Low-dose decitabine for refractory prolonged isolated thrombocytopenia after HCT: a randomized multicenter trial

Allogeneic hematopoietic cell transplantation (HCT) is a widely used treatment for hematologic malignancies with the potential for cure.¹  Prolonged isolated thrombocytopenia that is refractory to conventional treatments has remained a critical complication after allogeneic HCT since its recognition for 35 years.^2-4  The frequency of prolonged refractory thrombocytopenia after allogeneic HCT has been reported to be 20% to 37%, but the defining criteria of this disorder have been inconsistent.^3,5  Prolonged isolated thrombocytopenia post-HCT is usually associated with other post-HCT complications, including relapse of underlying disease, infection, graft-versus-host disease (GVHD), and thrombotic microangiopathy.^4-6  However, there are other patients with prolonged isolated thrombocytopenia without an identified cause who are often resistant to conventional treatments for thrombocytopenia, such as glucocorticoids, IV immunoglobulin, rituximab, interleukin-11 (IL-11), and recombinant human thrombopoietin (rhTPO) or TPO receptor agonists.^2,3  We define these patients by a platelet count <30 × 10⁹/L for >60 days post-HCT that is not associated with relapse of the underlying disease, severe GVHD, infection, or other recognized etiology and that is unresponsive to conventional treatments for thrombocytopenia. We describe these patients as having refractory prolonged isolated thrombocytopenia (RPIT). The cause of RPIT may be decreased platelet production.^2,3  Prognosis for these patients is poor.

Platelet production is a consecutive process beginning with the commitment of hematopoietic stem cells to the megakaryocyte lineage and ending with the fragmentation of megakaryocyte cytoplasm into platelets. Effective hematopoiesis depends on the bone marrow microenvironment where hematopoietic stem cells reside and interact with supportive cells.^7,8  Cross-talk between megakaryocytes and bone marrow endothelial cells (BMECs) in the marrow vascular microenvironment regulates megakaryocyte maturation and thrombopoiesis. In patients undergoing HCT, the interruption of homeostasis in the bone marrow microenvironment caused by the intensive conditioning treatments and alloreactivities associated with allogeneic transplantation may contribute to the etiology of thrombocytopenia.^9,10  Previous studies documented impaired BMECs, both in quantity and function, in patients with thrombocytopenia post-HCT.¹⁰ 

The pathogenesis of RPIT remains unclear, and effective treatment is lacking. It has been demonstrated that decitabine, a hypomethylating agent,¹¹  may increase platelet counts in mice by enhancing platelet release and megakaryocyte maturation.¹²  In addition, decitabine may ameliorate endothelial progenitor cell (EPC) impairment and moderate immune dysfunction in hematopoietic microenvironment.^13,14 

We previously reported a pilot study demonstrating the effectiveness of low-dose decitabine in increasing platelet counts in transplant recipients with RPIT.¹⁵  We have now conducted a randomized multicenter trial to validate the efficacy and safety of low-dose decitabine in patients with RPIT after HCT, and we have further explored the underlying mechanisms.

This prospective, open-label, 3-arm clinical trial was designed by investigators at the First Affiliated Hospital of Soochow University (Suzhou, China) and conducted at 6 hematology centers in China. The inclusion criteria were: (1) platelet count <30 × 10⁹/L for >60 days post-HCT, (2) recovered neutrophil and hemoglobin concentrations, (3) donor chimerism >95% detected by short tandem repeat polymerase chain reaction, and (4) no response to conventional treatments for thrombocytopenia (rhTPO alone or combined with IV immunoglobulin, rituximab, IL-11, or glucocorticoid) for at least 4 weeks. Exclusion criteria were: (1) relapse or progression of underlying disease, (2) active infection, (3) grade 3 to 4 acute GVHD or severe chronic GVHD according to National Institutes of Health criteria,^16,17  (4) severe organ damage (cardiac, renal, and/or hepatic dysfunction grade >2 according to the Common Terminology Criteria for Adverse Events [version 5.0]), (5) thrombosis requiring treatment, and (6) decitabine for any indication after the current transplantation. In addition, enrolled participants would be withdrawn from this trial if judged as having relapse or progression of the underlying disease before initial response evaluation at day 28.

