Ponatinib, but not the new Abl-kinase inhibitor asciminib, activates platelets, leukocytes, and endothelial cell TNF signaling to induce atherosclerotic plaque inflammation, myocardial infarction, and stroke

Abstract

Background:

Imatinib, the first Abl-tyrosine kinase inhibitor (TKI), improved leukemia outcomes without cardiovascular side effects. Newer agents, including ponatinib, addressed imatinib-resistance, improving cancer remission, but substantially increased arterial thrombotic events, including myocardial infarction (MI) and stroke. The mechanism behind ponatinib-induced thrombosis and the cardiovascular impact of asciminib, a newly approved Abl-TKI, remain unknown.

Methods:

The impact of clinically relevant plasma concentrations of imatinib, ponatinib, and asciminib were compared to vehicle in vivo using SR-BI-mut/LDLR-knockout (KO) mice to assess spontaneous MI and stroke risk. Mechanism was interrogated in C57BL/6J mice, assessing leukocyte trafficking and thromboinflammation by intravital microscopy (IVM) and flow cytometry, respectively, and in ApoE-KO mice, assessing plaque phenotype by flow cytometry and histology. In vitro impact on human umbilical vein (HUVEC) and coronary artery (HCAEC) endothelial cells was determined by flow cytometry, PCR, and immunoblotting. The role of TNF signaling was evaluated by pharmacologic inhibition and siRNA knockdown.

Results:

In SR-BI-mut/LDLR-KO mice, ponatinib significantly accelerated mortality from MI and stroke compared to vehicle, imatinib and asciminib. In human ECs, only ponatinib increased expression of TNF receptors (TNFR) and adhesion molecules (P-selectin, ICAM1, VCAM1). Ponatinib rapidly induced TNFR2 membrane trafficking and TNF signaling in HUVECs. TNFR inhibition or TNFR2 knockdown prevented ponatinib induction of EC adhesion molecules. In vivo, ponatinib increased mesenteric vessel adhesion molecules, leukocyte rolling and adhesion to vessels, leukocyte and platelet activation, and platelet-leukocyte aggregates. In ApoE-KO mice, ponatinib increased plaque necrotic core and inflammation, consistent with a rupture-prone phenotype. Asciminib-treated mice developed none of these in vitro or in vivo toxicities. In C57BL/6J mice, TNFR inhibition blocked ponatinib-induced mesenteric adhesion molecule expression and leukocyte trafficking, but not platelet-leukocyte aggregation. TNFR blockade prevented ponatinib-induced plaque inflammation in ApoE-KO mice and MI and stroke in SR-BI-mut/LDLR-KO mice.

Conclusion:

Ponatinib, a potent anti-cancer therapy, activates ECs, platelets, and leukocytes, driving plaque inflammation and mortality from MI and stroke in mice, mirroring clinical cardiotoxicities in cancer patients. Asciminib did not induce these effects, suggesting it might be a safer option for imatinib-resistant cancer patients. Inhibition of TNFR-mediated endothelial activation is sufficient to prevent ponatinib-induced major adverse cardiovascular events.

Introduction

While molecularly-targeted cancer therapies have dramatically improved cancer survival, many have cardiovascular side effects that limit cancer treatment and cause significant morbidity and mortality.^1,2 Determining the mechanisms driving these cardiovascular side effects provides the opportunity to predict toxicity prior to exposing cancer patients to risk, and further affords the possibility to identify mitigating strategies to protect cancer survivors from adverse cardiovascular outcomes. Based on the discovery that chronic myeloid leukemia (CML) is caused by a constitutively active BCR-Abl fusion kinase, the Abl tyrosine kinase inhibitor (Abl-TKI), imatinib, was developed as the first targeted cancer therapy in 2001.³ Imatinib has an excellent safety profile without serious adverse cardiovascular effects. However, lifelong Abl-TKI treatment selects for malignant cells with Abl-kinase mutations, resulting in imatinib-resistance in over 50% of patients and necessitating continued development of new Abl-TKIs.⁴ Ponatinib, an Abl-TKI approved in 2012, is a highly effective therapy for imatinib-resistant CML, including those with the common T315I mutation. As a class, Abl-TKIs have improved CML 5-year survival from 40% with chemotherapy to well over 90% with continuous oral treatment.^5,6 Unfortunately, long-term follow-up revealed that ponatinib, dasatinib and nilotinib elicit a nearly 4-fold increased risk of acute arterial thrombosis compared to imatinib-treated patients.⁷ These arterial events manifest as acute myocardial infarction (MI), ischemic stroke, and critical limb ischemia, typically occurring months to years after starting Abl-TKI treatment. This increased arterial thrombosis risk mitigates the overall survival benefit of ponatinib over imatinib.⁷

Asciminib, the first allosteric Abl-TKI, received accelerated approval in 2021 for CML patients who failed two Abl-TKIs or those with T315I mutations. In September 2024, the first randomized trial of asciminib as first-line treatment for CML revealed improved cancer efficacy over imatinib, with similar efficacy compared to other second-generation Abl-TKIs.⁸ Consequently, asciminib use is expected to rise since its approval as a first-line therapy in October 2024.⁸ While short term trial data suggest fewer acute adverse hematologic side effects with asciminib, long-term follow-up remains insufficient to establish the cardiovascular safety profile of asciminib relative to ponatinib or imatinib.⁸ Thus, physicians currently lack sufficient guidance to confidently select the safest agent for individual patients.

The mechanism underlying ponatinib-induced arterial thrombotic events is incompletely understood. Prior studies examining the direct effects of Abl-TKIs on platelets have yielded inconclusive results, suggesting additional cellular contributors are involved in the pro-thrombotic phenotype.^9,10 One study demonstrated that ponatinib accelerated vascular thrombosis in mice in association with vascular wall damage and enhanced platelet reactivity.¹¹ Microvascular angiopathy has also been implicated.¹² Multiple recent studies indicate that ponatinib also directly induces endothelial cell (EC) damage in vitro, resulting in upregulated leukocyte adhesion molecule expression.^13-18 Additionally, elevated serum tumor necrosis factor alpha (TNF) levels observed in Abl-TKI-treated CML patients suggest enhanced systemic inflammation.¹⁹ TNF signals through two TNF receptors (TNFR1 and TNFR2), to contribute to EC inflammation.²⁰ EC damage promotes plaque inflammation via expression of leukocyte adhesion molecules, including selectins and cellular adhesion molecules (CAMs), which bind to ligands on leukocytes, including p-selectin glycoprotein ligand-1 (PSGL1) to facilitate leukocyte trafficking into the vascular wall.²¹ Systemic inflammation and EC damage are known to contribute to MI and stroke pathophysiology by promoting atherosclerotic plaque inflammation resulting in plaque destabilization and rupture—the most common mechanisms underlying myocardial infarction and stroke. However, this mechanism has never been implicated in Abl-TKI-induced arterial thrombosis.^22,23

Here we test the hypothesis that ponatinib induces myocardial infarction (MI) and stroke by promoting atherosclerotic plaque inflammation and instability, explore the mechanism by examining the effects of ponatinib on ECs, leukocytes, and platelets, and compare these effects to imatinib, which has no cardiovascular side effects, and to the new drug asciminib, for which cardiovascular safety is unknown. Abl-TKIs are compared to vehicle at clinically relevant concentrations in vitro and in vivo, using the established ApoE-KO mouse atherosclerosis model and the new SR-BI-mut/LDLR-KO mouse model of spontaneous plaque rupture manifesting in acute MI and stroke.²⁴ We demonstrate that ponatinib, but not imatinib or asciminib, promotes EC inflammation, platelet and leukocyte activation, leukocyte trafficking, plaque inflammation, and major adverse cardiac events (MACE), a composite of MI, stroke and death. Mechanistically, ponatinib enhanced signaling through TNF receptor type 2 (TNFR2) on ECs to induce cell adhesion molecule expression. TNFR inhibition prevented EC inflammation but not blood cell activation, and this was sufficient to prevent plaque inflammation, MI, stroke, and death in mice.

