Functional Impairment of Cytomegalovirus Specific CD8 T Cells Predicts High-Level Replication After Renal Transplantation


  • Present address for F. M. Mattes: Department of Virology, Barts and the London Hospital NHS Trust, Whitechapel London E1 1BB.


Human cytomegalovirus (HCMV) remains an important cause of morbidity after allotransplantation, causing a range of direct effects including hepatitis, pneumonitis, enteritis and retinitis. A dominant risk factor for HCMV disease is high level viral replication in blood but it remains unexplained why only a subset of patients develop such diseases. In this detailed study of 25 renal transplant recipients, we show that functional impairment of HCMV specific CD8 T cells in the production of interferon gamma was associated with a 14-fold increased risk of progression to high level replication. The CD8 T-cell impairment persisted during the period of high level replication and was more prominent in patients above 40 years of age (odds ratio = 1.37, p = 0.01) and was also evident in dialysis patients. Threshold levels of functional impairment were associated with an increased risk of future HCMV replication and there was a direct relationship between the functional capacity of HCMV ppUL83 CD8 T cells and HCMV load (R2= 0.83). These results help to explain why a subset of seropositive individuals develop HCMV replication and are at risk of end-organ disease and may facilitate the early identification of individuals who would benefit from targeted anti-HCMV therapy after renal transplantation.


human cytomegalovirus


lymphocytic choriomeningitis virus


Historically, human cytomegalovirus (HCMV) infection has been a significant clinical problem following organ transplantation. The virus can lead to a range of overt clinical symptoms such as prolonged fever, hepatitis, enteritis, pneumonitis and retinitis (termed the ‘direct effects’) as well as a number of ‘indirect effects’ including acute and chronic organ rejection, accelerated cardiovascular disease, new onset of posttransplant diabetes mellitus and bronchiolittis obliterans (1,2). In recent years, the development of prophylactic and pre-emptive therapeutic patient management strategies has significantly reduced the mortality attributed to HCMV and impacted positively on morbidity (3–6). A variety of anti-HCMV agents (valaciclovir, ganciclovir [intravenous and oral] and most recently valganciclovir) have been shown in controlled clinical trials to be effective for prophylaxis of solid organ transplant recipients at high risk of HCMV disease, such as HCMV seronegative patients receiving an organ from a HCMV seropositive donor (7–10). However, HCMV disease still occurs in a subset of patients after prophylaxis is stopped, and disease at this stage can still be clinically severe (4). Pre-emptive therapy initiated on the basis of detection of high levels of virus in blood by PCR or antigenemia assays has also been shown to minimize disease and provide rapid control of HCMV replication (10–12), although it may not impact on the longer term consequences of HCMV infection highlighted above. In addition to the high-risk patients mentioned above, seropositive recipients are also at risk of HCMV disease, especially if they receive an organ from a seropositive donor and become reinfected (13). As these seropositive patients form the majority of patients in many areas of the world, the absolute numbers experiencing HCMV disease cannot be ignored.

In vivo, HCMV replication is highly dynamic with doubling times between 0.4 and 2 days (14). The basic reproductive number (Ro; the number of newly infected cells in the host arising from one infected cell) for HCMV in immune liver transplant recipients is ∼15 but is reduced to 2.4 in patients who are already HCMV seropositive (15). Several studies, including many from our group, have shown that high level HCMV replication, as detected by high viral loads in blood by quantitative virological assays, and persistant viral load with the equivalent viral turnover are a dominant risk factors for HCMV disease when examined in multivariable statistical analyses (16–20).

