Differentiation therapy in poor risk myeloid malignancies: Results of companion phase II studies

Abstract

Pre-clinical data in non-M3 AML supports the use of differentiation therapy, but clinical activity has been limited. Myeloid growth factors can enhance anti-leukemic activity of differentiating agents in vitro. We conducted companion phase II trials investigating sargramostim (GM-CSF) 125 μg/m²/day plus 1) bexarotene (BEX) 300 mg/m²/day or 2) entinostat (ENT) 4–8 mg/m²/week in patients with MDS or relapsed/refractory AML. Primary endpoints were response after at least two treatment cycles and toxicity. 26 patients enrolled on the BEX trial had a median of 2 prior treatments and 24 enrolled on the ENT trial had a median of 1. Of 13 response-evaluable patients treated with BEX, the best response noted was hematologic improvement in neutrophils (HI-N) seen in 4 (31%) patients; none achieved complete (CR) or partial remission (PR). Of 10 treated with ENT, there was 1 (10%) partial remission (PR) and 2 (20%) with HI-N. The secondary endpoint responses of HI-N with each combination were accompanied by a numerical increase in ANC (BEX: 524 to 931 cells/mm³, p = 0.096; ENT: 578 to 1 137 cells/mm³, p = 0.15) without increasing marrow blasts. Shared grade 3–4 non-hematologic toxicities included febrile neutropenia, bone pain, fatigue, and dyspnea. GM-CSF plus either BEX or ENT are well tolerated in resistant and refractory MDS and AML and showed modest clinical and biologic activity, most commonly HI-N.

1. Introduction

Standard therapies for myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) continue to result in poor long-term survival, especially among the elderly and unfit [1,2]. Given that a major attribute of cancer is a block in differentiation [3], many groups have attempted to harness differentiation strategies in diverse tumor types [4–8], the most successful example being all-trans retinoid acid (ATRA) and arsenic trioxide in APL [9–12]. Differentiation strategies are generally well tolerated, with less toxicity than AML induction or allogeneic hematopoietic cell transplantation. Unfortunately, clinical activity of ATRA in non-APL AML has been modest at best [13–17]. GM-CSF enhances the differentiation activity of agents with differing mechanisms of action such as interferon-alpha, ATRA, arsenic trioxide, and bryostatin-1 [18–23]. We previously demonstrated that bryostatin-1, a protein kinase C agonist, in combination with myeloid growth factors markedly enhanced differentiation [19]. We subsequently studied bryostatin in combination with GM-CSF in a phase I trial in patients with poor risk myeloid malignancies and saw improvements in neutrophil counts and rare clinical responses, but severe myalgias and arthralgias limited dose escalation [22].

The combination of differentiation agents and myeloid growth factors may preferentially impact early leukemia progenitors while sparing normal hematopoiesis [18–20,24], a strategy that fits well into the treatment paradigms of MDS and AML, which are characterized by ineffective or blocked differentiation and bone marrow (BM) failure. Leukemic stem cells maintain the self-renewal potential of the malignant clone while retaining the capacity to undergo differentiation, and their self-renewing potential is lost when they differentiate [21,25]. We hypothesized that forced terminal differentiation via the synergistic combination of a differentiation agent and sargramostim (GM-CSF) would lead to clinical responses through elimination of leukemic progenitors.

Bexarotene (BEX) is a retinoid FDA-approved for relapsed and refractory cutaneous T-cell lymphoma (CTCL) that selectively binds retinoid X receptor, mediating expression of genes that control differentiation, and inhibiting tumor growth in vitro [26,27]. A phase I trial in non-M3 AML patients showed that CTCL dosing (300 mg/m²/day) was well-tolerated and anti-leukemic activity was observed in a subset of patients [26]. Leukemic blast differentiation was suggested by the presence of the leukemic cytogenetic abnormality in mature circulating granulocytes in three patients and differentiation syndrome in two [26]. Likewise, ENT, a histone deacetylase (HDAC) inhibitor, has been shown to have antiproliferative effects on many cancer cell types [28]. A phase I trial using single agent ENT in relapsed and refractory acute leukemia showed good tolerability and differentiation activity, in terms of increased acetylation of CD34+ blasts and several patients with increased mature granulocytes [29]. We therefore studied the synergy of both BEX and ENT with GM-CSF in vitro and then embarked on companion phase II trials combining BEX and ENT with GM-CSF in patients with advanced myeloid malignancies.