Methods

Data Availability:

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Detailed descriptions of experimental methods are provided in the Detailed Methods section in the Supplemental Material.

Mouse Models:

All mouse studies were approved by the Tufts University Institutional Animal Care and Use Committee and conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. C57BL/6J (The Jackson Laboratory, Strain 000664), APOE^−/− (ApoE-KO, The Jackson Laboratory, Strain 002052), and SR-BI^ΔCT/ΔCT/Ldlr^−/− (SR-BI-mut/LDLR-KO)²⁴ mice, were maintained on standard laboratory diet or high fat western diet (HFD, Envigo TD.88137, 42% calories from fat) with 12-hour light-dark cycle and ad libitum access to food and water. Pharmacokinetic studies were performed to determine a dosing strategy for each Abl-TKI that achieves a C_max and area under the curve (AUC) that closely models human drug exposure based on data from cancer trials for imatinib²⁵, ponatinib²⁶, and asciminib²⁷, as confirmed in mouse serum by mass spectrometry (Figure S1). Mice were randomized to vehicle control or Abl-TKI +/− TNFR inhibitor, all given by oral gavage. Except for terminal intravital microscopy (IVM) studies, performed under ketamine/xylazine anesthesia, all other mice were euthanized under 4% isoflurane anesthesia by terminal blood and tissue collection.

Statistical Analysis:

All data were collected and analyzed by investigators blinded to treatment group. Statistical analyses were performed using Prism 10 (GraphPad). Survival Analyses used log-rank test and outcomes at 16 and 20 weeks were compared using a chi-square test. For more than 2 groups, data were analyzed by two-way ANOVA with Tukey’s posttest if variances were equal. For data with unequal variances, Brown-Forsythe ANOVA testing was used with Sidak’s multiple comparison testing. In vitro data with more than two groups were analyzed using one-way ANOVA with Tukey’s posttest. All data are presented as mean ± SEM. Statistical significance was defined as P<0.05.

Results

The Clinically Prothrombotic Abl-TKI, Ponatinib, Significantly Increases Death Due to MI and Stroke Compared to Imatinib and Asciminib in Mice

SR-BI-mut/LDLR-KO mice were fed a high fat diet (HFD) for 4 weeks starting at 8 weeks of age and then randomized to vehicle, imatinib, ponatinib, or asciminib at doses that generate clinically relevant plasma drug levels (Figure S1). Mice were then followed for up to 20 weeks of HFD or until death or euthanasia for humane endpoints and the heart and brain were analyzed to identify the presence of MI or stroke (Figure 1A). Ponatinib significantly accelerated the development of major adverse cardiovascular events (MACE), defined as the composite of death, MI, and stroke, compared to vehicle-, imatinib- and asciminib-treated mice (Figure 1B). At the time of euthanasia, the presence of MI or stroke was determined by TTC staining of serial sections of the heart and brain, which marks live tissue as red and dead ischemic tissue appears grey (Figure 1C). In a subset of animals, cardiac tissue histology was performed, confirming evidence of acute MI, including dense leukocytic infiltration and fibrosis (Figure 1D). Quantification of total animals with MI, stroke, or both events at 16 and 20 weeks of high fat diet reveals that ponatinib significantly increased the incidence of MACE (p<0.0001). By 16 weeks of HFD, there were no surviving ponatinib-treated mice, compared to greater than 85% survival at 16 weeks with 50% still alive at 20 weeks of HFD in the vehicle-, imatinib- or asciminib-treated animals (Figure 1E). These are the first preclinical data demonstrating spontaneous MI and stroke induced by ponatinib treatment in a mouse atherosclerosis model. Survival in imatinib-treated mice was not different from vehicle, consistent with the known cardiac safety of imatinib in humans. The rate of MACE in asciminib-treated mice was not different from vehicle or imatinib in the SR-BI-mut/LDLR-KO mouse model.

Fig.1: Ponatinib Induces Spontaneous Myocardial Infarction (MI) and Stroke in the SR-BI-mut/LDLR-KO Plaque Rupture Mouse Model:

(A) Experimental paradigm: 8-week-old male SR-BI-mut/LDLR-KO mice were fed high fat diet (HFD) for 4 weeks to model high risk factor burden followed by randomization to oral administration of vehicle (Veh), imatinib (Ima), ponatinib (Pon), or asciminib (Asc) at doses that achieve clinically relevant serum levels. (B) Kaplan-Meier Survival Curve: Pon significantly accelerates the development of major adverse cardiovascular events (MACE) defined as the composite of death and euthanasia for spontaneous MI or stroke. (C) Representative TTC stained heart and brain sections indicating MI (black arrow) or stroke (blue arrow). Red indicates live tissue; gray indicates areas of necrosis. (D) Representative histologic section of heart from a Pon-treated mouse with MI, showing fibrosis by picrosirius red and inflammation by H&E stain. (E) Quantification of the percent of mice still alive or dead with evidence of MI, stroke, or both after 16 weeks and 20 weeks of HFD. N=12-16 mice/drug. *p<0.05 via Log-rank test (B) and chi-square test (E). (A) made with Biorender.com.

Ponatinib, but Not Imatinib or Asciminib, Promotes Leukocyte Trafficking in vivo and Endothelial Cell Activation

Inflamed leukocyte-rich atherosclerotic plaques are prone to rupture in humans.²⁸ Thus, to begin to explore the mechanism for ponatinib-induced ischemic events, we compared the impact of the three Abl-TKIs on leukocyte trafficking by intravital microscopy (IVM). C57BL/6J mice were treated for 3 days with each Abl-TKI at clinically relevant doses, and IVM was performed to image leukocyte rolling and adhesion to the endothelium in mesenteric vessels (Figure 2A, Movies 1 – 4 in the online Data Supplement). Ponatinib significantly increased leukocyte rolling along the mesenteric vasculature by 7-fold (Figure 2B) and firm adhesion by 4-fold (Figure 2C) compared to vehicle. For comparison, the potent inflammatory stimulus prompted by injecting 0.3 micrograms of TNF directly into the peritoneum induced leukocyte rolling by 13-fold and adhesion by 8-fold (Figure S2), approximately double the impact of clinically relevant serum ponatinib exposure. Imatinib and asciminib did not impact leukocyte trafficking compared to vehicle. A complete blood count showed no significant effect of any of the Abl-TKIs on the number of circulating white blood cells, platelets, or the hemoglobin level (Figure S3). Leukocytes traffic to sites of vascular injury by first rolling on the surface of activated ECs via interaction with selectins, followed by firm adhesion and trans-endothelial migration mediated by interactions with ICAM1 and VCAM1.²¹ Thus, these EC adhesion molecules were quantified by immunoblotting of mesenteric vessel protein lysates from mice identically treated with Abl-TKIs for 3 days. P-selectin, ICAM1, and VCAM1 protein were significantly increased in mesenteric vessels from ponatinib-treated mice compared to vehicle (Figure 2D-E). Imatinib and asciminib did not impact expression of any of the EC adhesion molecules. E-selectin expression was not significantly changed by exposure to any of the Abl-TKIs.

Fig.2: Ponatinib Induces Leukocyte Trafficking and EC Adhesion Molecules:

6-week-old male C57BL/6J mice were treated for 3 days by oral gavage with vehicle (Veh), imatinib (Ima), ponatinib (Pon), or asciminib (Asc), and intravital microscopy (IVM) was performed to image leukocyte trafficking. (A) Representative IVM static images of mesenteric vessels. Quantification of (B) leukocyte rolling and (C) leukocyte adhesion. N=10-12 mice per group. (D) Representative immunoblots of mesenteric vessel lysates isolated from identically treated mice and (E) quantification of P-selectin, E-selectin, ICAM1, and VCAM1. N=3-4 sets of pooled (2 mice each) mesenteric vessels. (F) Representative immunoblots of lysate from human coronary artery endothelial cells (HCAECs) treated with Veh, Ima, Pon, or Asc for 24 hours and (G) quantification of E- and P-selectin, ICAM1, and VCAM1 protein. N=4 experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via one way ANOVA with Tukey’s posttest.

