Decitabine Induction with Myeloablative Conditioning and Allogeneic Hematopoietic Stem Cell Transplantation in High-Risk Patients with Myeloid Malignancies Is Associated with a High Rate of Infectious Complications

Abstract

Patients with high-risk myelodysplastic syndrome or acute myeloid leukemia have an increased risk of death following allogeneic hematopoietic stem cell transplantation (allo-HSCT). Decitabine has minimal non-hematologic toxicity and proven efficacy in myeloid diseases, and post-transplant cyclophosphamide (PTCy) has reduced rates of graft-versus-host-disease (GVHD). We hypothesized that decitabine induction with allo-HSCT and PTCy would improve outcomes in a high-risk myeloid disease population. We performed a phase-II trial of decitabine at 20 mg/m² for 10 days followed by allo-HSCT using a myeloablative regimen of fludarabine, IV busulfan and 4 Gy total body irradiation with PTCy for GVHD prophylaxis. Twenty patients underwent decitabine induction and 17 patients proceeded to transplant per protocol. Median overall survival from decitabine induction was 210 days (95% CI 122-not reached). All patients developed grade 4 neutropenia after decitabine, eleven patients (55%) developed grade 3–4 infections, and 5 cases were fatal. There were 5/20 (25%) long-term survivors with a median follow-up of 3.6 years. Decitabine induction followed by myeloablative allo-HSCT in a high-risk population was associated with a high risk of infection and mortality related to enhanced immunosuppression. Further exploration of decitabine conditioning on reduced intensity platforms and improved infectious prophylaxis and screening may better mitigate toxicity (ClinicalTrials.gov (NCT01707004)).

1.0. INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is the only curative therapy for myelodysplastic syndrome (MDS) and high-risk acute myeloid leukemia (AML). Patients with high-risk disease as defined by the disease risk index (DRI) have a markedly increased risk of relapse and death following transplant.^{1, 2} Although the majority of deaths in high-risk disease are due to disease progression, infection and organ dysfunction are also notable contributors, which highlights the cumulative toxicity and immune-suppressed state present in many high-risk patients at transplant.^{1, 2}

These issues highlight the importance of improved allo-HSCT platforms designed to reduce relapse without contributing to transplant-related morbidity. Decitabine is a methyltransferase inhibitor with demonstrated efficacy in myeloid disease.^{3, 4} Clinical studies have identified that intravenous administration of decitabine at a dose of 20 mg/m² over ten days is associated with complete response (CR) rates of up to 47% after a median of 3 cycles in AML patients, some entering a CR after only a single cycle.⁵ Additionally, phase 1 reports using a single cycle of decitabine prior to chemotherapy in relapsed AML (decitabine priming) demonstrated that this approach was safe and may even sensitize cancer cells to chemotherapy.^6–8 These studies identified minimal non-hematologic toxicity for decitabine, suggesting that pairing decitabine with allo-HSCT could translate into improved disease-related outcomes without additional toxicity.

Unfortunately, patients who remain disease-free after allo-HSCT are at risk of severe morbidity and mortality from graft-versus-host-disease (GVHD). Strategies targeting allo-reactive T-lymphocytes with post-transplant cyclophosphamide (PTCy) have resulted in low rates of grade 3–4 acute or chronic GVHD and have expanded the donor pool to include haplo-identical family members.^{9, 10}

We hypothesized that the combination of a single cycle of decitabine induction prior to transplant and PTCy would be safe and result in improved disease control, translating into improved survival in a high-risk cohort of myeloid malignancies.

2.0. METHODS

2.1. Study design and patients

This was a prospective, phase II, single-arm study registered at ClinicalTrials.gov (NCT01707004) prior to patient enrollment. We evaluated decitabine induction followed by allo-HSCT using myeloablative conditioning and PTCy for GVHD prophylaxis in 20 patients. Study accrual occurred from 2013–2017. This study was approved by our institution’s IRB prior to protocol execution. Eligible patients included adults aged 18–75 with either high risk MDS or AML according to the following definitions as well as adequate performance status and organ function. Eligibility for those with MDS required a World Health Organization Classification-based Prognostic Scoring System (WPSS) of >3 and/or transfusion-dependent or progressive disease despite hypomethylating agent therapy.¹¹ High-risk AML included primary induction failure, relapsed disease with or without remission, first CR and with high-risk features (therapy-related or secondary AML arising from a myeloproliferative disease, high-risk cytogenetics, or untreated secondary AML). All patients provided informed consent prior to any intervention on study. Patients were excluded for active CNS leukemia, new uncontrolled infection, or allergy to the study drugs.