2. Materials/patients and methods

2.1. Cell culture

Human cell lines U937 and HL-60 were obtained from American Type Culture Collection (Manassas, VA) and maintained in RPMI 1640 (Life Technologies, Inc., Rockville, MD) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Life Technologies, Inc.), 50 units/ml penicillin, 50 μg/ml streptomycin, and 2 mM L-glutamine in a humidified atmosphere at 37 ◦C and 5.0% CO₂. Cells (10⁵ cell/ml) were treated with BEX (Ligand Pharmaceuticals, San Diego, California), ENT (Syndax Pharmaceuticals, Waltham, MA), and/or recombinant human GM-CSF (200 units/ml; Immunex, Seattle, WA). Following 96 h of treatment, cells were washed twice with media, and then evaluated for clonogenic growth as previously described [19]. Briefly, 1000 cells were plated in quadruplicate in 1.0 ml of 1.2% methylcellulose, 30% FBS, 1% bovine serum albumin (BSA), 10⁻⁴ M 2-mercaptoethanol, and 2 μM L-glutamine, and following 7 days tumor colonies with >40 cells were counted using an inverted microscope. Results are presented as the percentage of colonies relative to untreated controls ± standard error of the mean (SEM).

2.2. Flow cytometry

Following 120 h of incubation, cell lines were analyzed for differentiation based on expression of CD11b and CD14 by flow cytometry as described previously [19]. Briefly, cells were washed with phosphate-buffered saline containing 0.2% BSA and then stained with fluorescein isothiocyanate-conjugated mouse antihuman CD11b and phycoerythrin-conjugated mouse antihuman CD14 antibodies for 30 min at 4 °C (BD Pharmigen, San Diego, CA). Control studies were carried out using mouse IgG1 and IgG2a isotype antibodies. Analysis was performed on a FACSCalibur flow cytometer (BD, Mountain View, CA). Results are presented as relative mean fluorescence intensity ± SEM as described previously [19].

2.3. Patient selection

We conducted sequential clinical trials studying the combinations of BEX + GM-CSF and ENT + GM-CSF in patients with advanced myeloid malignancies at the Sidney Kimmel Comprehensive Cancer Center (SKCCC) at Johns Hopkins. Adult patients were required to have a diagnosis per 2001 WHO criteria of (1) MDS, (2) AML resistant to or relapsed after prior induction, or (3) newly-diagnosed AML and ineligible for intensive chemotherapy. All subtypes of AML were included, with the exception of acute promyelocytic leukemia. Exclusion criteria included pregnancy, breastfeeding, renal or hepatic dysfunction, and Eastern Cooperative Oncology Group performance status >2. All patients signed a Johns Hopkins SKCCC IRB-approved informed consent prior to enrollment.

2.4. Study design

Patients underwent a screening BM aspirate and biopsy to confirm their diagnosis. BEX and GM-CSF were administered in 4 week cycles. BEX 300 mg/m² was given orally with food daily and GM-CSF 125 μg/m² was given subcutaneously daily for 28 days. Hydroxyurea was permitted concurrently for white blood count (WBC) >15,000/μl, given concerns for differentiation syndrome. Subsequent cycles could be initiated no longer than 42 days from the start of the previous cycle. Patients could receive up to 6 cycles of therapy.

ENT and GM-CSF was administered in 6 week cycles. ENT was given orally days 1, 8, 15 and 22 and GM-CSF 125 μg/m² was self-administered subcutaneously daily for all 42 days, but held one week prior to study-specified marrow assessments. ENT was supplied through the Cancer Therapy Evaluation Program of the National Cancer Institute. Hydroxyurea was permitted to maintain a WBC <30,000/μl through the second dose of ENT during cycle 1. Subsequent cycles were initiated no sooner than 42 days from the start of the previous cycle and no longer than 56 days. ENT was originally started at the maximum tolerated dose (MTD) of 8 mg/m² based on a prior phase I study [29], but was lowered to 6 mg/m² and then 4 mg/m² due to toxicities. Patients were eligible for up to 6 cycles of therapy.