Since ponatinib treatment is associated with a high risk of acute MI due to coronary artery thrombosis⁷, the impact of the 3 Abl-TKIs on adhesion molecule expression was also examined in primary human coronary artery endothelial cells (HCAECs) treated for 24 hours with each Abl-TKI at the clinical C_max concentration. Ponatinib significantly increased protein (Figure 2F-G) and mRNA (Figure S4A) expression of P-selectin, ICAM1, and VCAM1 compared to vehicle, imatinib, and asciminib in HCAECs. These same findings were observed in Abl-TKI-treated HUVECs (Figure S4B-C). Ponatinib did not impact E-selectin expression in either type of primary human EC. Imatinib and asciminib did not significantly induce expression of any of the adhesion molecules in HCAECs or HUVECs. Only in HCAECs, imatinib significantly decreased VCAM1 mRNA expression relative to vehicle control (Figure S4A). These data reveal that ponatinib, a drug with a high rate of arterial thrombosis in cancer patients, directly induces expression of inflammatory adhesion molecules in human coronary ECs in vitro and in mouse mesenteric vessels in vivo and induces leukocyte trafficking. Conversely, imatinib, which does not associate with vascular toxicity in patients, and asciminib, for which potential vascular side effects are not known, do not induce EC inflammation.

Ponatinib Induces Endothelial Cell TNF Receptor Expression Which Contributes to Induction of Adhesion Molecules

To begin to explore the mechanism by which EC adhesion molecules are induced by ponatinib, we examined the inflammatory signaling cascade linking TNF signaling to activation of the pro-inflammatory transcription factor NFkB, which regulates EC adhesion molecule transcription.²⁰ HUVECs were treated for 10 minutes with Abl-TKIs and the phosphorylation state of the adaptor protein TNF Receptor-Associated Factor 2 (TRAF2), its downstream effector endothelial tyrosine kinase (ETK)²⁰, and the p65 component of NFkB were quantified relative to each total protein. Ponatinib, but not imatinib or asciminib, significantly increased the relative phosphorylation of TRAF2 (Figure 3A), ETK (Figure 3B), and NFkBp65 (Figure 3C), compared to vehicle. This inflammatory signaling cascade is typically induced by inflammatory stimuli, such as TNF, signaling via TNF receptors. TNF itself was detectable in EC lysate, but not significantly altered by Abl-TKI treatment, even after 24 hours (Figure S5A). Thus, the impact of Abl-TKIs on TNF receptor expression was measured in vitro in human ECs and in vivo in mouse mesenteric vessels. Ponatinib treatment for 24 hours in vitro significantly increased protein expression of both TNFR1 and TNFR2 compared to vehicle in HCAECs (Figure 3D-F), HUVECs (Figure S5B), and in mesenteric vessels from mice treated for 3 days with each drug in vivo (Figure 3G-I). Imatinib and asciminib did not alter expression of either TNF receptor compared to vehicle treatment in HCAECs, HUVECs, or mouse mesenteric vessels. To determine if TNFR signaling is necessary for the subsequent induction of EC adhesion molecules, HUVECs were pre-treated with a non-selective (inhibits both TNFR1 and TNFR2) TNFR inhibitor (TNFRi) at a concentration that we first confirmed in HUVECs to effectively block induction of adhesion molecule mRNA by TNF (Figure S6). Pre-treatment of HUVECs with TNFRi prevented the ponatinib-induced increase in P-selectin, ICAM1 (Figure 3J), and VCAM1 (Figure S7) mRNA and surface expression (Figure 3K) by flow cytometry, confirming that TNFR signaling is necessary for ponatinib induction of adhesion in human ECs. Taken together, the data in Figure 3 support that ponatinib increases EC TNFR protein expression, activates NFkB signaling, and induces cell adhesion molecules in a TNFR-dependent manner, while asciminib and imatinib do not.

Fig. 3: Ponatinib Increases TNF Receptor Expression and Signaling in Human ECs which is Necessary for Adhesion Molecule Induction:

(A-C) Human umbilical vein endothelial cells (HUVECs) were treated with vehicle (Veh), imatinib (Ima), ponatinib (Pon), or asciminib (Asc) for 10 minutes. Representative immunoblots and quantification of; (A) phospho-TRAF2, (B) phospho-ETK, and (C) phospho-NFkB-p65 each normalized to their respective total protein level. N=4-6 experiments. (D-F) Human coronary artery endothelial cells (HCAECs) were treated with Abl-TKIs for 24 hours. Representative immunoblots and quantification of; (E) TNFR1 and (F) TNFR2. N=3-4 experiments. (G-I) Mice were treated for 3 days with Abl-TKIs. Representative immunoblots of mesenteric vessel lysates and quantification of; (H) TNFR1 and (I) TNFR2. N=4 sets of 2 pooled mouse mesenteric vessels. (J-K) HUVECs were pretreated for 15 minutes with Veh or TNFR inhibitor (TNFRi R-7050) and then treated with Veh or Pon for 24 hours. Quantification of (J) mRNA by PCR (N=6-9 experiments) and (K) surface P-selectin and ICAM1 protein by flow cytometry (N=4 experiments). *p<0.05, **p<0.01, ***p<0.001, via one way ANOVA (A-I), two-way ANOVA (J-K), with Tukey’s posttest.

Ponatinib Rapidly Induces Trafficking of TNFR2 to the Endothelial Cell Membrane Which is Necessary for EC Inflammation In Vitro

To further interrogate the mechanism of ponatinib-induced EC inflammation, a time-course evaluation was performed to examine the impact of ponatinib treatment from 15 minutes to 24 hours on HUVEC surface protein and mRNA expression of TNFR1, TNFR2, P-selectin, ICAM1, and VCAM1. Ponatinib induced a rapid early increase in TNFR2 expression on the EC surface as determined by flow cytometry after only 15 to 30 minutes with a later induction of expression at 4 and 24 hours (Figure 4A), with no impact on TNFR1 expression on the cell surface (Figure S8A). EC surface expression of P-selectin (Figure 4B) and ICAM1 (Figure 4C) were significantly increased later, after 8 and 24 hours, while surface VCAM1 was significantly induced only after 24 hours of ponatinib treatment (Figure S8B). At the mRNA level, ponatinib significantly increased mRNA for TNFR2, but not TNFR1, starting at 8 hours, followed closely by P-selectin, with the significant induction of ICAM1 becoming evident only after 16 hours, and VCAM1 after 24 hours of ponatinib treatment (Figure S9A-D). Consistent with Figures 2 and 3, neither imatinib nor asciminib significantly impacted mRNA expression of any TNFR or adhesion molecule at any time point and none of the Abl-TKIs impacted E-selectin mRNA expression (Figure S9E-F).

