Delayed-Onset Primary Cytomegalovirus Disease and the Risk of Allograft Failure and Mortality after Kidney Transplantation

Abstract

Background. During the contemporary era of antiviral prophylaxis, the impact of delayed-onset primary cytomegalovirus (CMV) disease on the outcome of kidney transplantation is not known. We evaluated the incidence, clinical features, risk factors, and outcomes of CMV disease among high-risk kidney transplant recipients.

Methods. The medical records of CMV-seronegative recipients of kidney transplants from CMV-seropositive donors were reviewed. Cox proportional hazards regression was used to identify factors associated with CMV disease and to assess its impact on allograft loss and mortality.

Results. None of the 176 CMV-seronegative recipients of kidney transplants from CMV-seropositive donors developed breakthrough CMV disease during a median of 92 days (interquartile range, 90–92 days) of oral ganciclovir or valganciclovir prophylaxis. Thereafter, 51 patients (29%) developed CMV disease at a median of 61 days (interquartile range, 40–143 days) after stopping antiviral prophylaxis. Early-onset bacterial and fungal infection (hazard ratio, 3.61; 95% confidence interval, 1.78–7.33; P<.001) and a Charlson comorbidity index ⩾3 (hazard ratio, 2.21; 95% confidence interval, 1.15–4.22; P=.011) were associated with a higher risk of delayed-onset primary CMV disease, and postrejection antiviral prophylaxis (hazard ratio, 0.29; 95% confidence interval, 0.09–0.94; P=.039) was associated with a lower risk of such CMV disease. A time-dependent Cox regression analysis revealed a statistically significant association between tissue-invasive CMV disease and allograft loss or mortality (hazard ratio, 2.85; 95% confidence interval, 1.22–6.67; P=.016).

Conclusion. This study of a large cohort of CMV-seronegative recipients of kidney transplants from CMV-seropositive donors illustrates the ongoing challenge of delayed-onset primary CMV disease and its impact on transplantation outcomes despite antiviral prophylaxis. Better strategies for CMV disease prevention after kidney transplantation are warranted.

Through its direct and indirect effects, cytomegalovirus (CMV) is associated with significant clinical illness, allograft loss, and mortality after kidney transplantation [1–5]. Beyond the direct clinical manifestations of CMV syndrome or tissue-invasive disease, the virus indirectly increases predisposition to allograft rejection and opportunistic infections [1, 6, 7]. The strongest risk factor for CMV disease is a lack of CMV-specific immunity. Thus, CMV-seronegative recipients of allograft from CMV-seropositive donors (CMV D⁺/R⁻ transplant recipients) are considered to be at highest risk of developing CMV disease [5, 8]. Without anti-CMV prophylaxis, 40%–58% of CMV D⁺/R⁻ kidney transplant recipients develop CMV disease, usually during the first 3 months after transplantation (early-onset CMV disease) [2, 9–15]. Because of anti-CMV prophylaxis, the incidence of early-onset CMV disease has decreased, but conversely, the rate of delayed-onset CMV disease has increased [6, 16–19].

The impact of delayed-onset primary CMV disease on the outcome of kidney transplantation is not known. We hypothesized that CMV disease, despite its delayed onset, will continue to negatively influence allograft and patient survival. Supporting this hypothesis is a recent study demonstrating a significant association between delayed-onset CMV disease and mortality during the first year after liver transplantation [17]. Accordingly, several strategies have been proposed to prevent delayed-onset CMV disease, including—but not limited to—prolonged antiviral prophylaxis [20]. However, the theoretical risks of drug resistance, toxicity, and cost incurred with prolonged prophylaxis have limited enthusiasm for this approach. A selective strategy of targeting CMV D⁺/R⁻ transplant recipients with clinical characteristics that place them at increased risk of delayed-onset primary CMV disease has been proposed. In this context, we performed this study to assess potential risk factors for delayed-onset CMV disease among CMV D⁺/R⁻ kidney transplant recipients. In addition, we questioned whether CMV disease, despite its delayed onset, influences allograft and patient survival after kidney transplantation.

