A retrospective study was performed for adult patients with hematologic malignancies who underwent haplo-PBSCT in our center from May 2017 to December 2021. All the patients received the low-dose ATG/PTCy-based regimen for GVHD prophylaxis. The graft source was from mobilized PBSCs with G-CSF. A single of unrelated cord blood cells was prescribed as the third-party cells for a minority of patients. Family members selected as haploidentical donors were defined on human leukocyte antigen (HLA)-A, -B, -C, -DRB1, and -DQB1 locus at the high-resolution level with the recipient -donor mismatched number (HLA) ≥3 (21). The study had ethical approval from hospital ethical committees (No: 2022KY023) and was conducted in accordance with the Declaration of Helsinki. All patients included in the study signed informed consent.

Reduced-intensity conditioning (RIC) regimens were prescribed for patients with advanced age (≥55 years) or hematopoietic cell transplantation- comorbidity index (HCT-CI) above 2, while myeloablative conditioning (MAC) regimens were designed for other patients ( Supplementary Figure 1 ) (32). For myeloid malignancies, the MAC regimen was composed of intravenous busulfan (Bu, 3.2 mg/kg/day for 4 days), fludarabine (Flu, 30 mg/m²/day), and cytarabine (Ara-C, 1-2 g/m²/day) both for 5 days; while the RIC regimen included Bu (3.2 mg/kg/d for 2 days), Flu and Ara-C for 5 days with the same doses as in the MAC regimen, total body irradiation (TBI, 3Gy on the day -1). For lymphoid malignancies, the MAC regimens included TBI-based and Bu-based regimens. The TBI-based regimen was composed of 10Gy fractioned TBI (FTBI), cyclophosphamide (Cy, 50 mg/kg/d for 2 days), and etoposide (VP-16, 10 mg/kg/d for 2 days); while Bu-based regimen consisted of Bu (3.2 mg/kg/d for 4 days) combined with the same doses of Cy and VP-16 as above. The RIC regimen included intravenous Bu (3.2 mg/kg/d for 2 days), Cy and VP16 with the same doses as in the MAC regimen, and TBI (3Gy).

All the patients received the low-dose ATG/PTCy-based regimen for prophylaxis of GVHD including ATG 2.5 mg/kg/d on day -2 to -1, Cy 50 mg/kg/d on day +3, CsA and MMF initiating on day +4. The starting infusion dose of CsA was 2 mg/kg, after which the dose was modified to obtain a nadir serum level between 200 and 300 ng/ml, eventually tapering from day +90 to day +180. MMF was administered orally at 15 mg/kg three times daily (maximum dose of 3 g per day) until day +34 and discontinued if no aGVHD (29).

G-CSF was given to all patients starting on day +5 at 5 µg/kg/day until neutrophil recovery. Prophylactic ganciclovir at 5 mg/kg/12h was given to patients during the conditioning period for 1 week. Prophylactic antifungals were used from conditioning until at least 3 months after transplantation. CMV-DNA in serum and EBV-DNA in whole blood were routinely monitored by quantitative polymerase chain reaction once a week until at least day +100.

Neutrophil engraftment was defined as obtaining an absolute neutrocyte (ANC) ≥ 0.5 × 10⁹/L for 3 consecutive days after transplantation without G-CSF. Platelet engraftment was defined as obtaining a platelet count ≥ 20 × 10⁹/L for the first of 7 consecutive days without platelet transfusion (33). Full donor chimerism was defined as ≥ 95% of donor T cells in BM samples (34). Graft failure was defined as failure of neutrophil engraftment on day 28 following transplantation (primary graft failure, PGF), or loss of donor chimerism after initial engraftment at any time without disease relapse (secondary graft failure, SGF) (33). aGVHD was diagnosed and graded in line with the modified Glucksberg criteria (35), and chronic GVHD (cGHVD) according to the 2014 National Institutes of Health consensus criteria (36). Morphologic complete remission (CR) was determined by the International Working Group (IWG) and National Comprehensive Cancer Network (NCCN)guidelines (version 3.2013) criteria (37–40) and patients not in morphologic CR were considered to have an active disease (41). Relapse was defined by the appearance of blasts in the peripheral blood (PB) or BM (>5%) after CR (21). Non-relapse mortality (NRM) was defined as death from any cause other than relapse. Overall survival (OS) was defined as the time from the day of stem cell infusion to death from any cause or follow-up. Relapse-free survival (RFS) was defined as survival in continuous CR. GVHD-free and relapse-free survival (GRFS) was defined as survival without the following events: grade III-IV aGVHD, severe cGVHD, disease relapse, or death from any cause after haplo-HSCT (42).

