A Phase I Trial Evaluating the Effects of Plerixafor, G-CSF, and Azacitidine for the Treatment of Myelodysplastic Syndromes

Abstract

Interactions between the bone marrow microenvironment and MDS tumor clones play a role in pathogenesis and response to treatment. We hypothesized G-CSF and plerixafor may enhance sensitivity to azacitidine in MDS. 28 patients with MDS were treated with plerixafor, G-CSF and azacitidine with a standard 3+3 design. Subjects received G-CSF 10 mcg/kg D1–D8, plerixafor D4–D8, and azacitidine 75 mg/m2 D4–D8, but the trial was amended to reduce G-CSF dose to 5 mcg/kg for 5 days after 2 patients had significant leukocytosis. Plerixafor was dose escalated to 560mcg/kg/day without dose limiting toxicity. Two complete responses and 6 marrow responses were seen for an overall response rate (ORR) of 36% in evaluable patients, and ORR of 53% in patients receiving the triplet. Evidence of mobilization correlated with a higher ORR, 60% vs 17%. Plerixafor, G-CSF and azacitidine appears tolerable when given over 5 days and has encouraging response rates.

INTRODUCTION

Myelodysplastic syndromes (MDS) represent a spectrum of clonal hematopoietic stem cell disorders characterized by ineffective hematopoiesis, malignant expansion of clonal myeloid cells, and progressive bone marrow failure.[1] Azacitidine can prolong survival and delay progression of leukemia in intermediate-2 and high risk MDS. However, it is not curative and only achieves remission in approximately 20–30% of patients, with a median duration of response of 8–10 months.[2, 3]

Normal hematopoietic stem cells (HSCs) migrate to and are anchored in the perivascular niche of the bone marrow by chemotactic signals, including CXCL12 (stem cell derived factor-1, or SDF-1), which is produced by cells in the marrow stroma and binds to its receptor, chemokine receptor 4 (CXCR4) on the HSC surface.[4] Leukemic stem cells also use CXCL12- and CXCR4- adhesion molecule-dependent mechanisms to home to this perivascular niche where they are maintained in a quiescent, protected state.[5, 6, 7, 8] The bone marrow niche has been implicated in leukemogenesis by upregulating production of CXCR4 and CXCL12, enhancing the survival of malignant stem cells and protecting them from cytotoxic chemotherapy in MDS and AML.[9, 10] While CXCR4 is an unfavorable prognostic marker in AML[11, 12, 13], disrupting CXCR4 signaling inhibits transmigration, survival and chemotherapy resistance in AML xenografts.[14, 15, 16] The use of G-CSF priming with cytotoxic therapy in AML and MDS decreases the duration of neutropenia, reduces infections and possibly improves responses.[17, 18, 19]

G-CSF can mobilize hematopoietic stem cells by reducing CXCL12 expression in osteoblasts in the bone marrow niche.[20, 21] Plerixafor is a bicyclam small molecule inhibitor of CXCR4 and a potent stem cell mobilizing agent that is approved for use with G-CSF for mobilization of patients undergoing autologous stem cell transplant for multiple myeloma or non-Hodgkin lymphoma[22, 23, 24] and has been studied as a single agent for mobilization of normal donors.[25, 26]

Increased expression of CXCR4 on MDS blasts is associated with worse outcomes and lower rates of apoptosis with chemotherapy; additionally, disrupting CXCR4 in vitro disrupts cell adhesion and migration.[27, 28, 29] In pre-clinical models, disruption of the CXCR4/CXCL12 axis leads to mobilization of leukemic cells and sensitization to chemotherapy.[30, 31] Several clinical studies demonstrated the feasibility using plerixafor with chemotherapy that resulted in mobilization of malignant cells and encouraging responses.[32, 33] We hypothesized that plerixafor in combination with G-CSF would mobilize MDS blasts and improve their sensitivity to azacitidine, improving the overall response rate. The goal of this trial was to determine the optimal dose of plerixafor in combination with G-CSF and azacitidine in patients with high risk MDS.

