In vivo activity of the dual PI3Kδ and PI3Kγ inhibitor duvelisib against pediatric acute lymphoblastic leukemia xenografts

Acute lymphoblastic leukemia (ALL) remains one of the most common causes of cancer‐related mortality in children. Phosphoinositide 3‐kinases (PI3Ks) are a family of lipid kinases, and aberrations in the PI3K pathway are associated with several hematological malignancies, including ALL. Duvelisib (Copiktra) is an orally available, small molecule dual inhibitor of PI3Kδ and PI3Kγ, that is Food and Drug Administration (FDA) approved for the treatment of relapsed/refractory chronic lymphocytic leukemia and small lymphocytic lymphoma. Here, we report the efficacy of duvelisib against a panel of pediatric ALL patient‐derived xenografts (PDXs).


INTRODUCTION
Acute lymphoblastic leukemia (ALL) is the most common childhood malignancy, and is one of the most common causes of cancer-related mortality in children. 1 While complete remission rates now exceed 95%, this is primarily due to optimization of chemotherapeutic regimens and stratification of patients, rather than the introduction of novel therapies. 1,2 Although the overall survival rate of pediatric ALL patients now approaches 90%, the prognosis for patients who relapse or present with chemoresistant or very high-risk disease remains dismal. 2 Therefore, there is an increasing need to develop novel therapies, particularly for relapsed and high-risk ALL. and PI3Kγ represent two protein isoforms. 3 PI3Ks are essential to signal transduction, relaying signals from cell surface receptors to downstream effectors, including those in the AKT and mammalian target of rapamycin (mTOR) pathways. [4][5][6] The PI3K/AKT pathway is essential for normal hematopoietic stem cell maintenance and the regulation of lineage development, and aberrations in this pathway are commonly associated with many hematological malignancies, including ALL. 5,7 Hyperactivation of the PI3K/AKT/mTOR pathway results in uncontrolled cell growth, which ultimately leads to metastatic competence, angiogenesis, and therapy resistance. 8,9 In particular, the PI3Kδ isoform contributes to hyperactivation of the PI3K/AKT/mTOR pathway leading to uncontrolled proliferation, and plays a crucial role in the development of pediatric B-cell precursor (BCP) ALL, where it correlates with glucocorticoid resistance and poor prognosis. 7,10 Moreover, PI3Kδ contributes to the adhesion of ALL cells to bone marrow stroma cells, triggering intercellular signals that contribute to cell adhesion-mediated drug resistance. 7,11 Conversely, the PI3Kγ isoform is associated with T-lymphocyte development and trafficking, and modulates the tumor microenvironment by increasing leukocyte chemotaxis. 11,12 While some PI3K isoforms are ubiquitous, PI3Kδ and PI3Kγ are preferentially expressed in leukocytes, which makes it possible to target them while minimizing off-target effects. 3 Given the non-overlapping functions of PI3Kδ and PI3Kγ, there is a strong rational for dual inhibition of these kinases. 3,13 Duvelisib (Copiktra) is an orally available, small molecule dual inhibitor of PI3Kδ and PI3Kγ developed by Verastem Oncology, Inc.
for the treatment of hematological malignancies. In 2018, duvelisib obtained Food and Drug Administration approval for the treatment of adults with relapsed or refractory chronic lymphocytic leukemia (CLL), small lymphocytic lymphoma (SLL), and follicular lymphoma. 14,15 Despite these advances, the in vivo efficacy of duvelisib in preclinical models of pediatric ALL is yet to be evaluated. Here, we describe the in vivo efficacy of duvelisib against a broad panel of pediatric ALL patient-derived xenografts (PDXs) with variable PI3Kδ and PI3Kγ expression and mutational status.

Analysis of PI3Kδ and PI3Kγ mRNA expression and mutational status
Whole-transcriptome RNA sequencing (RNA-seq) and whole-exome sequencing (WES) were conducted on 90 ALL PDXs as described previously. 19 These data were interrogated to investigate PI3Kδ (PIK3CD) and PI3Kγ (PIK3CG) mRNA expression, which was qualitatively assessed as low, moderate, or high expressing for each PDX

