2.1 Eligibility criteria and study design

The study (NCT01132586) was approved by the Institutional Review Board of The Ohio State University, conducted in accordance with the Declaration of Helsinki, and all patients provided written informed consent. Eligible patients were adults aged 18 to 65 with a confirmed pathologic diagnosis of AML or high‐risk MDS which had recurred or was refractory to at least one prior line of therapy. Patients who could not attain CR after one round of induction therapy containing anthracycline plus cytarabine (100‐200 mg/m² every 24 hours for 7 days) were considered to have refractory disease and were eligible for this study. High‐risk MDS was defined as International Prognostic Scoring System (IPSS) score of 1.5 or higher. ²² Cytogenetic risk for AML patients was defined per European LeukemiaNet 2017 criteria. ²³ Patients with secondary AML and therapy‐related myeloid neoplasms were eligible. Patients were required to have an Eastern Cooperative Oncology Group (ECOG) performance status of zero to two, and adequate organ function including a total bilirubin <2 mg/dL, ALT and AST <2.5× upper limit of normal, left ventricular ejection fraction ≥40%, creatinine <2 mg/dL and calculated creatinine clearance ≥50 mL/min. Patients who had previously received lenalidomide, cytarabine and/or idarubicin were eligible provided that the combination of the three agents had never been administered, and that no lenalidomide had been administered for at least 6 months. Patients were excluded if they had active central nervous system disease, uncontrolled systemic infections, and if any chemotherapy or radiotherapy within 2 weeks prior to enrollment. All participants had to agree to practice appropriate contraception. Hydroxyurea was permitted until day 5 if necessary to maintain white blood cell count (WBC) <40 000/μL.

Patients were treated with 21‐day course of oral lenalidomide in addition to 96‐hour continuous intravenous (IV) cytarabine infusion administered on days 5‐8 and idarubicin administered IV as 1‐hour infusion on days 5, 6 and 7 (Figure 1A). Lenalidomide dose was escalated in separate cohorts using a classic 3 + 3 phase one design to determine the MTD with this combination. No intrapatient dose escalation was allowed. After establishment of the MTD, a planned 12‐patient expansion group was included to better estimate the tolerability of this dose and to provide a preliminary efficacy assessment. All patients underwent bone marrow biopsy at baseline, on day 5 after single‐agent lenalidomide therapy, and on day 28 after the completion of induction cycle. Response was assessed according to the International Working Group criteria for AML and MDS. ²⁴ , ²⁵ Patients who achieved CR were evaluated for allogeneic HCT or a single cycle of consolidation therapy on this trial, which consisted of oral lenalidomide dosed as during the induction, cytarabine on days 5‐7, and idarubicin on days 5 and 6.

FIGURE 1

A, Study schema for 21‐day re‐induction cycle in relapsed/refractory acute myeloid leukemia or high‐risk myelodysplastic syndrome cohort. Patients underwent bone marrow (BM) biopsy prior to study entry and on day 5 after single‐agent lenalidomide therapy. Cytarabine was started on day 5 and administered as 96‐hour continuous infusion. Idarubicin (Ida) was given intravenously once a day on days 5, 6, 7 and each day the infusion lasts 1 hour. The table shows dosing schema for the study drugs at dose levels −1, 1 and 2. B, Best response to trial therapy according to dose level. CR, complete response; CRi, CR with incomplete count recovery. C, Kaplan–Meier overall survival and relapse‐free survival curves of all patients treated on this study, as well as comparison between patients who achieved remission on this study (responders) and patients who were refractory to study regimen (non‐responders) [Color figure can be viewed at wileyonlinelibrary.com]

2.2 Safety assessment

Adverse events were graded per the Common Terminology Criteria for Adverse Events (version 4.0). The DLT was defined during cycle one of therapy. It is recognized that drug‐related toxicity in this population may be difficult to ascertain given the aggressive nature of the disease. Nonhematologic grade four toxicities were defined as DLTs if they were drug‐related. Toxicity attributed to any of the three agents was considered dose‐limiting. Hematologic DLT was defined as failure to recover absolute neutrophil count (ANC <500/μL) or platelet count (<25 000/μL) by day 42 in the absence of persistent disease. If the DLT occurred in two or more patients at a single dose level, then that dose was deemed intolerable and the next lower dose level was expanded. Given the high prevalence of infectious complications associated with cytopenias, infection‐related complications were not considered DLT unless severity or duration was longer than expected with conventional treatment. Alopecia, nausea and vomiting controllable with antiemetic therapy, line‐associated venous thrombosis, transient and correctable electrolyte abnormalities were also not considered DLT.

