Effectiveness of Preemptive Therapy for Cytomegalovirus Disease in Pediatric Liver Transplantation

Abstract

Background

Most pediatric liver transplantation (LT) centers administer long courses of prophylaxis against cytomegalovirus (CMV) without evidence of benefit and with significant drug exposure and costs. We aimed at evaluating overall outcomes, direct and putative indirect effects of CMV, possible impact of viremia and risk factors for CMV infection in pediatric LT recipients managed with ganciclovir-based preemptive therapy (PET).

Methods

The records of all the children who underwent LT between 2008 and 2014 were retrospectively analyzed.

Results

One hundred children were included. Three children had CMV disease; no CMV-related death or graft loss was recorded. The only identified risk factor for CMV infection was the donor/recipient serostatus (odds ratio, 17.23; 95% confidence interval, 1.88-157.87; P = 0.012), while viremia per se did not worsen LT outcomes, such as the incidence of acute rejection, Epstein-Barr virus infection, sepsis, biliary and vascular complications, nor graft dysfunction/loss or death at 3 and 5 years after LT. When compared with a historical cohort of children receiving ganciclovir prophylaxis, PET did not differ from prophylaxis for any of the selected outcomes, but was rather associated with lower antiviral drug exposure (6.4 ± 13 days vs 38.6 ± 14 days, P < 0.0001) and cost per patient (2.2 ± 3.9 k€ vs 6.6 ± 8.2 k€, P = 0.001).

Conclusions

PET is effective in controlling CMV in children receiving LT, with lower costs and lower exposure to antivirals.

Cytomegalovirus (CMV) is a major threat in solid organ transplantation, representing the most frequent opportunistic infection. In absence of preventative measures, CMV disease incidence in liver transplantation (LT) depends mostly on the serologic donor (D) and recipient (R) status, ranging from 1% to 2% (D−/R−) to 44-65% (D+/R−).1 Besides the direct tissue injury, CMV disease has been associated with a threefold higher risk of death or graft loss at 5 years post-LT.2 CMV infection has also been linked to indirect effects, such as increased risk of other infections, acute and chronic graft rejection, vascular complications, graft failure and loss.3,4

Preemptive therapy (PET)—consisting in administering antiviral agents in patients with CMV viremia in the attempt to prevent the progression to disease—has proven comparable to universal prophylaxis (UP) in controlling incidence of CMV disease, acute rejection, and overall mortality in a recent meta-analysis of adult studies.5 Unlike in adult solid organ transplants, in children undergoing LT, there is no evidence that CMV viremia per se increases the risk of opportunistic infections, acute or chronic rejection, or vascular complications, thus making the need for prophylaxis even less compelling.5,6 Current guidelines by the Transplantation Society consider PET as an alternative option to prophylaxis, but conclusive evidence about safety and effectiveness remains to be assessed.7 Both strategies have been successfully used in pediatric LT so far, with overall incidence of CMV disease of 4.8%, 5%, and 10% with prophylaxis,1 PET,8 and a hybrid protocol of short prophylaxis course followed by PET,9 respectively.

In a previous randomized controlled trial at our Institution, PET demonstrated to be as effective as a sequential protocol with prophylaxis followed by monitoring (and possible PET) in preventing CMV disease, and no difference was found in the incidence of acute rejection and other infections, so that the study was interrupted because of the significantly higher drug exposure in the prophylaxis group.10 The study was limited to the first year after operation. Since then, universal PET has been adopted at our center.

Aim of our work was to retrospectively evaluate short and long-term safety and efficacy of the CMV PET protocol adopted at our Pediatric Liver Transplant Center from 2008 to 2014. We also sought to identify possible risk factors for the development of CMV viremia among children treated preemptively, and to assess the possible impact of viremia on LT outcomes. Finally, we provided a comparison with a historical cohort of LT children managed with anti-CMV UP protocol (2005-2006).

MATERIALS AND METHODS

Data Collection

We retrospectively reviewed the clinical records of all patients aged 0 to 17 years who had consecutively undergone LT at the Pediatric Liver Transplant Center of the Hospital Papa Giovanni XXIII in Bergamo, Italy, between January 2008 and December 2014.

Exclusion criteria were combined transplantation, retransplantation, liver tumors, HIV, or other immune deficiencies, a follow-up of less than 200 days or incompleteness of the clinical records.

The following data were recorded in an electronic database: patient ID, gender, age at LT, nationality, reason for LT, primary disease, D/R CMV serostatus match (D+/R−; D+/R+ or D−/R+; D−/R−).

