Clinical Predictors of Relapse after Treatment of Primary Gastrointestinal Cytomegalovirus Disease in Solid Organ Transplant Recipients

Authors


Corresponding author: Raymund R. Razonable, razonable.raymund@mayo.edu

Abstract

Primary gastrointestinal cytomegalovirus (CMV) disease after solid organ transplantation (SOT) is difficult to treat and may relapse. Herein, we reviewed the clinical records of CMV D+/R− SOT recipients with biopsy-proven gastrointestinal CMV disease to determine predictors of relapse. The population consisted of 26 kidney (13 [50%]), liver (10 [38%]) and heart (3 [12%]) transplant recipients who developed gastrointestinal CMV disease at a median of 54 (interquartile range [IQR]: 40–70) days after stopping antiviral prophylaxis. Except for one patient, all received induction intravenous ganciclovir (mean ± SD, 33.8 ± 19.3 days) followed by valganciclovir (27.5 ± 13.3 days) in 18 patients. Ten patients further received valganciclovir maintenance therapy (41.6 ± 28.6 days). The median times to CMV PCR negativity in blood was 22.5 days (IQR: 16.5–30.7) and to normal endoscopic findings was 27.0 days (IQR: 21.0–33.5). CMV relapse, which occurred in seven (27%) patients, was significantly associated with extensive disease (p = 0.03). CMV seroconversion, viral load, treatment duration, maintenance therapy and endoscopic findings at the end of therapy were not significantly associated with CMV relapse. In conclusion, an extensive involvement of the gastrointestinal tract was significantly associated with CMV relapse. However, endoscopic evidence of resolution of gastrointestinal disease did not necessarily translate into a lower risk of CMV relapse.

Introduction

In the contemporary era when antiviral prophylaxis is widely used, delayed-onset primary cytomegalovirus (CMV) disease has emerged as one of the most important clinical challenges in the management of CMV-seronegative recipients of solid organ allograft from CMV-seropositive donors (termed D+/R− patient) (1). A large number of CMV D+/R− solid organ transplant (SOT) recipients present with organ-invasive CMV disease at a delayed onset, and this most commonly involves the gastrointestinal (GI) tract (2). It is estimated that GI CMV disease accounts for 80% of all cases of organ-invasive disease in SOT recipients (3–5). Intravenous (IV) ganciclovir has been the cornerstone of therapy for CMV disease in general (6), and more specifically, for GI CMV disease, where absorption of an orally administered antiviral drug is a potential concern (6). Recently, oral valganciclovir was shown to be effective in the treatment of non life-threatening CMV diseases, including some cases of GI CMV disease, and the long-term outcome of this treatment approach was not significantly different from standard IV ganciclovir therapy (7).

The optimal duration of antiviral treatment of CMV disease is not known. Current guidelines recommend the use of CMV load in the blood as a guide to help define the duration of therapy (6,8). However, while CMV polymerase chain reaction (PCR) assay in the blood is very helpful in guiding the duration of treatment for CMV syndrome (9), its utility in organ-invasive CMV diseases has been less than satisfactory, particularly in cases of localized or compartmentalized diseases wherein the level of viremia does not necessarily reflect the severity of organ involvement. In this regard, our anecdotal clinical experience suggests that relapse of infection in cases of GI CMV disease is not uncommon even among SOT recipients with documented clearance of virus from the blood by CMV PCR. Because factors associated with success or failure in the treatment of GI CMV diseases are not well characterized, we adopted a clinical practice wherein follow-up endoscopic examination of the GI tract was performed to document resolution of GI CMV disease before discontinuing antiviral therapy. However, whether this practice translates into a lower risk of CMV relapse has not been systematically investigated. To address these issues, we performed a retrospective study investigating the outcomes of treatment of GI CMV disease among CMV D+/R− SOT recipients.

Patients and Methods

Patient population

This retrospective study was conducted during a 5-year period from May 1, 2000 to April 30, 2005. During this period, a total of 274 CMV D+/R− patients received kidney, kidney–pancreas, liver or heart transplants at the Mayo Clinic in Rochester, Minnesota. This group of patients was chosen based on known increased risk for the development of delayed-onset primary GI CMV disease. All patients received antiviral prophylaxis with oral ganciclovir or valganciclovir for 90–100 days as per our clinical practice protocols. Thirty-five (12.7%) of these patients developed symptoms suggestive of primary CMV disease involving the GI tract; however, only 26 (9.5%) patients had biopsy-confirmed tissue-invasive GI CMV disease. All patients consented to review of their medical records. The study was approved by the Institutional Review Board of the Mayo Foundation.