T-cell immunity is a fundamental effector process controlling HCMV replication in vivo (21–24). The CD8 T-cell immune response to HCMV appears to be dominated by responses to the tegument protein ppUL83 and to the immediate early protein pUL123 (22,25) although recent data using peptides representing the entire HCMV proteome reveal that many proteins are targeted by the host CD8 T-cell response (26). Studies using class I HLA tetramer reagents have shown that the human host devotes a high proportion of the total CD8 T-cell response to the control of HCMV. In healthy individuals this response averages ∼1% and can reach levels of 10–50% in acute infection in immunocompromised hosts (27–30). In transplant patients, CD4 T-cell help is a key factor to facilitate CD8 T-cell control of HCMV replication (31,32). While we have a substantial amount of qualitative data relating to the CD8 T-cell immune control of HCMV after transplantation, there are relatively few studies addressing whether specific quantitative malfunction of these CD8 T cells, in addition to their frequency and absolute number, also contributes to the failure to suppress replication and whether these assessments provide prognostic value. We, therefore, set out to investigate the functional capacity of HCMV specific ppUL83 specific T cells in patients who did, or did not, develop high level HCMV replication after renal transplantation. Our results show that patients who progressed to high level replication had a HCMV ppUL83 CD8 T-cell population that was functionally impaired compared to patients who kept HCMV replication suppressed to undetectable levels.

Materials and Methods

Study population and design

A prospective natural history study, investigating the role of HCMV specific CD8 T cells in controlling HCMV virus replication postrenal transplantation, was carried out in a single center (Royal Free Hospital, London, UK) between October 2000 and December 2001. In total, 25 renal transplant patients entered the study, with the following HLA class I types: HLA-A * 0201 (n = 14), HLA-B * 0702 (n = 5), HLA-B * 0801 (n = 7) and HLA-B * 3501 (n = 6). A further 10 patients were excluded since they did not have class I HLA types with appropriate HLA tetramer reagents available. Preservative free heparin blood samples were collected for isolation of peripheral blood mononuclear cells (PBMC) on a weekly basis. PBMCs were isolated on the same day and cryopreserved for immunological studies. In addition, we analyzed PBMCs taken from 10 healthy HCMV seropositive controls, 10 patients with common variable immunodeficiency to provide a suitable control group with a specific deficit in antibody production and 10 patients who were receiving dialysis but had not been transplanted. Citrated blood was collected for HCMV surveillance three times a week when the patient was hospitalized or whenever attending an outpatient clinic. The study was approved by the Local Research Ethics Committee (Institutional Review Board), and all clinical investigations were conducted according to the principles expressed in the Helsinki Declaration.

DNA extraction, qualitative and quantitative HCMV PCR

DNA was extracted from 200 μL whole blood using a Qiagen extraction kit (Minden, Germany) according to manufacturer's instructions. Qualitative HCMV PCR was carried out as described elsewhere (16), with a lower sensitivity level of 200 genomes/mL whole blood. Quantification of HCMV was performed with a TaqMan (ABI)–based method (Cheshire, UK) adapted from our previously published method (33). In the current study, high-level replication (henceforth referred to as viremia) was defined as viral loads above 200 genomes/mL whole blood.

Antiviral and immunosuppressive therapy

None of the patients received antiviral prophylaxis for HCMV. Patients with two consecutive positive HCMV PCR samples (>200 genomes/mL blood) received pre-emptive anti-HCMV therapy with either i.v. ganciclovir (5 mg/kg bd adjusted for creatinine clearance, n = 7 patients) or i.v. foscarnet plus i.v. ganciclovir each at half dose (n = 3) as part of a previously published randomized clinical trial which showed that these two treatments were equipotent (12). Antiviral therapy was continued until two consecutive PCR negative samples were obtained. Initial immunosupression consisted of prednisolone plus tacrolimus, alone or with sirolimus, basiliximab, or mycophenolate mofetil, prednisolone plus cyclosporin alone or in combination with either sirolimus or azathioprine. The majority of the patients (72%) received induction therapy with baxiliximab as part of their immunosuppressive regimen.

Isolation of PBMCs from whole blood

Whole blood (20 mL) was collected using the monovette system, containing preservative free heparin. PBMCs were isolated by centrifugation through Ficoll-Paque (600 g, 20 min at room temperature; Amersham-Pharmacia Biotec, Bucks, UK). The supernatant containing mononuclear cells was transferred to a new 50-mL tube and washed three times in RPMI 1640 supplemented with 2-mM glutamine. Purified PBMCs were counted and cryopreserved in fetal calf serum (FCS) containing 10% dimethyl sulphoxide, at a concentration of 3 × 106 to 6 × 106 per mL, and stored in liquid nitrogen.

