Breast Cancer Promotes Cardiac Dysfunction Through Deregulation of Cardiomyocyte Ca2+-Handling Protein Expression That is Not Reversed by Exercise Training.

Background Patients treated for breast cancer have a high incidence of cardiovascular complications. In this study, we evaluated the impact of breast cancer on cardiac function and cardiomyocyte Ca2+-handling protein expression. We also investigated whether exercise training (ET) would prevent these potential alterations. Methods and Results Transgenic mice with spontaneous breast cancer (mouse mammary tumor virus-polyomavirus middle T antigen [MMTV-PyMT+], n=15) and littermate mice with no cancer (MMTV-PyMT-, n=14) were studied. For the ET analysis, MMTV-PyMT+ were divided into sedentary (n=10) and exercise-trained (n=12) groups. Cardiac function was evaluated by echocardiography with speckle-tracking imaging. Exercise tolerance test was conducted on a treadmill. Both studies were performed when the tumor became palpable and when it reached 1 cm3. After euthanasia, Ca2+-handling protein expression (Western blot) was evaluated. Exercise capacity was reduced in MMTV-PyMT+ compared with MMTV-PyMT- (Pinteraction=0.031). Longitudinal strain (Pgroup <0.001) and strain rate (Pgroup=0.030) were impaired. Cardiomyocyte phospholamban was increased (P=0.011), whereas phospho-phospholamban and sodium/calcium exchanger were decreased (P=0.038 and P=0.017, respectively) in MMTV-PyMT+. No significant difference in sarcoplasmic or endoplasmic reticulum calcium 2 ATPase (SERCA2a) was found. SERCA2a/phospholamban ratio was reduced (P=0.007). ET was not associated with increased exercise capacity. ET decreased left ventricular end-systolic diameter (Pgroup=0.038) and end-diastolic volume (Pgroup=0.026). Other morphological and functional cardiac parameters were not improved by ET in MMTV-PyMT+. ET did not improve cardiomyocyte Ca2+-handling protein expression. Conclusions Breast cancer is associated with decreased exercise capacity and subclinical left ventricular dysfunction in MMTV-PyMT+, which is at least partly associated with dysregulation of cardiomyocyte Ca2+ handling. ET did not prevent or reverse these changes.

ejection time

extracellular signal‐regulated kinase

exercise training

exercise tolerance test

human epidermal growth factor receptor type 2

isovolumetric contraction time

isovolumetric relaxation time

interventricular septum

c‐Jun N‐terminal kinase

left ventricular end‐diastolic diameter

left ventricular end‐systolic diameter

left ventricular end‐systolic volume

posterior wall thickness

mitogen‐activated protein kinase

mouse mammary tumor virus

phosphoinositide 3‐kinase

phospho‐phospholamban

polyomavirus middle T antigen

sarcoplasmic or endoplasmic reticulum calcium 2 ATPase

Clinical Perspective

What Is New?

Breast cancer is associated with subclinical left ventricular dysfunction in transgenic mice with spontaneous breast cancer at the end stage of this disease.

This cardiac dysfunction is associated with dysregulation of cardiomyocyte Ca2+ handling.

Exercise training alone is not an effective strategy to prevent or reverse the cardiac function provoked by breast cancer.

What Are the Clinical Implications?

These findings highlight that cancer is a systemic disease that gradually compromises body homeostasis as the disease progresses.

Moreover, it is suggested that patients with cancer are more susceptible to organ dysfunction, including cardiac dysfunction.

Breast cancer is the most diagnosed cancer among women, and it was responsible for more than 2 million new cancer cases worldwide in 2018. Multimodality treatment, which includes surgical resection, radiation, and systemic therapy, is the current standard of care for this disease.

Systemic agents commonly used to treat breast cancer, such as anthracyclines, are known to be associated with cardiovascular events, a phenomenon known as cardiovascular toxicity. Endothelial dysfunction, acute coronary syndrome, supraventricular and ventricular arrhythmias, and ventricular dysfunction, leading to heart failure, have all been described in patients during or after chemotherapy treatment. However, little is known about the impact of breast cancer itself on cardiac function, which can be important to clarify the impact of cancer as a systemic disease that can negatively affect the cardiovascular physiology.

Exercise training (ET) promotes many benefits to the cardiovascular system both in healthy and chronically ill individuals. Previous studies from our group demonstrated that ET improves cardiac function in hypercaloric obese rats and in a mouse model of heart failure induced by sympathetic hyperactivity. , In addition, ET substantially increases physical capacity , and is considered a marker of ET adaptation.

In this study, we hypothesized that breast cancer would promote cardiac dysfunction and decrease exercise tolerance. We also hypothesized that ET would attenuate or prevent cancer‐induced cardiac dysfunction. To understand the effects of breast cancer on cardiac function, we used a mouse model of breast cancer. We used this same model to explore the effects of ET on cardiac function, and to test whether the changes in cardiac function were associated with deregulation of Ca2+ handling in the mouse cardiomyocyte.

