Insights and perspectives into the etiology of TKI-induced cardiotoxicity

Abstract

Tyrosine kinase inhibitors (TKIs) are a class of drugs that have significantly improved survival outcomes and revolutionized the treatment landscape for cancer patients. This is due to their ability to suppress the dysregulation of phosphorylation targets contributing to tumor progression; however, many TKIs approved to treat cancer are associated with adverse cardiac events, including potentially lethal cardiotoxicity. Despite extensive research on kinase signaling pathways, the mechanisms that regulate the movement of TKIs across cardiac cell membranes to interfere with the targets associated with TKI-induced cardiotoxicity remain to be fully elucidated. In this review, we focus on (1) summarizing the purported intracellular signaling pathways associated with TKI-induced cardiotoxicity, (2) the interaction of TKIs with membrane transporters, and (3) recent technological and methodological advances that can be leveraged to study the role of membrane transporters in the etiology of TKI-induced cardiotoxicity.

Significance Statement:

Tyrosine kinase inhibitors (TKIs) have transformed cancer treatment but are increasingly recognized for causing cardiotoxicity. Defining the interplay between TKIs and membrane transporters will improve our mechanistic understanding of cardiotoxicity, particularly how intracellular drug disposition may critically influence susceptibility to essential cardiac kinase targets. Improving our understanding of this mechanism will aid in refining cardiac safety assessments of TKIs and enable transporter-informed strategies to mitigate cardiac risk while maintaining anticancer efficacy.

1. Introduction

Severe injury to the heart, an organ with limited regenerative capacity, remains an important complication during preclinical or early phases of clinical drug development, as well as during postmarketing surveillance. Unfortunately, the development of ~30% of drug candidates are discontinued during clinical trials due to safety concerns including cardiac arrhythmias,¹ and many drugs have even been withdrawn from the market due to unacceptable cardiotoxicity.² Moreover, more than 2000 drugs currently in use for the treatment of various diseases are associated with adverse cardiovascular events.³

Over the past 25 years, the Food and Drug Administration (FDA) has approved more than 80 small-molecule tyrosine kinase inhibitors (TKIs) that have been shown to improve outcomes of disease states that include cancer, as well as autoimmune and neurodegenerative diseases. Unfortunately, many of these drugs can cause a broad spectrum of short- and long-term adverse effects associated with the cardiovascular system, including QT interval prolongation and arrhythmia, left ventricular dysfunction, congestive heart failure, ischemia, and myocardial infarction.^4–6 TKI-induced cardiotoxicity is also a recognized cause for early phase drug development discontinuation, because TKI adverse cardiac events are not always predicted with current preclinical safety pharmacology strategies.^7,8 Despite these cardiovascular abnormalities posing a significant increased mortality risk in patients and limiting clinical availability, there is a major gap in knowledge associated with the mechanism by which these events occur. This review outlines the recent advances made in our understanding of TKI-induced cardiotoxicity, the tools and models currently available to study these outcomes, as well as future opportunities in identifying predictive liabilities to develop safer treatment options and improving patient outcomes.

1.1. Forms of drug-induced cardiotoxicity

Drug-induced cardiotoxicity associated with cancer treatment is categorized under 2 classes.⁹ Type I cardiotoxicity is characterized by dose-dependent myocardial damage and is often irreversible, with structural myocyte loss and apoptosis that requires treatment discontinuation and causes permanent cardiac dysfunction. Examples of drugs that induce type I cardiotoxicity include cumulative exposure to anthracyclines (eg, doxorubicin, daunorubicin, epirubicin, and idarubicin)¹⁰ and high-dose alky-lating agents (eg, cyclophosphamide).¹¹ Type II cardiotoxicity is a form of myocardial damage through suppression of signaling cascades that are essential for normal cardiac function. Unlike type I cardiotoxicity, type II is often dose-independent, causes temporary functional impairment without structural damage, and is mostly reversible upon discontinuation of therapy.¹²

Cardiotoxic TKIs are classified as type II drugs; however, the exact underlying molecular mechanisms responsible remain unclear.¹³ Furthermore, not all TKIs targeting a particular protein kinase exert the same form of cardiotoxicity, as this event is driven often irrespective of the TKI’s primary intended target. As such, type II cardiotoxicity induced by TKIs may result from either the pharmacologic action of the drug on the intended therapeutic target involved in tumorigenesis (on-target effect), which is coincidentally also critical for homeostasis of cells within the myocardium and/or vasculature,¹⁴ or due to off-target effects on cardiac homeostasis signaling as a result of the unintended interaction of pleiotropic TKIs with the conserved ATP-binding domain of the kinase superfamily.¹⁵ In addition, TKIs may impair ion channels that play an important role in maintaining the electrical activity and rhythm of the heart. These channels regulate the transmembrane movement of essential ions, such as sodium (Na⁺), potassium (K⁺), and calcium (Ca²⁺), in cardiac tissues to coordinate action potentials responsible for each heartbeat. The disruption or blockade of these channels, such as the hERG (human ether-a-go-go-related gene) potassium channel, can lead to delayed repolarization and prolongation of the QT interval. This prolongation increases the risk of life-threatening arrhythmias and Torsades de Pointes, a potentially fatal form of ventricular tachycardia, which is another hallmark of drug-induced cardiotoxicity. Table 1 summarizes the cardiotoxicity of TKIs by their clinical relevance, and the following section highlights several purported mechanisms in TKI-induced cardiotoxicity.

Table 1.

