Sequential Azacitidine (Aza) and Lenalidomide (Len) for Patients (Pts) with Relapsed and Refractory (R/R) Acute Myeloid Leukemia (AML): Clinical Results and Predictive Modeling Using Computational Analysis and Serial Genomics

Introduction: Pts with R/R AML have few effective treatment options. Len has activity in myeloid malignancies including AML, and may have additive properties with aza. Based on encouraging data using sequential aza with high dose len in untreated elderly AML pts, we designed a phase 2 pilot study exploring this therapy (tx) in pts with R/R AML.

Methods: Adult pts with R/R AML or myelodysplastic syndromes (MDS) and preserved organ function with a WBC<10,000/μL (hydroxyurea permitted) were eligible. The tx consisted of aza 75 mg/m² d1-7, a bone marrow biopsy on cycle (C) 1 day 8, len 50 mg d8-28, and 2 weeks off tx, constituting a 42-day cycle. Response assessments were made per International Working Group (IWG) criteria. Cycles were repeated if tolerated and in the setting of response. Next generation sequencing (NGS) was performed with Illuimina MiSeq at repeated time points serially for each pt (baseline, C1 day 8, C1 day 42) through multiplex PCR amplification of 548 amplicons targeting mutational hot spots in 49 recurrently mutated genes, and analyzed with SoftGenetics NextGENe software. The UCSC hg19 and GRCh 37 reference genome were used as the reference sequence. Genomic information was entered into computational biology modeling (CBM) software (Cellworks), which generates disease specific protein network maps using PubMed and other online resources. Digital drug simulations of aza and len were conducted by quantitatively measuring drug effect on a cell growth score. Each pt specific protein network map was screened for the extent to which tx reduced simulated disease growth.

Results: 37 pts were eligible and received at least one day of tx. Median age was 73 (18-86); 34 had AML, 3 had high-risk MDS (Table 1). The most common possibly related/related > grade 2 adverse events (AEs) were thrombocytopenia (44%), neutropenia (44%), leukopenia (43%), fatigue (32%), anemia (30%) and febrile neutropenia (16%). During C1, 2 pts required len dose decreases and 11 pts did not receive the entire course of len, for infection, rash, neuropathy, renal insufficiency, GI toxicity or progression.

9 pts did not live to complete cycle 1; all died of AML. The overall response rate (ORR, complete remission [CR] + CR with incomplete count recovery [CRi] + partial remission [PR] + morphologic leukemia free state [MLFS]) was 14/37 (38%); 4 CR/CRi and 10 MLFS. 6 additional pts had blast reductions that did not meet IWG criteria for response. Median number of cycles was 1 (0-5); median time to achieve best response was 1 cycle (1-3). 16 pts were identified at baseline as potential candidates for a 1^st/2^nd allogeneic transplant or donor lymphocyte infusion; 4/16 (25%) achieved this endpoint and 1 remains alive. Median follow up time is 3 years. 36/37 patients have died; 33 from progression and 3 from treatment related causes. 5/36 (14%) died within 30 days of study initiation. Median progression free survival (PFS) was 112 days; 132 for responders and 98 for non responders (p=0.7). 1 year PFS was 7%. Median overall survival was 150 days, 216 for responders and 124 for non responders (P=0.5); 1 year OS was 25%. (Fig 1). No clinical features predicted response (Table 2).

CBM analysis has been performed on 5 pts (2 non responders and 3 responders). None were predicted to respond to len, 4/5 were predicted to respond to aza. 4/5 (80%) were predicted correctly; 3/3 responders and 1/2 non responder. Analysis of the entire cohort is ongoing.

Serial quantitative NGS was performed at multiple time points for most patients; results have been analyzed for 7 (3 responders, 4 non responders). These preliminary results do not reveal a correlation between response and specific mutations. However, there is a trend toward a correlation between increased total number of mutations at baseline and lack of response (p=0.099) (Fig 2). This is being further explored with clonal analysis to measure the heterogenous responses of individual pts over time.

Conclusion: Len has activity in a subpopulation of pts with AML, perhaps with additive value from sequential use with aza; responses are transient but can provide sufficient reduction in disease burden to allow for a transplant, which can result in long term remissions. However, the toxicity of this regimen is significant and typical clinical features are not predictive for response, limiting its widespread use. Therefore other tools, such as CBM or serial NGS, must be employed to select pts most likely to respond.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal