Sorafenib and Omacetaxine Mepesuccinate As a Safe and Effective Treatment for Acute Myeloid Leukemia Carrying Internal Tandem Du

Home , Azacitidine, Decitabine, Fludarabine, Venetoclax

Original Article Sorafenib and Omacetaxine Mepesuccinate as a Safe and Effective Treatment for Acute Myeloid Leukemia Carrying Internal Tandem Duplication of Fms-Like Tyrosine Kinase 3

Chunxiao Zhang, MSc1; Stephen S. Y. Lam, MBBS, PhD1; Garret M. K. Leung, MBBS1; Sze-Pui Tsui, MSc2; Ning Yang, PhD1; Nelson K. L. Ng, PhD1; Ho-Wan Ip, MBBS2; Chun-Hang Au, PhD3; Tsun-Leung Chan, PhD3; Edmond S. K. Ma, MBBS3; Sze-Fai Yip, MBBS4; Harold K. K. Lee, MBChB5; June S. M. Lau, MBChB6; Tsan-Hei Luk, MBChB6; Wa Li, MBChB7; Yok-Lam Kwong, MD 1; and Anskar Y. H. Leung, MD, PhD 1

BACKGROUND: Omacetaxine mepesuccinate (OME) has antileukemic effects against acute myeloid leukemia (AML) carrying an internal tandem duplication of Fms-like tyrosine kinase 3 (FLT3-ITD). A phase 2 clinical trial was conducted to evaluate a combination treatment of sorafenib and omacetaxine mepesuccinate (SOME). METHODS: Relapsed or refractory (R/R) or newly diagnosed patients were treated with sorafenib (200-400 mg twice daily) and OME (2 mg daily) for 7 (first course) or 5 days (second course on- ward) every 21 days until disease progression or allogeneic hematopoietic stem cell transplantation (HSCT). The primary endpoint was composite complete remission, which was defined as complete remission (CR) plus complete remission with incomplete hematologic recovery (CRi). Secondary endpoints were leukemia-free survival (LFS) and overall survival (OS). RESULTS: Thirty-nine R/R patients and 5 newly diagnosed patients were recruited. Among the R/R patients, 28 achieved CR or CRi. Two patients showed partial remission, and 9 patients did not respond. Among the 5 newly diagnosed patients, 4 achieved CR, and 1 achieved CRi. The median LFS and OS were 5.6 and 10.9 months, respectively. Prior Fms-like tyrosine kinase 3 (FLT3) inhibitor exposure (P = .007), 2 or more inductions (P = .001), and coexisting IDH2 (P = .008) and RUNX1 mutations (P = .003) were associated with lower CR/CRi rates. HSCT consolidation and deep molecular responses (defined as an FLT3-ITD variant allelic frequency [VAF] ≤ 0.1% or a nucleophosmin 1 [NPM1] mutant VAF ≤ 0.01%) were associated with better OS and LFS. Prior FLT3 inhibitor exposure and 2 or more inductions were associated with inferior LFS. CONCLUSIONS: SOME was safe and effective for R/R and newly diagnosed FLT3-ITD AML. Cancer 2019;0:1-10. © 2019 American Cancer Society.

KEYWORDS: acute myeloid leukemia (AML), internal tandem duplication of Fms-like tyrosine kinase 3 (FLT3-ITD), minimal residual disease, omacetaxine mepesuccinate, sorafenib.

INTRODUCTION Acute myeloid leukemia (AML) is a heterogeneous group of diseases characterized by distinct clinicopathologic, cytogenetic, and genetic features.1-3 Induction chemotherapy and allogeneic hematopoietic stem cell transplantation (HSCT) are the mainstays of treatment. Clinicopathologic and genetic features at diagnosis may predict treatment responses and long- term outcomes.4 Internal tandem duplication of Fms-like tyrosine kinase 3 (FLT3-ITD) is one of the most common mutations in AML2 and occurs in 25% to 35% of adult patients, particularly those with normal cytogenetics or rarely a t(6;9) translocation. Relapses after conventional chemotherapy are frequent in FLT3-ITD+ AML and portend an 5 extremely poor outcome. Salvage chemotherapy results in remission in <20% of cases, with a median duration of survival of merely 3 months.6,7 A combination of a tyrosine kinase inhibitor targeting Fms-like tyrosine kinase 3 (FLT3) with standard induction and consolidation chemotherapy has emerged as a new standard of care for newly diagnosed FLT3-ITD AML.8 In 9,10 relapsed or refractory (R/R) FLT3-ITD+ AML, the effects of FLT3 inhibitors have been modest and transient. In vitro drug sensitivity testing with primary leukemia samples has identified omacetaxine mepesuccinate (OME; also known as homoharringtonine or HHT) as an effective adjunct for the treatment of FLT3-ITD AML.11 OME acts by inhibiting

Corresponding author: Anskar Y. H. Leung, MD, PhD, Department of Medicine, Queen Mary Hospital, Pokfulam Road, Room K418, K Block, Hong Kong, China; ayhleung@ hku.hk 1 Division of Haematology, Department of Medicine, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong, China; 2 Division of Haematology, Department of Pathology, Queen Mary Hospital, Hong Kong, China; 3 Department of Pathology, Hong Kong Sanatorium and Hospital, Hong Kong, China; 4 Department of Medicine, Tuen Mun Hospital, Hong Kong, China; 5 Department of Medicine, Princess Margaret Hospital, Hong Kong, China; 6 Department of Medicine, Queen Elizabeth Hospital, Hong Kong, China; 7 Department of Clinical Oncology, Prince of Wales Hospital, Hong Kong, China We thank all the patients and their families for participating in the trial and contributing clinical samples. Additional supporting information may be found in the online version of this article.

