Introduction

Chronic myelogenous leukemia (CML) is typically associated cytogenetically with the presence of the Philadelphia chromosome [1]. This cytogenetic abnormality results from the fusion of the breakpoint cluster region (BCR) gene on chromosome 22 to the c-Abl gene on chromosome 9 [2]. The resulting oncogenic product, BCR–Abl, is thought to be responsible for the increases in cellular survival and proliferation observed in the chronic phase of CML. The multi-targeted kinase inhibitor, imatinib, has inhibitory activity toward BCR–Abl and is one of the first targeted therapies approved for oncology [3]. Since the discovery of imatinib, several multi-targeted kinase inhibitors have been approved for the treatment of various tumor types and kinase inhibitors continue to be of interest in pharmaceutical development [4–6].

Despite the enormous success of targeted therapies in treating cancer, recent preclinical and clinical observations have suggested potential untoward effects of these kinase inhibitors on the cardiovascular system [7–9]. In particular, imatinib has come under scrutiny with reports of preclinical and clinical cardiac dysfunction manifest as reduced ejection fraction and development of congestive heart failure (CHF) [10]. The incidence and severity of imatinib-induced cardiac dysfunction has been debated with various reports of the incidence ranging from 0.1 to 9.0 % [10–15] and the risk seemingly increasing in older patients, although the absolute risk may not be different from the standard population [12]. The current imatinib label lists CHF as a rare event occurring in around 0.7 % of patients similar to interferon treatment [10].

The mechanism of imatinib-induced cardiovascular dysfunction is also a subject of debate and several key hypotheses of potential mechanisms of action (MOA) have been suggested including endoplasmic reticulum (ER) stress [10, 16–18], perturbations of Akt and MAP kinase pathways [19], lysosomal [17] and mitochondrial toxicity as well as links to inhibition of Abl kinase [6]. Notably, the potential link of Abl kinase inhibition and CV dysfunction is a high concern as it suggests a potential class effect which would potentially extend to all drugs targeting Abl kinase.

Bosutinib (Bosulif®, SKI-606, PF-05208763) is a small molecule multi-targeted kinase inhibitor designed to potently inhibit BCR–Abl and sarcoma (Src) kinase for the treatment of CML. Bosutinib is of a distinct chemical structural class and, unlike the other marketed Abl inhibitors, lacks inhibitory activity toward c-Kit and PDGFR kinases [20–22]. The aim of the current set of experiments was to examine any potential effects of the novel c-Abl inhibitor, bosutinib, on cardiovascular structure and function at clinically relevant concentrations, and compare these to the effects of imatinib. A secondary aim of the experiments was to examine the potential role c-Abl inhibition plays in the cardiovascular effects of imatinib observed in rats and cardiomyocytes in vitro.

Materials and methods

In vivo studies

Three separate in vivo studies were conducted in male and/or female Sprague–Dawley rats starting at 7 weeks of age supplied by Charles River Laboratories, Inc. (Portage, MI, USA). In all studies, standard procedure and conditions were applied in a facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. All procedures involving laboratory animals were reviewed and approved by Pfizer Inc. or appropriate institution’s Institutional Animal Care and Use Committee associated with the testing facility.

In a 2-month study, rats (n = 10) were randomly assigned to 3 treatments groups: vehicle (0.05 % methyl cellulose, 2 % polysorbate 80, 0.06 % acetic acid), imatinib at 50 mg/kg/day (imatinib mesylate tablet lot # F0028) and bosutinib at 50 mg/kg/day (bosutinib free base, lot # 09-002). Satellite pharmacokinetic groups (n = 3) were also established for vehicle, imatinib and bosutinib. Groups were dosed orally once daily for 8 consecutive weeks with biweekly body weight collection for dose volume (10 mL/kg) adjustment. Blood was collected for pharmacokinetic analysis on days 1, 28 and 56 of the study. Echocardiography was performed once during the pre-dosing period and every 2 weeks during the course of the study. Heart and brain weights were collected during scheduled necropsy performed on day 56.