Cardiotoxicity of tyrosine kinase inhibitors

Tyrosine Kinase Inhibitor	Indication	Primary Target	Cardiac Event	Putative Mechanism	Transporter Interactions		References
					Inhibitor	Substrate
BCR-ABL tyrosine kinase inhibitors
Nilotiniba	Ph+ CML	BCR-ABL	QT prolongation	hERG block hERG regulation	ABCB1b, ABCG2b, SLC2A1b, SLC22A1b, SLC22A2, SLC22A3b, SLC29A1b, SLC47A1b, SLCO1B1, SLCO1B3	ABCB1b, ABCG2b, SLCO1B1, SLCO1B3, SLCO2B1	(16–24)
Ponatiniba	R/R CML Ph+ ALL	BCR-ABL	Arterial thrombosis Hypertension Arrhythmias Heart failure	VEGFR2 S100A8/A9-TLR4-NLRP3-IL1β	ABCB1b, ABCG2b, SLC29A1b, SLC47A1b	ABCB1b, ABCG2b	(17, 25, 26)
Imatiniba	Ph+ ALL GIST	BCR-ABL	Cardiac dysfunction (CHF and LVD)	PDGFR	SLC22A1b, SLC22A2, SLC22A3b, SLC47A1b	SLC22A2, SLC22A5, ABCB1b SLCO1A2b	(17, 18, 27–30)
Dasatiniba	Ph+ CML	BCR-ABL	Hypertension Cardiac dysfunction (CHF, LVD and MI)	SFK PDGFR	ABCB1b, ABCG2b, SLC22A1b, SLC22A2, SLC22A3b, SLC47A1b	ABCA3b, ABCB1b, ABCC3, ABCC4b, ABCG2, SLC22A2, SLC47A1b, SLCO1B3	(16, 17, 22, 31–35)
Vascular endothelial growth factor receptor tyrosine kinase inhibitors
Axitiniba	RCC	VEGFR	Hypertension	VEGFR	SLCO1B1	ABCB1b	(20, 36)
Lenvatiniba	DTC HCC RCC	VEGFR FGFR RET	QT prolongation Hypertension Cardiac dysfunction (CHF and LVD)	hERG block VEGFR PDGFR	ABCB1b	ABCB1b, ABCG2b	(37)
Pazopaniba	RCC	VEGFR PDGFR FGFR	QT prolongation Hypertension Cardiac dysfunction (CHF and LVD)	hERG block VEGFR PDGFR	ABCB1b, ABCG2b, SLCO1B1, SLC22A1b, SLC47A1b, SLC47A2	ABCB1b, ABCG2b, SLCO1B1, SLC22A1b	(20, 23, 31, 38, 39)
Sunitiniba	GIST RCC	VEGFR PDGFR	QT prolongation Cardiac dysfunction (CHF, LVD)	hERG block AMPK PDGFR RAF VEGFR	ABCB1b, ABCG2b, SLC22A1b, SLC22A2, SLC47A1b	ABCB1b, ABCG2b	(17, 31, 40, 41)
Vandetaniba	MTC	VEGFR EGFR RET	QT prolongation Torsades de Pointes	hERG block	ABCB1b, ABCC1b, ABCG2b, SLC22A1b, SLC22A2, SLC22A3b, SLC47A1b, SLC47A2	ABCB1b, ABCG2b, SLC22A2, SLC47A1b, SLC47A2, SLCO1B1, SLCO1B3	(17, 31, 42–44)
Rapidly accelerated fibrosarcoma proto-oncogene kinase inhibitors
Regorafeniba	CRC GIST HCC	RAF VEGFR	Hypertension Cardiac dysfunction (ischemia and MI)	PDGFR RAF VEGFR	ABCB1b, ABCG2b	ABCB1b, ABCC2, ABCG2b, SLCO1B1	(22, 31, 45–48)
Sorafeniba	HCC RCC DTC	RAF VEGFR	QT prolongation Hypertension Cardiac dysfunction (ischemia and MI)	RAF VEGFR	ABCB1b, ABCG2b	ABCB1b, ABCG2b, ABCC2, SLCO1B1, SLCO1B3	(30, 31, 49, 50–52)
Mitogen-activated protein kinase tyrosine kinase inhibitors
Cabozantiniba	RCC HCC	MET VEGFR RET ROS1	Hypertension	VEGFR	ABCB1b, ABCG2b, SLCO1B1	ABCC2	(20, 53–55)
Selumetiniba	NF1	MEK	Cardiomyopathy Tachycardia Cardiac dysfunction (LVD)	MEK/ERK	-	ABCB1b, ABCG2b	56
Anaplastic lymphoma kinase tyrosine kinase inhibitors
Alectiniba	NSCLC	ALK	Bradycardia	L-type Ca2+ block	ABCB1b	ABCB1b	57
Brigatiniba	NSCLC	ALK	Hypertension Bradycardia	L-type Ca2+ block EGFR PI3K	SLC22A3b, SLC29A1b	ABCB1b, ABCG2b	(57–59)
Ceritiniba	NSCLC	ALK ROS1	QT prolongation Bradycardia	hERG block PI3K	SLC22A3b, SLC29A1b, ABCB1b, ABCG2b	ABCB1b ABCG2b	(57, 58, 60, 61)
Crizotiniba	NSCLC	ALK	QT prolongation	PI3K	SLC2A4b, SLC22A3b, SLC29A1b	ABCB1b, ABCG2b, SLCO1B1, SLCO1B3	(57, 58, 62, 63)
Lorlatiniba	NSCLC	ALK ROS1	Hyperlipidemia Atrioventricular block (PR prolongation)	Nav1.5 block	ABCB1b, ABCG2b, SLC29A1b	ABCB1b	(57, 64, 65)
Epidermal growth factor receptor tyrosine kinase inhibitors
Lapatiniba	Breast cancer	EGFR HER2	QT prolongation Cardiac dysfunction (LVD)	EGFR HER2	ABCB1b, ABCG2b	ABCB1b, ABCG2b	66
Osimertiniba	NSCLC	EGFR	QT prolongation Cardiomyopathy Cardiac dysfunction (LVD)	hERG block EGFR HER2	ABCG2b	ABCB1b, ABCG2b	(67, 68)
Bruton tyrosine kinase inhibitors
Acalabrutiniba	CLL/SLL MCL	BTK	Atrial fibrillation Hypertension	TEC/PI3K/AKT	ABCG2b, SLC47A1b	ABCB1b, ABCG2b	69
Ibrutiniba	CLL/SLL MCL	BTK	QT shortening Arrhythmia Atrial fibrillation Hypertension	CSK SFK TEC	SLC22A1b, SLC22A2, SLC22A3b, SLC29A1b, SLC47A1b	SLC22A2, ABCB1b	(16, 17, 35, 57)
Zanubrutiniba	CLL/SLL MCL	BTK	Arrhythmia Hypertension	TEC/PI3K/AKT		ABCB1b	70

Note: “−” indicates literature relevant transporters not available at the time of writing.

AKT, protein kinase B; ALK, anaplastic lymphoma kinase; AMPK, AMP-activated protein kinase; BCR-ABL, breakpoint cluster region-Abelson fusion tyrosine kinase; CHF, congestive heart failure; CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; CRC, colorectal cancer; DTC, differentiated thyroid carcinoma; ERK, extracellular signal-regulated kinase; GIST, gastrointestinal stromal tumor; HCC, hepatocellular carcinoma; L-type Ca²⁺, L-type calcium channel; LVD, left ventricular dysfunction; MCL, mantle cell lymphoma; MEK, MAPK/ERK kinase; MET, hepatocyte growth factor receptor; MI, myocardial infarction; MTC, medullary thyroid carcinoma; Nav1.5, voltage-gated cardiac sodium channel; NF1, neurofibromatosis type 1; NSCLC, non-small cell lung carcinoma; Ph+ ALL, Philadelphia chromosome-positive acute lymphoblastic leukemia; Ph+ CML, Philadelphia chromosome-positive chronic myeloid leukemia; RAF, rapidly accelerated fibrosarcoma serine/threonine kinase; RCC, renal cell carcinoma; RET, rearranged during transfection proto-oncogene receptor; ROS1, c-ROS oncogene 1 receptor tyrosine kinase; R/R CML, relapsed/refractory chronic myeloid leukemia; SFK, Src family kinases; TEC, TEC family tyrosine kinases.

^aCardiac event identified based on prescribing information, grouped by boxed warning or warnings and precautions.

^bLocalization of transporter confirmed in the heart.

2. Signaling events associated with TKI-induced cardiotoxicity

Hyperactive epidermal growth factor receptor (EGFR, ErbB1) signaling pathways have been identified as a contributor to non-small cell lung cancers, leading to the development of EGFR-TKIs that counteract these aberrant kinase activities, yielding notable clinical advancements.^71,72 In recent years, concerns about the cardiac safety of EGFR-TKIs have risen, prompted by observations of osimertinib-associated cardiotoxicity and the discontinuation of rociletinib in phase 3 trials owing to QT prolongation.^73–76 It is presently unclear how or if EGFR itself is associated with cardiac function, and the only link of EGFR to cardiac outcome involves its association with protection from catecholamine-mediated cardiotoxicity.⁷⁷ Although osimertinib has been implicated in cardiac adverse events, the cardiotoxic risk associated with several other first and second generation EGFR inhibitors, including gefitinib, erlotinib, and afatinib, is low or negligible.^78–80 Lazertinib, the newest FDA-approved EGFR-targeting TKI, exhibits potent EGFR inhibition but reportedly has a low incidence of cardiotoxicity compared with osimertinib.⁸¹ Inhibition of the human epidermal growth factor receptor 2 (HER2, ErbB2) has instead been proposed as a major contributing factor to EGFR-TKIs-induced cardiotoxicity.⁸² This is largely based on the role of HER2 in cardiac development and emerging reports of adverse cardiac events with the HER2-targeting monoclonal antibody trastuzumab.⁸³ The HER2 signaling pathway is responsible for promoting cellular survival through repair and stress adaptations, maintaining mitochondrial integrity and sarcomere architecture.⁸⁴ Complete genetic deficiency of the ErbB2 gene in mice is embryonic lethal,⁸⁵ whereas conditional knockout of the ErbB2 gene in mouse ventricular cardiomyocytes is associated with the development of dilated cardiomyopathy.^84,86 Disruptions in this pathway also increase sensitivity to anthracycline-induced cardiac injury.⁸⁷ However, a direct role of HER2 in mediating TKIs cardiotoxicity remains questionable, considering that HER2-targeting TKIs, such as afatinib and neratinib, are associated with low risk of cardiotoxicity.⁷⁸ Given the discrepancy observed between the role of ErbB family kinases in cardiac development and the lack of cardiotoxicity from the majority of the ErbB-TKIs, future studies are needed to elucidate the potential on-target or off-target kinase(s) with which osimertinib interferes to understand the mechanism behind osimertinib-induced cardiotoxicity.

2.1. Direct on-target cardiotoxicity

Kinases that exacerbate tumorigenesis may also be coincidentally essential for homeostasis of cardiomyocytes and/or the vasculature. For example, vascular endothelial growth factor receptor (VEGFR) signaling plays a crucial role in maintaining vascular integrity, endothelial function, and myocardial repair.⁸⁸ Inhibiting the VEGF pathway has also been shown to disrupt nitric oxide production and increase endothelin-1 levels, which promotes vasoconstriction.⁸⁸ As such, VEGFR inhibitors, including sorafenib, regorafenib, sunitinib, axitinib, vandetanib, and pazopanib, widely used to suppress tumor angiogenesis, have been increasingly associated with cardiovascular toxicities such as hypertension, coronary artery disease, heart failure, and arterial thrombotic events.⁸⁹ Up to 73% of patients receiving VEGFR-TKIs for renal cell carcinoma experience cardiotoxicity, with hypertension accounting for 55% of these cases.⁹⁰ However, the extent and types of adverse cardiac events observed cannot be fully explained by VEGFR inhibition alone, suggesting that additional on- and off-target effects associated with VEGFR inhibitors may also contribute to cardiotoxicity. Furthermore, several discontinued and investigational VEGFR-TKIs, including semaxanib,⁹¹ motesanib,⁹² cediranib,⁹³ and vatalanib,⁹⁴ are also associated with adverse cardiac events consistent with approved VEGFR agents.

2.2. Indirect off-target cardiotoxicity

The conserved ATP binding domain of the kinase superfamily can lead to low specificity and unintended inhibition of kinases critical for cardiac homeostasis. For example, sunitinib, which displays dose-dependent cardiotoxicity,⁹⁵ is known to inhibit AMP-activated protein kinase (AMPK) activity in addition to its main target, VEGFR, at clinically relevant plasma concentration.^96,97 AMPK plays a critical role in cardiac homeostasis and exerts protective effects against ventricular hypertrophy and dysfunction, as evidenced by cardiac hypertrophy in mice genetically deficient in AMPKα2.⁹⁸ Angiogenesis inhibition by sunitinib in a rat model was found to increase blood pressure accompanied by elevation of circulating endothelin 1 level in a reversible manner.⁹⁹ In addition, mice expressing AMPK lacking its intracellular tyrosine kinase domain displayed cardiac remodeling and prolonged QRS duration.¹⁰⁰ Inhibition of platelet-derived growth factor receptor (PDGFR) signaling has also been suspected to be a contributing factor to sunitinib-induced cardiotoxicity. Investigation of cardiomyocyte-specific PDGFR-β knockout mice identified a role for PDGFR in regulating cardiac response to stress via the Akt and MAPK pathways.¹⁰¹ Cardiac performance in rats with myocardial infarction was also significantly improved in animals that received PDGF peptide injections compared to saline treatment.¹⁰² Collectively, the above studies demonstrate that a single TKI such as sunitinib can exert cardiotoxic effects through simultaneous impairment of multiple kinase pathways. Nonetheless, further investigation is warranted to accurately elucidate the specific roles of these individual kinases in maintaining cardiac homeostasis to improve our understanding of which kinase signaling pathways to avoid in drug development and mitigate TKI-induced cardiotoxicity.

