Cardiovascular Toxicities of Small Molecule

Tyrosine Kinase Inhibitors: An Opportunity for

Systems-Based Approaches
S-A Brown1, L Nhola1 and J Herrmann1

Once universally considered a rapidly fatal condition, cancer has increasingly become a chronic medical condition and
comorbidities and adverse effects of cancer therapies have become increasingly significant. One of the leading
advancements that has gained traction for the treatment of a variety of malignancies is the class of small molecule
tyrosine kinase inhibitors (TKIs). Although, in many aspects revolutionary, TKIs have their own profile of side effects,
including cardiovascular side effects, the most common being hypertension (HTN), congestive heart failure, corrected
QT (QTc) prolongation, and instances of premature coronary heart disease. Herein, we describe the mechanisms of
small TKI-induced cardiovascular toxicity and related intracellular signaling. In particular, systems-based approaches to the
understanding of small TKI-induced cardiovascular toxicity are addressed. The pathophysiology of synergic cardiovascular
toxicity from TKIs, anthracyclines, and monoclonal antibodies (e.g., trastuzumab, bevacizumab) is illustrated. Finally,
recommendations are made for implementing systems medicine in clinical practice, for individualized cardiovascular
wellness after cancer therapy.

Cancer-related mortality rates have been declining since the
1990s, boosted by the development of so-called targeted thera-
pies.1 The intention was that these drugs would be directed
against a molecular target perceived to be of crucial signicance
for a malignant disease process. Prime examples are trastuzumab,
a monoclonal antibody to the human epidermal receptor
(HER)2, and imatinib, a tyrosine kinase inhibitor (TKI) of
BCR-Abl, the oncogenic fusion product of the Philadelphia chro-
mosome. These two drugs are, in fact, the front-runners, the rst
representatives of the two leading subclasses of targeted therapies
monoclonal antibodies and small molecule TKIs. Contrary to
initial expectations, as assumed to be more specic to cancer cells
than nontargeted therapies, some of the targeted agents were
found to be associated with signicant cardiovascular morbidity
and mortality.2 Indeed, even in the current era, cancer survivors
who develop cardiovascular disease have signicantly worse out-
comes compared with cancer survivors without cardiovascular
disease.3
As illustrated for trastuzumab, likely one of the most well-
known targeted therapies associated with cardiotoxicity, elaborate
research efforts revealed that shared biologic pathways account
for the impact of targeted oncologic therapies on both cancer
and the cardiovascular system.4 Such understanding has also
1
Department of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota, USA. Correspondence: J Herrmann (herrmann.joerg@mayo.edu)
Received 30 September 2016; accepted 31 October 2016; advance online publication 2 November 2016. doi:10.1002/cpt.552
helped to guide the development of more selective oncologic
agents, to predict which patients will be at greatest risk of cardio-
vascular toxicity, and to develop protective strategies to minimize
cardiovascular toxicity.2,4 Herein, we will reect on how systems-
based approaches, such as genomics, epigenomics, transcriptomics,
microRNA regulomics, proteomics, metabolomics, microbiomics,
and mathematical and computational modelling can be of further
value in this regard. These technological tools can provide clues
about variations in different cardiovascular system components in
response to perturbation by TKI therapy, and how these variations
can affect the pathophysiological phenotype of the cardiovascular
system.
Small molecule tyrosine kinase inhibitor-induced
cardiovascular toxicity
Tyrosine kinases (TKs; as part of cellular receptors (receptor
tyrosine kinases (RTKs)) or individual molecules (nonreceptor
TKs) regulate a wide array of signaling pathways that control cell
growth, differentiation, migration, apoptosis, and survival. TKs
are enzymes that catalyze the transfer of phosphate from adeno-
sine triphosphate (ATP) to tyrosine residues of cellular proteins.
They thereby not only regulate but even more coordinate the
function of these proteins as TKs are operating in the broader

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017 65

Figure 1 Small tyrosine kinase inhibitor (TKI)-induced cardiotoxicity. Major components of intracellular signaling implicated in TKI-induced cardiotoxicity,
adapted from http://www.nexavar.com/ based on ref. 6 ERK, extracellular signal-regulated kinase; FGFR, fibroblast growth factor receptor; MAPK,
mitogen-activated protein kinase; MEK, methyl ethyl ketone; PDGF, platelet-derived growth factor; PDGFR, platelet-derived growth factor receptor; RAS,
renin angiotensin system; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor.

framework of signaling pathways governed by extracellular stimu-
li or ligands, such as growth factors (e.g., vascular endothelial
growth factor (VEGF) and platelet-derived growth factor
(PDGF)) or intracellular stimuli, such as oxidant stress (Figure 1).5,6
Hyperactivation of TK signaling has been shown to play a role in
the development of cancer and tumor angiogenesis, thus generating
an attractive venue for drug development.5
Importantly, most TKIs compete with ATP for binding to the
ATP-binding domain (or ATP-binding pocket), which is a rela-
tively conserved structure across several kinases, and for this rea-
son lack 100% specicity for their intended molecular target.4,5
This leads to inhibition of intended tumor pathways, which is
benecial in cancer therapy, but also inhibits off-target kinases.5
Although these off-target kinases may or may not have a role in
cancer cell survival and proliferation, they may be crucial to the
survival and function of other cells (e.g., cardiomyocytes), with
important clinical implications (i.e., off-target effects).5
A number of multitargeted TKIs have been approved by the
US Food and Drug Administration, such as sunitinib, sorafenib,
pazopanib, axitinib, vandetanib, cabozantinib, ponatinib, and
regorafenib, whose common target is interference with VEGF sig-
naling (Table 1).7 Although this is of signicance for tumor
angiogenesis and growth, VEGF signaling is of crucial importance
for normal cardiovascular function.8 This pertains primarily to
endothelial cells, in terms of proliferation, survival, and function.
Induction of endothelial cell death and rarefaction of resistance
vessels can be a consequence, as well as impaired vasodilation.9
Regulation of nitric oxide (NO) production is an important ele-
ment, intertwined with VEGF receptor signaling via the
phosphoinositide 3 kinase-AKT and RAS-RAF-mitogen-
activated protein kinase (MAPK)/extracellular signal-regulated
kinase-extracellular signal-regulated kinase pathways, as illustrated
in Figure 1.6,9
Related to these mechanisms, hypertension (HTN) has
emerged as the most common cardiovascular side effect of TKI.9
In addition to HTN, left ventricular dysfunction and heart fail-
ure (HF) have been associated with small-molecule multi-TKI
therapy.2,10 The incidence of left ventricular dysfunction and HF
varies in the literature (233%), which, in part, is due to differ-
ences in patient populations, drug regimens, and a lack of consen-
sus over denitions for cardiovascular toxicity (estimates for
different TKIs are listed in Table 1).2,10
The specic molecular mechanisms underlying the develop-
ment of cardiomyopathy with small molecule TKIs have not
been fully elucidated and varies based on the inhibitory spectrum
of the individuals TKIs. Sunitinib, for instance, inhibits at least
50 kinases, generating a high level of complexity. Prior studies,
however, outlined mitochondrial damage, energy rundown, and
off target impairment of the energy-restoring activity of AMP-
activated protein kinase (AMPK).11 Importantly, AMPK was
inhibited with an IC50 of 216 6 58 nM, comparable to other
molecular targets of sunitinib and at a concentration level that
matches trough blood levels of sunitinib plus its major active
metabolite, which are in the order of 125250 nM.12
Recently, the focus shifted to the primary targets of sunitinib,
namely VEGF and PDGF. Mice with cardiac or endothelial cell-
specic deletion of VEGF develop capillary rarecation, induc-
tion of hypoxia-inducible genes, and cardiomyopathy. Similarly,

66 VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt

 Table 1 Small molecule tyrosine kinase inhibitors (modified from ref. 7)
Stroke/
Malignancy targets Molecular targets HTN Chest pain MI TIA PAD DVT/PE Additional warnings
Bcr/Abl inhibitors
Imatinib Ph1 CML, Ph1 ALL, ABL-1, c-KIT, PDGFR-A, 4% 711% Edema 1186%
GIST, ASM, MDS PDGFR-B, FLT-3 Hypotension 11%
Off-label melanoma
Nilotinib Ph1 CML ABL-1, c-KIT, PDGFR-A, 1011% 59% 13% 34% Pericardial effusion 2%
Off-label GIST PDGFR-B, FLT-3
Ponatinib Ph1 CML, Ph1 ALL ABL-1, c-KIT, PDGFR-A, 5371% 12% 2% 8% 5% HF 615%
PDGFR-B, FLT-1, KDR, Pericardial effusion 13%
FLT-3, FLT-4, FGFR, Atrial fibrillation 4%
SRC, TIE2
Bosutinib Ph1 CML ABL-1, FGFR, FLT-1, Edema 14%
FLT-2, FLT-3, KDR, Pericardial effusion 13%
PDGFR-A, PDGFR-B, SRC
Dasatinib Ph1 CML, Ph1 ALL ABL-1, FGFR2, KIT, 4% IHD Edema 4%
Off-label GIST PDGFR-A, PDGFR-B, SRC Pericardial effusion 4%
Cardiac dysfunction 4%
QTc prolongation 1%
Conduction
abnormalities 7%

