Decitabine plus all-trans retinoic acid versus decitabine monotherapy for myelodysplastic syndromes with excess blasts: a multicenter, randomized controlled trial

Abstract

Despite standard treatment with hypomethylating agents, the prognosis of patients with higher-risk myelodysplastic syndrome (MDS) remains poor. All-trans retinoic acid (ATRA) has demonstrated promising efficacy in unfit patients with acute myeloid leukemia. This multicenter controlled trial randomized (1:1) untreated patients with MDS with excess blasts (MDSEB) to ATRA plus decitabine (ATRA at 25 mg/m2/day in 2 divided daily doses throughout the 28-day cycle plus decitabine at 20 mg/m2 on days 1-5) or decitabine alone (20 mg/m2 on days 1-5). The primary endpoint was the overall response rate within four treatment cycles. A total of 227 patients were randomized. Four patients who did not commence therapy were excluded from the modified intention-to-treat (mITT) analysis. The median patient age was 62 years (range, 19-81). The overall response rate was 78% (86/110) in the ATRA group versus 51% (58/113) in the decitabine group (odds ratio =3.40; 95% confidence interval [CI]: 1.90-6.09; P<0.001). The ATRA group also had a higher complete remission rate (23% vs. 12%; odds ratio =2.05; 95% CI: 1.02-4.25; P=0.042). With a median follow-up of 30.1 months, progression-free survival (PFS) was 14.9 months in the ATRA group versus 10.5 months in the decitabine group (hazard ratio [HR]=0.70; 95% CI: 0.51-0.97; P=0.03). The overall survival was 23.0 months and 19.3 months, respectively (HR=0.77; 95% CI: 0.54-1.09; P=0.137). The two groups did not differ in grade 3 or higher hematological adverse events. In conclusion, adding ATRA to decitabine increased the overall response rate and prolonged PFS in adult patients with MDS-EB without increasing hematological toxicity. This study was registered at Chinese Clinical Trial Registry (www.chictr.org.cn, identifier: ChiCTR1800018307).

Introduction

Myelodysplastic syndrome (MDS) is a clonal malignant disease of hematopoietic stem cells. Patients with higher-risk MDS, as classified by the International Prognostic Scoring System (IPSS) or Revised International Prognostic Scoring System (IPSS-R),¹ are at high risk of progression to acute myeloid leukemia (AML). Hypomethylating agents (HMA), including decitabine and azacytidine, are standard treatments for higher-risk MDS. However, patient outcomes remain poor despite HMA treatment. The rates of complete remission (CR) and overall response (OR) range from 7% to 17% and from 30% to 52%, respectively.^2-4 The median overall survival (OS) is 10.1-24.5 months.^2-11 Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is the only potentially curative treatment,¹² but many elderly patients are not appropriate candidates for allo-HSCT.

