Pre-engraftment clinically signicant CMV infection after allogeneic hematopoietic stem cell transplantation and its impact on engraftment

Pre-engraftment cytomegalovirus infection (CMVi) is a challenge in patients receiving allogeneic stem cell transplantation (Allo-HCT), as few data have been reported on its clinical importance. This study describes the clinical outcomes of pre-engraftment CMVi and compares them with those of episodes developing after engraftment in HCT patients. We performed a retrospective study of patients who underwent Allo-HCT from 2016 to 2020, including 111 recipients monitored by real-time PCR assay. Clinically signicant CMVi (csCMVi) was documented in 81 (73%) patients. There were 29 (26%) cases of pre-engraftment csCMVi. No signicant difference was observed regarding virological features, but patients with pre-engraftment csCMVi had a delayed start in treatment (24 vs. 12 days, p < 0.001) compared with those with postengraftment events. Pre-engraftment csCMVi was associated with a delay in engraftment (20 vs. 16 days, p = 0.02) and worse overall survival (54% vs. 73% 1-year OS; p = 0.020) than postengraftment events. In conclusion, pre-engraftment csCMVi occurred in 26% of our patients and was associated with engraftment delay and worse overall survival. Close monitoring of CMV DNAemia is necessary to identify these patients earlier. Prospective studies, including patients with letermovir prophylaxis, are necessary to dene the standard in the management and care of this population.


Introduction
Cytomegalovirus infection (CMVi) leads to high morbidity in patients undergoing allogeneic stem cell transplantation (Allo-HCT) and impacts clinical outcomes, including overall survival. (1,2,3) CMVi occurs in 70 to 80% of Allo-HCT patients after HCT. (4) Most CMV reactivations occur in the rst 100 days after HSCT at a median of 50 days post HCT, but some patients present very early CMV reactivation, sometimes before engraftment. (1,5) Approximately 25% of documented CMV reactivations have spontaneous clearance (i.e., without the need for therapy). (6) Nevertheless, in others, viral replication gives rise to clinically signi cant CMV infection (csCMVi), and a pre-emptive approach with antiviral therapy is needed to prevent CMV organ disease (7)(8)(9)(10) csCMVi documented pre-engraftment is challenging to treat. Anti-CMV drugs (usually intravenous ganciclovir or oral valganciclovir) have signi cant adverse effects, including myelotoxicity. (11)(12)(13) Foscarnet has severe side effects, especially nephrotoxicity, and in some countries, its access is di cult. (8) In this study, we describe the frequency pf and risk factors for pre-engraftment csCMVi in HCT patients and evaluate its impact on the time of engraftment and overall survival.