2.3
Drug treatments and assessment of in vivo drug activity Duvelisib was provided by Verastem Oncology, Inc. and formulated in 5% N-methyl-2-pyrrolidone + 95% polyethylene glycol 400. Tolerability testing was conducted in naïve female NSG mice at the following doses: 12.5, 25, 37.5, and 50 mg/kg, alongside a vehicle control group, with duvelisib or the vehicle administered at a maximum volume of 5 μL/g. Blood was analyzed weekly to monitor hematology and blood biochemistry parameters. For hematology analyses, whole blood from individual mice was diluted 1:2 with phosphate-buffered saline and analyzed using a Beckman Coulter AcT Diff Hematology Analyzer according to the manufacturer's instructions. For biochemistry analyses, whole blood from two to three mice within each treatment group was pooled and analyzed on a Vetscan Biochemistry Analyzer (Zoetis) according to the manufacturer's instructions.
The efficacy study was also conducted in female NSG mice and proceeded at the recommended maximum tolerated dose (MTD) of 50 mg/kg. For the efficacy study, a single mouse analysis 20 was performed whereby each mouse was inoculated with one PDX, and treatment began once the %huCD45 + for each individual mouse reached greater than or equal to 1% in the PB, as measured by flow cytometry. Events were predefined as the %huCD45 + reaching 25% in the PB, or mice exhibiting signs of leukemia-related morbidity with 50% or higher infiltration of at least two major organs. Individual mouse event-free survival (EFS) was calculated as days from treatment initiation until event, or until mice reached a humane endpoint with evidence of leukemia-related morbidity. The EFS was represented graphically by Kaplan-Meier analysis 21 for each subtype, and survival curves of each subtype were compared by the Gehan-Breslow-Wilcoxon test.
Response to treatment was evaluated by three methods: (i) the maximum decrease of the %huCD45 from baseline at any point following treatment initiation; (ii) the area over the curve, as determined by an analysis of %huCD45 + with respect to time; and (iii) objective response measures (ORM) modeled after stringent clinical criteria, which were assessed at day 42 post treatment initiation, as described previously 22 and in the Supporting Methods.

Evaluation of PI3K pathway inhibition via duvelisib ex vivo
ALL-84 cells were plated in triplicate at a density of 5 × 10 6 cells per well in QBSF media and were equilibrated for 3 hours prior to treatment with either QBSF media (untreated), vehicle (DMSO), or 1 μM of duvelisib for 1, 2, or 4 hours. Cells were harvested and fixed/permeabilised with 16% PFA and ice cold 80% methanol. Human Fc receptors were blocked using Human TruStain FcX (BioLegend) prior to staining with phospho-S6 (Ser235/236) APC conjugate antibody (Cell Signaling). Phospho-S6 levels were detected via flow cytometry.
Data were normalized to untreated cells.

PI3Kδ and PI3Kγ mRNA expression and mutational status
We have previously reported the molecular characterization (RNA-  (Table 1).  Initially, duvelisib was administered at a maximum volume of 10 μL/g, which induced severe weight loss across all treatment groups, including the vehicle control, resulting in termination of the study at day 6 ( Figure S1A,B). A second study testing the doses 50, 37.5, and 25 mg/kg, administered at a maximum volume of 5 μL/g alongside a vehicle control, allowed for the successful evaluation of the tolerability of duvelisib. Duvelisib was well tolerated, with only minor weight loss observed across all treatment groups ( Figure S1C,D), and hematology and blood biochemistry analyses did not indicate any acute toxicities (Tables S1 and S2).

In vivo efficacy of duvelisib against ALL PDXs
To assess the in vivo efficacy of duvelisib, each PDX (Table 1) was inoculated into one mouse, and treatment commenced once the %huCD45 + for each individual mouse reached 1% or higher in the PB. Duvelisib was well-tolerated, with a maximum weight loss of 4.2% observed across all PDXs (Table S3) Table 1). Reductions in the %huCD45 were also recorded in four PDXs, with the greatest remission observed in ALL-84, which had a 90.9% reduction in the %huCD45 + from baseline ( Figure 2D, Table S3). While it was predicted that duvelisib would elicit remissions in PDXs with high PI3Kδ or PI3Kγ expression, or with PI3K mutations, only ALL-84 achieved an objective response, scoring a partial response (Table S3). Duvelisib treatment did not induce significant differences in EFS between any of the subtypes ( Figure 2B), nor was there any significant correlation between mouse EFS and PI3Kδ or PI3Kγ mRNA expression (p = .384 and .413, respectively).
As the response to duvelisib treatment corresponded poorly to mRNA expression, we considered that protein expression may be a bet-  and PI3Kγ mRNA expression was not always indicative of protein expression in PDXs and was not associated with treatment response ( Figure S2). Finally, to confirm that duvelisib could inhibit the PI3K pathway in ALL-84, we used ex vivo phosphoflow to measure phosphorylated ribosomal protein S6 (pS6) as a downstream marker of PI3K pathway activation. Treatment of ALL-84 cells with duvelisib (1 μM) resulted in a rapid reduction in pS6 when compared to vehicle-treated cells ( Figure S3). This trend was observed over 1-, 2-, and 4-hour treatment periods, indicating that duvelisib does inhibit the PI3K pathway in ALL-84.

DISCUSSION
Signaling within the PI3K/AKT/mTOR pathway is regarded as a master regulator of cancer, 15  with a reduced splenic leukemic infiltration observed in seven of 10 PDXs, but idelalisib appeared unable to induce remissions in any of the models tested. 28 The PI3K/mTOR inhibitor gedatolisib was also tested by Tasian et al. and invoked a greater response than idelalisib, eliciting substantial in vivo activity in 10/10 Ph-like PDXs. 28 Given that simultaneous inhibition of PI3K and mTOR suppresses the feedback loop that activates AKT, 29