Mycophenolate mofetil is associated with inferior overall survival in cytomegalovirus-seropositive patients with acute myeloid leukemia undergoing hematopoietic cell transplantation

A combination of a calcineurin-inhibitor (CNI) with mycophenolate mofetil (MMF) or methotrexate (MTX) is commonly used as graft-versus-host disease (GvHD) prophylaxis for HLA-matched donor hematopoietic cell transplantation (HCT). However, the use of MMF, which targets both T- and B-lymphocytes, is associated with a higher risk of cytomegalovirus (CMV) reactivation as compared to MTX.¹ Moreover, CMV reactivation is associated with an increased risk of non-relapse mortality (NRM) and worse overall survival (OS).² We, therefore, postulated that the use of MMF as compared to MTX may increase these risks, especially in CMV-seropositive recipients. To investigate this hypothesis, we compared the HCT outcomes of four groups: MTX/ CMV– (n=916), MTX/CMV⁺ (n=1,527), MMF/CMV– (n=267), and MMF/CMV⁺ (n=395).

We used an existing Center for International Blood and Marrow Transplant Research (CIBMTR) publicly available dataset³ from a previous publication.⁴ Our study population included patients ≥18 years with acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), or myelodysplastic neoplasm (MDS) who underwent HCT with peripheral blood grafts from 10/10 HLA-matched unrelated donors between 2008 and 2017. All patients received tacrolimus with either MMF or MTX. Patients who received ex-vivo T-cell-depleted/CD34⁺ cell-selected grafts or post-transplant cyclophosphamide (PTCy) were excluded. Our primary outcome of interest was OS. Secondary outcomes included grades III-IV acute GvHD, chronic GvHD, relapse, and NRM. The Kaplan-Meier method was used to estimate OS probability. The cumulative incidence method accounting for competing risks was used to estimate the incidence of GvHD, NRM, and relapse. Competing risks considered were death or disease relapse for GvHD; disease relapse or relapse-related deaths for NRM; and death before relapse (NRM) for relapse. Predictors in univariate and multivariable analyses were evaluated using Cox proportional hazards regression for OS, and Fine-Gray regression accounting for competing risks for GvHD, NRM, and relapse. Factors significant at the 5% level (P≤0.05) were retained in the final model, except the main effect (GvHD prophylaxis/ recipient CMV serostatus), which was retained in the final model irrespective of the level of statistical significance. Bonferroni-adjusted P values are reported for multivariate analyses for the main effect to account for multiple testing.

Interaction effects between the main effect and other statistically significant co-variables were tested and accounted for as indicated in multivariable regression analyses. All reported outcomes are at 3 years, except acute GvHD which is at day 180. The proportionality of hazards assumption was tested graphically and statistically. All statistical analyses were performed using STATA/IC 16.1 (StataCorp LLC, College Station, TX, USA). As the data analysis was carried out at The MD Anderson Cancer Center, the local Institutional Review Board approved this study (protocol: 2022-0684), which was conducted in accordance with the Declaration of Helsinki. The dataset is publicly available³ for data sharing, in accordance with CIBMTR guidelines. The study population’s baseline characteristics are shown in Table 1. In patients with AML (n=1,572), OS was significantly (P<0.001) inferior in CMV⁺ recipients who received MMF prophylaxis. Overall survival at 3 years was only 31% (95% confidence interval [95% CI]: 25-37) in the MMF/CMV⁺ group, compared with 54% (95% CI: 45-63) in the MMF/ CMV– group, 51% (95% CI: 48-55) in the MTX/CMV⁺ group, and 58% (95% CI: 53-63) in the MTX/CMV– group (Figure 1A). This effect persisted in multivariate analysis: compared to the overall mortality in the MTX/CMV– group, the risk was 1.8-fold higher in the MMF/CMV⁺ group, while it did not differ significantly in other groups (Table 2, Online Supplementary Table S1). Inferior OS in the MMF/CMV⁺ group was driven by a higher risk of NRM (Figure 1B). The risks of relapse (Figure 1C) and grade III-IV acute and chronic GvHD (Figure 1D) did not differ significantly between the groups. In MDS patients (n=1,080), GvHD prophylaxis/recipient CMV serostatus had no significant association with OS, or the risk of relapse, but NRM varied by the Hematopoietic Cell Transplantation-specific Comorbidity Index (HCT-CI) score (Table 2, Online Supplementary Figure S1A). Among those with an HCT-CI score ≥3, the MTX/CMV⁺ group had a higher risk of NRM than had the MTX/CMV– group, while no statistically significant differences were seen in patients with an HCT-CI score 0-2. The risk of chronic GvHD was higher in the MMF/CMV– group than in the MTX groups.

In ALL patients (n=453), GvHD prophylaxis/recipient CMV serostatus had no significant associations with OS, NRM, or relapse (Table 2, Online Supplementary Figure S1B). The risk of chronic GvHD was higher in the MMF/CMV– group than in the MTX groups.

There were no significant predictors of grade III-IV acute GvHD in any disease type.