METHODS

Patients

This was a single institution open-label phase 1 clinical trial of plerixafor, G-CSF and azacitidine. Eligible patients were >18 years old with a histologically confirmed diagnosis of MDS according to WHO criteria with 5–20% blasts in the bone marrow aspirate and at least one cytopenia.[34] Previous hypomethylating agent (HMA) therapy was allowed and all patients were required to have a 4 week washout period from prior therapy. Patients were excluded if they had untreated 5q minus syndrome, if their life expectancy was < 2 months in the opinion of the treating physician, or if they had poor liver or kidney function.

Treatment

This phase 1 trial of adding plerixafor to G-CSF and azacitidine was performed using a standard 3 + 3 design with 3 dose escalation cohorts of plerixafor 320 µg/kg/day, 440 µg/kg/day, and 560 µg/kg/day. Seven additional patients were enrolled at the highest dose level of plerixafor and treated without G-CSF to evaluate the phenotype and preferential mobilization of blasts with plerixafor alone. Patients received G-CSF 5 µg/kg subcutaneously daily on days 1–5. Plerixafor was given subcutaneously at the prescribed dose level on days 1–5, 4 hours prior to azacitidine 75 mg/m² given as a subcutaneous injection (Figure 1). The trial initially called for patients to receive G-CSF 10 mcg/kg subcutaneously daily on days 1–8, plerixafor daily on days 4–8, and azacitidine on days 4–8, but was amended after the first 3 patients to reduce G-CSF dose to 5 mcg/kg/day and administration to 5 days.

Figure 1.

Trial schema showing cycle 1 of plerixafor, G-CSF and azacitidine. P plerixafor. V azacitidine (Vidaza).

The primary objective of this study was to determine the optimal dose and schedule of plerixafor and determine the safety and tolerability of plerixafor, G-CSF and azacitidine. Secondary objectives were to characterize the mobilization of MDS cells and determine the response rate and progression free survival of patients. Dose limiting toxicity was defined as grade 3 or higher non-hematologic toxicity and hematologic toxicity of leukostasis or tumor lysis. Myelosuppression, infection, grade 3 nausea, fatigue, weight loss, and electrolyte abnormalities were not considered dose limiting. Dose holding parameters for G-CSF and plerixafor were a white blood cell (WBC) count exceeding 40k/uL or absolute peripheral blood blast count exceeding 10k/uL.

Statistical Analysis

Demographic and clinical characteristics of the sample, as well as adverse events and complications, were summarized using frequencies and percentages for categorical variables and means, medians and standard deviations for ordinal and continuous variables. Responses to therapy were described using frequencies and percentages. Survival was estimated using the Kaplan-Meier method and univariate associations between clinical characteristics were evaluated with the log-rank test. All analyses were performed using Graphpad Prism, version 7.04 and R, version 3.5.1.

This study was conducted at the Siteman Cancer Center at the Washington University School of Medicine in St. Louis. It was approved by the Washington University School of Medicine Institutional Review Board and all patients voluntarily provided informed consent to participate. This trial was registered on clinicaltrials.gov as NCT01065129. Data Collection and statistical analyses were conducted by the Siteman Cancer Center.

RESULTS

Patient Characteristics

Twenty-eight patients were treated with plerixafor, G-CSF and azacitidine. Baseline characteristics are described in Table 1. The median age at time of registration was 67 years and 64% were male. Ten patients had refractory anemia with excess blasts (RAEB-1), while 18 patients had RAEB-2. The international prognosis scoring system (IPSS) was low risk in 1 patient, intermediate-1 in 6, intermediate-2 in 15 and high in 6. Thirteen patients had received prior chemotherapy, 8 of which were for a previous malignancy. Six patients had received prior HMA therapy (azacitidine = 3, decitabine = 3) with either no response or progressive disease. On baseline bone marrow biopsy, the median blast count was 11%. Seven patients had high risk cytogenetic abnormalities. Not all patients had baseline molecular data collected based on the year they were enrolled.

Table 1.

Baseline patient and disease characteristics.