Primary outcomes were the incidence of CMV infection and disease. Secondary outcomes were acute cellular rejection (ACR), sepsis and Epstein-Barr virus (EBV) infection in the first 200 days post-LT; biliary strictures and leakages and vascular complications by 3 years post-LT; graft dysfunction, graft loss or death at 3 and 5 years post-LT, when available. Moreover, length of hospitalization after the operation, type and cumulative duration of the antiviral treatment, days of hospitalization for antiviral administration, for the management of antiviral adverse effects and for CMV disease and relevant costs were calculated.

The local ethical committee has approved all retrospective studies provided the records are anonymized.

PET, Prophylaxis, and Immunosuppressive Protocols

All patients were monitored for viremia for 90 days after transplantation once a week, increasing the frequency to twice a week in case of positive CMV viremia. Subsequently, virologic monitoring was performed once a month through day 180, and then at least once every 3 months at days 270 and 360, unless CMV viremia recurred. The monitoring was performed with real-time PCR for measurement of CMV-DNA (detection threshold 650 copies/ml) in both the historical prophylaxis cohort,11 and in the PET cohort.

PET consisted in testing CMV and, when reaching the cutoff of 100000 copies/ml, administering ganciclovir (GCV) (5 mg/kg per dose intravenously [i.v.] every 12 hours) for at least 2 weeks and till its complete disappearance (2 consecutive negative results). After at least 2 weeks of i.v. GCV and upon reduction of viral load, oral valganciclovir (VGCV) (15 mg/kg per dose every 12 hours) was considered as an alternative in patients not requiring hospitalization.

Patients managed according to UP received GCV (5 mg/kg per dose i.v. every 12 hours) for 4 weeks and thereafter were monitored. If viremia reached the cutoff of 100 000 copies/mL, they were treated preemptively for at least 2 weeks and till 2 consecutive viral loads resulted negative. The cutoff used has been established in a previous study.10

Standard immunosuppression adopted in all patients consisted in tacrolimus and prednisone. Our standardized protocol indicates the following target plasma tacrolimus trough levels. First 3 months posttransplant, 8 to 12 ng/mL; 3 to 6 months, 6 to 10 ng/mL; 6 to 12 months, 4 to 8 ng/mL; thereafter, 2 to 5 ng/mL. No induction therapy was used. Mycophenolate mofetil was added in patients requiring tacrolimus-sparing regimen for renal impairment.

Calculated costs were relevant to the antiviral drugs used for prophylaxis, PET, and CMV disease, to the reimbursed cost of hospitalization specifically due to antiviral therapy administration and for the management of CMV disease and of antiviral drug adverse events, as well as of the virologic monitoring. We considered a cost per day of 95 € and 73 € for GCV and VGCV, respectively, which were extrapolated from the hospital pharmacy price list and did not vary with the patient weight or body surface; the cost of each CMV-DNA quantitative assay was 72 €.

Definitions

CMV infection and disease were defined as previously reported.12 CMV infection was defined as isolation of CMV or detection of viral proteins or nucleic acid in any body fluid or tissue specimen, regardless of symptoms.

CMV disease was defined as the evidence of infection in presence of consistent symptoms, such as CMV syndrome (fever, malaise, and myelosuppression) or proven CMV tissue-invasive disease (detection of CMV by culture, immunohistochemical analysis, or in situ hybridization in tissue specimen in presence of relevant histologic features).

EBV infection was defined as the presence of detectable EBV-DNA in whole blood.

ACR was histologically defined according to the Banff criteria.13 A Rejection Activity Index of 5 or greater was considered an indication to administer pulse dose steroids (10 mg/kg per day for 3 consecutive days). Graft dysfunction was defined as histologically proven graft injury found in liver biopsies carried out in case of raising or persistently abnormal liver function tests. Biliary strictures were confirmed by percutaneous transhepatic cholangiography. Vascular complications were suspected on liver ultrasound or on liver histology and confirmed by CT scan and direct angiography. Posttransplant lymphoproliferative disorder (PTLD) was defined by histology on a tissue biopsy according to previous studies.14

Statistical Analysis

The Student t test, the χ² method, or Fisher exact test were performed when appropriate for statistical analysis to compare continuous and categorical variables. Univariate logistic regression was performed to analyze risk factors of CMV infection, disease, and graft dysfunction; statistically significant variables were then included in a multivariate model adjusted for covariates to identify independent risk factors. Kaplan-Meier analysis and Log-rank test were used for cumulative proportion curve comparison. A P value less than 0.05 was chosen as cutoff for significance. Data were analyzed with SPSS (IBM Corp. Released 2011. IBM SPSS Statistics for Mac, Version 20.0. Armonk, NY: IBM Corp) and GraphPad Prism (GraphPad Prism version 5.00 for Mac, GraphPad Software, San Diego, CA) softwares.

