CMV reactivation following HSCT is a consequence of the massive immunosuppression and insufficient lymphocyte reconstitution, and occurs more frequently after T cell depletion of the graft or after transplantation of CMV-seropositive patients with grafts from seronegative donors. CMV reactivations still contribute significantly to post-HSCT morbidity, despite advancements made to reduce CMV disease by monitoring for viral reactivation and pre-emptive therapy.
Monitoring CMV-CTL reconstitution can be achieved in about 85% of all transplanted patients 
, using commercially available CMV-tetramers. A large cohort of 278 patients could be monitored with the tetramers, 198 (71%) actually developed detectable CMV-CTLs following HSCT. The CMV-CTL levels detected varied greatly (1–1235 per µl blood), and correlated only weakly with reconstitution of the CD3+
T cells, as described previously 
. Restoration of CMV-specific immunity is frequently analyzed using tetramer staining 
. However, absolute values for the CMV-CTL levels required to protect the patient from reactivation are still actively debated to date. Our data indicate that CMV-CTL numbers vary considerably for individual combinations of HLA molecules and CMV epitopes. Many authors focus on the HLA allele that is most common in the Caucasian population, HLA-A*0201. Using the HLA-A*0201-NLVP tetramer, 10 to 20 CMV-CTLs per µl blood by day +60 were described as being protective 
. Our data for CMV-CTL levels detected by HLA-A*0201-NLVP are similar, with a mean level of 10 CMV-CTL per µl blood (). However, analyzing additional HLA molecules in our large patient cohort showed that different HLA types yield quite different median values of detectable CMV-CTLs (). We and others observed that CMV-CTL levels detected by HLA-A*0101 (pp50243–255
) or HLA-B*3501 (pp65123–131
) were considerably higher or much lower than CMV-CTL levels detected by HLA-A*2402 (pp65341–349
), which yielded a mean of 4 per µl blood () 
. We detected no CMV-CTLs using the HLA-A*2402 tetramer in 73% of the tests. This is in agreement with data published by others, who detected low levels of CMV-CTLs using the HLA-A*2402 tetramer when monitoring CMV-CTL reconstitution 
. We also identified no correlation between the immune reconstitution of CD3+
T cells and CMV-CTL levels detected using the HLA-A*2402 tetramer. Since CMV reactivation did not occur more frequently in patients expressing the HLA-A*2402 molecule. We speculate that either lower levels of HLA-A*2402-corresponding CMV-CTLs are needed provide protection 
or the immune-response in HLA-A*2402-positive individuals is dominated by other epitopes 
. In addition to the different median levels of CMV-CTLs, we found that expression of different HLA alleles may interfere with the expansion of T cells specific for other HLA peptide combinations. For example, in individuals expressing both alleles HLA-A*0201 and HLA-B*0702, the HLA-B*0702-pp65 CMV-CTL response prevailed 
. Thus, in these patients CMV-CTLs binding to HLA-B*0702-pp65 may be the dominant response to CMV-CTLs recognizing HLA-A*0201-NLVP, since the level of CMV-CTLs binding to HLA-A*0201-NLVP is significantly higher in patients expressing only the HLA-A*0201 allele (). Our data imply that there may be more interference between the different HLA molecules and/or CMV epitopes expressed than monitoring techniques currently in use have detected. We documented that changes in the most abundant epitope-specific CMV-CTL population occurred over time. These alterations were solely dependent on particular HLA molecules. Interestingly, in 14 of 19 patients the response directed against the HLA-B*0801 epitope of the immediate early (IE)-1 protein prevailed. We hypothesize that a shift from CMV-CTLs recognizing pp65 epitopes to IE epitopes may delineate the shift from short-term to long-term protection against CMV reactivation 
. Taken together, the absolute number of CMV-CTLs detected in patient blood samples appears to be depend on the HLA expressed in the patient, the tetramers used for detection of CMV-CTLs and the time-point of the measurement, thus “protective” CMV-CTL levels vary considerably. Taken together, a “protective” CMV-CTL count cannot be defined, since the CMV-CTL level varies a) for the different tetramers used in detection, b) for different HLA combinations expressed in the patient (e.g. A0201 and B0702) and c) for a single tetramer over time. Thus, more information than an absolute number of all CMV-CTLs is necessary to define what is protective against reactivation, and levels defined as protective may vary considerably based on the tetramer or tetramer set used for monitoring.
The most important finding of our study was that significantly more patients who experienced no CMV reactivation had at least 1 CMV-CTL per µl blood before day +75 following HSCT (range: +50 to +75), compared to patients who experienced CMV reactivations (). Interestingly, increasing the threshold to 5–10 CMV-CTLs per µl blood did not yield better discrimination between patients experiencing or not experiencing CMV reactivation in our cohort. Lillieri et al.
and Tormo et al.
detected similarly low thresholds (1–3 CMV-CTLs per µl blood) using functional assays for IFN-γ and IL-2 secretion or ELISPOT assays 
, implying that, indeed, detection of 1 to 3 CMV-CTLs per µl blood may indicate the hallmark of a functional immunity against CMV.
In addition to the currently controversial CMV-CTL quantity that provides protection against viral reactivation, the influence of CMV reactivations on immune reconstitution is also widely debated. While Chen et al.
argue that CMV reactivation boosts the reconstitution of CMV-CTLs 
, others find no influence of CMV reactivation on CMV-CTL reconstitution 
. In our large prospective cohort, patients reactivating CMV (n
117/278) had higher median CMV-CTL numbers than patients without CMV reactivations (n
161/278), implying a significant influence of CMV reactivation on levels of CMV-CTLs detected. Since this expansion of CMV-CTLs after CMV reactivation does not allow for the definition of a minimal protective CMV-CTL level, even using only 1 tetramer, we searched for other means to differentiate between patients in whom CMV reactivation occurred only once, was recurrent or did not occur at all. We analyzed CMV-CTLs at two time-points, namely day +30 (+/−15 days) and +60 (+/−15 days), and measured the expansion of CMV-CTLs within that time period. Patients with a protective response after the first CMV reactivation showed a significantly increased expansion of CMV-CTLs within this time period compared to patients with recurrent CMV reactivations. Patients without CMV reactivation also expanded CMV-CTLs during this time interval, but to a lesser extent than those with 1 CMV reactivation. The inability of CMV-CTLs to expand after CMV reactivation may be due to even minor HLA incompatibilities between donor and recipient. Our results indicate that analyses using a single tetramer at only one time point, for instance on day +60, do not allow prediction of pending or, more importantly, recurrent CMV reactivations. Despite the fact that patients without CMV reactivation showed an earlier reconstitution of at least 1 CMV-CTL per µl blood, the wide range of CMV-CTL levels does not allow definition of definitive protective value that is broadly applicable for all HSCT patients. However, monitoring the level of CMV-CTL expansion between days +30 and +60 (+/−15 days) after CMV reactivation can indicate successful restoration of CMV immunity.
In summary, our results show that sequential tetramer monitoring rather single time point cut offs of the post-transplant CMV-CTL immune reconstitution allows a more accurate interpretation of an individual patient's response to CMV. In addition, CMV-CTL expansion after the first CMV reactivation indicates recurrence of CMV reactivation even in R+/D+ patients after allogeneic HSCT. Analysis of the CMV-CTL expansion rate may facilitate implementation of patient-specific antiviral strategies, including adoptive transfer of CMV-CTLs to recipients unable to respond to CMV reactivations.