A retrospective analysis of clinical data was performed on 65 patients diagnosed with relapse of myeloid malignancy after allo-HSCT, who were treated at the Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, between November 2013 and December 2022. The study included 23 patients who received VEN+HMA and 42 patients who received IC. Patients who were initially diagnosed with primary or secondary AML or MDS and experienced hematological relapse after allo-HSCT were included in the study, while patients with severe organic dysfunction were excluded. Risk stratification, diagnosis of relapse and response criteria were according to European Leukemia Network 2017 criteria (16). Overall response rate (ORR) was defined as CR+CRi+PR. MRD positivity is defined as >0.01% myeloid blasts detected by multiparameter flow cytometry or >0.001% leukemia-associated genes detected by RT-qPCR. This study was approved by the ethical committee of the Institute of Hematology and Blood Diseases Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, and informed consent forms were obtained from all patients.

Azacitidine (50mg/m²/d for 5 days) was used as prophylactic therapy in 4 patients after allo-HSCT. All relapsed patients discontinued immunosuppressants after diagnosis. In VEN+HMA group, venetoclax was gradually increased to a maximal dose of 400mg/d in 3 days and each treatment cycle was 14-28 days. Combined hypomethylating agents include azacitidine (75mg/m²/d for 7 days) or decitabine (20mg/m²/d for 5 days). Furthermore, eleven patients in VEN+HMA group received low-dose cytarabine (20 mg/m² twice daily) for 14 days. Patients in IC group received CLAG or FLAG (cladribine 5mg/m²/day or fludarabine 30mg/m² plus cytarabine 1-2g/m²/day plus G-CSF 5 ug/kg for 5 days) or IDAC, including cytarabine 1 g/m²/q12h plus mitoxantrone 8-10 mg/m²/d or idarubicin 8-12 mg/m²/d or daunorubicin 45-60 mg/m²/d or amsacrine 100 mg/m²/d for 3 days. Previous unsuccessful regimens were avoided in the selection of IC regimens. Donor lymphocyte infusion (DLI) was obtained from previously cryopreserved donor graft or donor’s peripheral blood. Concomitant DLI infusion was administered in 63 patients, and calcineurin inhibitor was administered in patients receiving DLI from haploidentical donors or matched unrelated donors (MUD) to prevent GvHD. GvHD prophylaxis was identical between the two groups. Median mononuclear cells and median CD34⁺ cells each dose were 2.13 (1.22-4.00) *10⁸/kg and 0.60(0.08-2.12)*10⁶/kg in VEN+HMA group, and were 2.76 (0.96-8.33) *10⁸/kg (P=0.144) and 0.65(0.17-4.27)*10⁶/kg (P=0.442) in IC group. Bone marrow aspiration was performed after each treatment course and then continued monthly to evaluate efficacy in patients achieving complete response (CR)/CR with incomplete recovery of normal neutrophil or platelet counts (CRi). Overall survival (OS) was recorded from initiation of venetoclax or IC to last follow-up or death. Relapse-free survival (RFS) was defined as time from CR/CRi to the date of hematologic relapse or last follow-up. And Data cutoff date was January 31^th, 2023. Treatment-related mortality (TRM) was defined as death not directly caused by relapse.

During treatment session, blood routine examination, kidney and hepatic functions were monitored in all patients. Patients with neutropenic fever underwent blood culture for pathogenic microorganisms, chest imaging examination and antimicrobial therapy. Adverse events were evaluated according to CTCAE v5.0. Acute GVHD (aGVHD) and chronic GvHD (cGVHD) were diagnosed according to Glucksberg (17) and NIH (18) criteria, respectively.

The statistical analysis was performed using IBM SPSS (v.26) and R programming language (v 4.21). Quantitative variables were expressed as median (range), categorical variables were presented as rate and percentage. Mann-Whitney U test was performed for non-normally distributed quantitative data, Chi-square test and Fisher exact probability test were used for comparison of categorical variables. Survival analysis was conducted using Kaplan-Meier method and compared using log-rank test. Univariable and multivariable analyses were calculated via Cox proportional hazards regression model. Co-variables were selected using a stepwise forward procedure, and clinical factors with a P<0.1 in univariable analysis were selected to fit the multivariable model. A P<0.05 was considered statistically significant.