Fig. 4: Ponatinib Rapidly Induces TNF Type 2 Receptors (TNFR2) on the EC Surface Followed by Increased Adhesion Molecule Expression:

(A-C) Time course of HUVECs treated with vehicle (Veh) or ponatinib (Pon) for 15 minutes to 24 hours with flow cytometry to quantify the fold change in surface mean fluorescence intensity (MFI) of; (A) TNFR2, (B) P-selectin, and (C) ICAM1 compared to vehicle. N=8 experiments. (D) Experimental schematic showing the site of action of the TNFR inhibitor (TNFRi) and the vesicle-mediated trafficking inhibitor brefeldin (BRE) used to test the role of TNF signaling versus TNFR membrane trafficking upregulation of TNFR on the EC surface. (E-F) HUVECs were pretreated for 15 minutes with (E) TNFRi or (F) Brefeldin followed by Veh or Pon for 30 minutes and the fold change in the MFI of TNFR2 surface expression compared to vehicle was quantified. (G) HUVECs were pretreated for 15 minutes with TNFRi followed by Veh or Pon for 24 hours and the fold change in the MFI of TNFR2 surface expression compared to vehicle was quantified. N=4 experiments. (H-K) HUVECs were treated with siRNAs specific for (H-I) TNFR1 or (J-K) TNFR2 compared to scrambled (scr) siRNA control for 24 hours followed by Veh or Pon for an additional 24 hours. (H, J) representative immunoblots and (I, K) quantification of P-selectin, E-selectin, ICAM1 and VCAM1. N=5-6 experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via two-way ANOVA with Sidak’s posttest (A-C), two-way ANOVA with Tukey’s posttest (E-K). (D) made with Biorender.com.

Since ponatinib treatment rapidly increased EC surface levels of TNFR2 within 15 minutes, but induction of mRNA for TNFR2 and adhesion molecules did not occur for hours, we next tested the hypothesis that ponatinib rapidly induces TNFR2 translocation to the cell membrane, independently of TNFR signaling. HUVECs were pre-treated for 15 minutes with either a TNFRi, to inhibit TNFR signaling, or brefeldin, to inhibit trafficking of protein-containing vesicles to the membrane. Inhibitor-treated HUVECs were then exposed to vehicle or ponatinib for 30 minutes (Figure 4D). Pretreatment with a TNFRi had no impact on the rapid induction of TNFR2 on the surface of HUVECs after 30 minutes of exposure to ponatinib (Figure 4E). Rather, blockade of vesicle trafficking with brefeldin prevented the rapid ponatinib-induced increase in surface TNFR2 (Figure 4F). Only after 24 hours did TNFRi blockade prevent the ponatinib-induced longer-term rise in cell surface TNFR2 expression (Figure 4G). Similarly, the ponatinib induced increase in TNFR2 mRNA observed at 24 hours was prevented by TNFRi pretreatment (Figure S10).

To further clarify which TNFR mediates the impact of ponatinib on EC inflammation, each TNF receptor was individually knocked down (KD) in HUVECs using receptor-specific siRNA, confirming that TNFR1 siRNA specifically decreased EC TNFR1 protein without impacting TNFR2 expression, and that TNFR2-targeted siRNA significantly decreased TNFR2 with no change in TNFR1 (Figure S11). KD of TNFR1 did not prevent ponatinib induction of the adhesion molecules P-selectin, ICAM1, and VCAM1 (Figure 4H-I). Rather, KD of TNFR2 completely prevented the ponatinib-induced increase of P-selectin, ICAM1, and VCAM1 protein in HUVECs compared to scrambled control siRNA (Figure 4J-K). As such, ponatinib-treated TNFR2 KD ECs had significantly less P-selectin, ICAM1, and VCAM1 compared to ponatinib-treated ECs pretreated with scrambled siRNA control. The basal level of only ICAM1 expression was significantly decreased by TNFR2 KD compared to scrambled siRNA in the absence of ponatinib. Thus, TNFR2 specifically is necessary for induction of EC adhesion molecule expression by ponatinib in vitro. Together, the data in Figures 3-4 support a model in which ponatinib induces rapid TNFR2 trafficking to the EC membrane, resulting in a rapid increase in EC surface TNFR2, increased TNFR signaling to NFkB, followed by increased mRNA expression of TNFR2 and of leukocyte adhesion molecules, and this is necessary for the ponatinib-induced increase in surface protein expression of P-selectin, ICAM1, and VCAM1 on human ECs.

Ponatinib Induces Inflamed Atherosclerotic Plaques and Aortic Adhesion Molecules in ApoE-KO Mice in vivo

We next examined whether ponatinib impacts plaque inflammation and aortic adhesion molecule expression in an atherosclerotic mouse model in vivo. Male and female ApoE-KO mice were fed high fat diet (HFD) for 4 weeks and then randomized to vehicle, imatinib, ponatinib or asciminib for an additional 4 weeks along with HFD (Figure 5A). This experimental paradigm was utilized to model the high cardiovascular disease burden in older CML patients at the time of Abl-TKI initiation. Measurement of traditional cardiovascular risk factors in males revealed no significant drug impact on weight gain or systolic blood pressure. Imatinib modestly reduced fasting glucose compared to ponatinib and asciminib. Ponatinib decreased fasting cholesterol compared to imatinib and asciminib (Table S1). In females, there were no changes in blood pressure, fasting glucose, or cholesterol, but imatinib- and ponatinib-treated mice gained less weight while on drug compared to asciminib-treated mice (Table S2).

Fig. 5: Ponatinib Induces Rupture Prone Plaque Phenotypes and Plaque Inflammation in ApoE-KO Mice:

(A) Experimental paradigm: ApoE-KO mice were fed HFD for 4 weeks followed by randomization to oral administration of vehicle (Veh), imatinib (Ima), ponatinib (Pon), or asciminib (Asc) for 4 additional weeks at doses that achieve clinically relevant serum levels. (B) Representative Oil Red O-stained aortic root cryosections and (C) Quantification of plaque cross-sectional area and the percentage of plaque area composed of neutral lipids and necrotic core. (D) Quantification by flow cytometry of aortic arch total CD45+ leukocytes, CD3+/CD19− T-cells, Ly6G−/CD19−/F4/80+ macrophages, and Ly-6G−/CD19−/F4/80− monocytes. N=13-19 mice per group. (E) Representative descending aorta lysate immunoblots and quantification of; (F) P-selectin, (G) E-selectin, (H) ICAM1, (I) VCAM1, (J) TNFR1, and (K) TNFR2. N=6 pairs of pooled aortas per group. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via one-way ANOVA with Tukey’s posttest. (A) made with Biorender.com.

Histologic analysis of aortic root plaque phenotype from male mice demonstrates that ponatinib treatment results in smaller plaques with significantly increased lipid content and necrotic core (characteristics of an unstable phenotype) when compared to mice treated with vehicle or imatinib, (Figure 5B-C). Furthermore, ponatinib significantly increased leukocyte content of aortic arch plaques as determined by flow cytometry (Figure 5D). Ponatinib significantly increased the number of total CD45+ leukocytes as well as subsets of T-cells, macrophages, and monocytes in the plaque compared to mice treated with vehicle, imatinib, or asciminib (Figure 5D). Imatinib or asciminib did not significantly impact atherosclerotic plaque size, phenotype (Figure 5B-C), or inflammation (Figure 5D) compared to vehicle-treated ApoE-KO mice.

The impact of Abl-TKIs on adhesion molecule expression was further quantified by immunoblotting of descending aorta lysates from these same mice. Ponatinib increased aortic expression of P-selectin, E-selectin, ICAM1, VCAM1, TNFR1, and TNFR2 compared to vehicle-, imatinib-, and asciminib-treated mice (Figure 5E-K). In female ApoE-KO mice, ponatinib also significantly increased the plaque necrotic core content, leukocyte infiltration, and aortic expression of P-selectin, ICAM1, VCAM1, TNFR1, and TNFR2 (Figure S12). Only the ponatinib-induced increase in E-selectin and plaque lipid content observed in male mice was not reproduced in female ApoE-KO mice. Together, these data demonstrate in vivo that ponatinib, but not imatinib or asciminib, induces vascular expression of TNF receptors and cell adhesion molecules and produces atherosclerotic plaques with greater necrotic core area and infiltrating inflammatory cells, consistent with a phenotype that is predisposed to rupture.