Definition of GI CMV disease

Tissue-invasive GI CMV disease was defined as the presence of characteristic CMV inclusions, a positive CMV-specific immunoperoxidase stain, or positive in situ hybridization for CMV in tissue specimens obtained from SOT recipients presenting with signs and symptoms consistent with GI tissue-invasive disease (10). The occurrence of CMV viremia alone was not sufficient to establish the diagnosis even in a patient with GI symptoms if the virus was not detected in tissue specimens. Tissue-invasive GI CMV disease was classified as upper GI disease if it involved the esophagus, stomach or duodenum, and lower GI disease if it involved the terminal ileum or colon. Simultaneous involvement of both upper and lower GI tract was termed extensive GI CMV disease.

Clinical follow-up

The medical records of all 26 patients with biopsy-proven GI CMV disease were reviewed. Data collected included demographic characteristics, clinical presentation, endoscopic and histological findings, CMV seroconversion after diagnosis, development of acute rejection and its treatment, CMV load and decline during therapy, induction and maintenance antiviral therapy and outcome. The primary outcome of interest in this study was CMV relapse, which was defined as recurrence of CMV viremia and/or disease during or early (<3 months) after induction antiviral therapy. CMV viremia and disease were considered together as a composite outcome because both conditions were retreated with induction doses of antiviral therapy.

All patients with CMV disease were followed-up clinically and virologically using CMV PCR on a weekly basis. Induction anti-CMV therapy was administered at least until resolution of clinical symptoms and CMV viremia. In addition, it was also a routine clinical practice in our institution to perform follow-up endoscopy to document endoscopic and histopathologic resolution of GI CMV disease before stopping induction antiviral therapy. Biopsy specimens were obtained from the involved site, usually after 3–4 weeks of induction anti-CMV therapy. In cases of persistent GI CMV disease, induction therapy was extended until another repeat endoscopy confirmed clearance of the virus from infected tissue.

Statistical analysis

The study cohort was characterized using descriptive statistics. Continuous parameters were expressed by mean and standard deviation (SD) or median and interquartile range (IQR) as appropriate, while categorical data were represented by the number of patients and percentages. Univariate analysis using logistic regression model was performed to evaluate variables associated with the primary outcome of relapse of viremia or disease during or early (<3 months) after finishing treatment for GI CMV disease. JMP 7 (Cary, NC) was used to perform logistic regression analysis. Hazard ratios (HRs) with 95% confidence intervals (95% CI) were computed for variables significantly associated with outcome. A p value of <0.05 was considered statistically significant.

Results

GI CMV disease and its treatment

Twenty-six of 274 (9.5%) CMV D+/R− patients developed biopsy-proven GI CMV disease after kidney (13 [50%]), liver (10 [38%]) or heart (3 [12%]) transplantation. The clinical and demographic characteristics of all 26 patients are summarized in Table 1. The mean (±SD) age of the patients at the time of GI CMV disease diagnosis was 47.5 (±15.0) years. The majority (73%) of patients were male.

Table 1.  Clinical and demographic characteristics of 26 CMV D+/R− solid organ transplant recipients who developed CMV gastrointestinal disease
Variablesn (%)1
  1. 1Values are presented as number of patients (percentage), unless otherwise specified. CMV = cytomegalovirus; GI = gastrointestinal; IQR = interquartile range; PCR = polymerase chain reaction.

Age at transplant in years, mean (±SD)47.5 ± 15
Male gender19 (77)
Caucasian24 (92)
Transplanted organ
 Kidney13 (50)
 Liver10 (38)
 Heart 3 (12)
Maintenance immunosuppressive therapy
 Prednisone24 (92)
 Tacrolimus23 (88)
 Mycophenolate mofetil20 (77)
 Sirolimus 3 (12)
 Cyclosporine2 (8)
 Azathioprine2 (8)
Median duration of CMV primary prophylaxis in days (IQR)92 (91–92)
Median time-to-CMV disease after prophylaxis in days (IQR)54 (40–70)
Median viral load at the time of diagnosis (copies/mL)154 700
Extent of GI disease
 Upper GI disease 7 (27)
 Lower GI disease16 (62)
 Upper and lower GI disease 3 (12)
Treatment for acute rejection 4 (15)
Initial induction therapy
 Intravenous ganciclovir25 (96)
 Valganciclovir1 (4)
Maintenance therapy10 (38)
Median time to negative CMV PCR in days (IQR)23 (16–31)
Follow-up endoscopy21 (81)
Relapse during or early after finishing therapy for GI CMV disease 6 (23)
Death1 (4)