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017

VEGFR inhibitors
Sorafenib RCC, thyroid cancer, B-Raf, FLT-1, FLT-3, FLT-4, 941% 3% 1% HF in 2%
HCC, off-label: KDR, KIT, PDGFR-A, (grades 3
angiosarcoma PDGFR-B, FGFR, c-fms or 4 10%)
and GIST
Sunitinib RCC, GIST, PNET, ABL-1, c-KIT, PDGFR-A, 2734% 13% 3% each LVEF drop 1627%
off-label: STS, PDGFR-B, FLT-1, KDR, (grade 3
thyroid cancer FLT-3, FLT-4, FGFR, or 4 10%)
SRC, c-smc
Pazopanib RCC, STS, thyroid ABL-1, c-KIT, PDGFR-A, 40% (grade 510% 2% 1% 15% HF 11%
cancer PDGFR-B, FLT-1, KDR, 3 or 4 7%) QTc prolongation 2%
FLT-4, FGFR, c-fms Bradycardia 219%
Axitinib RCC c-KIT, PDGFR-A, PDGFR-B, 40% (grade 2% 1% 13%
FLT-1, KDR, FLT-4, 3 or 4 16%)
Regorafenib GIST, colorectal PDGFR-B, FLT-1, KDR, 3059% 1%
cancer FLT-4, TIE2, RET, (grade 3 or
c-KIT, RAF 4 828%)
Lenvatinib Thyroid cancer PDGFR-B, FLT-1, KDR, 73% (grade 3% LVEF drop 2%
FLT-4, RET, c-KIT 3 or 4 44%) QTc prolongation 9%
Table 1 Continued on next page

67
68
Table 1 Continued

Stroke/
Malignancy targets Molecular targets HTN Chest pain MI TIA PAD DVT/PE Additional warnings
EGFR inhibitors
Erlotinib NSCLC, pancreatic EGFR (HER1) N/a 12% (2%) (3%) N/a (4%) (Cardiac arrhythmia 5%,
(1gemcitabine) cancer syncope 5%,
thrombosis 11%)
Vandetanib Thyroid cancer EGFR, KDR, 33% (grade 1% QTc prolongation 14%
FLT-4, RET 3 or 49%) Heart failure 2%
ALK inhibitors
Crizotinib NSCLC ALK, MET, 6% QTc prolongation 56%
ROS1 Bradycardia 515%
Syncope 13%
Edema 3149%
Ceritinib NSCLC ALK, IGF-1R, QTc prolongation 4%
ROS1, InsR Bradycardia 3%
JAK inhibitors
Ruxolitinib Myelofibrosis JAK1, JAK2, 6%
Polycythemia TYK2
vera
ALK, anaplastic lymphoma kinase; ALL, acute lymphoblastic leukemia; ASM, aggressive systemic mastocytosis; CML, chronic myeloid leukemia; DVT, deep vein thrombosis; FGFR, fibroblast growth factor receptor; FLT, Fms-
like tyrosine kinase; GIST, gastrointestinal stromal tumor; HCC, hepatocellular carcinoma; HER, human epidermal receptor; HF, heart failure; HTN, hypertension; IHD, ischemic heart disease; LVEF, left venrticular ejection
fraction; MDS, myelodysplastic syndrome; MET, mesenchymal-epithelial transition factor; MI, myocardial infarction; N/A, not applicable; NSCLC, nonsmall cell lung cancer; PAD, peripheral arterial disease; PDGFR, platelet-
derived growth factor receptor; PE, pulmonary embolism; Ph1, Philadelphia chromosome-positive; PNET, pancreatic neuroendocrine tumors; QTc, corrected QT.

VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt

deletion of the PDGF receptor beta leads to decreased capillary
density, hypoxia, and cardiac dysfunction and remodeling upon
pressure overload. This has led to the theory that induction of hyp-
oxia and hypoxia response elements is the underlying mechanism
of cardiomyopathy observed with TKIs that target the VEGF and
PDGF signaling pathway. Accordingly, the presentation would be
more consistent with myocardial hibernation.4 Indeed, ultrastruc-
tural changes, including those of the mitochondria, and cardiac dys-
function are reversible in most cases of sunitinib cardiotoxicity.
Moreover, a recent study pointed out the detrimental role of peri-
cyte depletion via inhibition of PDGF-beta signaling upon micro-
vascular and cardiac function.13 This has generated the concept of
pericyte-endothelial-myocardial coupling as an important element
in the induction of small molecule TKI-induced cardiotoxicity, as
illustrated in the example of sunitinib.14
Systems-based approaches
Cardiovascular toxicity is mainly a pleiotropic or off-target effect
of small molecule TKIs, for which every patient treated with
small molecule TKIs should be at risk. Yet, there is variability in
observed cardiovascular toxicity. This suggests that there may be
unique characteristics of genetics and interaction with lifestyle
and therapeutic environment that result in a small percentage of
individuals experiencing HTN, congestive heart failure, prema-
ture coronary heart disease, prolonged corrected QT (QTc), or
other forms of cardiovascular toxicity. Systems-based approaches
to evaluating cardiovascular toxicity in response to small molecule
TKI therapy could accelerate understanding of pathophysiology
and development of safe and effective drugs. Systems-based
approaches are hallmarks of systems biology, which uses biotech-
nological tools to study whole cells, tissues, or organisms. Systems
biology studies integrate disparate cellular processes and signaling
pathways at different levels of the organism, and in multiple
dimensions. Systems medicine is the application of systems biolo-
gy to elds of medicine, including cardio-oncology.
The mechanisms of cardiovascular toxicity from small mole-
cule TKI therapy are not yet fully understood. Reviewing the
current state of our ndings from systems medicine tools may help
set the stage for further analyses to inspire inferences about the
steps underpinning toxicity. In the following sections, we review
systems-based approaches to studying genomic, transcriptomic,
pathway, and other variations resulting from perturbation of the
cardiovascular system by TKI therapy. Integrating such data from
diverse systems medicine tools could potentially advance our under-
standing of the pathophysiology of TKI-induced cardiotoxicity and
point to high-yield effective therapeutic avenues. Ultimately, incor-
porating these systems-level data into individualized comprehensive
predictive modeling should help personalize the study of these per-
turbations for each patient as a whole system. As HTN is the most
common cardiovascular adverse effect of inhibitors of the VEGF
pathway,15,16 most studies in the following sections focus on HTN
as the main form of toxicity.
Genomics
For simplicity, this section focuses on cardiovascular toxicity
from sunitinib therapy. Several genomic variants underlying
CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017
sunitinib-induced cardiovascular toxicity (e.g., HTN, arterial
thromboembolic events) have been discovered1528 (Figure 2, top
left), and are described in the following sections. Interestingly, the
variants are relevant to almost every major step of sunitinib phar-
macodynamics or pharmacokinetics. Many variants are related to
transcription factors that bind VEGF promoter regions.15
Pharmacokinetics
Absorption/efflux. The ATP-binding cassette transporter protein
ABCG2 mediates absorption and/or excretion of sunitinib
through the wall of the intestine and also efux of sunitinib from
varied cells targeted by the drug.17 The ABCG2 variant 421C>A
(rs2231142) associates with high intracellular concentrations of
sunitinib, with consequent adverse events, such as HTN.18 The
ABCG2 TT haplotype (15622C/T and 1143C/T; rs2622604)
associates with grade >2 sunitinib-induced toxicities, including
HTN.19 The ABCB1 variants rs1128503C>T (variation
G412G) and rs2032582G>T (variation A893S) show a trend
toward protection from sunitinib-induced HTN.16
Metabolism. The CYP3A5 variant rs776746 associates with dose
reductions due to grade 2 toxicities, including HTN.20 This
variant at the CYP3A5*1 allele is thought to increase CYP3A5
expression, thereby increasing conversion of sunitinib to its more
metabolite SU12662.21 The metabolite has a longer half-life than
sunitinib, and is thought to therefore lead to increased exposure
to the effect of sunitinib therapy.21,22
Pharmacodynamics
Receptor substrate. The vascular endothelial growth factor recep-
tor type 2 (VEGFR2) receptor variant 1191T>C (rs2305948)
also associates with grade >2 sunitinib-induced toxicities, including
HTN.19 The VEGFR2 variant rs1870377T>A (variation Q472H)
trends toward association with sunitinib-induced HTN.16
Receptor ligand substrate. The VEGF variant -634G>C
(rs2010963) associates with sunitinib-induced HTN.23 The
VEGF variant rs1570360 associates with increased risk for
sunitinib-induced HTN.20 The VEGF-A variant rs699947A>C
(promoter) trended toward association with sunitinib-induced
HTN.16 In addition, the VEGF variants -1498C>T (rs833061)
and -634G>C (rs2010963) associates with changes in VEGF
expression levels and bevacizumab-induced HTN.29,30 In fact,
the variant -634G>C (rs2010963) is critical for expression of a
VEGF isoform.31
The interleukin-8 variant rs1126647A>T associates with
sunitinib-induced HTN,24 presumably by increasing interleukin-
8 expression levels,25 which may upregulate VEGF levels through
NFkappaB.26,27 The interleukin-13 variant rs1800925C>T also
associates with grade >2 sunitinib-induced toxicities, including
HTN.24
Downstream effector. In response to stimulation of VEGFR2,
NO is produced.32,33 This is mediated by promoting expression
of NO synthases,34 such as endothelial NO synthase, which is
found in endothelial cells and cardiomyocytes, among other cell
69
Figure 2 A systems medicine approach to prediction, prevention, and therapy. Left: Numerous systems biology tools are available for use in cardio-
oncology, to yield and analyze systems medicine data, such as results from genomic, transcriptomic, metabolomics, network medicine, and computational
biology assessments. Right: Patients with cancer will have the capability to undergo testing with a number of systems medicine technologies. Test results
will suggest the patients individual probabilistic low or high risk for developing cardiotoxicity. A therapeutic plan should then be devised to minimize other
contributors to risk, such as concomitant hormonal or radiotherapy. Protective cardiovascular drugs could also be initiated, guided by the patients likeli-
hood of being a responder or nonresponder to this medication. Sunitinib cardiotoxicity genetic variants not shown are: ABCB1, ATP binding cassette sub-
family B member 1 (efflux pump/transporter); ABCG2, ATP-binding cassette sub-family G member 2 (efflux pump/transporter); CYP3A5, cytochrome P450
family 3 subfamily A member 5 (pharmacokinetics/metabolism); IL8/13, interleukin-8/interleukin-13 (inflammation/cytokine); NR1I2, nuclear receptor
subfamily 1, group I, member 2 (nuclear transcription/localization, regulates CYP3A and ABCB1); VEGFA, vascular endothelial growth factor A
(angiogenesis).

types.35 In concert with prostaglandins, endothelial cell-dependent
vasodilation of the arterial and venous systems ensue to control
blood pressure.15,32,33 Consistent with this, the variant endothelial
nitric oxide synthase rs2070744 associates with grade 3 sunitinib-
induced HTN.28
Inherent differences among agents
There are inherent differences between agents,15 as they have dif-
ferent interaction sites on their substrates, which affects potency
and afnity of binding (e.g., to VEGFR).9,15 Therefore, a variant
associated with toxicity from one drug may not associate toxicity
from another. For example, the VEGF variants -1498C>T
(rs833061) associates with bevacizumab-induced HTN, but not
sunitinib-induced HTN.15 Additionally, TKIs can be categorized
into at least four main groups.36 Some TKIs are ATP-competitive
inhibitors. These have an afnity for the ATP-binding site of
RTKs in their active conformation.36 Some TKIs have an afnity
for the ATP-binding site of RTKs in their nonactive conforma-
tion.36 In this case, activation of the RTK becomes energetically
unfavorable. Some TKIs function as allosteric inhibitors and have
an afnity for areas of the RTK that are located outside of the
ATP-binding site.36 These TKIs modify the RTKs tridimension-
al structures and disrupt the ATP-kinase pocket interaction. Oth-
er TKIs function as covalent inhibitors. These have an afnity for
the ATP-binding site of RTKs, but bind irreversibly by covalently
bonding.36 As a result, a generalized class effect cannot be inferred
for the inuence of small molecule TKIs on outcome, efcacy, or
cardiovascular toxicity.37
Variant overlap for toxicity and outcome/efficacy
Authors have noted overlap in variants that associate with toxici-
ty and outcome or efcacy. As one might expect, variants that
associate with more toxicity often also associate with more cyto-
toxic efcacy or better outcome in terms of progression-free sur-
vival or overall survival, and the converse is also true.15,16
Transcriptomics
Gene expression proles, or transcriptomes, have been used to
uncover pathways involved in cardiovascular toxicity. For exam-
ple, results from a transcriptomics study illustrated the role of
AMPK in survival of cardiomyocytes during small molecule TKI