2.2. Study Intervention

The study schema is depicted in Figure 1. Decitabine was administered at 20 mg/m² IV daily for ten days as a single cycle, starting between 17–24 days prior to the conditioning regimen. Fludarabine was administered at 50 mg/m² IV daily over 30 minutes day −5 to −2 before transplant as part of the preparative conditioning regimen. Busulfan was administered at 3.2 mg/kg IV daily over 3 hours on day −5 to −2 with dosing modified by pharmacokinetic modeling to achieve a steady state concentration of 800 ng/mL for patients aged 18–60, or dosed at 2.4 mg/m² IV daily over 3 hours day −5 to −2 with dosing modified to achieve a steady state concentration of 600 ng/mL for patients aged 60–75. Total body irradiation (TBI) was given at 4 Gy on day −1 followed by the infusion of unmanipulated bone marrow on day 0. GVHD prophylaxis consisted of cyclophosphamide administered at 50 mg/kg IV on days +3 and +4 for all patients. Haplo-identical or mismatched recipients also received tacrolimus administered at 0.12 mg/kg/day orally divided into 2 doses adjusted to maintain a serum level of 5–15 ng/mL given on days +5-+180, and mycophenolate mofetil (MMF) administered at 15 mg/kg orally three times a day given on days +5-+35.

Figure 1: Treatment Schema.

Treatment doses: Decitabine: 20 mg/m² IV x 10 days for 1 cycle, Fludarabine: 50 mg/m²/day, Busulfan: 3.2 mg/kg/day age ≤60, Busulfan: 2.4 mg/kg/day age >60. Steady-state busulfan target levels listed above. Flu: fludarabine CTX: cyclophosphamide MMF: mycophenolate mofetil. TBI: total body irradiation.

2.3. Supportive Care

All subjects underwent routine infectious screening methods including viral serology testing and CT imaging within 30 days of transplant. Subjects underwent standard infection prophylaxis including granulocyte colony-stimulating factor (G-CSF) and levofloxacin for neutropenic prophylaxis, beginning after conditioning for an absolute neutrophil count <500 cells/uL. Additional prophylaxis included fluconazole or an extended spectrum azole, acyclovir, and trimethoprim/sulfamethoxazole after engraftment. Cytomegalovirus (CMV) screening occurred weekly following transplant through day +100, and those with reactivation received valganciclovir until clearance.

2.4. Statistical Considerations

Overall survival (OS), progression-free survival (PFS), times to acute GVHD (aGvHD) chronic GVHD (cGVHD), and time to neutrophil recovery after transplant were analyzed using the Kaplan-Meier (KM) method. Day-100 OS and GVHD rates and median time to neutrophil recovery were estimated along with 95% confidence intervals. OS from start of decitabine was also analyzed for all patients enrolled. Competing risk methods were used to analyze non-relapse mortality (NRM), where the risk of NRM was quantified using a cumulative incidence function that accounted for one competing event, death due to disease. Time was calculated from date of transplant until death. Patients who were alive at last follow-up were censored. Competing risk methods were also used to analyze platelet recovery, where the cumulative incidence function accounted for one competing event, death without platelet recovery. Time was calculated from date of transplant until platelet recovery. Patients who had no platelet recovery at last follow-up were censored. Binary outcome variables such as grade 3 or worse non-hematologic toxicity were summarized with a proportion. Statistical analysis was performed using R 3.5.1, including the “survival”, “cmprsk”, “survminer” and “Hmisc” packages.¹²

Planned enrollment included 20 patients. With 20 patients, the trial had 0.71 power to detect the day-100 OS probability of ≥0.7 according to a one-tailed test of the null hypothesis that it is ≤0.5 at significance level of 0.1. Because of safety concerns, day-100 OS, day-100 aGVHD grade 3–4, and day-30 engraftment failure were monitored continuously using sequential probability ratio tests of the hypotheses noted above for these outcome measures.