	All Patients (n = 28)	Responders (n=8)	Non-responders (n=14)
Median Age (range)	67 years (32–79)	67 years (60–71)	66 years (32–79)	p = 0.49
Sex (% male)	18 (64%)	3 (38%)	11 (79%)	p = 0.054
Diagnosis				p = 0.67
RAEB-1	10 (36%)	3 (38%)	4 (29%)
RAEB-2	18 (64%)	5 (63%)	10 (71%)
Therapy-related	8 (29%)	3 (38%)	1 (7%)
IPSS				p = 0.58
Low	1 (4%)	1 (13%)	0 (0%)
INT-1	6 (21%)	1 (13%)	2 (14%)
INT-2	15 (54%)	4 (50%)	9 (64%)
High	6 (21%)	2 (25%)	3 (21%)
Prior HMA exposure	6 (21%)	2 (25%)	4 (29%)	p = 0.86
Azacitidine	3 (11%)	2 (25%)	3 (21%)
Decitabine	3 (11%)	0 (0%)	1 (7%)
Cytogenetics:				p = 0.71
Poor Risk	8 (29%)	2 (25%)	4 (29%)
Intermediate	6 (21%)	3 (38%)	3 (21%)
Favorable	13 (46%)	3 (38%)	7 (50%)
Not evaluated	1 (4%)	0 (0%)	0 (0%)

RAEB: refractory anemia with excess blasts; IPSS: international prognostic scoring system; INT: intermediate; HMA: hypomethylating agent.

Study Results

The addition of plerixafor at the lowest dose level (plerixafor 320 µg/kg/day) to G-CSF 10 mcg/kg/day and azacitidine resulted in brisk leukocytosis. Two of 3 patients enrolled in the first cohort experienced significant leukocytosis to 80k/uL and 100k/uL (Figure 2A). One patient developed symptoms concerning for leukostasis. These patients had treatment held, and their counts returned to normal. After this, the trial was amended to reduce the G-CSF dose from 10 mcg/kg/day to 5 mcg/kg/day and to reduce the duration from 8 to 5 days, concurrent with plerixafor and azacitidine. After this protocol amendment, 5 of 18 additional patients receiving plerixafor and G-CSF developed leukocytosis of WBC >40k/uL requiring treatment to be held (Figure 2C).

Figure 2. WBC and Blast Mobilization.

(A) WBC count of the original study cohort of 3 patients treated with G-CSF 10 mcg/kg daily on days 1–8, plerixafor 320 µg/kg/day days 4–8, and azacitidine (arrow indicates start of plerixafor). (B) Blast fold increase in peripheral blood (by morphology) for the original study cohort (arrow indicates plerixafor administration). (C) Mean daily WBC count by plerixafor dose during cycles 1 and 2 in the revised cohorts treated with G-CSF 5 mcg/kg given daily on days 1–5, azacitidine, and plerixafor 320 µg/kg/day, 440 µg/kg/day, and 560 µg/kg/day in dose level (DL) 1, DL2, and DL3 as well as the expansion cohort treated with plerixafor 560 µg/kg/day and azacitidine without G-CSF (error bars indicate range). (D) Mean absolute peripheral blood blast count by dose level with ranges. (E) Baseline and peak WBC count during cycle 1 of assigned therapy by dose level. (F) Baseline and peak peripheral absolute blast count during cycle 1 of assigned therapy by dose level.

MTD Determination and Toxicity

Dose-limiting toxicities seen in the revised cohort for dose level 1 were atrial fibrillation in 1 patient and thrombocytosis in 1 patient at dose level 2 (Table 2). No DLTs were seen at dose level 3 or in the expansion cohort at dose level 3 treated with plerixafor and azacitidine. Grade 3–4 adverse events are listed in Table 3. The most common grade 3–4 nonhematologic events included hyponatremia (25%), febrile neutropenia (25%), hypophosphatemia (14%), hyperglycemia (14%), pneumonia (14%), hypoxia (11%), fatigue (11%) and dyspnea (11%). Other common adverse events included fatigue (68%), nausea (64%), diarrhea (61%), dyspnea (54%), constipation (46%), generalized weakness (46%), dizziness (42%), edema (39%), hypotension (39%) and injection site reactions (39%). Two patients died on study, 1 from respiratory failure and a second from sepsis. These events occurred after cycle 3 and cycle 2 at dose level 3 and in the expansion cohort, respectively, and were thought to be unrelated to treatment with plerixafor and/or G-CSF.

Table 2.