70 VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt

therapy.38 Analysis of gene expression patterns indicated that a
lapatinib analog led to activation of AMPK, with subsequent
effects on p53 and apoptosis. Although lapatinib has overall car-
diotoxic effects, this pathway seemed to be cardioprotective and
may function as a homeostatic mechanism. This is in contrast to
studies showing that sunitinib-induced cardiovascular toxicity is
mediated by direct inhibition of AMPK.12 It is thus important to
recognize that different mechanisms underlie toxicity from differ-
ent TKIs, limiting generalizations about class effects. Another
study reviewed assorted pathways (determined in part from tran-
scriptomic studies) that mediate the cardioprotective effects of
aerobic exercise therapy in the setting of small molecule TKI
therapy.39 Some groups have used gene expression patterns to
establish the utility of distinct cell lines for examining the efcacy
of small molecule TKI therapy, and suggested that these could
also potentially be used to assess mechanisms of cardiovascular
toxicity.40 Other groups have used transcriptomics to delineate
signaling pathways in response to RTKs,41 and have established
cell lines that could potentially also be used to evaluate cardiovas-
cular toxicity from TKI therapy.42 One such cell line is composed
of human-induced pluripotent stem cell-derived cardiomyocytes
derived from broblasts.41 These cell lines hold great promise for
personalizing therapy.43 For instance, a recent study showed that
pluripotent cells from individuals with breast cancer exhibit dif-
ferential gene expression proles if they develop doxorubicin-
induced cardiovascular toxicity.43 Similar studies could be pur-
sued to identify a gene expression prole unique to individuals
who develop TKI-induced cardiovascular toxicity. This could
then be used to guide therapy in precision medicine.
Other systems medicine tools
In addition to genomics and transcriptomics, there are different
systems biology tools that have the potential to someday contrib-
ute to the conversation about TKI-induced cardiotoxicity. Three
of these have been associated with ndings in cardiovascular dis-
ease or cancer, namely microRNA regulomics, metabolomics, and
microbiomics.
MicroRNAs (or miRNAs) are small (1825 nucleotides),
single-stranded, noncoding RNAs that repress translation of
mRNAs by binding to imperfectly complementary-specic
mRNA sequences.44,45 The miRNAs that have been found to be
critical for heart function are involved in vascular remodeling,
coronary artery disease, and heart failure.46 Although several stud-
ies have implicated miRNAs in cancer pathophysiology, no stud-
ies have yet elucidated the role of miRNA regulomics in
cardiovascular toxicity following small molecule TKI therapy.
Nevertheless, expression of miRNA-34 has been shown to pro-
mote apoptosis of pro-angiogenic cells and reduce the expression
of Bcl-2 and other anti-apoptotic proteins in patients with car-
diovascular disease.44 The miRNA expression and function may,
therefore, become an attractive area of study in cardiovascular
toxicity from cancer therapy in the near future.
Metabolic signatures in the blood and strategies to examine
alterations in metabolic states could potentially become useful for
early identication of cardiovascular toxicity,42 and even for pre-
dicting and preventing toxicity. Metabolic proles from blood
CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017
can be indicative of metabolism, stress, and injury, and cardiac
magnetic resonance imaging and 18-uorodeoxyglucose positron
emission tomography can detect energetic alterations, but the
latter has not yet been well developed for cost-effective care of
individuals with cardiovascular toxicity,42 or for those at risk of
toxicity.
When assessing the microbiome, a bacteriomic signature was
discovered for individuals undergoing TKI therapy, relative to
healthy individuals.47 DNA analysis of fecal microbiota further
revealed a unique bacterial signature for individuals at high vs.
low risk of developing diarrhea in response to TKI therapy.47
Potentially, microbiomics could also be used to evaluate and
predict TKI-induced cardiovascular toxicity.
Other systems biology tools, such as epigenomics (or methylo-
mics),48 proteomics,49 interactomics,49,50 network medicine,49 and
mathematical or computational biology49 have also provided useful
information regarding mechanisms of adverse drug effects, develop-
ment of drug targets,50 optimization of drug delivery,51 determi-
nants of drug resistance and sensitivity,52 and pathophysiology of
cancers4852 and cardiovascular disease.46,48 Network biology or
medicine is the consolidation of multiple networks of proteins,
cells, or other units for comprehensive analysis of the relationships
among these entities. Mathematical or computational biology or
medicine is the use of mathematical or computational methods for
studying biology or systems in medicine. Several of these tools have
been combined to enhance investigations.49 These systems medi-
cine tools in diverse combinations are now primed for investigating
small molecule TKI-induced cardiovascular toxicity. Using systems
medicine tools, we may approach an era in which optimal dosing
can be selected to minimize cardiovascular toxicity and maximize
efcacy and outcomes.
SYNERGY OF CHEMOTHERAPEUTICS
Anthracyclines and monoclonal antibodies to receptor
tyrosine kinases
The prime example of synergy in chemotherapeutic cardiotoxicity
is that of anthracyclines and trastuzumab. First in its class, trastu-
zumab became the frontrunner for mAb-based cancer therapy,
especially for patients with cancers that overexpress the HER2.53
In keeping with the initially expressed thought that molecularly
targeted therapy would equate highly cancer-specic therapy;
reports of cardiotoxicity came as a complete surprise. This includ-
ed a 2% incidence of cardiotoxicity with trastuzumab monother-
apy, and a 27% incidence of cardiotoxicity when trastuzumab
was combined with anthracyclines.4,54 Trastuzumab is very effec-
tive in blocking the activation of extracellular signal-regulated
kinase, phosphoinositide 3 kinase/AKT, and JNK-STAT signal-
ing pathways in HER2-positive breast cancer, and it was subse-
quently realized that it also interfered with important signaling
pathways in the heart. A clinical observation, therefore, led to
further investigations into the previously unknown nuances of
HER2 signaling in cardiomyocytes. This signaling pathway is rap-
idly downregulated in the heart after birth, but is re-expressed
under conditions of stress.55 This can include afterload stress,
ischemic stress, or chemical stress (e.g., by anthracyclines). Con-
ceivably, anthracyclines, by their acute accumulation in
71
mitochondria, trigger a stress response in cardiomyocytes that
involves the HER2 signaling pathway and abates over time.
Indeed, this model explains very well the detrimental effects of
concomitant exposure and the substantially lower risk of cardio-
toxicity with trastuzumab with increasing time from anthracy-
cline exposure. Moreover, stimulation of this pathway with
recombinant neuregulin-1, the endogenous ligand of HER2 nor-
mally released by cardiac endothelial cells, can rescue cardiomyo-
cytes in culture from the myobrillar disarray caused by
anthracyclines.55
There has also been recent recognition that increases in sympa-
thetic tone may be a mechanism of HER2-targeted therapy alter-
ing cardiac function. In preclinical models, the b-adrenergic
receptor expression and activity is coupled to HER2 expression56
and HER2 activation is a critical modulator of toxicity to chronic
b-adrenergic receptor agonist exposure.56 Concordantly, patients
with breast cancer who are treated with HER2-targeted therapy
demonstrated increased levels of norepinephrine along with
increased blood pressure and heart rate.56 Accordingly, there may
be provocation of the very mechanisms that activates an HER2-
mediated stress response in patients who undergo treatment with
HER2 inhibitors and increased sensitivity to any concomitant
alternate chemotherapy, especially anthracycline combination
therapy.
Small molecule tyrosine kinase inhibitors, anthracyclines, and
monoclonal antibodies to receptor tyrosine kinases
In a clinical trial, patients with histologically proven advanced
hepatocellular carcinoma who were treated with doxorubicin and
sorafenib combination therapy vs. doxorubicin monotherapy
showed improvement in time to progression, overall survival, and
progression-free survival.57 The trial authors speculated that
improvement may have been due to synergism between doxorubi-
cin and sorafenib.57 Eight patients treated with doxorubicin and
sorafenib combination therapy but not doxorubicin monother-
apy developed HTN. In another study, patients with histological-
ly proven advanced hepatocellular carcinoma who were treated
with doxorubicin and sorafenib combination therapy vs. sorafe-
nib monotherapy showed no improvement in overall survival,
and progression-free survival.58 Although cardiovascular toxicity
was not explicitly delineated, the authors noted grade 3 or 4
hematologic adverse events in 37.8% of individuals receiving
doxorubicin and sorafenib combination therapy, compared with
8.1% in individuals receiving sorafenib monotherapy.58 This sug-
gested synergistic hematologic toxicity in those receiving doxoru-
bicin and sorafenib combination therapy. It will be interesting to
see a more detailed report of this study to determine whether
indeed cardiovascular adverse effects were adequately assessed and
compared between the two groups.
TKIs inhibit multiple receptors, including VEGFR, platelet-
derived growth factor receptor (PDGFR), and the ERBB2/
ERBB4 heterodimer (Figure 3). This antagonizes the MAPK,
phosphoinositide 3 kinase-AKT, and JAK-STAT intracellular
signaling pathways. Inhibition of the MAPK and phosphoinosi-
tide 3 kinase-AKT pathways decreases MDR1 availability on the
cell membrane.59 This limits efux of drugs, such as doxorubicin
72
from the cell. Inhibition of all three pathways decreases the levels
or activity of transcription factors that regulate growth, survival,
and cell cycle progression.59 In addition, inhibition of the MAPK
pathway increases levels of reactive oxygen species (ROS), leading
to apoptosis. Monoclonal antibodies to RTKs, such as bevacizu-
mab, can synergize disruption of these intracellular signaling
pathways. Further, upon entering the cell, doxorubicin binds
topoisomerase alpha or beta.59 The doxorubicin/topoisomerase
complex induces double-stranded DNA breaks, inuences tran-
scription in the nucleus, and increases ROS in the mitochondri-
on, contributing to mitochondrial dysfunction and apoptosis.59
The extent of HTN occurring in response to therapy with
VEGF signaling pathway inhibitors is dose-dependent,15 consis-
tent with synergistic activation of downstream effectors leading
to amplied toxicity. Thus, small molecule TKIs, anthracyclines,
and monoclonal antibodies to RTKs can stimulate or inhibit
overlapping intracellular signaling pathways with potential for
synergy. This would support ndings suggestive of synergy in
clinical trials.
MULTIPLE HIT HYPOTHESIS
The multiple hit hypothesis was originally developed as a concep-
tual model to explain carcinogenesis as the consequence of multi-
ple DNA mutations that, over time, critically alter the expression
and activity of proto-oncogenes and tumor suppressor genes. The
idea was rejuvenated by Jones et al.60 in efforts to provide a
framework for the emerging eld of cardio-oncology. According-
ly, every patient with cancer presents with a given cardiovascular
reserve, which can be reduced by any pre-existing cardiovascular
condition and is challenged by the cancer and its treatments. The
very fact that cardiovascular risk factors and diseases are predic-
tive of cardiovascular toxicity is supportive of this idea, although
it has never been directly proven experimentally.
An indirect illustration, however, is provided by studies on a
particular class of chemotherapeutics: proteasome inhibitors.61
Although coronary atherosclerosis and cardiac dysfunction was
also noted in normal animals undergoing proteasome inhibitor
treatment, the degree of cardiovascular injury was more profound
in animals with concomitant hypercholesterolemia. The potential
explanation was that proteasome activity is upregulated under
stress conditions as a compensatory response for the increase in
proteolytic demand and the consequences of proteasome inhibi-
tion are thus much more severe under these circumstances. Simi-
larly, one may postulate that a reduction of baseline proteasome
function as seen with aging may pose a greater risk. Thus, where-
as proteasome inhibition could induce cardiovascular disease as
one single hit depending on dose and duration even under nor-
mal circumstances, any additional pre-existing or concomitant hit
on the proteasome system is an aggravating factor for potential
cardiovascular toxicity.
TKIs are a typical example of cancer therapeutics that by them-
selves can induce multiple hits. Sunitinib, for instance, inhibits
not only VEGF signaling but also PDGF signaling. These multi-
ple hits have multiple profound consequences on the vasculature.
First of all, PDGF-beta, produced by endothelial cells, stimulates
an essential survival signaling pathway in pericytes, and VEGF,
VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt
Figure 3 Overlapping intracellular signaling pathways in synergistic cardiotoxicity. Mechanisms of cardiotoxicity resulting from interactions of small
tyrosine kinase inhibitor (TKI) signaling pathways with those of other therapeutics, such as anthracyclines and human epidermal receptor 2 (HER2)
receptor antibodies. HER2 receptor antibodies bind to the HER2 (also known as ERBB2, named for a similarity to the avian erythroblastosis viral
oncogene) and HER4 (also known as ERBB4) heterodimer (ERBB2, ERBB4). HER2/ERBB2 5 human epidermal growth factor receptor 2, HER4/ERBB4 5
human epidermal growth factor receptor 4. MAPK, mitogen-activated protein kinase; PDGFR, platelet-derived growth factor receptor; VEGF, vascular
endothelial growth factor; VEGFR, vascular endothelial growth factor receptor.