The main endpoints of this study included the CIs of aGVHD, cGVHD, relapse, and NRM, and the probabilities of OS, RFS, and GRFS. Survival curves were plotted using the Kaplan-Meier method, and subgroups were compared by log-rank tests. Relapse, NRM, and GVHD were calculated using a CI estimate to accommodate the following competing events (death for relapse, relapse for NRM, and both death and relapse for GVHD), and subgroups were compared by the Fine and Gray test. The prognostic significances of covariates affecting OS, RFS, and GRFS were determined by the Cox proportional hazards regression model. The prognostic significances of covariates affecting the CIs of relapse, NRM, and GVHD were determined using Fine-Gray proportional hazards regression for competing events. Multivariate analyses were performed using variables with a P value < 0.10 in prior univariate analyses. Continuous variables and percentages for categorical variables were expressed via median values and ranges. The Mann-Whitney test was used to analyze continuous variables. All statistical analyses were performed using ‘R’ software version 4.2.1. Statistical significance was set at P value < 0.05.

A total of 260 patients were enrolled in the study. Details of the patient, donor, and allograft characteristics are summarized in Table 1 . In total, 162 patients with myeloid malignancies were enrolled in the study, including 130 with acute myeloid leukemia (AML), 6 with chronic myelomonocytic leukemia (CMML), and 26 with myelodysplastic syndrome (MDS), while 98 patients with lymphoid malignancies were enrolled, including 62 with acute lymphoblastic leukemia (ALL), 35 with non-Hodgkin’s lymphoma (NHL), and 1 with multiple myeloma (MM). The last enrolled patient underwent HSCT at least 12 months before the initiation of follow-up. The median follow-up time was 27.0 months (range, 0.2 to 67.9 months). We performed a transplant conditioning intensity (TCI) score for the conditioning regimen (43). The TCI score ranged from 1.5 to 5.5 (median 4.5) with a median of 4.5 (range, 2.5–5.5) in the MAC group and 3.0 (range, 1.5–4.0) in the RIC group (p=0.000).

Graft failure was observed in nine patients (3.46%), of which only one (1.02%) occurred in lymphoid malignancies and the remaining eight (4.94%) developed in myeloid malignancies (p=0.050). PGF accounted for six (2.31%) and SGF for three (1.15%). Five out of the nine patients received a second transplant, of which two patients achieved long-term survival. The median time for neutrophil engraftment was 12 days (range, 9–28 days), while the median time for platelet engraftment was 13 days (range, 9–87 days).

For the entire cohort, the CIs of grade I-IV, II-IV, and III-IV aGVHD by 28 days were 26.15% (95% confidence interval [CI], 20.97%-31.64%), 9.23% (95% CI, 6.10%-13.14%), and 4.23% (95% CI, 2.24%-7.18%), respectively. The 180-day CIs of grade I-IV, II-IV, and III-IV aGVHD were 35.00% (95% CI, 29.24%-40.81%), 13.46% (95% CI, 9.64%-17.92%), and 5.77% (95% CI, 3.37%-9.07%), respectively ( Figure 1A ). The CIs of grade II-IV (p=0.273) and III-IV aGVHD (p=0.838) were similar between the myeloid and lymphoid malignancies ( Figure 1B ), although the CI of grade I-IV aGVHD in myeloid malignancies was significantly lower than that in lymphoid malignancies [29.01% (95% CI, 22.22%-36.14%) vs 41.84% (95% CI, 31.94%-51.41%), p=0.003)]. The CIs of total and moderate/severe cGVHD in all patients within 2 years after transplantation were 30.97% (95% CI, 25.43%-36.66%) and 18.08% (95% CI, 13.68%-22.98%), respectively ( Figure 1C ). The CIs of total and moderate/severe cGVHD between myeloid and lymphoid malignancies were similar (p=0.398 and p=0.160, respectively) ( Figure 1D ). In our long-term follow-up, the proportion of patients with cGVHD requiring second-line therapy was 28.05% (23/82).