2.3 Mutational analysis

Viably preserved patient bone marrow and blood samples were sequenced using targeted amplicon sequencing on the HiSeq (Illumina) platforms to detect common myeloid mutations as described before. ²⁶ Genomic DNA was isolated from viably preserved patient bone marrow and blood samples using the QIAmp DNA Mini Kit (QIAGEN, Hilden, Germany). Genomic DNA fragmentation was done via focused ultrasonication (Covaris, Woburn, Massachusetts). The NGS libraries were prepared using a KAPA HyperPlus Kit (Roche, Pleasanton, CA). Target Enrichment was performed with xGen Lockdown Probes (IDT, Coralville, IA). Libraries were sequenced using the Illumina HiSeq 4000 (Illumina, San Diego, CA). Sequenced reads were aligned to the hg19 genome build using the Burrows‐Wheeler Aligner (BWA). Picard Tools was used to perform UMI‐consensus calling on the aligned reads. The Genome Analysis Toolkit (GATK) was used to realign insertions and deletions in the aligned reads and to perform base quality score recalibration for those realigned regions. Also GATKʼs MuTect2 was used to perform variant calling. After variant calling, variants were annotated using SnpEff and vcfanno along with the dbsnp, COSMIC, 1000 genomes, and 6500 exomes variant databases. The Mucor3 algorithm was used as the baseline for integrative mutation assessment. All called variants underwent visual inspection of the aligned reads using the Integrative Genomics Viewer. Variants in regions of high discrepancy, low quality, tandem repeats, or mononucleotide runs were excluded. Variants that were annotated as a likely germline variant by 1000 genomes or 6500 exomes and visually suspected of being germline were also removed. We used variant allelic frequency (VAF) ≥5% as cutoff to call a mutation. Both NPM1 and CEBPA mutation results were confirmed in a clinical PCR/Sanger assay. The FLT3‐internal tandem duplications (FLT3‐ITD) were evaluated using capillary electrophoresis fragment analysis. Pretreatment and posttreatment samples were analyzed similarly and monitored for changes in VAF.

2.4 Outcomes and statistical analysis

The primary objective was to determine the MTD of lenalidomide in combination with conventional chemotherapy in patients with R/R AML and high‐risk MDS, and recommend starting doses for phase two studies of this combination. Secondary objectives were to define toxicities associated with this combination and assess therapeutic response.

Data are presented with percentage proportions for categorical variables and medians for continuous variables. The chi‐square test and Fischerʼs exact test were used to compare categorical variables. Survival estimates were calculated using the Kaplan‐Meier method and differences between curves were assessed using the log‐rank test. The overall survival (OS) time was calculated from the time point of study entry until death or last follow‐up. Relapse‐free survival (RFS) was calculated from the time point of CR/CRi until the time of relapse, death or the last follow‐up. Statistical analyses were performed using JMP software v.14.0.0 (SAS Inc, Cary, NC) and P ≤ .05 was considered statistically significant.

3.1 Toxicity profile and MTD assessment

Toxicity data were collected for 33 patients and ≥grade three non‐hematologic adverse events are shown in Table 2. A total of eight DLTs (24%) were observed with four being hematologic DLTs. At dose level one, three patients were initially enrolled, of whom one of them experienced hematologic DLT (neutropenia). Three more patients were enrolled at this dose level, one had grade four confusion (requiring intensive‐care unit stay) which was likely multifactorial and not drug‐related. We subsequently enrolled three patients at dose level 2, one had prolonged grade four neutropenia and one patient died on day 5 of the study due to pulmonary hemorrhage after bronchoscopic procedure. Three more patients were enrolled at this dose level, one developed heart failure symptoms due to new onset reduced left ventricular ejection fraction, thought possibly due to anthracycline toxicity. Therefore, the dose was reduced back to dose level 1 with three more patients enrolled, two of whom had prolonged grade four neutropenia and one patient died from intracranial hemorrhage. This led to further de‐escalation to dose level −1, which delivered the same lenalidomide dose (25 mg daily) as the previous dose level, but both idarubicin and cytarabine were dose‐reduced (see Figure 1). The initial six patients enrolled at this level did not experience DLT, therefore this dose level was determined to be the MTD and recommended to pursue for the phase II dose level. Twelve subsequent patients were enrolled for pre‐planned dose expansion to further assess safety and tolerability. Only one patient experienced serious adverse event (SAE) with acute renal failure, septic shock and respiratory failure.

