Ex Vivo Drug Sensitivity Assay Correlates with Clinical Response and Identifies Panobinostat and Bortezomib As a Potential Novel Drug Combination for Pediatric AML

Background

Pediatric acute myeloid leukemia (AML) is a rare disease with roughly 600 cases diagnosed in the United States each year with minimal improvement in clinical outcomes over the last few decades. We previously demonstrated that an ex vivo drug sensitivity assay (DSA) can predict clinical response in myelodysplastic syndrome (Spinner et al. Blood Adv 2020). Here we investigated whether the DSA performed on pre-induction pediatric AML samples correlates with clinical response and can identify potent novel drug combinations.

Methods

Pre-induction blood or bone marrow samples were assayed from 20 de novo pediatric AML patients diagnosed at Texas Children's between 5/2015 and 10/2020. All patients consented to research (82% enrolled in clinical trial identifier NCT03568994) and received ADE (Cytarabine, Daunorubicin, and Etoposide), and next-generation sequencing was done as part of clinical care. Risk stratification was per AAML1831 guidelines. Drug sensitivity data was analyzed from 13/20 samples that passed quality control with matched treatment conditions: 9/13 (69%) patients had M1/M2 histology, 3/13 (23%) were M4/M5 and 1/13 (8%) was M7 with a median age of 12.3 years. For the ex vivo DSA, samples were incubated in conditioned media and treated with a single dose of up to 25 unique compounds and up to 149 drug combinations. After 72 hours, changes in tumor blast populations were assessed by flow cytometry using an 11-marker panel to identify blasts. For each treatment condition, drug sensitivity was calculated based on the number of blasts remaining after treatment compared to DMSO control. Clinical response data, including minimal residual disease (MRD) percentage by flow cytometry, and 1-year relapse-free survival (RFS), were correlated with drug sensitivity results. Log odds ratios (OR) were calculated with the Haldane-Anscombe correction. ORs were used to quantitatively measure the association between clinical attributes and the DSA to the clinical response data. For evaluation of ORs, a normalized blast score of 70% viability was used to maximize the separation between high and low drug sensitivity.

Results

Ex vivo drug sensitivity correlated with both MRD (r=0.63) and 1-year RFS (r=0.59) in the de novo patient subset (Fig A). Three patients with an MRD >1% exhibited low ex vivo sensitivity to ADE, and among these 3 patients, 2 did not achieve 1-year RFS. Results from the DSA predicted increased odds of having an MRD >1% compared to demographic and mutational clinical attributes that showed weaker associations with MRD (Fig B).

Of the 77 treatment conditions that were tested in 13 patient samples, Bortezomib in combination with Panobinostat (B/P) was the most efficacious treatment in the DSA, where drug sensitivity ranged from low (>100% blast viability) to high (0% blast viability). Separation of patient samples into two distinct low and high DSA response groups was observed with B/P, whereas ADE and single agents showed a graded distribution (Fig C). Within these response groups, pAML3 showed low sensitivity to ADE in the ex vivo DSA and the patient did not respond to ADE. In contrast, pAML8 showed high sensitivity to ADE ex vivo and the patient responded to ADE treatment. While pAML3 and pAML8 showed similar ex vivo sensitivity to B/P as for ADE (Fig D), pAML4 showed preferential sensitivity to ADE and not B/P, and conversely pAML6 showed sensitivity to B/P and not ADE.

Conclusion

Ex vivo drug sensitivity to ADE correlates with both MRD and 1-year RFS in a cohort of 13 de novo pediatric AML patients. These results suggest that clinical response in pediatric AML may be assessed prior to treatment using an ex vivo drug sensitivity assay. Compared to demographic and mutational clinical characteristics queried, ex vivo drug sensitivity to ADE has the potential to be a more predictive measure compared to clinical attributes alone. Combining genomics with functional ex vivo drug sensitivity data could further enhance precision medicine and biomarker discovery in pediatric AML. The DSA also highlights Bortezomib/Panobinostat as a potential novel drug combination for pediatric AML, and the ability to identify a patient sample that is insensitive to ADE and sensitive to Bortezomib/Panobinostat ex vivo supports the use of the DSA to not only predict clinical response but also to possibly inform treatment decisions for pediatric AML patients.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal