Romidepsin enhances the efficacy of cytarabine in vivo, revealing histone deacetylase inhibition as a promising therapeutic strategy for KMT2A-rearranged infant acute lymphoblastic leukemia

Acute lymphoblastic leukemia (ALL) in infants diagnosed at less than 12 months of age is an aggressive malignancy with a poor prognosis. Rearrangements of the KMT2A gene (KMT2A-r) are present in up to 80% of cases, with 5-year event-free survival (EFS) less than 40%.1 Dose intensive chemotherapy has been incorporated into contemporary treatment regimens; however, this has increased the burden of toxicity during therapy and late effects in survivors.21 There is a desperate need to identify novel therapies to improve outcome.

Histone deacetylase inhibition appears to be a promising therapeutic strategy for KMT2A-r infant ALL, with our recent chemo-genomic profiling identifying the potential for romidepsin. Romidepsin was shown to enhance the in vitro activity of cytarabine, a key component of infant ALL therapy, with an in vivo signal identified when combined with high-dose cytarabine.3 In this study, we investigate the in vivo synergy between romidepsin and cytarabine, determine the in vivo toxicity of this combination, and further explore the effect of romidepsin on the DNA damage-response to cytarabine. All in vivo experiments were approved by the Animal Ethics Committee, Telethon Kids Institute, Perth, Australia.

To determine drug toxicity and efficacy upon completion of therapy, 7-week old female NOD/SCID mice were inoculated with 1x10 PER-785A cells. PER-785A is a genetically characterized cell line harboring the t(4;11) translocation.3 For all in vivo studies, drug treatment was commenced when the percentage of human CD19 or CD45 cells reached 1% in the bone marrow (BM), identified from extensive mapping of the leukemia cell kinetics for each model (Online Supplementary Figure S1). Treatment was scheduled to mimic the concepts of contemporary clinical trial design. Currently, novel agents are being investigated following induction therapy on the Interfant chemotherapy backbone,4 thus are introduced in the setting of low disease burden rather than at diagnosis where overt disease is evident in the BM. Mice were randomized into five groups of five mice and drug treatment was commenced on day 12 by intraperitoneal (i.p.) injection. Treatment groups comprised of vehicle control; 1.5 mg/kg romidepsin twice/week on Mondays and Thursdays; low-dose cytarabine (5 mg/kg) five times/week Monday to Friday; high-dose cytarabine (100 mg/kg) five times/week Monday to Friday; combination therapy with 1.5 mg/kg romidepsin and low-dose cytarabine. Treatment was administered for three weeks. Three days following completion of therapy mice were sacrificed and leukemia burden was determined by measuring the percentage of human CD19 cells in the BM by flow cytometry with anti-human CD19-APC antibody. At the time of sacrifice, 0.2 mL of blood was obtained from each mouse by cardiac puncture and a complete blood count performed to determine the degree of myelosuppressive toxicity for each cohort. Single-agent activity was observed with the mean percentage of leukemic cells in the BM of 66.6% (P<0.01) for romidepsin and 27.3% (P<0.01) for low-dose cytarabine (Figure 1A). The reduction in leukemic burden was significantly enhanced with combination therapy with a reduction to a mean of 3.8% infiltration (P<0.0001) (Figure 1A). Although treatment with high-dose cytarabine achieved clearance of leukemic cells from the BM, mice treated with high-dose cytarabine developed severe myelosuppression in comparison to the other cohorts (Figure 1B). In particular, there was a statistically significant reduction in mean hemoglobin (98 vs. 42.5 g/L; P<0.0001), white blood cell (2.43 vs. 0.13x10/L; P<0.0001) and platelet (757 vs. 294x10/L; P<0.0021) count between the mice treated with romidepsin and low-dose cytarabine combination therapy compared to those treated with high-dose cytarabine.