TABLE 2

Summary of grade 3 or higher non‐hematologic toxicities according to dose level

Toxicity description	Frequency				Any grade ≥3 toxicity by dose level
	Grade 3 n (%)	Grade 4 n (%)	Grade 5 n (%)	Total n (%)	Level 2 (N = 6)	Level 1 (N = 9)	Level –1 (N = 18)
Febrile neutropenia	10 (30)	0	0	10 (30)	3	3	4
Rash	8 (24)	0	0	8 (24)	1	2	5
Sepsis	1 (3)	2 (6)	3 (9)	6 (18)	2	2	2
Elevated creatinine	3 (9)	1 (3)	0	4 (12)	2	1	1
Diarrhea	3 (9)	0	0	3 (9)	0	1	2
Confusion	1 (3)	1 (3)	0	2 (6)	0	2	0
Reduced LVEF	0	1 (3)	0	1 (3)	1	0	0
Intracranial hemorrhage	0	0	1 (3)	1 (3)	0	1	0
Pulmonary hemorrhage	0	0	1 (3)	1 (3)	1	0	0
Hypertension	1 (3)	0	0	1 (3)	0	1	0
Atrial fibrillation	0	1 (3)	0	1 (3)	0	0	1

Abbreviation: LVEF, left ventricular ejection fraction.

The most commonly observed non‐hematologic SAEs (ie, ≥grade three toxicities) were febrile neutropenia (30%), rash (24%), sepsis (15%) and acute kidney injury (12%). Death from any cause within 30 days of trial entry occurred in five (15%) of 33 patients. Death was a result of sepsis in the setting of active leukemia in three patients, pulmonary hemorrhage in one patient and progressive disease in one patient.

Overall, 15 out of 33 (45%) of patients were able to receive pre‐planned 21 days of lenalidomide therapy. In 18 patients, lenalidomide was held due to side effects with a median duration of therapy of 13 days (range, 5‐19 days of treatment). Among 18 patients treated at MTD level, eight (44%) were able to receive 21 days of treatment, while 10 patients had a median treatment duration of 14 days (range, 10‐19 days).

3.2 Clinical responses

Twenty‐nine out of 33 enrolled patients were evaluable for efficacy (Figure 1B). Response could not be assessed in four patients due to complications and mortality within the first 7 weeks of enrollment, with inability to perform a bone marrow disease assessment. Eighteen patients achieved CR (n = 14) or CRi (n = 4) with a total CR/CRi rate of 56% in the entire cohort. Among three MDS patients, one achieved CRi, one was refractory, and response could not be assessed in the other patient due to early death after bronchoscopic procedure. The CR/CRi rate was significantly lower for patients treated at dose level − 1 (MTD) when compared to patients treated at dose levels 1 and 2 combined (39% vs 73%, P = .04). In addition, all six patients who had a long duration of remission before trial enrollment (>1 year) achieved CR/CRi on this study, while CR/CRi rate was 47% for patients with shorter duration of last remission before study treatment (<1 year, n = 19) and 50% for patients with primary refractory disease (n = 6). There were two patients with del(5q), and both of them achieved CR. The median duration of CR/CRi was 5.1 months (range, 1‐93.5 months).

Seven patients enrolled in this study had AML relapse after allogeneic HCT. Among them, five (71%) achieved CR, with response durations of 1, 2.5, 4, 27 and 72 months. One patient experienced skin and liver graft‐vs‐host disease (GVHD) after lenalidomide therapy, requiring initiation of systemic steroids while another patient had mild skin GVHD. Only one patient received donor leukocyte infusion for relapsed disease after study treatment.

The median OS for the entire cohort was 6 months, and the median RFS for patients who achieved CR/CRi was 4.4 months (Figure 1C). The 1‐year and 2‐year OS were 24% and 10%, respectively. Patients who achieved CR/CRi after study treatment (ie, responders) had longer OS when compared to patients who were refractory to this regimen (ie, non‐responders) (median OS, 10 months vs 2 months, P = .001) (Figure 1C). Of the 18 patients who achieved CR/CRi, 10 underwent allogeneic HCT. The median OS and RFS of patients who had HCT were 11.4 and 7.3 months, respectively, and with a 1‐year OS of 40%.

3.3 Molecular profile and response

Next‐generation sequencing for recurrent somatic mutations was performed at study entry in 24 patients (Figure 2A). There was no clear association between mutation patterns and response to therapy. Twelve patients also had serial pre‐treatment and post‐treatment samples available for analysis (Figure 2B). Among seven patients with serial pre‐treatment and CR samples, three patients had no detectable mutations at remission suggesting eradication of mutated myeloid clones at the time of morphologic CR. Duration of response in these patients were 1, 28, and 31 months. An additional three patients had reduction in the size of their pre‐treatment clones and one patient did not have any detectable mutations at any time point. Among five patients who had serial pre‐treatment and post‐treatment refractory disease samples, though changes in the size of the clones was observed, all persisted after therapy.

FIGURE 2

A, Oncoprint showing myeloid mutation profile of study patients stratified based on response to treatment. B, Serial pretreatment and posttreatment samples obtained from study patients were analyzed with targeted next‐generation sequencing panel. Variant allelic frequencies (VAF) of mutations detected before treatment and at the time of complete remission (CR) or persistent disease are illustrated [Color figure can be viewed at wileyonlinelibrary.com]