DOI: 10.1002/cncr.32534, Received: May 16, 2019; Revised: August 16, 2019; Accepted: August 28, 2019, Published online Month 00, 2019 in Wiley Online Library (wileyonlinelibrary.com)

Cancer Month 0, 2019 1 Original Article global messenger RNA translation. FLT3-ITD+ AML reaction would require cessation of sorafenib until clini- appears to be particularly sensitive to the antileuke- cal improvement, and sorafenib would be reinstated at mia effect of OME. We showed previously that signif- 200 mg twice daily. icant synergism occurred between FLT3 inhibitors and OME in leukemic samples in vitro and in vivo as murine Definition of Responses xenografts.11 Importantly, preliminary clinical data have A bone marrow (BM) examination was performed on day indicated that such synergism occurs in patients with R/R 21 after the commencement of treatment. Patients who did 11 FLT3-ITD+ AML. In this report, we describe the entire not achieve CR/CRi received a second course of SOME cohort of patients with FLT3-ITD+ AML treated with (7 days of OME), and the BM examination was repeated sorafenib and omacetaxine mepesuccinate (SOME). on day 21 of the second cycle. The best BM response was documented, and those who did not achieve CR/CRi after the second cycle exited the trial. CR was defined as 5% MATERIALS AND METHODS < blasts in BM or blood with an absolute neutrophil count Patients 9 9 ≥1 × 10 /L and a platelet count ≥100 × 10 /L. CRi was Adult patients (age ≥ 18 years) with FLT3-ITD+ AML defined as <5% blasts in BM or blood but with incom- who relapsed or were refractory to induction chemother- plete hematologic recovery. Partial responses (PR) and no apy as well as newly diagnosed patients with FLT3-ITD+ responses (NR) were defined as ≥50% and <50% reduc- AML who were unfit for or declined chemotherapy were tions of circulating or BM blasts, respectively, and in both recruited. All patients gave informed consent for treat- cases, the absolute blast counts were >5% of nucleated cells. ment. The study was approved by the institutional review board in accordance with the Declaration of Helsinki and Transfusion Dependence was registered at ClinicalTrials.gov. Assuming a 20% to The numbers of packed cell units and platelet transfusions 30% increase in the response rate in comparison with (1 adult therapeutic dose for each platelet transfusion) for sorafenib monotherapy,12 a type I error of 0.05, and a sta- patients who achieved CR/CRi were summed on a monthly tistical power of 0.8, we needed approximately 40 patients basis from the time they achieved CR/CRi to the time of for this study (https://clincalc.com/stats/samplesize.aspx). death, disease progression, or allogeneic HSCT.

SOME Regimen Mutation Profile Sorafenib was administered continuously from day 1 Mutation profiling was performed at diagnosis or before (400 mg twice daily orally). OME was administered from SOME treatment. Genomic DNA was extracted with day 1 to day 7 (2 mg/d intravenously) in the first 21-day the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, cycle. After the first cycle, patients achieving complete Germany) and was analyzed with MiSeq next-generation remission (CR) or complete remission with incomplete sequencing (Illumina, San Diego, California) based on hematologic recovery (CRi) received OME from day 1 to the TruSight myeloid sequencing panel (Illumina). The day 5 for subsequent 21-day cycles. For patients achieving panel targeted 54 genes and covered the full coding se- a partial response, OME continued to be administered quence of 15 genes and exonic hotspots for the other from day 1 to day 7 per 21-day cycle until CR/CRi, after 39 genes. The workflows of MiSeq sequencing library which OME was given from day 1 to day 5 for subse- preparation, variant calling, and annotation as well as the quent cycles. Patients with nonremission were withdrawn detection of FLT3-ITD by ITD seek have been described from the study. SOME was continued until allogeneic elsewhere.11,13 Complex insertions and deletions were HSCT, a loss of response, or withdrawal from the study detected by an in-house designed algorithm (INDEL due to adverse events. seek) on a Cray XC30 supercomputer.

Adverse Events Measurement of the FLT3-ITD Variant Allelic Investigator-assessed adverse events were graded Frequency and Detection of Tyrosine Kinase according to the National Cancer Institute’s Common Domain Mutations Terminology Criteria for Adverse Events (version 4.03). FLT3-ITD was evaluated by polymerase chain reaction In the event of treatment-related pancytopenia, gran- (PCR) with primers covering exons 14 to 15 followed by ulocyte colony-stimulating factor and prophylactic capillary electrophoresis. The size of FLT3-ITD clones, antifungal therapy were given. Patients who developed a represented as the variant allelic frequency (VAF), was grade 2 or higher maculopapular rash or hand-foot skin evaluated by their area under the curve with reference

