Anti-cytomegalovirus (CMV) prophylaxis is recommended in D+R− kidney transplant recipients (KTR), but is associated with a theoretical increased risk of developing anti-CMV drug resistance. This hypothesis was retested in this study by comparing 32 D+R− KTR who received 3 months prophylaxis (valganciclovir) with 80 D+R− KTR who received preemptive treatment. The incidence of CMV infections was higher in the preemptive group than in the prophylactic group (60% vs. 34%, respectively; p = 0.02). Treatment failure (i.e. a positive DNAemia 8 weeks after the initiation of anti-CMV treatment) was more frequent in the preemptive group (31% vs. 3% in the prophylactic group; p = 0.001). Similarly, anti-CMV drug resistance (UL97 or UL54 mutations) was also more frequent in the preemptive group (16% vs. 3% in the prophylactic group; p = 0.05). Antiviral treatment failures were associated with anti-CMV drug resistance (p = 0.0001). Patients with a CMV load over 5.25 log10 copies/mL displayed the highest risk of developing anti-CMV drug resistance (OR = 16.91, p = 0.0008). Finally, the 1-year estimated glomerular filtration rate was reduced in patients with anti-CMV drug resistance (p = 0.02). In summary, preemptive therapy in D+R− KTR with high CMV loads and antiviral treatment failure was associated with a high incidence of anti-CMV drug resistance.
Management strategies for cytomegalovirus (CMV) infection in kidney transplant recipients (KTR) remain a controversial issue. During the first year posttransplantation, both prophylactic and preemptive approaches appear equally effective in controlling CMV disease, but the long-term impact of either therapy is more elusive. In the recently published international consensus guidelines on the management of CMV infection in solid-organ transplantation (1), the majority of expert participants favored the use of universal prophylaxis over preemptive therapy in the highest risk recipients (D+R−), based on better clinical outcomes (lower rates of opportunistic infections, prevention of rejection), improved graft survival, ease of use and lower monitoring costs.
Whatever the chosen therapy, the current drugs of choice for the treatment of CMV infection are ganciclovir and its oral prodrug valganciclovir. Ganciclovir requires pUL97-mediated phosphorylation for its antiviral activity. All currently available anti-CMV drugs (including ganciclovir, foscarnet and cidofovir) inhibit the CMV DNA polymerase (product of the UL54 gene). Therefore, mutations within UL97 and/or UL54 can confer anti-CMV drug resistance (2,3). Theoretically, the risk of developing anti-CMV drug resistance, leading to treatment failure, is higher in prophylactic therapy (1) since prolonged antiviral drug exposure has been identified as one of the predisposing factors (4–6), as well as other interrelated factors, such as lung transplantation, D+R− status, potent immunosuppression, oral ganciclovir use, suboptimal antiviral drug levels, prolonged viral replication and high CMV load (2,7,8). Conversely, preemptive therapy would facilitate the development of significant CMV-specific cell-mediated immunity by promoting the encounter between the virus and the immune system. However, whether or not prophylaxis and preemptive strategies lead to the emergence of antiviral resistance in KTR remains unclear (9).
Between 2005 and 2009, these two different strategies were successively used in our center for the management of D+R− patients. In this retrospective single-center study, we first compared the incidence of anti-CMV drug resistance between the prophylaxis and the purely preemptive therapy groups. We next identified in this cohort the risk factors associated with the occurrence of anti-CMV drug resistance.
Materials and Methods
Study design and patients
As shown in Figure 1, 570 kidney transplantations were performed in our department between January 1, 2005 and November 30, 2009. KTR who were included in another research protocol with mandatory antiviral therapy were excluded from this retrospective analysis (n = 30).
From January 1, 2005 to November 30, 2006, 172 patients underwent kidney transplantation. All D+R− KTR (n = 32), except three who were excluded from the study, received valganciclovir prophylaxis for 3 months (900 mg per day). From December 1, 2006 to November 30, 2009, 368 patients underwent kidney transplantation without prophylaxis. All D+ R– KTR (n = 80) were preemptively followed.
From 2005 to 2009, all of the patients were monitored for CMV using a whole-blood quantitative CMV real-time PCR (WB CMV PCR) assay. The assay was performed once a week for the first 3 months, then once a month between months 3 and 6, and finally every 2 months up until 1 year. Intravenous (IV) ganciclovir (5 mg/kg b.i.d.) or oral valganciclovir (900 mg b.i.d.) treatments were initiated when the WB CMV PCR result reached 2000 copies/mL. IV ganciclovir was administered for 2–3 weeks and was always followed by valganciclovir treatment. The anti-CMV treatment was stopped after two consecutive negative WB CMV PCR results. The daily total doses of IV ganciclovir or valganciclovir for patients with impaired renal function was adjusted according to the manufacturer's recommendations (using the Cockroft–Gault formula).