Treatment Cohorts

Cohorts	Dose Limiting Toxicity	Best Response	Reason for study withdrawal
DL1 (n=3) Plerixafor 320 µg/kg D4–8 G-CSF 10 mcg/kg D1–8 Azacitidine 75mg/m2 D4–8	Leukocytosis, grade 4 (n=1) Leukocytosis, grade 2 (n=1)	Not evaluable n=3	DLT n=1 Transplant n=1 Hospice n=1
Revised DL1 (n=6) Plerixafor 320 µg/kg D1–5 G-CSF 5 mcg/kg D1–5 Azacitidine 75 mg/m2 D1–5	Atrial fibrillation, grade 3 (n=1)	CR/mCR n=2 Progressive disease n=3 Not evaluable n=1	Disease progression n=4 DLT n=1 Hospice n=1
DL2 (n=6) Plerixafor 440 µg/kg D1–5 G-CSF 5 mcg/kg D1–5 Azacitidine 75 mg/m2 D1–5	Thrombocytosis (n=1)	CR/mCR n=2 Stable disease n=2 Not evaluable n=1 Progressive disease n=1	Disease progression n=3 Transplant n=1 DLT n=1 Withdrew consent n=1
DL3 (n=6) Plerixafor 560 µg/kg D1–5 G-CSF 5 mcg/kg D1–5 Azacitidine 75 mg/m2 D1–5	None	CR/mCR n=4 Stable disease n=1 Not evaluable n=1	Withdrew consent n=3 Transplant n=1 Death n=1 Disease progression n=1
DL3 Expansion Cohort (n=7) Plerixafor 560 µg/kg D1–5 Azacitidine 75 mg/m2 D1–5	None	Stable disease n=5 Progressive disease n=2	Transplant n=4 Disease progression n=2 Death n=1

DL: dose level, CR: complete response, mCR: marrow complete response

Table 3.

Grade 3–4 Adverse Events Seen in >1 Patient (N = 28)

AE	Patients with Gr III/IV AE	%
Leukopenia	16	57%
Neutropenia	11	39%
Anemia	10	36%
Thrombocytopenia	9	32%
Hyponatremia	7	25%
Febrile Neutropenia	7	25%
Pneumonia	4	14%
Hypophosphatemia	4	14%
Hypoxia	3	11%
Fatigue	3	11%
Dyspnea	3	11%
Urinary Tract Infection	2	7%
Syncope	2	7%
Cellulitis	2	7%
Sinusitis	2	7%
Hypokalemia	2	7%
Hypertension	2	7%
Headache	2	7%
Decreased ejection fraction	2	7%
Diarrhea	2	7%
Acute kidney injury	2	7%
C. Diff Colitis	2	7%

As expected with HMA therapy in patients with MDS, hematologic toxicity was common. Grade 3–4 lymphopenia, anemia, and thrombocytopenia were seen in 64%, 36% and 32% of patients, respectively. As expected with G-CSF and plerixafor, leukocytosis was relatively common with 4 instances of grade 2 leukocytosis and one instance of grade 4 leukocytosis. This led to study discontinuation in two patients and treatment delays in three patients.

Efficacy Assessments

The 28 patients treated received a median of 3 cycles of treatment (range < 1 to 10) with 22 patients evaluable after cycle 2. The best response achieved in these 22 patients was a complete response (CR) in 2 patients, marrow complete response (mCR, with incomplete count recovery) in 6, stable disease in 8, and progressive disease in 6 (Table 2). The overall response rate (ORR, CR + mCR) of evaluable patients was 36% and the intent-to-treat overall response rate was 29%. Of the 6 patients with mCR, 4 had hematologic improvement; with 3 in platelet count, 2 in hemoglobin, and 1 in neutrophil count.

Of the 6 patients that received prior HMA therapy, there was 1 CR, 1 mCR, 1 stable disease, and 3 progressive diseases, for an ORR of 33% (p = 0.93, versus HMA naïve patients). Of the patients treated at dose level 3 of plerixafor 560 µg/kg/day with G-CSF, the ORR was 4/5 evaluable patients, with 1 patient having a CR, 3 with marrow CR, and 1 stable disease. For the expansion cohort of 7 patients treated with plerixafor 560 µg/kg/day and azacitidine, without concurrent G-CSF, there were no responders. Five patients had stable disease and 2 had disease progression after 2 cycles. Of these 7 patients, 2 had prior HMA therapy and 5 were treatment-naïve. Excluding the expansion cohort, the ORR for all evaluable patients receiving plerixafor (any dose), G-CSF, and azacitidine was 53% (8 of 15 evaluable patients).