Ponatinib Induces Leukocyte and Platelet Activation and Aggregation in vivo

Abl-TKIs are systemically administered and hence may impact other cells, in addition to ECs, that could contribute to plaque inflammation and arterial thrombosis. Thus, we next compared the impact of imatinib, ponatinib, and asciminib on leukocyte and platelet markers that drive thromboinflammation. Wild type C57BL/6J mice were treated for 3 days with Abl-TKIs and flow cytometry performed on circulating leukocytes and platelets. Ponatinib significantly increased leukocyte and platelet surface activation markers (Figure 6). Specifically, on CD45+ leukocytes, ponatinib significantly increased surface expression of PSGL1, the ligand for P-selectin, compared to vehicle and imatinib (Figure 6A-B). Ponatinib also induced platelet activation as measured by surface CD41 (also known as glycoprotein IIb/IIIa, Figure S13A-B) and significantly increased P-selectin on single platelets compared to vehicle, imatinib, and ponatinib (Figure 6C-D). Imatinib and asciminib did not impact these markers of platelet or leukocyte activation. PCR of leukocyte mRNA reveals no impact of ponatinib on PSGL1 mRNA (Figure 6E), suggesting a non-transcriptional mechanism of the increase in surface expression. Ponatinib also significantly increases leukocyte TNF mRNA (Figure 6F) and serum TNF levels by ELISA (Figure 6G), consistent with increased leukocyte inflammatory activation. Finally, since ponatinib increased surface P-selectin on platelets and the respective receptor PSGL1 on leukocytes, we quantified platelet-leukocyte aggregates (PLAs). After first gating on all CD45+ cells, thereby including all flow cytometry events containing leukocytes (Figure 6H), we quantified the percent of CD45+ events that also express the platelet marker CD41. Ponatinib more than doubled the percent of these platelet-leukocyte aggregates in the blood from mice compared to vehicle-, imatinib-, and asciminib-treated animals (Figure 6I). Overall, these data reveal that in addition to activation of EC adhesion molecules in vivo in mice, treatment with ponatinib, but not imatinib or asciminib, activates platelets and leukocytes to increase platelet surface P-selectin, the P-selectin ligand PSGL1 on circulating leukocytes, leukocyte-derived TNF, and formation of platelet-leukocyte aggregates.

Fig. 6: Ponatinib Activates Blood Platelets, Leukocytes and Induces Platelet-Leukocyte Aggregate Formation In Vivo:

C57BL/6J mice were treated for 3 days with vehicle (Veh), imatinib (Ima), ponatinib (Pon) or asciminib (Asc) and blood flow cytometry and PCR performed to quantify; (A-B) PSGL1 expression on CD45+ single leukocytes and (C-D) P-selectin expression on CD41+ single platelets, each expressed as the fold change in mean fluorescence intensity (MFI) compared to vehicle. (E) PSGL1 mRNA and (F) TNF mRNA by PCR. N=4-6 mice per group. (G) Serum TNF quantified by ELISA. N=8-15 mice per group. (H-I) CD45+ events were gated for CD41 expression to quantify platelet-leukocyte aggregates. N=4-6 mice per group. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via one-way ANOVA with Tukey’s posttest.

Pharmacologic TNFR Inhibition Prevents Ponatinib-Induced EC Activation and Leukocyte Trafficking without Blocking Leukocyte-Platelet Activation in vivo

We next examined whether TNFR signaling contributes to the impact of ponatinib on EC, leukocyte, and platelet activation to induce leukocyte trafficking and platelet-leukocyte aggregation in vivo. C57BL/6J mice were treated with vehicle or ponatinib, each alone or with TNFRi for 3 days (Figure 7) at a TNFRi dose sufficient to significantly inhibit TNF-induced leukocyte trafficking (see Figure S2). Mesenteric vessel protein expression of EC adhesion molecules was quantified by immunoblotting, revealing that co-treatment with TNFRi prevents ponatinib induction of P-selectin, ICAM1, and VCAM1 in vivo. As such there is significantly less adhesion molecule expression in vessels from mice co-treated with ponatinib and TNFRi compared to ponatinib alone (Figure 7A-B, comparing red bars). Using the identical treatment paradigm, IVM was also performed (Figure 7C, Movies 5 – 8 in the online Data Supplement). TNFRi co-treatment completely prevented ponatinib-induced leukocyte rolling and adhesion, resulting in a significant decrease in leukocyte trafficking when TNFRi was co-administered with ponatinib compared to ponatinib alone (Figure 7D-E). ELISA of serum TNF revealed this reduction in leukocyte trafficking occurred despite no reduction in circulating TNF (Figure 7F).

Fig. 7: TNFR Inhibition Prevents Ponatinib-Induced EC Activation and Leukocyte Trafficking In Vivo:

C57BL/6J mice were treated with vehicle (Veh) or ponatinib (Pon) each with Veh or TNF receptor inhibitor (TNFRi) for 3 days. (A-B) Representative immunoblots of mesenteric vessel lysate with quantification of P-selectin, E-selectin, ICAM1, and VCAM1. N=3 sets of 2 pooled mesenteric vessel lysates. (C-E) Representative intravital microscopy (IVM) static images of mesenteric vessels. Quantification of (D) leukocyte rolling and (E) leukocyte adhesion. (F) Serum TNF measured by ELISA. N=5-6 mice per group. (G-K) Mice were treated with Veh or Pon, each with Veh or TNFRi, for 3 days and blood collected for flow cytometry. All CD45+ single leukocytes were gated and (G-H) PSGL1 expression was quantified and expressed as fold change in MFI compared to vehicle. All CD41+ single platelets were gated and (I-J) P-selectin expression was quantified and expressed as the fold change in MFI compared to vehicle. (K) CD45+ leukocyte-containing events were gated for CD41 expression to quantify platelet-leukocyte aggregates. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via two-way ANOVA with Tukey’s posttest.

In contrast to the inhibitory impact of TNFRi on ponatinib-induced EC activation, TNFRi co-treatment did not significantly attenuate ponatinib-induced surface leukocyte PSGL1 expression (Figure 7G-H). Similarly, co-treatment of TNFRi along with ponatinib did not significantly decrease the level of platelet surface P-selectin compared to ponatinib alone (Figure 7I-J, comparing red bars). TNFRi alone did significantly increase platelet activation, as demonstrated by an increase in surface expression of CD41 (Figure S13C-D) and P-selectin on platelets from TNFRi-treated mice compared to vehicle alone (Figure 7I-J, comparing grey bars). We confirmed that circulating leukocytes express TNFR2, which is modestly but significantly increased by ponatinib compared to imatinib (Figure S14A-B) and that TNFRi treatment alone also causes a modest increase in TNFR2 expression on leukocytes and platelets (Figure S14C). Despite the presence of TNFR2 on circulating blood cells, TNFRi did not attenuate the ponatinib-induced increase in platelet-leukocyte aggregation (Figure 7K), consistent with the finding that TNFRi does not prevent ponatinib activation of leukocyte PSGL1 or platelet P-selectin. Together, these data in Figure 7 demonstrate that TNFR signaling is necessary for ponatinib induction of EC adhesion molecules and leukocyte trafficking but does not contribute to ponatinib activation of leukocytes or leukocyte-platelet aggregate formation.