CMV manifestations were defined, based on standardized criteria, as CMV infection (CMV load proven by two consecutive positive WB CMV PCR results) or CMV disease, either as CMV syndrome (unspecific clinical symptoms and detectable CMV load in blood) or CMV tissue-invasive disease (proven CMV-related organ dysfunction or failure) (11). Late-onset CMV infection was defined as a CMV infection occurring after 3 months post-transplantation. Treatment failure was defined as the presence of a positive WB CMV PCR result 8 weeks (or at day 50) after the initiation of anti-CMV treatment (based on the VICTOR study ). Anti-CMV drug resistance was investigated when a significant increase of CMV load (>1 log10 copies/mL) was observed during antiviral therapy with valganciclovir or IV ganciclovir and was defined as the presence of resistance-associated mutations in the UL97 and/or UL54 genes. Recurrent CMV infection was defined as a new CMV infection occurring after a successfully treated CMV infection or disease. Expanded criteria donors (ECD) were defined as previously described (12). All rejection episodes were biopsy proven and classified according to the last Banff Classification (13). The 1-year estimated glomerular filtration rate (e-GFR) was calculated using the MDRD formula.
The detection of CMV IgG in serum was performed following the manufacturer's recommendations (Enzygnost anti-CMV/IgM and IgG, Dade Behring, Marburg, Germany). WB CMV PCR was performed as previously described (14). Genotypic resistance testing was performed at the French National Cytomegalovirus Reference Center (Limoges, France). The full-length UL97 and UL54 genes were sequenced from DNA extracted from clinical samples. Sequences were compared with the AD169 reference sequence using the Gene Librarian 3.2TM software (Visible Genetics Inc., Siemens, France) for the identification of known resistance-related mutations, polymorphisms and new mutations, whether or not they were within suspected resistance loci (15,16).
Analyses were performed using conventional statistical methods. The Mac Nemar χ2 test was used for qualitative variables; Student's t-test and the Mann–Whitney U test were used when appropriate. The variables potentially associated with anti-CMV drug resistance were subjected to univariate and multivariate analysis. Analyses were performed with the Statview software (Abacus Concepts, Berkeley, CA, USA).
Table 1. Baseline characteristics of D+R− patients
Prophylactic (n = 32)
Preemptive (n = 80)
Age (mean, years)
50.8 ± 12.9
47.5 ± 14.2
Mean PRA (%)
15.4 ± 33.7
7.8 ± 18.1
Binephrectomy for cancer
Age (mean, years)
47.6 ± 13.3
46.1 ± 16.0
Expanded criteria donors
3.3 ± 1.2
3.8 ± 1.3
Total ischemia time (mean, h)
17.6 ± 7.6
16.9 ± 6.0
Delayed graft function
Anti-IL2 receptor antibody/antithymocyte globulin
No significant differences in the baseline characteristics of patients were observed between the prophylactic and preemptive strategies, except for human leukocyte antigen (HLA) mismatches and the type of calcineurin inhibitor. A cyclosporine-based therapy was mainly used until the end of 2006 and was changed for a tacrolimus-based regimen in 2007.
CMV manifestations according to the type of therapy (Table 2)
Table 2. Direct effects of CMV and anti-CMV treatment
Prophylactic (n = 32)
Preemptive (n = 80)
CMV infection (%)
CMV disease (%)
Time of CMV infection (median, days)
Late-onset infection (%)
Baseline viral load (mean, log10 copies/mL)
4.3 ± 1.6
3.7 ± 1.1
Peak viral load (mean, log10 copies/mL)
4.2 ± 1.1
5.0 ± 1.0
Prophylaxis: valganciclovir for 3 months
Initial anti-CMV therapy for CMV infection (curative not prophylactic)
IV ganciclovir (%)
Treatment failure (%)
Anti-CMV drug resistance (%)
Recurrent CMV (%)
The incidence of CMV infections was almost twice as high in the preemptive group than in the prophylactic group, whereas the number of CMV diseases was similar between the two groups. The percentage of late-onset CMV infections was higher in the prophylactic group. The mean peak viral load was higher during preemptive therapy compared with prophylaxis therapy, but this difference was not statistically significant. No significant differences in mean baseline viral load were observed between the two groups.
Treatment of CMV infection and incidence of anti-CMV drug resistance (Table 2)
Eight patients in the prophylactic group had a WB CMV PCR result above 2000 copies/mL (all late-onset infections) and were mainly treated with IV ganciclovir (n = 7), with valganciclovir only being used in one case. CMV infections in the preemptive group were initially treated with either valganciclovir (n = 31) or IV ganciclovir (n = 17). Thus, when comparing the two strategies, CMV infections were more often treated with valganciclovir in the preemptive group than in the prophylactic group.