METHODS

The data, analytical methods, and study materials that support the conclusions of this study will be available to other researchers for the purpose of replicating the procedures and reproducing the results through reasonable request by contacting the corresponding author.

Animal Model

Female mice carrying the polyomavirus middle T antigen (mouse mammary tumor virus [MMTV]‐PyMT+), a transgenic mouse model of spontaneous breast cancer with C57BL6/J genetic background, from University of São Paulo Medical School, were used in the study, and their respective healthy offspring (MMTV‐PyMT−) served as controls. In these mice, the development of spontaneous breast cancer in 4 distinctly identifiable stages of tumor progression occurs in a single primary tumor focus and it resembles the pathology of human breast cancer in relation to hyperplasia, adenoma, and early and late carcinoma. , , In addition, the loss of estrogen and progesterone receptors and the overexpression of human epidermal growth factor receptor type 2 (HER2) are similar to HER2‐amplified human breast cancer.

Mice were kept in the animal facilities of the University of São Paulo Medical School, placed in a standard mouse polypropylene box (30×20×13 cm) with a meshed lid and lined with sterile wood shavings. The animals were housed 4 per cage and kept at controlled temperatures of 22 to 24°C, with relative humidity of 45% to 50%, and reversed light‐dark cycle in 12‐hour day and night periods. Meals included standard mice chow and water ad libitum. All animal procedures were approved by the Scientific Committee for Animal Research from the University of São Paulo Medical School (N°061/16). The experiment was conducted in concordance with the National Council for Animal Experimentation Control.

Experimental Protocol

In a pilot study to test the hypothesis regarding cardiac function analysis, the sample size (n=4 per group) was reconsidered because of the variability in the results. In addition, some histological and molecular techniques widely used in previous studies highlight a minimum number of 5 animals for tissue biodistribution. The sample size calculation based on the longitudinal strain of the first 5 animals of each group from the substudy 1 (pilot) showed the average value of −10.3% with 1.9% (SD) variability in the group with cancer (MMTV‐PyMT+) after the period of tumor development and the average value of −12.9% with 2.6% (SD) variability in the group without cancer (MMTV‐PyMT−) after the corresponding period of the experimental protocol. Expecting that these differences would be maintained in the total sampling with a power of 80% and confidence of 95% (α of 5%), the sample size would be 13 animals in each group. Thus, 14 and 15 mice were included in the substudy 1. In the complementary study (substudy 2), 10 to 12 mice were involved in the experiment. All animals were monitored by high‐resolution echocardiographic imaging and exercise tolerance test (ETT), with the exception of 1 animal in the sedentary group, which died during the experimental protocol.

The mice were matched for age under the same environmental conditions. This experiment began when the tumor became palpable, which occurred around the 12th week of age and 20 g body weight and ended when the tumor reached 1 cm3 (average: 22 weeks), according to regulations from the Scientific Committee for Animal Research of the University of São Paulo Medical School. Body weight was assessed every week and tumor growth every 2 to 3 days. Echocardiography and ETT were conducted at the beginning and at the end of the experiment. When tumors reached 1 cm3, the animals were euthanized after lethal dose by isoflurane (5%) inhalation. After loss of consciousness and motor activity, there was exsanguination and removal of tissues for analysis. Lungs, liver, spleen, kidneys, white and brown fat, anterior tibial and soleus muscles, and tumors were weighed and banked (Figure 1). The heart was only deposited.

Figure 1

Experimental protocol.

ETT indicates exercise tolerance test; PNM, palpable and not measurable tumor; Pre, pre phase tumor growth (BASELINE); and Post, post phase tumor growth (END).

Next, we randomized MMTV‐PyMT+ mice into 2 groups: sedentary (n=10) and exercise‐trained (n=12). The animals were matched for age and selected in a systematic random manner. ECG and ETT evaluation were performed when the tumor became apparent and when it reached 1 cm3. In addition, the ETT was performed every 4 weeks during the experimental protocol to readjust training intensity. The test was performed in the control group in the same period. Body weight was assessed every week and tumor growth every 2 to 3 days. At the end of the protocol, all animals were euthanized after lethal dose by isoflurane (5%) inhalation. After loss of consciousness and motor activity, there was exsanguination and removal of tissues for analysis. Heart, liver, spleen, kidneys, white and brown fat, anterior tibial and soleus muscles, and tumors were weighed and banked for analysis (Figure 1). The lungs were only inflated and stored for further analysis.

The researchers analyzing the study samples were blinded to the experimental arms to assure both animal experiments were performed in a blinded manner.

Measures and Procedures

During the experimental protocol, measurements were performed in the morning, while the functional procedures were performed in the afternoon.