2 Cancer Month 0, 2019 Sorafenib and Omacetaxine in FLT3-ITD AML/Zhang et al to that of the sum of FLT3-ITD and FLT3 wild-type Study Endpoints and Statistical Analysis clones. Where multiple FLT3-ITD clones existed, their The primary endpoint was the composite complete areas under the curve were summed. The FLT3-ITD VAF remission rate, which was defined as CR plus CRi at the was measured in BM samples before SOME, at the time time of the best BM response. Secondary endpoints were of the best BM response, and thereafter when the BM leukemia-free survival (LFS), which was calculated from examination was performed as clinically indicated. The CR/CRi to the time of relapse, death, or last follow-up, detection limit was determined by serial dilutions of plas- and overall survival (OS), which was calculated from mids encoding the FLT3-ITD (260 bp) and FLT3 wild- the beginning of treatment to death or latest follow-up. type (239 bp) alleles and was set at 0.1% (Supporting Survival was evaluated with the Kaplan-Meier method Fig. 1A,B). All patients in this study had a BM FLT3-ITD and compared with the log-rank test. Patients’ age, sex, VAF ≥ 1%. Tyrosine kinase domain (TKD) mutation cytogenetic subtypes, gene mutations, white cell counts, was examined with Sanger DNA sequencing.12 BM blasts, and FLT3-ITD VAF as well as treatment before SOME were examined for their impact on the compos- Determination of Minimal Residue Disease ite complete remission. These factors, together with the For patients with a type A nucleophosmin 1 (NPM1) mu- achievement of MRD negativity and HSCT, were exam- tation at diagnosis, quantification of the NPM1 mutant ined for their impact on LFS and OS. Nominal variables VAF was used as a surrogate for minimal residue disease were compared with the chi-square or Fisher exact test. (MRD) measurement. Genomic DNA extracted from Continuous variables were compared with the Student serial BM samples was analyzed by droplet digital poly- t test for parametric parameters or the Mann-Whitney merase chain reaction (ddPCR).14 TaqMan Liquid Biopsy U test for nonparametric parameters. A Cox proportional digital PCR assays for NPM1 (Thermo Fisher Scientific, hazards model was used in univariate analyses. A P value Waltham, Massachusetts) were used to quantify the NPM1 less than .05 was considered statistically significant. All type A mutant and wild-type allele. Each ddPCR reaction analyses were performed with IBM SPSS Statistics (ver- contained the DNA to be tested (200 ng), 2× ddPCR sion 18). Supermix for probes (no deoxyuridine triphosphate; Bio- Rad, Hercules, California), 20× TaqMan Liquid Biopsy RESULTS AND DISCUSSION digital PCR assays for NPM1 (Thermo Fisher Scientific), Patients and an HaeIII restriction enzyme (10,000 U/mL; New Eighteen men and 26 women at a median age of 50 years England Biolabs, Ipswich, Massachusetts). The ddPCR (range, 19-76 years) were treated (Table 1). At recruit- reaction was made up to 20 µL and was performed with ment, all patients showed an FLT3-ITD VAF ≥ 1% in the QX200 ddPCR system (Bio-Rad). Droplet genera- their BM (Supporting Table 1). Patients had relapsed tion was performed with the QX200 droplet generator (n = 27), refractory (n = 12), or newly diagnosed disease (Bio-Rad). PCR was performed with the C1000 Touch (n = 5). Seven patients had previously experienced failure thermal cycler, and droplets were read with the QX200 with FLT3 inhibitors, including sorafenib (n = 4) and droplet reader (Bio-Rad). The detection of fluorescence quizartinib (n = 3). Their mutation profile is shown in signals (FAM for an NPM1 type A mutation/VIC for Figure 1A. The most frequent concurrently mutated genes the NPM1 wild type) was performed for each individual were NPM1, DNA methyltransferase 3A (DNMT3A), droplet. Results were analyzed with QuantaSoft software Wilms tumor 1 (WT1), IDH2, and RUNX1. (version 1.7.4; Bio-Rad), which enumerated the number of positive and negative droplets arising from each Treatment Responses sample, and the frequency of NPM1 mutants was deter- Among the 39 R/R patients, 28 (71.8%) achieved CR/ mined on the basis of Poisson statistics. Each experiment CRi (CR, 2; CRi, 26) after 1 (n = 26) or 2 courses of was performed in duplicate. For calibration of sensitivity, SOME (n = 2), and they received a median of 4 courses an OCI-AML3 cell line carrying NPM1 type A muta- of treatment (range, 1-13; Fig. 1B). Thirteen eligible tion was used. The NPM1 MRD status was expressed as patients underwent allogeneic HSCT after remission; VAF, which was calculated on the basis of the number of 10 of these patients relapsed, and none of them received NPM1 type A copies with respect to the sum of mutant FLT3 inhibitor maintenance after HSCT. Fourteen pa- and wild-type NPM1. Serial dilutions of mutant NPM1 tients who achieved CR/CRi but did not undergo HSCT versus wild-type NPM1 was performed, and the detection relapsed despite continuous SOME treatment. One limit was defined at 0.001% (Supporting Fig. 1C,D). patient died of pneumonia at CRi. Eleven patients