Treatment failure occurred in 31% of patients from the preemptive group and only 3% of patients from the prophylactic group (p = 0.001). Among the patients receiving curative therapy for CMV infection, the incidence of treatment failure was still higher in the preemptive group (25/48, 52%) than in the prophylactic group (1/8, 13%, p = 0.04). Anti-CMV drug resistance was detected more frequently in the preemptive group than in the prophylactic group (16% vs. 3%, respectively; p = 0.05). However, among the patients receiving curative therapy for CMV infection, the incidence of anti-CMV drug resistance was not significantly different between the preemptive group (13/48, 27%) and the prophylactic group (1/8, 13%, p = 0.4).
Secondary effects of viral infection according to the therapy (Table S1)
The incidences of biopsy-proven acute rejections and other infections did not differ between the two groups. Graft survival, patient survival and the e-GFR at 1 year were also similar between the two groups.
Risk factors associated with anti-CMV drug resistance in D+R− patients
The incidences of anti-CMV drug resistance among the 112 D+R− KTR and the 56 D+R− KTR treated for CMV infection were 13% and 25%, respectively. We continued our analysis by focusing on the 14 D+R− KTR displaying anti-CMV drug resistance. All these patients, except the one that received prophylaxis, showed signs of CMV infection during the first 90 days posttransplantation (Figure 2). Mutations in the viral strains were identified 170 ± 56 days after transplantation and 134 ± 61 days after the first positive CMV PCR result. All of the patients exhibiting drug-resistant virus were administered foscavir as an alternative treatment. The mean duration between the first positive CMV PCR result and viral eradication was 266 ± 180 days.
All these 14 patients displayed antiviral treatment failure that always occurred before the identification of a mutation (Figure 2). Among the 56 D+R− KTR treated for CMV infection, 54% (14/26) of antiviral treatment failure was associated with anti-CMV drug resistance. By contrast, none of the 30 KTR who displayed viral eradication within 8 weeks after the initiation of anti-CMV treatment developed anti-CMV drug resistance (p = 0.0001).
Univariate analysis was next performed on the 56 D+R− KTR treated for CMV infection. Among the parameters evaluated, only the peak viral load was found to be significantly associated with anti-CMV drug resistance (OR = 5.01, p = 0.002) (Table 3). Using multivariate analysis including the type of strategy (prophylaxis vs. preemptive), the type of anti-CMV treatment (valganciclovir vs. IV ganciclovir) and the peak viral load we found that the only parameter independently associated with anti-CMV resistance in this subgroup of patients remained the peak viral load (OR = 28.72, p = 0.0004). Peak viral loads were observed before the identification of resistance-related CMV mutations. Figure 3 shows the peak viral load according to the subsequent occurrence of anti-CMV drug resistance in the D+R− KTR with CMV infection. It was evident that 75% of patients with a peak viral load over 5.25 log10 copies/mL (equivalent to 180 000 copies/mL) experienced anti-CMV drug resistance, whereas 75% of patients with a peak viral load below 5.25 log10 copies/mL did not exhibit resistance. Consequently, patients with a viral load over 5.25 log10 copies/mL were at the highest risk of developing anti-CMV drug resistance (OR = 16.91, p = 0.0008) (Table 3).
Table 3. Factors associated with the occurrence of anti-CMV drug resistance among the 56 D+R− KTR treated for CMV infection
Peak viral load (>5.25 log10 copies/mL vs. <5.25 log10 copies/mL)
Prophylaxis with valganciclovir
Valganciclovir for CMV infection (vs. ganciclovir)
Finally, a peak viral load over 5.25 log10 copies/mL was more frequent in patients displaying treatment failure (16/26, 62%) than in other patients (7/30, 23%, p = 0.0001). Using univariate analysis, we were unable to identify any risk factors associated with treatment failure, neither the use of valganciclovir as the initial anti-CMV treatment (OR = 1.04, p = 0.9), nor the use of prophylaxis (OR = 0.13, p = 0.07), nor the baseline viral load level (OR = 0.94, p = 0.8) (data notshown).
Clinical outcome of D+R− patients exhibiting anti-CMV drug resistance (Table S2)
KTR exhibiting drug-resistant virus did not have a higher rate of acute rejections or other opportunistic infections than other KTR. The 1-year e-GFR was significantly lower in KTR exhibiting resistant virus than in other KTR (41 ± 24 vs. 58 ± 20, respectively; p = 0.02). Three patients with confirmed resistance (two mutations in UL97, one in UL54) lost their graft following acute rejection (n = 1) or foscavir nephropathy (n = 2). However, the rates of patient and graft survival after 1 year were not significantly different between the two groups.