Patient information is summarized in Table 1 . No significant differences were observed between VEN+HMA and IC groups concerning age, gender, initial disease types, risk stratification, therapies before allo-HSCT, donor types and disease status at transplantation. Patients in VEN+HMA group carried FLT3-ITD (n=7), RUNX1 (n=2) and c-KIT (n=1) mutations, while those receiving IC regimen had FLT3-ITD (n=3), TP53 (n=4), ASXL1 (n=3), GATA2 (n=2) and c-KIT (n=2) mutations. Additionally, complex karyotypes were presented in 1 patient in VEN+HMA group and 6 patients in IC groups. All patients received myeloablative conditioning and GvHD prophylaxis before allo-HSCT. The majority of patients in both VEN+HMA (n=11) and IC (n=24, P=0.471) groups used matched-sibling donors (MSD). One patient in VEN+HMA group and 3 patients in IC group received azacitidine maintenance after transplantation.

Relapse and treatment information is displayed in Table 2 . One patient in VEN+HMA group suffered from skin involvement and received radiation therapy. In IC group, orbital chloroma (n=1), invasion of skin (n=2), ribs (n=1), lymph nodes (n=1) and vertebras (n=2) were observed. Two patients were treated with radiotherapy and four patients with either skin or vertebras invasion presented with concurrent bone marrow relapse. Notably, 17.4% of relapsed patients (n=4) in VEN+HMA group suffered from concomitant GVHD or lung infection (aGvHD=1, cGVHD=1, pneumocystis pneumonia =1, mycoplasma pneumonia with decreased oxygen saturation=1), while only 9.5% patients in IC group (n=4, P=0.597) had similar diseases (aGvHD=2, cGvHD=1, pulmonary mycosis=1). In VEN+HMA group, four patients used VEN+HMA as second (n=3) or third line (n=1) treatment, two of whom received previous IC regimen without response and switched to venetoclax-based regimen. Twenty-one patients received azacitidine and 2 patients used decitabine in combination with venetoclax. In addition, 11 patients in VEN+HMA group received 14-day low-dose cytarabine. In IC group, IC was the first-line treatment in 37 patients, second-line in 4 patients and third-line in 1 patient. IDAC (n=15), FLAG (n=16) and CLAG (n=11) were used. IDAC treatment included cytarabine combined with mitoxantrone (n=7) or idarubicin (n=4) or daunorubicin (n=3) or amsacrine (n=1).

Treatment efficacy was shown in Table 3 . All treatment responses were achieved in one cycle. Patients who did not respond, but were medically fit and willing to receive further therapies, were switched to a different regimen. In VEN+HMA group, twelve patients (52.2%) achieved CR/CRi (CR=2, CRi=10), with 4 patients (17.4%) reaching MRD negativity. However, eight of the 12 CR/CRi patients (66.7%) relapsed later. One of the 2 patients who failed prior IC achieved CRi, MRD+. Of the eight CR/CRi patients who continued with venetoclax-based treatment, one proceeded to second allo-HSCT and was alive until last follow-up. The other 4 responders all relapsed and were either treated successfully with FLAG (n=1) or died (n=3). Of the 11 non-responders, five switched to intensive (n=2) or low-dose chemotherapy (n=3), and allo-HSCT was performed in 1 NR patient, who later died of relapse. In the IC group, twenty-five (59.5%) patients achieved CR/CRi (CR=8, CRi=17), and 12 patients (28.6%) achieved MRD negativity. Eleven of the 25 patients (40.7%) who responded later relapsed. Eleven responders continued treatment with azacitidine (n=4), venetoclax (n=3), or DLI (n=4), and 6 of 17 non-responders were treated with azacitidine (n=1), DLI (n=4) or intensive chemotherapy (n=1). No statistical significance was observed between two groups regarding response, relapse after response, treatment-related mortality and early mortality.