Tetrameric complex and surface marker staining

HCMV specific tetrameric complexes were constructed for HLA-A * 0201, HLA-B * 0702, HLA-B * 0801 and HLA-B * 3501, as described elsewhere (34), using the appropriate ppUL83-derived immunogenic peptides (A2: NLVPMVATV, B7: TPRVTGGGA, B8: DANDIYRIF, B35: IPSINVHHY). All samples from an individual patient were analysed in the same assay to minimize variation. PBMCs were thawed by placing the vial in a 37°C water bath. Lymphocytes were washed in RPMI and R10 (RPMI 1640 containing 10% FCS), resuspended in 2 mL of R10 and incubated for 2 to 3 h at 37°C prior to staining. Lymphocytes were counted and the cell number adjusted to 2 × 106–4 × 106 cells/mL. Fifty microliters of the cell suspension were transferred into a FACS tube (Becton Dickinson) together with 0.3 μg (in PBS) of the HLA-A2 or HLA-B35 tetramer or 1 μg (in PBS) of the HLA-B7 or HLA-B8 ppUL83-specific tetrameric complexes. After incubation for 30 min at 37°C, cells were subsequently stained for the surface marker CD8 by adding 5 μL of antihuman Tri-color labeled CD8+ antibody (Becton Dickinson, Oxford, UK). Surface marker staining was carried out on ice for 30 min and the cells then washed with PBS/0.1% sodium azide and resuspended in 200 μL of 2% freshly prepared paraformaldehyde. All results were acquired on a FACScalibur (Becton Dickinson) and analyzed with the CellQuest software (Becton Dickinson).

Enzyme-linked immunospot (ELISPOT) for interferon-gamma producing CD8+ T cells

A multiscreen, 96-well filtration plate (Millipore, Watford, UK) was coated with 50 μL of 1:66 dilution of antihuman IFN-γ antibody (D1K, Mabtech, Nacka Strand, Sweden) in filtered PBS and incubated overnight at 4°C. Unbound antibody was removed by washing the plate six times with filtered PBS. Cryopreserved PBMCs were thawed as described above and incubated for at least 3 h at 37°C before peptide stimulation. Either 5 × 104 or 1 × 105 of lymphocytes (in a 90 μL volume) were added to each well together with 10 μL of the ppUL83 peptide (200 μM stock in RPMI) or R10 medium for the control well. All ELISPOT assays were carried out in triplicate. After 16 h incubation at 37°C / 5% CO2, cells were removed by washing the plates four times with PBS containing 5% Tween 20 and twice with PBS. Fifty microliters of biotinylated anti-IFN-γ antibody was added (1:1000 dilution, 7-B6-1-biotin; Mabtech) and incubated for 3 h at room temperature. The ELISPOT plate was washed a further six times with PBS/Tween 20 and incubated for 2 h with streptavidin-ALP substrate (Mabtech, Hamburg, Germany) followed by the addition of an alkaline phosphatase conjugate substrate (50 μL; Bio-Rad, Hemel Hempstead, UK). The resulting spots were counted semi-automatically with an ELISPOT reader. Results were expressed as percentage of cells secreting IFN-γ after subtracting the number of spots due to spontaneous IFN-γ release (measured in the control wells) from the number of spots obtained in the wells incubated with the ppUL83 peptide. Recent data obtained using the ELISPOT assay for IFN-γ and an intracellular cytokine assay have shown high correlation between the two assays and similar functional impairment of HCMV specific CD8 T cells as observed in the current study (data not shown).