RESULTS

One hundred eighty-six children underwent LT at our center between 2008 and 2014 and were managed with PET for CMV. Eighty-six of them were excluded for the aforementioned criteria, therefore 100 recipients (50:50 male/female, age at LT 0.25-17 years, mean follow up 42.7 ± 23 months) were included in the analysis (Figure 1). Eighty-five patients had chronic liver insufficiency (biliary atresia, 62; sclerosing cholangitis, 8; familial cholestasis, 4; Alagille syndrome, 4; cirrhosis due to metabolic disease, congenital hepatic fibrosis, hepatopulmonary syndrome in portal cavernoma, intestinal failure-associated liver disease, 1 patient each; cirrhosis of unknown etiology, 3); 7 patients were transplanted for decompensated metabolic diseases. Eight patients were transplanted for acute liver failure, which was of unknown etiology in 4, and due to autoimmune hepatitis, Wilson disease, and Human Herpesvirus 6 infection in 2, 1, and 1 patients, respectively. Baseline characteristics of the patients are depicted in Table 1 (PET group).

FIGURE 1.

Patient selection and inclusion in the study.

TABLE 1.

Baseline features of the patients enrolled

Sixty-one patients had CMV infection, and 19 required PET. Among infected patients who required PET, D+/R+ serostatus was more frequent (64% vs 15%, P = 0.001), and the maximum viral load was higher (276 517 ± 313 471 vs 47 957 ± 68242 copies/mL, P = 0.0001) than in those who did not. The remaining baseline features and outcomes did not vary. Patients who remained below the treatment threshold spontaneously resolved the infection reaching negative viremia after a median time of 30 (range, 7-105) days. Only 3 cases of CMV disease occurred at 15, 27, and 32 days after LT, and in all, viremia was detected at disease onset with a load above the threshold of 100 000 copies/mL. Two patients had CMV-related colitis (both in D+/R+ pairs, with intestinal perforations in 1 patient) while 1 patient (D+/R−) presented with CMV hepatitis and fever. Antiviral therapy was needed for 20, 30, and 49 days, respectively, only 1 patient had a viral relapse not needing retreatment, and all of them eventually did well. Overall, 3 patients died, and 4 lost the graft, but none of the causes were related to CMV. The outcomes of the patients are shown in Table 2 (PET group). Twenty-five children had graft dysfunction 3 years after LT, which was related to chronic rejection/de novo autoimmune hepatitis, recurrent biliary obstruction, and vascular complications in 11, 9, and 5 cases, respectively.

TABLE 2.

Outcomes of the patients according to the CMV preventative strategy

Impact of CMV Infection on LT Outcomes

Comparing patients with and without the occurrence of CMV infection during the virologic monitoring according to the PET protocol, the risk of short-term and long-term complications of LT (ACR, EBV infection, sepsis, PTLD, graft dysfunction, biliary, and vascular complications) was not found to differ significantly in the 2 groups, as shown in Table 3. Namely, only 18 (58%) of the acute rejection episodes occurred in patients who were CMV infected, and out of these only 4 (22%) occurred after CMV infection. Recipients’ anti-EBV IgG positivity did not affect the occurrence of EBV infection (odds ratio [OR], 0.81; 95% confidence interval [CI], 0.34-1.88; P = 0.625) and of PTLD (OR, 0.29; 95% CI, 0.06-1.31; P = 0.110).

TABLE 3.

Impact of CMV infection on graft outcomes in LT children managed with PET

The proportion of ACR-free patients by 200 days after LT (P = 0.655), the graft dysfunction-free survival (P = 0.464 at 1 year; P = 0.362 at 3 years), and the proportion of patients without biliary (P = 0.760 at 1 year; P = 0.449 at 3 years) and vascular complications (P = 0.869 at 1 year; P = 0.421 at 3 years) were not associated with the occurrence of CMV infection at Kaplan-Meier analysis and Log-rank test (Figure S1, SDC, http://links.lww.com/TP/B367).