Exercise Tolerance Test

ETT was performed on a motor treadmill as previously described. In brief, exercise started with a 6‐m/min speed increment of 3 m/min every 3 minutes up to the maximal speed tolerated. Exhaustion was determined when the mice could no longer run despite further speed increment. The maximal distance, maximal velocity, and duration of the exercise were registered.

ET Protocol

ET was performed on a treadmill at 50% to 60% of the maximal speed achieved in the ETT. Training sessions occurred 5 times a week, with a session duration of 60 minutes.

High‐Resolution Echocardiography

To assess cardiac morphology and function, transthoracic high‐resolution echocardiography was performed using a Vevo 2100 system (Visual Sonics) with an MS400 (30‐MHz centerline frequency) probe. In brief, mice were anesthetized with isoflurane (Zoetis IsoFlo; induction 3.0% and maintenance 1%–3%) and hair was removed from the chest. Mice were subsequently placed in the supine position on a heated table with embedded ECG leads and the core temperature maintained at 37°C. Anesthesia was maintained with 1% to 3% isoflurane throughout the study. Image acquisition was performed by 2‐dimensional mode in the parasternal long‐axis view followed by the short‐axis and apical 4‐chamber views. Two‐dimensional guided M‐mode imaging was used to measure left ventricular (LV) end‐diastolic diameter (LVEDD), LV end‐systolic diameter (LVESD), interventricular septum (IVS), and posterior wall thickness (LVPW), corrected by body weight. Also, the LV end‐diastolic volume (LVEDV), LV end‐systolic volume (LVESV) and mass , both corrected by body weight, fractional shortening , and ejection fraction by Teichholz method were calculated from M‐mode echocardiography. Left atrium (LA) diameter was acquired from 2‐dimensional parasternal long‐axis view. Diastolic function was evaluated using conventional echocardiography with tissue Doppler imaging and pulsed‐wave Doppler techniques. Transmitral inflow velocities were recorded from an apical long‐axis view by setting the sample volume in the tip of the mitral leaflets. From the pulsed‐wave Doppler spectral waveforms, the peak early‐ and late‐diastolic transmitral velocities (E and A waves) to obtain the E/A ratio were measured. E‐wave deceleration time and isovolumetric relaxation time (IVRT) were corrected by heart rate dividing them by the square root of the R‐R interval. Isovolumetric contraction time (IVCT) and ejection time (EjT) were measured to obtain myocardium performance index . From the tissue Doppler spectral waveforms, E′ (early diastolic mitral annular peak velocity at the septum) was measured and E/E′ ratio was calculated.

Speckle‐tracking imaging analyses were performed on parasternal long‐axis and short‐axis view (papillary level) using a Visual Sonics Vevo 2100 system (Visual Sonics). Image depth, width, and gain settings were optimized to improve image quality. Two‐dimensional loops with a frame rate ≥200 frame/s were utilized for speckle‐tracking imaging. All images were digitally stored in cine loops consisting of 300 frames, acquired, and then analyzed using Vevo Strain Software (Vevo LAB1.7.1). Strain, which evaluates length changes relative to initial length, was calculated either in the radial short‐ and long‐axis (thickening of the LV wall along radius axis) or longitudinally (shortening of the LV wall along long axis) and circumferentially (reduction in the circumference of the LV cavity). Strain rate, defined as the rate of change of this deformation over time, was also measured.

Cardiac Morphological and Structural Analysis

Heart sections were fixed in 10% buffered formalin for 48 hours. After routine histologic processing, paraffin‐embedded tissue was sectioned (5‐µm‐thick sections) and the histologic slides were stained with hematoxylin‐eosin and Sirius red for quantitative analysis of cardiomyocyte diameters and of collagen rate, respectively. Measurement of the transverse diameter of the cardiomyocytes was performed in a computerized system (LEICA QUANTIMET 500) coupled with a ×40 magnification optical microscope. Cardiomyocyte diameter was calculated from an average of all fields of a single cut of the LV per animal, considering the fibers intact. The degree of myocardial fibrosis was analyzed by means of a complete ventricular wall scan performed under a polarized light optical microscope (OLYMPUS BX‐41) at ×40 magnification. The type I collagen (yellowish orange and reddish orange) and type III collagen (green or yellowish green) were evaluated by staining. The analysis was performed by ImageJ software (National Institutes of Health) using the threshold color plug‐in to obtain the collagen percentage by analyzing the stained pixels according to the area measurement. Threshold color values for each collagen type were standardized: matrix 0 to 40 for collagen type I red color and 45 to 120 for collagen type III green color, with saturation 0 to 255 and brightness 5 to 225 for both.

