Background: Post-transplant cyclophosphamide (PTCy) is widely used for graft-versus-host disease (GVHD) prophylaxis in haploidentical hematopoietic stem cell transplantation (haplo-HSCT). This study aimed to compare CMV reactivation kinetics among patients receiving different substandard PTCy doses (less than 50 mg/kg D+3, +5) versus the standard dose as an optimization approach for PTCy dosage, and to develop predictive model for CMV reactivation and treatment response.

Methods: We conducted a single-center retrospective cohort study of 200 patients who underwent haplo-HSCT between December 2020 and December 2023. Patients were stratified into four groups based on PTCy dosing: 25 mg/kg (n=50), 30 mg/kg (n=50), 40 mg/kg (n=50), and standard dose 50 mg/kg (n=50), all administered on days +3 and +5. Primary endpoints were incidence to CMV reactivation, peak CMV viral loads by PCR, and duration of viremia within 100 days post-HSCT. Secondary endpoints included; 6-month overall survival (OS), incidence of acute GVHD, hemorrhagic cystitis (HC) incidence, relapse rate, engraftment kinetics, and Graft-versus-host disease-free-relapse, free survival (GRFS). Mathematical model was constructed to predict patient response to antiviral treatment and identify optimal timing for therapy initiation. Machine learning models were used to develop a predictive score for CMV reactivation.

Results: Median follow-up was 6 months. CMV reactivation rates were 50%, 53%, 70%, and 60% in the 25, 30, 40, and 50 mg/kg groups, respectively (p=0.62). Median time to CMV reactivation was significantly delayed in the 25 mg/kg group (41 days) compared to other groups (25, 27, and 30 days for 30, 40, and 50 mg/kg, respectively; p=0.03). Peak viral loads and duration of viremia were highest in the 25 mg/kg group (p=0.04 and p=0.02, respectively). Six-month OS was 47%, 59%, 79%, and 47% for the 25, 30, 40, and 50 mg/kg groups, respectively (p=0.89). NRM at 6 months was 12%, 11%, 10%, and 11% across the groups (p=0.97). Incidence of grade II-IV acute GVHD was higher in the 25 mg/kg group (73% vs. 47%, 42%, 71%; p=0.0001). Multivariate analysis identified PTCy dose, recipient CMV seropositivity, use of anti-thymocyte globulin and acute GVHD grade II-IV as independent risk factors for CMV reactivation and high-level viremia. Mathematical modeling revealed that initiating antiviral therapy within 48 hours of CMV detection improved treatment response across all PTCy dose groups. Our symbolic classification model demonstrated high accuracy in predicting CMV reactivation (AUC 0.85, 95% CI 0.79-0.91), with a sensitivity of 88% and specificity of 82%.

Conclusion: Low-dose PTCy interval between 30 to 40 mg/kg seems to offer a sweet spot for an optimum PTCy dose with similar CMV reactivation kinetics and promising 180-days outcomes compared to the standard 50 mg/kg dose in haplo-HSCT. ⁤Nevertheless, 25 mg/kg dose was associated with delayed timing of CMV reactivation and longer duration of viremia with higher PCR peaks. ⁤⁤Our findings could help identifying patients at higher risk of CMV reactivation and optimizing timing of antiviral therapy in a prophylactic and/or pre-emptive approach. ⁤ ⁤⁤However, Prospective validation of these findings is needed. ⁤

Disclosures

No relevant conflicts of interest to declare.

This content is only available as a PDF.
2024
Sign in via your Institution