TNFRi Prevents Ponatinib-Induced Atherosclerotic Plaque Inflammation, and MI, Stroke, and Death in Mice

Since pharmacologic TNFR inhibition protects from ponatinib-induced leukocyte trafficking and EC adhesion molecule expression, we next examined the impact of TNFRi on ponatinib-induced atherosclerotic plaque inflammation. Male and female ApoE-KO mice were treated with vehicle or ponatinib in the presence or absence of TNFRi (Figure 8A). As before, measurement of traditional cardiovascular risk factors in males revealed no drug-induced changes in weight gain or systolic blood pressure, though ponatinib alone did reduce serum cholesterol compared to both TNFRi treated groups (Table S3). Female animals had no significant changes in any traditional risk factors across treatment groups (Table S4). TNFRi co-treatment prevented the ponatinib-induced increase in EC adhesion molecules P-selectin, E-selectin, ICAM1, and VCAM1 in aortic lysate from ApoE-KO mice (Figure 8B). Flow cytometry of the aortic arches from the same animals revealed that TNFRi prevented the ponatinib-induced increase in aortic plaque total leukocytes, T-cells, macrophages, and monocytes (Figure 8C). This resulted in significantly fewer leukocytes in aortic arches from mice co-treated with TNFRi and ponatinib compared to ponatinib alone. Consistent with prior studies²⁹, ApoE-KO mice on HFD have higher serum TNF levels (>5 pg/ml, Figure 8D) compared to wild type mice (<0.5 pg/ml, Figure 6G) but Abl-TKIs or TNFRi treatment did not significantly alter serum TNF levels in ApoE-KO mice (Figure 8D and Figure S15).

Fig. 8: TNFR Inhibition Prevents Ponatinib-Induced Atherosclerotic Plaque Inflammation and Rescues Ponatinib-induced MACE in Mice:

(A-D) Experimental paradigm for atherosclerosis study: Male and female ApoE-KO mice were fed high fat diet (HFD) for 4 weeks and then randomized to vehicle (Veh) or ponatinib (Pon) alone or with TNF receptor inhibitor (TNFRi) for 4 additional weeks. (B) Representative immunoblots and quantification of adhesion molecules P-selectin, E-selectin, ICAM1, and VCAM1 in descending aortas. (C) Quantification of aortic arch total CD45+ leukocytes, CD3+/CD19− T cells, Ly6G−/CD19−/F4/80+ macrophages, and Ly-6G−/CD19−/F4/80− monocytes by flow cytometry. Data are represented as fold change compared to vehicle for each sex and males and females are combined. (D) Blood TNF measured by ELISA. N=5-8 mice per group. (E-F) Experimental paradigm for plaque rupture: 8-week-old male SR-BI-mut/LDLR-KO mice were fed HFD for 4 weeks and then randomized to Veh or Pon alone or with TNFRi and followed until death or euthanasia for MI or stroke. (F) Kaplan-Meier Survival Curve: TNFRi co-treatment prevented ponatinib induction of major adverse cardiovascular events (MACE), the composite of death, MI and stroke. N=5-8 mice per group. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, via two-way ANOVA with Tukey’s posttest (B-D) and Log-rank test (F). (A) and (E) made with Biorender.com.

Finally, we tested whether TNFRi inhibition could rescue the profound increase in MACE (the composite of MI, stroke, and death) induced by ponatinib in SR-BI-mut/LDLR-KO mice. As in Figure 1, after 4 weeks of HFD, SR-BI-mut/LDLR-KO mice were randomized to vehicle or ponatinib, this time, in the presence or absence of TNFRi (Figure 8E). Ponatinib again profoundly increased the rate of MACE, with 100% mortality from the composite of MI, stroke, and death by 14 weeks. TNFRi co-treatment significantly improved survival compared to ponatinib alone, completely rescuing the phenotype, resulting in survival in the ponatinib plus TNFRi group that was not different from vehicle or TNFRi alone (Figure 8F). Together, the data in Figures 7-8 reveal that TNFR inhibition in vivo prevents ponatinib-induced EC activation, but does not prevent leukocyte or platelet activation, and that EC protection is sufficient to prevent leukocyte trafficking, atherosclerotic plaque inflammation, MI, stroke, and death in vivo.

Discussion

This study interrogated mechanisms by which the potent anti-cancer Abl-TKI ponatinib causes arterial thrombosis, a treatment-limiting cardiovascular side effect in cancer patients. The studies were performed using drug doses and concentrations consistent with clinical drug levels in cancer patients and all findings directly compared ponatinib with the first-generation drug imatinib, which lacks clinical cardiovascular side effects, and the newly approved first line agent, asciminib, with limited long term safety data. The in vitro results reveal that in human ECs: (1) ponatinib rapidly induces surface expression of the type 2 TNF receptor by a mechanism that is dependent on vesicular trafficking; (2) ponatinib rapidly activates TNF signaling pathways, as indicated by phosphorylation of TRAF2, ETK, and NFkB; and (3) these signaling events are followed by a later increase in mRNA, protein, and cell surface expression of cell adhesion molecules P-selectin, ICAM1, and VCAM1 that is specifically dependent on TNFR2 in ECs. The in vivo results reveal that: (1) ponatinib induces expression of TNF receptors and EC adhesion molecules P-selectin, ICAM1, and VCAM1 in mesenteric vessels from wild type mice and in aortas from ApoE-KO atherogenic mice; (2) ponatinib induces platelet activation as demonstrated by increased surface expression of CD41 (GPIIb/IIIa) and P-selectin; (3) ponatinib activates leukocytes to produce the inflammatory cytokine TNF and to increase surface expression of the P-selectin ligand, PSGL1; (4) this induction of adhesion molecules on ECs and platelets, and of adhesion ligands on leukocytes, associates with increased leukocyte trafficking visualized by intravital microscopy and with formation of platelet-leukocyte aggregates; (5) in ApoE-KO atherosclerosis mice, ponatinib increases plaque inflammation and produces an unstable plaque phenotype; (6) ponatinib dramatically increases the rate of MI, stroke, and death in the SR-BI-mut/LDLR-KO plaque rupture mouse model; (7) a small molecule inhibitor of TNF receptors blocks ponatinib-induced EC vascular adhesion molecule expression, leukocyte trafficking, plaque inflammation, MI, stroke, and death, but does not prevent platelet or leukocyte activation and aggregation; (8) imatinib, an Abl-TKI without cardiovascular side effects, has none of these in vitro or in vivo effects, and finally; (9) the new Abl-TKI asciminib, for which clinical cardiovascular safety remains unknown, does not activate ECs, platelets, or leukocytes, nor does it induce leukocyte trafficking, plaque inflammation, MI, or stroke in vivo.

Arterial thrombosis can arise due to inappropriate clotting triggered by endothelial injury, platelet activation or hypercoagulable states. As hematologists were the first to appreciate the high risk of arterial thrombosis in ponatinib-treated leukemia patients, initial investigations focused on the effects of Abl-TKIs on platelet function, with initially divergent findings that appeared to depend on the study design.¹⁰ This was clarified by one study of nilotinib, another pro-thrombotic Abl-TKI, showing that platelets from mice treated with nilotinib demonstrated enhanced adhesion to collagen, whereas direct exposure of platelets to nilotinib ex vivo had no impact on platelet adhesiveness.⁹ Those data support a role for non-platelet cells in Abl-TKI-mediated platelet activation in vivo. This is consistent with another study that treated mice with ponatinib in vivo, demonstrating platelet activation along with arterial injury¹¹. Subsequently, multiple studies have demonstrated direct toxic effects of the prothrombotic Abl-TKIs on ECs, including induction of adhesion molecules³⁰. Increasingly, inflammation is recognized as a contributor to thrombosis, with systemic inflammation activating platelets and leukocytes and damaging endothelial cells, together increasing thrombotic risk. Indeed, conditions including sepsis, COVID-19, and atherosclerosis are strongly associated with heightened thromboinflammation characterized by endothelial activation, leukocyte recruitment to the vasculature, and heightened thrombosis risk.²² The current study reveals that ponatinib induces multiple hallmarks of thromboinflammation in vivo, including activation of leukocytes to produce and release inflammatory cytokines, particularly TNF, along with increased surface expression of PSGL1, the leukocyte ligand that mediates interactions with platelets and ECs via the receptor P-selectin. Simultaneously, ponatinib induces P-selectin expression on ECs and platelets along with increased platelet surface expression of CD41 (GPIIb/IIIa), crucial for platelet aggregation. Together, the impact of ponatinib on these 3 cell types combines to promote platelet-leukocyte aggregate formation and leukocyte-endothelial interactions that drive leukocyte trafficking and vascular inflammation.