Expression of Protein Levels Analysis by Western Blot

The protein levels of sodium/calcium exchanger, sarcoplasmic or endoplasmic reticulum calcium ATPase cardiac isoform 2a (SERCA2a), and phospholamban and phospho‐phospholamban (p‐phospholamban) in the heart muscle lysate were analyzed by Western blot. Frozen samples were homogenized in cell lysis buffer (100 mmol/L TrisHCl, 50 mmol/L NaCl, 1% Triton X‐100) and protease and phosphatase inhibitor cocktail (1:100; Sigma‐Aldrich). Heart tissue debris was removed by centrifugation at 3000g, 4°C, for 10 minutes. Samples were loaded and subjected to SDS‐PAGE on polyacrylamide gels (8%–15%) depending on the protein molecular weight. After electrophoresis, proteins were electrotransferred to a nitrocellulose membrane (BioRad Biosciences). Equal loading of samples (30 μg) and even transfer efficiency were monitored with the use of 0.5% Ponceau staining of the blot membrane. The blot membrane was then incubated in a blocking buffer (5% bovine serum albumin, 10 mmol/L Tris‐HCl [pH 7.6], 150 mmol/L NaCl, and 0.1% Tween 20) for 2 hours at room temperature and then incubated overnight at 4°C with anti‐sodium/calcium exchanger (Thermo Fisher, #MA3‐926), anti‐SERCA2 ATPase (Thermo Fisher, #MA3‐919), anti‐phospholamban (Thermo Fisher, #MA3‐922), and anti–p‐phospholambanSer16,Thr17 (Thermo Fisher, #711401) antibodies. Binding of the primary anti‐sodium/calcium exchanger, anti‐SERCA2 ATPase, anti‐phospholamban, anti–p‐phospholambanSer16,Thr17, and anti‐GAPDH was detected by secondary antibody (IRDye 800CW anti‐mouse [#926‐32210] or IRDye 800CW anti‐rabbit [#926‐32211], and LI‐COR Biosciences), and detection was performed by LI‐COR (LI‐COR Biosciences). The bands were analyzed using Image Studio Lite software (LI‐COR Biosciences). The results of protein expression were normalized by Ponceau staining. They are expressed as a percentage of control expression for each group. Of note, the Western blot assays were run in triplicate for all of the Ca2+‐handling proteins, with the exception of the p‐phospholamban protein, performed in duplicate.

Citrate Synthase Activity

Activity of citrate synthase enzyme in the anterior tibial and soleus muscles was evaluated as a marker of ET effectiveness. The enzyme activity was determined by means of commercial citrate synthase assay kit (Sigma‐Aldrich).

Statistical Analysis

Values are expressed as means±SD or medians and interquartile range (IQR). Animal body weights were evaluated weekly, but only the values of every 2 weeks were used for the analysis. The echocardiogram and speckle tracking measures were compared according to group and time of evaluation using generalized estimation equations with normal marginal distribution and identity link function, assuming a first‐order autoregressive correlation matrix between moments. In case of significant difference, the results were followed by Bonferroni multiple comparisons to identify between group and time differences. The nonparametric parameters or those with a sample size of <15 animals were compared by using Mann‐Whitney test. The analyses were performed using SPSS for Windows version 20.0 software (IBM) and tabulated using Microsoft Excel 2003 software. The tests were performed with a 5% significance level.

RESULTS

Breast Cancer Provokes Physical Characteristic Changes

Body weight increased in both MMTV‐PyMT− and MMTV‐PyMT+ mice (Figure S1, P time<0.001). Despite significant interaction (P interaction=0.013), it was not possible to identify difference between groups throughout the study. Body weight after euthanasia was higher in MMTV‐PyMT+ mice (P<0.001). This difference was caused by tumor burden, since body weight without tumor was similar between groups (Figure S1). Anterior tibial muscle, lung, and brown and white fat weight were not different between MMTV‐PyMT+ and MMTV‐PyMT− mice. Soleus muscle and kidney weight was lower and liver and spleen weight were higher in MMTV‐PyMT+ mice compared with MMTV‐PyMT− mice (Table S1).

Breast Cancer Changes Physical Capacity

Time to exhaustion (P interaction=0.031) and treadmill velocity (P interaction=0.037) in the maximal exercise test were significantly lower in the MMTV‐PyMT+ mice. Distance in the maximal exercise test was statistically lower in the MMTV‐PyMT+ mice (P group=0.003) regardless of the time evaluated. These results are summarized in Table 1.

Table 1

Physical Capacity Obtained in the Maximal Exercise Test in MMTV‐PyMT− and MMTV‐PyMT+ Mice at the Beginning and End of Experimental Protocol

Variable	MMTV‐PyMT−		MMTV‐PyMT+		P Group	P Time	P Interaction
	Pre	Post	Pre	Post
Distance, m	703.4±128.5	599.1±224.7	653.1±95.9	384.9±187.6	0.003*	<0.001*	0.079
Time, min	35.9±3.4	32.9±6.2	34.6±2.5	25.7±6.4	0.001*	<0.001*	0.031*
Velocity, m/min	36.2±3.3	33.7±6.3	34.9±2.8	26.6±6.6	0.002*	<0.001*	0.037*

Generalized estimation equations with normal distribution and identity link function using AR(1) correlation matrix between evaluations. Values are expressed as means±SD. One mouse in the mouse mammary tumor virus–polyomavirus middle T antigen (MMTV‐PyMT)− group and 1 mouse in the MMTV‐PyMT+ group were not included in exercise tolerance test. The MMTV‐PyMT− mouse refused to run on the treadmill, while the MMTV‐PyMT+ mouse had its stride motion compromised by the tumor. Thus, 13 mice in the MMTV‐PyMT− and 14 mice in the MMTV‐PyMT+ groups were included in this study.

Significant difference.

Breast Cancer Alters Cardiac Morphology and Function

The pooled data (beginning and end of the study) of the LVEDD (PyMT− pre: 0.185±0.019 and post: 0.168±0.01 mm/g versus PyMT+ pre: 0.177±0.014 and post: 0.151±0.014 mm/g [P group<0.001, P interaction=0.417]) and LVEDV (PyMT− pre: 2.99±0.41 and post: 2.64±0.38 μL/g versus PyMT+ pre: 2.61±0.35 and post: 2.45±0.34 μL/g [P group=0.003, P interaction=0.437]) were lower in the MMTV‐PyMT+ mice when compared with MMTV‐PyMT− mice (Figure 2). LVESD and LVESV were similar between MMTV‐PyMT +and MMTV‐PyMT− mice. The LA responses were statistically different between groups throughout the experimental protocol (P interaction=0.004). The comparison between groups showed that at the beginning of the study, the LA was not different between the MMTV‐PyMT+ and MMTV‐PyMT− groups; however, at the end of the study, the LA was lower in MMTV‐ PyMT+ mice (0.079±0.001 versus 0.091±0.008 mm/g, P=0.030) (Figure 2 and Table S2). The pooled data (beginning and end of the study) of LV mass was lower in MMTV‐PyMT+ mice than in MMTV‐PyMT− mice (PyMT− pre: 4.297±0.798 and post: 3.396±0.496 mg/g versus PyMT+ pre: 3.454±0.649 and post: 2.968±0.601 mg/g [P group<0.001, P interaction=0.365]) (Figure 2 and Table S2). E/E′ wave, E‐wave deceleration time, and IVRT values were not different between groups. Similarly, heart rate, fractional shortening, ejection fraction, and myocardium performance index were not different in MMTV‐PyMT− and MMTV‐PyMT+ mice.

Figure 2

Morphological and functional cardiac parameters in mouse mammary tumor virus–polyomavirus middle T antigen (MMTV‐PyMT)− and MMTV‐PyMT+ mice in the pre‐experimental and post‐experimental protocol.

Left‐ventricular (LV) end‐diastolic diameter (LVEDD) (A); LV end‐diastolic volume (LVEDV) (B); LV end‐systolic diameter (LVESD) (C); LV end‐systolic volume (LVESV) (D); left atrium (LA) (E); and LV mass (F). Parameters were corrected by weight. LVEDD: P group<0.001, P interaction=0.417; LVEDV: P group=0.003, P interaction=0.437; LA: P interaction=0.004; post hoc test: * P=0.030, between group difference and †† P<0.001, within MMTV‐PyMT+ group difference; and LV mass: P group<0.001, P interaction=0.365. Generalized estimation equations with normal distribution and identity link function using AR(1) correlation matrix between evaluations followed by Bonferroni multiple comparison. Values are expressed as means and SD. Four mice in the MMTV‐PyMT− and 2 mice in the MMTV‐PyMT+ were excluded from this analysis because of poor imaging quality on echocardiography. Thus, 10 mice in the MMTV‐PyMT− and 13 mice in the MMTV‐PyMT+ were involved in this analysis. Of note, 2 LA measures were not obtained in the MMTV‐PyMT+ mice.

Longitudinal strain and strain rate at the beginning and at the end of the study are shown in Figure 3. The pooled data (beginning and end of the study) of the longitudinal strain (PyMT− pre: −14.4±4.8 and post: −14.5±2.7% versus PyMT+ pre: −12.4±2.4 and post: −9.8±1.7% [P group<0.001, P interaction=0.126]) and strain rate (PyMT− pre: −5.07±2.27 and post: −6.03±3.56 s−1 versus PyMT+ pre: −4.39±1.47 and post: −3.66±0.92 s−1 [P group=0.030, P interaction=0.168]) were impaired in the MMTV‐PyMT+ mice when compared with MMTV‐PyMT− mice. Both the longitudinal strain and strain rate were statistically lower in absolute value in animals with cancer than in control animals. Radial strain and strain rate, radial and circumferential short systolic axis strain, and strain rate were similar between MMTV‐PyMT− and MMTV‐PyMT+ mice (Table S2).

