Pre-transplantation cytoreduction does not benefit advanced myelodysplastic syndrome patients after myeloablative transplantation with grafts from family donors

2.1 Patient selection

We enrolled consecutive MDS patients with excess blasts undergoing HSCT with grafts from family donors at our institution between January 2015 and December 2018. We included patients with ≥5% bone marrow (BM) blasts at peak, since this cut-off represents a potential trigger for pre-HSCT cytoreduction. Other inclusion criteria included sufficient cardiac function, lung diffusion capacity of at least 80%, transaminases level less than three times of the upper limit of normal, and creatinine clearance greater than 70 mL/min. Patients were excluded from HSCT if they had any active infections, transformation to acute myeloid leukemia, received URD HSCT, or received IC only before HSCT. Disease stages [refractory anemia with excess blasts (RAEB)-1 or RAEB-2] were classified based on the French-American-British (FAB) criteria [17]. Patients with ≥20% myeloblasts, defined as RAEB-t by the FAB criteria [17] but defined as acute leukemia by the world health organization [18], were excluded. Cytogenetic risk was defined based on the International Prognostic Scoring System (IPSS) criteria [19] and the revised IPSS (IPSS-R) criteria [20]. A monosomal karyotype (MK) was defined according to a published report [21]. All patients gave informed consent in accordance with the Declaration of Helsinki and according to the rules directed by the Institutional Review Board of Peking University People's Hospital.

The decision on the administration of pre-HSCT cytoreduction or BSC was made individually for each patient, mainly depending on the referral time. Patients referred to our center at diagnosis were intended to undergo upfront HSCT. In contrast, for patients referred to our center during the disease course, the referring physician made the decision on pre-HSCT treatment. In other words, most patients in the cytoreduction cohort received pre-HSCT cytoreduction at other hospitals prior to referral.

2.2 Prior therapy and response evaluation

Patients in the BSC cohort received only BSC (blood transfusions and growth factors). Patients in the cytoreduction cohort were given at least one cycle of HMA or HMA in combination with IC. IC regimens included at least the joint use of cytarabine and an anthracycline. Patients receiving IC only were excluded. According to the IWG criteria [22], complete remission (CR) and marrow complete remission (MCR) at the time of HSCT were grouped together as overall response (OR), while partial remission (PR), stable disease (SD), and progressive disease (PD) at the time of HSCT were grouped together as non-response (NR).

2.3 Donor selection and HLA typing

An MSD was the first choice for allo-HSCT. If an MSD was unavailable, subjects without a suitable closely HLA-matched URD (>8 of 10 matching HLA-A, B, C, DR, and DQ loci and > 5 of 6 matching HLA-A, B, and DR loci) or whose disease state left insufficient time for a URD searching were eligible for HIDT. HLA typing has been previously described in details [3].

2.4 HSCT procedure

For HIDT patients, the uniform conditioning regimen consisted of cytarabine (4 g/m², intravenous infusion, days -10 to -9; Pfizer Pharmaceuticals Ltd., New York, NY, USA), busulfan (3.2 mg/kg, intravenous infusion, days -8 to -6; Otsuka Pharmaceutical Co. Ltd., Naruto, Tokushima, Japan), cyclophosphamide (1.8 g/m², intravenous infusion, days -5 to -4; Baxter Oncology GmbH, Frankfurt, Hesse-Darmstadt, Germany), semustine (250 mg/m², oral administration, day -3; ZheJiang Ruixin Pharmaceutical Co. Ltd., Lishui, Zhejiang, China), and rabbit antithymocyte globulin (2.5 mg/kg, intravenous infusion, days -5 to -2; Imtix Sangstat, Lyon, France). MSDT patients received hydroxycarbamide (80 mg/kg, oral administration, day -10; Qilu Pharmaceutical Co. Ltd., Jinan, Shandong, China) and a lower dose of cytarabine (2 g/m², intravenous infusion, day -9) without antithymocyte globulin, and received the same other agents as in the HIDT regimen. GVHD prophylaxis was comprised of cyclosporine (2.5 mg/kg, intravenous infusion, since day -9), mycophenolate mofetil (1.0 g, oral administration, since day -9), and short-course methotrexate (15 mg/m², intravenous infusion, on day 1 and 10 mg/m² on days 3, 6, and 11) for all patients. Granulocyte colony stimulating factor (5 μg/kg) mobilized bone marrow and peripheral blood grafts were infused on the day of collection for HSCT. The total target mononuclear cell count for bone marrow and peripheral blood grafts was ≥6 × 10⁸/kg recipient weight. Details of the HSCT procedure have been previously described [23].