2.5. Response assessment

BM aspirate and biopsies were performed after cycles 1, 2, and every two subsequent cycles, between days 5 and 7 off therapy. Complete blood count and chemistries were followed twice weekly for the first month and weekly thereafter. For the BEX trial, fasting lipids, triglycerides, and thyroid function tests were evaluated every 2 weeks for the first month and every 8 weeks thereafter. Packed red blood cell and platelet transfusions were tracked weekly for both studies. Transfusion indications were individualized, but general targets were hematocrit >25–28% and platelets >10 × 10⁹/L. Response to treatment was assessed after two cycles, according to International Working Group (IWG) criteria [30,31]. Cytogenetic responses were monitored in patients with abnormalities at baseline. Toxicities were recorded using NCI Common Toxicity Criteria, Version 3.0.

2.6. Statistical considerations

Pre-clinical data were analyzed as two-tail Student’s t-tests to compare differences between drug treatment groups. The primary endpoint for both clinical trials was response per IWG criteria after 2 treatment cycles (complete remission (CR) + partial remission (PR)). Simon two-stage designs [32] were used with a two-sided α of 0.10 and ß of 0.10, assuming the null hypothesis response rate of 10% and alternative hypothesis response rate of 35%. Thus, we planned enrollment of 11 patients on the first stage, and if 2 or more responded, an additional 7 patients would be accrued for a total 18 evaluable patients. If 4 or more of 18 responded, the null hypothesis would be rejected. Secondary clinical endpoints included hematologic improvement (HI) per IWG criteria for MDS [31], transfusion requirements, and toxicity, with assessment of frequency and severity of adverse events for all patients who received drug. In addition, biologic activity was assessed through changes in peripheral blood and BM blast counts and cytogenetics. Clinical data were analyzed with dependent samples t-tests to compare baseline to post cycle 2. Prior to conducting the analysis, the assumption of normally distributed difference was examined. A non-parametric equivalent of dependent samples t-test (Wilcoxon signed-rank test) was used in the case when the assumption of normality was not met. Baseline counts were averaged over 2 weeks prior to treatment and post cycle 2 counts were averaged over 8 weeks after cycle 2. Survival functions were estimated by the Kaplan-Meier method, censoring from date of last follow-up. Analyses were done using SAS 9.4 (SAS Institute, Inc., Cary, North Carolina) at a significance of 0.05.

3. Results

3.1. Differentiation of AML cells in vitro

We treated U937 and HL60 cells with BEX and ENT with or without GM-CSF and found enhanced phenotypic differentiation as evidenced by increased expression of myeloid surface antigens CD11b and CD14 with the combinations (Fig. 1A and B). Terminal differentiation is associated with loss of clonogenic growth potential and both BEX and ENT limited colony formation in methylcellulose, with this effect being significantly enhanced by co-treatment with GM-CSF (Fig. 1C and D).

Fig. 1.

A Relative expression of CD11b and CD14 of U937 cells detected by flow cytometry following treatment with entinostat (ENT) with or without GM-CSF. B. Relative expression of CD11b by U937 or HL60 cells following treatment with bexarotene (BEX) with or without GM-CSF. C. Relative clonogenic growth of U937 or HL60 cells following treatment with ENT or BEX with or without GM-CSF. Values represent mean ± SEM of 3 experiments. *P < 0.05.

3.2. Patient demographics

Between July 6, 2007 and November 11, 2008, 24 patients were enrolled on the ENT + GM-CSF trial and between February 13, 2009 and January 6, 2011, 26 patients were enrolled on the BEX + GM-CSF trial (Table 1). Median ages for BEX and ENT trials were 72 (range: 49–89) and 71 (range: 52–84) years, respectively, and most patients had AML (61.5% BEX and 79.2% ENT). The BEX trial had 16 AML patients (13 out of MDS, 1 therapy-related, and 2 refractory de novo) and 10 patients had MDS. 17 patients on BEX (65%) had an abnormal karyotype. Median number of prior regimens was 2 (range, 0–5) and median ECOG PS 1 (range, 0–2). The ENT study had 19 patients with AML (7 arising from MDS, 3 therapy-related, 8 relapsed/refractory de novo, and 1 newly diagnosed) and 5 patients had MDS. 13 (54%) on ENT had an abnormal karyotype. Median prior therapies was 1 (range 0–5) and median ECOG PS 1 (range, 0–2).