Statistical analyses

The ppUL83 specific HCMV CD8+ T-cell frequency was recorded as a percentage of CD8 cells or total lymphocytes. Multiple samples derived from the same patient were expressed as a median ppUL83 specific HCMV CD8+ T-cell frequency. The relative ppUL83 specific CD8+ T-cell frequency was calculated as a percentage of CD8+ T cells or as a percentage of lymphocytes for comparison with the ELISPOT data. Samples with less than 20 000 total events on the FACS plots were excluded from the analysis. The number of cells secreting IFN-γ was expressed as percentage of cells secreting IFN-γ relative to the total number of cells used in the assay. These percentages were log transformed to provide a normal distribution before performing further statistical tests. Data were stratified according to whether patients experienced high level replication (viremia >200 genomes/ mL blood) with a sub-stratification to samples taken before, during and after the episode of high level replication. Samples from patients who did not experience HCMV viremia (<200 genomes/mL blood) during the study period were stratified into those taken before day 50 posttransplantation and those taken after day 50 to match the typical time that viremia was first detectable. Samples in different strata were compared using ANOVA. Mann-Whitney U test was used to compare the tetrameric complex to ELISPOT ratio between viremic and non-viremic patients. A linear regression model was used to compare the frequency of HCMV specific tetrameric positive T cells and cells responding in an IFN-γ assay. Logistic regression models, using viremia as a binary outcome, were used to identify and quantify risk factors associated with becoming HCMV viremic post renal transplantation. Except where indicated, all summaries of data are expressed as the median with the range in brackets. A p-value ≤ 0.05 was regarded as significant.


Patient demographics

A total of 25 (17 male, 8 female) consecutive patients who had an HLA type, for which a ppUL83 class I HLA tetramer was available, were recruited to the study. The median follow-up time was 363 days (range 86–680 days) with 13 (range 7– 22) blood samples per patient available for the immunological studies. The median age of the group was 42 years (range 19–67 years). Ten patients (40%) had at least one episode of high-level replication (HCMV viremia; HCMV load >200 genomes/mL blood) and were given pre-emptive antiviral therapy. Four patients were HCMV IgG negative at transplantation and received an organ from a HCMV seronegative donor. None of these became HCMV viremic. In contrast, the two seronegative patients who received a kidney from a HCMV seropositive donor became viremic. Eight of the 19 HCMV seropositive patients, who received either a HCMV negative or positive kidney, became viremic. This HCMV immune experienced patient group (n = 19) was studied in more detail to determine factors associated with appearance of HCMV viremia. A more comprehensive summary of patient demographics is shown in Table 1. Recipients who experienced high level HCMV replication (viremic) were significantly older than those who maintained viral replication below detectable levels (viremic patients: 56 [range 22–67] years versus nonviremic patients: 26 [range 19–59] years; p = 0.02). This age effect was also observed if the analysis was restricted to the 19 patients with pre-existing immunity to HCMV. There were no significant differences in the immunosuppressive therapy (baseline or induction) received by patients who did or did not experience high level HCMV replication. In addition, there were no significant differences in the donor/recipient HCMV serostatus and incidence of viremia in the 10 patients excluded from the study (70% R+, 40% viremia) compared to the patients who comprised the study cohort.

Table 1. Demographic characteristics of the study population
CharacteristicHigh level replication
No (n = 15)Yes (n = 10)
Sex (male/female)10/57/3
Renal graft (deceased/living donor)13/29/1
Median Follow up (days)363397
Number of samples 
 (per patient) 13 13
 Cyclosporin based 13  4
 Tacrolimus based  2  4
 Other  0  2
 Induction therapy 10  8
Donor (D) and recipient (R) HCMV serostatus
 D–R–  4  0
 D+R+ 11  8
 D+R–  0  2

HCMV specific CD8+ T-cell frequencies in HCMV seropositive patients using class I HLA tetramers

In the 11 HCMV seropositive recipients who remained PCR-negative (HCMV load < 200 genomes/mL blood) following transplantation, the median ppUL83 specific CD8+ T-cell frequency was 0.28% (0.02–3.35) prior to day 50 and 0.37% (0.01–6.92) after day 50 (p nonsignificant). In the eight HCMV seropositive patients who experienced high-level replication, the HCMV specific CD8+ T-cell frequency was elevated prior to viremia (0.71%[0.29–3.65]) during viremia (1.19%[0.27–5.2]) and following viremia (0.5%[0.03–6.46]) compared with patients not experiencing viremia (p = 0.008). In all samples analyzed, patients with viremia had a significantly higher ppUL83 specific CD8+ T-cell frequency by HLA tetramer analysis than patients who remained PCR negative (p = 0.008) (Figure 1).

Figure 1.