2.5 Follow-up

The follow-up period ended by March 1, 2020. Bone marrow examinations were scheduled at 1, 2, 3, 4.5, 6, 9, and 12 months after HSCT and at 6-month intervals thereafter.

2.6 Study endpoints and definitions

The primary endpoint was relapse-free survival (RFS). The secondary endpoints included rates of successful engraftment, relapse, non-relapse deaths, acute and chronic GVHD (aGVHD and cGVHD), and overall survival (OS). Assessments of endpoints have been previously described in details with all data calculated from the day of graft infusion [3]. Relapse was identified as hematologic recurrence of MDS based on standard criteria [22, 24].

2.7 Statistical analysis

The probabilities of OS and RFS were calculated using the Kaplan-Meier estimator with the log-rank test. Cumulative rates of successful engraftment, GVHD, relapse, and non-relapse deaths were estimated, taking into account competing risks. Competing events were defined as follows: for GVHD, successful engraftment, relapse, and death from any cause; for non-relapse deaths, relapse. Multivariate analyses were performed to identify independent predictors of non-relapse deaths, relapse, RFS, and OS. Pre-HSCT cytoreduction was included in the Cox proportional hazards model with forced entry modeling. Backward elimination with a criterion of P < 0.10 for retention was used to select a final model. The following variables were analyzed in the multivariate analysis: patient age, sex, disease characteristics (IPSS score, disease stage at peak, MK, blast count at HSCT), and HSCT-related variables (interval between diagnosis and HSCT, donor source, donor-recipient sex match). MK and complex karyotype (CK) were also included into the models due to their potential to distinguish risk groups for relapse, RFS, and OS in a linear fashion. Because pre-HSCT treatment with cytoreduction or BSC was not allocated through randomization, a 1:1 ratio PSM analysis was conducted using nearest neighbor or exact matching of age, sex, WHO disease stage, cytogenetics, interval between diagnosis and HSCT, and donor source. The SPSS software package (version 22.0, SPSS Inc., Chicago, IL, USA) and R statistical software (Bell Labs, New York, NY, USA) were used for data analyses.

3.1 Patients and pre-HSCT treatment

A total of 228 patients were enrolled. Among them, 162 (71.1%) underwent HIDT, and 66 (28.9%) underwent MSDT. Fifty-nine of the 162 patients receiving HIDT were previously reported [2] and further followed in this study. The median age was 42 years (range, 2-66 years). Among the 228 patients, 131 (57.5%) received solely BSC; 49 (21.5%) were given HMA, and 48 (21.1%) received both HMA and IC (HMA+IC). Before therapy, 42 (18.4%), 132 (57.9%), and 54 (23.7%) patients had intermediate-1, intermediate-2, and high-risk disease, respectively, per the IPSS criteria. Seventeen (7.5%) achieved CR, 56 (24.6%) achieved MCR, 135 (59.2%) had SD, and 20 (8.8%) had PD at the time of HSCT as defined by IWG criteria. In the cytoreduction cohort, the CR rate after pre-HSCT cytoreduction was 17.5% (4.1% for the HMA group and 31.3% for the HMA+IC group, P < 0.001), and the OR (CR+MCR) rate was 49.5% (36.7% for the HMA group and 62.5% for the HMA+IC group, P = 0.015).

Details on the characteristics of the patients and donors in the BSC and the cytoreduction cohorts are presented in Table 1. Patients in the cytoreduction cohort had a higher rate of refractory anemia with excess blasts subtype 2 and higher BM blast counts both at peak and at HSCT than patients in the BSC cohort (all P < 0.05).

3.2 Post-HSCT outcomes according to pre-HSCT therapy

As of the last follow-up, 50 patients had died, and 4 were lost to follow-up. The median follow-up period for all patients was 25.1 months (range, 0.1-61.1 months). The estimated rates of 3-year OS, 3-year RFS, relapse, and non-relapse deaths were 77.5% [95% confidence interval (CI): 71.9%-83.1%], 74.8% (95% CI: 69.0%-80.6%), 7% (95% CI: 3%-10%), and 17% (95% CI: 12%-22%), respectively.