Table 1.

Patient demographics and baseline characteristics.

Characteristic	Bexarotene n = 26	Entinostat n = 24
Age, years
Median	72	71
Range	49–89	52–84
Male, n (%)	19 (73)	15 (62.5)
Diagnosis, n (%)
AML	16 (61.5)	19 (79.2)
MDS	10 (38.5)	5 (20.8)
AML subgroups, n (%)
MDS/AML	13 (81.3)	7 (36.8)
t-AML	1 (6.3)	3 (15.8)
AML, other	2 (12.5)	9 (47.4)
Favorable cytogenetics	1 (6.3)	0 (10)
Intermediate cytogenetics	9 (56.3)	11 (57.9)
Poor cytogenetics	6 (37.5)	9 (47.4)
MDS subgroups, n (%)
IPSS Low	1 (10)	0 (0)
IPSS Int-1	5 (50)	4 (80)
IPSS Int-2	4 (40)	1 (20)
Favorable cytogenetics	5 (50)	5 (100)
Intermediate cytogenetics	3 (30)	0 (0)
Poor cytogenetics	2 (20)	0 (0)
Previous therapy, n (%)
None	1 (3.8)	5 (20.8)
1 therapy	10 (38.5)	13 (54.2)
2 or more therapies	15 (57.7)	6 (25)
ECOG PS, n (%)
0	8 (30.8)	10 (41.7)
1	17 (65.4)	11 (45.8)
2	1 (3.8)	3 (12.5)

Abbreviations: AML: acute myeloid leukemia; ECOG PS: Eastern Cooperative Oncology Group performance status; Int: intermediate; IPSS: International Prognostic Scoring System; MDS: myelodysplastic syndrome; t-AML: therapy-related AML.

3.3. Treatment administered

BEX + GM-CSF were given to 26 patients, with 13 (50%) completing 2 or more cycles (Fig. 2A). The most common reasons for treatment discontinuation prior to completing cycle 2 were progressive disease (PD), physician discretion, and elevated WBC, with the majority being unrelated to toxicity. Of note, only three total patients (only one who completed 2 cycles) required concurrent hydroxyurea for leukocytosis at doses of 500–1000 mg twice daily. ENT + GM-CSF were given to 24 patients, only one of whom received hydroxyurea at a dose of 1000 mg twice daily. 10 (42%) completed 2 or more cycles (Fig. 2B), none of whom received hydroxyurea. Patients came off trial early initially due to dose-related toxicity and no patients receiving 8 mg/m² ENT and only 1 of 4 patients treated with the 6 mg/m² dose completed 2 cycles. At the final treatment dose of 4 mg/m², PD and patient withdrawal of consent were the main reasons for treatment discontinuation.

Fig. 2.

Patients completing 2 or more cycles of therapy. Reasons for treatment discontinuation prior to cycle 2 on (A) bexarotene + GM-CSF trial and (B) entinostat + GM-CSF trial.

3.4. Adverse events

The ENT dose was modified after the originally planned dosage of 8 mg/m² resulted in toxicities in the first 4 patients, with 3 serious adverse events (SAEs) – hypoxia, pneumonitis, and death from progressive fungal pneumonia. Initially, the dose was reduced to 6 mg/m², but notable toxicities, including one death from renal failure and another from fungal pneumonia, resulted in further reduction to 4 mg/m² (Table 2), and this dose was well tolerated.

Table 2.

Dose levels of entinostat and select non-hematologic grade 3/4 toxicities.