After enrollment, patients were randomly assigned to 1 of 3 intervention arm: arm A, 15 mg/m² of decitabine (Chia Tai Tianqing Pharma, Lianyungang, China) daily IV for 3 consecutive days (days 1-3) followed by 300 U/kg of rhTPO (SanSheng Pharma, Shenyang, China) daily subcutaneously from day 4 until response or initial evaluation (day 28); arm B, 15 mg/m² of decitabine daily IV for 3 consecutive days (days 1-3) alone; or arm C, conventional therapies with recommended options, including IV immunoglobulin, IL-11, and glucocorticoids, alone or in combination. Decitabine was administered for 1 cycle in both arms A and B, followed by 28 days of evaluation and additional 24 weeks of follow-up. Dosage adjustment of rhTPO in arm A was not permitted unless participants withdrew from this study.

Hematologic adverse events (AEs) were assessed by both duration and maximum deviation from baseline level of white blood cells and hemoglobin. Nonhematologic AEs were graded according to Common Terminology Criteria for Adverse Events (version 5.0), and severe AEs were defined as grade ≥3 AEs or AEs resulting in unplanned hospital stays. From day 1, all patients were evaluated for safety every day until the day of evaluation (day 28).

Donors for allogeneic HCT were selected based on the results of HLA typing, age, sex, and health status. Generally, HLA-matched related younger male donors were preferred. HLA haploidentical donors and HLA-matched unrelated donors were selected according to institutional guidelines.

The protocol was approved by the institutional review board at each participating center. All participants provided written informed consent. The trial was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines.

Bone marrow aspirate was performed at baseline and day 28 after treatment. The total number of megakaryocytes as well as the platelet-shedding megakaryocytes of bone marrow smears (per cm²) was counted and cross-checked by blinded observers as previously described.¹⁸  To measure ploidy distribution, megakaryocytes were labeled with PE-Cy5–conjugated CD41 (BD Biosciences) and incubated with propidium iodide (Abcam). The samples were then evaluated using an ACEA NovoCyte flow cytometer.

Data from preliminary experiments performed before this trial suggested that patients with RPIT had decreased EPCs, BMECs, and related cytokines reflecting the function of BMECs. These included vascular endothelial growth factor (VEGF), endothelia-1, C-X-C chemokine receptor type 4 (CXCR4), stromal cell–derived factor-1 (SDF-1), and fibroblast growth factor-4 (FGF-4). Endogenous nitric oxide (NO), another biomarker of BMEC impairment, was increased. Therefore, the levels of these cytokines and chemokines were measured for patients enrolled in this trial. EPCs and BMECs were detected in mononuclear cells and stained with mouse anti-human CD45, CD34, CD133, and VEGF receptor monoclonal antibodies (BD Biosciences). SDF-1, VEGF, endothelial-1, and endogenous NO levels were assessed by ELISA kits (R&D Systems) and CXCR4 and FGF-4 by reverse transcription polymerase chain reaction (Thermo Fisher Scientific).

Evaluation of the primary end point was conducted at day 28; platelet response was defined as sustained increase (stable or increasing level) of at least 30 × 10⁹/L independent of transfusion for 3 days.¹⁵  Counts of blood cells were monitored daily during the first 4 weeks after decitabine treatment and every 4 weeks until the 28th week. Secondary end points included megakaryocyte counts 4 weeks after decitabine treatment and survival during follow-up of 24 weeks. Additional follow-up after 24 weeks was conducted to observe long-term survival. Overall survival was calculated from the date of enrollment to date of death or end of follow-up.