Frequency of ppUL83 specific tetramer+(tet+) CD8+ T cells in patients with and without HCMV high level HCMV replication according to time posttransplantation or relative to the period of viremia. Data are expressed as a proportion of total CD8 T cells. The median value of each dataset is shown as a horizontal within the box which encompasses the 25th and 75th percentiles. The whiskers extending from either end of the box represent the extent of the data. Using ANOVA, there was a significant difference between the groups stratified according to high level replication (yes or no, p = 0.008).

IFN-γ secretory capacity of HCMV specific CD8+ T cells from HCMV seropositive patients

In the 11 HCMV seropositive patients who remained PCR-negative, the median frequency of IFN-γ positive ppUL83 epitope specific CD8+ T cells (as a percentage of the total PBMC population) was 0.023% (0.005–1.08) before day 50 and 0.032% (0.005–4.66) after day 50 (Figure 2). For comparison with the HLA tetramer data, the ELISPOT frequency prior to day 50 is equivalent to a frequency of 0.3% in the specific CD8 T-cell compartment. In patients with high-level replication, a frequency of 0.027% (0.005–0.009) (∼0.14% of the CD8 T cells) was observed before viremia, 0.082% (0.005–0.87) during viremia and 0.17% (0.005–0.82) after viremia (Figure 2). In contrast to the results obtained with the ppUL83 HCMV specific tetramers, there was no significant difference in the frequency of IFN-γ producing CD8 T cells after peptide stimulation between patients who did or did not experience high-level replication (p = 0.35).

Figure 2.

Frequency of ppUL83 specific PBMCs producing IFN-γ following peptide stimulation in patients with or without high level HCMV replication according to time posttransplantation or relative to the period of viremia. It should be noted that data are expressed as a proportion of total PBMCs. The data are presented as described in the legend to Figure 1. Using ANOVA, there were no significant differences between the groups (p = 0.35).

Functional impairment of ppUL83 specific CD8+ T cells and failure to control HCMV replication

There was a strong correlation between the frequency of ppUL83 specific IFN-γ secreting CD8 cells and CD8 cells identified using ppUL83 specific tetrameric complexes in patients with or without viremia (R2= 0.49 viremic patients, R2= 0.41 non-viremic patients, p = 0.01; Figure 3). However, patients who experienced high-level replication had significantly fewer cells able to secrete IFN-γ following peptide stimulation. This functional impairment is illustrated in Figure 4, where the ratio of the HCMV CD8+ T cells identified using class I HLA tetramers and using the IFN-γ ELISPOT assay are compared. The difference in peptide responsiveness of the ppUL83 CD8 T cells stratified according to high-level HCMV replication as a binary variable was highly significant (p = 0.016). We also performed similar analyses on a group of healthy HCMV seropositive individuals (n = 10) and also a group of patients with common variable immune deficiency (CVID; n = 10). These data revealed that 65.3% (±26.5%) of the tetramer+ cells from healthy seropositive individuals could produce IFN-γ after peptide stimulation, which was comparable to the HCMV specific CD8 T cells present in patients with CVID (59.4 ± 22.2%, p = 0.6). In a group of dialysis patients (n = 10), a similar distribution was observed with an average of 47.5% (± 38.4%) of the HCMV CD8 tetramer positive T cells being able to secrete IFN-γ although this ranged from 0.71 to 107% for individual patients (p = 0.21 and p = 0.38 when compared with healthy individuals and CVID patients, respectively). Comparison of these data with the average capacity of the HCMV CD8 tetramer+ T cells to produce IFN-γ in the renal transplant patients showed that there were no significant differences in this ratio between renal transplant patients, who remained HCMV PCR negative, and the healthy controls/ CVID patients (p = 0.89) and dialysis patients (p = 0.08); whereas the ratio was significantly lower in renal transplant patients who experienced HCMV viremia and the healthy controls and CVID patients ((p = 0.006) but not against the dialysis patients (p = 0.09).

Figure 3.

Scatterplot showing the correlation between ppUL83 CD8+ T cells identified by class I HLA tetramer staining and those secreting IFN-γ following peptide stimulation in patients with and without viremia.

Figure 4.