Cancer Month 0, 2019 3 Original Article

TABLE 1. Clinicopathologic Characteristics of 44 (median; range, 1-19 months). The 2 patients who Patients achieved CR did not require a transfusion. Of the 5 newly Characteristic Value diagnosed patients, none required packed cells or platelet transfusions after the achievement of CR/CRi. Relapsed or refractory (n = 39) Male/female, No. 16/23 Age, median (range), y 50 (19-76) Clinicopathologic Factors Associated With 9 Presenting WCC, median (range), ×10 /L 85 (1-411) CR/CRi Cytogenetics, No. (%) Normal 27 (69) The clinical and genetic factors of patients were correlated Complex 2 with responses to SOME treatment in the 39 patients Trisomy 3 Miscellaneous 3 with R/R disease (Supporting Table 2). Among the factors Poor growth 4 analyzed, prior exposure to FLT3 inhibitors (P .007), 2 Previous regimens of chemotherapy, No. (%) = 1 31 (79) or more prior induction regimens (P = .001), and coexist- ≥2 8 (21) ing IDH2 (P = .008) and RUNX1 mutations (P = .003) NPM1 type A mutated, No. (%) 23 (59) Pre-SOME FLT3-ITD VAF, median (range) 0.71 (0.014-0.99) were associated with lower CR/CRi rates. Age, presenting Prior exposure to single-agent TKI treatment, No. (%) white blood cell counts, pretreatment BM blast counts, No 32 (82) Sorafenib 4 (10) and FLT3-ITD VAFs as well as karyotype had no effects Quizartinib 3 (8) on the CR/CRi rate. Upfront SOME treatment (n = 5) Male/female, No. 2/3 Survival Age, median (range), y 59 (54-69) 9 Presenting WCC, median (range), ×10 /L 68.3 (0.77-189.8) The survival of the entire cohort of 44 patients is shown Cytogenetics, No. Normal 4 in Figure 1C. The 11 patients who did not achieve CR/ Tetraploidy 1 CRi showed dismal outcomes with survival shorter than NPM1 type A mutated, No. (%) 4 (80) 11 months (median, 4.6 months; range, 0.13-10.8 months). Pre-SOME FLT3-ITD VAF, median (range) 0.43 (0.38-0.78) Those who achieved CR/CRi had median OS and LFS Abbreviations: FLT3-ITD, internal tandem duplication of Fms-like tyrosine ki- times of 10.9 and 5.6 months, respectively. Patients who nase 3; NPM1, nucleophosmin 1; SOME, sorafenib and omacetaxine mepesuccinate; TKI, tyrosine kinase inhibitor; VAF, variant allelic frequency; WCC, underwent allogeneic HSCT (n = 16) at CR/CRi had white blood cell count. median OS and LFS times of 20.9 and 12.5 months, respectively. Seventeen patients who achieved CR/CRi but showed no response (n = 9) or a partial response (n = 2). did not undergo HSCT did not survive beyond 20 months Five patients with newly diagnosed FLT3-ITD AML were (Fig. 1D). Clinicopathologic features associated with LFS treated with SOME. Four achieved CR after 1 (n = 3) or and OS are shown in Table 2. HSCT and achieving an FLT3- 2 courses of SOME treatment (n 1), and 1 achieved CRi = ITD VAF ≤ 0.1% and an NPM1 mutant VAF ≤ 0.01% as her best response. Two patients underwent allogeneic during the course of treatment were associated with longer HSCT at the first CR. One patient relapsed 7.5 months LFS and OS. A high white blood cell count and a high after remission, achieved a second CR by induction 9 FLT3-ITD VAF before SOME, arbitrarily set as 30 × 10 /L chemotherapy, and underwent allogeneic HSCT. These and 90%, respectively, as well as DNMT3A and WT1 3 patients remained in remission 14, 6, and 2 months mutations were associated with shorter LFS and OS. after HSCT. One patient was lost to follow-up, and 1 patient relapsed 3 months after CRi and died. Molecular Monitoring Molecular monitoring based on FLT3-ITD and NPM1 Transfusion Dependence mutations was performed for 119 serial peripheral blood Among the 26 of 28 R/R patients who achieved CRi, 1 or BM samples from 33 patients who achieved CR/CRi (unique patient number [UPN] 1668) did not require a with SOME (28 R/R patients and 5 newly diagnosed packed cell or platelet transfusion after the achievement of patients). remission. Four patients (UPNs 2265, 2523, 2606, and 2612) did not require a packed cell transfusion during Changes in FLT3-ITD VAF remission. Twenty-one patients required 3 units of packed FLT3-ITD VAF tests were performed successfully in cell transfusions per month (median; range, 1-5 units) for 27 patients but failed in 5 patients because of an insuf- 2 months (median; range, 1-18 months). Twenty-five ficient quantity of DNA. A remission sample was not patients required platelet transfusions of 4 adult therapeu- available for 1 patient. Fifteen patients had undetectable tic doses per month (median; range, 1-14) for 2 months FLT3-ITD (VAF ≤ 0.1%) before or without HSCT

4 Cancer Month 0, 2019 Sorafenib and Omacetaxine in FLT3-ITD AML/Zhang et al

(n = 9) or only after HSCT (n = 6). Twelve patients better survival than those with persistently positive FLT3- showed persistently detectable FLT3-ITD (VAF > 0.1%; ITD (median OS, 20.9 vs 7.89 months; median LFS, Fig. 2A). Patients with undetectable FLT3-ITD showed 17.70 vs 3.12 months; Fig. 2B,C).