Although sunitinib cardiotoxicity may be attributed to both on-target and off-target inhibition of VEGFR, PDGFR, and AMPK, similar concerns have emerged with other multitargeted VEGFR-TKIs, such as sorafenib and regorafenib. In addition to VEGFR, sorafenib and regorafenib interfere with the RAF/ERK pathway. RAF1, a key kinase of this pathway, is essential for maintaining cardiac structure and function by regulating the MAPK signaling cascade.¹⁰³ The role of RAF1 in cardiac physiology has been extensively studied, particularly due to genetic variants in RAF1 being associated with cardiomyopathy.¹⁰⁴ Cardiac-specific Raf1 deficient mice also exhibit dilated cardiomyopathy and diminished myocardial integrity.¹⁰⁵ Furthermore, B-RAF, a member of the RAF family, has been found to upregulate hERG protein abundance as well as its activity in oocytes, while chemical inhibition of B-RAF downregulates the membrane expression of hERG.¹⁰⁶ Despite this evidence, there is insufficient published work linking RAF signaling inhibition to the mechanism underlying sorafenib and regorafenib cardiotoxicity, especially considering that cardiac events reported with other RAF inhibitors, such as vemurafenib, dabrafenib, and encorafenib, are complicated by combination with MEK inhibitors, such as trametinib, cobimetinib, and binimetinib.^103,107

ABL1 is a ubiquitous non-receptor tyrosine kinase essential for cell cycle regulation, DNA damage response, and cytoskeletal dynamics¹⁰⁸ that also appears to be essential for normal cardiac growth and development, as well as adaption and repair. Inhibition or dysregulation of ABL1 can therefore prompt cardiomyocyte susceptibility to injury and apoptosis. Although BCR-ABL-TKIs such as nilotinib, ponatinib, and dasatinib are associated with cardiotoxicity, mechanistic studies using imatinib or structural analogs lacking ABL1 binding ability did not preserve isolated neonatal cardiomyocytes from injury.¹⁰⁹ Therefore, aside from nilotinib, which has potent and direct electrophysiologic effects on hERG channels, causing the risk of QT interval prolongation, cardiotoxicity associated with BCR-ABL-TKIs is likely multifactorial (eg, ponatinib also potently inhibits VEGFR). Indeed, the multifactorial effect of BCR-ABL-TKIs is consistent with the diversity of the adverse cardiac events among these TKIs that have a wide range of biochemical potency against ABL1.^110–112 Notably, among the BCR-ABL-TKIs, bosutinib displays fewer cardiovascular-related toxicity events than imatinib, despite its higher potency against ABL1.^112,113 This lower incidence is thought to be due to lack of activity against PDGFR and Kit.¹¹³ Although these observations suggest that inhibition of ABL1 does not fully explain clinical cardiotoxicity, the precise mechanism remains unclear, and the diverse cardiac events should be further explored.

Cardiotoxicity associated with the use of ibrutinib has also been attributed to off-target effects rather than the effects on its main target, Bruton tyrosine kinase (BTK), due to its expression being mainly in hematopoietic and in B cells and that loss of Btk in mice causes no signs of cardiotoxicity.^114,115 Ibrutinib appears to induce atrial fibrillation via off-target inhibition of the c-terminal Src kinase (CSK).¹¹⁴ This finding is supported by observations that a pan Src family kinase inhibitor and genetic knockdown of Csk alters hERG electrophysiology and increases neonatal rat cardiomyocyte susceptibility to apoptosis.^116,117 Consistent with these effects, cardiac-specific knockout of Csk has been shown to increase cardiac interstitial fibrosis, resembling clinical phenotypes associated with ibrutinib cardiotoxicity.¹¹⁴ Moreover, second and third generation BTK inhibitors (eg, acalabrutinib, zanubrutinib, and pirtobrutinib) that have improved selectivity for BTK and fewer off-target effects, including reduced potency against CSK, are associated with lower incidence of atrial fibrillation.

Direct inhibition of the hERG channel has been shown to prolong cardiac repolarization, resulting in an extended action potential, which clinically manifests as QT interval prolongation. Although many TKIs have been shown to exhibit direct hERG blockade in in vitro hERG assays (Table 1), it is important to note that expression and activity of ion channels, including hERG, are dependent on protein kinases.¹¹⁸ For example, acute and chronic exposure to nilotinib and vandetanib produced proarrhythmic effects in human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) that were associated with reduced membrane protein expression of hERG. Although the underlying off-target kinase(s) impacted by nilotinib and vandetanib that mediated this outcome remains to be fully elucidated, a serum and glucocorticoid kinase 1 (SGK1) activator has been shown to reverse proarrhythmic effects and restore membrane expression.^118,119 Because cancer patients require long-term treatment with these TKIs, these findings underscore the need to consider chronic exposure conditions when assessing the liabilities of hERG block by TKIs in humans.

In addition to the above kinases, the cardiomyocyte-restricted deletion of the insulin receptor (IR) has also been shown to reduce the transcript and protein expression of multiple potassium channels that are critical for ventricular repolarization.¹²⁰ As such, the combined deletion of the insulin receptor and the insulin-like growth factor 1 receptor (IGF1R) in cardiomyocytes can lead to reduced cardiomyocyte viability and heart failure.^121,122 IGF1R activation and phosphorylation has also been associated with exposure and resistance to the cardiotoxic TKI, ponatinib, which could implicate this kinase as a regulator of a compensatory pathway that provides protection from ponatinib.⁸⁰ Notably, cotreatment with insulin growth factor 1 (IGF1) and insulin has been shown to mitigate ponatinib-induced toxicity, suggesting a protective role for IGF1R and IR signaling.⁸⁰ The downstream signaling of IGF1R, IR, and the IGF1R/IR complex converges on a common pathway that begins with the phosphorylation of insulin receptor substrates (IRS1/2), followed by activation of the phosphoinositide 3-kinase (PI3K) and AKT pathway.^121,123 Studies in rodents have demonstrated that cardiomyocyte-specific deficiency of Irs1 confers protection against heart failure, whereas deficiency of Irs2 leads to ventricular arrhythmias.^124,125 Additionally, mice lacking PI3K signaling exhibit longer QT intervals compared with wild-type controls.¹²⁶ Considering the pleiotropic role of the PI3K pathway in maintaining action potentials in cardiac tissues, and that various TKIs known to interfere with the PI3K pathway (such as dasatinib, crizotinib, and sunitinib) exhibit QT interval prolongation in addition to hERG block,^127–129 further studies are warranted to investigate the complexity of TKIs that target kinases within this pathway.