Dose level (mg/m2 )	Patients per dose	Cycles per dose	Infectious (n)	Pain (n)	Other adverse events (n)
4	16	38	Neutropenic infection (3)	Bone (2)	Fatigue (1) Pericardial effusion (1) Weakness (1)
6	4	8	Neutropenic infection (1)	–	Anorexia (1) Fatigue (1) Hypoxia (1) Vasovagal episode (1)
8	4	5	Neutropenic infection (1)	Bone (1) Joint (1)	Cough (1) Hypoxia (1) Pneumonitis (1)

Grade 3–4 non-hematologic toxicities for both therapies were similar (Table 3, BEX%, ENT%), and included neutropenic infection (8%, 25%), dyspnea (15%, 4%), generalized muscle weakness (8%, 13%), bone pain (4%, 13%), and fatigue (4%, 13%). BEX combination had unique grade 3–4 toxicities of hypertriglyceridemia (8%), elevated ALT (12%), and elevated bilirubin (4%). ENT resulted in grade 3–4 hyponatremia (8%), hypoxia (8%), hyopcalcemia (4%), pericardial effusion (4%), and pneumonitis (4%).

Table 3.

All grade 3 and 4 non-hematologic adverse events.

Adverse event	Bexarotene (n = 26)	Entinostat (n= 24)
Neutropenic infection	2 (8%)	6 (25%)
Dyspnea	4 (15%)	1 (4%)
Elevated ALT	3 (12%)	-
Hypertriglyceridemia	2 (8%)	-
Muscle weakness (gen)	2 (8%)	3 (13%)
Bone pain	1 (4%)	3 (13%)
Fatigue	1 (4%)	3 (13%)
Confusion	1 (4%)	2 (8%)
Back pain	1 (4%)	1 (4%)
Hyponatremia	-	2 (8%)
Hypoxia	-	2 (8%)
Ankle tendonitis	1 (4%)	-
Atrial flutter/fibrillation	1 (4%)	-
Bilirubin	1 (4%)	-
Edema	1 (4%)	-
Flank pain	1 (4%)	-
Hyperglycemia	1 (4%)	-
Lethargy	1 (4%)	-
Respiratory failure	1 (4%)	-
Somnolence	1 (4%)	-
Sweet syndrome	1 (4%)	-
Ureteral obstruction	1 (4%)	-
Cough	-	1 (4%)
Anorexia	-	1 (4%)
Joint pain	-	1 (4%)
Hypocalcemia	-	1 (4%)
GI bleed	-	1 (4%)
Pericardial effusion	-	1 (4%)
Pneumonitis	-	1 (4%)
Vasovagal episode	-	1 (4%)

Abbreviations: ALT: alanine aminotransferase; gen: generalized.

3.5. Response and survival

After completing two cycles, none of the 13 patients on the BEX trial achieved the primary endpoint of CR or PR. However, the combination showed evidence of biologic activity as 4 (31%), 3 with MDS and 1 with MDS/AML, experienced HI-N (Table 4). In 2 of these patients, there was a >50% reduction in marrow blasts, one of which constituted a morphologic leukemia-free state (MLFS) with <5% blasts, suggesting differentiation of the malignant clone. Four additional patients (31%) had stable disease (SD), 2 of whom had a > 50% reduction in marrow blasts, one with a MLFS, without count recovery. Two MDS patients maintained SD after 6 cycles. Five patients (38%) had PD. Overall, there was a trend towards improved neutrophil count with mean ANC ± SEM of 524 ± 95 cells/mm³ at baseline and 931 ± 244 cells/mm³ after 2 cycles (p = 0.096), with no increase in marrow blasts (p = 0.48). In the 4 patients with HI-N, there was a significant improvement in neutrophil count from 616 ± 124 at baseline to 2074 ± 282 cells/mm³ after 2 cycles (p < 0.005) (Fig. 3), but no impact on platelet count, transfusion needs, or cytogenetics.

Table 4.

Clinical responses in patients completing 2 or more cycles of therapy.

Therapy	Dose level (mg/m2)	Clinical Response
		PR	HI	SD	PD
Bexarotene	300	0	4	4	5
Entinostat	4	1	2	4	2
	6	-	-	1	-
	8	-	-	-	-

Abbreviations: PR: partial remission; HI: hematologic improvement; SD: stable disease; PD: progressive disease.

