to that observed in transplant recipients after CMV reactivation [13, 19]. Similarly, a study in 13 seropositive children aged 1.5–4 years found that the percentage of CMV-reactive CD4 T cells was lower than in adults . Interestingly, these differences of specific CD4 T-cell immunity may account for the prolonged shedding of CMV in young age groups [15, 20], although this is generally not associated with clinical symptoms. This may be due to the concomitant induction of robust CMV-specific CD8 T cells [15-18] and to the presence of maternal antibodies that may contribute to limitation of overt viral replication. In general, CMV infection is a major driving force for typical immune deviations of cellular immune responses that manifests as clonal expansions and altered phenotypes of overall effector and memory T cells [14, 21, 22]. Interestingly, this already seems to be evident in adolescents, where bulk T cells showed distinct phenotypical properties in close association with the CMV serological status . In young infants, CMV infection may have a quantitative effect on the size of the overall effector T-cell pool, as we show that frequencies of polyclonal T cells among young infants are higher in those with detectable CMV-specific immunity as compared with those without. This suggests that the emergence of polyclonal effector T cells in early childhood may be considered to mirror an individual's natural exposure to infectious agents, whereas this effect is less prominent in older age groups, where a number of additional infections and vaccination responses may have accumulated over time.
The low extent of cumulative pathogen exposure in young infants is important when interpreting test results of T-cell assays in young infants. In recent years, commercial IFN-γ based in vitro assays for the determination of Mycobacterium tuberculosis-specific cellular immune responses have been developed . In most assays, the cytokine IFN-γ is used as a readout in an enzyme-linked immunosorbent assay (ELISA) or an enzyme-linked immunosorbent spot (ELISPOT) assay to identify a specific immune response . As in the present study, these IFN-γ release assays (IGRAs) are performed along with a polyclonal mitogen stimulation to control for general T-cell reactivity. In this context, an indeterminate result is defined as a poor mitogen control reaction in the absence of an antigen-specific T-cell response. Notably, IGRAs may also be used in children [25-29], although many studies reported a high number of indeterminate results, especially in children below the age of 5 years [30-32]. Therefore, the value of IGRAs in young children is discussed controversially . Together with a known bias of neonatal immunity toward Th2 cells , this has stimulated discussions as to whether cytokines other than IFN-γ may be more suitable as a readout system in young children. In the context of CMV, we now show that CMV-specific T cells do produce other cytokines such as IL-2 and TNF-α, but neither IL-2 nor TNF-α was suitable to identify CMV-specific T-cell responses from samples that did not produce any IFN-γ. In the absence of severe signs of immunodeficiency, this suggests that the lack of IFN-γ reactivity after antigen-specific or polyclonal stimulation does not imply test failure but is rather indicative of a limited infectious history and hence a truly negative CMV infection status. Conversely, the capability to respond toward an antigen-specific or polyclonal stimulus increases with age and largely reflects the extent of natural exposure to various infections, including CMV.
In line with a previous study in adults , we now show that the determination of CMV-specific T cells may be comparable or even superior to serology for defining the CMV status in adults, children, and infants. We even show that CMV-specific T cells were detectable in two adults in the absence of positive CMV-IgG titers, although the sensitivity of serology may have been increased by using additional ELISA tests . In children with potential passive antibody titers, other tests to identify evidence for CMV infection may include the analysis of CMV viral load from urine or saliva, assessment of IgM or IgG avidity testing. Although viral load analysis may be a valuable approach due to prolonged viral shedding [15, 34], a negative test may result from intermittent viral shedding and thus does not exclude history of infection. CMV–IgM testing was of insufficient sensitivity to identify culture-proven CMV-infected infants [35-37]. CMV–IgG avidity testing could be performed to potentially differentiate between maternally derived IgG from low-avidity IgG newly generated by the infant . However, given that both types of IgG may concomitantly exist in samples from young infants with actual infection, the sensitivity does not seem to be sufficiently high to detect low-avidity IgG, especially in the young age group . Unfortunately, restrictions in sample volume due to ethical obligations precluded this type of analysis in our study.