3. Drug accumulation in cardiac tissues

Small-molecule TKIs target protein kinases at intracellular domains, and therefore, intracellular drug concentrations within sensitive cardiac cell types are expected to be a critical determinant of the onset and severity of TKI-induced cardiotoxicity. This is consistent with the fact that TKIs can induce a broad spectrum of hERG-independent abnormalities and activate cell death pathways, even in isolated cardiomyocytes.¹³⁰ hERG blockers gain access to their binding site on the α-subunit of the ion channel intracellularly^131,132; therefore, intracellular concentrations of such agents also represent a key determinant of drug action on cardiac repolarization. Past studies have relied on measuring systemic plasma concentrations to predict cardiotoxicity by correlating drug- and concentration-responses to free plasma levels in animals and humans to estimate a cardiovascular safety margin. However, this approach is limited due to systemic plasma concentrations not always accurately reflecting distribution and intracellular concentrations in different cardiac cell types. Thus, there is an urgent need to understand the molecular mechanisms of TKI-induced cardiotoxicity, not only by identifying kinase targets but also regulators of intracellular drug accumulation that promote toxicity when sufficient concentrations are reached.

In many prior studies evaluating the transmembrane movement of TKIs, it was largely surmised that the predominant mechanisms of uptake and efflux occur through direct movement of the non-ionized drug through the phospholipid bilayer (passive diffusion).¹³³ However, the saturable uptake process of many TKIs in cell-based models and their ionizable properties at physiological pH instead support the involvement of one or more unknown members of the solute carrier (SLC) and ATP binding cassette (ABC) family of membrane transporters,^134,135 which contains 65 gene families with 458 different human transporter genes that can be highly diverse in structure, function, and tissue expression. The expression of these SLCs and ABCs can drive the accumulation or removal of TKIs in cardiac tissues to reach levels that prompt unwanted cardiac adverse events. In addition, cancer patients are at a particularly higher risk for drug-drug interactions (DDIs) due to concomitant medications for disease treatment, symptom management, and other comorbid conditions, together with the chronic daily administration of TKIs. As such, these DDIs may increase patients’ risk for deleterious cardiac injury associated with the use of TKIs due to potential inhibition of efflux from cardiac tissues, or inhibition of uptake transporters involved in clearance pathways, leading to excess plasma concentrations. Together, the intricate balance between drug uptake and efflux sets intracellular drug concentrations and thus affects their activity on cardiac kinases or ion channels that are essential for homeostasis. As such, membrane transporters as regulators, or prerequisite mechanism, of tissue disposition of TKIs could prompt discrepancies between plasma and tissue concentrations that would lead to poor prediction of TKI-associated cardiotoxicity. This has been reported for other xenobiotic agents that accumulate extensively in the heart temporally before the development of cardiotoxicity.^31,136–139

3.1. Role of cardiac transporters in xenobiotic disposition

To prompt a cellular response, cardiotoxic molecules must accumulate to sufficient concentrations that enable interaction with intracellular targets. The cellular accumulation of molecules can be highly dependent on SLC and ABC transporters. Numerous SLC and ABC transporters are expressed in cardiac tissues and provide essential roles in energy metabolism, nutrient and ion homeostasis, and handling of xenobiotics. This includes several cationic-type transporters in human heart cardiomyocytes and vasculature,^140–143 such as SLC22A1 (OCT1), SLC22A3 (OCT3), SLC22A4 (OCTN1), SLC22A5 (OCTN2), SLC22A16 (OCT6), and SLC47A1 (MATE1). Evidence supporting a role for some of these transporters as mediators of selective cardiac distribution already exists. For example, despite ubiquitous expression in mice, deficiency of Oct3 disrupts uptake of the neurotoxin 1-methyl-4-phenylpyridinium (MPP⁺) into cardiac tissue,¹⁴⁴ and similar observations have been reported for dehydrocorydaline,¹⁴⁵ metformin,¹⁴³ doxorubicin,¹⁶ and meta-iodobenzylguanidine (mIBG).^146,147 In vitro and clinical genetic in vivo studies have shown that diminished OCTN1 function promotes quinidine intracellular accumulation, hERG inhibition, and increased occurrence of Torsades de Pointes.^142,148 Genetic or pharmacologic perturbations of the efflux transporter, MATE1, which is also highly expressed in cardiomyocytes,¹⁴³ has been shown to exacerbate dofetilide proarrhythmia by enhancing intracellular retention in isolated myocytes ex vivo and QT interval prolongation in vivo.¹³⁶

Multiple members of the organic anion-transporting polypeptide (OATP) family are also reportedly expressed in the heart, including but not limited to SLCO1A2 (OATP1A2), SLCO2A1 (OATP2A1), SLCO2B1 (OATP2B1), SLCO3A1 (OATP3A1), SLCO4A1 (OATP4A1), and SLCO5A1 (OATP5A1).^149,150 OATPs are predominantly expressed in vascular endothelial cells,^149,151,152 suggesting a localized role in controlling uptake of xenobiotics in the heart. For example, elevated doxorubicin systemic concentration and a modest but lower heart to plasma ratio has been reported in mice genetically deficient in all isoforms of Oatp1a/1b compared with wildtype animals, possibly due to the drug being a substrate for OATP1A2.¹⁵³ Consistent with these findings, loss of OATP1A2 in cardiac-derived induced pluripotent stem cells reportedly prevents doxorubicin-induced cardiac cell death, and OATP1A2 overexpression exacerbates doxorubicin-induced cell death.¹⁵⁰

Many ABC efflux transporters are also expressed in cardiomyocytes, atria, ventricles, and endothelial cells, including ABCB1 (P-glycoprotein; P-gp), ABCG2 (breast cancer resistance protein; BCRP), ABCC1/4/5/6/9 (multidrug resistance-associated protein; MRP1/4/5/6/9).^154–158 Independent of mediating drug accumulation, deficient ABC transport activity can be associated with adverse cardiovascular outcomes due to dysregulated cholesterol and lipid homeostasis.¹⁵⁹ Consistent with the latter, ABCC6 and ABCC9 deficiency is associated with cardiac complications due to disrupted cellular homeostasis.^160–162 Abcg2-deficient mice are more susceptible to cardiac hypertrophy after aortic constriction¹⁶³ and suffer from incomplete recovery after myocardial infarction,¹⁶⁴ indicating that the transporter mediates angiogenic repair after mechanical stressors. Deficiency of Abcb1a/b transport does not exhibit pronounced cardiovascular dysfunction alone; however, it is associated with increased cardiac exposure to QT interval-prolonging drug substrates such as romidepsin and doxorubicin.^138,139