In a 6-month study, male and female rats (n = 15) were randomly assigned to 3 treatments groups: vehicle (0.05 % methyl cellulose, 2 % polysorbate 80, 0.06 % acetic acid), imatinib at 50 mg/kg/day (imatinib mesylate, LC Labs, Boston, MA) and bosutinib at 50 mg/kg/day (bosutinib free base, lot # 09-002). Satellite pharmacokinetic groups (n = 4) were also established for vehicle, imatinib and bosutinib. Groups were dosed orally once daily for 6 consecutive months with weekly body weight collection for dose volume (10 mL/kg) adjustment. To maintain target plasma exposures, imatinib male rats were dosed adjusted to 25 mg/kg on day 101 by reducing dosing volume to 5 mL/kg. Blood was collected for pharmacokinetic analysis on days 30, 90, 135 (imatinib males only) and day 180 of the study. Echocardiography was performed once during the pre-dosing period and every 8 weeks during the course of the study. Heart and brain weights were collected during scheduled necropsy performed on day 180.

In a 2-year carcinogenicity study, male and female rats were treated with vehicle or bosutinib for up to 23 months at a high dose of 25/15 (week 78 dose reduction) and 15 mg/kg, respectively. During necropsy hearts were collected and processed for histological analysis.

Pharmacokinetic analysis

Plasma EDTA-K3 samples frozen at −80 °C were analyzed for the presence of imatinib or bosutinib. Briefly, 25 μL of plasma was extracted using 100 μL of a 75/25 mix of acetonitrile and methanol in a 1.2-mL 96-well plate. All tubes were vortexed for 10 min followed by centrifugation at 3000 rpm for 10 min. 80 μL of each supernatant was transferred to a clean 96-well plate containing 20 μL of 10 % ethylene glycol and then analyzed by LC–MS/MS. Samples and standards were run in duplicate, while quality control samples were run in triplicate.

Echocardiography

Rats were anesthetized with 1.5 % isoflurane (Baxter, Inc., Deerfield, IL) and underwent fur removal from the thorax by shaving followed by treatment with a depilatory agent and then placed on their left side to collect echocardiographic images in the parasternal long axis, parasternal short axis and apical four-chamber view. Echocardiograms were performed as previously described using the Philips SONOS 5500 system equipped with a 15-MHz linear-array transducer (Philips, Andover, MA) and included the collection of multiple 2D, m-mode and Doppler images [23]. All measurements were performed according to the recommendations of the American Society for Echocardiography leading-edge method from three consecutive cardiac cycles. Echocardiographic images, analysis and data were peer reviewed.

Measurement of cardiac fetal gene expression using qRT-PCR

From the 2-month study, heart samples from the pharmacokinetic animals in each treatment group were collected and flash frozen in liquid nitrogen on day 56. RNA was extracted using the RNeasy kit per manufacturer’s instructions (Qiagen, Valencia, CA). RNA quality was assessed using Bioanalyzer (Agilent, Sunnyvale, CA) and spectrophotometer. Total RNA isolated from in vivo heart samples and in vitro neonatal cardiomyocytes was reverse-transcribed and amplified using qScript one-step qRT-PCR kit (Quanta Biosciences, Gaithersburg, MD). The following primers and probes were custom designed and ordered from Integrated DNA Technology (San Diego, CA): natriuretic peptide precursor A (NPPA) (forward: ACC TCT CAG TGG CAA TGC, reverse: GGT AGG ATT GAC AGG ATT GGA, probe: TCG AGC AGA TTT GGC T), b-type natriuretic peptide (NPPB) (forward: CTT TTC CTT AAT CTG TCG CCG, reverse: GTC TCT GAG CCA TTT CCT CTG, probe: TCC TAG CCA GTC TCC AGA ACA ATC CA), myosin heavy chain 7 (MYH7) (forward: CGC CTG TCA GCT TGT AAA TG, reverse: ACA ACC CCT ACG ATT ATG CG, probe: CCT TTG ATG TGC TGG G), and c-Abl (forward: GGA TCA ACG GCA GCT TCT TAG T, reverse: CTT GCC ATC AGA GGC AGT GTT, probe: CCA GAG GTC CAT CTC GCT GCG GTA T). The quantification of the target mRNA was performed on the LightCycler® 480 Real-Time PCR System (Roche Applied Science, Indianapolis, IN), and data were analyzed using LightCycler 480 SW1.5 software.

Rat neonatal ventricular cardiomyocyte isolation and c-Abl gene knockdown

Cardiomyocytes were isolated from post-natal days 3–5 Wistar-Han rats (Charles River, Wilmington, MA), according to the method described in detail previously [17]. The cardiomyocytes were counted on a hemocytometer and plated in collagen-coated 6-well plates (BD Biosciences) at 8.8 × 105 cells per well. After 24 h, the media were changed to serum-free DMEM with 25 mM glucose and cells were cultured at 37 °C, 5 % CO2 and 95 % relative humidity.

Two independent shRNA oligos specific to the rat c-Abl1 target gene (Gene ID 311860) were designed and cloned in the pLKO plasmid (sigma, St. Louis, MO). In addition, a separate shRNA oligo sequence that does not target any known human or rodent gene to a high degree of homology was also cloned and served as a non-specific negative control. The shRNA sequences have been described previously [17]. Neonatal cardiomyocytes were transduced with virus in a 1:1 ratio with culture media in the presence of 8 μg/mL polybrene (Sigma, St. Louis, MO). The media were changed to regular growth media 8–10 h post-transfection. The c-Abl gene knockdown was verified at 48 h post-virus infection using qRT-PCR procedure described above.

Imatinib analogs

Imatinib structural analogs were designed to have minimal c-Abl inhibition activity, but to maintain similar physicochemical properties as imatinib. The structures, synthetic procedures, kinase inhibition profiles and physicochemical properties have been described previously [17].

Statistical analysis

Body weight and organ weight data were analyzed using one-way analysis of variance (ANOVA) followed by Dunnett’s test to compare all groups to vehicle.

Echocardiography data were analyzed to highlight treatment-related changes of ultrasound parameters from baseline measurements. Mixed effect modeling with appropriate contrasts were carried out to determine any time point-specific treatment. For each time point that demonstrated a significant effect of treatment, follow-up pair-wise comparisons against the vehicle were performed. The alpha = 0.05 level of significance was used.

The gene expression data were analyzed using LightCycler 480 SW 1.5 software. The data were normalized to a housekeeping gene (β-actin) and statistical significance was calculated using ANOVA.

Results

Organ weights and heart histopathology

A summary of the effects of each compound on body weight gain and heart and brain weights is presented in supplementary Table 1. In males treated for 2 and 6 months, statistical analysis revealed significantly increased absolute and relative heart weight for imatinib-treated rats (22 %) when compared with vehicle-treated animals, whereas bosutinib-treated rats did not differ from vehicle-treated animals (Figs. 1a, 2a, b). In females treated for 6 months at exposures roughly fivefold clinically efficacious AUC, statistical analysis revealed significantly increased absolute and relative heart weight for imatinib- (24 %) and bosutinib-treated females when compared with vehicle-treated females, albeit bosutinib-treated females had a smaller magnitude of change (13 %) (Fig. 2b). The heart weights were not collected as part of the 2-year carcinogenicity study. However, full histopathology examination revealed no significant effects of bosutinib exposure at 1.5-fold (males) or threefold (females) on heart histopathology. The incidence and severity of progressive murine cardiomyopathy, which is a well-documented background finding in Sprague–Dawley rats were similar among all treatment groups and controls (Bosutinib Carcinogenicity Study, supplemental Table 2) [24]. Examples of grade 2 cardiomyopathy are illustrated in Fig. 3a–c, comparing representative samples from water control, vehicle control and 15 mg/kg treated female bosutinib rats.

Fig. 1
figure 1

Individual and mean heart weight (HW) to brain weight (BrW) ratio (normalized HW) after 2 months (8 weeks) of treatment (a), indicator of hypertrophy. Time course averages of ejection fraction (EF) taken at pre-dose, 2, 4, 6 and 8 weeks during treatment (b), measure of systolic function. Time course averages of estimated left ventricular (LV) mass taken at pre-dose, 2, 4, 6 and 8 weeks during treatment (c), indicator of hypertrophy. Bars represent standard error of the mean. Differences from vehicle are denoted by an asterisk at p < 0.05. Measurement of cardiac fetal gene activation in the ventricular samples of rats treated with imatinib or bosutinib for 8 weeks (d). The normalized intensity values of cardiac fetal genes NPPA, NPPB and MYH7 measured using quantitative RT-PCR. Error bars represent standard error of the mean from three animals per group. Significant changes from the vehicle are denoted by an asterisk at p < 0.05 or double asterisk at p < 0.01

Full size image
Fig. 2
figure 2

Individual and mean heart weight (HW) to brain weight (BrW) ratio (normalized HW) after 6 months of treatment in males (a) and females (b), indicator of hypertrophy. Time course averages of ejection fraction (EF) taken at pre-dose, 2, 4 and 6 months during treatment (c males, d females), measure of systolic function. Individual and mean ejection fraction (EF) after 6 months of treatment in males (e) and females (f), indicator of systolic function demonstrating individual animal response to treatment. Bars represent standard error of the mean. Differences from vehicle are denoted by an asterisk at p < 0.05

Full size image
Fig. 3
figure 3

H & E-stained left ventricle (×10 magnification) demonstrating grade 2 background cardiomyopathy in water (a), vehicle (b) or bosutinib (c 15 mg/kg) -treated female rats at approximately 25 months of age. Similar presence of infiltrate and myocardial necrosis was observed for all treatments

Full size image

Pharmacokinetics

Pharmacokinetic (PK) data for imatinib and bosutinib are summarized in Table 1 and supplemental Table 3. PK data were within the expected concentration range and comparable to internal and published studies with both imatinib and bosutinib in the rat (18, 25, 26, Pfizer internal data). Due to similar protein binding in rat and human blood for both imatinib and bosutinib, only total drug concentrations are reported. Reported imatinib human exposures (AUC) were taken from literature sources [11, 18] and used to calculate the relative fold over clinical exposure achieved in the rats, which was similar for both compounds. The definitive clinical AUC for bosutinib was 3650 ng/mL taken from steady-state exposure data in a phase 3 trial for first-line treatment of CML at a dose of 500 mg QD [26]. Based on a CML cellular tumor model, half maximum inhibitory concentrations (IC50) of 309 nM (imatinib) and 12.5 nM (bosutinib) exposures achieved in the current study likely provided near maximal inhibition of C-Abl kinase over the entire time course of the individual study duration for both compounds.

Table 1 Pharmacokinetic parameters of imatinib and bosutinib including therapeutic index
Full size table

Echocardiography and cardiac fetal gene expression

2-Month study results

The key echocardiography results for the 2-month study are illustrated in Fig. 2, while the complete group averages and statistical summary of parameters are presented in supplemental Table 4. None of the treatments resulted in changes in ejection fraction (Fig. 1b). Significant increases in the following parameters were observed with imatinib treatment: interventricular septal (IVS) diastolic and systolic wall thickness, left ventricular (LV) posterior systolic and diastolic wall thickness, LV diastolic and systolic diameter, LV end diastolic (EDV) and end systolic volume (ESV), stroke volume, LV endocardial and epicardial area and LV mass (Fig. 1c). Two months treatment with imatinib significantly induced the expression of cardiac NPPA, NPPB and MYH7, members of the fetal gene program in myocardium which are known to be reactivated during cardiac hypertrophy. Bosutinib treatment for 2 months did not result in significant changes in the expression levels of NPPA, NPPB or MYH7 in rat heart (Fig. 1d), consistent with echocardiography findings. Overall, echocardiographic analysis indicated that imatinib treatment caused hypertrophy, whereas bosutinib had no discernable echocardiography effect over the 2-month study.

6-Month study: male results

Group averages and statistical summary of key echocardiography parameters for the 6-month study are presented in supplemental Table 5, separated by sex. Treatment of male rats with imatinib resulted in significant increases in the same parameters as in the 2-month study. The prevalent phenotype in male rats treated with imatinib in the 6-month study remained hypertrophy coupled with diminished function in individual animals (Fig. 2c, e). Two individual male rats in the imatinib treatment groups demonstrated signs of transition from hypertrophy to decompensation. The transition into decompensation was supported by increased EDV (LV dilation), decrease in EF greater than 15 % compared to baseline and/or vehicle, decreased relative wall thickness and increased E/A ratio occurring from the 3- to the 6-month time point. No biologically or toxicologically relevant changes were identified in bosutinib-treated male rats.

6-Month study: female results

Treatment of female rats with imatinib resulted in significant increases in the following parameters first observed at the 2-month time point and persisting until the end of the study: IVS diastolic and systolic wall thickness, LV posterior systolic and diastolic wall thickness, LV diastolic diameter, LV diastolic and systolic volume, stroke volume, LV endocardial and epicardial area, LV mass and mitral valve A velocity (supplemental Table 5). Ejection fraction was not significantly different in any treatment group (Figs. 2d, 3f) Treatment with bosutinib resulted in intermittent statistical increases in LV diastolic volume, LV endo and epicardial areas and LV mass starting at 2 months and persisting until 6 months. A comparison of the percent change from baseline for vehicle, imatinib- and bosutinib-treated female rats following 6 months of treatment is shown in supplemental Table 6. Both imatinib- and to a lesser extent bosutinib-treated female rats expressed a hypertrophic phenotype.