produced by pericytes, is an essential signaling pathway for
endothelial cell survival and function.14,62 Accordingly, a dual,
anti-PDGF and anti-VEGF, effect is much more detrimental
than either one alone, and profound consequences in terms of
microvascular dysfunction and reduced contractile reserve have
been shown.62 Furthermore, there is not only coupling of
pericytes and endothelial cells but also of endothelial cells with
vascular smooth muscle cells or cardiomyocytes.14 The multiple
hit theory on the endothelial cell level is further illustrated below.
As outlined above, endothelial cells produce neuregulin-1, and
with their impairment, major alterations of the neuregulin-1 sig-
naling pathway can be expected as well. Reduction of NO pro-
duction is equally profound, affecting muscular tone, stiffness,
and pressure.63,64 Accordingly, multitargeted TKIs, like sunitinib,
can have a profound impact on the cardiovascular system.
With regard to cancer therapies, one can distinguish between
those drugs that induce a fairly limited scope of injury and may
require additional factors to generate clinical consequences and
those agents that induce multiple injuries sufcient to become
clinically evident. For the vasculature, the response to injury theo-
ry proposes that atherosclerosis develops as a response to any type
of endothelial injury.65 Indeed, endothelial dysfunction is noted
frequently under diverse clinical conditions and is the rst step.
However, it is still reversible, and no permanent damage is
induced until structural changes occur. These are more likely to
evolve the more severe and the more chronic the exposure. This
may explain why atherosclerosis is not more common in patients
with cancer who are undergoing limited rounds of chemotherapy.
In distinction, patients receiving radiotherapy sustain concurrent
hits, similar to the original theory of cancer related to radiation.66
Cardiovascular risk factors and pre-existing coronary artery dis-
ease are potent determinants of ischemic events after chest radio-
therapy in individuals with breast cancer.67 These observations as
well as the general considerations outlined herein provide impe-
tus for adoption of a mnemonic: awareness and aspirin, blood
pressure, cholesterol and cigarette use, diet and diabetes mellitus,
exercise (ABCDE).68 The mnemonic reects the multifactorial
nature of (and hits in) cardiovascular disease and is geared to lim-
it the extent of injury to minimize any untoward long-term
consequences.
Multiple hits in vascular endothelial cells
In the endothelial cell, TKIs inhibit diverse RTKs, which disrupt
regulation of angiogenesis and metabolism, and adaptation to
stress, as well as growth, survival, and homeostasis (Figure 4).
Inhibition of VEGFR also decreases NO levels, which limits
relaxation of adjacent vascular smooth muscle cells (VSMCs) and
contributes to HTN.63,69 As shown in Figure 4, disruption of
regulation of metabolism and adaptation to stress are linked to
mitochondrial dysfunction.63,64 The cardiovascular system is
critically dependent on normal mitochondrial morphology, distri-
bution, DNA regulation, and function.63 Mitochondrial dysfunc-
tion increases levels of cytotoxic ROS. The ROS levels also
increase in the setting of oxidized low density lipoprotein

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017 73

Figure 4 Proposed mechanisms for the multiple hit hypothesis. Underlying susceptibility to cardiovascular injury, based on an individuals systems
medicine makeup and any relevant personal/family medical history, may predispose to cardiotoxicity, including premature coronary heart disease, subse-
quent to small molecule tyrosine kinase inhibitor (TKI) therapy. ABCDE, awareness and aspirin, blood pressure, cholesterol and cigarette use, diet and
diabetes mellitus, exercise; PDGF, platelet-derived growth factor; PDGFR, platelet-derived growth factor receptor; RTK, receptor tyrosine kinase; VEGF,
vascular endothelial growth factor; VSMC, vascular smooth muscle cell.

cholesterol, triglycerides, hyperglycemia, and free fatty acids.63
ROS disrupts endothelial nitric oxide synthase, with subsequent
reduction in NO levels.63 NO is considered an antihypertensive,
antithrombotic, and anti-atherosclerotic protein.63 Under normal
conditions, mitochondria also regulates intracellular calcium lev-
els by buffering and releasing calcium near calcium inux or
release channels, respectively.63 This mitochondrial regulation of
calcium modulates cell metabolism and survival, and also mito-
chondrial biogenesis.63 Besides TKI therapy and other causes,
doxorubicin therapy also contributes to mitochondrial dysfunc-
tion.59 Mitochondrial dysfunction then results in dysfunction of
VSMCs and adhesion and activation of inammatory cells.64
The outcome of these concurrent and collaborative processes is
atherosclerosis. Patient characteristics, such as age, personal medi-
cal history, and family history, also contribute to atherosclerosis.
Modiable cardiovascular risk factors captured by the mnemonic
ABCDE can also contribute, as host susceptibility and character-
istics help to determine development of atherosclerosis. Other
cardiotoxic therapies (e.g., radiation), can also impact preserva-
tion of growth, survival, or homeostasis.
Feedback/feedforward loops
Figure 5 shows the substrate for several potential feedback and
feedforward loops involved in cardiovascular toxicity resulting
from small molecule TKI therapy, particularly premature coro-
nary heart disease. As in other circumstances of
atherosclerosis,63,64 TKI-induced atherosclerosis is likely mediat-
ed in part by mitochondrial dysfunction and dysfunctional angio-
genesis. Mitochondrial dysfunction leads to increased production
of ROS and decreased levels of calcium and NO and thereby
endothelial dysfunction.63,64 Reduced calcium concentration
worsens mitochondrial dysfunction, and also independently
decreases NO levels further.63,64 Decreased levels of NO impair
vascular relaxation and aggravate HTN, perpetuating the dynam-
ics.70 ROS worsens mitochondrial dysfunction, and also leads to
VSMC dysfunction, adhesion, and activation of inammatory
cells, and macrophage apoptosis, all of which potently pro-athero-
genic in effects.63,64,71
Close inspection of the TKI signaling pathways that lead to
atherosclerosis reveal a myriad of potential feedback and feedfor-
ward loops. Positive feedforward loops propagate the signal to a
particular downstream protein through two separate and comple-
mentary signaling pathways, whereas negative feedforward loops
activate a downstream protein in one arm and block activity in
the same downstream protein in the other arm of the loop. Simi-
larly, positive feedback loops strengthen the signals from
upstream proteins, whereas negative positive feedback loops
antagonize the signals from upstream proteins.
TKI-induced atherosclerosis potentially boasts at least 10 posi-
tive feedforward loops and 2 positive feedback loops. Eight of
these feedforward loops involve HTN. All of these feedback and
feedforward loops are identied in Figure 5. Several loops, which

74 VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt

Figure 5 Potential feedback and feedforward loops inherent in small molecule tyrosine kinase inhibitor (TKI) cardio-oncology. Small molecule TKI therapy
may function as a molecular switch to turn on premature atherosclerosis by inducing several positive feedforward loops (1FFLs). Each 1FFL has two
arms. One arm proceeds from X to Z, with X leading to an impact on Z (upper left inset); the second arm proceeds from X to Y to Z, with X impacting Y,
which also impacts Z. In this way, X influences Z through two separate and complementary pathways. In small molecule TKI cardio-oncology,
mitochondrial dysfunction is a conduit for many of these 1FFLs, along with positive feedback loops (1FBLs). These 1FFLs and 1FBLs subsequent to
mitochondrial dysfunction may also be common to anthracyclines and monoclonal antibodies to receptor tyrosine kinases (RTKs). However, the
anti-angiogenesis pathway is unique to small molecule TKI therapy and may facilitate the molecular switch.

involve intracellular signaling relationships and are well supported
in the literature,5,10,11,63,64,69 are extracted from Figure 5 and
illustrated in isolation in Figure 6 for further clarity. In Figure 5,
arm one of the rst feedforward loop is as follows. TKI therapy
inhibits the PDGFR (among other multiple kinases). Interference
with the PDGFR has been shown to lead to mitochondrial dys-
function.11 Mitochondrial dysfunction results in decreased NO
levels,63,64 which leads to HTN.63,64 In arm two of this rst feed-
forward loop, mitochondrial dysfunction reduces intracellular cal-
cium concentration, which also induces decreased NO levels and
aggravates HTN,63,64 and risk for atherosclerosis. This could be
written more succinctly as: TKI therapy ! mitochondrial dys-
function11,63,64 ! decreased NO levels63,64 ! HTN63,64 ! ath-
erosclerosis, whereas arm two is TKI therapy ! mitochondrial
dysfunction11,63,64 ! reduced calcium concentration63,64 !
decreased NO levels63,64 ! HTN63,64 ! atherosclerosis. Similar-
ly, Figure 5 shows arm one of the second feedforward loop as
TKI therapy ! mitochondrial dysfunction11,63,64 ! decreased
NO levels63,64 ! HTN63,64 ! atherosclerosis, whereas arm two
is TKI therapy ! mitochondrial dysfunction11,63,64 ! endothe-
lial dysfunction (mediated by increased ROS levels, decreased
calcium levels, and decreased NO levels)63,64 ! atherosclerosis.
In Figure 5 and Figure 6a, arm one of the third feedforward loop
is TKI therapy ! mitochondrial dysfunction11,63,64 ! decreased
NO levels63,64 ! HTN63,64 ! atherosclerosis, whereas arm two is
TKI therapy ! decreased angiogenesis5,10 ! HTN63,64 ! athero-
sclerosis. In Figure 5 and Figure 6b, arm one of the fourth feedfor-
ward loop is TKI therapy ! mitochondrial dysfunction11,63,64 !
endothelial dysfunction (mediated by increased ROS levels,
decreased calcium levels, and decreased NO levels)63,64 ! athero-
sclerosis, whereas arm two is TKI therapy ! decreased angiogene-
sis5,10 ! HTN63,64 ! atherosclerosis.
In Figure 5, arm one of the fth feedforward loop is TKI
therapy ! mitochondrial dysfunction11,63,64 ! VSMC
dysfunction (mediated by increased ROS levels)63,64,69 ! athero-
sclerosis, whereas arm two is TKI therapy ! mitochondrial

