PDF
Citation
Print
INTRODUCTION
Cytomegalovirus (CMV) infection is a common occurrence following solid-organ transplantation and represents the most prevalent opportunistic pathogen among renal graft recipients [1]. The infection can arise through endogenous reactivation of the virus in the recipient, transmission of donor-derived infection through the transplanted kidney, or acquisition of the infection through contact. Clinically, CMV infection presents in three forms: asymptomatic; as a viral syndrome characterized by fever, malaise, and leukopenia; or as a specific tissue-invasive disease. The last of these can lead to conditions such as pneumonia, hepatitis, encephalitis, retinitis, nephritis, or myocarditis. Furthermore, CMV syndrome is associated with an increased risk of graft rejection [2] and chronic graft injury [3].
Risk categories are determined by the presence or absence of specific anti-CMV antibodies in the recipient (R) and donor (D). The group at highest risk is D+R−; this is followed by D+R+ and D−R+, both with an intermediate risk, and D−R−, with a low risk [4].
CMV infection control in transplant cases involves prophylactic therapy, preventive therapy, or monitoring [5]. Prophylactic therapy consists of administering antivirals to all patients in a specific risk category. This treatment starts within 10 days posttransplant and continues for a defined period, typically 3 to 6 months. Preventive therapy entails monitoring the patient’s CMV viral load posttransplant using polymerase chain reaction (PCR); if the viral load exceeds a certain threshold, antivirals are administered to prevent disease development. Monitoring typically occurs after the prophylactic period to ensure that no late infection arises and is performed weekly for 8 to 12 weeks. Prophylactic or preventive therapy is generally reserved for patients at intermediate or high risk, employing medications such as valacyclovir (ValA), valganciclovir (ValG), or intravenous ganciclovir. While patients at low risk (D−R−) usually do not receive prophylaxis or preventive therapy, they are subject to monitoring or antiviral prophylaxis against other types of herpes infections.
Regarding prophylaxis, although ValG is most widely used [5], it has dose-limiting side effects and carries a black box warning for teratogenicity [6,7]. Alternatively, ValA represents an attractive option for some centers due to economic considerations and/or its lower impact on bone marrow suppression. The current literature indicates similar rates of CMV DNAemia between these two drugs [8]. Additionally, ValA has been associated with a lower incidence of acute rejection episodes in several studies [9–11]. This debate has practical implications, particularly in the development of guidelines [5], as few randomized controlled trials (RCTs) have directly compared the two drugs. A recent Cochrane review [12] classified the evidence regarding outcomes such as CMV disease, all-cause mortality, CMV infection, acute rejection, and graft loss between ValA and ValG or ganciclovir as low-level. In this context, we aimed to conduct a systematic review and meta-analysis to clarify the safety and efficacy of ValG and ValA for CMV prophylaxis after kidney transplantation.
METHODS
This systematic review and meta-analysis was performed and reported in accordance with the recommendations outlined in the Cochrane Collaboration Handbook for Systematic Reviews of Interventions and adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement guidelines [13,14].
Eligibility Criteria
This meta-analysis included studies comparing ValG and ValA for CMV prophylaxis in patients who underwent renal transplantation. To minimize the risk of bias from case series reports, we excluded studies with fewer than five patients and those that did not report the outcomes under investigation. Furthermore, non-English papers, reviews, letters to the editor, and conference abstracts were also excluded.
Search Strategy and Data Extraction
The Medline, Cochrane, Web of Science, Embase, and Ovid databases were comprehensively searched from inception to April 2024 using the following search strategy: (valacyclovir) AND (valganciclovir) AND (cytomegalovirus OR CMV). The references from all included studies were also examined manually to identify additional relevant studies. The data were independently extracted by two authors (LJCC and KAMM) based on predefined search criteria. This study was not prospectively registered.
Endpoints
The efficacy outcomes assessed included the incidence rates of CMV disease, CMV viremia, acute rejection, and infections caused by other herpesviruses (HVs) and the BK polyomavirus. The safety outcome of interest was leukopenia/neutropenia. In accordance with the included studies, CMV disease was defined as symptomatic CMV DNAemia, manifesting either as “CMV syndrome”—characterized by fever, malaise, or leukopenia—or as organ dysfunction in the absence of other documented causes. CMV viremia was identified as the presence of CMV DNA detected by PCR, regardless of the load. Acute rejection was determined based on the criteria outlined in each study, which may or may not have required biopsy confirmation.
We opted to analyze the data on leukopenia and neutropenia collectively, as certain studies did not distinguish between these outcomes. For studies that presented separate data for leukopenia and neutropenia, we selected leukopenia, as defined by the individual study, to represent the outcome. This approach was taken because some studies documented leukopenia without reporting on neutropenia. This choice did not influence the pooled result, as these adverse events typically present together. Furthermore, when studies reported varying grades of leukopenia, we selected the highest grade for our analysis.
Given the variations in risk across the study populations, we conducted a subanalysis limited to the RCTs, termed the “RCT-only” analysis, in which the risk proportions were comparable between groups. This analysis evaluated CMV viremia and the incidence of leukopenia/neutropenia.
Quality Assessment
Nonrandomized studies were assessed using the ROBINS-I (Risk Of Bias In Non-randomized Studies of Interventions) tool [15]. This instrument is designed to evaluate the risk of bias in observational studies and provides a detailed framework for assessing the methodological quality of nonrandomized intervention studies. The quality of RCTs was evaluated using version 2 of the Cochrane risk of bias tool for randomized trials (RoB 2) [16]. This tool categorizes studies as having a high, low, or unclear risk of bias across five domains: selection, performance, detection, attrition, and reporting bias. Two authors (LJCC and PVM) independently conducted data extraction and quality assessment of the studies, with any disagreements resolved by a third author (KAMM). Publication bias was investigated using funnel plot visual inspection and the Egger regression test.
Statistical Analysis
Risk ratios (RRs) with 95% confidence intervals (CIs) were utilized to compare treatment effects for categorical endpoints. The I2 statistic was used to assess heterogeneity, with P-values less than 0.10 and I2 values exceeding 25% considered to indicate significant heterogeneity. Leave-one-out analysis was conducted to identify outlier studies. Due to the retrospective and nonrandomized nature of the included studies, a DerSimonian and Laird random-effects model was applied. Statistical analyses were performed using R ver. 4.3.2 (R Foundation).
RESULTS
Study Selection
Our initial search yielded a total of 429 papers (PubMed, 66; Embase, 111; Cochrane, 38; Web of Science, 101; Ovid, 110; and additional records obtained through snowballing, 3). After removing duplicates (n=145), we screened titles and abstracts, excluding papers that did not meet all eligibility criteria (n=256). This process resulted in 75 articles for full-text screening. One paper was unavailable due to a lack of institutional access. Exclusion criteria for the screened papers included studies with overlapping populations (n=10), studies comparing one of the drugs of interest to another drug not relevant to our review (e.g., ganciclovir or acyclovir in most cases) or to preventive therapy (n=11), and studies involving patients who underwent bone marrow transplantation (n=1). A detailed recording of the selection stages and the process is presented in Fig. 1. Ultimately, six articles [8,17–21] were included in this systematic review and meta-analysis.
Baseline Characteristics of Included Studies
Our analysis included a total of six studies [8,17–21], comprising two RCTs and four retrospective cohorts and involving 888 patients. Table 1 presents the baseline characteristics of the included studies. Two studies [17,18] contained groups of patients who received either preemptive therapy (i.e., PCR follow-up with drug administration only in the case of viremia) or no prophylaxis. These groups were excluded from our analysis. Among the enrolled patients, 438 (49%) received ValA for CMV prophylaxis following renal transplantation, while the remaining 450 (51%) were administered ValG. Across the five studies that reported sex distribution, the male/female ratio was 1.70 in the ValA group and 1.88 in the ValG group. The study populations were similar in age, apart from one study [17] that focused on a pediatric population. In that study, patient allocation to the different prophylactic regimens was determined by CMV serostatus: high-risk patients (D+/R−) received ValG, while those at intermediate risk (R+) received ValA. CMV serostatus was also employed to determine patient allocation in another study [19], in which all but four patients at low risk (D−/R−) received ValA while all those at intermediate risk (R+) received ValG. In comparison, in a nonrandomized study [18], allocation was determined by the period of treatment: patients treated from 1996 to 2000 did not receive anti-CMV prophylaxis, individuals treated from 2000 to 2004 received ValA, and those treated from 2004 to 2009 received ValG. The risk of bias assessment appropriately addressed such issues with intervention allocation observed in nonrandomized studies.
Regarding CMV serostatus, except for the previously mentioned cases, the distribution of high-risk patients was similar across the groups. All studies reported on the immunosuppressive regimens used, with the specific drugs listed in Table 1. Antithymocyte globulin was included in the induction regimens of four studies [8,18,19,21]. The proportion of patients receiving antithymocyte globulin was comparable between groups in the randomized studies, though minor differences were noted in the nonrandomized studies. In five of the six total studies, medication doses were adjusted based on renal function. The remaining study [20] prescribed a low dose of ValA at 2 g/day and a fixed dose of ValG at 900 mg/day. The duration of prophylaxis was set at 3 months in three studies and at 6 months in another. In one study [21], the ValA group received 3 months of prophylaxis, while the ValG group had a variable duration ranging from 3 to 6 months. The duration of prophylaxis in the remaining study was adjusted according to serostatus and age, varying from 3 months to 1 year.
Outcomes
The included studies provided sufficient data for a pooled analysis of several outcomes: the incidences of CMV disease and CMV viremia; acute rejection; leukopenia; and infections caused by other HVs and the BK polyomavirus. The analysis of CMV disease incidence revealed no significant difference between the ValA and ValG groups (RR, 0.74; 95% CI, 0.09–5.97; P=0.8; I2=71%) (Fig. 2A). Similarly, the incidence of CMV viremia was comparable between groups (RR, 0.70; 95% CI, 0.31–1.57; P=0.4; I2=68%) (Fig. 2B). Given the considerable heterogeneity observed in these results, additional sensitivity analyses were conducted for each included study. In the analysis of disease incidence, excluding the study by Leone et al. [18] resulted in the lowest overall heterogeneity (RR, 0.25; 95% CI, 0.04–1.44; I2=0%) (Supplementary Fig. 1). No changes in the classification of heterogeneity were noted upon the exclusion of any study in the CMV viremia analysis (Supplementary Fig. 2).
Regarding acute rejection, no significant differences were observed between the groups (RR, 0.97; 95% CI, 0.50–1.91; P=0.8; I2=49%) (Fig. 3A). Sensitivity analysis employing the leave-one-out method identified Reischig et al. [8] as an outlier that contributed to the heterogeneity (Supplementary Fig. 3). In the analysis of leukopenia/neutropenia, a significantly lower incidence of this adverse event was noted in the ValA group (RR, 0.57; 95% CI, 0.42–0.77; P<0.01; I2=0%) (Fig. 3B).
The exploratory endpoints were associated with infections by other viruses in both groups. No significant difference was noted in the incidence of other HV infections (RR, 1.43; 95% CI, 0.61–3.38; P=0.4; I2=64%) (Fig. 4A) or of BK viremia (RR, 0.67; 95% CI, 0.24–1.87; P=0.4; I2=43%) (Fig. 4B). However, leave-one-out analyses for these endpoints did reveal significant findings. When the study by Reischig et al. [8] was excluded from the analysis of other HV infections, the results favored the ValG group with minimal heterogeneity (RR, 2.29; 95% CI, 1.24–4.25; I2=0%) (Supplementary Fig. 4). Conversely, excluding the study by Velioğlu et al. [20] from the analysis of BK viremia produced a finding favoring the ValA group, again with minimal heterogeneity (RR, 0.49; 95% CI, 0.27–0.90; I2=0%) (Supplementary Fig. 5).
Subanalysis
In a subanalysis of RCT-only data, involving a total of 256 patients, no significant difference was observed between the groups in CMV disease incidence (RR, 0.99; 95% CI, 0.41–2.37; P=0.98; I2=53%) (Supplementary Fig. 6). Regarding the incidence of leukopenia/neutropenia, the subanalysis of RCT-only data corroborated the findings of the overall analysis. Specifically, it demonstrated a lower incidence of leukopenia/neutropenia in the group treated with ValA (RR, 0.65; 95% CI, 0.46–0.92; P=0.01; I2=0%) (Supplementary Fig. 7).
Risk of Bias Assessment
Fig. 5 presents the risk of bias assessment results for the six studies included in this review, which compared ValA and ValG for CMV prophylaxis following kidney transplantation. All RCTs were deemed to exhibit a low risk of bias. In contrast, all nonrandomized studies were categorized as having a serious risk of bias, predominantly due to confounding factors, which are common in such studies. Velioğlu et al. [20] were assessed as having a serious risk of bias in the domain of missing data because the study failed to report some critical information for the ValG group, such as CMV viremia. Stamps et al. [19] were determined to have a serious risk of bias in the domain of deviations from intended interventions, due to unbalanced cointerventions across the study groups.
Regarding publication bias, visual inspection of the funnel plots (Supplementary Fig. 8) for the analyzed outcomes revealed no asymmetry across all outcomes, except CMV disease (Supplementary Fig. 8A). For that outcome, we observed a slight asymmetry, indicative of a small study effect, and the Egger test corroborated the possibility of publication bias (P<0.05). Notably, however, the accuracy of funnel plot analysis is limited when fewer than 10 studies are included [13].
DISCUSSION
This systematic review and meta-analysis compared the use of ValA and ValG as prophylactic therapies for CMV in renal transplant recipients, drawing on data from six studies involving a total of 888 patients. Our results revealed no statistically significant differences between the two drugs in terms of CMV disease incidence, CMV viremia, acute rejection rates, and the incidences of BK viremia and other HV infections. However, the ValA group exhibited a significantly lower incidence of leukopenia/neutropenia. This systematic review and meta-analysis is valuable in that it addresses the current gap in robust evidence concerning the efficacy of various CMV prophylaxis protocols [5,12].
The comparison between ValA and ValG showed comparable efficacy in preventing CMV disease ValA versus ValG demonstrated similar efficacy in preventing CMV disease. The incidence of CMV viremia and the development of CMV disease were comparable between the two groups, as confirmed by a subgroup analysis of only RCTs. However, Leone et al. [18] reported diverging findings. The investigators reported a lower incidence of CMV during each treatment period in the group receiving ValG prophylaxis. This reduction, coupled with the delayed onset of CMV disease, may be attributed to the unique approach of the study, which involved an extended 6-month duration of ValG prophylaxis. The use of 6-month ValG prophylaxis in renal transplant recipients has been linked to a decrease in CMV disease when compared to 3-month prophylaxis, as shown in the IMPACT trial [22]. A similar observation was made in lung transplant recipients [23]. Additionally, the longer follow-up period of this study (48 months) in comparison to others (6–21 months) may have contributed to these results. Nevertheless, the other studies included in the analysis individually support the overall conclusion. Reischig et al. [8] stated that both protocols effectively reduced the incidence of CMV. Velioğlu et al. [20] and Verghese et al. [21] found that no CMV disease was observed with either medication in their samples during the follow-up period. Supporting the potential benefits of ValA, another meta-analysis [12] indicated that this drug reduces the risk of CMV disease and all-cause mortality, which includes mortality from CMV disease. This finding was obtained despite the limited evidence available on ValA at the time, which restricted its direct comparison with ValG.
Our meta-analysis revealed that acute rejection rates were not significantly different between the two prophylaxis regimens. This may be attributed to the similar rates of CMV disease, which is associated with rejection [5]. Another meta-analysis [24], which primarily assessed rejection rates among various antiviral agents, determined that ValG, ValA, and ganciclovir significantly reduced the incidence of biopsy-proven acute rejection compared to placebo. ValG was noted to have the greatest risk reduction, with a significant difference compared to other groups, including ValA. However, that meta-analysis included only one study directly comparing ValG and ValA (Reischig et al. [8]), which was also part of our analysis. This limitation in the data may explain the observed results.
Leukopenia/neutropenia was more frequently reported in the ValG group than among the ValA recipients, a finding exhibiting statistical significance and confirmed by subgroup analysis. Verghese et al. [21] found that leukopenia prompted a reduction or cessation of ValG in all but one participant. In contrast, Reischig et al. [8] reported a nonsignificant increase in the incidence of leukopenia, with no difference in severity. The inclusion of pediatric patients in the study by Verghese et al. [21] may have influenced these results, given that a recent review suggests neutropenia is the most common adverse event in pediatric patients treated with ValG [25]. Supporting this, a meta-analysis by Kalil et al. [26] showed that ValG was significantly associated with the risk of absolute neutropenia (<1,500/mm3) when compared with all therapies (ganciclovir and non-ganciclovir therapies); however, the study did not directly compare ValG and ValA.
Despite these effects, ValG remains the primary choice for CMV prevention, as supported by the International Consensus Guidelines on the Management of Cytomegalovirus in Solid-Organ Transplantation [5]. According to this guideline, it is strongly recommended—albeit with a low level of evidence—that in cases of ValG-induced myelotoxicity, a granulocyte colony-stimulating factor should be added and/or other myelosuppressive therapies should be discontinued before switching to an alternative antiviral therapy. Another approach to managing this toxicity is the use of “mini-dose” ValG prophylaxis (450 mg orally), which has been trialed by some centers and evaluated through a meta-analysis [27]. However, this strategy is associated with the emergence of drug-resistant CMV, particularly in patients with a CMV D+/R− serostatus [28].
Regarding other opportunistic infections, Leone et al. [18] demonstrated a higher incidence of polyoma BK virus infection in the ValG prophylaxis group, a finding corroborated by Reischig et al. [8], who observed a significantly higher incidence of polyoma BK viremia in the same group and linked it to the clinically relevant immunosuppressive potential of ValG. However, our pooled analysis indicated no significant difference in BK viremia between the groups. The leave-one-out analysis revealed that this result was influenced by the inclusion of the study by Velioğlu et al. [20] That study, conducted in two different centers, was judged to exhibit serious risk in the domain of bias due to missing data, as it reported CMV viremia only for the ValA group. We believe that its dual-center design could have led to the implementation of different cointerventions across the groups, resulting in disparate results compared to other studies on the incidence of BK viremia.
Our results are also potentially applicable to a financial context. Kacer et al. [29] conducted a study comparing the economic impact of ValA with ValG for CMV prevention. They found that for patients undergoing kidney transplantation, prophylaxis with ValA is a more cost-effective strategy than that with ValG. Specifically, ValA was associated with a significantly lower cost—44% less in the first year following kidney transplantation—and could be considered a preferred alternative in situations involving limited resources.
These findings must be interpreted considering several limitations. First, only six studies were included in this meta-analysis; four were retrospective with a serious risk of bias, while only two were low-risk RCTs. To mitigate this concern, a subanalysis was performed using data exclusively from the randomized studies, which supported the overall results. Nonetheless, the data from the two RCTs in this subanalysis are limited by their small sample sizes. The performance of additional RCTs would contribute to more robust evidence. A subgroup analysis for different risk levels was not feasible, as the studies did not report separate data for these groups regarding most of the evaluated outcomes. Another limitation was the heterogeneity of the results, which complicated extrapolation of the findings. This heterogeneity could be attributed to variations in dosing, differences in therapy allocation, age (with one study exclusively involving a pediatric population), and an unbalanced distribution of risk across the studies. A leave-one-out analysis for outcomes with high heterogeneity was conducted to identify outlier studies, thus facilitating the interpretation of the results. Finally, most studies did not report bacterial or fungal infections, precluding an assessment of the direct consequences of leukopenia.
In conclusion, the two drugs were equally effective in preventing CMV disease. However, ValG was associated with a significantly higher risk of leukopenia and exhibited more pronounced immunosuppressive effects compared to ValA. These findings suggest that ValA may be a suitable alternative to ValG in clinical settings where reducing hematological side effects is a priority or when a more cost-effective option is required.
XML Download