The median overall survival (OS) for the entire study population was 452 days (95% confidence interval 271–794 days) (Figure 4). For responders, the median OS was 511 days (95% CI 423 days with the upper 95% CI not reached) versus 338 days (95% CI 149–880 days) for nonresponders (p=0.7). The median duration of response was 392 days (95% CI 230-NR). The OS of patients who were previously treated with a HMA therapy was 510 days, which was not significantly different from those who were treatment-naïve (p = 0.50). Fifteen of the 28 patients went on to receive an allogeneic hematopoietic cell transplant (HCT) at a median of 90 days after their last study treatment (range 12–184 days). Patients who received a HCT had a median OS of 616 days versus 270 days for those who did not (p < 0.01).

Figure 4.

OS of all patients treated with 95% confidence intervals.

Mobilization of Leukocytes and Blasts

The peripheral blood WBC count increased briskly in the first 3 patients in the original study cohort treated with G-CSF 10 mcg/kg on days 1–8 prior after starting plerixafor 320 µg/kg on day 4 (Figure 2A). In the revised study cohort treated with G-CSF 5 mcg/kg/day on days 1–5, concurrent with the dose escalation of plerixafor, there was a more modest increase in WBC count with no clear dose-response relationship with plerixafor (Figure 2). In the plerixafor 560 µg/kg/day expanded cohort that did not receive G-CSF, there was minimal increase in WBC count or circulating blasts.

Blast mobilization was analyzed by peripheral blood morphology and by FISH using commercially available probes on whole blood mononuclear cells in patients with baseline cytogenetic abnormalities. In the original study cohort of 3 patients treated with G-CSF 10 mcg/kg on days 1–8 prior to starting plerixafor on day 4, there was a substantial increase in blast percentage between 15–50 fold after three days of G-CSF, which then increased an additional 11–40 fold after plerixafor administration on day 4 (Figure 2B). The revised study cohort received G-CSF 5 mcg/kg on day 1–5, concurrent with plerixafor and had a more variable increase in blast count with 10 patients having a greater than 2-fold increase in blast mobilization. Blast mobilization did not appear to correlate with plerixafor dose (p = 0.60) (Figure 2D). Blast mobilization during treatment was associated with response, with patients that mobilized >2-fold having an ORR of 60% versus 17% for those that did not mobilize (p = 0.035). However, this improved response rate did not result in improved OS for patients that mobilized (p = 0.20). Blast mobilization was not associated with patient IPSS score (p = 0.36) and there was a non-significant trend toward better mobilization in patients not previously treated with prior HMA therapy, 52.9% versus 16.7% (p = 0.12).

Thirteen patients had informative cytogenetics and FISH was performed at 0, 2, 6 and 24 hours after the day 3 dose of plerixafor in an attempt to evaluate for preferential mobilization of malignant blasts. The percent of blasts with cytogenetic abnormalities in the peripheral blood did not appear to change significantly after plerixafor injection (Figure 3), suggesting that while the absolute number of peripheral blasts increased in response to plerixafor in some patients, the malignant cells did not preferentially mobilize.

Figure 3.

FISH of mononuclear cells in the peripheral blood on Cycle 1 Day 3 after plerixafor administration in 13 patients with informative cytogenetics to evaluate for preferential mobilization of leukemic blasts. Lines depict the percentage of a representative cytogenetic abnormality in an individual patient.