Risk Factors for CMV Infection

Risk factors for CMV infection in patients managed preemptively were assessed with univariate analysis. As shown in Table S1, SDC (http://links.lww.com/TP/B367), D+/R− CMV serostatus (OR, 11.66; 95% CI, 1.45-93.72; P = 0.021) and donor 40 years or older (OR, 3.70; 95% CI, 1.29-10.62; P = 0.015) were associated with the onset of CMV infection, whereas sex, age at LT, reason for LT, and underlying disease were not. In multivariate analysis, only D+/R− CMV serostatus match was an independent risk factor for developing CMV infection (OR, 17.23 vs D^any/R+; 95% CI, 1.88-157.87; P = 0.012). The proportion of patients without CMV infection by 200 days after LT was significantly associated with donor/recipient serostatus match also when assessed by Kaplan-Meier analysis and Log-rank test (Figure 2; P = 0.0004).

FIGURE 2.

Kaplan-Meier curve comparison of patients managed with preemptive therapy: CMV infection-free survival according to donor (D)/recipient (R) serostatus: D−/R− (N = 5); Dany/R+ (N = 64); D+/R− (N = 16).

Comparison Between PET and UP Followed by Monitoring

Sixteen patients managed with 30 days antiviral UP followed by virologic monitoring and possible preemptive treatment (6:10 male/female; age at LT, 0.58-10.7 years; mean follow-up, 94.9 ± 30 months) between 2005 and 2006 were included for comparison (Table 1, UP group). The 2 groups did not differ for age at LT, sex, CMV donor/recipient serostatus risk, reason for LT, and underlying disease. The donor age was significantly higher in the PET group (30 ± 16 years) than in the UP cohort (19 ± 13 years, P = 0.018), confirming a universally observed trend over time. Also, the patients’ origin differed between the 2 groups, because since 2008, the transplant program of our center has been extended to children from foreign countries by institutional agreements.

Two death and 3 graft losses in the UP group were observed, none directly related to CMV. As shown in Table 2, no difference was observed in the proportion of CMV infections, relapses, ACR, sepsis, EBV infections, PTLD, graft dysfunction, graft loss or death at 1, 3, and 5 years post-LT, nor in biliary and vascular complications. The onset of CMV infection was earlier in the PET than in UP group (41.2 ± 35.8 vs 70.2 ± 41.2 days, P = 0.019), but the proportion of patients not infected by 200 days after LT did not differ between the 2 groups, as shown by Kaplan-Meier analysis and Log-rank test (Figure 3A; P = 0.410). Likewise, the proportion of patients without graft dysfunction (P = 0.275 at 1 year; P = 0.749 at 3 years), biliary (P = 0.142 at 1 year; P = 0.455 at 3 years), and vascular complications (P = 0.562 at 1 year; P = 0.362 at 3 years) did not differ between the 2 groups at curve comparison (Figure 3B-D).

FIGURE 3.

Kaplan-Meier curve comparison of patients managed with preemptive therapy and prophylaxis. A, CMV infection-free survival at 200 days post-LT; proportion of patients without graft dysfunction (B), biliary complications (C), and vascular complications (D) at 3 years after LT.

On the other hand, antiviral drug exposure (6.4 ± 13 vs 38.6 ± 14 days, P < 0.0001) and total costs per patient (2.2 ± 3.9 vs 6.6 ± 8.2 k€, P = 0.001) were significantly lower in the preemptive group than in the prophylaxis cohort.

Three children had leukopenia (2 in the UP and 1 in the PET group) and 1 had thrombocytopenia (PET group) as adverse events of antiviral therapy.

DISCUSSION

The PET strategy used at our center resulted in 61% of CMV infections and only 3% of diseases, which are comparable with previously described outcomes obtained with prophylaxis, PET and hybrid approach.1,8,10 Moreover, preemptive strategy did not result in an increased incidence of putative CMV indirect effects, such as ACR, EBV infection and PTLD, sepsis, nor was associated with worse outcomes in terms of graft dysfunction. On the other hand, prophylaxis did not reduce the number of infections by 200 days post-LT, suggesting that putative long-term effect of CMV infection should not differ in the 2 approaches. Noteworthy, preemptive strategy proved to be associated to much lower costs per patient and to a significantly lower antiviral drug exposure, a not negligible advantage considering the potential long-term carcinogenicity and reproductive toxicity of these compounds.15,16 Another potential advantage of the lower exposure to antiviral drugs is the expected lower incidence of viral resistance to these compounds, an emerging problem in solid organ transplantation.17 As for the need of virologic surveillance that the PET implies, the weekly monitoring that we adopted for the first 90 days seems to be appropriate, because the majority of the infections occurred by this period.