Kaplan-meier survival analysis showed that achieving CR/CRi significantly improved patients’ prognosis (median OS 524 days in CR/CRi versus 130 days in PR/NR, P=0.004) ( Figure 1A ). Patients reaching MRD negativity also had significantly prolonged median OS (742 days in MRD negativity versus 169 days in MRD positivity, P=0.014) ( Figure 1B ). The median time of post-transplantation follow-up was not significant different (614 days in VEN+HMA group versus 377 days in IC group, P=0.347). Median OS was 209 days for VEN+HMA group and 211 days for IC group (P=0.491) ( Figure 1C ). In VEN+HMA group, ten patients died due to no response to regimen (n=8), relapse after CR/CRi (n=1) or severe pneumonia (n=1). In IC group, lack of response and relapse led to the death of 10 and 9 patients, respectively, and 8 patients died of infection (n=4) or GvHD (n=1) or multiorgan failure (n=3).

The univariable and multivariable analysis using Cox proportional hazards regression model ( Table 4 ) revealed that certain characteristics of patients’ initial diseases, including age, baseline ECOG score and adverse mutations did not significantly impact survival. TP53 mutation (HR=3.077 (95%CI 1.055-8.972), P=0.040), Grade III/IV aGvHD after treatment (HR=4.011 (95%CI 1.689-9.525), P=0.002) and time from allo-HSCT to relapse>1 year (HR=0.214 (95%CI 0.093-0.491), P<0.001) were found to have significant effects on survival in univariable analysis. Furthermore, multivariable analysis confirmed that late-onset relapse (HR=0.083 (95%CI 0.020-0.339), P=0.001) and treatment-induced grade III/IV aGvHD (HR=3.534 (95%CI 1.141-10.953), P=0.029) significantly impacted survival. In addition, multivariable analysis identified male gender (HR=4.406 (95% CI 1.599-12.140), P=0.004), FLT3-ITD mutation (HR=3.523 (95% CI 1.091-11.376), P=0.035), concomitant pulmonary infection (HR=4.060 (95% CI 1.027-16.056), P=0.046) and WBC>10,000/microL at relapse (HR=4.720 (95%CI 1.561-14.271), P=0.006) as posing significant risks. The subgroup analysis of survival was displayed in Figure 2 , demonstrating the positive trending effect of VEN+HMA regimen in multiple subgroups, with significance observed in patients with Hgb < 110g/L at relapse.

Detailed information of adverse events and GvHD is shown in Table 5 . All patients experienced grade 3-5 adverse events during their initial course of treatment. Thrombocytopenia was the most common event in both treatment groups, but the incidence was significantly higher in IC group than in VEN+HMA group (95.2% versus 73.9%, P=0.035). Pneumonia was the most common infection, with a significant higher incidence rate in IC group (50.0% versus 17.4%, P=0.010). The incidence of bacteremia was comparable between VEN+HMA group (17.4%) and the IC group (21.4%, P=0.948), and sepsis occurred in 4.3% and 4.8% patients, respectively (P=0.588). No cases of tumor lysis syndrome, patient intolerance or medication reduction were recorded, except the reduction of venetoclax to 100mg when combined with azoles. Of the 11 patients receiving further venetoclax therapy in VEN+HMA (n=8) and IC group (n=3), grade 3-5 adverse events were observed, including thrombocytopenia (n=5), neutropenia (n=3), upper respiratory infection (n=1), urinary tract infection (n=1), elevated aminotransferase (n=1).

After treatment of relapse, aGvHD incidence was significantly lower in VEN+HMA group (13.0% versus 50.0% in IC group, P=0.003). Grade III/IV aGvHD was observed in one patient (4.3%) in the VEN+HMA group and five patients (11.9%) in the IC group (P=0.577). Among patients with concomitant aGvHD at relapse in VEN+HMA (n=1) and IC group (n=2), one patient in each group suffered aGvHD progression. The disease severity of the 2 patients with concomitant cGvHD did not progress during treatment.