By the end of the follow-up, 104 patients died. Relapse (n=43) and infection (n=34) accounted for 74.04% of the death. The 2-year OS, RFS, and GRFS for the entire cohort were 60.70% (95% CI, 54.80%-67.10%), 58.10% (95% CI, 52.20%-64.50%), and 50.60% (95% CI, 44.80%-57.10%), respectively. There were no significant differences in OS (p=0.683, Figure 2A ), RFS (p=0.995, Figure 2B ), and GRFS (p=0.990, Figure 2C ) between myeloid and lymphoid malignancies. All results of the univariate analyses are included in the Supplementary Material . In the multivariate Cox analysis ( Table 2 ), disease status at transplantation was an independent prognostic factor for OS (HR,1.9; 95% CI,1.3-2.9; p=0.002), RFS (HR, 1.8; 95% CI, 1.2-2.7; p=0.005), and GRFS (HR, 1.7; 95% CI, 1.2-2.5; p=0.004). The patients with CR at transplantation had significantly higher 2-year OS (67.70% vs 45.40%, p=0.000, Figure 2D ), RFS (64.30% vs 44.50%, p=0.000, Figure 2E ), and GRFS (57.30% vs 36.00%, p=0.000, Figure 2F ) than those of patients with NR. Meanwhile, patients without cGVHD had significantly lower OS (HR, 0.55; 95% CI, 0.35-0.87; p=0.010) and RFS (HR, 0.57; 95% CI, 0.37-0.87; p=0.009) than those with cGVHD. The 2-year OS and RFS for patients without cGVHD were 57.83% and 54.21%, respectively, while for patients with cGVHD, these were 67.85% and 66.50%, respectively (p=0.028, Figure 2G for OS; p=0.037, Figure 2H for RFS). ECOG scores also had a strong trend toward lowering the OS (HR, 1.9; 95% CI, 0.97-3.7; p=0.060), RFS (HR, 1.9; 95% CI, 0.99-3.8; p=0.054), and GRFS (HR, 1.7; 95% CI, 0.92-3.1; p=0.089).

The 2-year OS, RFS, and GRFS for myeloid malignancies were 59.70% (95% CI, 52.40% - 68.10%), 59.40% (95% CI, 52.20% - 67.50%), and 51.40% (95% CI, 44.10% - 59.80%), respectively. In multivariate analysis, disease status at transplantation was the independent prognostic factor for OS (HR, 1.9; 95% CI, 1.1-3.2; p=0.018), RFS (HR, 1.8; 95% CI, 1.1-3.0; p=0.021), and GRFS (HR, 1.7; 95% CI, 1.1-2.7; p=0.029) ( Table 3 ). Patients with CR at transplantation had significantly higher 2-year OS (69.50% vs 46.20%, p=0.001, Figure S2A ), RFS (68.90% vs 46.20%, p=0.001, Figure S2B ), and GRFS (61.50% vs 37.30%, p=0.001, Figure S2C ) than those with NR. Higher ECOG scores (2–4) were associated with inferior survival outcomes in OS (HR, 2.4; 95% CI, 1.2-4.8; p=0.014), RFS (HR, 2.4; 95% CI, 1.2-4.8; p=0.014), and GRFS (HR, 1.9; 95% CI, 1.7-2.7; p=0.050) ( Table 3 ). The 2-year OS, RFS, and GRFS for patients with ECOG scores of 2-4 were 33.30%, 33.30%, and 26.70%, respectively, while for patients with ECOG scores of 0-1, these were 62.50%, 62.20%, and 54.00%, respectively (p=0.005, Figure S2D for OS; p=0.004, Figure S2E for RFS; and p=0.010, Figure S2F for GRFS, respectively). cGVHD was associated with better outcomes for 2-year OS (72.50% vs 53.40%; p=0.019, Figure S2G ) and 2-year RFS (71.90% vs 53.00%; p=0.038, Figure S2H ).

The 2-year OS, RFS, and GRFS were 62.30% (95% CI, 53.10%-73.00%), 56.0% (95% CI, 46.70%- 67.10%), and 49.3% (95% CI, 40.20% - 60.50%), respectively. In multivariate analysis, disease status at transplantation was the independent prognostic factor for OS (HR, 2.6; 95% CI, 1.1-5.8; p=0.023), RFS (HR, 2.4; 95% CI, 1.1-5.2; p=0.029), and GRFS (HR, 2.1; 95% CI, 1.0-4.4; p=0.037) ( Table 4 ). Patients with CR at transplantation had significantly higher 2-year OS (65.80% vs 42.90%, p=0.095, Figure S3A ), RFS (59.1% vs 35.7%, p=0.053, Figure S3B ), and GRFS (52.70% vs 28.60%, p=0.037, Figure S3C ) than those with NR. RIC regimen was also associated with inferior survival outcomes in OS (HR, 5.3; 95% CI, 2.0-14.0; p=0.001), RFS (HR, 4.8; 95% CI, 1.8-13.0; p=0.002), and GRFS (HR, 4.6; 95% CI, 1.9-11.0; p=0.001). The 2-year OS, RFS, and GRFS for patients with the RIC regimen were 28.57%, 28.57%, and 28.60%, respectively, while for patients with the MAC regimen, these were 64.90%, 58.20%, and 52.70%, respectively (p=0.005, Figure S3D for OS, p=0.010, Figure S3E for RFS and p=0.001, Figure S3F for GRFS, respectively). Higher HCT-CI scores (≥2) were associated with worse outcomes for 2-year OS (20.00% vs 64.50%; p=0.019, Figure S3G ) and 2-year RFS (20.00% vs 57.90%; p=0.040, Figure S3H ).