Patients and Methods

Patient population. This retrospective study was conducted among patients who underwent kidney—with or without other solid organ (i.e., liver, heart, or pancreas)—transplantation at the Mayo Clinic (Rochester, MN) from 1 January 2000 through 31 December 2004. Because our major aim was to define clinical features, risk factors, and outcome of delayed-onset primary CMV disease, we studied all adult patients who were CMV D⁺/R⁻ kidney transplant recipients, received oral ganciclovir or valganciclovir prophylaxis, and had allograft and patient survival of ⩾3 months. All patients consented to the review of medical records. This study was approved by the Institutional Review Board of the Mayo Foundation.

Clinical follow-up. The medical records of all patients were reviewed for demographic and clinical characteristics from the time of transplantation to the last day of follow-up, death, allograft loss, retransplantation, or end of data abstraction (May 2006). The data included patient age and sex, indication for transplantation, underlying comorbidity (as assessed by Charlson comorbidity index [21] 3 months after transplantation), antiviral prophylaxis, acute graft rejection, immunosuppressive drugs, bacterial and fungal infections, and the outcomes of CMV disease, allograft loss, and mortality. Per protocol, allograft biopsy was performed for all patients at the 3–4-month follow-up visit, yearly thereafter, and when clinically indicated. The Charlson comorbidity index [21] contains 19 components, which are defined using International Classification of Diseases, Ninth Revision code, and includes such conditions as myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcer disease, liver disease, diabetes, hemiplegia, renal disease, solid tumor, leukemia, lymphoma, and AIDS. Each component was scored on the basis of severity, as described elsewhere [21].

CMV prophylaxis. Consistent with current guidelines [22], our protocol has been to provide CMV D⁺/R⁻ kidney transplant recipients with anti-CMV prophylaxis with ganciclovir (1 g orally 3 times daily; prior to October 2001) or valganciclovir (900 mg orally once daily; since October 2001) for 3 months after transplantation. In addition, CMV D⁺/R⁻ transplant recipients who received treatment for acute allograft rejection began receiving antiviral prophylaxis for 1–3 months, beginning at the time of acute graft rejection. The dose of medications was adjusted on the basis of renal function. Routine surveillance using CMV PCR assay was not performed.

Definitions of CMV infection and disease. CMV disease was classified either as CMV syndrome or tissue-invasive disease, according to published guidelines [23]. CMV syndrome denotes the detection of CMV accompanied by fever, fatigue, malaise, leukopenia, thrombocytopenia, and/or arthralgias. CMV syndrome required that other causes had been ruled out and that there was no indication of tissue-invasive disease. In contrast, tissue-invasive CMV disease was diagnosed when there were clinical symptoms and signs of tissue invasion, accompanied by detection of CMV in a blood specimen and virologic and/or histologic detection of CMV in a biopsy specimen. The detection of CMV in blood or other body fluid specimens in the absence of symptoms was defined as CMV infection.

Induction and maintenance immunosuppressive treatments. The majority of patients received induction immunosuppressive therapy with antithymocyte globulin (63% of patients), anti-CD3 monoclonal antibody (6%), or IL-2 receptor blocker (13%). Patients who underwent positive cross-match or ABO blood group-incompatible transplantation underwent plasmapheresis and received antithymocyte globulin, intravenous immunoglobulin, and rituximab (12% of patients). Patients received maintenance immunosuppressive therapy with prednisone (99% of patients), tacrolimus (89%), and mycophenolate mofetil (95%). A few patients were treated with sirolimus, azathioprine, and cyclosporine.

Statistical analysis. Descriptive statistics were used to describe demographic and clinical characteristics. Incidence was estimated using the Kaplan-Meier method. Potential risk factors for delayed-onset CMV disease, such as age, sex, comorbid disease, type of immunosuppressive therapy, invasive bacterial and fungal infection, and acute graft rejection, were analyzed using Cox proportional hazard regression analysis. Time-dependent factors were used when appropriate. We analyzed the association between CMV disease and immunosuppressive therapy regimen at the end of antiviral prophylaxis. CMV disease as a risk factor for allograft loss or patient mortality was analyzed using Cox proportional hazard regression, with CMV as a time-dependent factor. Statistical significance was set at P⩽.05.