Patients And Methods
Consecutive adult patients who underwent Allo-HCT at Complexo Hospitalar de Niteroi between 2016 and 2020 were included in the observational retrospective cohort. CMV-seronegative recipients allografted with CMV-seronegative donors, those receiving anti-CMV prophylaxis, or patients who died before D + 30 after transplant were excluded. CMV viral loads (time after HCT to reactivation, initial viral load, highest viral load within the event, duration of viremia, and viremia clearance) were obtained for the rst episode of following HCT for each patient. For this study, outcomes were analyzed until D + 100. The local Institutional Review Board approved the study (CEP Hospital 9 de Juho, CAAE number 54941216.0.3001.5455), and the study was conducted according to guidelines of the Declaration of Helsinki.
Patients underwent weekly surveillance for CMV viremia by a CMV real-time polymerase chain reaction (PCR) kit (produced by Qiagen -CMV ampli cation Reagent kit -Abbott) in plasma. (14) This commercial test provides a limit of detection of 31 IU/mL(95% CI). In this study, DNAemia was reported in quantitative international units (IU) per milliliter. The screening started in the rst week after SCT and was repeated weekly until D + 100 and after D + 100 if immunosuppression was maintained.
Preemptive therapy was initiated at the treating physician's discretion, and it was based upon positive surveillance tests and the patient's risk group. Suggested thresholds were CMV viremia of more than 150 IU/mL for high-risk (cord-blood, haploidentical, HLA-mismatched or T-cell depleted Allo-HCT or receiving > 1mg/kg corticosteroids) patients and more than 500 IU/mL for low-risk patients (all HCT that do not meet high risk criteria). (10) (Val)gancyclovir is standard therapy unless the episode occurs before engraftment or during neutropenia. Foscarnet is the drug of choice if ganciclovir is contraindicated. (8) All the patients received antiviral prophylaxis with acyclovir, and none of them received anti-CMV prophylaxis. Quinolone prophylaxis is not routinely used. Antifungal prophylaxis is recommended in highrisk patients (acute leukemia, previous invasive fungal infection, or long estimated neutropenia).
Cefepime is the frontline empiric treatment for febrile neutropenia unless there is a multiresistant pathogen history (colonization or previous infection) or a clinically complicated presentation. In these situations, empirical therapy is adapted by risk and de-escalated when possible and at the discretion of the treating physician.
Data collection CMV episodes were described considering the time after SCT to reactivation, initial viral load, highest viral load within the event, duration of viremia, treatment, and viremia clearance after the start of anti-CMV therapy.
We collected HCT characteristics (baseline disease and its status, preconditioning absolute lymphocyte count), donor type and stem cell source, conditioning, and graft versus host disease (GVHD) prophylaxis.
In this analysis, patients with csCMVi detected pre-engraftment were compared with those with csCMVi detected post engraftment. Engraftment was de ned as absolute neutrophil count > 500 cels/mm 3 on three consecutive days.
For outcome analysis, we collected data regarding CMV episode outcomes (CMV clearance) and HCT outcomes: time to engraftment and overall mortality at D100 and 1 year post HCT.
De nitions CMV episode was de ned as detecting CMV DNA at any level in one or more plasma samples. Clinically signi cant CMV infection (csCMVi) was de ned as CMV disease or CMV viremia leading to preemptive treatment. (9) CMV clearance was de ned as the documentation of 2 negative PCR values. The overall duration of CMV viremia was between the rst detection and the rst negative (indetectable) PCR result. Viremia clearance was de ned as the interval from the date of the initiation of anti-CMV therapy to clearance. (6) aGVHD was diagnosed and graded as reported by the modi ed Glucksberd criteria. (15) All-cause mortality was calculated at D + 100 and 1 year post HCT.

Statistical analyses
Categorical and continuous numeric variables were expressed as frequencies and medians and explored using the chi-square or Fisher exact test, Wilcoxon rank-sum test, or Spearman correlation test. Time to event was estimated by Kaplan-Meier analysis and analyzed by the log-rank test. Risk factors for preengraftment csCMVi were assessed by logistic regression analysis. Cox proportional hazard regression was used to assess risk factors for time to engraftment and overall survival. A p-value lower than 5% was considered signi cant. All analyses were performed with the IBM SPSS statistical package (SPSS Statistics for MAC, Version 27, by IBM Corporation).
Therapy for pre-engraftment csCMVi was mainly started after engraftment (79%). The drug was initiated before engraftment in only three patients, all of whom received foscarnet. In the other 3, ganciclovir was started concomitantly with engraftment. The treatment of pre-engraftment csCMVi was started a median of 24 days after CMV DNAemia was rst documented. In patients with postengraftment csCMVi, the median time to start therapy was 12 days (p < 0.001). The CMV viral load at the start of therapy was similar in pre-and postengraftment events (2184 vs. 1742 IU/ml; p = 0.637).
The patients' clinical characteristics and virological features of csCMVi are detailed in Table 2. There were signi cant differences in the source of stem cells, CD34 cellularity of donor progenitors, time to engraftment, and overall mortality between pre-engraftment and postengraftment csCMVi patients.
Regarding virological features, no signi cant difference was observed. Pre-engraftment csCMVi was associated with prolonged time to engraftment (median time to engraftment was 20 vs. 16 days; p = 0.02, by log-rank) and worse overall survival (OS) than postengraftment csCMVi (Figs. 1 and 2). In a Cox regression model, including stem cell source and CD34 cellularity, pre-engraftment csCMVi was still a signi cant risk factor for later engraftment (Table 3).

