A dual-center, retrospective study of patients undergoing HCT for SAA was conducted at Vanderbilt University Medical Center (VUMC) and the associated Veterans Affairs hospital, Tennessee Valley Healthcare System (TVHS). The VUMC and TVHS Institutional Review Board approved the study. Patients with a SAA diagnosis who underwent first allogeneic HCT using FCR conditioning regimen at VUMC or TVHS between January 2016 and May 2022 were included in the study. Patients were excluded if they were younger than 18 or had not completed all planned treatments at the time of data collection.

All patients received conditioning per established protocol as determined by degree of HLA-matching with their designated donor (Fig. 1). Patients received PBSC or BM grafts per treating physician’s discretion with allogeneic HCT performed on day 0. In patients with matched related, matched unrelated, or 1-allele mismatched donors, fludarabine (30 mg/m2) was given intravenously for four days (–7 to –4, i.e., 7 to 4 days before transplantation) in combination with cyclophosphamide (750 mg/m2) given intravenously for three days (–6 to –4) and anti-thymocyte globulin (rabbit) (3.75 mg/kg) for two days (–2 and –1). They were also given rituximab (375 mg/m2) on days –13, –7, +1, and +8.

Doses and timing of each agent in FCR conditioning regimens for patients undergoing HCT for SAA with matched related, matched unrelated, or 1-allele mismatch donor (a) or haploidentical donor (b). Time in days is represented along the horizontal axis progressing from left to right, with day of transplantation depicted as day 0.

Patients who underwent HCT from a haploidentical donor received fludarabine (30 mg/m2) intravenously for five days (–6 to –2) and cyclophosphamide (14.5 mg/kg) for two days (–6 to –5). They also received rituximab (200 mg/m2) on day +5. All patients received total body irradiation at a dose of 200 cGy on day 0 for those with matched related, unrelated, or 1-allele mismatched donors, or on day –1 for those with haploidentical donors.

Prophylaxis for GVHD in patients with matched related, matched unrelated, or 1-allele mismatch donors consisted of tacrolimus starting on day –3 and methotrexate, 5 mg/m2 intravenously, on days +1, +3, and +6 after transplantation. Patients with haploidentical donors received standard post-transplant cyclophosphamide (50 mg/kg on days +3 and +4), and tacrolimus and mycophenolate mofetil starting on day +5. Tacrolimus levels were adjusted to a goal range of 5–15 ng/mL per institutional standard, and administered for 180 days, at which point a taper was initiated provided absence of GVHD. Mycophenolate mofetil was administered until day +35 in patients with haploidentical donors.

The primary outcome of interest was GVHD-free/relapse-free survival. A patient was considered to have an event if they experienced moderate or severe GVHD (including both acute GVHD [aGVHD] and cGVHD), relapse, or death. If a patient experienced multiple events, the earliest event date was used as the time to event (e.g., if a patient had a diagnosis of both aGVHD and cGVHD, the earliest date of diagnosis was used). If a patient did not experience an event until the end of follow-up time (i.e., the last date the patient was seen in the clinic or lost to follow-up [unable to contact, transitioned care to another city, etc.]), it was censored. Acute and chronic GVHD were graded according to Glucksburg and 2014 National Institutes of Health consensus criteria, respectively [17, 18]. Secondary outcomes included time to engraftment, incidence of graft failure, incidence of GVHD, rate of viral reactivation, post-HCT disease status, and number of inpatient hospital days.

Descriptive statistics were used to summarize the patient characteristics. Medians and interquartile ranges (IQRs) were used for continuous variables, while frequencies and percentages were used for categorical variables. Differences in patient characteristics were tested for using Wilcoxon rank sum tests for continuous variables and chi-square tests for categorical variables. Probability of GRFS over time was estimated using the Kaplan-Meier estimation method.