Results

Patient characteristics. During the 5-year study period, a total of 1127 patients received a kidney (with or without other solid organ) transplant at the Mayo Clinic, including 193 patients (17.1%) who received CMV D⁺/R⁻ transplants. After excluding 8 patients who experienced allograft loss or died within 3 months after transplantation, 3 patients for whom CMV-specific antiviral prophylaxis was not confirmed, and 6 patients who denied research participation, there was a total of 176 CMV D⁺/R⁻ kidney transplant recipients who qualified for this study. One hundred fifteen (65%) of the patients were men, and 61 (35%) were women. The median age was 47 years (range, 18–75 years). The majority (91%) of patients were white. In addition, the majority of patients (133 patients [76%]) received a kidney from living donors. Fourteen patients (8%) received a positive cross-match transplant, and 7 (4%) were ABO blood group incompatible with the donor. Thirteen patients (7.4%) underwent kidney transplantation, in addition to pancreas (4 patients), liver (8), or heart (1) transplantation. The patients' characteristics are shown in table 1.

Incidence and onset of delayed-onset primary CMV disease. All patients received prophylaxis with oral ganciclovir (34 patients [19%]) or valganciclovir (142 [81%]) for a median duration of 92 days (interquartile range, 90–92 days) after transplantation. During the period of antiviral prophylaxis, no patient developed breakthrough CMV disease. Thereafter, 51 patients (29%) developed CMV disease at a median of 157 days (interquartile range, 127–235 days) after transplantation and a median of 61 days (interquartile range, 40–143 days) after completing prophylaxis. Kaplan-Meier estimates for CMV disease were 0%, 18%, 27%, 29%, 30%, 30%, and 30% at 3, 6, 12, 18, 24, 30, and 36 months after transplantation, respectively (figure 1).

Clinical features and treatment of delayed-onset primary CMV disease. There were similar proportions of patients with CMV syndrome (49% [25 patients]) and tissue-invasive disease (51% [26 patients]). The majority of cases of tissue-invasive CMV disease involved the gastrointestinal tract (21 cases [41%]), including 1 case involving concurrent pneumonitis. There were 2 cases of CMV nephritis (4%), and there was 1 case each of retinitis (2%), pancreatitis (2%), and hepatitis (2%).

Twenty-four patients (47%) received treatment with intravenous ganciclovir followed by oral valganciclovir, 14 (27%) received valganciclovir treatment alone, 7 (14%) received intravenous ganciclovir alone, and 6 (12%) received various drug combinations including foscarnet, cidofovir, or CMV hyperimmunoglobulin. Relapse of CMV disease occurred in 5 (9.8%) of 51 patients.

Ganciclovir resistance was clinically suspected in 8 (16%) cases of CMV disease. However, only 4 patients (8%) with CMV disease had genotypic mutations known to confer antiviral resistance; 3 patients had UL97 mutations (ganciclovir resistance), and 1 had a UL54 mutation that conferred resistance to ganciclovir and foscarnet but not to cidofovir.

Potential risk factors for CMV disease. In a univariate Cox proportional hazard model (table 2), a Charlson comorbidity index of ⩾3 was associated with delayed-onset CMV disease (hazard ratio [HR], 2.207; 95% CI, 1.155–4.218; P=.016). The occurrence of bacterial and fungal infection during the receipt of antiviral prophylaxis was also associated with subsequent CMV disease (HR, 2.022; 95% CI, 0.992–4.121; P=.052), especially when bacterial and fungal infections occurred during the first (HR, 5.865; 95% CI, 2.711–12.689; P<.001) and second (HR, 3.612; 95% CI, 1.780–7.328; P=.004) months after transplantation. Other potential factors, such as patient age and sex, diabetes mellitus, and specific induction and maintenance immunosuppressive drugs, were not statistically significantly associated with delayed-onset CMV disease, although a trend was observed with ABO blood group-incompatible transplantation (HR, 2.54; 95% CI, 0.915–7.074; P=.115).

In a time-dependent variable Cox proportional hazard model, acute graft rejection was associated with a lower risk of delayed-onset CMV disease (HR, 0.335; 95% CI, 0.120–0.993; P=.036). Notably, the majority (93%) of patients received 1–3 months of antiviral prophylaxis after treatment for acute graft rejection. When treated acute graft rejection (which was followed by antiviral prophylaxis) was analyzed (i.e., postrejection antiviral prophylaxis), a lower risk of CMV disease was observed (HR, 0.292; 95% CI, 0.095–0.940; P=.039).

Impact of CMV disease on graft loss and death. During the median follow-up period of 3.0 years (range, 0.31–6.1 years; mean duration±SD, 3.0±1.42 years), 34 patients (20.7%) died or experienced allograft failure. Twenty patients died (including 8 patients who experienced allograft failure prior to death), and 14 patients required the initiation of hemodialysis. Causes of death were bacterial and fungal infections (in 15% of patients), cardiovascular causes (25%), malignancy and posttransplantation lymphoproliferative disease (10%), and unspecified causes (50%). Kaplan-Meier estimates of allograft failure or death at 1, 2, and 3 years after transplantation were 7%, 12% and 20%, respectively. Using time-dependent Cox proportional hazard model, delayed-onset primary tissue-invasive CMV disease was statistically significantly associated with allograft loss or death (HR, 2.852; 95% CI, 1.218–6.675; P=.016), and CMV syndrome was not associated with these outcomes (HR, 0.739; 95% CI, 0.258–2.117; P=.573). The association between tissue-invasive CMV disease and allograft loss or death remained statistically significant (HR, 2.519; 95% CI, 1.068–5.941; P=.035), even after adjusting for Charlson comorbidity index (HR, 2.682; 95% CI, 1.101–6.536; P=.030).

Discussion

The major finding of this study was the significant association between delayed-onset tissue-invasive primary CMV disease and the outcome of allograft failure and mortality after kidney transplantation. This finding is clinically relevant, because CMV disease remains common—albeit at a delayed onset—among CMV serostatus-mismatched kidney recipients, despite antiviral prophylaxis. Notably, almost all cases of delayed-onset tissue-invasive CMV disease involved the gastrointestinal tract, a large number of cases of delayed-onset primary CMV disease were clinically suspected to be attributable to drug-resistant virus, and bacterial and fungal infections and a higher comorbidity rate were associated with delayed-onset primary CMV disease. Collectively, these observations illustrate the current challenges of post-kidney transplantation CMV disease and highlight the opportunity to improve strategies for CMV management.

The current recommendation for preventing CMV disease in CMV D⁺/R⁻ kidney transplant recipients is antiviral prophylaxis [22]. However, as illustrated in our study and in other studies [10, 16, 19, 24], this approach does not completely eliminate the risk of CMV disease in the population at highest risk of CMV disease. On the basis of our study, we estimate that roughly 1 in 3–4 CMV D⁺/R⁻ kidney transplant recipients will subsequently develop delayed-onset primary CMV disease, with the vast majority of cases occurring during the first 6 months after stopping prophylaxis (and, very rarely, thereafter). Although this incidence is already a remarkable improvement, compared with the rate of ∼50% among those who did not receive prophylaxis [10], it highlights the need for further improving our preventive strategy. In this context, a clinical trial is ongoing to compare 100 days with 200 days of prophylaxis in kidney recipients. Foreshadowing what may be expected from this trial is a recent study of 68 CMV D⁺/R⁻ kidney transplant recipients that demonstrated a significantly lower incidence of CMV disease among those who received 24 weeks, compared with 12 weeks, of oral ganciclovir prophylaxis [25]. The potential major drawbacks to prolonged antiviral prophylaxis are the risk of antiviral resistance [26], an increase in drug toxicity (particularly of myelosuppressive drugs), overall drug cost [20, 24], and the concern that this approach will only further delay the incubation and onset of disease in a subset of patients [27]. Indeed, delayed-onset primary CMV disease was seen even after 5 years of antiviral prophylaxis [27], suggesting that patients will continue to be at risk of primary CMV disease as long as they lack or have insufficient CMV-specific T cell immunity [28].

A strategy of identifying clinical variables that could stratify CMV D⁺/R⁻ kidney transplant recipients according to their postprophylaxis risk of CMV disease so that they can be targeted for prevention has been proposed [20]. In previous studies, female sex [29], a lower creatinine clearance [29], and allograft rejection [2] were associated with delayed-onset primary CMV disease among a heterogenous cohort of solid organ (including kidney) transplant recipients. However, risk factors that are specific only to CMV D⁺/R⁻ kidney transplant recipients have not been assessed. This study, which investigated the largest cohort of CMV D⁺/R⁻ kidney transplant recipients to date, demonstrates that Charlson comorbidity index (a method of classifying prognostic comorbidity in longitudinal studies [21]) may be significantly associated with delayed-onset primary CMV disease, although its significance was noted only when analyzed as a dichotomous (but not as a continuous) variable. Bacterial and fungal infections that occur early after kidney transplantation were significantly associated with delayed-onset primary CMV disease [30, 31]. This was possibly due to the ability of these infections to cause the release of inflammatory mediators, such as TNF-α, that transactivate CMV (also known as the cytokine storm) [32]. Indeed, prior studies have demonstrated the relationship between bacterial infections and CMV disease after transplantation [33].