In conclusion, T-cell analysis in individuals with potential passive immunity may represent a novel area of application for cell-mediated immunity assays that are increasingly evaluated as monitoring tools after transplantation . As a limitation of our study, we did not perform any T-cell analyses in pediatric candidates for transplantation to actually assess whether this approach may also be applicable in children with comorbidities, although a small study recently provided evidence that CMV-specific T cells may be detectable in pediatric transplant recipients . Thus, larger cohorts of pediatric transplant candidates are required to clearly evaluate the performance and potential limitations of T-cell analyses as alternative to serology in infants in the setting of transplantation. In general, the assay used in this study can be applied in a clinical setting, as CMV-specific T cells can be quantified directly from small volumes of less than 1 mL of whole blood by the use of standard flow cytometers within one day. Consequently, it may be applied in prospective studies to guide antiviral management decisions in adults and children with potential passive immunity. In addition, as shown in adults [5, 11], this assay may also be used for the individual monitoring of the maintenance of CMV-specific immunity in the posttransplant period.
Materials and methods
Demographical data of the study cohorts are summarized in Table 1. The study was conducted among 59 mothers (mean age 29.67 ± 5.78 years) and 168 immunocompetent children below the age of 18 years (mean age 7.02 ± 6.37 years). Among those, 29 healthy term neonates born after uneventful pregnancy were tested at the day of birth (cord blood samples), 30 were below the age of 18 months (mean age 0.48 ± 0.51 years), and 109 were above 18 months (mean age 10.68 ± 4.90 years). Infants below the age of 18 months were considered as potentially passively seropositive based on current guidelines . None of the individuals showed signs or symptoms of active cytomegalovirus infection. Infants and children were recruited from pediatric patients of various ages who were fully immunocompetent and had no chronic diseases. The study was approved by the local ethics committee and all individuals and in the case of children their parents gave written informed consent.
Table 1. Demographic characteristics of study populations
Female gender (%)
Mothers of either newborns or children below the age of 18 months; age is given as mean ± SD.
Quantification and characterization of CMV-specific T cells
T cells from whole blood were stimulated for 6 h in vitro as previously described [4, 6, 13, 39] using titered amounts of CMV-antigen (32 μL/mL, Virion, Würzburg), and antigen-specific T cells were flow cytometrically quantified and phenotypically characterized by induction of the activation marker CD69 as well as Th1 cytokines IFN-γ, IL-2, and TNF-α. In addition, samples were analyzed for the expression of IL-4 and IL-17 (all antibodies from BD Biosciences, Heidelberg, Germany). Samples incubated with control antigen (Virion) served as negative controls. Stimulation with SEB (Sigma, Deisenhofen, Germany) served as positive control. All stimulations including negative and positive controls were carried out in the presence of 1 μg/mL of costimulatory antibodies anti-CD28 and anti-CD49d. After 2 h of stimulation, 10 μg/mL brefeldin A (Sigma) was added to accumulate cytokines intracellularly. After a total of 6-h stimulation, leukocytes were fixed and erythrocytes were lysed using BD lysing solution according to the manufacturer's instructions. Thereafter cells were processed for staining with fluorochrome-conjugated monoclonal antibodies as previously described [4, 6, 13, 39].
Quantification of CMV-specific antibody responses
CMV-specific antibody titers were analyzed using a quantitative IgG-ELISA according to the manufacturer's instructions (Euroimmun, Lübeck, Germany).
Statistical analysis was carried out using GraphPad Prism 5.03. The nonparametric Mann–Whitney test was applied to analyze differences in T-cell frequencies among children and mothers. The nonparametric Kruskal–Wallis test with Dunn's posttest was used to analyze respective differences among more than two groups. Kappa statistics were calculated as described before .
The authors thank Candida Guckelmus and Katharina Schmitt for excellent technical assistance. We also acknowledge the support of the staff of the Department for Pediatrics, and all volunteers for their participation in the study. The study was supported in part by grants from HOMFOR and the Roche Organ Transplant Research Foundation (ROTRF) to M.S.
Conflict of interest
M.S. acts as a consultant for Genentech, and M.S. and U.S. have received honoraria from Roche Pharma and Astellas; B.C.G. has received honoraria from Roche Diagnostics and Euroimmun. All other authors declare no financial or commercial conflict of interest.