3.0. RESULTS

3.1. Patient Distribution and Characteristics

A diagram of patient distribution throughout the study is depicted in Figure 2. All 20 patients enrolled received decitabine. Seventeen of 20 (85%) proceeded to transplant on study; two patients did not receive a transplant due to development of infection after receiving decitabine (fungal pneumonia, respiratory syncytial virus) precluding transplantation and later died of complications of fungal pneumonia and progressive leukemia, respectively. One other patient was removed from study for an active infection and later underwent a transplant off-protocol.

Figure 2: Subject Flowchart.

Study flowchart for patient outcomes and reporting.

Baseline patient characteristics are displayed in Table 1. Fifteen of twenty enrolled patients (75%) had AML, and five patients (25%) had MDS. The majority of patients had active disease at time of enrollment and carried a high or very high-risk Disease Risk Index score (18/20, 90%) The cohort was elderly with a median age of 64 and also featured significant comorbidity with many (8/20, 40%) patients having a hematopoietic cell transplantation-specific comorbidity index (HCT-CI) of 3+ or more. Ten (50%) patients had prior exposure to a hypomethylating agent. The majority of subjects (13/20, 65%) were significantly neutropenic (ANC <1000 cells/uL) prior to decitabine exposure with a low mean absolute neutrophil count for the cohort (962 +/− 1321 cells/uL, mean +/− standard deviation). Twelve of 17 (71%) received a haplo-identical donor transplant with the remainder using a matched-related donor (5/17, 29%).

Table 1:

Baseline Demographics

Baseline Demographics	Total (%)
N	20
Age in years at Transplant
Median Age (range)	64 (29–73)
Race
White	20 (100)
Non-white	0 (0)
Gender
Male	14 (70)
Female	6 (30)
Donor type*
Matched Related	5 (28)
Haplo-identical	13 (72)
Disease Risk by DRI
Low Risk	0 (0)
Intermediate Risk	2 (10)
High Risk	13 (60)
Very High	5 (25)
HCT-CI Score
0	5 (28)
1–2	6 (33)
3+	7 (39)
Disease Status
AML	15 (75)
Primary Induction Failure	6 (30)
Relapsed Disease	7 (35)
CR1 with high risk features	2 (10)
MDS	5 (25)
Prior HMA Exposure	10 (50)
ANC (cell/uL) Pre-Decitabine**	962 (1321)
ANC (cell/uL) Pre-Transplant**	121 (208)
IgG (g/dL)**	917 (274)

^*For those undergoing transplant (n=17). AML: acute myeloid leukemia; ANC: absolute neutrophil count; CR1: first complete response; DRI: disease risk index; HCT-CI: hematopoietic cell transplantation-comorbidity index;. HMA: hypomethylating agents; IgG: immunoglobulin G; MDS: myelodysplastic syndrome.

^**Mean (standard deviation)

3.2. Engraftment, Immune Reconstitution, and Regimen Toxicities

All 17 patients who underwent transplant on study demonstrated engraftment of neutrophil and platelet compartments with a median time to neutrophil recovery of 16 days (95% confidence interval (CI) 16–19). The cumulative incidence of platelet recovery at day 30 was 58.8% (95% CI 31–78%). Decitabine had a profound and significant effect on neutrophil counts prior to conditioning chemotherapy with all subjects (17/17, 100%) undergoing transplant displaying average absolute neutrophil counts (ANC) below 500 (post-decitabine, pre-conditioning) and a mean ANC of 121 (+/− 208) cells/uL compared to a mean pre-decitabine level of 962 (p = 0.007). Lymphocyte populations were also notably deficient post-transplantation although no pre-transplant data were collected for comparison. At day 30, mean CD4 lymphocyte populations were 68 (+/− 70) cells/uL and improved to 168 (+/− 122) cells/uL by day 100 for patients surviving to this timepoint. CD8 lymphocyte populations were similarly low at day 30 (410 +/− 893 cells/uL) and declined further by day 100 (259 +/− 299 cells/uL). IgG levels declined from pre-transplant level of 962 (+/− 274) to 504 (+/− 212) mg/dL by day 100.