The diagnosis of GI CMV disease was established at a median of 54 days (IQR: 40–70) after cessation of antiviral prophylaxis. CMV disease involved the upper GI tract in seven (27%) patients, the lower GI tract in 16 (62%) and both upper and lower GI tracts in three (12%) patients. Mucosal erosions and ulcerations were the most common endoscopic findings at the time of diagnosis (66.7%), although some patients had grossly normal mucosa (12.5%) or mucosal hyperemia only (20.8%). Histopathologic examination demonstrated cytomegalic cells with characteristic inclusions in the majority of patients (80%); however, a positive immunoperoxidase stain or in situ hybridization was required to demonstrate tissue-invasive CMV disease in 20% of patients. The median CMV load at the time of diagnosis was 154 700 copies/mL (IQR: 55 375–288 850).

All patients except one received induction therapy with IV ganciclovir for a mean (±SD) duration of 33.8 (±19.3) days; the remaining patient received oral valganciclovir for 30 days. Eighteen of the 25 patients who received IV ganciclovir were eventually transitioned to renally adjusted induction dose of valganciclovir (equivalent of 900 mg twice daily) for a mean (±SD) duration of 27.5 (±13.3) days. Ten patients subsequently received renally adjusted maintenance dose of valganciclovir (equivalent of 900 mg once daily) for a mean (±SD) duration of 41.6 (±28.6) days.

Clearance of CMV from blood and gastrointestinal tract

The median half-life of viral load decline during antiviral therapy was 4.7 days (IQR: 3.5–9.1). The median time-to-CMV PCR negativity in the blood was 22.5 days (IQR: 16.5–30.7; mean ± SD, 23.8 ± 9.45) after the start of induction therapy. Follow-up endoscopy was performed in 20 (77%) of the 26 patients. Clearance of CMV from tissue specimens was documented in 17 (85%) of 20 patients. Among 17 patients with negative follow-up endoscopy, the median time-to-normal endoscopic finding was 27.0 days (IQR: 21.0–33.5; mean ± SD, 28.6 ± 10.6) after start of induction therapy. Three of 20 patients had evidence of persistent GI CMV disease on follow-up endoscopy performed at a median of 32 days after the start of induction antiviral therapy.

Outcomes and predictors of CMV relapse

During the mean (±SD) follow-up period of 979 (±550) days, relapse of CMV viremia or disease was observed in seven (27%) patients. In one of these seven patients, CMV viremia relapsed while the patient was still receiving induction IV ganciclovir therapy. In this patient, CMV viremia was low with viral load copies that ranged from 2500 to 5000 copies/mL. Ganciclovir resistance was suspected but genotype testing could not be performed because viral DNA could not be amplified. Because of low viral load, the patient was continued on therapy with intravenous ganciclovir with eventual resolution of viremia 1 month later. Five of the seven patients had relapse of CMV viremia (n = 3) or disease (n = 2) within 3 months after completing antiviral treatment. Finally, the seventh patient had relapse of asymptomatic CMV viremia >3 months after finishing antiviral therapy. All patients with CMV relapse were retreated with induction doses of valganciclovir or IV ganciclovir.