Cancer Month 0, 2019 5 Original Article

Figure 1. Mutation profile of 44 patients and their treatment outcomes. (A) Mutation profile of the entire cohort of 44 patients with AML. The upper panels indicate patient categories (upfront treatment, n = 5; treatment in an R/R state, n = 39) and treatment responses. Shaded boxes at the bottom panel indicate mutations with VAFs ≥ 5%. (B) Treatment outcomes of the entire cohort of 44 patients with AML. (C) (Left) OS for 44 patients and (Right) LFS for 33 patients who achieved CR/CRi with SOME. (D) Effects of HSCT on (Left) OS and (Right) LFS among the 33 patients who achieved CR/CRi with SOME. Statistical significance was calculated with the log-rank test. AML indicates acute myeloid leukemia; Ara-C, cytosine arabinoside; CR, complete remission; CRi, complete remission with incomplete hematologic recovery; DNMT3A, DNA methyltransferase 3A; FLAG, fludarabine, cytarabine, and granulocyte colony-stimulating factor; FLT3-ITD, internal tandem duplication of Fms-like tyrosine kinase 3; HiDAC, high-dose cytarabine; HSCT, hematopoietic stem cell transplantation; ICE, idarubicin, cytarabine, and etoposide; LFS, leukemia-free survival; NPM1, nucleophosmin 1; NR, nonremission; OS, overall survival; PBSC, peripheral blood stem cell; PR, partial remission; QUIZOM, quizartinib and omacetaxine mepesuccinate; R/R, relapsed or refractory; SOME, sorafenib and omacetaxine mepesuccinate; VAF, variant allelic frequency; WT1, Wilms tumor 1.

TABLE 2. Univariate Analysis of LFS and OS for Patients Who Achieved CR/CRi (n = 33) LFS OS

95% CI 95% CI

P Hazard Ratio Lower Upper P Hazard Ratio Lower Upper

Age ≥ 50 y .954 0.975 0.410 2.319 .549 0.760 0.310 1.865 9 Pre-SOME WBCs ≥ 30 × 10 /L .031 0.353 0.138 0.908 .017 0.308 0.117 0.809 Pre-SOME BM blasts ≥ 70% .931 0.961 0.394 2.345 .508 0.736 0.296 1.828 Pre-SOME FLT3-ITD VAF ≥ 90% .007 0.292 0.119 0.715 .007 0.283 0.113 0.705 Normal cytogenetics .720 1.190 0.460 3.076 .911 1.045 0.481 2.275 HSCT <.001 6.668 2.369 18.771 .001 7.103 2.320 21.747 FLT3-ITD VAF ≤ 0.1% at DR <.001 11.778 3.076 45.101 .001 8.738 2.351 32.477 NPM1 VAF ≤ 0.01% at DR .013 15.427 1.760 135.185 .030 6.187 1.190 32.170 Mutations NPM1 .705 0.859 0.374 1.939 .703 0.876 0.437 1.746 DNMT3A .039 2.176 1.067 4.880 .099 1.762 0.906 3.479 WT1 .221 2.064 0.550 14.21 .009 2.825 1.570 171.71

Abbreviations: BM, bone marrow; CR, complete remission; CRi, complete remission with incomplete hematologic recovery; DNMT3A, DNA methyltransferase 3A; DR, deepest response; FLT3-ITD, internal tandem duplication of Fms-like tyrosine kinase 3; HSCT, hematopoietic stem cell transplantation; LFS, leukemia-free survival; NPM1, nucleophosmin 1; OS, overall survival; SOME, sorafenib and omacetaxine mepesuccinate; VAF, variant allelic frequency; WBC, white blood cell; WT1, Wilms tumor 1. 9 9 9 Bold P values represents the statistical significance. Different levels of WBCs (≥10 × 10 /L, ≥15 × 10 /L, and ≥30 × 10 /L), BM blasts (≥50%, ≥70%, and ≥85%), and pre-SOME FLT3-ITD VAFs (≥30%, ≥50%, ≥60%, and ≥90%) were tested, and the cutoffs that defined significant differences in LFS and OS are shown.

NPM1 MRD 22 relapsed samples. FLT3-TKD mutations were detected Among the 23 patients who showed an NPM1 type A in 5 samples at D835; they included 1 patient who carried mutation at diagnosis, NPM1 MRD tests were performed the same mutation before SOME treatment (Supporting successfully in 16 patients. Nine patients showed an Fig. 2). No D691 mutation was detected. NPM1 mutant VAF ≤ 0.01% as their deepest response during the course of treatment either before or without Toxicities Related to SOME HSCT (n = 3) or only after HSCT (n = 6). Seven pa- Most patients developed cytopenia after SOME and re- tients showed a VAF > 0.01% throughout treatment quired granulocyte colony-stimulating factor, antifungal (Fig. 2D). There were significant differences in median prophylaxis, and blood product support. Nonhematologic OS (not reached vs 10.70 months; P = .0493) and LFS toxicities associated with SOME were minimal and were (24.67 vs 3.97 months; P = .0077) between the 2 groups limited to hand-foot skin reactions and grade 2 or lower (Fig. 2E,F). sorafenib-associated rashes that resolved upon a sorafenib dose reduction (Table 3). Repeated treatments were possi- FLT3-TKD Mutation ble, and the modest side effects make it uniquely suitable Three patients carried FLT3-TKD mutations before for elderly patients or those unfit for conventional treat- receiving SOME treatment, and all achieved CR/CRi ment. Specifically, patient UPN 2251 was diagnosed with (Fig. 1A). Sanger sequencing of the FLT3 gene, corre- AML with trisomy 8 and an NPM1 type A mutation. sponding to the TKD activation loop (D835 and D836) FLT3 was the wild type. He was given 7+3 induction and and the gate keeper position (D691), was performed in achieved remission but relapsed 24.5 months afterward