Next, we tested whether the use of PTCy prophylaxis altered these observations. We used a separate CIBMTR matched unrelated donor peripheral blood HCT cohort⁵ of patients with AML (n=136), ALL (n=42) or MDS (n=64) who received PTCy/CNI/MMF prophylaxis (Online Supplementary Table S2). In this population too, in multivariate analysis, CMV⁺ recipients had a significantly worse OS than CMV^– recipients, but only if they had AML (hazard ratio [HR]=2.7, 95% CI: 1.1-6.4; P=0.03) and not if they had ALL (HR=0.4, 95% CI: 0.1-1.5; P=0.2), or MDS (HR=0.7, 95% CI: 0.3-1.5; P=0.3). The risk of relapse and NRM did not differ by CMV serostatus in any disease group, although these analyses were limited by the small number of events.

In summary, in the CNI study population of patients who all received tacrolimus for GvHD prophylaxis with either

Table 1.Baseline characteristics.

Figure 1.Outcomes of patients according to their cytomegalovirus status and whether they received methotrexate or mycophenolate mofetil for graft-versus-host disease prophylaxis. Among patients with acute myeloid leukemia, four groups were studied: patients given methotrexate (MTX) who were cytomegalovirus (CMV) seronegative (MTX/CMV^–, dotted black), patients given MTX who were CMV seropositive (MTX/CMV⁺, solid black), patients given mycophenolate mofetil (MMF) who were CMV seronegative (MMF/CMV^–, dotted maroon) and those given MMF who were CMV seropositive (MMF/CMV⁺, solid maroon). Outcomes include (A) overall survival, (B) non-relapse mortality, (C) relapse, and (D) chronic graft-versus-host disease. AML: acute myeloid leukemia; GvHD: graft-versus-host disease.

MMF or MTX and underwent matched unrelated donor HCT with peripheral blood grafts, we showed that MMF was associated with a significantly higher risk of NRM and worse OS in CMV⁺ individuals but only in AML patients, and not in those with MDS or ALL. The underlying mechanism of this observation is unclear, but one hypothesis is as follows. It is known that CMV induces an exaggerated proliferation of natural killer cells, γ/δ T cells, and cytotoxic T cells.^6,7 As MMF inhibits both B- and T-lymphocytes,¹ it is associated with broader immunosuppressive effects than MTX, and CMV-induced lymphocytic expansion^8,9 is suppressed by MMF more than by MTX.¹⁰ Moreover, MMF is administered for an extended period after HCT, while MTX is given for a short course. It is, therefore, conceivable that MMF may thwart CMV-driven expansion of both innate and adaptive immune cells, which may not only negate any potential protective effects of CMV on relapse but may also increase the risk of infections and NRM. Several studies have shown that CMV is associated with a reduced risk of relapse, but predominantly in AML patients^11-13 and not in those with lymphoid malignancies,^11,13,14 or even MDS.^12,13 The additional inhibition of humoral and T-cell immunity in the presence of MMF would be predicted to promote CMV reactivation with its associated highly suppressive immunological imprinting and infectious sequelae.¹⁵

Table 2.Multivariate analysis.*

The effect of MMF when used with PTCy remains unknown. In our ad hoc analysis of the PTCy cohort, our findings were similar to those in the CNI cohort (not given PTCy), i.e., increased mortality in CMV⁺ recipients with AML, but not in those with ALL or MDS. However, the independent effects of MMF and CMV on this association could not be determined as all patients received PTCy/CNI/MMF. These findings are hypothesis-generating and provide foundational data for further studies to assess the interaction between MMF and CMV serostatus in patients with AML, versus other hematologic malignancies, receiving PTCy-based prophylaxis.

Limitations of our study are the lack of data on CMV reactivation and causes of death. Because we lacked these data, we do not know the precise reason for the worse OS in the MMF/CMV⁺ group. Of note, almost all patients in the CNI cohort and a majority in the PTCy cohort underwent HCT in the era before letermovir was approved by the US Food and Drug Administration. Whether or not the use of letermovir alters these conclusions also remains to be investigated.

In conclusion, our data suggest that the use of MMF with CNI prophylaxis should be avoided in CMV-seropositive AML patients undergoing matched unrelated donor HCT. Further studies are needed to assess the interaction

between MMF and CMV serostatus when PTCy is used for GvHD prophylaxis, and in the setting of letermovir prophylaxis.

Footnotes

• Received October 19, 2023
• Accepted February 16, 2024

Correspondence

R.S. MEHTA - rmehta@fredhutch.org

Disclosures

No conflicts of interest to disclose.

Contributions

RMS performed the statistical analysis, interpreted data, and wrote the statistical section of the manuscript. SJL, PAC, JRH, CJL, AA, GC, MD, REC, KR, and EJS reviewed and interpreted the data, reviewed the manuscript, and provided critical feedback. RSM and RMS conceptualized the study design and interpreted data. RSM wrote the manuscript with the help of RMS. RMS and RSM had full access to the raw data which is publicly available. All authors approved the manuscript. The corresponding author had the final responsibility for submitting the manuscript for publication.

Funding

Acknowledgments

We thank the Center for International Blood and Marrow Transplant Research (CIBMTR) staff for providing the dataset used in this study.