The 3-year relapse rate was significantly lower in the BSC cohort than in the cytoreduction cohort [2.7% (95% CI: 0-5.7%) vs. 14.7% (95% CI: 6.7%-22.7%), P = 0.001], but no significant differences were observed in the rates of non-relapse deaths, 3-year RFS, and 3-year OS (all P > 0.05; Figure 1A-D). When stratifying the cytoreduction cohort into the HMA and HMA+IC groups, similar differences between the BSC cohort and the two groups were observed (overall P = 0.269 for RFS; P = 0.108 for BSC vs. HMA, and P = 0.610 for BSC vs. HMA+IC; Figure 1E.

3.3 Outcome predictors

For patients with CK, the 3-year RFS rate was comparable between the BSC and cytoreduction cohorts [69.0% (95% CI: 53.8%-84.2%) vs. 62.5% (95% CI: 28.3%-96.7%), P = 0.381]. Among the 37 patients with CK, only 9 were tested for TP53 mutations: 3 with TP53 mutation (2 achieved RFS until the last follow-up) and 6 without (2 achieved RFS until the last follow-up). Regarding the donor type, the 3-year RFS rate was not significantly different between the MSDT and HIDT groups [83.3% (95% CI: 74.1%-92.5%) vs. 71.1% (95% CI: 63.7%-78.5%), P = 0.080]. Multivariate analysis confirmed that BSC was better than cytoreduction regarding OS and RFS with lower relapse and comparable non-relapse deaths (Table 2). Older age, MK, and longer interval between diagnosis and HSCT reduced both RFS and OS. Older age and longer interval between diagnosis and HSCT were related to increased non-relapse death rates.

3.4 PSM analysis

We pair-matched 91 BSC with 91 pre-HSCT cytoreduction cases. The 3-year RFS rate was comparable between the two cohorts (81.9% vs. 70.1%, P = 0.070, Figure 2A) and among three treatment groups (BSC 81.9%, HMA 67.4%, and HMA+IC 72.7%, P = 0.137, Figure 2B).

3.5 Post-HSCT outcomes according to response to pre-HSCT treatment or blast count in the PSM dataset

Of the 91 patients who received pre-HSCT cytoreduction, 16 achieved CR at HSCT (CR group), and 75 did not (non-CR group). We compared the post-HSCT outcomes of these two groups with those of the BSC cohort in the PSM dataset. The 3-year RFS rates were 81.9%, 87.5%, and 66.9% in the BSC, CR, and non-CR groups, respectively (overall P = 0.051). The RFS was shorter in the non-CR group than in the BSC cohort (HR 2.003; 95% CI, 1.064-3.772; P = 0.031), while no difference was noted between the CR group and the BSC cohort (HR 0.725; 95% CI, 0.166-3.155; P = 0.668). Furthermore, in the cytoreduction cohort, 44 patients achieved OR (CR + MCR) at HSCT (OR group), and 47 had no response (NR group). The 3-year RFS rates were 81.9%, 77.3%, and 65.2% in the BSC, OR, and NR groups, respectively (overall P = 0.108; P = 0.411 for BSC vs. OR, and P = 0.033 for BSC vs. NR).

Since a panel of international experts has recently advised to conduct pre-HSCT cytoreduction at least in those MDS patients with a BM blast count of ≥10% [1], we compared the outcomes between patients with a BM blast count of <10% (n = 30) and ≥10% (n = 61) at peak in the BSC cohort. We found no difference in 3-year RFS rate (81.4% vs. 82.0%, P = 0.840). By dichotomizing the patients according to the cut-off of 5% blasts at HSCT, we observed no difference in 3-year RFS rate in the pooled population (80.0% vs. 74.3%, P = 0.575), while the difference was significant in the cytoreduction cohort (82.2% vs. 59.8%, P = 0.046).

3.6 Subgroup analysis in the PSM dataset

To further examine the influence of pre-HSCT cytoreduction on post-HSCT outcomes, we performed subgroup analyses with stratification by age, donor source, IPSS risk, and cytogenetic risk. The 3-year RFS rate was significantly higher in the BSC cohort than in the cytoreduction cohort among patients younger than 45 years, in patients with IPSS intermediate-2/high risk, and in patients with IPSS intermediate/poor-risk cytogenetics (all P < 0.05, Table 3).