Phase I study of the LSD1 inhibitor tranylcypromine (TCP) in combination with all-trans retinoic acid (ATRA) and low-dose cytarabine (LDAC) in elderly, medically non-fit patients with AML or high-risk MDS (TRANSATRA trial)

Background Elderly, non-fit AML patients (pts) failing hypomethylating agent (HMA) based treatment because of secondary resistance represent a highly unmet clinical need, necessitating development of age-appropriate therapies with an acceptable safety profile. Histone demethylase LSD1 (KDM1A) is a rational pharmacological target in AML, particularly when LSD1 inhibition is combined with retinoic acid (Schenk et al., Nat. Med. 2012). To pursue this hypothesis clinically, others (Wass et al., Leukemia 2021, Tayari et al., Clin. Cancer Res. 2021) as well as ourselves have embarked on phase I trials with tranylcypromine (TCP), a first generation LSD1 inhibitor, in relapsed/refractory AML pts. Patients, Materials and Methods Primary objective of this trial (NCT02717884) is the determination of the maximum tolerated dose (MTD) of TCP, in combination with ATRA and low-dose cytarabine (LDAC), in AML/MDS pts who failed HMA treatment (or no standard therapy available). Secondary objectives: to determine overall safety and tolerability, objective response rate (ORR), disease stabilization rate, survival, and quality of life (QoL). Using the rolling-six design, four dose levels (DLs) of TCP (20 mg, 40 mg, 60 mg, 80 mg p.o. on days 1-28), with a ramp-up dosing (in 10-mg steps q 4 days) were examined in combination with ATRA (45 mg/m2 p.o. on days 10-28) and LDAC (40 mg s.c. on days 1-10), based on own in vitro data demonstrating cooperative effects of this 3-drug combination (Barth et al., Leukemia 2019). The MTD of TCP in combination with ATRA and LDAC (determined at day 28 of cycle 1, with > 80% of ATRA and LDAC administered, and no relapse/progression) was defined as the highest level that results in DLT in < 1 of 6 pts. Safety was described by Medical dictionary for regulatory activities (MedDRA) coding of (serious) adverse events ((S)AEs). Responses were scored according to Cheson et al. (J. Clin. Oncol. 2003). QoL was captured with the EORTC-QLQ-C30, depression with the HADS instrument (administered before treatment start and on day 28 of cycle 1). TCP serum levels were determined on days 10, 20, 28 and 56. To determine treatment-induced transcriptome changes in vivo, CD34+ CD117+ peripheral blood blasts were serially isolated, and subjected to RNA-sequencing. Results Between August 2015 and May 2020, 25 pts (56% male, 23 AML, 2 MDS) were enrolled. Their median age was 75 years (range, 62-84), 5 pts were > 80 years of age. ECOG performance status was 0 / 1 / 2 in 4%, 56% and 40% of pts, respectively. ELN 2010 risk of AML pts (87% relapsed/refractory, 13% untreated) was favorable, int-I/II or unfavorable in 21.7%, 39.1% and 39.1% of pts, respectively, 43.5% had secondary AML. Both MDS pts had high-risk disease by IPSS-R. The median blast count was 15% and 46% in peripheral blood and bone marrow, respectively (ranges, 0-89; 5-95). Pts had received a median of 2 prior treatment lines (range, 1-5), with HMAs in 23 of 25. Three, 6, 6 and 10 pts were allocated to DL 20 / 40 / 60 / 80 mg, respectively, with 3 pts on each DL being fully evaluable for MTD. TCP was administered for a median of 39.5 days (range, 11-228), and 30% of pts received > 60 days of study drugs. No DLT was observed on any of the 4 DLs. At the highest DL (80 mg), further recruitment was terminated after the 10th pt (slow accrual). Among all pts who started TCP, adverse events > grade 3 (MedDRA preferred term) occurring in >10% of pts were: thrombocytopenia (45.8%), neutropenia (20.8%), anemia (16.7%), pneumonia (16.7%), dyspnea and febrile neutropenia (both 12.5%). Importantly, neurological, skin and gastrointestinal toxicities were infrequent and of lower grade. Two partial remissions were attained in 2 of 25 pts; disease stabilization was achieved in 10 of 25 pts. Median OS was 62 days (range, 14-325). RNA-seq analyses in serially acquired purified AML blasts (with up to 4 on-treatment timepoints) identified differentially expressed genes in 5 of 5 pts under treatment. Conclusions In this cohort of elderly, medically non-fit AML/MDS patients resistant to prior HMA treatment, the combination of the oral LSD1 inhibitor TCP with ATRA and LDAC was well feasible, even at the highest TCP dose level. While the ORR was modest, 48% of pts benefited by attaining partial remission or stable disease with this 3-drug combination, indicating limited cross-resistance with prior azanucleosides treatment in these pts. Studies with more powerful LSD1 inhibitors in this indication appear warranted.