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017 75

Figure 6 One feedback and five feedforward loops in small molecule tyrosine kinase inhibitor (TKI) cardio-oncology. Several positive feedforward loops
(1FFLs) and one positive feedback loop (1FBL) are isolated here for visual clarification. Each positive feedforward loop has two arms. (a) Arm one of the
third feedforward loop (1FFL3) is TKI therapy ! mitochondrial dysfunction ! decreased nitric oxide (NO) levels ! hypertension (HTN) ! atherosclerosis,
whereas arm two is TKI therapy ! decreased angiogenesis ! HTN ! atherosclerosis. (b) Arm one of the fourth feedforward loop (1FFL4) is TKI therapy !
mitochondrial dysfunction ! endothelial dysfunction (mediated by increased reactive oxygen species (ROS) levels, decreased calcium levels, and
decreased NO levels) ! atherosclerosis, whereas arm two is TKI therapy ! decreased angiogenesis ! HTN ! atherosclerosis. (c) Arm one of the sev-
enth feedforward loop (1FFL7) is TKI therapy ! mitochondrial dysfunction ! vascular smooth muscle cell (VSMC) dysfunction (mediated by increased
ROS levels) ! atherosclerosis, whereas arm two is TKI therapy ! decreased angiogenesis ! HTN ! atherosclerosis. (d) Arm one of the eighth feedfor-
ward loop (1FFL8) is TKI therapy ! mitochondrial dysfunction ! adhesion/activation of inflammatory cells (mediated by increased ROS levels) ! athero-
sclerosis, whereas arm two is TKI therapy ! mitochondrial dysfunction ! decreased NO levels ! HTN ! atherosclerosis. (e) Arm one of the 10th
feedforward loop is (1FFL10) TKI therapy ! mitochondrial dysfunction ! adhesion/activation of inflammatory cells (mediated by increased ROS levels) !
atherosclerosis, whereas arm two is TKI therapy ! decreased angiogenesis ! HTN ! atherosclerosis. (f) The first feedback loop (1FBL1) is composed of
mitochondrial dysfunction and increased ROS levels.
dysfunction11,63,64 ! decreased NO levels63,64 ! HTN63,64 !
atherosclerosis. In Figure 5, arm one of the sixth feedforward
loop is TKI therapy ! mitochondrial dysfunction11,63,64 !
VSMC dysfunction (mediated by increased ROS levels)63,64,69 !
atherosclerosis, whereas arm two is TKI therapy ! mitochondrial
dysfunction11,63,64 ! endothelial dysfunction (mediated by
increased ROS levels, decreased calcium levels, and decreased NO
levels)63,64 ! atherosclerosis. In Figure 5 and Figure 6c, arm one
of the seventh feedforward loop is TKI therapy ! mitochondrial
dysfunction11,63,64 ! VSMC dysfunction (mediated by increased
ROS levels)63,64,69 ! atherosclerosis, whereas arm two is TKI ther-
apy ! decreased angiogenesis5,10 ! HTN63,64 ! atherosclerosis.
Figure 5 and Figure 6d show arm one of the eighth feedforward
loop as TKI therapy ! mitochondrial dysfunction11,63,64 ! adhe-
sion/activation of inammatory cells (mediated by increased ROS
levels)63,64 ! atherosclerosis, whereas arm two is TKI therapy !
mitochondrial dysfunction11,63,64 ! decreased NO levels63,64 !
HTN63,64 ! atherosclerosis. In Figure 5, arm one of the ninth
feedforward loop is TKI therapy ! mitochondrial dysfunc-
tion11,63,64 ! adhesion/activation of inammatory cells (mediated
by increased ROS levels)63,64 ! atherosclerosis, whereas arm two is
TKI therapy ! mitochondrial dysfunction11,63,64 ! endothelial
dysfunction (mediated by increased ROS levels, decreased calcium
levels, and decreased NO levels)63,64 ! atherosclerosis. In Figure 5
and Figure 6e, arm one of the tenth feedforward loop is TKI thera-
py ! mitochondrial dysfunction11,63,64 ! adhesion/activation of
inammatory cells (mediated by increased ROS levels)63,64 ! ath-
erosclerosis, whereas arm two is TKI therapy ! decreased angio-
genesis5,10 ! HTN63,64 ! atherosclerosis.
In Figure 5 and Figure 6f, the rst feedback loop is composed
of mitochondrial dysfunction and increased ROS levels.63,64 In
Figure 5, the second feedback loop is composed of mitochondrial
dysfunction and decreased calcium levels.63,64 Both loops would
worsen mitochondrial dysfunction and ultimately atherosclero-
sis.63,64 Both feedback loops are likely common to small molecule
TKIs, monoclonal RTK antibodies, and anthracyclines, as all
three can target pathways that induce mitochondrial dysfunction
and resultant downstream effects.4,11,43,53 For the same reason, of
the feedforward loops, six should be common to small molecule
TKIs, monoclonal RTK antibodies, and anthracyclines. Four
would be unique to small molecule TKI-induced atherosclerosis
the loops that include decreased angiogenesis. This has the
potential to produce a molecular switch, in which TKI therapy
switches on atherosclerosis (in some patients), and in the absence
of the four unique feedforward loops involving decreased angio-
genesis, the switch is off. This systems biology phenomenon may
help to explain why small molecule TKIs, relative to monoclonal
RTK antibodies and anthracyclines, may show more of an associ-
ation with atherosclerosis and cardiovascular disease. This pro-
vides a great example of how systems biology and likely network
biology can aid in the interpretation of clinical or phenotypic
observations.
Influence of small molecule tyrosine kinase inhibitors
on efflux pumps
Small TKIs also interact with doxorubicin therapy via ATP-
binding cassette transporters or efux pumps, such as MDR1
CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017
(Figure 3). Studies have shown that TKIs can inhibit these
pumps.36,72,73 In fact, there is potentially a bell-shaped curve dis-
tribution of the ability of several TKIs at different concentrations
to activate or inhibit these efux pumps.72 Thus, inhibition or
activation of efux pumps by small molecule TKIs can contribute
to homeostasis within cardiac or vascular cells, while at the same
time having the capacity to worsen toxicity. For example, inhibi-
tion of doxorubicin efux pumps by small molecule TKIs can
increase intracellular doxorubicin levels, which can worsen double-
strand breaks and ROS and subsequently apoptosis, as well as ath-
erosclerosis. In this way, small molecule TKIs inhibiting efux
pumps adds another arm to the feedforward loops that instigate
the molecular switch. In fact, a number of small molecule TKIs
(e.g., cediranib, lapatinib, and sunitinib) have been used to inhibit
ABC transporters and efux pumps and thereby reverse drug
resistance.36 Although this improves outcomes and efcacy, this
methodology could also worsen toxicity. A ne balance would
need to be achieved between efcacy and toxicitythe same
desired balance that drives the eld of cardio-oncology.
IMPLEMENTATION OF SYSTEMS MEDICINE IN
CLINICAL PRACTICE
Integrating various systems-based approaches into clinical prac-
tice will likely provide the most meaningful outcomes for preci-
sion medicine in cardio-oncology (Figure 2). High-powered
computers should be used to analyze the plethora of data, or as
commonly referred to big data, resulting from distinct systems-
based approaches. Computational and mathematical modeling
should be leveraged to assess combinatorial interactions among
various components of the cardiovascular system in response to
perturbation by TKIs. Such interactions often lead to complex
phenotypes that are not additives, but rather can be synergistic or
antagonistic. Predictive computational modeling could fuse data
from the available systems medicine tools to determine actionable
feedback for implementation in precision medicine. The output
should be combed through to help determine which individual
patients maybe at high vs. low risk for TKI-induced cardiovascu-
lar toxicity. At the same time, data should be evaluated to also
determine which individuals might be considered responders or
nonresponders to cardioprotective therapy.
The predictive P*4 pathway would be useful to streamline sys-
tems medicine in clinical practice.59 The pathway includes sys-
tems medicine counseling, akin to genetic counseling, prior to
systems medicine testing. A composite systems medicine score
would categorize individuals as high-risk, low-risk, or potentially
intermediate-risk for developing TKI-induced cardiovascular tox-
icity. Adjustments in cancer therapy and initiation of cardiopro-
tective therapy would be guided by the composite systems
medicine score. Based on the progress of studies on cardiovascular
toxicity from small molecule TKI therapy, the most specic
approach with robust data is currently genomics (Figure 2, top
left). Clinical trials should be designed to determine whether
genomic testing prior to small molecule TKI therapy could
decrease the incidence of cardiotoxicity. This would be facilitated