3.2. Interaction of TKIs with cardiac membrane transporters

Among cardiotoxic TKIs, most are recognized substrates or inhibitors of ABCB1 and ABCG2 (Table 1). This substrate-inhibitor duality is not surprising given the broad substrate recognition of these transporters, in part due to polyspecific binding at low and high concentrations, multiple binding sites, and conformational plasticity.^165–167 Interference of such transporter activity, either by genetic polymorphisms or DDIs, can consequentially increase intracellular TKI concentrations and further disrupt key signaling pathways. Indeed, ABCB1-overexpressing cells can reduce the inhibitory potential of imatinib due to inadequate concentration accumulation at target kinases.¹⁶⁸ Increased ABCG2 abundance has also been shown to reduce gefitinib interaction with EGFR,¹⁶⁹ while variants with reduced function have resulted in greater cellular sensitivity to erlotinib, gefitinib, and lapatinib.^170,171 Additionally, ABCC4 reportedly recognizes dasatinib as a substrate, and deficient activity reduces not only tumor accumulation, but also alters the pharmacokinetic profile of dasatinib in mice.¹⁷² Despite the above evidence, the role of ABC transporters in mediating TKI concentrations and response in cardiac tissue remains largely understudied. This includes a lack of clarity of whether ABC transporter drug interactions contribute to TKI cardiotoxicity in patients subjected to polypharmacy. Moreover, the role of other transporter families in mediating TKI cardiotoxicity also needs to be considered.

Many TKIs are positively charged or can be ionized into cations under physiological pH, yielding optimal substrates for cation transporters. In fact, several TKIs are reported substrates or inhibitors of cation transporters (Table 1). For example, imatinib has been identified as a substrate for MATE1, which mediates cellular sensitivity to the drug,²⁷ and MATE1 transport activity has also been shown to be sensitive to TKI exposure,^17,32 although the nature of this inhibition remains unclear and requires further study. Several cardiotoxic TKIs, including ibrutinib and vandetanib, are also recognized substrates or inhibitors of OCT2 (SLC22A2). OCT2 is not expressed in cardiac tissue but does share a large overlap in substrate specificity with OCT1/3, which are expressed in both human and rodent hearts.^140,142 Therefore, it is conceivable that the cardiotoxic potential of ibrutinib and vandetanib could be dependent on OCT3-mediated uptake, a notion partly supported by the fact that ibrutinib and vandetanib inhibit OCT3 uptake of tetraethylammonium.¹⁶ Crizotinib, brigatinib, ceritinib, dasatinib, nilotinib, and sunitinib, which are associated with cardiotoxic risk, are also capable of inhibiting OCT3 (Table 1).^16,32,58

In addition to cation SLCs, OATP transporters have been associated with regulating disposition of several cardiotoxic TKIs (Table 1). For example, imatinib has been identified as a substrate of OATP1A2, and patients with genetic variants of this transporter show reduced imatinib clearance.^173,174 Larotrectinib was also found to be a substrate of OATP1A2 (but not OATP1B1, OATP1B3, or OATP2B1) in overexpressing cells, and deficiency of Oatp1a/b transporters in mice increased larotrectinib plasma exposure.¹⁷⁵ Beyond OATP1A2, OATP2B1 has been shown to mediate disposition of erlotinib in vitro in a pH dependent manner, although no significant change in erlotinib or its main metabolite plasma concentrations were observed in vivo in Oatp2b1 deficient rats.^165,176 Whether deficiency of OATP2B1 alters accumulation within specific tissues remains unclear. Erlotinib and crizotinib have also been reported to be OATP1B1 and OATP1B3 substrates in vitro using overexpressing cells,¹⁶⁵ and a recent study using a competitive counter flow assay identified pazopanib and many other FDA-approved TKIs as potential OATP1B1 substrates.³⁸ Consistent with such observations, deficient Oatp1a/b transport in mice is associated with reduced hepatic pazopanib concentrations³⁸ or increased plasma concentration of sorafenib-glucuronide, which is also a substrate of OATP1B1 and OATP1B3.⁴⁹ Regardless of these findings, evidence supporting a direct role of OATP transporters in mediating TKI accumulation in cardiac tissues is lacking. Therefore, further studies are needed to verify the contribution of OATPs in regulating TKI distribution into cardiac tissue.

3.3. Membrane transporters essential to cardiac homeostasis

Cardiac function requires extensive energy to maintain systemic blood flow and is consequentially dependent on many essential nutrient transporters, including the glucose transporters (GLUTs) GLUT1 and GLUT4, which are highly expressed in heart tissue.¹⁷⁷ In a humanized cardiac-specific model, GLUT1 overexpression was shown to be protective against cardiac dysfunction after exposure to mechanical stress.¹⁷⁸ Conversely, cardiac-specific deletion of Glut4 has been shown to predispose the heart to ischemic injury.¹⁷⁹ Studies have shown that imatinib, gefitinib, nilotinib, and pazopanib reduce glucose uptake, possibly through reduced cell surface expression of GLUT1.^18,180,181 Reduced GLUT4-mediated glucose uptake by crizotinib has also been reported.⁶² Nonetheless, further studies are required to elucidate whether TKI-mediated disruption of glucose uptake contributes to clinically observed cardiotoxicity.

Concentrative nucleoside transporter (CNT2/3) and equilibrative nucleoside transporter (ENT1/2/4) transcript expression have been reported in cardiac tissue.^182–186 Although the precise cardiac distribution of CNT2/3 remains unclear, sex-dependent differences in overall transcription products have been identified in rodents.¹⁸⁷ ENT1 and ENT4 transcripts have been characterized in different regions of cardiomyocytes, cardiac smooth muscle, as well as in the sinoatrial node and right atrium.^184,188,189 Each of these transporters can regulate cardiac concentrations of adenosine, which maintains cardiac homeostasis through interaction with the A1 and A2 receptors that regulate heart rhythm and blood flow in the vascular system.^186,189,190 Given the importance of adenosine in maintaining cardiac homeostasis, disturbance of nucleoside transporters have been shown to elevate extracellular adenosine levels leading to vasodilation.¹⁹¹ Several TKIs associated with hypotension, including ceritinib, crizotinib, lorlatinib, neratinib, nilotinib, ponatinib, imatinib, and ibrutinib, have been shown to reduce the function of ENTs and CNTs in vitro^19,57,192; however the mechanism, if any, by which these transporters contribute to TKI-induced cardiotoxicity remains unclear. Indeed, further investigation is required considering that deficient activity of Ent1 alone could have benefits in protecting from cardiac damage, given that Ent1-null mice are protected from ischemia—reperfusion injury.^183,193

Finally, OATP transporters are known to mediate cellular accumulation of endogenous prostaglandins and thyroid hormones that are essential to cardiac homeostasis.^194,195 Therefore, in addition to mediating disposition of TKIs, as described above, disruption of OATP natural substrate disposition could also contribute to TKI-induced cardiotoxicity, given that many TKIs are inhibitors of OATPs (Table 1).^20–22,196 Nonetheless, the observations outlined above indicate a role of SLC and ABC transporters in regulating TKI cardiotoxicity, but evidence is limited, and there is an urgent need to clarify their contributions.

4. Current and emerging technologies and PKPD models in predicting cardiotoxicity

The standard battery of investigational new drug-enabling cardiovascular safety pharmacology studies includes (1) in vitro tests of the inhibitory potential on channels using cell-based models and (2) in vivo tests on electrophysiological and hemo-dynamic endpoints using telemetry-based rat or nonrodent (commonly dog or nonhuman primate) models. However, the clinical incidence of cardiovascular-related liabilities has frequently been underestimated, even in registration trials.¹ This has included underestimation of TKI cardiotoxicity.

In vitro hERG and ion channel inhibition assays are critical early tools in nonclinical development to identify molecules with potential to cause QT interval prolongation and cardiac arrhythmias. These studies use heterologous expression models in HEK293 or CHO cells to test concentrations required for half maximal inhibition (IC₅₀) on cardiac action potential and repolarization. This IC₅₀ value quantifies the potency of channel blockade, with lower IC₅₀ values indicating stronger inhibition.^197–200 In vivo studies are then performed to confirm whether in vitro findings (positive or negative) translate into functional electrophysiological effects. These studies evaluate the electrocardiograms (ECGs) and hemo-dynamic endpoints from conscious telemetered animals after receiving escalating single or multiple doses. In tandem, concurrent pharmacokinetic samples are collected to quantify plasma concentrations and determine concentration-QT relationships. Taken together with the anticipated unbound concentrations in humans, a safety margin is estimated to predict the likelihood of QT interval prolongation and proarrhythmic risk in humans at efficacious and clinical exposures.^197–200 However, these current models prioritize detecting cardiotoxicity while ignoring the TKI disposition. For example, in vitro HEK293-hERG expression models do not natively express kinases involved in TKI-induced cardiotoxicity or cardiac membrane transporters affecting their cardiac disposition. Furthermore, the translational findings of electrophysiological effects in nonclinical animal species may be affected by species-dependent transport affinity and substrate recognition by transporters.^201–203 As a result, the cardiac liabilities in humans may be overpredicted if TKI distribution and accumulation is lower than plasma levels or underestimated if distribution and accumulation is greater than plasma levels.

Recent efforts have been made to integrate hiPSC-CMs as a more accurate model of human cardiac physiology that enables recreation of pathological phenotypes for early detection and mitigation of drug-induced cardiotoxicity.^204,205 hiPSC-CMs not only recapitulate the full complement of cardiac-specific regulatory signaling pathways and ion channels targeted by TKIs but also cardiac membrane transporters involved in distribution and accumulation in cardiomyocytes. Furthermore, for TKIs with unknown propensities for cardiac membrane transporters, hiPSC-CMs can be used in transportome-wide gene knockout strategies using high-throughput genetic screening libraries (eg, SLC CRISPR/Cas9 library) coupled with single-cell RNA sequencing (scRNA-seq) to determine which members of the transporter families play a role in the cardiac uptake or efflux of TKIs.^16,206 Although hiPSC-CMs provide an in vitro platform to understand the mechanistic insight of TKI-induced cardiotoxicity, their predictive power can be enhanced by integrating with mathematical models to facilitate hypothesis generation, predict individual-specific responses, and identify potential cardioprotective strategies. Indeed, improving the toolbox of nonclinical assays to predict cardiac liabilities will enable optimal clinical use of TKIs with known or unknown propensity for cardiotoxicity in humans.

4.1. Mathematical models to study TKI cardiotoxicity

The complexities underlying TKI-associated cardiotoxicity, such as cardiomyocyte signaling processes related to cellular survival, energy homeostasis, and excitation-contraction coupling, require a holistic platform for studying and predicting drug safety.²⁰⁷ Systems modeling approaches, such as the one used by Grabowska et al,²⁰⁸ can be employed to generate hypotheses underlying signaling responses. The developed logic-based differential equation model links TKI-induced signaling responses to intrinsic apoptosis in cardiomyocytes, thereby identifying potential regulators of TKI-induced apoptosis, such as reactive oxygen species, as potential therapeutic targets to mitigate cardiotoxicity. In another example, Shim et al²⁰⁹ combined transcriptomic data with a systems model of electrophysiology and contraction in human-induced pluripotent stem cell-derived cardiomyocytes hiPSC-CMs to investigate individual differences in response to 26 TKIs. Gene expression data collected from 2 cell lines after each treatment were used to scale model parameters, enabling individual predictions of how TKIs alter action potentials, intracellular Ca²⁺ transients, and sarcomere shortening. Model predictions were in good agreement with experimental data, providing confidence that this framework can be applied to investigate individual-specific responses to drug-induced cardiotoxicity. Such modeling approaches demonstrate how systems modeling can generate testable hypotheses, prioritize follow-up experiments, and identify potential cardioprotective targets. However, the lack of concentration-response relationships limits the direct translation of these findings to the clinic and could be considered in the future.

Quantitative systems pharmacology (QSP) and quantitative systems toxicology (QST) merge principles of systems biology and pharmacokinetics/pharmacodynamics to gain a quantitative understanding of how drugs modulate biological systems.²¹⁰ By integrating drug exposure, cellular/organ physiology, drug pharmacology, and toxicodynamic (TD) biomarkers, QST models can be used to increase translational confidence and bridge the gap between preclinical results and clinical outcomes.²¹¹ Wan et al²¹² employed a QSP-pharmacokinetic/toxicodynamic framework to extrapolate in vitro hiPSC-CMs data and predict the decline in left ventricular ejection fraction for 21 TKIs. They incorporated measures of cell viability, mitochondrial membrane potential, and contractility into a QST model of the cardiovascular system to predict cardiac dysfunction. Model predictions were substantially improved by correcting for determinants of drug exposure, such as protein binding, tissue-plasma partitioning, and a heterogeneity coefficient. However, the model might be further improved by accounting for temporal dynamics, variability in cellular responses, and broader mechanisms of cardiovascular toxicities.