According to the previous study, among the 3 arms, we supposed the biggest effective rate to be 80% (P_max) and the lowest rate to be 35% (P_min). Under the condition of α = 0.05 (2 tailed), β = 0.20 (1 tailed), and ν = 2, so λ = 9.63, the theoretic sample size for each group was calculated with the following formula: n = λ/(2[arcsin √P_max − arcsin √P_min]²) = 9.63/(2[arcsin √0.80 − arcsin √0.35]²) = 21.42. Considering an expulsion rate of 20%, the sample size for each group was at least 25.Completely randomized block design method was performed to assign patients to 1 of the 3 arms. For each study site, we produced the random series and corresponding assigning arms A, B, and C, in which the block equaled 5 and length equaled 6. If the length was ≤2, the arm was A; if length was 3 to 4, arm was B; and if length was 5 to 6, arm was C. Numeric data are presented as medians with interquartile ranges (IQRs), and categorical data are shown as proportions. One-way analysis of variance was used to compare the difference of numeric variables among 3 treatment arms, and χ² or Wilcoxon rank test was used to compare the categorized variables. Survival was estimated by Kaplan-Meier method and compared by log-rank test. Cox and logistic regression models were used to evaluate the impact of variables on outcomes. Significance level was set as a 2-tailed P value of <.05. All analyses used SPSS (version 19.0; SPSS, Inc.).

From October 2015 to April 2019, of 2616 allogeneic transplant recipients screened at 6 participating centers, 256 patients were identified as having thrombocytopenia (platelet count <30 × 10⁹/L) for >60 days post-HCT (Figure 1). Ninety-seven patients met the inclusion and exclusion criteria and were randomly allocated to arm A (n = 32), B (n = 33), or C (n = 32; Figure 1). Subsequently, 2 patients in arm A, 3 in arm B and 1 in arm C were excluded because of relapse before initial evaluation. Therefore, 30 patients in arm A, 30 in arm B, and 31 in arm C were included in the analyses. The main treatments for RPIT after HCT in arm C included glucocorticoids, rhTPO, and immunoglobulin, alone or in combination. Baseline characteristics were comparable among the 3 arms (Table 1).

Patients in arms A and B had a significant increase in platelet count beginning in the third week after initiation of treatment (Figure 2A). Response rates were similar between arms A (n = 20 patients; 66.7%) and B (n = 22; 73.3%; P = .779); both arms A and B were superior to arm C (n = 6; 19.4%; P < .001). For the intention-to-treat data set (n = 97 participants in total), response rates in arms A (62.5%) and B (66.7%) were both higher than that in arm C (18.8%; P < .001). Median response durations for responding patients were comparable among arms A (25 weeks; IQR, 1-26 weeks), B (24 weeks; IQR, 1-27 weeks), and C (25 weeks; IQR, 3-27 weeks; P = .471 for comparison among 3 arms; P = .314 between arms A and B). Median use of platelet transfusions was 3, 3, and 5 units per week in arms A, B and C, respectively (P = .001).

Total megakaryocyte counts, platelet-shedding megakaryocytes, and megakaryocyte polyploidy of patients in arms A (P = .017, .027, and .015, respectively) and B (P = .042, .013, and .017, respectively) were significantly elevated after treatment (Figure 2B), suggesting improved megakaryocyte proliferation and maturation in patients who received decitabine. There was no increase of these megakaryocyte parameters in arm C (P = .836, .875, and .485, respectively). The increase in megakaryocytes after decitabine was also visually apparent on bone marrow biopsies (Figure 2C).

All variates described in Table 1 were tested in risk factor analyses for response, which revealed the trial intervention (decitabine) to be the only significant variate. Compared with arm C, both arms A (hazard ratio [HR], 8.333; 95% confidence interval [CI], 2.585-26.864; P < .001) and B (HR, 11.458; 95% CI, 3.439-38.181; P < .001) led to better response rates (Table 2).