Figure 1.Romidepsin and low-dose cytarabine combination therapy reduces leukemia burden and improves survival in mice. (A and B) Mice injected with PER-785 leukemia cells were treated with vehicle control, 1.5 mg/kg romidepsin (ROM) twice weekly, low-dose (5 mg/kg) cytarabine (AraC) five times a week, high-dose (100 mg/kg) AraC five times a week, or a combination of 1.5 mg/kg ROM and low-dose AraC for three weeks. Mice were sacrificed three days following completion of therapy (n=4-5 mice/group). (A) Leukemia burden in the bone marrow and (B) Complete blood count measurements. (C) Kaplan-Meier survival curves of mice injected with PER-785, MLL-5, or MLL-14 leukemia cells that were treated with vehicle control, 1.5 mg/kg ROM, 5 mg/kg AraC, or combination of 1.5 mg/kg ROM and 5 mg/kg AraC for three weeks (n=7-8 mice/group). Two-tailed unpaired Student t-test was used to analyze leukemia burden and complete blood count measurements, and log-rank test for Kaplan-Meier survival analysis. Error bars represent mean±Standard Deviation. **P<0.01; ***P<0.001; ****P<0.0001.

Three xenograft models, PER-785, MLL-5 and MLL-14, were used to determine the response to drug treatment by EFS. MLL-5 and MLL-14 are well characterized patient-derived xenografts which harbor t(10;11) and t(11;19) translocations respectively.5 MLL-5 and MLL-14 were selected to test whether findings could be validated in independent models with distinct translocation partners. For MLL-5, drug treatment commenced 11 days following injection of 1x10 cells, and for MLL-14, treatment commenced seven days following injection of 2x10 cells, corresponding to 1% of human CD45 cells in the BM. For each xenograft model, mice were randomized into four groups of eight mice prior to commencing therapy by i.p. injection. Given that the premise was to identify novel agents that would abrogate the toxicity of conventional chemotherapeutic agents, high-dose cytarabine was not further investigated. The rest of the treatment groups and schedules remained the same. Individual mouse EFS was calculated as the time in days from treatment initiation until mice reached a humane end point with evidence of leukemia-related morbidity. For MLL-5 and MLL-14, the anti-human CD45-PE antibody was used to determine the percentage of human CD45 cells by flow cytometry. EFS curves were compared using log-rank test.

Compared to vehicle control, a significant survival benefit was seen with low-dose cytarabine alone for PER-785 and MLL-5 but not for MLL-14 (Figure 1C and Online Supplementary Table S1). This finding of differential sensitivity to cytarabine was validated in vitro (Online Supplementary Table S2). Our models thus provide representation of the known inter-patient variability in response to conventional chemotherapeutic agents. A significant survival advantage was not demonstrated for romidepsin monotherapy; however, combination therapy with romidepsin and low-dose cytarabine resulted in a profound and highly significant increase in median survival in all three xenograft models, compared to both vehicle control and low-dose cytarabine alone (Figure 1C and Online Supplementary Table S1). As expected, at the time the mice succumbed to leukemia, the disease burden was similar in all treatment groups (Online Supplementary Figure S2).

Next, we determined whether romidepsin enhanced the DNA-damage response of cytarabine in six of our previously characterized infant ALL cell lines.3 Phospho-H2A.X levels were determined by flow cytometry and western blotting (Online Supplementary Methods). For all cell lines, romidepsin was shown to significantly enhance the DNA double-strand break response of cytarabine (Figure 2A and B and Online Supplementary Figure S3). However, between cell lines there was variability in the extent to which DNA-damage was enhanced; this may reflect a potential for differences in the magnitude of clinical effect between patients. Romidepsin was not shown to increase the expression of other components of the DNA-damage signaling pathway (Online Supplementary Figure S4) or increase the inhibition of DNA repair (Online Supplementary Figure S5) when combined with cytarabine.

Figure 2.Romidepsin enhances the DNA double-strand break response of DNA damaging chemotherapeutic agents. (A and B) Leukemia cell lines were treated with 2 nM romidepsin (ROM), 1 μM cytarabine (AraC), or combination of 2 nM ROM and 1 μM AraC for 24 hours (h). (A) Cells were stained with PI and phospho-H2A.X and analyzed by flow cytometry. (B) Western blot analysis of phospho-H2A.X protein levels in the treated cell lines (left). Bar graphs depict phospho-H2A.X/β-actin densitometry ratios that were normalized to ratios calculated for non-treated controls (right). (C) Leukemia cell lines were treated with 2 nM ROM, 30 nM Daunorubicin (DNR), or combination of 2 nM ROM and 30 nM DNR for 24 h. (D) Leukemia cell lines were treated with 2 nM ROM, 2 μM 4-hydroperoxycyclophosphamide (4HPC), or combination of 2 nM ROM and 2 μM 4HPC for 24 h. (C and D) Western blot analysis of phospho-H2A.X protein levels in the treated cell lines (top). Bar graphs depict phospho-H2A.X/b-actin densitometry ratios that were normalized to ratios calculated for non-treated controls (bottom). Data were collected from three independent experiments and analyzed using the two-tailed unpaired Student t-test. Error bars represent mean±Standard Deviation. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

Using our panel of infant ALL cell lines, we subsequently determined whether romidepsin exhibited synergy with other conventional agents used in the Interfant chemotherapy backbone. In vitro synergy experiments were performed as previously described, with synergy scored using Chalice software, applying the Bliss-independence model.3 Synergy was demonstrated across all six cell lines for romidepsin combined with daunorubicin or 4-hydroperoxycyclophosphamide (Table 1). Romidepsin was also shown to significantly enhance the DNA double-strand break response of daunorubicin and 4-hydroperox-ycyclophosphamide at the protein level (Figure 2C and D). Mixed responses across the cell line panel were observed when romidepsin was combined with vincristine (synergy in four cell lines) or dexamethasone (synergy in three cell lines), whereas antagonism was observed in four out of six cell lines when romidepsin was combined with methotrexate or L-asparaginase (Table 1). Our data provide support for romidepsin to be scheduled prior to or during cytarabine-and cyclophosphamide-based therapy within the current standard Interfant chemotherapy backbone, namely protocol IB and the second half of OCTADAD.

Table 1.Total and maximum in vitro synergy scores between romidepsin and conventional chemotherapy agents.

There is a desperate need to identify novel agents for integration into up-front therapy to improve the dismal outcome of infants with KMT2A-r ALL. The KMT2A-fusion protein has altered histone-methyltransferase activity with characteristic changes to chromatin and transcriptome.6 Several components of the KMT2A-complex can be targeted by novel agents. However, in order to integrate any novel drug into infant therapy protocols, the drug needs to be available in a preparation suitable for administration to infants, i.e. as an oral suspension or formulated for intravenous use. Furthermore, current infant ALL protocols contain up to ten different chemotherapeutics, given in an intricate drug delivery scheme over 24 months. Evidence is required as to where in the standard chemotherapy backbone novel drugs can be safely administered with maximal effect. Simultaneous administration with a conventional chemotherapeutic agent would only be considered for clinical use if the drug combination achieves more than each drug individually, with a tolerable toxicity profile, and if antagonism can be excluded. Our study focused on KMT2A-r infant ALL; however, use of romidepsin is not specific for this indication, with current extensive clinical investigation in combination regimens for several adult hematologic malignancies, including anthracycline based-therapies, ICE (ifosfamide, carboplatin and etoposide) and gemcitabine-containing regimens.7 Our study provides strong evidence that romidepsin can be combined to augment the effect of the conventional chemotherapeutic agent, cytarabine, in infants with KMT2A-r ALL without undue myelosuppression, which is one of the predominant dose-limiting side effects of chemotherapy. The potential for histone deacetylase inhibition in ALL has also been highlighted in several recent studies,108 including strong in vivo efficacy of panobinostat against human KMT2A-r ALL.8 Taken together, there is mounting evidence to support the use of histone deacetylase inhibitors in the next generation of clinical trials for infants with KMT2A-r ALL.