This was a retrospective longitudinal study based on the maribavir CUP database. This database was compiled using prospective data collection (from 13 October 2021 to 31 August 2023), in accordance with the therapeutic use protocol (Protocole d’Utilisation Thérapeutique - Recueil des données; PUT-RD) validated by the French Health Authority (ANSM). All French centers had access to the CUP. Requests were submitted by physicians from transplant units and evaluated on a case-by-case basis by ANSM according to the PUT-RD. After inclusion in the CUP, no invasive or specific examinations were performed. All patients were managed by their transplant physicians in accordance with local procedures.

According to French legislation, this study protocol, based on the retrospective analysis of CUP data, was declared to the French authorities under reference methodology MR-004. Patients provided informed consent regarding the use of their data as part of the CUP, and for the reuse of these data for the purposes of this research.

Processing of personal data was performed in accordance with Regulation (EU) No 2016/679 on the General Data Protection Regulation (GDPR) of natural persons.

The patients included in this study were participating in the CUP and completed both treatment initiation and end-of-treatment forms. According to the PUT-RD, the main eligibility criteria to be included in the CUP were: age >18 years, history of a HCT or SOT, ongoing CMV refractory infection (i.e., failed to reach 1 log10 reduction in blood or plasma viral load after 14 days of standard antiviral treatment) with or without antiviral resistance, and estimated glomerular filtration rate (eGFR) >15 mL/min/1.73 m² (calculated using the Modification of Diet in Renal Disease Equation). The definition of refractory infection used in the present study is consistent with recent recommendations [5, 6, 15]. The complete list of inclusion criteria in CUP program is provided in the Supplementary Material.

Maribavir was administered orally at the recommended dose of 400 mg (two 200-mg tablets) twice daily (i.e., total daily dose of 800 mg). Although the recommended duration according to the maribavir label is 8 weeks, the treatment duration could be adjusted by the physician based on the clinical characteristics of each patient.

Data on baseline variables were collected at the time of CUP request or at treatment initiation, then on follow-up visits: at week 4 (W4), W8, W12, and W20. Data on the use of maribavir were collected at each follow-up visit. Given the real-life setting of the study, there were no mandatory visits or interventions. In the event of inconsistencies regarding clinical values, quality controls were conducted by contacting the centers.

Baseline patient characteristics were collected on demographics (age, sex), biological parameters (hemoglobin, leukocyte, and platelet counts, eGFR), transplantation parameters (type and timing), and CMV-related variables (viral load, CMV drug resistance mutations, presence of CMV disease), when available. The collection of all CMV-related variables was not mandatory in routine clinical care.

Viral load levels were measured in local laboratories using a highly sensitive polymerase chain reaction (PCR) with a limit of detection of 200–500 IU/mL to quantify CMV deoxyribonucleic acid (DNA) in blood or plasma, i.e., DNAemia. DNAemia categories were defined as follows:

High: ≥91,000 IU/mL (4.95 log10) for plasma, or ≥273,000 IU/mL (5.4 log10) for whole blood;

These thresholds were those used in the pivotal SOLSTICE study [14].

No standardized definition of CMV disease was provided in the CUP protocol (PUT-RD); physicians were expected to rely on international recommendations and clinical judgment [16].

Co-prescriptions were recorded at treatment initiation and during follow-up. Treatment duration was measured from therapy start to discontinuation.

For effectiveness assessment, patients were considered treatment responders if they met either of the following criteria: they achieved viremia clearance, defined as a viral load concentration below the lower limit of quantification of <137 IU/mL of plasma, or <411 IU/mL of whole blood); or 2) the viral load concentration was reported as below a threshold determined by the local laboratory or “not detectable” or “not quantifiable”. The proportion of responders was estimated both across the full maribavir treatment period (i.e., between treatment initiation and treatment termination), regardless of treatment duration; and by treatment duration according to three categories: <8 weeks, 8 weeks ±6 days and ≥8 weeks.

Patients were followed for viral load up to 20 weeks after maribavir initiation, which allowed the identification of CMV recurrence after an initial response, defined as a PCR value above reference values for plasma or whole blood. Known CMV drug resistance mutations [17] were identified by Sanger sequencing genotyping before, during, and after maribavir treatment, at the physician’s request.

Descriptive statistics on patient characteristics, maribavir treatment use, and effectiveness were conducted within the overall study population. Kaplan Meier curves were provided for descriptive purposes. The median time to first clearance and time to treatment discontinuation were calculated, along with their 95% confidence intervals (CI).

Regression models were performed to identify potential predictive factors of treatment effectiveness. The construction of this model included the following steps: 1) selection of clinically relevant variables, 2) univariate analyses, 3) initial multivariate model and 4) final multivariate model. First, clinically relevant variables at treatment request or initiation were considered in univariate analyses [i.e., presence of CMV disease, CMV drug resistance mutations, biological measures (categorized according to population distribution, > or ≤ median), and viral load levels (categorized according to low, intermediate, and high levels)]. Then, univariate analyses were conducted using Chi2 tests or Fisher’s exact test, with variables retained being those with p < 0.05. Statistically significant factors identified in the univariate analyses were then modeled using a Cox regression model to estimate the hazard ratio (HR) adjusted for significantly associated factors (after having verified that they were not highly correlated, i.e., R < 0.7; in case of highly correlated variables, the most significant variables in the univariate analysis were retained in the model). At the end, a final model was obtained considering model performance based on the Akaike information criterion, given the limited sample size.

Between October 2021 and April 2023 (18 months), 79 patients initiated treatment with maribavir across 61 transplant departments and completed both initiation and end-of-treatment forms. The completeness rate of collected data exceeded 90% (CUP report ANSM, 2024). At baseline, patients had a mean age of 56.2 years and were predominantly male (69.6%). The majority had received SOT (67 patients; 84.8%), mainly kidney (52; 77.6%), followed by heart (7; 10.4%), lung (7; 10.4%), and pancreas (1; 1.5%). HCT recipients (12 patients) accounted for 15.2% of the study population. Additionally, 18.9% (15 patients) had cytopenia (neutrophil <1,000/mm³, platelet <25,000/mm³ or hemoglobin <8 g/dL), 16.5% (13 patients) had severe renal impairment (eGFR ≤30 mL/min/1.73 m²), and 48.1% (38 patients) had viral loads (DNAemia) in the intermediate to high range (Table 1). Of the 79 infected patients, 57 (72.2%) were asymptomatic, and 22 (27.9%) had CMV disease. Co-prescription of maribavir with other anti-CMV agents was observed in 12 patients, including anti-CMV immunoglobulins Cytotect^® (9 patients), foscarnet (2 patients), and cidofovir (1 patient).

Among the 79 patients with refractory CMV, 57 (72.2%) had at least one known CMV drug resistance mutation at baseline (resistant CMV), whereas 22 (27.9%) had no drug resistance mutations detected. One patient presented a mutation conferring resistance to maribavir, while the remaining mutations were associated with resistance to standard antiviral agents. The majority of detected CMV drug resistance mutations were in the UL97 gene, with additional mutations in the UL54 gene, conferring multidrug resistance (Table 1).

The mean treatment duration of maribavir was 65.2 ± 4.4 days (median 59 days, [95% CI: 57–61] days). Most patients had a complete 8-week treatment (40.5%, 32 patients) or a treatment prolongation ≥8 weeks (38.0%, 30 patients), while 21.5% of patients (17 patients) received treatment for less than 8 weeks.

Overall, 53.2% of patients (42/79) responded to maribavir at any time during treatment. The median time to first CMV clearance was 59 days [95% CI: 39–69] days (Figure 1A). Moreover, as shown in Figure 1B and Supplementary Table S1, the median time to first CMV clearance according to treatment duration (<8 weeks, 8 weeks ± 6 days, or >8 weeks) was 39, 56, and 63 days, respectively. These data indicate that patients may continue to achieve initial CMV clearance beyond the 8-week mark. Notably, 66.7% (20/30) of patients treated for >8 weeks experienced their first clearance after the 8th week.

Regarding potential predictive factors of treatment effectiveness, clinically relevant variables that could be related to viremia clearance under treatment were first considered in univariate analyses. These analyses showed that only three variables had p-values <0.05: the presence of CMV disease, leucocyte levels, and viral load levels at baseline (Table 2). These variables were then considered in multivariate models. In the final model, the presence of CMV disease at baseline was statistically significantly associated with viremia clearance (Table 3), indicating that patients with CMV disease at baseline were less likely to achieve CMV viremia clearance under maribavir treatment (adjusted HR = 0.44 [95% CI: 0.20–0.96]). However, baseline leucocyte levels (>4,180 mm³ vs. ≤4,180 mm³) and viral load levels (low vs. intermediate/high) were not statistically associated with response to maribavir: adjusted HR = 1.76 [95% CI: 0.92–3.37] and adjusted HR = 1.56 [95% CI: 0.82–2.97], respectively.

Post-treatment, emergent maribavir resistance was detected in 13.9% (11/79) of patients. All CMV maribavir resistance mutations were located in the UL97 gene and included: T409M (5 patients), F342Y (3 patients), C480F (2 patients), C480R (1 patient), H411L (1 patient), and H411Y (1 patient). Of note, a single patient could present with multiple mutations. The median time to first maribavir resistance mutation occurrence was 45 days. Maribavir resistance mutations occurred during treatment in 8 patients and post-treatment in 3 patients. Among the 42 patients who achieved a treatment response, 19 (45.2%) experienced CMV recurrence within the 20-week follow-up period. The median time to recurrence was 91 days.