Because PET only prevents CMV disease but allows viral infection or reactivation, we also evaluated whether viremia per se was associated with indirect CMV effects on LT outcomes. The development of CMV viremia in our cohort of LT children was not associated with an increased risk of possible CMV indirect effects. Particularly, children developing viremia did not have more ACR episodes, with an overall incidence of about 30%. This finding replicates that of other previous adult and pediatric studies,6,18 but it is in contradiction with others.19 The fact that most rejection episodes in infected patients actually preceded the onset of viremia seems to confirm the little importance of infection in determining rejection, conversely suggesting a possible causative role of pulsed steroid treatment as a trigger for CMV infection or reactivation.

We also investigated the possible risk factors for CMV infection in children managed with PET after LT. Only the donor/recipient serostatus match played an undisputable role in predicting viremia, as reported elsewhere.1,3,7 D+/R− status was independently associated with a 17-fold higher risk to develop infection in comparison to the D^any/R+ “standard risk” patients. Nevertheless, the D/R serostatus should be interpreted with caution in pediatric age, due to the possible confounding factor of the maternal antibodies in very young recipients. The lack of discrimination between maternal and infant antibodies—potentially leading to the underestimation of the CMV risk in a subgroup of patients—is also a limitation of our study as well as of other previous pediatric studies. Among our “standard risk” patients, about a half were at risk of having passive CMV antibodies of maternal origin whereas the risk related to transfusions of blood products was unclear in many foreign patients. This subset of patients might have accounted for the high prevalence of CMV-positive recipients among those who needed PET.

Moreover, because the D+/R− status does not imply a worse course of CMV infection, we do not recommend prophylaxis for these children nor a different threshold to start PET.

The present study is the first to compare preemptive and prophylactic approach in a large cohort of pediatric patients, including short-term and long-term outcomes evaluation and exploring CMV direct and putative indirect effects. Only 2 other comparative studies exist, 1 on aciclovir prophylaxis versus PET with GCV i.v.,20 and our center’s previous pediatric experience on the same hybrid strategy of 4 weeks GCV i.v. prophylaxis followed by PET versus PCR-guided PET alone.10 Both studies, although with important limitations, suggested that preemptive strategy is as safe as prophylaxis, and the recent metanalysis of adult studies of Mumtaz et al5—that only provided indirect comparison of the 2 approaches—seems to support this conclusion. In this scenario, our observations strengthen the rationale for test/preemptive protocols in children after LT.

Nevertheless, the present study has different limitations. First, there is a disproportion between the cohort of children managed by PET and the smaller comparative historical cohort treated with prophylaxis. These children where not randomly allocated to UP or PET, but were consecutively managed in the same center. The numerical diversity and the different period in which the children were operated make the comparison between the 2 groups of difficult interpretation. The different distribution of the serostatus categories between the 2 groups—although not statistically significant—might have played a role in the outcomes’ interpretation. The comparison of the 2 strategies remains to be more extensively evaluated in larger and more homogeneous cohorts or with randomized studies. Nevertheless, PET per se, in the studied group, demonstrated to be safe and cost-effective.

Also importantly, it is noteworthy that the prophylaxis protocol that we used is a hybrid schedule with shorter antiviral administration with respect to the previous literature, and it is followed by surveillance and possible preemptive interventions.

Another limitation concerns the definition of graft dysfunction. Graft dysfunction was defined in the presence of biochemical abnormalities of the liver function tests that led to perform liver biopsies. It has been described that 70% of children developing liver graft fibrosis have unaltered liver enzymes.21 Long-term effect of CMV infection on graft injury may be further explored by the adoption of serial protocol biopsies. Nevertheless, the fact that the rate of infected patients at 200 days is the same suggests that long-term impact on graft outcomes should not vary in patients managed with PET or prophylaxis.

Finally, the decision about the optimal CMV preventative approach should take into account the availability and the economic burden of CMV monitoring, which is even more relevant in case of foreign patients that have to be discharged to a resource-limited home country setting. On the other hand, even in these countries, the use of commercial kits for CMV monitoring may become at some point more economically and clinically sustainable than administering UP or leaving the patient abroad for a longer period. However, in centers having no availability of CMV monitoring, prophylaxis treatment is the only option and may therefore be recommended.

Even with the aforementioned limitations of this retrospective study, our findings ultimately provide evidence that PET is valuable, drug-sparing, and cost-effective and should be preferred to prophylaxis against CMV in children undergoing LT in settings where viral monitoring is available.