Table 2 shows the results of univariate analysis of potential predictors of the primary outcome of CMV relapse. Of all factors evaluated, only the extent of GI disease at the time of diagnosis was significantly associated with a higher risk of CMV relapse. Transplant recipients with extensive CMV disease (with involvement of both the upper and lower GI tracts) at the time of diagnosis had a significantly higher risk of recurrent CMV relapse after completion of antiviral therapy (hazard ratio [confidence interval], 4.8 [1.14–20.7]; p = 0.031). In contrast, CMV seroconversion at the time of diagnosis, the peak viral load, the rate of viral load decline and treatment for acute cellular rejection during the 2 months prior to and 1 month following the diagnosis of CMV disease were not significantly associated with the primary outcome. An initial endoscopic finding of a grossly normal mucosa or mucosal erythema, as opposed to mucosal erosions and ulcerations, were not significantly associated with a lower risk of CMV relapse. Transplant recipients whose GI CMV disease diagnosis was based on a positive CMV-specific immunoperoxidase staining or in situ hybridization only were not significantly less likely to experience CMV relapse compared to those who had clearly visible CMV inclusions on histopathology. A longer duration of induction antiviral therapy was not associated with a lower risk of CMV relapse. Likewise, administration of valganciclovir maintenance therapy was not significantly associated with protection from CMV relapse. The incidence of CMV relapse was 25% among patients who received maintenance valganciclovir and 25% among those who did not receive maintenance valganciclovir therapy. Finally, endoscopic evidence of resolution of GI disease before stopping induction antiviral therapy was not significantly associated with protection from CMV relapse.

Table 2.  Univariate analysis of predictors of CMV viremia or disease relapse during or early after finishing treatment for GI CMV disease
Variablep-Value
  1. 1Endoscopic findings were stratified based on the presence of mucosal ulcerations and erosions, mucosal erythema or grossly normal mucosa.

  2. 2Histopathologic findings were stratified based on whether the diagnosis of CMV disease was made based on characteristic CMV inclusions alone whether immunoperoxidase stain or in situ hybridization was necessary for diagnosis.

  3. 3Upper and lower gastrointestinal tract disease versus upper or lower gastrointestinal tract disease; CMV = cytomegalovirus; PCR = polymerase chain reaction.

Peak CMV viral load0.33
Half-life of viral decline0.43
Time-to-negative CMV PCR0.18
Endoscopic findings10.39
Histopathologic findings20.11
Negative pathology at the end of therapy0.10
Positive CMV serology or seroconversion after diagnosis0.97
Treatment of acute rejection0.94
Duration of induction therapy0.52
Maintenance valganciclovir therapy1.00
Duration of maintenance therapy0.34
Extent of GI disease30.03

Discussion

This study highlights the difficulty and complexity of treating primary GI CMV disease in CMV D+/R− SOT recipients. Even with a very prolonged course of treatment, recurrence of CMV infection and clinical disease is not uncommon. This study identified extensive CMV disease, as indicated by the involvement of both the upper and lower GI tracts, as a significant predictor of relapsing CMV viremia or disease. In contrast, the degree of viral replication (as measured by the peak viral load in the blood), the rate of viral decline during antiviral therapy, treatment for acute cellular rejection and CMV seroconversion at the time of diagnosis were not significantly associated with CMV relapse. Likewise, endoscopic and histopathologic findings at the time of diagnosis, endoscopic evidence of resolution of GI CMV disease at the end of induction antiviral therapy, a longer duration of induction antiviral therapy and the administration of maintenance therapy were not significantly associated with CMV relapse.

Previous studies have indicated that 23–33% of all cases of primary CMV disease relapse after a defined course of treatment (9,11). The propensity of CMV infection or disease to relapse led to the search for and adoption of strategies aimed at minimizing this risk. Foremost among these strategies is the use of viral quantification as a guide to assess the duration of treatment (6,9,12,13). This standard practice was based on studies demonstrating that persistent CMV viremia at the end of treatment, the rate of viral load decline and peak viral load were significantly associated with the risk of CMV relapse (9,11). However, this approach may not be optimal in some cases of tissue-invasive CMV disease, because it is not uncommon for the virus to be cleared from the systemic circulation while it persists in tissue. Indeed, this study of biopsy-proven GI CMV disease did not find significant associations between the degree of viral replication and decay and the risk of CMV relapse after treatment. Furthermore, some cases of GI CMV disease could be compartmentalized and may not be accompanied by detectable viremia, especially among CMV seropositive recipients with reactivation disease. In these situations, viral load surveillance in the blood is not useful for clinical follow-up.