Figure 3

Cardiac function by speckle tracking in mouse mammary tumor virus–polyomavirus middle T antigen (MMTV‐PyMT)− and MMTV‐PyMT+—longitudinal long‐axis strain (A) and strain rate (B).

Longitudinal strain (P group<0.001, P interaction=0.126) and strain rate (P group=0.030, P interaction=0.168) were lower in MMTV‐PyMT+ mice when compared with MMTV‐PyMT− mice. Pre: pre phase tumor growth and post: post phase tumor growth. Generalized estimation equations with normal distribution and identity link function using AR(1) correlation matrix between evaluations. Values are expressed as means and SD. Three mice in the MMTV‐PyMT− and 3 mice in the MMTV‐PyMT+ groups were excluded from the speckle‐tracking analysis because of poor imaging quality. Thus, 11 mice in the MMTV‐PyMT− and 12 mice in the MMTV‐PyMT+ were involved in this analysis.

Cardiac myocyte diameter and cardiac collagen volume fraction were similar between MMTV‐PyMT− and MMTV‐PyMT+ mice (Table 2).

Table 2

Cardiac Morphology in MMTV‐PyMT− and MMTV‐PyMT+

Variable	MMTV‐PyMT−		MMTV‐PyMT+		P Value
	Median (IQR)	N	Median (IQR)	N
Myocyte diameter, µm	16.9 (15.8–20.7)	7	16.7 (16.3–18)	8	>0.999
Type 1 collagen, µm2	339.3 (93.2–488.1)	7	381.7 (222.5–497.3)	7	0.535
Type 3 collagen, µm2	403.4 (107–921.4)	7	902.5 (285–1000.8)	7	0.535
Type 1+3 collagen, µm2	911 (194.9–1261.6)	7	1125 (621.2–1498.1)	7	0.383

Mann‐Whitney test. Values are expressed as medians and interquartile range (IQR). MMTV‐PyMT indicates mouse mammary tumor virus–polyomavirus middle T antigen.

Breast Cancer Changes Myocyte Ca2+‐Handling Protein Expression

In an attempt to understand the molecular basis of the alterations in the longitudinal strain and strain rate in the MMTV‐PyMT+ mice at end of the study protocol, we studied Ca2+‐handling proteins in the cardiac myocyte (Figure 4). In the MMTV‐PyMT+ mice, phospholamban protein expression was increased (247.8% [IQR, 192.3%–360.4%] versus 105.4% [IQR, 76.8%–123.7%], P=0.011), whereas sodium/calcium exchanger protein expression was decreased (40.5% [IQR, 11%–66.6%] versus 101.5% [IQR, 64.2%–143.5%], P=0.017) compared with MMTV‐PyMT− mice. SERCA2a protein expression was not different between the MMTV‐PyMT+ and MMTV‐PyMT− groups (51.1% [IQR, 37%–84.5%] versus 91.3% [IQR, 55.2%–159.7%], P=0.165); however, the SERCA2a/phospholamban ratio was reduced in the MMTV‐PyMT+ mice (0.23% [IQR, 0.15%–0.61%] versus 0.86% [IQR, 0.65%–1.41%], P=0.007). We also observed that p‐phospholamban was lower in the MMTV‐PyMT+ compared with the MMTV‐PyMT− mice (31.2% [IQR, 3.5%–69.6%] versus 83.8 [IQR, 47.9%–163.3%], P=0.038), but because of the lack of a sufficient sample we were able to test this marker only 2 times.

Figure 4

Cardiac myocyte Ca2+‐handling proteins in mouse mammary tumor virus–polyomavirus middle T antigen (MMTV‐PyMT)− and MMTV‐PyMT+ mice—phospholamban (A); SERCA2a (sarcoplasmic or endoplasmic reticulum calcium 2 ATPase) (B); phosphorilated phospholamban (p‐phospholamban) (C); sodium/calcium exchanger (NCX) (D); and SERCA2a/phospholambam ratio (E).

Western blot bands are shown for each protein. In the MMTV‐PyMT+ mice, phospholamban protein expression was increased (P=0.011) and SERCA2a protein expression was not different between the MMTV‐PyMT+ and MMTV‐PyMT− groups (P=0.165). SERCA2a/phospholamban ratio was reduced in the MMTV‐PyMT+ mice (P=0.007). p‐Phospholamban and NCX protein expression were decreased (P=0.038 and P=0.017, respectively) compared with MMTV‐PyMT− mice. Mann‐Whitney test. Values are expressed as medians and interquartile ranges (IQRs). Animals per group: 7 MMTV‐PyMT− and 7 MMTV‐PyMT+.