Our findings are similar to those of three previous randomized studies comparing universal prophylaxis and preemptive therapies (17–19). As reported by Kliem et al. (18), we observed an elevated incidence of CMV diseases (26%) in the preemptive group. As expected, we observed a low incidence of late-onset infections with preemptive treatment. Late-onset CMV diseases are a well-recognized drawback of the prophylactic strategy (20,21) that prompted the recent proposal to extend the duration of prophylaxis to 6 months (22).
Anti-CMV drug resistance mainly occurs in D+R− KTR (5,6,16,23), with an incidence usually between 5 and 10% (1). In our study, since no systematic screening for resistance at specific time points was performed, the incidence of resistance (13%) could be underestimated. All UL97 and UL54 mutations identified in the drug-resistant viral strains of our patients were drug-resistant associated mutations that had been previously confirmed by marker transfer or recombinant phenotyping (3). Interestingly, all mutations were detected after treatment failure as previously demonstrated by Boivin et al. (23). Prolonged exposure to ganciclovir therapy can lead to anti-CMV drug resistance (7), probably because of its poor bioavailability and failure to reach the target tissues in sufficient concentration (2). However, some recent papers have also reported the emergence of anti-CMV drug resistance in patients who received prolonged valganciclovir therapy (4,24).
The second risk factor associated with anti-CMV drug resistance was a peak viral load over 5.25 log10 copies/mL. Boivin et al. (23) previously described a trend between a high baseline viral load (>10 000 copies/mL) and the subsequent emergence of resistance mutations. These high peak viral loads could be explained by (i) the replication dynamics of the resistant virus, which are especially rapid in the absence of efficient treatment (8,25) and (ii) our choice to initiate treatment at a threshold of 2000 virus copies/mL, which may have been too high for D+R− KTR. In our opinion, such a threshold could be clinically useful.
Importantly, we confirmed that the use of valganciclovir as posttransplant CMV prophylaxis was not a risk factor for anti-CMV drug resistance (15,26), probably because the treatment was initiated when there was no active viral replication (27). The choice of initial anti-CMV therapy has also not been associated with anti-CMV drug resistance (28,29). However, persistent subclinical CMV loads in the presence of relatively low ganciclovir serum concentrations over a long period can lead to the selection of ganciclovir-resistant CMV mutants (6,8). Therefore, we believe that the choice and the duration of the initial treatment (oral or IV) should be guided by the viral load kinetics and the dosage of ganciclovir in order to optimize CMV replication eradication.
It could be questioned whether the change in therapy over time, from prophylaxis to preemptive, has contributed to the occurrence of drug resistance. However, a direct comparison should be made with caution, because our study is a retrospective, single-center, nonrandomized study involving a succession of different immunosuppressive protocols and donor characteristics that may have influenced the outcome of CMV treatment (30). Moreover, among the 56 D+R− KTR who received anti-CMV therapy, the percentage of anti-CMV drug resistance was similar between the prophylaxis and the preemptive groups. However, since preemptive therapy was associated with a higher incidence of CMV infection, it was rather logical to observe a higher incidence of resistance in these patients. The increased incidence of anti-CMV drug resistance observed in the preemptive group could also be explained by the increased incidence of treatment failures (31) and high peak viral loads encountered. The three above-mentioned prospective randomized studies comparing the two strategies did not report any anti-CMV drug resistance. Our results are different probably because the peak viral loads in their preemptive group remained below 5.3 log10 copies/mL in two of the studies (17,19), while the mean peak viral load in the third study was also low (4.4 log10 copies/mL) (18). Moreover, in two of these studies, the median time to clear CMV DNAemia was similarly short in the prophylactic and preemptive groups (17,18). In summary, the increased incidence of treatment failure (31) and high peak viral load observed in our study could explain the difference in resistance rate between our study and previous studies.
Finally, we identified a reduction of more than 10 mL/min in the e-GFR at 1 year in patients exhibiting resistant virus compared with patients displaying no CMV resistance. This major change in the e-GFR could be associated with the secondary effects of viral infection (18) or linked to the nephrotoxic effect of antiviral drugs, such as foscavir (4). In contrast to previous studies (32,33), patient survival rates were excellent in the resistant virus-infected KTR.
In conclusion, preemptive therapy is associated with high incidence of anti-CMV drug resistance in D+R− KTR displaying treatment failure and a peak viral load over 5.25 log10 copies/mL.
We thank Catherine Rio for her help as a nurse coordinator. We also acknowledge the technicians from the Laboratory of Virology in Bordeaux Hospital and the Centre National de Référence des Cytomégalovirus for their significant contribution to this study. There was no study sponsor.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplanation.