Economic burden of preemptive treatment of CMV infection after allogeneic stem cell transplantation: a retrospective study of 208 consecutive patients

Cytomegalovirus (CMV) infection and diseases are global concerns for the transplant community after allogeneic hematopoietic stem cell transplantation (HSCT). CMV-seropositive recipients of HSCT have a higher rate of mortality than CMV-seronegative patients, especially in transplants from unrelated or mismatched donors, and even more if the donor is CMV-seronegative [1, 2]. In agreement with this finding, the occurrence of CMV reactivation before day 100 is associated with higher non-relapse mortality [3]. There are many mechanisms by which the CMV may affect survival, including CMV infection and disease, but also indirect effects relating to graft-versus-host disease (GVHD), side-effects of antiviral drugs, and concomitant non-viral infections [2]. Preemptive treatment of CMV infection is currently the most common strategy to reduce the risk of CMV disease and the survey of Pollack et al. showed that the CMV practices are not different between North America and Europe [4]. This preference of many programs for preemptive over prophylactic treatment is mainly because of the side effects of the most common antiviral drugs (ganciclovir and foscavir) available to treat CMV. Indeed, these drugs are not safe enough to be routinely given for several weeks [1]. In three randomized trials to evaluate early prophylactic ganciclovir use after allogeneic HSCT, the benefit of preventing CMV infection, and possibly CMV disease, was offset by a greater risk of bacterial or fungal infections due to neutropenia [5‐7]. Long-term administration of foscavir also raises safety issues, exposing patients to unacceptable renal toxicity [8].

New anti-CMV drugs such as maribavir [9], brincidofovir [10], letermovir [11], and CMV vaccines [12, 13] are currently being, or have been recently evaluated in phase 2 or 3 studies. On many aspects, the new antiviral drugs have safer profiles than available anti-CMV drugs, and the results of the ongoing trials may encourage CMV prophylaxis after HSCT [1], especially with letermovir which has been shown, in a large placebo-controlled phase 3 study, to decrease CMV infection in CMV seropositive patients [14]. However, considering a minimal 3-month duration of the at-risk period for CMV reactivation, these prophylactic approaches will be probably more expensive than the actual preemptive approach whose duration of treatment is usually limited to 2–4 weeks.

In this study, we aimed to evaluate the overcost related to CMV infection and disease over the first year of transplant, using a preemptive strategy, and to know whether the transplant cost can be predicted from the CMV serological status of the donor and recipient.

The study included 208 consecutive adult patients; of these, six were censored at the time of second transplant. The median follow-up was 18 months. The characteristics of the patients are summarized in Table 1. Most patients received transplants because of acute leukemia. The graft was from an HLA-identical sibling (40%), an unrelated donor (51%) or a cord-blood unit (9%). The CMV donor/recipient serologies are shown in Table 2. One hundred and thirty recipients (62%) were CMV seropositive.

The patients were managed for CMV risk according to local written procedures, established according to the EBMT [20], and international [21] guidelines. Briefly, patient’s blood, except for CMV seronegative patients transplanted with a seronegative donor, was screened weekly for pp65 Ag (CMV Brite Turbo kit, IQ Products) until September 2009, or plasma CMV DNA (Artus® CMV RG PCR Kit, Qiagen) from October 2009, until day 100 or for longer in cases of persistent immunosuppression. No patient received CMV prophylaxis or prophylactic intravenous immunoglobulins. After discharge following the transplant, all patients were seen in our day-care center at least once a week until day 100, then every two weeks from day 100 to day 180, then monthly, or more often if required. All infected patients received ganciclovir (5 mg/kg × 2/d) or foscavir (60 mg/kg × 2/d) for 14 days, either in the ward or at home, according to their clinical status, blood counts, renal function and the concomitant therapies. Patients who had a viral load below these cut-offs did not receive anti-CMV antiviral drugs. Valganciclovir was not routinely used as first^t line treatment of CMV infection. Maintenance treatment of 14 additional days was given at the discretion of the physician but was recommended for cases of severe GVHD and/or transplant from an unrelated donor. Patients with CMV disease were treated with the same doses of ganciclovir as those used to treat CMV infection, but with higher doses of foscavir (90 mg/kg ×2 /d). Patients with CMV pneumonia additionally received intravenous immunoglobulins.

We show that having two or more CMV episodes after allogeneic HSCT increased the transplant cost of 25–30% whereas having only one episode did not. Additionally, our estimate of the average cumulative 12-month costs of allogeneic HSCT for more than 200 consecutive patients is the largest, single-institution cost study in Europe. The average hospital cost per transplant patient, as paid by the third party payer, was €104,000 and corresponded to a total of 96 days in hospital, and higher costs were significantly driven by the number of CMV episodes.

Despite the use of various costing methods, previous, smaller, cost studies of stem cell transplantation have generated consistent results showing an average 12 month cost of about €100,000, in the same range in the United States [22‐25] and in Europe [26]. Eighty percent of this cost is incurred during the first 6 months [22‐26]. Few cost predictors have been identified, i.e. pretransplant patient characteristics, preferably those that can be altered by medical intervention. Saito et al. [24] explored cost predictors and found only GHVD prevention among the actionable factors. Lee et al. [23] found no relationship between pre-transplant information and total costs except for mismatched donors. Khera et al. [22] reported graft source, donor type and HLA match as pre-transplant predictors, none of which are easily actionable. In our study, we did not find any pre-transplant predictors of cost, even though we specifically tested CMV D/R serology.

Most multivariate cost models, however, found that GVHD and infections were important cost drivers [22‐24, 26, 27]. Consistently, in a previous French study [27], we showed that the main cost drivers were acute grade II-IV GVHD and infections which are usual causes of prolongation of hospitalization or re-hospitalization. However, we did not look specifically at CMV episodes. In fact, few studies have specifically focused on CMV as a cost driver. Pre-transplant CMV serology of the donor or of the recipient has never been found to predict costs significantly, although a trend for lower costs in the seronegative donor/recipient groups has sometimes been found [22, 24]. CMV infection was identified as a cost driver in the 90’s, at a time when no anti-CMV drugs were available and the CMV-related mortality was very high [28]. In a recent US study, patients who received preemptive treatment for CMV infection incurred an additional cost of US $58,000–74,000 [29]. In our study, the occurrence of CMV episodes was significantly associated with higher length of stay and higher costs. Due to the strong association and bidirectional effects between GVHD and CMV [2, 30], both factors increasing the risk of other infections, it is possible that the over cost associated with CMV episodes only reflects the selection of the most severe transplant population. Indeed, it will be a challenge for future anti-CMV prophylactic approaches to be cost-effective by reducing both the direct and indirect effects of CMV, possibly including GVHD.

Until now, most stem cell transplant centers have been using a preemptive strategy for CMV infection [4, 31] with a significant risk of recurrence when CMV-specific immunity is not recovered when antivirals are stopped. However, new anti-CMV compounds have recently or are currently being studied for prophylaxis [9‐11]. It may be expected that some of these compounds could prevent the occurrence of CMV episodes, and eventually the indirect effects of CMV. As most of these drugs have safe profiles and are consistent with outpatient management, the transplant community could prefer a prophylactic approach for the first 3 months of transplant to reduce hospitalization. By showing that the occurrence of two or more CMV episodes increases the cost of a transplant by 25–30% (€25–30,000), our study gives a reasonable assessment of the costs which could be saved if any efficient CMV prophylaxis would dramatically decrease CMV episodes. Roughly half (n = 70) of our 130 seropositive patients developed a CMV episode. If new CMV prophylactic treatments are not associated with better survival than current strategies, they would be beneficial if their costs were not higher than half of this over-cost. If such CMV prophylaxis improves overall survival, it should be cost-effective.

Our study has limitations: (1) it is monocentric. However, our center is representative of French practice and all our patients were managed according to common procedures, closely followed in one center, such that we did not miss any significant complication. This design with a systematic check of CMV episodes in the original medical charts is more precise than large studies based on CMV coding in healthcare databases which may have underestimated the CMV episodes when not coded [32] (2) We did not take into consideration non-viral infections that are often associated with GVHD and CMV episodes. However, most fungal or bacterial infections in such patients require hospitalization and are, thus, indirectly captured in the length of stay. (3) During the study period and until now, we did not use valganciclovir, the oral pro-drug of ganciclovir, for first line pre-emptive therapy of CMV infection. Although associated with hematological toxicity and especially neutropenia which is a limiting factor to its prolonged administration [33], the use of valganciclovir may facilitate outpatient management and should reduce costs, but is not recommended as an “AI” first line treatment of CMV infection after stem cell transplant, especially in high risk patients [20]. Despite these limitations, our study is the largest cost study so far run in Europe, and while most cost analyses are commissioned by drug companies [34], our study is academic.

In summary, the costs of allogeneic HSCT increase by 25–30% when there are CMV episodes. The cost difference between patients who develop CMV episodes and those who do not could be reduced by appropriate use of prophylactic strategies.