Discussion
In this study, we addressed the frequency and clinical impact of pre-engraftment CMVi. We noted that 26% of our patients had csCMVi before engraftment, and this event had a negative effect on time to engraftment and overall survival. In this study, we analyzed only csCMVi. In the study of CMV disease or CMVi that requires preemptive therapy, csCMVi is the preferred primary endpoint of studies reporting CMV post HCT. (9) When quantitative PCR was not feasible, the documentation of CMV reactivation during neutropenia was a challenge. Episodes were mostly determined through tissue biopsy. (16, 17) Currently, the monitoring of CMVi by quantitative PCR is routine, including during the pre-engraftment HCT phase and early CMVi, as quantitative PCR can be easily performed. (8, 16) Therefore, many questions about pre-engraftment CMVi need answers, including the frequency of pre-engraftment CMVi, its clinical importance, and how to treat this condition.
Few studies have previously addressed pre-engraftment CMVi, and none have reported csCMVi. (18,19) Two publications addressed pre-engraftment csCMVi in the last two years. (18, 19) In both, patients were closely monitored by quantitative PCR starting within the rst week after HCT, but the incidence of preengraftment CMVi was in these studies was different: Solano and colleagues reported a pre-engraftment CMVi incidence of 19%, whereas Martin and colleagues reported an incidence of 6.5%. Our cohort had an overall csCMVi incidence (73%) that was similar to that of Solano et al. but a higher incidence of preengraftment csCMVi (26%), even when analyzing only csCMVi episodes. Several differences in terms of HCT population characteristics can be noted, including the frequency of CMV seropositivity in our recipients, transplant modality, and baseline disease. (4,20) These differences probably contributed to the variation in pre-engraftment CMVi incidences.
Similar to Solano et al., we could not identify risk factors for pre-engraftment csCMVi by considering pre-HCT characteristics, the cellularity of donor progenitor infusion, conditioning strategy, or aGVHD prophylaxis. In addition, no differences in terms of viral features comparing pre-vs. postengraftment episodes were observed in Solano's cohort and our cohort. In our study, the frequency of tissue-invasive CMV disease was higher than that in other studies, but the frequencies of pre-and postengraftment csCMVi were similar.
Our series observed that pre-engraftment csCMVi was associated with a delay in engraftment compared to postengraftment csCMVi. This could not be attributed to the myelotoxicity of therapy, as only 3 of 29 pre-engraftment csCMVi patients started the treatment before engraftment. All of them were treated during neutropenia with foscarnet. Anti-CMV therapy was delayed in pre-engraftment compared with postengraftment events. Martin and colleagues also observed a delayed start of CMV treatment in patients with pre-engraftment CMVi, but no difference in engraftment was observed between those who started therapy before engraftment and those who started therapy after engraftment. It is important to highlight that Martin and colleagues did not compare pre-engraftment to postengraftment events. We hypothesize that the virus had an important role in the engraftment delay in our cohort, as the adjusted model maintained the signi cance of pre-engraftment csCMVi as a risk factor, and CMV is a well-known myelotoxic pathogen. (21,22) Cytomegalovirus replication is a possible contributor to acute GVHD development, as demonstrated by Cantoni et al. (23) In our series, the frequency of aGVHD was not different in pre-and postengraftment events.
CMVi has been associated with higher nonrelapse mortality in Allo-HCT. (1, 5) Unlike Solano, we observed a negative impact of pre-engraftment csCMVi on overall survival, and this effect was sustained months after the csCMVi episode. Overall survival post HCT depends on several factors, but it is feasible that preengraftment csCMVi can serve as an indicator of high-risk patients.
Our study has limitations. The rst is the retrospective and single-center design. As there is a lack of CMV treatment standardization and threshold viremia to indicate treatment, our results may not be extrapolated to other transplant centers.
This study described a potentially negative impact on outcome (engraftment delay and worse overall survival) in HCT patients who experienced pre-engraftment csCMVi. Close and early monitoring of CMV DNAemia after transplantation is needed to identify these patients. Prospective studies, including patients with letermovir prophylaxis, are necessary to de ne this population's standard in management and care. Currently, letermovir prophylaxis is the standard of care for high-risk patients (8, 24,25), and an early start of prophylactic therapy could be an effective intervention to reduce pre-engraftment csCMVi, especially in the context of high incidences of csCMV. Declarations