Use of physiologically based pharmacokinetic (PBPK) models to better characterize and predict drug concentrations at the site of action rather than simple systemic exposures is another major tool to advance understanding of TKI-induced cardiotoxicity. These models incorporate drug-specific properties and physiological parameters to predict tissue exposure. In a recent PBPK model developed for osimertinib, EGFR binding kinetics was incorporated to improve the prediction of drug exposure in target-rich tissues.²¹³ Accounting for target-binding kinetics enabled the prediction of target occupancy in tissues of interest, allowing for the evaluation of how variability in EGFR expression influences drug disposition. Moving forward, the influence of transporter and enzyme-mediated cardiac uptake remains underexplored, and the addition of these processes might provide a more comprehensive framework for studying the pharmacokinetic/toxicodynamic variability in response to genetic polymorphisms and drug-drug interactions. Integrating PBPK and QSP modeling with toxicity and efficacy data can provide a mechanistic framework connecting drug exposure, tissue distribution, toxicity, and efficacy. These efforts demonstrate the growing role of mathematical modeling to investigate TKI cardiotoxicity. Further development of QSP and PBPK models using high-quality temporal and omics data is needed to test the underlying mechanisms associated with drug response, off-target effects, and interindividual variability. Such models will be critical for improving predictions of TKI-induced cardiotoxicity and guide safer individualized treatment strategies.

Lastly, artificial intelligence (AI)-based tools are expected to transform how drug-induced cardiotoxicity is detected and predicted, as well as aid in discovery research to de-risk molecules with potential for cardiac liabilities in humans. Current AI models combine phenotypic (eg, electrophysiological) and physiochemical (chemical structures) data and converts these features into numerical fingerprints or descriptors that can train machine or deep learning models (eg, Random Forest, XGBoost, Graph Neural Networks) to classify cardiotoxic versus noncardiotoxic compounds. Examples of these approaches include AttenHERG and CardioGenAI, which have applied these concepts to predict hERG block liability.^214,215 AI and machine learning can also assist in detecting subtle changes in morphological (eg, sarcomere organization) and contractility (eg, beat patterns, calcium transients, and action potential tracing) in hiPSC-CMs,²⁰⁶ enabling phenotypic screening with high sensitivity. These approaches can similarly be used to predict transporter-drug interactions and identify substrate or inhibition classifications, such as identification of novel inhibitory scaffolds, binding affinity, and regulation of transporters.^216–219 Taken together, the dual integration of these technological advances can enable predictive toxicology that considers regulators responsible for accumulation across cardiac cell membranes to improve drug safety.

5. Conclusions

TKIs have revolutionized the treatment landscape for targeted cancer therapy; however, their clinical success and utility is limited by TKI-induced cardiotoxicity arising from both on-target and off-target effects. Although our current understanding of the etiology of TKI-induced cardiotoxicity has identified essential kinases in cardiac homeostasis (Fig. 1), the extent of toxicity is expected to be shaped by membrane transporters that regulate intracellular TKI accumulation or endogenous molecules in cardiac tissue. Although past studies have uncovered several transporters linked to TKI cardiotoxicity (Fig. 1), the current knowledge is inadequate. Understanding the intersection between kinase signaling and transporter-mediated drug disposition is essential to refining drug safety and informing the next-generation kinase inhibitors with improved cardiac safety.

Fig. 1.

Transporters shown are implicated or potentially involved in contributing to TKI-induced cardiotoxicity through loss of transport function or as regulators of TKI disposition. Transporters in yellow or red represent confirmed roles using in vitro or in vivo studies, while those in blue indicate roles that are not well understood. hERG is also shown. Asterisks (*) denote confirmed events in cardiac tissue. TKIs may also inhibit cardiac tissue kinases if sufficient accumulation occurs. Red dots indicate TKIs interacting with specific transporters or protein kinases. Arrows represent transport direction.

Abbreviations

ABC

ATP-binding cassette

ABCB1

ATP-binding cassette subfamily B member 1 (P-glycoprotein)

ABCG2

ATP-binding cassette subfamily G member 2 (breast cancer resistance protein)

AI

artificial intelligence

AKT

protein kinase B

AMPK

AMP-activated protein kinase

BCRP

breast cancer resistance protein

BTK

Bruton tyrosine kinase

CNT2/3

concentrative nucleoside transporter 2/3

CSK

C-terminal Src kinase

DDI

drug-drug interaction

EGFR

epidermal growth factor receptor (ErbB1)

ERK

extracellular signal-regulated kinase

FDA

Food and Drug Administration

FGFR

fibroblast growth factor receptor

GLUT1/4

glucose transporter type 1/4

HER2

human epidermal growth factor receptor 2 (ErbB2)

hERG

human ether-à-go-go-related gene potassium channel

hiPSC-CM

human induced pluripotent stem cell-derived cardiomyocyte

IGF1R

insulin-like growth factor receptor 1

IR

insulin receptor

MAPK

mitogen-activated protein kinase

MEK

MAPK/ERK kinase

MET

hepatocyte growth factor receptor

PBPK

physiologically based pharmacokinetic

PDGFR

platelet-derived growth factor receptor

PI3K

phosphoinositide 3-kinase

QSP

quantitative systems pharmacology

QST

quantitative systems toxicology

RAF

rapidly accelerated fibrosarcoma kinase

RET

rearranged during transfection proto-oncogene receptor

ROS1

c-ROS oncogene 1 receptor tyrosine kinase

SLC

solute carrier

SLC22A5

solute carrier family 22 member 5 (OCTN2)

SLC29A1

solute carrier family 29 member 1 (ENT1)

SLC47A1

solute carrier family 47 member 1 (MATE1)

SLC47A2

solute carrier family 47 member 2 (MATE2-K)

SLCO2B1

solute carrier organic anion transporter family member 2B1 (OATP2B1)

TEC

TEC family tyrosine kinase

TKIs

tyrosine kinase inhibitors

VEGFR

vascular endothelial growth factor receptor

Data availability

This review article is based on previously published studies; no new datasets were generated or analyzed. All data are located within cited studies.