A summary of grade 3–5 non-hematologic adverse events is provided in Table 2. Infectious complications were common throughout the study, with a majority of patients experiencing a grade 3–5 infectious complication ranging from culture-negative neutropenic fever to pneumonia and sepsis as well as five infection-related deaths, reflecting the marked immune suppression observed. As described above, three patients were removed from study prior to transplant due to infectious complications (neutropenic fever, respiratory syncytial virus infection, fungal pneumonia) developing after decitabine and only one subject went on to receive transplant off-protocol. Other transplant-related toxicities were observed including hypertension (10/20, 50%), mucositis (12/20, 60%), diarrhea (5/20, 25%), and elevated bilirubin (3/20, 15%).

Table 2:

Cause of Death and Grade 3–5 Adverse Events

Type	Number Affected (%)
Cause of Death (n=15)
Transplant-related	6/15 (40)
Infection	5/15 (33)
CMV-related	2/15 (13)
Mold-related Pneumonia	2/15 (13)
HHV-6 dissemination	1/15 (7)
Organ dysfunction	1/15 (7)
Relapse/Refractory Disease	9/15 (60)
Adverse Events (n=20)
Infection
G3–4	11 (55)
G5	5 (25)
Mucositis (G3–4)	12 (60)
Hypertension (G3–4)	10 (50)
Hypokalemia (G3–4)	9 (45)
Febrile Neutropenia (G3–4)	7 (35)
Hypophosphatemia	7 (35)
Hypoalbuminemia	6 (30)
Anorexia	6 (30)
Diarrhea (G3–4)	5 (25)
Fatigue	5 (25)
Elevated bilirubin (G3–4)	3 (15)
Weight Loss	3 (15)
Hypoxia	3 (15)
Dehydration	3 (15)
Rash	3 (15)

Adverse events occurring above 10% are depicted above. CMV: cytomegalovirus. Mold-related pneumonia includes 1 case of mucor and 1 case of aspergillus pneumonia. HHV-6: human herpes virus 6. Adverse effects compiled for all events occurring in at least 2 patients, excluding expected hematologic toxicities occurring with transplant. Grading according to NCI Common Terminology Criteria for Adverse Events version 4.0. G3–4: grade 3 or 4 toxicity, G5: grade 5 toxicity.

3.3. Acute and Chronic GVHD

All patients surviving beyond day 30 underwent assessment for acute GVHD. The day-100 cumulative incidence of grade 2–4 aGVHD was 43.5% (95% CI 18–66%). There were four cases of grade 3–4 aGVHD with a day 100 cumulative incidence of 27.8% (95% CI 8 – 52%). All aGVHD cases developed in either the skin, gut, or oropharynx. There were three cases of chronic GVHD, with a cumulative incidence at 1 year of 40% (95% CI 7–74.0%).

3.4. Disease Response and Cause of Death

There were 17 evaluable patients with bone marrow blast counts before and after decitabine, and there were no responses according to the International Working Group criteria for response assessment.¹³ Following allo-HSCT, fourteen patients developed a complete response at day 30 (14/17, 82%). Overall survival results are depicted in Figure 3. Of the 17 patients transplanted on protocol, the OS at day 100 after transplant was 65% (95% CI 46–92%). The median OS after start of decitabine for all 20 patients was 210 days (95% CI 122-not reached). The cumulative incidence of non-relapse mortality at day 100 was 24% (figure 3). Median overall survival for haplo- and MRD recipients was 190 and 87 days (p=0.62) respectively, and not statistically significant. There were 5/20 (25%) long-term survivors who remain disease-free at last follow-up with a median follow-up of 3.6 years. Rates of GVHD in this group of long-term survivors were low with 1/5 developing moderate grade cGVHD of the skin, oropharynx, and liver.

Figure 3: Overall and Non-Relapse Survival.

A: Overall survival (OS) from time of transplant for those undergoing transplant on protocol (n=17). B: OS from initiation of decitabine for all patients enrolled (n=20). Shaded area represents the 95% confidence interval for the K-M estimates. C: Incidence of non-relapse mortality using death from disease as a competing risk from day of transplant (n=17 evaluable).