Fig. 3.

Effect on counts in patients completing 2 or more cycles. (A) ANC, (B) platelet count, and (C) bone marrow blast count before and after 2 cycles of therapy on the bexarotene trial; (D)–(F) for the entinostat trial. Trends were evaluated in patients with HI-N or PR responses and those with SD or PD. Values represent mean ± SEM. *P < 0.005.

Of 10 patients on the ENT trial completing 2 cycles, 1 MDS patient achieved the primary endpoint of PR. Two further patients, both with AML, achieved the secondary endpoint of HI-N, of whom one also had HI in platelets, without improvement in marrow blasts, and the other achieved a MLFS. Five additional patients had SD, one of whom with MDS achieved a MLFS. Two patients had PD although one with AML initially had a CR with incomplete platelet recovery (CRp) after the first cycle, with improved ANC from 407 to 3883 cells/mm³. Two MDS patients remained with SD or better for 5 cycles; one maintained a PR after cycle 4, but returned to SD after cycle 5. Overall, there was a trend toward improvement in neutrophil counts from a mean ANC ± SEM 578 ± 126 cells/mm³ at baseline to a mean of 1137 ± 384 cells/mm³ after 2 cycles (p = 0.15), without an impact on marrow or peripheral blasts (Fig. 3). Furthermore, 2 AML patients who achieved HI-N normalized their abnormal karyotypes by marrow cytogenetics and FISH. The first patient had 2/20 cells with +Y, dup [6] (p21.3p25) at baseline, which reverted to normal after 2 cycles of therapy and remained normal after a third cycle. The other had baseline FISH abnormalities of +5 (30%), +7(22%), +8 (35.5%), +11 (34.5%), and +20 (24.5%), which resolved to normal after 2 cycles. Unfortunately, this patient was unable to continue onto cycle 3 due to interim complications.

The median overall survival (OS) of 13 evaluable patients on the BEX trial was 7.7 (range 3.1–80.5) months; 6.4 (range 3.1–16.6) months in 7 AML patients and 22.6 (6.1–80.5) months in 6 MDS patients. One patient with MDS who achieved HI-N after 2 cycles and remained on therapy for 6 months with SD is still living and maintained on transfusion support alone, after a brief course of cyclosporine in 2013. The median OS was 19.2 (range 5.6–89.2) months for the 10 evaluable patients on the ENT combination; 8.0 (range 5.6–9.6) months in 5 AML patients and 72.6 (range 28.9–89.2) months in 5 MDS patients. Two patients with MDS were living at the date of last contact. One had SD until the 6th cycle of therapy and then progressed, and eventually developed AML, but on last follow-up was in remission over 3 years after DLI. The second patient had PD after 2 cycles of therapy, but remains on transfusion support alone. Finally, the numbers of patients on each trial were small and differences in OS were not observed among patients with HI-N or better responses versus non-responders.

4. Discussion

The addition of GM-CSF to BEX or ENT increases biologic activity in vitro, and in vivo, the combination therapies lead to improved neutrophil counts and a reduction in marrow blasts in some patients with resistant disease. Treatment of patients with relapsed AML and MDS is challenging, as both the disease and traditional treatments can worsen the marrow failure. This becomes a particularly precarious situation in elderly patients [1,2]. An advantage of differentiation agents is their potential to improve marrow function with less toxicity when compared to standard cytotoxic treatments.

BEX + GM-CSF were well tolerated with the best response of HI-N seen in 31% of patients after two cycles of therapy and an overall trend toward higher ANC. The lack of increase in marrow blast count is suggestive of differentiation, as opposed to growth factors purely promoting normal hematopoiesis. Moreover, half of the patients with HI-N had excess blasts at baseline and experienced a BM blast response, supporting the combination’s activity against resistant leukemia. One MDS/AML patient with complex cytogenetics died 3 days after receiving one dose of BEX of presumed septic shock and multi-organ failure, but had features of differentiation syndrome, with an increasing WBC, coagulopathy, and diffuse pulmonary infiltrates. These clinical features are non-specific and no other patients experienced them, but this could reflect the differentiation potential of the therapy. There were no CRs or PRs given the lack of trilineage count recovery, but 62% had at least SD after 2 cycles, and two patients remained with SD following cycle 6.