We had initially hypothesized that repeat endoscopic examination of the GI tract with collection of tissue specimens to confirm the resolution of invasive CMV disease would be useful to define the duration of treatment and to prevent CMV relapse. With this clinical approach, induction antiviral therapy was continued until there was no endoscopic or histopathologic evidence of persistent CMV disease in the GI tract. However, this study demonstrates that viral clearance from the GI tract per se, as indicated by repeat endoscopy and histopathologic examination, does not necessarily translate into protection from CMV relapse. Indeed, the rate of relapse of CMV viremia or disease remains high (27%) and comparable to previous reports (23–33%) (9,11). Accordingly, if a repeat invasive endoscopic procedure, with the risk and cost that it entails, does not serve as a useful guide to assess the risk of CMV relapse, this practice should not be recommended as a standard of care for all patients with GI CMV disease. Moreover, because GI CMV disease is often patchy and multifocal, a false-negative histopathologic examination may occur as a result of sampling error when tissue biopsies are obtained from normal-looking mucosa during follow-up endoscopy.

Follow-up endoscopic examination in the majority of our patients provided unique insights into the kinetics of viral eradication from the GI tract. This study demonstrates an earlier clearance of CMV from blood compared to the GI tract. While this observation is limited by the infrequent use of follow-up endoscopy (i.e. performed once at 3–4 weeks, and thereafter as indicated), it was evident that some patients had cleared their peripheral viremia but remained with active gastrointestinal disease. Hence, clearance of circulating virus should not be used as a surrogate marker for CMV eradication from infected GI tissue. The potential clinical implication of this finding is the need to treat patients with GI CMV disease for a duration which extends beyond the time of documented viral eradication in the blood. Indeed, patients in this study received prolonged courses (4–8 weeks) of induction therapy with IV ganciclovir or oral valganciclovir. The prolonged duration of therapy correlates with the mean time to negative endoscopic findings of 4 weeks. However, despite the prolonged course of antiviral treatment, relapse of CMV disease or viremia was not uncommon. Consequently, this raises the question as to whether maintenance treatment may be beneficial (7). This study observed similar rates of CMV relapse between patients who received or those who did not receive maintenance valganciclovir therapy. One likely reason for relapse despite the use of maintenance valganciclovir therapy (and even after documented clearance of CMV from the GI tract) is a persistent defect in CMV-specific immunity (14,15). Host immunity, as measured by CMV seroconversion, has not been shown to be significantly associated with CMV relapse (13). Whether measurement of CMV-specific T cells would be a better marker will need to be addressed in future studies (14,15).

This study has limitations related to its retrospective and single-center design. The number of patients was limited by the requirement for biopsy-proven cases, because part of our aim was to assess the utility of follow-up endoscopy. Nonetheless, we believe that this is the largest collection of cases of biopsy-proven primary GI CMV disease with follow-up endoscopy to monitor treatment response. Because of our requirement for tissue diagnosis, this inclusion criterion may have biased our population by selecting patients with severe disease, and may have excluded mild cases of GI CMV diseases that did not require endoscopy and biopsy. We emphasize, however, that even patients with grossly normal endoscopic findings but with CMV detected only histopathologically were included in our cohort. Hence, it is likely that the spectrum of GI CMV disease is well represented in this study. Our study population consisted of only high-risk CMV D+/R− patients, and therefore, the extrapolation of the results to CMV R+ patients with GI CMV disease may not be appropriate. Conversely, one of the strengths of this study is that it presents a large cohort of biopsy-proven cases of primary GI CMV disease with follow-up endoscopy. The availability of follow-up endoscopy in this study sheds unique insights into the kinetics of viral clearance from the GI tract. In this regard, we believe that this is the first and only study that has addressed this issue in the clinical setting.

In conclusion, our study illustrates the difficulties of treating primary CMV GI disease in SOT recipients. This study, which we believe is the first to simultaneously evaluate the kinetics of CMV clearance in the blood and GI tissue, provides solid evidence of CMV persistence in GI tissues long after the virus has been eliminated from peripheral blood. This study therefore has direct implications in clinical practice, as our data would support more prolonged antiviral therapy in patients with GI CMV disease. However, the optimal duration is not defined. Despite prolonged antiviral therapy, relapse of CMV viremia or disease occurred in several of our patients. In this regard, this study shows that simultaneous involvement of both the upper and lower GI tract is significantly associated with CMV relapse. However, follow-up endoscopy with tissue biopsies to document clearance of virus from GI tissue does not necessarily translate clinically into protection from CMV relapse. Therefore, considering the expense and invasive nature of endoscopy, such a clinical practice should not be routinely recommended in SOT recipients with limited primary GI CMV disease.

Ancillary