Ultimately, there were 15 deaths on study out of the 20 subjects enrolled. Of the 15 deaths, six were deemed transplant-related and five of these were secondary to infection. Of the five deaths related to infectious complications, two were related to CMV, with one patient developing disseminated CMV with respiratory complications and one patient developing CMV colitis leading to Escherichia coli bacteremia and death. Two separate patients died secondary to fungal pneumonia (mucor and likely aspergillosis). The remaining patient developed a disseminated Human Herpes Virus 6 infection complicated by profound encephalopathy resulting in death. Separately, one patient died after developing respiratory failure of unclear etiology (presumed GVHD vs undiagnosed infection) on post-transplant day 81. All other deaths (9/15) were related to disease relapse or progression.

4.0. DISCUSSION

In this study, we hypothesized that including a 10-day treatment course of decitabine prior to a myeloablative HSCT would be well tolerated and improve subsequent patient outcomes by reducing relapse risk. We used a decitabine induction approach based on evidence that decitabine “priming” may sensitize cancer cells to subsequent chemotherapy and noted that significant responses can occur with even 1 cycle.^{1, 6–8, 14} We predicted that decitabine given as a single course would carry little non-hematologic toxicity, would not accentuate toxicity of myeloablative conditioning, and might improve disease control. We made the decision to proceed to HSCT contingent on at least stable disease because of the possibility of a priming effect that would accentuate the response to the preparative regimen. We used a myeloablative regimen of fludarabine and busulfan with an additional 4 Gy TBI, based on the favorable impact of myeloablative conditioning on AML and noting that a low-dose TBI strategy did not confer additional toxicity and could enhance the efficacy of the regimen.¹⁵

We successfully enrolled an elderly and comorbid population with a median age of 64 and 40% scoring 3+ by HCT-CI. We also captured a high-risk disease population by multiple measures including the presence of active disease (18/20, 90%) and high-risk DRI (18/20 90%). This study provides important outcome information as this population is frequently encountered in practice but often under-represented in clinical trials of AML that commonly enroll younger and healthier patients with better disease control.^16–19

Ultimately, our study did not meet our planned primary endpoint of day +100 OS >70% for those transplanted, likely due to excessive toxicity resulting from the addition of the 10-day course of decitabine. We also used a myeloablative regimen that carried significant toxicity for a group of heavily pre-treated patients with significant comorbidities. Patients over age 60 years did receive a 25% busulfan-dose reduction, but this did not appear to be sufficient to avoid significant toxicity. We observed a higher than expected incidence of NRM (24%) which was primarily related to a high rate of infection. Importantly, we observed a very high degree of grade 4 neutropenia after only decitabine exposure, which is much greater than reported in decitabine studies for AML.^3–5 Furthermore, multiple mechanisms of decitabine-mediated immunesuppression exist beyond neutropenia, and a high risk of infectious complications was reported by Ali et al. in a cohort analysis of patients with AML or MDS receiving a 10-day decitabine schedule.^{20, 21} Our data further supports an association between a 10-day decitabine schedule and infections in a high-risk myeloid disease population as 3 patients (15%) were unable proceed to transplant due to infection. Two of these patients never received a transplant and the third patient was taken off study but did eventually undergo HSCT at a later date.

Delayed immune reconstitution with low CD4 or CD8 lymphocyte subsets is associated with increased NRM.²² Here, we note profound effects on immune reconstitution where CD4 and CD8 lymphocyte subsets were much lower than previously reported for myeloablative conditioning but consistent with a prior report following PTCy.^{23, 24} Furthermore, a recent large retrospective study identified higher rates of CMV and invasive fungal infection in patients treated with PTCy.²⁵ In our study, nearly all of the infectious complications including death occurred prior to day +100, which may indicate a cumulative effect between decitabine, myeloablative conditioning, and PTCy above what has been reported previously for each of these regimens. These findings suggest that improvements to infectious screening or antibiotic prophylaxis could help mitigate the increased infectious risk. In our study, all patients were screened for occult infection with CT scans of the chest, abdomen, pelvis, and sinuses within 30 days prior to transplant per program standards. Importantly, although these scans were performed within the 30 day window, nearly all of them were performed prior to decitabine exposure, suggesting that occult infection could have developed following decitabine exposure that would not have been identified. Additionally, serum measurement of galactomannan and 1,3 beta-D-glucan for early identification of invasive fungal infection has been tested in allo-HSCT recipients with high negative predictive values (≥99%) and modest positive predicitive values (~65%) for both assays.^{26, 27} For patients at a higher risk of infectious complications after allo-HSCT, particularly those receiving conditioning augmented with decitabine, these approaches could certainly be considered recognizing the absence of robust evidence from a randomized controlled trial. Such an approach could also assist with tailoring anti-mold prophylaxis or treatment to those with identified infections or at high risk, optimize care based on local infection patterns, and would be in line with infectious disease guidelines on the use of anti-fungal and anti-mold prophylaxis for allo-HSCT recipients.²⁸