ET Does Not Reverse Morphologic and Physiologic Changes in MMTV‐PyMT+ Mice and Does Not Delay Time to Tumor Growth

Next, we assessed the impact of ET in modulating the changes in exercise capacity, cardiac function, and tumor growth in the MMTV‐PyMT+ mice. ET had no effect on tumor growth rate. Time of tumor growth to reach 1 cm3 was not significantly different between sedentary and exercise‐trained mice (11.0±1.3 versus 12.8±0.6 weeks, P=0.215).

ET did not affect animal body weight (Figure S2; P interaction=0.930) or weight of the heart, anterior tibial and soleus muscles, liver, spleen, kidneys, and brown and white fat (Table S3; P>0.05). ET did not seem to prevent the reduction in physical capacity provoked by cancer (Figure S3). There was a significant reduction in the distance covered in the ETTs performed during the experimental protocol in both groups (P time<0.001), although no significant difference in running distance was observed between the sedentary and exercise‐trained groups at 0, 1, 2, and 3 months (P interaction=0.069). This occurred despite a significant increase in citrate synthase enzyme activity in both anterior tibial and soleus muscle (Figure S4; P<0.001).

Data of cardiac morphology and function in sedentary and exercise‐trained MMTV‐PyMT+ mice are shown in Figure S5 and Table S4. ET decreased LVESD (P group=0.038, P interaction=0.226) and LVEDV (P group=0.026, P interaction=0.431), but did not change the other parameters associated with cardiac morphology (Table S4). Despite the average difference in longitudinal strain rate responses throughout the experimental protocol (P interaction=0.032; Table S4), it was not possible to identify differences between sedentary and exercise‐trained MMTV‐PyMT+ mice.

ET did not improve phospholamban protein expression. Likewise, ET did not change SERCA2a, p‐phospholamban, and sodium‐calcium exchanger protein expression. In addition, ET did not alter SERCA2a/phospholamban ratio (Figure S6). Additional results can be found in Data S1.

DISCUSSION

Previous studies have shown that chemotherapy agents used for treatment of breast cancer are associated with cardiotoxicity. , Changes in QT interval, , ventricular arrhythmias, , acute coronary syndrome, , and systolic and diastolic LV dysfunction have been reported in patients undergoing chemotherapy. However, the impact of cancer progression in cardiovascular function remains poorly addressed. In this regard, our study shows for the first time that breast cancer itself, in an advanced stage, is associated with myocardial dysfunction. In MMTV‐PyMT+ mice, we show that the presence of breast cancer was associated with worse longitudinal long‐axis strain and strain rate, known subclinical markers of cardiac dysfunction, and impaired exercise capacity, both at least partly related to Ca2+protein‐handling deregulation.

The evaluation of ventricular function by speckle‐tracking imaging has been highly used in clinical practice. Ventricular longitudinal deformation alterations have been used for early detection of ventricular impairment before changes in circumferential and radial deformations in most cardiac diseases with high sensitivity and specificity. In addition, longitudinal deformation changes are prognostic of cardiotoxicity even in the presence of preserved ejection fraction. Our study shows for the first time that breast cancer is associated with impairment in LV longitudinal strain and strain rate, both consistent with subclinical myocardial dysfunction. Furthermore, these myocardial alterations are associated with an increase in phospholamban expression and a decrease in sodium/calcium exchanger, p‐phospholamban expression, and SERCA2a/phospholamban ratio in the cardiac myocyte of MMTV‐PyMT+ mice, all consistent with altered myocardial Ca2+ handling. Ca2+‐related proteins play an important role in myocardial contraction and relaxation. Phospholamban binds SERCA2a in a 1:1 heterodimeric regulatory complex, which results in inhibition of SERCA2a activity. , Reduction in phosphorylation of phospholamban at Ser16 by protein kinase A and Thr17 by calmodulin‐dependent protein kinase II increases the inhibition of phospholamban on SERCA2a. These responses reduce Ca2+ reuptake into the sarcoplasmic reticulum during the muscle relaxation phase and the load of Ca2+ necessary for myocardial contraction in the subsequent cardiac cycle, which could explain the impairment in LV longitudinal strain and strain rate. Although our findings do not establish a definitive causal effect, they strongly suggest that Ca2+‐handling deregulation plays at least a partial role in breast cancer–induced cardiac dysfunction.

Another important finding in our study was the association between breast cancer and changes to cardiac morphology. The reductions in LV diameter and volume associated with lower LV mass and LA size is suggestive of cardiac remodeling in consequence of breast cancer. These morphological changes can predispose to progression of ventricular dysfunction and malignant arrhythmias. Cancer cachexia has been associated with loss of cardiac mass and, consequently, cardiac dysfunction and sarcopenia. The sarcopenia has been described in 25% of patients with metastatic breast cancer. , In fact, we found decreased soleus muscle in MMTV‐PyMT+ mice. Despite the changes in cardiac morphology, we found no changes in cardiomyocyte diameter and fibrosis. The histological adaptations related to heart mass loss seem to occur as a late compensation event, along with manifestations of overt clinical cardiac dysfunction, such as a fall in ejection fraction and fractional shortening.

Other than cardiac dysfunction, our findings in the MMTV‐PyMT mouse model highlight that the systemic effects of cancer can affect other organs as well. Some findings include increased liver and spleen dimensions and decreased kidneys size compared with MMTV‐PyMT− mice. Despite no functional studies having been performed in these organs, hepatomegaly and kidney atrophy are well‐establish signs of organ dysfunction. , In addition, splenomegaly has been associated with changes in immune function , that can further affect the prognosis. Spleen enlargement was associated with an increase in the percentage of CD3+T lymphocytes and CD8+T lymphocytes. Of note, ET did not impact the weight of these organs in MMTV‐PyMT+ mice.

Our group has previously reported that ET improves the net balance of cardiac Ca2+‐handling–related proteins in heart failure, and similar findings were observed in obese rat models. This improvement seems to be related to a reduction in cardiac phospho–Thr17‐phospholamban and phospho–Ser2808‐Ryanodine protein expression, which ultimately improves cardiac function. Based on these findings, we hypothesized that ET would prevent cancer‐induced cardiac dysfunction, but this was not the case. ET decreased LVESD and LVEDV, but physical capacity was not overall changed. Likewise, ET did not change cardiomyocyte Ca2+‐handling protein expression. It is likely that the lack of anti‐tumor effect of ET alone in the absence of specific anti‐cancer therapy to halt cancer progression is at least partly if not completely associated with these findings. This is supported by the fact that ET alone did not impact the tumor growth rate in the MMTV‐PyMT+ mice. Another possibility is that our exercise paradigm was not adequate to prevent the changes in cardiac function in MMTV‐PyMT+ mice. Despite forced treadmill exercise being an accepted model of animal training, it may be insufficient or inadequate for animals with ongoing cancer progression. We are convinced that studies involving ET in cancer models should continue. However, we recognize that these studies should be performed in the context of anti‐cancer treatment to understand how ET may sinergize with specific anti‐cancer therapies to promote cancer control and/or preservation of body homeostasis. This is an area of ongoing investigation in our group as well as others.

Study Limitations

The animals were euthanized when the tumor reached 1 cm3, ie, at the late stage of breast cancer. Thus, our study does not provide information on the effects of breast cancer on cardiac function during the development of the disease. The reduction in caveolin‐1 following transformation of activated oncogenes, including the polyomavirus middle T antigen (PyMT), and the known role of caveolin‐1 in several cardiovascular alterations could lead someone to raise the hypothesis that the cardiac alterations in the present study were due, in part, to the reduction in caveolin‐1. The activation in mitogen‐activated protein kinase (MAPK) and phosphoinositide 3‐kinase (PI3K) pathways associated with PyMT , might have also contributed to the cardiac alterations in the present study. The role of extracellular signal‐regulated kinase (ERK), c‐Jun N‐terminal kinase (JNK), and p38 MAPK has been established in animal models and humans with heart failure. , The increase in total PI3K activity can negatively regulate Ca2+ transients, which results in decreasing of positive inotropic effects of β‐adrenergic agonists. Unfortunately, the caveolin‐1, MAPK, and PI3K were not investigated in our study. This is an interesting topic for further investigations.

The study was entirely conducted in animals, which limits its extrapolation to humans with breast cancer. In addition, we analyzed a breast cancer model (MMTV‐PyMT) that only resembles HER2‐positive breast cancer in humans. Thus, our findings cannot be generalized to other tumor types. However, the MMTV‐PyMT mouse is a unique transgenic model of multifocal breast adenocarcinoma because cancer development occurs similarly to that of invasive ductal carcinoma of the human breast. Cancer spontaneously develops and gradually progresses from hyperplasia, dysplasia, adenoma, and carcinoma because of expression of the polyoma middle oncoprotein (PyMT). Therefore, we believe that our study will provide important information that will be useful for future investigations in the clinical setting.

Conclusions

Breast cancer is associated with decreased exercise capacity and subclinical LV dysfunction in MMTV‐PyMT+ mice, which is directly associated with dysregulation of cardiomyocyte Ca2+‐handling. ET alone is not an effective strategy to prevent or reverse these changes.

Sources of Funding

This work was supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP#2015/22814‐5). Gomes‐Santos and Passos received scholarship from FAPESP (#2014/13690‐8 and #15/19076‐2, respectively). Brum, Oliveira, Salemi and C. E. Negrao were supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPQ#306261/2016‐2; #313479/2017‐8; #307227/2018‐9 and; #303573/2015‐5).

Disclosures

None.

Data S1

Tables S1–S4

Figures S1–S6

Click here for additional data file.