DISCUSSION

In this article, we describe the safety and tolerability of adding plerixafor to G-CSF and azacitidine. While the safety of adding plerixafor to G-CSF is well-established in healthy donors and in patients with myeloma and non-hodgkin’s lymphoma, there remains concerns of leukocytosis when using this combination in patients with high grade hematologic malignancies. Indeed, when plerixafor 320 µg/kg/day was given after priming with 3 days of G-CSF 10 mcg/kg, brisk leukocytosis was observed in 2 of the first 3 patients treated at dose level 1, prompting a protocol amendment. One of these patients had clinical evidence of leukostasis, which resolved after holding further therapy. After the reduction in dose and duration of G-CSF to 5 mcg/kg/day for 5 days, the combination still caused frequent leukocytosis that could be effectively managed with dose delays in 5 patients. In the expanded cohort treated with plerixafor 560 µg/kg/day without G-CSF, minimal leukocytosis or blast mobilization was observed, suggesting the combination of G-CSF with plerixafor is more potent at mobilizing MDS patients than plerixafor alone. Since normal hematopoietic cells can also be mobilized and may be sensitized to the effects chemotherapy, there is a possibility of prolonging cytopenias with this combination, although this was not observed in this study. While other studies in AML have suggested FISH-positive blasts may be preferentially mobilized with the combination of plerixafor and G-CSF, this was not observed in this trial.[33] A number of explanations could account for this finding, first being the small numbers in this trial, the effect of different chemotherapies on mobilization when used with plerixafor and G-CSF, or the lower blast burden seen in MDS compared to AML.

This study supports our hypothesis that adding plerixafor to G-CSF mobilizes MDS blasts in patients with high risk MDS. The concept of G-CSF priming in hematopoietic malignancies has been extensively studied over the last few decades, where it was shown to reduce duration of neutropenia and infectious complications.[18, 35, 36, 37] Although the reported anti-leukemic efficacy of G-CSF priming has been mixed, there are several studies demonstrating improvements in response rate and DFS when used with intensive chemotherapy for induction or salvage of myeloid malignancies.[17, 18, 19] The mechanism of action is presumed to be G-CSF activation of metabolic processes and blast proliferation, thereby sensitizing them to cytotoxic therapy. G-CSF mobilizes normal and leukemic stem cells by down-regulating CXCL12 mRNA and protein expression in the perivascular niche of the marrow, disrupting homing and retention of stem cells and the interaction with other cell surface molecules that cause stromal-mediated chemo-resistance. Plerixafor acts synergistically with G-CSF to mobilize stem cells in autologous donors but also has a second potential mechanism of action by preventing stem cell factor (SCF)- induced internalization of CXCR4 thus disrupting downstream signaling through the PI3K-ADK and MAPK pro-survival pathways.[32]

Fifty three percent of evaluable patients responded to plerixafor, G-CSF and azacitidine in this study, which compares favorably to azacitidine alone.[2] An association between blast mobilization and response was observed that has been previously reported by other groups using plerixafor with decitabine.[38] Interestingly, responses were seen at all dose levels of plerixafor with G-CSF, but no responses were seen in the seven patients enrolled at the MTD of plerixafor 560 µg/kg/day without G-CSF. While these are small numbers and most responses were mCR, it suggests clinical response may be improved with dual targeting of the CXCR4/CXCL12 axis with both G-CSF and plerixafor. Other studies have demonstrated increased surface expression of CXCR4 on AML blasts after the administration of plerixafor, which is felt to be due to the ability of plerixafor to block ligand induced internalization of CXCR4.[32, 33] This increase in CXCR4 expression may be abrogated with G-CSF administration, leading to improved response.

Another group has administered plerixafor plus ten day decitabine in elderly patients with AML, showing an ORR of 43%, which was similar to the response published by the same group using decitabine without plerixafor.[38, 39] Interestingly, when plerixafor was combined with decitabine, blast mobilization was observed that was also significantly associated with clinical response. However, the largest predictor of response to decitabine and plerixafor was prior HMA exposure. In our study, responses were also seen in patients who were previously treated with HMAs, although comparison of response rates is limited due to small numbers. While this may not be a real difference, the mechanisms of chemoresistance in MDS are poorly understood and may be different between different HMAs. These mechanisms are likely at least in part due to the complex interactions between MDS blasts and the bone marrow microenvironment and may be susceptible to dual targeting to the CXCR4/CXCL12 axis by plerixafor and G-CSF, restoring HMA-sensitivity to some individuals.

Combining the CXCR4/CXCL12 inhibitor plerixafor with G-CSF and azacitidine is an intriguing approach to improving responses in high grade MDS. This trial demonstrates the safety and feasibility of this strategy, and its ability to mobilize MDS blasts. The response rates observed in this study are encouraging and warrant further study.

KEY POINTS.

Plerixafor and G-CSF can be safely combined with azacitidine for 5 days in patients with MDS.

The overall response rate of 53% for evaluable patients with this regimen is higher than expected and more responses were seen in patients with blast mobilization.