The devastating clinical manifestations of ponatinib-induced cardiovascular toxicity are acute MI, stroke, and critical limb ischemia. In recent cancer trials, 31% of those randomized to ponatinib experienced arterial thrombosis, and in a study of ponatinib added to chemotherapy, MI-related deaths necessitated dose reduction.³¹ CML typically presents in the sixth decade of life, hence, over 60% of CML patients have a history of hypertension, diabetes, or known coronary heart disease at initiation of Abl-TKI therapy.³² The high risk of arterial thrombosis associated with ponatinib is further exacerbated 2.2 fold in those with at least one CV risk factor verses no risk factors⁷, suggesting synergy between pre-existing CV risk and Abl-TKI treatment. When non-cancer patients with underlying risk factors experience acute MI or stroke, the predominant cause is rupture or erosion of an unstable plaque. Unstable plaques are characterized by increased inflammatory cells, lipid content and necrotic core.²⁸ Here, we demonstrate in the ApoE-KO proatherogenic mouse, that ponatinib induces an unstable plaque phenotype, with more inflammatory cells and a greater proportion of necrotic core. One prior study examined the impact of ponatinib in the APOE*3Leiden/CETP mouse model and found that ponatinib decreased plasma cholesterol as well as atherosclerotic lesion area, but with no assessment of plaque inflammation.³³ In the ApoE-KO model, we also found that specifically in males, ponatinib decreased cholesterol relative to imatinib- and asciminib-treated mice and produced smaller plaques, consistent with that prior study. However, we show in both males and females, that ponatinib substantially increased plaque inflammation and necrotic core area, suggesting plaque instability. Since a major limitation of traditional mouse atherosclerosis models is that the plaques do not spontaneously rupture to cause ischemic events, the field has been limited to using only surrogate markers of plaque vulnerability until very recently. Using a new model of spontaneous plaque rupture on the LDL receptor-KO background²⁴, we demonstrate for the first time that ponatinib profoundly increases the rate of MI, stroke, and death, compared to vehicle-, imatinib-, and asciminib-treated mice. This finding provides the first preclinical evidence supporting that increased rupture of inflamed plaques may underly the arterial thrombosis risk induced by ponatinib and potentially by other pro-thrombotic Abl-TKIs.

Endothelial dysfunction, characterized by upregulation of adhesion molecules, is known to promote leukocyte adhesion and infiltration into plaques to contribute to vascular inflammation.²⁸ In vivo, we show that ponatinib alone is sufficient to induce vascular adhesion molecules and leukocyte trafficking. This mechanism parallels prior studies where increased expression of adhesion molecules has been linked to plaque progression and rupture in coronary disease.³⁴ The magnitude of the ponatinib-induced increase in adhesion molecules observed in human endothelial cells in vitro appears to be about half the response to the potent inflammatory stimulus TNF and this correlates with a similar relative impact on leukocyte trafficking in vivo, supporting that the magnitude of the effects on EC inflammation are physiologically meaningful. Whether the pro-inflammatory effects of ponatinib are independent of, or synergistic with TNF, cannot be determined from these studies and await further testing. However, ponatinib-induced endothelial activation translates to functional consequences, with ponatinib-induced adhesion molecule upregulation in the ApoE-KO model being sufficient to induce plaque inflammation, and in the SR-BI-mut/LDLR-KO plaque rupture model, ponatinib treatment culminates in a very significant increase in spontaneous myocardial infarction and stroke events, further supporting the relevance of these findings.

Mechanistically, our in vitro data reveals that ponatinib induces a rapid increase in TNFR surface expression, mediated by vesicle trafficking, and activation of TNF signaling cascades leading to phosphorylation of the transcription factor NFkB, which is known to transcriptionally regulate EC inflammatory genes.²⁰ This acute effect of ponatinib is temporally followed by increased mRNA, total protein, and cell surface expression of TNFRs, selectins, and CAMs. We further examined the impact of ponatinib in the different models on the level of TNF, the traditional ligand for TNFR activation. In cultured human ECs, TNF is present but unaffected by Abl-TKIs, while in wild type mice, serum TNF levels are low and induced only by ponatinib (likely from the leukocytes in which ponatinib increases TNF mRNA), while in ApoE-KO mice on HFD, TNF levels are broadly elevated and unaffected by treatment with any of the Abl-TKIs. We propose that by rapidly inducing surface TNFR trafficking, ponatinib increases the sensitivity of ECs to the circulating level of TNF, which may differ depending on the inflammatory state in each study. Clinical data in CML patients supports a high level of systemic inflammation, characterized by elevated TNF levels, even during clinical cancer remission.¹⁹ In addition, as CML patients also have high rates of pro-inflammatory coronary risk factors, this mechanism of ponatinib-induced endothelial sensitization to TNF, could also contribute to the observed synergy between coronary risk factors and thrombotic risk in ponatinib-treated cancer patients. Recent work has also shown that ponatinib activates the NLRP3 inflammasome and induces systemic and cardiac inflammation in pressure-overload models³⁵, highlighting the potential for multiple inflammatory pathways to converge in mediating cardiovascular toxicity. Since inflammasome activation has been implicated in atherosclerosis³⁶, whether this may also contribute to the plaque inflammation, MI, and stroke phenotypes described here remains an important area for future study.

Administration of a small-molecule TNFR inhibitor robustly mitigated ponatinib-induced endothelial adhesion molecule expression, leukocyte recruitment, plaque inflammation, and most importantly, completely rescued the elevated risk of myocardial infarction, stroke, and death in ponatinib-treated SR-BI-mut/LDLR-KO mice. TNF signaling inhibition has been shown to reduce cardiovascular thrombotic events in chronic inflammatory diseases such as rheumatoid arthritis and inflammatory bowel disease.^37,38 However, despite these promising findings, TNFR inhibition is unlikely to be a feasible clinical strategy for preventing ponatinib-induced vascular events in CML patients given its potent immunosuppressive effects and the already heightened infection risk in this population. In our study, TNFR blockade did not reduce ponatinib-induced platelet or leukocyte activation, indicating that these components of thromboinflammation may be regulated independently of endothelial injury. Notably, the prevention of vascular events by TNFR inhibition, despite persistent immune cell and platelet activation, suggests that endothelial protection alone may be sufficient to avert arterial thrombosis. While empirical use of anti-platelet agents is sometimes considered in CML patients receiving Abl-TKIs, clinical trials supporting this approach are lacking, and the associated bleeding risk is significant.³⁹ These results instead raise the possibility that alternative endothelial-protective strategies could offer a more targeted and safer approach to mitigate cardiovascular risk without exacerbating immune suppression or bleeding complications.