In contrast to previous observations [2], this study did not demonstrate a positive association between CMV disease and acute graft rejection, which is also characterized by a proinflammatory environment. We believe that this negative association was accounted for by our practice of providing anti-CMV prophylaxis to CMV D⁺/R⁻ transplant recipients receiving treatment for acute graft rejection. Indeed, this practice was adapted as a result of our previous study demonstrating the association between graft rejection and CMV disease [2]. Thus, our current finding suggests that the anti-CMV prophylaxis, which was provided to CMV D⁺/R⁻ transplant recipients who received treatment for acute graft rejection, reversed the risk of CMV disease associated with acute graft rejection. Whether this practice will also benefit CMV D⁺/R⁻ transplant recipients who develop bacterial and fungal infections early after kidney transplantation should be studied.

Anecdotal experience suggests that delayed-onset primary CMV disease may be less severe in the contemporary era of antiviral prophylaxis. In this cohort, the distribution was similar between CMV syndrome and the more aggressive tissue-invasive CMV disease (which predominantly included gastrointestinal involvement). In contrast, CMV hepatitis, nephritis, pancreatitis, retinitis, and pneumonitis were less common. This finding illustrates a changing pattern of CMV disease in the era of antiviral prophylaxis and is in contrast to the more common occurrence of pneumonitis, hepatitis, and nephropathy among kidney recipients prior to the era of antiviral prophylaxis [10, 34]. Also contributing to the changing pattern of posttransplantation CMV is the emergence of drug-resistant virus [20]. Recent studies have suggested that ganciclovir-resistant CMV occurs most commonly among CMV D⁺/R⁻ lung or pancreas transplant recipients and less frequently among recipients of kidney transplants alone [26]. Thus, it was surprising to observe that as many as 16% of CMV cases in our study were suspected clinically to be attributable to ganciclovir-resistant virus. Although only one-half of these clinically suspected cases were found to have genotypic mutations that confer drug resistance, this observation illustrates the potential difficulty of treating delayed-onset CMV disease. Prolonged suboptimal systemic ganciclovir concentration in an environment that permits rapid viral replication (such as in a CMV D⁺/R⁻ transplant recipient) has been a suggested mechanism for ganciclovir-resistant CMV [5]. In our patients with genotypically confirmed ganciclovir-resistant CMV, the initial renally adjusted dose of valganciclovir prophylaxis was not further adjusted when renal function improved. Efforts to prevent the emergence of drug-resistant CMV through judicious use of antiviral drugs, including dose adjustments, may circumvent this morbid complication [35].

Importantly, our study highlights the continued negative impact of tissue-invasive CMV disease on allograft and patient survival. Experimental studies have supported the role of CMV in allograft nephropathy [36–40]. Likewise, clinical studies conducted prior to the era of antiviral prophylaxis have shown the impact of CMV on allograft loss and mortality after kidney transplantation [4, 15]. The mechanism underlying these clinical observations has been attributed to the indirect viral effect, possibly through immune modulation [7]. In our cohort, none of deaths due to known causes were directly related to CMV. However, many were due to infections, and at least 1 was due to posttransplantation lymphoproliferative disease, which has been described as an indirect CMV effect. These observations suggest that the current standard of antiviral prophylaxis does not completely protect against the indirect CMV effect of allograft loss and mortality.

In conclusion, delayed-onset tissue-invasive CMV disease was significantly associated with allograft loss and mortality after kidney transplantation. Delayed-onset CMV disease is more common among patients with a high comorbidity index and early bacterial and fungal infections. In contrast, postrejection antiviral prophylaxis decreases the risk of delayed-onset primary CMV disease. Manifested most commonly as a febrile syndrome and gastrointestinal disease, in some cases, CMV disease may be caused by a ganciclovir-resistant virus. These observations highlight the current challenges and the opportunities to better define the management of CMV in kidney transplant recipients who are at high risk of CMV disease.

Acknowledgments

Potential conflicts of interest. All authors: no conflicts.

References