During follow-up of additional 24 weeks after the initial evaluation, 4 patients had delayed platelet recovery (at the 6th, 11th, 14th, and 18th week during follow-up, respectively) in arm A, 3 in arm B (at the 5th, 10th, and 17th week during follow-up, respectively), and 6 in arm C (at the 5th, 10th, 11th, 18th, 25th, and 26th week during follow-up, respectively). No statistical difference was found in comparison among the 3 arms (P = .655).

At the end of 28 weeks after treatment, survival was 80.0% for arm A, 83.3% for arm B, and 58.1% for arm C (P = .057). Because survival rates in arms A and B were similar (P > .99), we combined arms A and B and found superior 28-week survival for patients treated with decitabine compared with those receiving conventional treatments (82.5% vs 60.0%; P = .22). With a median total follow-up of 11 months, the estimated 1-year survival rate was 64.4% ± 9.1% for arm A, 73.4% ± 8.8% for arm B, and 41.0% ± 9.8% for arm C (P = .025; Figure 3A). Decitabine treatment (arms A + B) was associated with significantly greater 1-year survival (68.2% ± 6.4% vs 41.0% ± 9.8%; P = .008; Figure 3B).

Six patients relapsed before the primary end point evaluation (day 28) and were therefore excluded from this study. An additional 3 patients in arm A, 2 in arm B, and 2 in arm C relapsed during long-term follow-up (median, 11 months). No difference in overall incidence of relapse was found among the 3 arms (P = .807).

Thirty-five patients died during follow-up, including 10 in arm A, 8 in arm B, and 17 in arm C. More patients died as a result of bleeding (n = 6 [35.3%] of 17) in arm C compared with arms A (n = 2 [20.0%] of 10) and B (n = 1 [12.5%] of 8), although the difference was not statistically significant (P = .543). Other causes of death were severe GVHD (n = 1 in arm A, 2 in B, and 3 in C), infection/sepsis (n = 3, 3, and 5, respectively), relapse (n = 3, 2, and 2, respectively), and multiple organ failure (n = 1, 0, and 1, respectively).

Univariate analysis revealed that the trial intervention and platelet response were related to survival (Table 2). Because the intervention was strongly associated with response, we excluded platelet response from multivariate analysis for survival to avoid a conjugation interaction. As a result, the intervention was the only independent factor affecting survival for all patients.

Factors related to response and survival in patients in arms A and B were further analyzed (supplemental Table 1). We did not identify the predictor for response after analyzing the impact of the combination of rhTPO, diagnosis, age, sex, time of enrollment, ABO compatibility, conditioning, cell counts in graft, neutrophil recovery, and megakaryocytes before treatment. Similarly, none of these variates showed an independent impact on survival in Cox regression, except response to decitabine, which was associated with prolonged survival with marginal significance (HR, 2.316; 95% CI, 0.964-5.556; P = .060).

Percentages of EPCs and BMECs in bone marrow increased significantly in arms A (P = .001 and .006, respectively) and B (P = .024 and .045, respectively) compared with arm C (P = .635 and .869, respectively; supplemental Figure 1A). After decitabine treatment, secreted endothelin-1 (P = .004 and .011) and VEGF (P = .003 and .006) levels increased and endogenous NO level (P = .01 and .019) decreased significantly in patients in arms A and B, respectively (supplemental Figure 1B). Cytokines and chemokines reflecting megakaryocyte migration and adhesion, including SDF-1 (P = .003 and .01), CXCR4 (P = .014 and .04), and FGF-4 (P = .016 and .007) were also increased after treatment in both arms A and B, respectively (supplemental Figure 1C).

The main AEs were hematologic and related toxicities, including neutropenia, anemia, infection, and bleeding events (Table 3). Slight fluctuations in white blood cells and hemoglobin were observed in arm C, although no cytotoxic agents were administered to these patients. Greater myelosuppression was found in arms A and B, presenting as more severe and prolonged leukopenia and anemia, when compared with arm C, but the incidence of infection was similar among the 3 arms. Despite the increased platelet counts in arms A and B, the frequency of bleeding events was comparable. Other AEs included fatigue, decreased liver function, and chest pain, but few of these events were grade ≥3. Nonhematologic toxicities were not different among the 3 arms (P = .702).