In vitro c-Abl knockdown and imatinib analog treatment

To assess the role c-Abl plays in cardiac hypertrophy, targeted gene silencing of c-Abl in neonatal cardiomyocytes (NCM) was performed using lentiviral-mediated shRNA transfection. Compared to the non-treated control, the c-Abl shRNA treatment in NCM resulted in significant decrease in c-Abl gene transcript level (82 and 75 % decrease by c-Abl oligo #1 and #2, respectively) as assessed using real-time RT-PCR (Fig. 4a). To compare the effect of c-Abl gene knockdown with compound treatment, cardiomyocytes were also treated with imatinib at 10 and 20 µM and bosutinib at 2.8 µM for 24 h. The concentrations selected were approximately fivefold over total plasma concentrations at therapeutic doses for both compounds and therefore comparable between imatinib and bosutinib. Phenylephrine has been known to induce cardiac hypertrophy both in vitro and in vivo and was used as a positive control in this experiment [27]. Phenylephrine at 50 μM induced NPPA and MYH7 gene expression as expected. Imatinib treatment in NCM resulted in concentration-dependent increase in both NPPA and MYH7 gene expression level, but no change was observed with bosutinib treatment at 2 μM (Fig. 4a). NCM transfected with c-Abl shRNA showed c-Abl suppression, but did not induce cardiac hypertrophy genes (Fig. 4a). This result supports the hypothesis that c-Abl kinase inhibition is not involved in the induction of cardiac hypertrophy and the imatinib-induced hypertrophic phenotype is likely through an off-target mechanism.

Fig. 4
figure 4

Targeted c-Abl knockdown in rat neonatal cardiomyocytes and measurement of cardiac fetal gene expression in c-Abl deficient cells as well as compound-treated cells. a Rat neonatal cardiomyocytes (NCMs) were transfected with lentivirus containing two shRNA oligos specific for rat c-Abl 1 target gene or non-specific shRNA sequence. Non-transfected NCMs were treated with imatinib at 10 and 20 μM, bosutinib at 2 μM or phenylephrine at 50 μM. The expression of c-Abl and cardiac fetal genes, including NPPA and MYH7, was measured using quantitative real-time RT-PCR at 48 h post-transfection and 24 h post-compound treatments. b Neonatal cardiomyocytes were treated with imatinib and the two structural analogs without appreciable c-Abl inhibition for 16 h. The normalized intensity values of cardiac fetal genes NPPA, NPPB and MYH7 were measured using quantitative RT-PCR. Error bars represent standard error of the mean from triplicate samples. Significant changes from the vehicle are denoted by an asterisk at p < 0.05 or double asterisks at p < 0.01

Full size image

To further support this hypothesis, NCMs were treated for 16 h with the two c-Abl inactive imatinib structural analogs, and the expression of cardiac hypertrophy genes was evaluated using RT-PCR (Fig. 4b). As reported previously, the imatinib analogs do not show appreciable c-Abl inhibition, but maintain similar physicochemical properties as imatinib [17]. Analog 1, with a c-Abl IC50 of 5.9 μM, resulted in induction of cardiac hypertrophy gene NPPA, NPPB and MYH7 to the same magnitude as imatinib. Analog 2, which has a c-Abl IC50 higher than 30 μM and is more lipophilic, induced cardiac hypertrophy gene expression to a greater extent (Fig. 4b). These findings further support the notion that c-Abl does not play a direct role in imatinib-induced cardiac hypertrophy.

Discussion

The current studies were designed to examine the effects of longer-term treatment of imatinib and bosutinib on cardiovascular structure and function in rats and to investigate the potential role of c-Abl inhibition in any cardiac-related findings. Treatment with imatinib and bosutinib at 50 mg/kg resulted in exposure margins ranging from 0.6 to 5.0 times the clinical AUC for both compounds dependent upon the clinical dose comparator and sex of the animal. In male rats at exposures approximately onefold the clinical AUC, there were no observed effects of bosutinib on cardiac structure or function as measured by echocardiography or organ weights following 2 or 6 months of treatment. In contrast, imatinib treatment in male rats near onefold clinical AUC for 2 or 6 months had significant effects on multiple echocardiographic and organ weight parameters. Echocardiographic analysis suggests that rats treated with imatinib start developing hypertrophy of the heart after approximately 1 month of treatment, which progresses throughout the duration of treatment up to 6 months. Further corroboration of imatinib-induced hypertrophy is found in increases in absolute heart weight and heart weight to brain weight ratio taken during necropsy. Gene expression analysis following 2 months treatment demonstrated increased expression of hypertrophic markers with imatinib, but not bosutinib treatment. Interestingly, there does not appear to be a loss of cardiovascular function in imatinib-treated rats as EF is maintained during the entirety of the 2-month treatment. However, in the 6-month study a decrease in EF of greater than 10 % is observed in two individual animals suggesting that longer treatment may result in loss of function. Overall, there are clear hypertrophic effects of imatinib on the heart, while bosutinib treatment appears devoid of any cardiovascular effects in male rats. At exposures near fivefold the clinical AUC, female rats treated with imatinib and to a lesser extent bosutinib displayed hypertrophic phenotypes corroborated by organ weight and echocardiography data. The overall magnitude of the hypertrophic effect in females was greater for imatinib compared to bosutinib at similar therapeutic indices (four to fivefold). Finally, the histological examination of male and female rat hearts following up to 23 months exposure to bosutinib demonstrated no significant morphological or pathological findings associated with treatment at up to threefold clinical exposures. Imatinib-induced hypertrophy in the rat has been reported previously [11, 18, 25] and may over time result in a decompensated heart leading to LV dilation and eventually overt heart failure and a reduction in longevity. While not conducted as any part of the current studies, imatinib-related cardiomyopathy has been reported previously to be observed during a similar 2-year carcinogenicity study [11].

From the current data set, it appears that most of the treated rats were able to compensate for imatinib-induced morphological and functional effects on the heart and maintain normal cardiovascular function during the duration of the study. Due to the multifaceted nature and complexity of heart failure, the etiology may take months to years to fully develop. Therefore, it is reasonable to assume that duration of treatment would be an important factor in the switch from functional compensated heart to a dysfunctional compromised phenotype [28]. The current study identified two potential responders with decreased ejection fraction and cardiac output, in addition to the observed hypertrophy, following 6 months of treatment of imatinib at therapeutic exposures. Although the clinical incidence of imatinib-induced heart failure is debatable, there is evidence to suggest that imatinib treatment can lead to heart failure in a select patient population, particularly the elderly [16]. Perhaps, increased age or underlying cardiovascular disease in combination with prolonged use of imatinib is a sufficient set of circumstances for the cardiovascular dysfunction to manifest in at-risk patients. The current and published studies would suggest that there may be early signals of potential cardiac dysfunction, and that these signals can be detectable in patients by standard diagnostic tools such as echocardiography or multi-gated acquisition scan (MUGA). To date, there have been no clinical or pre-clinical evidence that the heart is a target organ for bosutinib. Multiple long-duration toxicological studies examining bosutinib treatment in rats and dogs have been completed including 2-year carcinogenicity rat studies in which male (1.5-fold) and female (threefold) exposures exceeded clinical efficacious concentrations and no histological evidence of cardiovascular abnormality was found (Fig. 1). Additionally, MUGA and or echocardiographic analysis of bosutinib-treated patients have not demonstrated a robust or reliable clinical signal of cardiovascular effects including decreased ejection fraction (Pfizer unpublished data).

The mechanism of imatinib-induced cardiovascular dysfunction is not currently fully understood and several hypotheses have been suggested including both on and off-target kinase inhibition, as well as the role of physicochemical properties of the compound [10, 16–19]. The first paper to note the preclinical and clinical effects on the heart suggested that c-Abl inhibition may be responsible for ER stress and mitochondrial toxicity leading to the CV phenotype [10]. The in vitro preclinical findings were not corroborated by Wolf and colleagues [18] who concluded that the cardiovascular effects of imatinib were not related to c-Abl inhibition and occurred at concentrations above the therapeutic rage. Recent investigations confirmed that imatinib could lead to ER stress and proposed that the mechanism again was not c-Abl inhibition, rather it was related to the physicochemical properties of imatinib and accumulation in lysosomes leading to disruption of autophagy [17]. Similar to the current study, in vivo studies conducted by Wolf [18] also report imatinib-induced hypertrophy in the rat; however, these studies were conducted at higher doses and for shorter duration than the current study, resulting in higher plasma exposure. Other studies examining potential mechanisms did not find imatinib to be toxic to rat cardiomyocyte mitochondria at concentrations achieved in the present study [29]. Collectively, the data suggest that other factors or mechanisms other than c-Abl inhibition play a key role in imatinib-induced hypertrophy.

A secondary aim of the current experiments was to further investigate the potential role of c-Abl inhibition as a key factor in imatinib-induced cardiac hypertrophy or toxicity. To this end, in vitro experiments utilizing two shRNA sequences targeting c-Abl, imatinib, bosutinib and appropriate controls were used to explore the role of c-Abl in the observed cardiac hypertrophy. The hypertrophic gene markers NPPA and MYH7 were concentration-dependently upregulated in cardiomyocytes treated with imatinib or the positive control phenylephrine. In contrast, treatment with bosutinib or suppression of c-Abl mRNA up to 80 % did not result in the induction of hypertrophic genes. The in vitro data support findings of the 2-month study in which imatinib- but not bosutinib-treated hearts had increased expression of the hypertrophic genes NPPA, NPPB and MYH7. Further evidence that inhibition of c-Abl is likely not involved in the development of the hypertrophic cardiovascular phenotype is demonstrated by the effects of imatinib analogs 1 and 2, which retain similar physicochemical properties of imatinib while having much lower affinity for c-Abl. The analog compounds demonstrated similar induction of the hypertrophic genes suggesting that c-Abl inhibition was not necessary to induce hypertrophy. Since bosutinib and imatinib possess similar physicochemical properties as both are basic and lipophilic, the distinction between the two compounds may reflect differences in therapeutic index; bosutinib is roughly 10× more potent toward the BCR–Abl kinase and requires approximately 10× lower free plasma levels for the treatment of CML than imatinib [18, 20]. Thus, the threshold for lysosome accumulation and autophagy dysfunction in the heart leading to hypertrophy is not exceeded with bosutinib treatment in the male rat at approximately onefold clinically therapeutic plasma concentrations. However, as bosutinib exposure increases to approximately fivefold as observed in the female rats, perhaps the threshold of accumulation into the lysosomes is exceeded and the initial pathophysiology begins as evidenced by the mild hypertrophy observed. Collectively, the current in vitro and in vivo data suggest that c-Abl inhibition is not the cause of the cardiovascular phenotype observed in the rat.

Although the current studies suggest a preclinical differentiation of imatinib and bosutinib and differing therapeutic indices for cardiovascular effects in the rat, which is corroborated by distinct clinical safety profiles, several limitations of the current study must be considered. First, with cardiovascular events a seemingly low adverse event for imatinib and bosutinib, the role of species differences must be considered. The differentiation of the compounds was assessed in the rat which, although is a typical species for cardiovascular and toxicological research, may not fully represent the human condition, phenotype and clinical outcome [30]. The translation from normal rat to diseased human is difficult to fully delineate; however recent publications and internal data suggest that the rat is an acceptable cardiovascular model for human disease and toxicology. Rat models should be considered on a case by case basis [30–32]. Additional cardiovascular functional data in a second species or taken directly from clinical trials would strengthen the idea of distinct cardiovascular profiles for imatinib and bosutinib. Moreover, even though small animal echocardiography has become more standardized in research, there may be limitations in the sensitivity of the techniques used in the current study. Alternative study paradigms such as stress echocardiography or newer technologies such as strain imaging may have increased sensitivity and thus may identify additional potential effects of imatinib and bosutinib [33].

The data presented here suggests a preclinical cardiovascular differentiation of the c-Abl inhibitors imatinib and bosutinib at clinically relevant concentrations of each of the respective drug. Two to six months of imatinib treatment at approximately clinically therapeutic exposures induced a hypertrophic phenotype in normal male Sprague–Dawley rats, whereas bosutinib treatment did not. Additionally, early signs of functional deficits were observed in imatinib-, but not bosutinib-treated animals. Furthermore, in the rat bosutinib displays a higher threshold for hypertrophic effects compared to imatinib requiring approximately four to fivefold the clinical exposure to observe the effect. Moreover, the current in vivo and in vitro data sets suggest that c-Abl inhibition is not involved in the hypertrophic response to imatinib and that potency for primary kinase targets coupled with physicochemical properties may play a key role in the pathophysiology of imatinib-induced cardiac hypertrophy.