While not addressed in this study, prior literature suggests that TNF signaling may promote the survival of leukemia stem cells in CML, raising the possibility that TNFR inhibition could also potentially enhance anti-leukemic therapy under certain conditions.⁴⁰ In a mouse model of CML, TNF blockade with infliximab improved Abl-TKI mediated cancer cell death.⁴¹ Nonetheless, rather than proposing TNFR inhibition as a therapeutic intervention, our findings motivate the exploration of safer, clinically available agents with known endothelial-protective effects. For example, one study demonstrated that pre-treatment of mice with diltiazem reduced ponatinib-induced pro-thrombotic and pro-inflammatory effects, suggesting that repurposing approved cardiovascular drugs could offer potential translational value.^42,43

The upstream mechanism by which ponatinib induces TNFR membrane trafficking is not addressed in this study, yet the differences between the Abl-TKIs supports a role for off-target kinase inhibition. We previously demonstrated that all the Abl-TKIs used to treat CML also inhibit Abl-kinase activity in human ECs, yet imatinib and asciminib, which are more Abl-kinase specific, do not cause EC activation and inflammation¹⁴. This suggests that off target effects mediate the vascular toxicity of some Abl-TKIs. Ponatinib is known to inhibit multiple kinases, including the Src family and VEGFR2, which may contribute to endothelial dysfunction. Indeed, studies mapping the array of kinases inhibited by each Abl-TKI, confirm that drugs that associate with CV ischemic events inhibit a broader kinome^44,45 than those without such toxicity. Yet, a specific kinase has not been implicated that is only inhibited by the vasculotoxic Abl-TKIs (ponatinib, dasatinib, nilotinib), and not by those that lack CV side effects¹⁰. This suggests that some combination of off target kinases causes EC toxicity. Future studies are needed to determine the specific signaling events upstream of ponatinib-induced TNFR2 trafficking. Clarification of the culprit off-target kinases could lead to safer drug development or identification of EC protective strategies for cancer patients.

Biomarkers to predict cancer patients at risk for thrombotic events could help physicians monitor and prevent vascular toxicity in Abl-TKI-treated patients. Serum biomarkers of EC activation have been measured in CML patients, with ponatinib-treated patients found to have significantly higher levels of soluble E-selectin, VCAM1, and ICAM1 compared to imatinib-treated patients, although asciminib was not included in that study⁴⁶. The current study also raises the possibility of considering platelet-leukocyte aggregates (PLAs) as a potential biomarker, which may be expected to be increased in ponatinib-treated patients. PLAs are established markers of platelet activation that have prognostic significance in acute and chronic coronary syndromes⁴⁷. However, since TNFR inhibition did not prevent PLA formation, but did prevent MI and stroke, the data in this study suggests that PLAs may not be expected to predict events in patients. Whether elevated TNF levels or soluble EC adhesion molecules could predict events has not yet been tested in cancer outcome studies.

Asciminib, a new drug recently approved as a first-line therapy for CML but without long term safety data, is the first allosteric Abl-kinase inhibitor, targeting the myristoyl pocket of the kinase, thereby having fewer off-target effects ^8,48. Here we demonstrate that asciminib does not induce endothelial activation, platelet, or leukocyte activation, nor does it promote leukocyte trafficking, plaque inflammation, MI, stroke, or death in mice, supporting its potential advantage over ponatinib. Clinical trial data confirms superior cancer efficacy of asciminib over imatinib with a favorable safety profile relative to other Abl-TKIs, but without sufficient long-term data to determine cardiovascular risk.^8,49 The preclinical data in this study suggests that the EC inflammatory phenotype is necessary for ponatinib-induced plaque rupture, which would predict that asciminib will not promote arterial thrombosis in cancer patients with longer follow-up.

In CML patients undergoing Abl-TKI therapy, the overall frequency of adverse cardiovascular events is comparable between males and females, with some emerging evidence suggesting that underlying cardiovascular risk factors may have greater predictive significance in women.⁵⁰ Consistent with this clinical literature, the impact of ponatinib was similar between the sexes for most of the significant findings. In both male and female ApoE-KO mice, ponatinib significantly increases plaque necrotic core content, leukocyte infiltration, and aortic expression of P-selectin, ICAM1, VCAM1, TNFR1, and TNFR2. Only the modest ponatinib-induced increase in E-selectin and plaque lipid content observed in male mice was not reproduced in female mice.

This study has limitations, including the absence of cancer-specific models. Further research incorporating mouse models of CML would enhance translational relevance. Indeed, preclinical models can never fully recapitulate all the inherent variables in the human condition. We attempted to mitigate this limitation by presenting data from human cells treated in vitro along with in vivo data in multiple models including healthy mice, ApoE-KO mice on high fat diet, to model the risk factor and preexisting coronary disease burden in CML patients, and SR-BI-mut/LDLR-KO mice, to model the spontaneous acute ischemic events we seek to prevent. In each model, we directly compared the 3 Abl-TKIs to vehicle and confirmed drug exposures in vitro and in vivo that are consistent with serum levels in human cancer patients. Since the Abl-TKIs circulate systemically, they are in direct contact with ECs, platelets and leukocytes, hence the cellular effects are more likely to reflect the impact on human cells in patients.

Despite these inherent limitations, the data in this study substantially advances mechanistic understanding of ponatinib-associated arterial thrombosis. We demonstrate that, at clinically relevant concentrations, ponatinib activates ECs, platelets, and leukocytes to express adhesion molecules, inducing leukocyte-platelet aggregation, leukocyte trafficking, and atherosclerotic plaque inflammation. Mechanistically, ponatinib induces endothelial TNFR2 signaling to promote inflammation. We show for the first time that ponatinib induces acute MI, stroke, and death in a mouse model while imatinib, a drug with known vascular safety, does not. Pharmacologic TNFR inhibition prevented the EC activation but not the effects of ponatinib on platelets and leukocytes, yet EC protection was sufficient to prevent leukocyte trafficking, plaque inflammation, MI, stroke, and death in vivo. Finally, asciminib, a newly approved potent Abl-TKI does not induce any of these adverse effects on ECs, platelets, or leukocytes and does not exacerbate acute MI or stroke risk in vivo in mice. The insights provided have immediate clinical relevance, potentially guiding safer treatment decisions for CML patients and paving the way for future vascular-protective interventions.

Supplementary Material

ARRIVE Checklist

Detailed Methods

Major Resources Table

Supplemental Figures S1 – S15

Supplemental Tables S1 – S4

Video Files Movie 1 – Movie 8

References 51 – 62

Clinical Perspective.

What is new?

• The vascular effects of imatinib, ponatinib, and asciminib were compared in vitro and in vivo at serum concentrations consistent with human exposure.

• Ponatinib increased major adverse cardiac events (MACE), including MI, stroke, and mortality in a new mouse model of plaque rupture, mimicking its clinical side effects in cancer patients.

• Ponatinib induced TNFR2-dependent endothelial inflammation in human endothelial cells (ECs) in vitro and enhanced leukocyte trafficking, platelet-leukocyte aggregation, and plaque inflammation in vivo.

• TNF receptor inhibition in vivo prevented ponatinib activation of EC inflammation, but not leukocyte-platelet activation, and EC protection was sufficient to prevent MI and stroke in mice.

What are the clinical implications?

• Ponatinib causes endothelial cell (EC) damage and induces atherosclerotic plaque inflammation and MACE in mice, providing a mechanism for the high risk of myocardial infarction and stroke in ponatinib-treated cancer patients.

• Asciminib, a novel alternative to ponatinib, does not damage ECs or promote plaque inflammation and MACE, and hence may be a safer option for long-term therapy.

• Anti-inflammatory endothelial-protective treatments could be explored for potential to mitigate ponatinib-induced cardiovascular risk.

Nonstandard Abbreviations and Acronyms

Asc

Asciminib

CAM

Cellular adhesion molecule

CML

Chronic myeloid leukemia

EC

Endothelial cell

HCAECs

Human coronary artery endothelial cells

HFD

High fat diet

HUVECs

Human umbilical vein endothelial cells

Ima

Imatinib

IVM

Intravital microscopy

KO

Knockout

MACE

major adverse cardiovascular events

MI

myocardial infarction

NFkB

Nuclear factor kappa-light-chain-enhancer of activated B cells

PSGL1

P-selectin glycoprotein ligand-1

PLA

Platelet - leukocyte aggregate

Pon

Ponatinib

TKI

Tyrosine kinase inhibitor

TNF

Tumor necrosis factor

TNFR

Tumor necrosis factor receptor

TNFRi

Tumor necrosis factor receptor inhibitor

TTC

2,3,5-Triphenyltetrazolium chloride

Veh

Vehicle