Isolated thrombocytopenia is a frequent and severe complication of HCT that is associated with worse outcomes.⁴  Thrombocytopenia post-HCT has been attributed to the recurrence of underlying disease, treatment history, transplant donor, and transplantation complications, such as GVHD or infection.^4,5  However, the pathogenesis of thrombocytopenia in patients without an identified risk factor remains uncertain. Because of the lack of a consistent definition for prolonged isolated thrombocytopenia, published studies have enrolled diverse patients according to various criteria, which makes it difficult to evaluate incidence, treatment efficacy, and long-term outcomes.

After transplantation, relapse and transplantation complications such as GVHD, infection, or microangiopathy may lead to incomplete platelet recovery, which requires appropriate interventions to resolve the underlying conditions. For those without an identified cause, supportive treatments for increasing platelet counts are usually administered. Therefore, in our study, we enrolled patients with isolated thrombocytopenia after 60 days post-HCT to allow conventional treatments for thrombocytopenia to be effective so as to ensure that the recruited patients were refractory to the usual treatments. As recommended by the 2019 immune thrombocytopenia guideline of the American Society of Hematology, patients with platelet count <30 × 10⁹/L were enrolled, because these patients have a significantly higher risk of severe bleeding.¹⁹  Therefore, RPIT in our study was defined as platelet count <30 × 10⁹/L after 60 days post-HCT without an identified cause and refractoriness to conventional supportive treatments. Patients with RPIT deserve more intensive interventions, because both our data on patients in arm C and previously published data suggest that these patients have poor outcomes despite continuous administration of conventional treatments.^3,5 

Patients with RPIT after HCT had lower megakaryocyte counts and abnormal megakaryocyte maturation,²  which may be correlated with the damaged bone marrow microenvironment, including endothelial cells and cytokines, which are required to support megakaryocytopoiesis and platelet production. It has been reported that the administration of vascular disrupting agents could lead to prolonged and selective thrombocytopenia after chemotherapy, which continued even if other hematopoietic lineages were restored.²⁰  Damage to microvascular endothelial cells is unavoidable for HCT patients because of the radiochemotherapy conditioning regimen administered before transplantation as well as the potential immune disorders caused by alloreactivities.^9,21,22  Endothelial-associated adhesion molecules such as SDF-1 and CXCR4 have an essential role in regulating megakaryocyte migration and platelet release.^23,24  A previous study indicated that administration of SDF-1 and FGF-4 caused a threefold increase in platelet count as a result of enhanced interactions between megakaryocyte progenitors and the bone marrow vascular endothelium in thrombocytopenic, TPO-deficient mice.²⁵  In posttransplantation settings, impaired EPCs and related chemokines have been found in patients with poor graft function.¹⁰ 

Optimal treatment remains a challenge for patients with RPIT after HCT. Decitabine, a well-recognized hypomethylating agent, has been reported to induce megakaryocyte differentiation and maturation in vitro.¹²  In vivo, a 30% platelet count increase in mice after the injection of decitabine was observed, suggesting rapid platelet release from the bone marrow.¹²  Recently, a multicenter study in refractory immune thrombocytopenia patients demonstrated that low-dose decitabine (3.5 mg/m² IV for 3 consecutive days per cycle for 3 cycles, with a 4-week interval between cycles) improved the platelet count in 51% participants,²⁶  possibly by increasing the number of mature polyploid megakaryocytes and promoting platelet production. Moreover, there is evidence suggesting that hypomethylating agents may help alleviate endothelial damage¹³  and modulate the microenvironment in niche by upregulating CXCR4, the SDF-1 ligand on CD34⁺ cells.²⁷  In accordance with previous observations, we found that RPIT patients responding to decitabine had increased levels of EPCs, BMECs, and chemokines related to megakaryocyte migration, as well as improved megakaryocyte maturation status (data supplement).

