Background

Myelodysplastic syndromes (MDS) are heterogeneous myeloid neoplasms characterized by ineffective hematopoiesis, dysplasia, cytopenias, and a variable risk of progression to acute myeloid leukemia (AML). The biologic heterogeneity of MDS and related myeloid neoplasms relates to the genomic complexity of these disorders with different combinations of cytogenetic abnormalities and mutations associated with distinct clinical phenotypes. Ex vivo drug sensitivity screening (DSS) is a promising tool that may inform personalized therapy in MDS, particularly in patients refractory to standard therapies such as hypomethylating agents (HMAs). We report our updated experience using a fully automated ex vivo DSS platform in 64 patients with MDS and related myeloid neoplasms and identify correlations between clinical and genomic features and ex vivo drug sensitivity.

Methods

Patients: Patients were evaluated at the Stanford MDS Center between September 2016 and May 2020 and had a diagnosis of MDS, MDS/MPN, or secondary AML. Bone marrow (BM) aspirate and peripheral blood (PB) samples were procured for mutation testing (164-gene panel) and ex vivo DSS (Notable Labs, Foster City, CA).

Ex vivo DSS: Fresh BM aspirate and PB specimens were RBC-lysed and resuspended in serum-free media with cytokines as previously described (Spinner et al, Blood Adv 2020;4(12):2768-78). Samples were plated in 384-well microtiter plates and screened against a collection of up to 74 drugs and 36 drug combinations in triplicate. Specimens were treated for 72 hours and assayed using high-throughput, multi-parametic flow cytometry, gating on the blast population (expressing CD34, CD33, and/or HLA-DR) to assess for blast viability.

Patient clustering & statistical analysis: The Euclidean distance metric and Ward minimum variance method were used to identify patient clusters with distinct ex vivo drug sensitivity patterns. A 1-way repeated measures ANOVA was used to identify drug classes with variable sensitivity across clusters and to identify associations between clusters and clinical variables, including blast count, mutations and cytogenetic groups, IPSS-R risk group, and prior HMA exposure. A generalized estimation equation model was used to identify associations between mutations and ex vivo drug sensitivity.

Results

Ex vivo DSS was performed in 64 patients with myeloid neoplasms including 43 with MDS (67%), 11 with MDS/MPNs (17%), and 10 with secondary AML (16%). The median age was 75 years (range 23-90) and 78% were male. The majority of patients had higher risk disease with IPSS-R >3.5 (66%), excess blasts (58%), and adverse cytogenetics or mutations (53%) by IPSS-R or the ELN classification. Patients had a median of 2 pathogenic mutations (range 0-7), with the most frequent including TET2, ASXL1, DNMT3A, SF3B1, RUNX1, STAG2, SRSF2, NRAS, KRAS, BCOR, TP53, and EZH2. The majority of patients (64%) had prior HMA exposure.

Ex vivo DSS defined three distinct patient clusters with differential sensitivity to numerous drug classes (Figure 1). Cluster 1 (N=13) demonstrated the greatest ex vivo sensitivity to HMAs, HMA/venetoclax combinations, cytotoxic agents, kinase inhibitors, mTOR inhibitors, HDAC inhibitors, and PARP inhibitors, while cluster 3 (N=19) demonstrated the greatest ex vivo resistance (p<0.0001 for all comparisons). Correlating clinical variables with drug sensitivity clusters, only IPSS-R score differed significantly among clusters, with fewer higher risk patients in cluster 3 (p=0.02).

Correlating specific mutations with ex vivo drug sensitivity, STAG2 mutations were associated with greater ex vivo sensitivity to HMAs (p=0.002), HMA/venetoclax combinations (p=0.003), kinase inhibitors (p=0.002), and PARP inhibitors (p=0.003). TP53 mutations were associated with greater ex vivo sensitivity to proteasome inhibitors (p=0.0002).

Conclusions

Ex vivo DSS defined distinct patient clusters with differential sensitivity to numerous drug classes. Specific mutations, such as STAG2 and TP53, were associated with greater ex vivo sensitivity to specific drug classes. A larger sample size is needed to evaluate combinations of mutations and better define associations between genotype and drug sensitivity phenotype. Ultimately, combining both genomics and functional screening may further refine personalized therapy selection for patients with MDS and related myeloid neoplasms.

Disclosures

Spinner:Notable Labs: Honoraria. Schaffert:Notable Labs: Current Employment. Aleshin:Notable Labs: Consultancy. Santaguida:Notable Labs: Current Employment. Tada:Notable Labs: Current Employment, Current equity holder in private company. Greenberg:BMS: Research Funding; Aprea: Research Funding; H3 Biotech: Research Funding; Notable Labs: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.