Comparison of the proportion of ppUL83 CD8+ T cells able to produce IFN-γ following peptide stimulation between patients with or without high level HCMV replication. The data are presented as described in the legend to Figure 1.

Factors associated with high-level HCMV replication after renal transplantation

A series of univariable and multivariable logistic regression models were used to determine the relative importance of the risk factors identified in this study, which predispose renal transplant recipients to high-level HCMV replication. Importantly, these models were based on samples taken before patients became HCMV PCR positive, that is when viral loads exceeded 200 genomes/mL blood. Univariable models showed that each 10-fold increase in ppUL83 specific CD8 T-cell frequency was associated with an odds ratio of 2.15 for high-level replication. In contrast, a lower frequency of IFN-γ secreting ppUL83 CD8 T cells was associated with high-level HCMV replication (odds ratio = 3.03, 96% CI 1.25–7.14). However, it was the functional impairment of these CD8 T cells (the proportion of tetramer positive cells that could secrete IFN-γ following peptide stimulation) that was associated with the greatest risk (odds ratio = 6.7) for the development of high-level replication. Recipient age above 40 years was also associated with an elevated risk of developing high-level replication (odds ratio = 1.07). In bivariable and multivariable additive models, both age and functional impairment of ppUL83 CD8 T cells remained independent risk factors for high level HCMV replication after renal transplantation. A summary of the models is shown in Table 2.

Table 2. Univariable and multivariable logistic regression models relating immune parameters and age with future appearance of high level replication
ModelVariable1Odds ratio95% Confidence intervalp-Value
  1. 1Model based on log10 transformed frequencies.

Age (per 1 year increase) 1.121.07–1.170.02
ppUL83 tet+ CD8+ frequency (per 10-fold increase) 2.150.99–4.700.05
ppUL83 specific IFN-γ secreting cells (per 10-fold decrease) 3.031.25–7.140.01
Proportion of HCMV tetramer+ CD8 T cells secreting IFN-γ (per 10-fold decrease)6.71.5–33.3 0.01
 Proportion of HCMV tetramer+ CD8 T cells secreting IFN-γ (per 10-fold decrease)14.2 2.7–100 0.02
Age (above 40 years) 1.371.07–1.750.01

Following these logistic regression analyses, we constructed graphs relating the probability of a patient developing high-level replication based upon the functional impairment of their ppUL83 specific CD8 T cells (Figure 5, top panel). The graphs revealed a rapid increase in the probability of viremia once critical levels of functional impairment became evident. Indeed, the probability can increase substantially over relatively small decreases in functional capacity. For example, the probability of high-level replication is 10% when the functional CD8 impairment is 50% of normal but increases to 60% when only 10% of the HCMV ppUL83 tetramer+ CD8 T cells can secrete IFN-γ. The effect of age both changes the shape of the probability curve and shifts it to the left indicating that a patient above 40 years of age has a significantly increased risk of high-level replication since their HCMV tetramer+ CD8 T cells are already impaired in their ability to secrete IFN-γ following peptide stimulation (Figure 5, lower panel).

Figure 5.

Probability plot illustrating the risk of experiencing high level replication based on the ability of tet+ ppUL83 CD8+ T cells to produce IFN-γ following peptide stimulation for the entire study population (upper panel) or stratified according to age (above or below 40 years; lower panel).

Finally, we investigated the relationship between functional impairment of HCMV CD8 T cells prior to onset of high-level replication and the initial virus load attained during the period of DNAemia. A linear relationship between initial level of HCMV DNAemia and proportion of tetramer+ HCMV CD8 T cells that can secrete IFN-γ following peptide stimulation, was apparent (R2= 0.83, p = 0.008; Figure 6). A similar relationship was observed when the analysis was repeated using the maximum virus load attained by individual patients (R2= 0.48, p = 0.03).

Figure 6.

Relationship between the maximum HCMV load observed posttransplantation and the ability of the tet+ ppUL83 CD8+ T cells to produce IFNγ. This analysis was restricted to patients where quantification of HCMV loads were available.


The underlying reasons why only a subset of patients who receive the same immunosuppressive drugs to permit allo-transplantation develop high-level HCMV replication or HCMV end-organ disease have always been an enigmatic. It is known that receipt of augmented immunosuppression for the treatment of rejection episodes is associated with a higher risk of HCMV disease yet, even in this setting, many patients do not develop HCMV disease. In contrast to the extensive data showing that HCMV replication parameters provide prognostic information for HCMV disease (14–20), very few studies have assessed whether quantitative immune parameters are useful in predicting patients who will ultimately experience high level HCMV replication. In the present study we have shown that the quantitative function of CD8 T cells directed against a dominant CD8 T-cell target within HCMV (ppUL83) is directly related to the risk of developing high-level HCMV replication in HCMV seropositive recipients and to the viral load attained during infection. The frequency of HCMV specific T cells measured by HLA tetramer analysis or IFN-γ ELISPOT assays showed considerable overlap before, during and after HCMV viremia and provided a poor discriminator for patients destined to experience viremia. However, when we investigated the proportion of the tetramer+ cells that could secrete IFN-γ after peptide stimulation, a more striking and significant association was revealed. The functional impairment was consistently observed across a range of CD8 T-cell frequencies as measured by the class I HLA tetramers implying a broad impact on CD8 T-cell function irrespective of relative CD8 population size. The impairment was evident before the detection of HCMV DNAemia indicating that either the impairment was present prior to transplantation or that it was driven by early HCMV replication in the target organ in the first weeks following transplantation. The patients who did not experience DNAemia had a similar proportion of tetramer reactive T cells that could secrete IFN-γ as cells obtained from healthy volunteers and patients with CVID. However, individual members of these populations showed marked differences in this ratio implying that the baseline functionality of HCMV UL83 specific T cells may be an important factor that predisposes individuals to HCMV replication if they proceed to renal transplantation. Consistent with this observation was the finding that pretransplant dialysis patients showed a broad distribution in the proportion of tetramer staining CD8 T cells secreting IFN-γ, which is consistent with a subset of patients posttransplant being at highest risk for high level HCMV replication. A recent longitudinal study of HIV-infected patients progressing to HCMV retinitis has also observed an increase in the tetramer+ CD8 T cells and a decrease in their capacity to secrete IFN-γ prior to disease occurrence (32). In addition, a study of IFN-γ positive HCMV CD8 T cells in heart and heart/lung patients has shown that patients who developed an IE1 CD8 T-cell response irrespective of their ppUL83(s) were protected against HCMV disease whereas HCMV disease developed only in those with a ppUL83 response (35). Although we have not formally investigated IE1 CD8 T-cell responses in our study, preliminary data suggest that the functional impairment we observed in ppUL83 specific T cells is also evident in other HCMV T-cell antigen specificities (unpublished data). It will also be interesting to determine whether this functional impairment in CD8 T cells extends to other persistent viral infections such as EBV following transplantation. It is noteworthy that a recent longitudinal study of HCMV immunity in D+R– liver transplant patients after antiviral prophylaxis did not find an association between the quantity of functional T cells and occurrence of HCMV viremia or HCMV disease, although the study did not investigate the proportion of functional T cells as we have done in this study (36). In contrast, the study by Crough et al. (37) has reported the same IFN-γ deficit in HCMV CD8 T cells as our study in a mixed cohort of 15 solid organ transplant recipients although, unlike our study, they addressed neither the prognostic significance of the deficit nor its relationship with HCMV replication levels.

The linear relationship between logarithm of viral load and functional impairment of the ppUL83 specific CD8 T cells provides direct evidence for the role that T cells play in limiting the level of replication and is also consistent with our previous work on viral replication dynamics following liver transplantation (15) illustrating the substantial reduction in peak virus load and replication rate associated with pre-existing functional immunity. Such combinatorial approaches provide tools to further determine which subcomponents of natural immunity protect against high-level replication. Furthermore, the broader phenotype of these functionally impaired HCMV specific CD8+ T cells needs to be investigated especially as recent data suggest that the addition of interleukin-2 at the time of ex vivo stimulation of PBMCs derived from HIV-1 infected patients could overcome the poor proliferative capacity of both HIV and HCMV specific CD8+ T cells (38) and that PD-1 expression on LCMV and HIV specific CD8 and CD4 T cells correlates with poor suppression of replication, lack of proliferative capacity and inability to produce IFN-γ following peptide stimulation (39–42). In the latter studies, PD-1 expression on HCMV specific cells, whilst lower than that on HIV specific cells, was variable, suggesting that expression of this molecule could be involved in determining the functional capacity of these HCMV CD8-specific cells or, indeed, CD4 T-helper cells. Indeed, very recently, elevated PD-1 on HCMV CD8 T cells has been observed in liver transplant patients experiencing HCMV DNAemia and disease (43). Although we have not addressed the mechanism by which these CD8 T cells are impaired in their capacity to produce IFN-γ, it is noteworthy that the phenotype of these cells were CD57hi, which is a cell surface marker associated with replicative T-cell incompetence especially in the aged human host (see later). In a subset of samples from patients included in this study, CD57 expression itself was not associated with a substantial increased risk of HCMV replication (relative risk 1.04, 95% CI 0.94–1.14, p = 0.461) (data not shown) although further detailed prospective monitoring is required for this marker to assess its prognostic value. The functional impairment of HCMV CD8 T cells may also reflect the latent burden of HCMV in the kidney, that is a higher burden may be associated with increased reactivation and a higher frequency of exhausted T cells. In this context, HCMV DNA can be frequently detected in renal tubular epithelial cells and endothelial cells soon after infection (unpublished data) and has been associated with long-term organ dysfunction (44). The immunosuppressive regimen could also impact on the functional capacity of the HCMV CD8 T cells although we could not uncover specific associations in this study due the cohort size and range of immunosuppressive regimens. However, we have previously shown in liver transplant recipients that a complex interaplay exists between immunosuppressive drugs, used alone and in combination, and the risk of HCMV viraemia (45).

Using pre-emptive therapy to control HCMV replication and minimize disease invariably results in a proportion of patients (up to 30% in some studies) presenting with a second discrete episode of high-level replication (12). It is known that high viral load and a slow decline rate following antiviral therapy are risk factors for second episodes of viremia which may be secondary to the immune impairment we describe in this study (12). In addition, D+R+ transplants may be at risk of reactivation of latent HCMV strains, re-infection or both. The present study was underpowered to investigate the relationship between donor seropositivity for HCMV and CD8 T-cell function. Further work is needed to assess these issues and also to identify whether a similar impairment may contribute to the appearance of late infection and, more importantly late HCMV disease, observed following the cessation of 100 days of prophylactic therapy (4,5,10). Nevertheless, the potential for identifying a high-risk group subgroup of HCMV seropositive transplant recipients pretransplant based upon a functional defect in their CD8 T cells could allow rational deployment of antiviral prophylaxis for this group similar to the approaches adopted for the high-risk HCMV seronegative recipients receiving organs from seropositive donors (10).

In our studies, recipient age remained a risk factor for high-level replication even in the multivariable analyses. It is known that HCMV seropositive individuals have an immune risk phenotype associated with a decreased life expectancy (44, reviewed in 46). Cross sectional and longitudinal studies have also shown that functional HCMV CD8 T cells are reduced in ageing populations and CD85j (also known as LIR-1) expression on HCMV specific CD8 T cells increases with age (47–49). Our data revealed that older age (>40 years) shifted the probability of HCMV viremia–CD8 functional impairment curves, so that, for any given level of CD8 impairment a patient above 40 years of age had a substantially increased probability of progressing to high-level HCMV replication compared with patients under the age of 40 years. These data are consistent with previous studies and suggest that the immune risk phenotype associated with HCMV may be particularly relevant when the host is therapeutically immunocompromised.

In conclusion, our data provide compelling evidence linking functional impairment of CD8 T cells in HCMV infection with the future development of high-level replication after renal transplantation. The data will facilitate the development of alternative prognostic markers to complement real time PCR monitoring of viral replication to identify patients at high risk of developing HCMV disease.


This work was supported by a Wellcome Trust programme Grant, Wellcome Trust Clinical Fellowship grant (to Dr. Frank Mattes) and Prize studentship award (to Dr. Jakub Kopycinski) and by a UK Medical Research Council Centre award.

The authors have no conflicting financial interests.