Keys to drug sensitivity from updated functional work flows

A primary goal of cancer therapy is to match patients with the most appropriate drug regimens. Identifying characteristics of patients who respond to therapies and devising alternative strategies for non-responsive patients are important clinical considerations. Next generation sequencing (NGS) has provided a central technology to reveal genetic alterations and guide this process. Additionally, the development of cancer therapies that target specific signaling pathways and subcellular components has increased the opportunity for matching patients with molecularly targeted drugs. In practice, however, limitations in understanding the relationship between cancer genotypes and their corresponding phenotypes have hindered this process; somatic cancer mutations do not always reliably suggest therapies. Indeed, in some cases, targeted drugs have shown clinical utility when matched to cell phenotypes rather than somatic genotype. In this way, the use of orthogonal technologies, such as functional testing and immune-profiling, integrated with NGS holds promise to improve outcomes by better matching therapies to individual patients.1 In addition, lack of durable efficacy of many categories of therapies has sometimes been attributed to inadequate elimination or targeting of leukemic stem/progenitor cells (reviewed by Rossi et al.2).

In this issue, Majumder et al. report on the manner by which understanding of innate drug sensitivities in healthy hematopoietic cells advances both the identification of lineage-specific anti-cancer therapies as well as off-target drug effects in treating acute myeloid leukemia.3 Underlying this work is the well-characterized biology of hematopoiesis whereby multipotent stem cells and precursors differentiate through distinct signaling pathways to generate a set of blood cell types with discrete phenotypes and functions. The authors surmise that malignant hematopoietic cells use the same signaling pathways; consequently, they leverage specific pathways from normal cells as a means to identify cancer therapies for their malignant counterparts. Conversely, the authors note that drug responses seen in healthy cells may reveal potential adverse effects.

The authors augment their established cell-based screening platform for identifying anti-leukemia drugs4 with high capacity flow cytometry (Figure 1A). This technological development permits the simultaneous evaluation of drug responses from multiple hematopoietic cell populations based on their respective surface antigens. Drug responses are mapped to proteome and cell type specific signaling profiles using mass spectrometry and mass cytometry. In this study, sensitivities to 71 small molecules were simultaneously assessed using multi-parametric flow cytometry and then mapped to proteomic and signaling profiles to characterize the spectrum of drug responses in various hematopoietic cell types. Across healthy cell types for B cells, natural killer (NK) cells, helper T cells, cytotoxic T cells and monocytes, the authors identify cell lineage-specific drug responses to define a global view of response profiles. By comparing drug responses between healthy and neoplastic cells, they show that healthy cell responses predict drug responses in corresponding malignant cells. The authors evaluate this screening approach on a large cohort of primary samples obtained from healthy donors and patients with myeloid and lymphoid leukemias, providing evidence that this method identifies new applications for the tested drugs.

Figure 1.High-throughput flow cytometry functional screening strategy identifies drug effects on discrete cell lineages. (A) Schematic for screening workflow whereby cells from leukemia patient samples are exposed to a library of drugs and then subjected to flow cytometry using antibodies to distinguish discrete cell populations, such as T cells (CD3), B cells (CD19), monocytes (CD14), hematopoietic stem/progenitor cells (CD34), etc. (B) Results from the workflow in (A) have shown that drugs have differential effects on cell lineage states, which are often conserved between malignant and healthy settings. As an example, the BCL2 inhibitor, venetoclax, was shown to have more effect against less differentiated cells as well as mature B cells and less impact on mature monocytes and granulocytes in cells from healthy donors. This observation extended to these same cell states in leukemia patient samples, where venetoclax was more effective in killing mature B cells in chronic lymphocytic leukemia and leukemic progenitor cells in acute myeloid leukemia (AML), but was less effective against malignant monocytes in the same AML samples.

A key highlight of this study is the profile observed for the BCL2 inhibitor, venetoclax, which revealed dose-dependent sensitivities across the hematopoietic cell types (Figure 1B). At the ends of this spectrum, B cells (CD19) were the most sensitive whereas monocytes and granulocytes were the least sensitive to venetoclax. Moderate sensitivities were observed on cytotoxic and helper T cells (CD3CD4 and CD3CD4), NK cells (CD56), and NK-T cells (CD3CD56). Venetoclax had similar cell-specific effects regardless of disease status (healthy vs. malignant) indicating the variable nature of response to venetoclax is lineage specific. In addition, the study found an inverse relationship between venetoclax sensitivity and levels of phosphorylated STAT3. Monocytes and granulocytes have the highest levels of phosphorylated STAT3 and the lowest venetoclax sensitivity, perhaps reflecting the different transcriptional programs defining these two cell types.

Previous work by these authors and others indicated that BCL2 is differentially expressed in subpopulations of AML cells enriched for malignant stem/progenitor cells compared to more differentiated tumor cells5 and that venetoclax sensitivity in primary AML cells with a monocytic phenotype is reduced.6 These observations led to the hypothesis that clinical features of AML indicative of myeloid differentiation status may correlate with reduced BCL2 dependence in AML patients. Indeed, the venetoclax response profile in this study is consistent with recent findings correlating venetoclax sensitivity with stages of AML disease differentiation as defined by flow cytometry.7 In the context of venetoclax-based therapies, phenotypically primitive AML is sensitive whereas monocytic AML is more resistant, due to intrinsic properties of monocytic AML cells including loss of BCL2 expression and reliance on MCL1 to mediate oxidative phosphorylation and survival.

Cumulatively, these findings raise the possibility of new definitions for stem/progenitor cells in hematologic malignancies; definitions that would be based on propensity of cell types of any maturation state to persist in the face of a selective pressure. In some ways, these findings may also call into question the long-held notions that targeting of more primitive leukemic cell populations will hold the key to durable disease control. It seems possible that most cancer therapeutics utilized to date are simply more active against more mature cell types based on targeting of biological programs that are more prominent in these more mature cells. The identification of drugs that show inversely preferential activity against the more primitive cell states is an example of the enormous benefit that can be derived from this updated platform for flow cytometric drug sensitivity assessment.

Majumder et al. speculate that their profiling will open new opportunities for other disease indications. As examples, they note that dexamethasone and midostaurin targeted NK cells as effectively as B cells, suggesting their potential clinical use in NK-cell malignancies. They further envision incorporating cell lineage specific drug responses into the regimen for preclinical drug development will identify unexpected therapeutic niches for small molecules and enhanced therapeutic precision. Their study provides additional support for the concept of using multiple diagnostic technologies to enhance precision therapy. Indeed, the use of this technology platform to identify drug combinations that can simultaneously target the undifferentiated leukemic cell populations as well as the more mature myeloid lineages (especially monocytes) may represent a powerful way to prioritize the most promising drug combinations for pre-clinical study and for clinical development. Clearly, this report demonstrates just the beginning of utility of this exciting drug screening platform.