A variety of novel treatment strategies have been attempted for higher-risk MDS. The molecular targets of these treatments include somatic genetic mutations, cell-surface antigens, and anti-apoptotic pathways. Despite encouraging results in the early development phases, subsequent trials with larger sample sizes have generally been disappointing.^13-17 In a phase III trial of the anti-TP53 agent eprenetapopt plus azacitidine in patients with TP53-mutated MDS, the CR rate did not differ from that of azacitidine monotherapy.¹⁵ The PANTHER trial compared azacitidine alone versus azacitidine in combination with the NEDD8 inhibitor pevonedistat and failed to show improvement in event-free survival (EFS). The ENHANCE trial compared azacitidine alone versus azacitidine in combination with the CD47 antibody magrolimab but failed to improve CR and OS.¹⁸ The STIMULUS trial compared HMA alone with those combined with the TIM-3 inhibitor sabatolimab but also failed to improve CR and progression-free survival (PFS).¹⁹ All-trans retinoic acid (ATRA) induces blast differentiation in acute promyelocytic leukemia (APL) and is recommended for APL treatment. ATRA may be effective in AML without the PML-RARA fusion protein. Preclinical studies have indicated that HMA increase the sensitivity of AML cells to ATRA.^20,21 Reciprocally, ATRA can potentiate the cytotoxic effects of decitabine.^22,23 Lübbert and colleagues demonstrated the cooperative antileukemic effects of decitabine and ATRA on the AML cell lines U937 and MOLM-13. In addition to the derepression of tumor suppressor genes such as HIC1, the induction of transposable elements was observed with both drugs, triggering a “viral mimicry” response.²³ Our preclinical data revealed that decitabine induced Nrf2 activation and downstream anti-oxidative responses and inhibited reactive oxygen species (ROS) generation, thus leading to decitabine resistance. The addition of ATRA blocked Nrf2 activation by activating the RARα-Nrf2 complex, leading to ROS accumulation and ROS-dependent cytotoxicity.²² Two single-arm trials in the 2000s reported the promising efficacy of a therapeutic regimen that included HMA, ATRA, and valproate in mixed groups of AML and MDS patients.^24,25 The more recent DECIDER trial compared decitabine alone versus decitabine in combination with valproate, ATRA, or valproate plus ATRA in AML patients and demonstrated a higher objective response rate in patients treated with regimens that included ATRA but not valproate.²⁶ Notably, 56 of the 200 patients in the DECIDER trial fulfilled the criteria for oligoblastic AML (20-30% blasts) according to the current classification system but were diagnosed with refractory anemia with excess blasts in transformation (RAEB-T) according to the French-American-British classification at the time. A post hoc analysis of this subgroup revealed that the combination of ATRA and decitabine increased ORR and OS rates.²⁷ Based on these findings, we hypothesized that adding ATRA to HMA could improve treatment response in patients with MDS with excess blasts (MDS-EB) and conducted a prospective trial to compare decitabine in combination with ATRA versus decitabine alone in patients with newly diagnosed MDS-EB.

Methods

Participants and eligibility

This multicenter, randomized controlled trial was conducted in eight centers in China. Adult patients (≥18 years of age) with morphologically confirmed MDS-EB-1 or MDS-EB-2, as defined by the World Health Organization classification (WHO 2016), were eligible. The trial was conducted following the principles of the Declaration of Helsinki and the Good Clinical Practice guidelines of the International Council for Harmonization. The Institutional Review Boards approved the trial protocol (master protocol ID: 2018850). Written informed consent was obtained from all patients before enrollment. This trial is registered at the Chinese Clinical Trial Registry (Identifier: ChiCTR1800018307).

Study design and procedures

Patients were randomized at a 1:1 ratio to receive ATRA plus decitabine or decitabine alone. The randomization was stratified based on centers.

Each cycle lasted 28 days, and the study treatment was planned for at least four cycles. For each cycle, all patients received 20 mg/m² intravenous decitabine infusion over 60 minutes during the first 5 days. For the first four cycles, patients in the ATRA group also received oral ATRA at 25 mg/m²/day in two divided daily doses throughout the 28-day cycle (see page 16 of the Online Supplementary Appendix). Starting from the fifth cycle, ATRA was given at 25 mg/m²/ day only for the first 14 days of the 28-day cycle. The daily ATRA dosage was based on its recommendation in the Chinese clinical practice guidelines for APL.²⁸ Treatment continued until disease progression, unacceptable toxicity, or patient decision to withdraw from the trial (see pages 49-50 of the Online Supplementary Appendix).

Outcomes

The primary endpoint was the overall response rate (ORR), which was calculated based on the best response within four cycles and included CR, partial remission (PR), marrow CR (mCR), and hematologic improvement (HI) according to the modified International Working Group (IWG) 2006 response criteria in the MDS.²⁷

The secondary endpoints included the bone marrow blast response (CR+mCR), hematological response (CR+PR+HI), progression-free survival (PFS), OS, and treatment-emergent adverse events.

Statistical analysis

The sample size requirement was estimated via PASS 11.0 software (NCSS, Kaysville, UT) based on the following assumptions: (i) an ORR of 50% in the decitabine group¹¹ and 70% in the ATRA group; (ii) an α of 0.05; and (iii) a power (1-|3) of 80%. The calculation yielded 180 subjects (90 per group). Assuming a 20% dropout rate, we planned to enroll 226 patients. All efficacy endpoints and treatment-emergent adverse events were calculated using a modified intent-to-treat (mITT) population that included all patients who received at least one dose of the assigned treatment. Categorical variables were analyzed via the χ² test or Fisher’s exact test, as appropriate. Continuous variables with a normal distribution were analyzed via Student’s t test and are presented as the means ± standard deviations. Continuous variables not normally distributed were analyzed via the Mann-Whitney U test and are presented as the median and interquartile range (IQR). Survival outcomes were estimated via the Kaplan-Meier method and compared via the log-rank test. All the statistical analyses were conducted via SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results

Patient characteristics

Between September 11, 2018, and September 10, 2022, 243 patients were assessed for eligibility, and 227 were enrolled (113 and 114 patients in the ATRA and decitabine groups, respectively; Figure 1). The disease subtype was MDS-EB-1 in 109 patients and MDS-EB-2 in 118 patients. 28 (25%) patients in the ATRA group and 25 (22%) in the decitabine group received allo-HSCT. Twelve (11%) patients in the ATRA group and ten (9%) in the decitabine group received allo-HSCT within four cycles. At the time of data cutoff (May 31, 2023), 13 patients in the ATRA group and ten in the decitabine group were still receiving the assigned treatment.

Figure 1.Patient flow through the trial. CMML: chronic myelomonocytic leukemia; AML: acute myeloid leukemia; MDS: myelodysplastic syndromes; MPN: myeloproliferative neoplasms; ECOG: Eastern Cooperative Oncology Group; ATRA: all-trans retinoic acid; DEC: decitabine; HSCT: hematopoietic stem cell transplantation.

The median patient age was 62 years (IQR, 54-68; range, 19-81; Table 1). The IPSS-R category was high or very high in 181 (81%), intermediate in 33 patients (15%), and low in the remaining patient (0.4%). Mutated genes were generally balanced between the two groups (Figure 2). Post hoc reclassification of the patients with next-generation sequencing data based on the IPSS-M is shown on page 17 of the Online Supplementary Appendix.

The median number of decitabine cycles was 4 (range, 1-17) in the ATRA group and 3 (range, 1-20) in the decitabine group. The reasons for treatment discontinuation within four cycles are shown in Figure 1. ATRA was reduced in dosage in 13 patients and permanently discontinued in four patients before the completion of four cycles; decitabine was continued in all 17 patients.

Treatment response

Owing to the outbreak of COVID-19, 38 of the 110 patients in the ATRA group and 32 of the 113 patients in the decitabine group missed at least one scheduled bone marrow examination during the follow-up.

The ORR was 78% (86/110) in the ATRA group versus 51% (58/113) in the decitabine group (odds ratio =3.40; 95% confidence interval [CI]: 1.90-6.09; P<0.001; Table 2). The ATRA group also had a higher CR rate (23% vs. 12%; odds ratio =2.05; 95% CI: 1.02-4.25; P=0.042), CR+mCR rate (72% vs. 49%; odds ratio =2.69; 95% CI: 1.54-4.68; P<0.001), and CR+PR+HI rate (48% vs. 34%; odds ratio =1.84; 95% CI: 1.07-3.15; P=0.027). In patients with TP53 mutation, ORR was 79% (11/14) and 54% (7/13) in the ATRA+decitabine and decitabine alone groups, respectively (odds ratio =3.14; 95% CI: 0.59-16.85; P=0.181). The CR rate was 29% (4/14) in ATRA group and 23% (3/13) in decitabine group (odds ratio =1.33; 95% CI: 0.24-7.56; P=0.745). The median time to initial response did not differ between the two groups (1 cycle in both groups). The rate of response type after one treatment cycle was 3.6% (4/110) CR, 12.7% (14/110) mCR+HI, 38.2% (42/110) mCR without HI, and 6.4% (7/110) HI only in the ATRA plus decitabine group versus 2.7% (3/113) CR, 10.6% (12/113) mCR+HI, 23.9% (27/113) mCR without HI, and 3.5% (4/113) HI in the decitabine alone group. The rate of leukemic transformation was 58% (64/110) in the ATRA group versus 66% (75/113) in the decitabine group (odds ratio =1.42, 95% CI: 0.82-2.44; P=0.207). In patients with abnormal cytogenetic data at baseline, the rate of cytogenetic response was 76% (22/29) in the ATRA group and 57% (16/28) in the decitabine group (odds ratio =2.36; 95% CI: 0.76–7.32; P=0.134). The main effect (higher ORR in the ATRA group) was observed across all subgroups, except for women (P for interaction =0.032; Figure 3). The cause of the difference in efficacy between men and women needs further exploration. Subgroup analysis of overall response (CR+PR+mCR+HI) in patients with mutations is shown on page 9 of the Online Supplementary Appendix.

Table 1.Demographic and baseline characteristics.

Figure 2.Distribution of genes associated with poor prognosis and frequently mutated genes in patients by treatment group. Mutational analysis was conducted on 158 bone marrow aspirate samples collected at screening (N=81 in the all-trans retinoic acid [ATRA] group and N=77 in the decitabine [DEC] group).

Survival outcomes

Within a median follow-up of 30.1 months (interquartile range [IQR], 17.7-44.8), a total of 129 (58%) patients died. The number of patients with disease progression and relapse was 68 (62%) in the ATRA group and 82 (73%) in the decitabine group. The median PFS was 14.9 months (95% CI: 9.5-20.3) in the ATRA group and 10.5 months (95% CI: 7.2-13.7) in the decitabine group (hazard ratio [HR]=0.70; 95% CI: 0.51-0.97; P=0.032; Figure 4). The two groups did not differ in median time to AML transformation (18.9 vs. 15.8 months; HR=0.81; 95% CI: 0.58-1.14; P=0.221) or OS (23.0 vs. 19.3 months; HR=0.77, 95% CI: 0.54-1.09; P=0.137). PFS censored for HSCT also supported the favorable results of the ATRA group (shown on page 33 of the Online Supplementary Appendix). For patients with TP53 mutation, median PFS was 8.3 months (95% CI: 1.88-14.71) in the ATRA group and 9.4 months (95% CI: 2.47-16.33) in the decitabine group (HR=0.80; 95% CI: 0.34-1.86; P=0.605 and OS was 10.7 months (95% CI: 3.7-17.7) in the ATRA group and 10.5 months (95% CI: 3.0-18.0) in decitabine group (HR=0.68; 95% CI: 0.27-1.69; P=0.401) was numerically shorter than in the overall patient. Subgroup analyses of PFS, OS, and time to AML transformation are shown on pages 18-20 of the Online Supplementary Appendix. Post hoc subgroup analyses to assess survival outcomes according to IPSS-R and IPSS-M scores are shown on pages 21-32 of the Online Supplementary Appendix.

Adverse events

An overview of treatment-emergent adverse events is shown in Table 3. The most common treatment-emergent adverse events of any grade were thrombocytopenia (75% in the ATRA group vs. 70% in the decitabine group), followed by leukopenia (71% vs. 71%), neutropenia (59% vs. 55%), anemia (37% vs. 31%), dry skin (35% vs. 5%), febrile neutropenia (28% vs. 27%), fatigue (25% vs. 27%), headache (19% vs. 6%), hypertriglyceridemia (17% vs. 5%), and pneumonia (14% vs. 17%) (Table 3). The most common treatment-emergent adverse events of grade 3 or higher were thrombocytopenia (66% in the ATRA group vs. 63% in the decitabine group) and leukopenia (66% vs. 63%), followed by neutropenia (55% vs. 54%), febrile neutropenia (28% vs. 27%), and anemia (26% vs. 24%). The majority of the non-hematologic treatment-emergent adverse events were grade 1-2. The ATRA group had higher rates of dry skin (35% vs. 5%; P<0.001), headache (19% vs. 6%; P=0.020), and hypertriglyceridemia (17% vs. 5%; P=0.019) (Table 3). No differentiation syndrome was observed. The rate of serious hematological treatment-emergent adverse events was 86% (94/110) in the ATRA group and 80% (90/113) in the decitabine group (P=0.254). The rate of non-hematological treatment-emergent serious adverse events was 35% (38/110) in the ATRA group and 23% (26/113) in the decitabine group (P=0.057). Details of serious treatment-emergent adverse events with a ≥3% frequency in either group are listed on page 8 of the Online Supplementary Appendix.