6 Cancer Month 0, 2019 Sorafenib and Omacetaxine in FLT3-ITD AML/Zhang et al

Figure 2. Molecular monitoring based on FLT3-ITD and minimal residual disease monitoring based on NPM1 mutations among patients who showed morphologic remission upon SOME treatment. (A) Fifteen patients (red) achieved FLT3-ITD VAF ≤ 0.1% at DR, whereas 12 patients (blue) had VAF > 0.1%. (B) OS and (C) LFS of patients based on the FLT3-ITD VAF at DR. (D) Nine patients (red) achieved NPM1 VAF ≤ 0.01% at DR, whereas 7 patients (blue) showed VAF > 0.01%. (E) OS and (F) LFS of patients based on the NPM1 VAF at DR. The detection limits of the assays for the FLT3-ITD and NPM1 mutant were 0.1% and 0.001%, respectively, as shown by the gray area in panels A and D. Statistical significance was calculated with the log-rank test. DR indicates deepest response; FLT3-ITD, internal tandem duplication of Fms-like tyrosine kinase 3; ITD, internal tandem duplication; LFS, leukemia-free survival; NPM1, nucleophosmin 1; OS, overall survival; Pre, before treatment; SOME, sorafenib and omacetaxine mepesuccinate; VAF, variant allelic frequency.

when he presented with severe multilobar pneumonia with time, he was recruited into the SOME trial and (Fig. 3A). Repeat molecular testing showed that his AML achieved CR after 1 course. Both the FLT3-ITD VAF and became FLT3-ITD+. In view of profound neutropenia the NPM1 VAF became undetectable before he received due to persistent leukemia that was unlikely to improve HSCT (Fig. 3C). His neutrophil count recovered, and his

Cancer Month 0, 2019 7 Original Article

TABLE 3. Summary of Adverse Events

Adverse Effects Grade No. of Patients %

Grade 3 or higher Neutropenia 3 17 39 Thrombocytopenia 4 16 36 Hematuria 3 1 2 Grade 1 or 2 Fever 1 23 52 Rash, maculopapular 2 14 32 Anemia 2 8 18 Hypotension 2 2 5 Bone pain 1 2 5 Diarrhea 1 2 5 Headache 1 1 2 Hypocalcemia 1 1 2 Hypokalemia 1 1 2 Nausea 1 1 2 Oral hemorrhage 1 1 2 Palpitations 1 1 2 pneumonia resolved (Fig. 3B). He underwent allogeneic Figure 3. The modest side effects of SOME have made it suitable for patients unfit for conventional treatment. This figure HSCT in September 2018 and has remained in remission presents a patient (UPN 2251) who received SOME at leukemia since then. relapse with severe pneumonia. Computed tomography of the thoracic cavity (A) before SOME and (B) after 3 courses of SOME is shown. (C) Serial changes in the FLT3-ITD VAF (blue; DISCUSSION left y-axis) and NPM1 VAF (red; right y-axis) during the course of treatment are shown. FLT3-ITD indicates internal tandem We have reported SOME to be a safe and effective treat- duplication of Fms-like tyrosine kinase 3; NPM1, nucleophosmin ment regimen for FLT3-ITD AML. This is predicated on 1; Pre, before treatment; SOME, sorafenib and omacetaxine mepesuccinate; UPN, unique patient number; VAF, variant the premise that OME inhibits global protein synthesis allelic frequency. and is particularly effective in FLT3-ITD AML, which survives on short-lived signaling proteins.11 The CR/CRi rate for R/R patients was 72%. Four patients with newly diagnosed FLT3-ITD AML in this cohort also achieved bridged to HSCT. The therapeutic efficacy of OME as CR, and another patient achieved CRi. The median LFS an adjunct to sorafenib also appeared superior to that of and OS for the complete responders in this cohort were another low-intensity treatment. Specifically, azacitidine 5.6 and 10.9 months, respectively. Sixteen patients were in combination with sorafenib has been shown to induce bridged to allogeneic HSCT, and the achievement of CR and CRi in 16% and 27% of patients, respectively, deep molecular responses was correlated with improved with mostly R/R FLT3-ITD AML with a median du- outcomes. SOME was well tolerated with minimal ration of response of 2.3 months.16 Among untreated organ toxicities. These observations might provide novel patients, CR and CRi/CRp were achieved by 26% and insights into the development of therapeutic strategies for 44%, respectively, with a median duration of response of FLT3-ITD AML. 14.5 months.17 On the basis of these observations, the First, we demonstrated the efficacy of SOME in combination of second-generation FLT3 inhibitors with adult patients with R/R AML. Conventional chemo- OME should be tested and compared in clinical trials to therapy was largely ineffective, with CR/CRi achieved ascertain their clinical efficacy and impact on long-term in only 20% to 30%.6,7,15 FLT3 inhibitor monotherapy survival. induced CR/CRi in 30% to 50% of cases with a dura- Second, a deep molecular response based on tion of remission of 2 to 7 months (Supporting Table 3). NPM1 and FLT3-ITD VAFs as surrogates for MRD Specifically, LFS and OS in this cohort compared favor- could be accomplished in some patients. The achieve- ably with the respective times of 3.0 and 6.2 months for ment of undetectable FLT3-ITD and an NPM1 mutant 9 patients who were treated with quizartinib and 5 and VAF ≤ 0.01% during the course of treatment was associ- 10.2 months for patients who were treated with gilter- ated with longer patient survival. The considerable het- itinib and responded.10 At our institution, SOME has erogeneity in molecular responses among patients was also shown a higher response rate and longer remission intriguing, and the underlying mechanisms have to be than sorafenib monotherapy,12 with more patients being further investigated. Because MRD measurement was not