Importantly, the main cause of death remained relapsed disease, suggesting that further attempts to refine this platform are necessary.²⁹ Decitabine priming is becoming an increasingly tested strategy, particularly in MDS.³⁰ Recently reported outcomes of decitabine and myeloablative conditioning in a retrospective Chinese cohort demonstrated low rates of NRM (12%) and favorable survival, although this population was younger and featured lower-risk disease.³¹ Importantly, their approach used 5 days of decitabine versus 10 days and had fewer infectious episodes, suggesting that alternative dosing of decitabine may help reduce toxicity. In addition, a 5-day decitabine schedule combined with cytarabine followed by umbilical cord blood infusion as a consolidative approach demonstrated a low rate of infection and high rate of overall survival at 2 years (68%), suggesting another safe and potentially efficacious alternative to matching a 10-day decitabine schedule to myeloablative conditioning.³²

Given the high rate of NRM and infectious complications using our decitabine and a myeloablative conditioning (MAC) platform, it is possible that an alternative therapy on a reduced intensity platform may be able to augment the disease response to standard reduced-intensity conditioning (RIC) approaches without excess risk of toxicity. One example by Baker et al. features 5-day decitabine paired with a RIC regimen of fludarabine and busulfan for myeloid disease and demonstrated both lower rates of infection and a higher rate of overall survival (94%) by day +100 post-transplant.³³ Sequential conditioning with thiotepa, etoposide, and cyclophosphamide followed by a RIC regimen using thymoglobulin (TEC-RIC) was reported to be safe, reporting reduced infection rates and non-transplant mortality with improved efficacy in a similar population of high-risk disease and comorbidities.³⁴ Although this study cannot be adequately compared to our own, our cohort did feature an older population and we are reporting a prospective study as opposed to the retrospective analysis applied for TEC-RIC. Additionally, sequential conditioning with fludarabine, amsacrine, and cytarabine followed by a RIC regimen (FLAMSA) has also been utilized for a similar population with favorable outcomes.^{35, 36} A recent meta-analysis of FLAMSA conditioning incorporating 12 studies and over 2,300 patients similarly reported lower rates of 1 year NRM (~18%) and higher 1 year OS rates (~59%) compared to our study, although, once again, our cohort tended to be older and also featured much higher rates of active disease at transplant.³⁶

Our study is limited by its non-randomized single center approach featuring a limited sample of 20 patients. These limitations restrict our interpretation to a well-defined cohort of patients with myeloid diseases and precludes meaningful sub-group analysis.

4.1. Conclusion

In conclusion, the use of decitabine induction and immediate myeloablative conditioning with PTCy in an elderly and comorbid population with high-risk myeloid disease was associated with significant immunosuppression, a high risk of morbidity and mortality from infectious complications, and failed to suggest any improvement in disease control. Future studies incorporating decitabine into the conditioning regimen should recognize the increased risk of infectious complications and consider earlier use of antibiotic prophylaxis, increased screening for occult infection, alternative decitabine dosing, or reduced-intensity conditioning approaches.

Highlights.

• Decitabine plus myeloablative allogeneic transplant was studied in myeloid cancers

• Phase II clinical trial did not meet hypothesized improvement in overall survival

• Decitabine and myeloablative combo led to high rates of infection and mortality

• High rates of neutropenia and immunosuppression was the likely cause of infections