The rates of death due to serious treatment-emergent adverse events were 1% (1/110) in the ATRA group and 3% (3/113) in the decitabine group (P=0.622). The death in the ATRA group was attributed to intracranial bleeding. The cause of death for the three patients in the decitabine-only group was infection.

Discussion

In this multicenter, randomized controlled trial, we demonstrated a higher ORR (78% in the ATRA group vs. 51% in the decitabine group), a higher rate of hematological response (CR+PR+HI rate 48% vs. 34%, response not including mCR) a higher rate of CR (23% vs. 12%), and improved PFS (14.9 months vs. 10.5 months) in the ATRA group. Hematological treatment-emergent adverse events did not differ between the two groups.

Table 2.Hematological responses.

The ORR in the decitabine group in this trial was 51%. This rate was generally consistent with the ORR as a composite of CR, PR, HI, and mCR without HI in MDS-RAEB patients receiving monotherapy with decitabine reported in previous studies.¹¹ The modified ORR, which included CR, PR, and HI, was exclusive of mCR without HI in this trial (34%) and was also similar to that reported for decitabine monotherapy in MDS patients (30%).⁴ The CR rate in this trial (12%) was also consistent with that reported in a previous study (9%).⁴ The median time to progression to AML or death in this trial (15.8 months) was also similar to that reported in a previous study (12.1 months).⁴ These results demonstrated that the patient sample in this trial was representative, which highlights the potential for applicability to a general population of patients with MDS-EB-1 or MDS-EB-2.

Since mCR (a reduction in BM blast count to ≤5% without HI) has not been shown to be associated with prolonged OS, including mCR without HI in the ORR could result in overestimation of treatment response and bias in the group comparison. We therefore analyzed treatment response by excluding mCR (i.e., CR+PR+HI), and the results continued to show a higher response rate in the ATRA group versus the decitabine alone group (48% vs. 34%; P=0.027). Further more, since reduction in marrow blasts may be beneficial in patients who would be bridged to allo-HSCT, higher rate of marrow response (i.e., CR+mCR) in the ATRA group also supported its potential benefits in bridging patients to allo-HSCT.^29-31 However, very-high risk MDS patients exhibited reduced responsiveness to decitabine-ATRA. Therefore, future therapeutic strategies should focus on more intensified or rationally designed treatment regimens. Promising strategies may include: triplet combinations incorporating venetoclax, immunomodulatory agents (such as anti-CD47 or anti-TIM-3 antibodies), or targeted inhibitors (e.g., IDH1/2 inhibitors); optimization of ATRA and/or HMA dosing based on bone marrow niche targeting; and biomarker-guided patient selection for ATRA-based combination therapies. The median PFS was 14.9 months in the ATRA group versus 10.5 months in the decitabine group (P=0.032) in this trial. OS did not differ between the two groups. This discrepancy between PFS and OS outcomes may be partly explained by the variety of subsequent treatments administered to patients in both the combination therapy and control groups after disease progression. These treatments included HSCT, other hypomethylating agents, and conventional chemotherapy. Such factors are likely to act as significant confounders, thereby interfering with an accurate assessment of the independent survival benefit attributable to the decitabine-ATRA combination regimen itself. Additionally, the relatively small sample size may have contributed to limited statistical power. Even if the combination therapy holds potential clinical value, this limited power could have resulted in the failure to detect a statistically significant difference in OS between the two groups.

Figure 3.Forest plot of the subgroup analysis. Effect of treatment on the overall response rate in subgroups with different demographic and disease characteristics. P values are for interactions. ATRA: all-trans retinoic acid; OR: odds ratio; CI: confidence interval; ECOG: Eastern Cooperative Oncology Group; EB: excess blasts; IPSS-R: revised International Prognostic Scoring System.

Table 3.Treatment-emergent adverse events.

Figure 4.Kaplan-Meier curves for survival outcomes. (A) Progression-free survival (PFS), (B) overall survival (OS) and (C) transformation to acute myeloid leukemia (AML) in the modified intention-to-treat population at the final analyses. ATRA: alltrans retinoic acid; HR: hazard ratio; CI: confidence interval.

Hematologic treatment-emergent adverse events did not differ between the two groups in this trial, supporting the hematologic safety of ATRA. However, the ATRA group had a higher rate of several non-hematologic treatment-emergent adverse events, including dry skin, headache, and hypertriglyceridemia. This profile is generally consistent with previous trials of ATRA in humans.^26,32-36 In addition, consistent with results previously reported in AML patients,²⁶ most of the non-hematologic treatment-emergent adverse events reported in this trial were of low grade and reversible upon dose reduction or delay.

Interestingly, while the phase III SELECT-MDS-1 trial recently failed to demonstrate a significant improvement in CR rates with azacitidine plus tamibarotene, a selective RARA agonist, compared to azacitidine plus placebo in higher-risk MDS patients overexpressing RARA,³⁷ our study showed that ATRA, also an RARA agonist, significantly improved response rates and PFS when combined with decitabine. This discrepancy suggests that ATRA may exert anti-tumor effects through mechanisms beyond RARA agonism. Notably, emerging evidence indicates that ATRA can degrade β-catenin in the bone marrow niche independently of genetic mutations, thereby disrupting the β-catenin-Jag1 axis and potentially contributing to therapeutic efficacy in non-APL myeloid malignancies.³⁸ This mechanism, which is not shared by all RARα agonists, may help explain the enhanced activity observed in our trial and highlights the multifaceted nature of ATRA’s mode of action. In addition, the two different drugs and schedules (in combination with azacitidine instead of decitabine) used in the two studies made it quite difficult to compare.

This trial has several limitations. First, the primary endpoint was the ORR rather than survival outcomes. Future trials with either PFS or OS as the primary endpoint are needed to verify the potential survival advantage of ATRA plus decitabine. Second, bone marrow follow-up has deviated substantially from the plan due to the disruption caused by the COVID-19 pandemic. Accordingly, PFS and time to leukemia transformation might have been overestimated. Third, quality of life was not examined as planned. Fourth, genetic testing for mutation panels varied significantly across trial centers. In conclusion, treatment with decitabine plus ATRA resulted in a higher ORR and prolonged PFS than decitabine monotherapy in patients with MDS-EB without increasing hematologic treatment-emergent adverse events. No new non-hematologic treatment-emergent adverse events were reported. These findings encourage phase III trials using survival outcomes as the primary endpoint.

Footnotes

• Received July 10, 2025
• Accepted November 14, 2025

Correspondence

*XZ, YL, YG, ZG and LH contributed equally as first authors

H. Tong tonghongyan@zju.edu.cn

Disclosures

No conflicts of interest to disclose.

Contributions

HT and XZ were involved in the study design, data acquisition and interpretation, manuscript writing and critical review, and accessed and verified the data. HT, XZ, YL, YG, ZG, LH, JZ, HC, GO, YT, YK, FZ, LS, WJ, GX, LM, LY, CM, JJ and FM enrolled and treated patients and gathered data. JL conducted the statistical analysis. XZ drafted the manuscript. All the authors reviewed the final manuscript and took responsibility for the data. All the authors approved the article for publication.

Funding

This trial was supported by research grants from the National Health Commission Major Disease Prevention and Control Science and Technology Action Plan, the Special Research Projects for Cancer Prevention and Control (2018ZX-07S-007), the National Natural Science Foundation of China (82270146), and the Key R&D Program of Zhejiang Province (2024C03164).

Acknowledgments

The authors thank all the patients and their families who participated in this study. LongFine Pharmaceuticals Co., Ltd., provided ATRA. Dr. Kehong Zhang from Ivy Medical Editing (Shanghai, China) provided language editing (compensated by LongFine Pharmaceuticals Co., Ltd.).