by initiation of cardioprotective drugs for those deemed to be at
high risk (Figure 2, right), as well as minimizing other risk
77
factors or concomitant cardiotoxic therapy, in order to avoid
multiple hits (Figure 4). As the eld of cardio-oncology pro-
gresses and the state of the art advances, iterative clinical trials
could be pursued to rene a composite systems medicine score, as
comprehensive data from additional systems-based approaches
become available.
CONCLUSION
Opportunities for systems-based approaches to precision medi-
cine have been described, using the example of small molecule
TKIs in cardio-oncology. Primarily, ndings were presented from
genomic studies that have revealed variants associated with
sunitinib-induced cardiovascular toxicity. These variants affect
angiogenesis and TKI pharmacokinetics, metabolism, regulation,
transport, and efux. The potential for useful inferences from
other systems biology tools was also described. Such tools includ-
ed transcriptomics, epigenomics, miRNA regulomics, proteomics,
metabolomics, microbiomics, interactomics, and network biology
or medicine. Mathematical and computational biology or medi-
cine could also be used to model interactions among various car-
diovascular system components implicated in TKI-induced
toxicity. Robust predictive models spanning multiple scales
(e.g., genomics, transcriptomics, and proteomics) would augment
clinical information available to physicians and scientists. The
functional and phenotypic corollaries of variations in omics
induced by TKI therapy could be integrated in predictive compu-
tational models to inuence patient care and outcomes. Further,
synergic toxicity likely occur when TKIs are used with anthracy-
clines or monoclonal antibodies to RTKs, with consolidation of
a myriad of feedback and feedforward loops and a molecular
switch. Our understanding of small molecule TKI intracellular
signaling and how this overlaps with other therapeutic pathways
will be continually shaped by results from systems biology tools
in the future and knowledge of interaction with other therapeu-
tics, such as anthracyclines and HER2 receptor antibodies. The
potentially bell-shaped curve for activation and inhibition of
MDR1 by numerous TKIs at dissimilar concentrations suggests
that several TKIs will impact ATP-binding cassette efux pumps
differently. Although some individuals with no risk factors
will experience premature coronary heart disease, others develop
cardiovascular toxicity in the setting of underlying risk factors.
Contribution of multiple factors can lead to mitochondrial
dysfunction and ROS in the vasculature, resulting in athero-
sclerosis. Thus, cardiovascular toxicity occurs from multiple
hits at multiple levels, involving multiple cells, pathways, and
injuries. A similar systems biology approach will also be
extremely valuable to dene the mechanism of accelerated ath-
erosclerosis with the BCR-Abl inhibitors nilotinib and ponati-
nib, which has remained, in large part, unexplained so far.7 To
minimize the impact of small molecule TKI therapy on the
cardiovascular system, the ABCDE steps are an easy approach
to remember for heart and vascular wellness of individuals
with cancer. Cardio-oncologic care can be further individual-
ized by pursuing novel targets for drug development from
molecules implicated in vascular endothelial metabolism and
signaling pathways elucidated in systems medicine studies.
78
Such drugs will alleviate injury to the heart, while facilitating
cancer remission or cure.
ACKNOWLEDGMENTS
The Joerg Herrmann was supported by NIH grant HL 116952.
C 2016 American Society for Clinical Pharmacology and Therapeutics
V
1. Hurwitz, H. et al. Bevacizumab plus irinotecan, fluorouracil, and
leucovorin for metastatic colorectal cancer. N. Engl. J. Med. 350,
23352342 (2004).
2. Lenneman, C.G. & Sawyer, D.B. Cardio-oncology: an update on
cardiotoxicity of cancer-related treatment. Circ. Res. 118, 1008
1020 (2016).
3. Armenian, S.H. et al. Cardiovascular disease among survivors of
adult-onset cancer: a community-based retrospective cohort study. J.
Clin. Oncol. 34, 11221130 (2016).
4. Bellinger, A.M. et al. Cardio-oncology: how new targeted cancer
therapies and precision medicine can inform cardiovascular
discovery. Ciculation 132, 22482258 (2015).
5. Krause, D.S. & Van Etten, R.A. Tyrosine kinases as targets for cancer
therapy. N. Engl. J. Med. 353, 172187 (2005).
6. Wilhelm, S. et al. Discovery and development of sorafenib: a
multikinase inhibitor for treating cancer. Nat. Rev. Drug Discov. 5,
835844 (2006).
7. Herrmann, J. Tyrosine kinase inhibitors and vascular toxicity: impetus
for a classification system? Curr. Oncol. Rep. 18, 33 (2016).
8. Carmeliet, P. Angiogenesis in life, disease and medicine. Nature
438, 932936 (2005).
9. Bhargava, P. VEGF kinase inhibitors: how do they cause hypertension?
Am. J. Physiol. Regul. Integr. Comp. Physiol. 297, R1R5 (2009).
10. Ghatalia, P. et al. Congestive heart failure with vascular endothelial
growth factor receptor tyrosine kinase inhibitors. Crit. Rev. Oncol.
Hematol. 94, 228237 (2015).
11. Mercurio, V. et al. Models of heart failure based on the cardiotoxicity
of anticancer drugs. J. Card. Fail. 22, 449458 (2016).
12. Kerkela, R. et al. Sunitinib-induced cardiotoxicity is mediated by off-
target inhibition of AMP-activated protein kinase. Clin. Transl. Sci. 2,
1525 (2009).
13. Chintalgattu, V. et al. Cardiomyocyte PDGFR-beta signaling is an
essential component of the mouse cardiac response to load-induced
stress. J. Clin. Invest. 120, 472484 (2010).
14. Herrmann, J. et al. Vascular toxicities of cancer therapies: the old and
the new--an evolving avenue. Circulation 133, 12721289 (2016).
15. Vaziri, S.A., Kim, J., Ganapathi, M.K. & Ganapathi, R. Vascular
endothelial growth factor polymorphisms: role in response and toxicity
of tyrosine kinase inhibitors. Curr. Oncol. Rep. 12, 102108 (2010).
16. Garcia-Donas, J. et al. Single nucleotide polymorphism associations
with response and toxic effects in patients with advanced renal-cell
carcinoma treated with first-line sunitinib: a multicentre, observational,
prospective study. Lancet Oncol. 12, 11431150 (2011).
17. Yuasa, T., Takahashi, S., Hatake, K., Yonese, J. & Fukui, I.
Biomarkers to predict response to sunitinib therapy and prognosis in
metastatic renal cell cancer. Cancer Sci. 102, 19491957 (2011).
18. Mizuno, T. et al. ABCG2 421C>A polymorphism and high exposure of
sunitinib in a patient with renal cell carcinoma. Ann. Oncol. 21,
13821383 (2010).
19. van Erp, N.P. et al. Pharmacogenetic pathway analysis for determination
of sunitinib-induced toxicity. J. Clin. Oncol. 27, 44064412 (2009).
20. Junker, K., Ficarra, V., Kwon, E.D., Leibovich, B.C., Thompson, R.H. &
Oosterwijk, E. Potential role of genetic markers in the management of
kidney cancer. Eur. Urol. 63, 333340 (2013).
21. van der Veldt, A.A. et al. Genetic polymorphisms associated with a
prolonged progression-free survival in patients with metastatic renal cell
cancer treated with sunitinib. Clin. Cancer Res. 17, 620629 (2011).
22. Diekstra, M.H. et al. Association analysis of genetic polymorphisms
in genes related to sunitinib pharmacokinetics, specifically clearance
of sunitinib and SU12662. Clin. Pharmacol. Ther. 96, 8189 (2014).
23. Kim, J.J. et al. Association of VEGF and VEGFR2 single nucleotide
polymorphisms with hypertension and clinical outcome in metastatic
VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt
 clear cell renal cell carcinoma patients treated with sunitinib. Cancer
118, 19461954 (2012).
24. Diekstra, M.H. et al. Association of single nucleotide polymorphisms
in IL8 and IL13 with sunitinib-induced toxicity in patients with
metastatic renal cell carcinoma. Eur. J. Clin. Pharmacol. 71,
14771484 (2015).
25. Hacking, D. et al. Increased in vivo transcription of an IL-8 haplotype
associated with respiratory syncytial virus disease-susceptibility.
Genes Immun. 5, 274282 (2004).
26. Petreaca, M.L., Yao, M., Liu, Y., Defea, K. & Martins-Green, M.
Transactivation of vascular endothelial growth factor receptor-2 by
interleukin-8 (IL-8/CXCL8) is required for IL-8/CXCL8-induced
endothelial permeability. Mol. Biol. Cell. 18, 50145023 (2007).
27. Martin, D., Galisteo, R. & Gutkind, J.S. CXCL8/IL8 stimulates
vascular endothelial growth factor (VEGF) expression and the
autocrine activation of VEGFR2 in endothelial cells by activating
NFkappaB through the CBM (Carma3/Bcl10/Malt1) complex. J. Biol.
Chem. 284, 60386042 (2009).
28. Eechoute, K. et al. Polymorphisms in endothelial nitric oxide synthase
(eNOS) and vascular endothelial growth factor (VEGF) predict
sunitinib-induced hypertension. Clin. Pharmacol. Ther. 92, 503510
(2012).
29. Schneider, B.P. et al. Association of vascular endothelial growth
factor and vascular endothelial growth factor receptor-2 genetic
polymorphisms with outcome in a trial of paclitaxel compared with
paclitaxel plus bevacizumab in advanced breast cancer: ECOG 2100.
J. Clin. Oncol. 26, 46724678 (2008).
30. Pander, J., Gelderblom, H. & Guchelaar, H.J. Pharmacogenetics of
EGFR and VEGF inhibition. Drug Discov. Today 12, 10541060
(2007).
31. Bastide, A. et al. An upstream open reading frame within an IRES
controls expression of a specific VEGF-A isoform. Nucleic Acids Res.
36, 24342445 (2008).
32. Hood, J.D., Meininger, C.J., Ziche, M. & Granger, H.J. VEGF
upregulates ecNOS message, protein, and NO production in human
endothelial cells. Am. J. Physiol. 274(3 Pt 2), H1054H1058 (1998).
33. Horowitz, J.R. et al. Vascular endothelial growth factor/vascular
permeability factor produces nitric oxide-dependent hypotension.
Evidence for a maintenance role in quiescent adult endothelium.
Arterioscler. Thromb. Vasc. Biol. 17, 27932800 (1997).
34. Facemire, C.S., Nixon, A.B., Griffiths, R., Hurwitz, H. & Coffman, T.M.
Vascular endothelial growth factor receptor 2 controls blood pressure
by regulating nitric oxide synthase expression. Hypertension 54,
652658 (2009).
35. Villanueva, C. & Giulivi, C. Subcellular and cellular locations of nitric
oxide synthase isoforms as determinants of health and disease. Free
Radic. Biol. Med. 49, 307316 (2010).
36. Sierra, J.R., Cepero, V. & Giordano, S. Molecular mechanisms of
acquired resistance to tyrosine kinase targeted therapy. Mol. Cancer
9, 75 (2010).
37. Diekstra, M.H., Swen, J.J., Gelderblom, H. & Guchelaar, H.J. A
decade of pharmacogenomics research on tyrosine kinase inhibitors
in metastatic renal cell cancer: a systematic review. Expert Rev. Mol.
Diagn. 16, 605618 (2016).
38. Shell, S.A., Lyass, L., Trusk, P.B., Pry, K.J., Wappel, R.L. & Bacus,
S.S. Activation of AMPK is necessary for killing cancer cells and
sparing cardiac cells. Cell Cycle 7, 17691775 (2008).
39. Scott, J.M., Lakoski, S., Mackey, J.R., Douglas, P.S., Haykowsky,
M.J. & Jones, L.W. The potential role of aerobic exercise to modulate
cardiotoxicity of molecularly targeted cancer therapeutics. Oncologist
18, 221231 (2013).
40. Mirza, Z. et al. Molecular interaction of a kinase inhibitor midostaurin
with anticancer drug targets, S100A8 and EGFR: transcriptional
profiling and molecular docking study for kidney cancer therapeutics.
PLoS One 10, e0119765 (2015).
41. Eldridge, S., Guo, L., Mussio, J., Furniss, M., Hamre, J. & Davis, M.
Examining the protective role of ErbB2 modulation in human-induced
pluripotent stem cell-derived cardiomyocytes. Toxicol. Sci. 141, 547
559 (2014).
42. Force, T. & Kolaja, K.L. Cardiotoxicity of kinase inhibitors: the
prediction and translation of preclinical models to clinical outcomes.
Nat. Rev. Drug Discov. 10, 111126 (2011).
43. Burridge, P.W. et al. Human induced pluripotent stem cell-derived
cardiomyocytes recapitulate the predilection of breast cancer
CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 101 NUMBER 1 | JANUARY 2017
patients to doxorubicin-induced cardiotoxicity. Nat. Med. 22, 547
556 (2016).
44. Xu, Q. et al. Micro-RNA-34a contributes to the impaired function of
bone marrow-derived mononuclear cells from patients with
cardiovascular disease. J. Am. Coll. Cardiol. 59, 21072117 (2012).
45. Tang, Y., Poteh, N.A., Chiang, A. & Kim, I. MicroRNA therapeutics for
cardiac disease. Cardiovasc. Pharmacol. 2, 1000e113 (2013).
46. De Rosa, S., Curcio, A. & Indolfi, C. Emerging role of microRNAs in
cardiovascular diseases. Circ. J. 78, 567575 (2014).
47. Pal, S.K. et al. Stool bacteriomic profiling in patients with metastatic
renal cell carcinoma receiving vascular endothelial growth factor-
tyrosine kinase inhibitors. Clin. Cancer Res. 21, 52865293 (2015).
48. Peng, X., Pentassuglia, L. & Sawyer, D.B. Emerging anticancer
therapeutic targets and the cardiovascular system: is there cause for
concern? Circ. Res. 106, 10221034 (2010).
49. Burkard, T.R., Rix, U., Breitwieser, F.P., Superti-Furga, G. & Colinge, J.
A computational approach to analyze the mechanism of action of the
kinase inhibitor bafetinib. PLoS Comput. Biol. 6, e1001001 (2010).
50. Buchwald, F., Richter, L. & Kramer, S. Predicting a small molecule-
kinase interaction map: a machine learning approach. J. Cheminform.
3, 22 (2011).
51. Kim, M., Gillies, R.J. & Rejniak, K.A. Current advances in
mathematical modeling of anti-cancer drug penetration into tumor
tissues. Front. Oncol. 3, 278 (2013).
52. Lin, F., Li, Z., Hua, Y. & Lim, Y.P. Proteomic profiling predicts drug
response to novel targeted anticancer therapeutics. Expert Rev.
Proteomics 13, 411420 (2016).
53. Hudis, C.A. Trastuzumab mechanism of action and use in clinical
practice. N. Engl. J. Med. 357, 3951 (2007).
54. Onitilo, A.A., Engel, J.M. & Stankowski, R.V. Cardiovascular toxicity
associated with adjuvant trastuzumab therapy: prevalence, patient
characteristics, and risk factors. Ther. Adv. Drug Saf. 5, 154166
(2014).
55. Pentassuglia, L. & Sawyer, D.B. The role of neuregulin-1b/ErbB
signaling in the heart. Exp. Cell Res. 315, 627637 (2009).
56. Sysa-Shah, P. et al. Bidirectional cross-regulation between ErbB2 and
b-adrenergic signalling pathways. Cardiovasc. Res. 109, 358373
(2016).
57. Abou-Alfa, G.K. et al. Doxorubicin plus sorafenib vs doxorubicin alone
in patients with advanced hepatocellular carcinoma: a randomized
trial. JAMA 304, 21542160 (2010).
58. Abou-Alfa, G. et al. Phase III randomized study of sorafenib plus
doxorubicin versus sorafenib in patients with advanced
hepatocellular carcinoma (HCC): CALGB 80802 (Alliance). J. Clin.
Oncol. 34(suppl. 4S), abstract 192 (2016).
59. Brown, S.A., Sandhu, N. & Herrmann, J. Systems biology approaches
to adverse drug effects: the example of cardio-oncology. Nat. Rev.
Clin. Oncol. 12, 718731 (2015).
60. Jones, L.W., Haykowsky, M.J., Swartz, J.J., Douglas, P.S. & Mackey,
J.R. Early breast cancer therapy and cardiovascular injury. J. Am. Coll.
Cardiol. 50, 14351441 (2007).
61. Herrmann, J. et al. Chronic proteasome inhibition contributes to
coronary atherosclerosis. Circ. Res. 101, 865874 (2007).
62. Chintalgattu, V. et al. Coronary microvascular pericytes are the
cellular target of sunitinib malate-induced cardiotoxicity. Sci. Transl.
Med. 5, 187ra69 (2013).
63. Tang, X., Luo, Y.X., Chen, H.Z. & Liu, D.P. Mitochondria, endothelial
cell function, and vascular diseases. Front. Physiol. 5, 175 (2014).
64. Hulsmans, M., Van Dooren, E. & Holvoet, P. Mitochondrial reactive
oxygen species and risk of atherosclerosis. Curr. Atheroscler. Rep.
14, 264276 (2012).
65. Ross, R., Faggiotto, A. Bowen-Pope, D. & Raines, E. The role of
endothelial injury and platelet and macrophage interactions in
atherosclerosis. Circulation 70(5 Pt 2), 1117711182 (1984).
66. Weintraub, N.L., Jones, W.K. & Manka, D. Understanding
radiation-induced vascular disease. J. Am. Coll. Cardiol. 55,
12371239 (2010).
67. Darby, S.C. et al. Risk of ischemic heart disease in women after
radiotherapy for breast cancer. N. Engl. J. Med. 368, 987998 (2013).
68. Bhatia, N. et al. Cardiovascular effects of androgen deprivation
therapy for the treatment of prostate cancer: ABCDE steps to reduce
cardiovascular disease in patients with prostate cancer. Circulation
133, 537541 (2016).
69. Ku, D.D., Zaleski, J.K., Liu, S. & Brock, T.A. Vascular
endothelial growth factor induces EDRF-dependent relaxation in
79
 coronary arteries. Am. J. Physiol. 265(2 Pt 2), H586H592
(1993).
70. Liu, Y., Li, H., Bubolz, A.H., Zhang, D.X. & Gutterman, D.D.
Endothelial cytoskeletal elements are critical for flow-mediated
dilation in human coronary arterioles. Med. Biol. Eng. Comput. 46,
469478 (2008).
71. Madamanchi, N.R. & Runge, M.S. Mitochondrial dysfunction in
atherosclerosis. Circ. Res. 100, 460473 (2007).
80
radi, A., Sarkadi, B. & Ke
72. Hegedus, T., Orfi, L., Seprodi, A., Va ri, G.
Interaction of tyrosine kinase inhibitors with the human multidrug
transporter proteins, MDR1 and MRP1. Biochim. Biophys. Acta
1587, 318325 (2002).
73. Wang, Y.J., Zhang, Y.K., Kathawala, R.J. & Chen, Z.S. Repositioning
of tyrosine kinase inhibitors as antagonists of ATP-binding cassette
transporters in anticancer drug resistance. Cancers (Basel) 6,
19251952 (2014).
VOLUME 101 NUMBER 1 | JANUARY 2017 | www.wileyonlinelibrary/cpt