8 Cancer Month 0, 2019 Sorafenib and Omacetaxine in FLT3-ITD AML/Zhang et al a prerequisite for HSCT in this study, transplant-eligible prolong the remission duration of an FLT3 inhibitor and patients who achieved morphologic CR/CRi underwent OME combination should be further investigated. HSCT regardless of their MRD status. Therefore, the The mechanisms of relapse after SOME are cur- impact of SOME per se on MRD could not be ascer- rently unclear. An emergence of clones carrying FLT3- tained. However, in UPN 2251, whose HSCT was post- TKD mutations was identified in 5 of 22 patients who poned because of multilobar pneumonia, MRD became relapsed after CR/CRi. This might be related to the undetectable with repeated courses of SOME, and this intrinsic resistance of FLT3-TKD clones to sorafenib, suggested that at least in some patients, the combination which is a type II FLT3 inhibitor.22 Intriguingly, all of an FLT3 inhibitor and OME could bring about a deep 3 patients with TKD mutations before treatment achieved molecular response. remission after SOME in this study, and this demon- Third, SOME induction was effective in patients strates the potential ability of OME to overcome TKD- with newly diagnosed FLT3-ITD AML. At present, the mediated resistance. In vitro drug screening based on standard of care for young, fit, and transplant-eligible primary AML samples or cell lines carrying FLT3-ITD patients entails midostaurin with induction and consol- with concomitant TKD mutations showed that OME idation chemotherapy; the latter are associated with sig- was equally effective in FLT3-ITD with or without TKD nificant treatment morbidity and mortality. In this study, mutations.11 Other mechanisms of drug resistance to morphologic remission was achieved in all 5 patients: a sorafenib include changes in intracellular pH,12 protec- significant reduction in MRD occurred in 4 patients, and tion by a leukemia niche, and overexpression of antia- 2 of them remained in remission 13 and 8 months after poptotic proteins.23 HSCT at the time of the writing of this article. Because On the basis of our observations, elderly patients of the lack of organ toxicities, a patient who relapses after or patients with FLT3-ITD who are unfit for induction SOME can be readily salvaged by standard induction che- chemotherapy and those who have experienced failure or motherapy. This occurred for UPN 2660, who achieved relapsed after induction should consider a combination a second CR with standard induction and underwent al- of an FLT3 inhibitor and OME as a safe and effective logeneic HSCT. The high safety profile also allows the induction or salvage regimen. Monitoring treatment re- treatment of elderly and unfit patients hitherto ineligi- sponses by MRD detection should be considered as a ble for conventional chemotherapy. This occurred for re- prognostic indicator. The relative efficacy of different ported patients with severe pneumonia (UPN 2251) and FLT3 inhibitors when they are combined with OME and intracranial hemorrhage (UPN 1560). the therapeutic benefits of adjunctive treatment are cur- Fourth, our data have demonstrated the benefit of rently unclear. Allogeneic HSCT should be considered allogeneic HSCT for patients who have achieved CR/CRi for transplant-eligible patients once CR/CRi has been with SOME. The benefit of allogeneic HSCT for patients achieved, and MRD monitoring is needed to evaluate the with FLT3-ITD AML has been reported.18 However, depth of the response. The optimal post-HSCT mainte- post-HSCT relapse has been frequent. In this cohort, post- nance remains to be determined, and the benefits of FLT3 HSCT maintenance was not incorporated into the study inhibitors should be tested by clinical trials. design, and only 3 recent patients (UPNs 2650, 2251, In conclusion, a combination of an FLT3 inhibi- and 2757) received sorafenib maintenance after HSCT; tor and OME was safe and effective for R/R and newly they were all in remission at the time of the writing of diagnosed FLT3-ITD AML. A deep molecular response, this article. Effective maintenance strategies after HSCT, defined by negative MRD, was achieved in select patients. including FLT3 inhibitors,19 have to be further evaluated. The regimen was tolerable and allowed effective treatment This study has generated a number of testable in patients unfit for conventional chemotherapy. hypotheses. For instance, coexisting IDH2, RUNX1, DNMT3A, and WT1 mutations were associated with FUNDING SUPPORT inferior responses and clinical outcomes after SOME This work was supported by the S. K. Yee Medical Foundation (project treatment. The molecular mechanisms by which these 2151209), the Croucher Foundation, and the Li Ka Shing Faculty of Medicine. mutations might affect the response of FLT3-ITD AML to SOME should be further investigated. In addition, the optimal regimen for transplant-ineligible patients has not CONFLICT OF INTEREST DISCLOSURES Anskar Y. H. Leung is the Li Shu Fan Medical Foundation Professor in been determined. Additional therapeutic agents such as Haematology and receives funding from the foundation’s endowment. The 20,21 venetoclax that might improve treatment efficacy and other authors made no disclosures.