rhTPO and TPO receptor agonists (eltrombopag and romiplostim) are alternative approaches to treat thrombocytopenia caused by various settings, including allogeneic HCT.^28-30  TPO has proved to be the principal physiologic cytokine for megakaryogenesis. Our previous study showed rhTPO after allogeneic HCT could promote platelet engraftment and significantly improve the prognosis of patients with myelodysplastic syndrome and aplastic anemia.²⁹  Because neither eltrombopag nor romiplostim were available in China when this trial began, we designed a combined regimen of decitabine and rhTPO (arm A) to investigate the potential synergistic effect. No benefit of this combination was apparent when compared with decitabine alone.

For patients with RPIT after HCT, it remains to be investigated whether additional decitabine may be beneficial. Delayed recovery of platelets was also observed in all 3 arms, although it is difficult to determine whether this was a delayed response to treatment (either decitabine or control) or spontaneous platelet recovery. Meanwhile, death during follow-up may have contributed to bias when analyzing this group of patients, because there was a competing risk between death and delayed recovery.

Superior survival was found in patients receiving decitabine of both arms A and B, which may be attributed to the improved bone marrow microenvironment that facilitates platelet recovery and immune reconstitution. Because of the poor platelet recovery in arm C, more patients died as a result of bleeding compared with patients in arms A and B. Although the difference was not significant, probably because of the small number of patients, this result suggests that the reduced survival in arm C can be mainly attributed to increased fatal hemorrhage, because other causes of death (eg, GVHD, infection, and relapse) were comparable among the 3 groups. Although the antitumor property of decitabine potentially contributed to the prolonged overall survival, relapse incidence was comparable among the 3 arms during long-term follow-up.

Limitations of our study include the small and diverse sample and heterogeneous treatment options in arm C. Although our data suggests that decitabine can increase megakaryocyte numbers and function, additional studies are required to accurately define the effect of decitabine on platelet production. Our data suggest that the benefit of decitabine can be attributed to the promotion of megakaryocyte maturation and migration, as well as the repair of endothelial cells. Additional studies are warranted to determine the role of decitabine regarding EPCs and megakaryocytes.

In conclusion, this multicenter randomized study demonstrates the efficacy and safety of decitabine for patients with RPIT after HCT, with improved response and prolonged survival.

The authors thank James George and Lijun Xia for their critical reading and comments, as well as Yu Hu at Wuhan Union Hospital (Wuhan, China), Xiaojun Huang at Peking University People’s Hospital (Beijing, China), and Ming Hou at Qilu Hospital of Shandong University (Jinan, China) for assistance with developing the clinical trial protocol.

This work was supported by grants from the National Natural Science Foundation of China (81873432, 81670132, 81730003, and 81700173), National Key R&D Program of China (2019YFC0840604), National Science and Technology Major Project (2017ZX09304021), Jiangsu Province of China (18KJA320006, SBE2016740635, BE2019798, and ZDRCA2016047), and Priority Academic Program Development of Jiangsu Higher Education Institutions.

Contribution: Y.H. and D.W. designed the study; Q.L., X.F.H., F.C., T.Y., X.M., X.W., S.H., X.C., X.H., J.H., and Y.L. recruited patients; Q.L., Y.H., D.W., and C.R. provided administrative support; Y.T., J.C., T.C., T.P., J.L., J.Q., and Y.S. collected the data and performed the analyses; Y.T., J.C., and Y.H. interpreted the data and wrote the manuscript; and all authors contributed to data interpretation, reviewed and provided their comments on the manuscript, and approved the final version.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Yue Han, Jiangsu Institute of Hematology, First Affiliated Hospital of Soochow University, 188 Shizi St, Suzhou, Jiangsu 215006, China; e-mail: hanyue@suda.edu.cn; and Depei Wu, Jiangsu Institute of Hematology, First Affiliated Hospital of Soochow University, 188 Shizi St, Suzhou, Jiangsu 215006, China; e-mail: wudepei@suda.edu.cn.