For the entire cohort, the 1-year and 2-year NRMs were 20.77% (95% CI,16.07%-25.90%) and 23.04% (95% CI, 18.06%-28.40%), respectively. In total, 61 patients died from NRM, which accounted for 58.65% (61/104) of all deaths. Infection was the most common cause of NRM, accounting for 42.62% (26/61), followed by cGVHD at 19.67% (12/61), organ failure at 18.03% (11/61), hemorrhagic diseases at 8.20% (5/61), graft failure at 6.55% (4/61), and aGVHD at 4.92% (3/61). The 2-year NRMs between myeloid and lymphoid malignancies were similar (22.99% vs 23.08%, p=0.670; Figure 3A ). In multivariate analysis, older patients (>median age) (HR, 2.42; 95% CI, 1.266-3.96; p=0.006) and grade II-IV aGVHD (HR, 1.81; 95% CI, 1.003-3.26; p=0.049) were the unfavorable independent prognostic factors for NRM ( Table 2 ). The 2-year NRMs were significantly increased in older patients (32.76% vs 13.43%; p=0.000, Figure 3B ) and patients with grade II-IV aGVHD (37.65% vs 20.78%; p=0.045, Figure 3C ). For myeloid malignancies, the multivariate analysis results showed that older patients had a significantly increased NRM compared with younger patients (12.05% vs 34.75%; p=0.003, Figure 3D ) ( Table 3 ). For lymphoid malignancies, higher HCT-CI scores (≥2) significantly increased the NRM (50.00% vs 15.84%; p=0.023, Figure 3E ) ( Table 4 ).

In total, 43 patients relapsed in the entire cohort and the median time of relapse was post-transplant 16 months (range, 1-36 months). For the entire cohort, the 1-year and 2-year CIs of relapse were 15.77% (95% CI, 11.64%-20.46%) and 18.90% (95% CI, 14.33%-23.97%), respectively. CIs of relapse were similar between myeloid and lymphoid malignancies (p=0.560; Figure 4A ). In multivariate analysis for the entire cohort, the disease status was the independent prognostic factor for relapse, while cGVHD had a strong trend toward lowering the relapse rate ( Table 2 ). CR status at transplantation significantly lowered the 2-year CI of relapse as compared with NR status (15.66% vs 25.98%; p=0.019, Figure 4B ). NR status at transplantation also had strong trends of increasing the risk of relapse for myeloid (HR, 1.95; 95% CI, 0.957-3.97; p=0.066) and lymphoid malignancies (HR, 2.662; 95% CI, 0.990-7.16; p=0.052). The 2-year CI of relapse of patients with NR status at transplantation was higher than that with CR status, whether for myeloid (22.60% vs 14.08%; p=0.066, Figure 4C ) or for lymphoid malignancies (42.86% vs 17.41%; p=0.036, Figure 4D ). The rate of HLA loss in relapsed patients was 14.28% (2/14). We carried out maintenance treatments for patients with tyrosine kinase inhibitors such as sorafenib for FMS-like tyrosine kinase 3 internal tandem duplication (FLT3-ITD) mutated AML and dasatinib for Philadelphia chromosome-positive ALL (Ph⁺ ALL). Preemptive treatment was adopted for patients with measurable residual disease (MRD, flow cytometry, and/or reverse transcription-polymerase chain reaction) relapse. Azacytidine in combination with interferon α was given to 49 patients with MRD-positive myeloid malignancies including AML and MDS, of which 20/24 (83.33%) achieved MRD-negative after 6 cycles of therapy.

The median time of CMV and EBV reactivation was 114 days (range,15-1041) and 429 days (range, 41 - 939) after transplantation, respectively. The 1-year CIs of CMV and EBV reactivation were 43.46% (95% CI, 37.39%-49.37%) and 18.08% (95% CI, 13.68%-22.98%) in the entire cohort. The 1-year CIs of CMV and EBV reactivation were similar between myeloid and lymphoid malignancies (p=0.147, p=0.482). In the entire cohort, 28 patients (10.77%) had CMV disease, and 7 (2.69%) developed post-transplant lymphoproliferative disorder (PTLD), of which 5 were associated with EBV reactivation. No statistically significant differences were found between the 2-year CIs of PTLD between myeloid and lymphoid malignancies [2.47% (95% CI, 0.81%-5.80%) vs 3.06% (95% CI, 0.82-7.98%), p=0.764]. The CI of BKV-related hemorrhagic cystitis was 16.54%, which had no difference between myeloid and lymphoid malignancies (p=0.600). A total of 10 patients showed activation of human herpesvirus-6 B (HHV-6B) after transplantation. All patients were included in immune reconstitution studies and 81 cases were analyzed at each endpoint. On days +120, median CD3⁺, CD4⁺, CD8⁺, CD19⁺, and CD56/CD16⁺ counts were 954 (95-4891), 180 (8-743),754 (73-4272), 34 (1-276), and 215 (19-2539)/μl, respectively.