ENT + GM-CSF also were well-tolerated at the final ENT dose of 4 mg/m², and similar to BEX, resulted in clinical improvement in select patients. Of patients able to complete two cycles, one had a PR that was sustained following 4 cycles and two had HI-N, one with multi-lineage improvement, and the other with a MLFS. The patient with cytogenetically normal AML who achieved the MLFS had suggestions of differentiation on his post-cycle 2 BM, with young myeloid cells present, but considerably improved maturation and 1–2% blasts by flow cytometry (from 10–15% at baseline), which were unable to be distinguished from reactive blasts. Furthermore, 2 patients with AML had cytogenetic responses, raising the possibility that the differentiation treatment was capable of eliminating malignant clones in the BM. Despite no sustained CRs following 2 cycles of therapy, 80% had at least SD. Two remained on study for 6 cycles given minimal adverse effects and an additional 3 patients remained on therapy for 3–4 cycles.

It is hard to assess the differentiating activity of such approaches, as the kinetics of differentiation and the eventual elimination of the malignant clone are not well understood. We know that with ATRA and arsenic trioxide in APL, it takes months to achieve complete molecular responses [9,33]. It is possible that assessing for response after two cycles in highly refractory and relapsed patients may not be enough time for the effects of differentiation to be seen clinically. However, the improvement in marrow blasts, with complete marrow responses in some patients, and increasing peripheral blood neutrophils, is encouraging. Per results presented by Silverman et al. at ASCO 2015, blast response/stabilization in the absence of hematologic improvement may be associated with improved survival in MDS patients [34]. The combinations could thus have a role in either palliation for older adults unfit for intensive chemotherapy, as adjunctive treatments with chemotherapy, or as maintenance therapy.

There are several limitations of our studies to consider. First, given that many patients were unable to complete at least 2 cycles of therapy for reasons including older age, advanced disease, and medical co-morbidities, the clinical effects of the combinations may be underestimated. Based on the Simon two-stage designs, enrollment on both trials ceased after 13 (BEX) and 10 (ENT) patients completed 2 cycles. Next, ENT was only given for 4 weeks out of each 6 week cycle, and enhanced responses may have been seen with continuous dosing. Furthermore, assays to prove that the combinations result in terminal differentiation are limited and improvements seen in neutrophil counts in the setting of stable marrow blasts do not definitively confirm the mechanism of action. Assessment of impact on leukemia stem cells (LSC) is difficult, but emerging techniques have been developed to isolate LSC populations [35], and these could be utilized in future trials to determine elimination of leukemia progenitors.

Given the biologic activity but limited sustained clinical activity of differentiation approaches, strategies to improve these treatments should be entertained. One possible direction is to combine them with other treatments that might enhance their differentiating activity, such as demethylating agents. There are several studies suggesting synergy between retinoids and epigenetic agents [36–38]. It is also possible that BEX and ENT did not reach their target due to mechanisms of drug resistance in the BM microenvironment. Our lab demonstrated that expression of cytochrome P450 enzymes in BM stromal cells, including CYP3A4, which metabolizes BEX [39], and CYP26, which metabolizes ATRA [40], serves as a key mechanism for drug resistance in AML [41]. While ATRA is highly efficacious in the differentiation and elimination of AML cells in vitro, its effect is eliminated in stromal co-culture, and restored when combined with a CYP26 inhibitor [42]. Our next step moving forward will be to study CYP resistant retinoids in AML to see if we can garner enhanced responses.

These companion phase II studies demonstrate modest clinical and biologic activity, including many patients with SD, improvements in ANC, and lower BM blast counts. HI-N was seen in patients without evidence of increased marrow or peripheral blood blasts; in fact, several showed a decrease in marrow blast percentage, supporting differentiation as a possible mechanism of action of the combinations. Both therapies were well tolerated in our patient population. Differentiation approaches in AML deserve further study as frontline therapy in combination with chemotherapy, palliation in older and unfit adults, and as post-remission maintenance.