Cancer Month 0, 2019 9 Original Article

AUTHOR CONTRIBUTIONS leukaemia (QuANTUM-R): a multicentre, randomised, controlled, Chunxiao Zhang: Collection of patient samples, analysis of the data, open-label, phase 3 trial. Lancet Oncol. 2019;20:984-997. and writing of the manuscript. Stephen S. Y. Lam: Collection of patient 10. Perl AE, Altman JK, Cortes J, et al. Selective inhibition of FLT3 by samples, analysis of the data, and writing of the manuscript. Garret M. gilteritinib in relapsed or refractory acute myeloid leukaemia: a mul- K. Leung: Collection of patient samples, analysis of the data, and writ- ticentre, first-in-human, open-label, phase 1-2 study. Lancet Oncol. ing of the manuscript. Sze-Pui Tsui: MRD testing and analysis of the re- 2017;18:1061-1075. sults. Ning Yang: MRD testing and analysis of the results. Nelson K. L. 11. Lam SS, Ho ES, He BL, et al. Homoharringtonine (omacetaxine me- Ng: MRD testing and analysis of the results. Ho-Wan Ip: MRD testing pesuccinate) as an adjunct for FLT3-ITD acute myeloid leukemia. Sci and analysis of the results. Chun-Hang Au: MRD testing and analysis of Transl Med. 2016;8:359ra129. the results. Tsun-Leung Chan: MRD testing and analysis of the results. 12. Man CH, Fung TK, Ho C, et al. Sorafenib treatment of FLT3-ITD(+) Edmond S. K. Ma: MRD testing and analysis of the results. Sze-Fai Yip: acute myeloid leukemia: favorable initial outcome and mechanisms Recruitment and treatment of the patients. Harold K. K. Lee: Recruitment of subsequent nonresponsiveness associated with the emergence of a and treatment of the patients. June S. M. Lau: Recruitment and treatment D835 mutation. Blood. 2012;119:5133-5143. of the patients. Tsan-Hei Luk: Recruitment and treatment of the patients. 13. Au CH, Wa A, Ho DN, Chan TL, Ma ES. Clinical evaluation of panel Wa Li: Recruitment and treatment of the patients. Yok-Lam Kwong: testing by next-generation sequencing (NGS) for gene mutations in Recruitment and treatment of the patients. Anskar Y. H. Leung: Design of myeloid neoplasms. Diagn Pathol. 2016;11:11. the experiments, treatment of the patients, analysis of the data, and writing 14. Bill M, Grimm J, Jentzsch M, et al. Digital droplet PCR–based absolute of the manuscript. quantification of pre-transplant NPM1 mutation burden predicts relapse in acute myeloid leukemia patients. Ann Hematol. 2018;97:1757-1765. 15. Levis M, Ravandi F, Wang ES, et al. Results from a randomized trial of REFERENCES salvage chemotherapy followed by lestaurtinib for patients with FLT3 1. Dohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N mutant AML in first relapse. Blood. 2011;117:3294-3301. Engl J Med. 2015;373:1136-1152. 16. Ravandi F, Alattar ML, Grunwald MR, et al. Phase 2 study of azacyti- 2. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classi- dine plus sorafenib in patients with acute myeloid leukemia and FLT-3 fication and prognosis in acute myeloid leukemia. N Engl J Med. internal tandem duplication mutation. Blood. 2013;121:4655-4662. 2016;374:2209-2221. 17. Ohanian M, Garcia-Manero G, Levis M, et al. Sorafenib combined 3. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World with 5-azacytidine in older patients with untreated FLT3-ITD mutated Health Organization classification of myeloid neoplasms and acute leu- acute myeloid leukemia. Am J Hematol. 2018;93:1136-1141. kemia. Blood. 2016;127:2391-2405. 18. Schlenk RF, Kayser S, Bullinger L, et al. Differential impact of allelic 4. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management ratio and insertion site in FLT3-ITD–positive AML with respect to of AML in adults: 2017 ELN recommendations from an international allogeneic transplantation. Blood. 2014;124:3441-3449. expert panel. Blood. 2017;129:424-447. 19. Chappell G, Geer M, Gatza E, et al. Maintenance sorafenib in FLT3- 5. Stirewalt DL, Radich JP. The role of FLT3 in haematopoietic malignan- ITD AML following allogeneic HCT favorably impacts relapse and cies. Nat Rev Cancer. 2003;3:650-665. overall survival. Bone Marrow Transplant. 2019;54:1518-1520. 6. Ravandi F, Kantarjian H, Faderl S, et al. Outcome of patients with 20. Wei AH, Strickland SA Jr, Hou JZ, et al. Venetoclax combined with FLT3-mutated acute myeloid leukemia in first relapse. Leuk Res. low-dose cytarabine for previously untreated patients with acute 2010;34:752-756. myeloid leukemia: results from a phase Ib/II study. J Clin Oncol. 7. Wagner K, Damm F, Thol F, et al. FLT3-internal tandem dupli- 2019;37:1277-1284. cation and age are the major prognostic factors in patients with re- 21. DiNardo CD, Pratz K, Pullarkat V, et al. Venetoclax combined with lapsed acute myeloid leukemia with normal karyotype. Haematologica. decitabine or azacitidine in treatment-naive, elderly patients with acute 2011;96:681-686. myeloid leukemia. Blood. 2019;133:7-17. 8. Stone RM, Mandrekar SJ, Sanford BL, et al. Midostaurin plus chemo- 22. Larrosa-Garcia M, Baer MR. FLT3 inhibitors in acute myeloid therapy for acute myeloid leukemia with a FLT3 mutation. N Engl J leukemia: current status and future directions. Mol Cancer Ther. Med. 2017;377:454-464. 2017;16:991-1001. 9. Cortes J, Khaled S, Martinelli G, et al. Quizartinib versus salvage 23. Leung AY, Man CH, Kwong YL. FLT3 inhibition: a moving and evolv- chemotherapy in relapsed or refractory FLT3-ITD acute myeloid ing target in acute myeloid leukaemia. Leukemia. 2013;27:260-268.

10 Cancer Month 0, 2019