Measurement of Ejection Fraction by Cardiac Magnetic Resonance Imaging and Echocardiography to Monitor Doxorubicin-Induced Cardiotoxicity

Abstract

Doxorubicin is a standard treatment option for breast cancer, lymphoma, and leukemia, but its benefits are limited by its potential for cardiotoxicity. The primary objective of this study was to compare cardiac magnetic resonance imaging (CMRI) versus echocardiography (ECHO) to detect a reduction in left ventricular ejection function, suggestive of doxorubicin cardiotoxicity. We studied eligible patients who were 18 years or older, who had breast cancer or lymphoma, and who were offered treatment with doxorubicin with curative intent dosing of 240 to 300 mg/m ² body surface area between March 1, 2009 and October 31, 2013. Patients underwent baseline CMRI and ECHO. Both imaging studies were repeated after four cycles of treatment. Ejection fraction (EF) calculated by both methods was compared and analyzed with the inferential statistical Student's t test. Twenty-eight eligible patients were enrolled. Two patients stopped participating in the study before undergoing baseline CMRI; 26 patients underwent baseline ECHO and CMRI. Eight of those 26 patients declined posttreatment studies, so the final study population was 18 patients. There was a significant difference in EF pre- and posttreatment in the CMRI group ( p  = 0.009) versus the ECHO group that showed no significant differences in EF ( p  = NS). It appears that CMRI is superior to ECHO for detecting doxorubicin-induced reductions in cardiac systolic function. However, ECHO is less expensive and more convenient for patients because of its noninvasive character and bedside practicality. A larger study is needed to confirm these findings.

Anthracyclines are the standard choice for the primary treatment of malignant tumors and hematologic abnormalities. Doxorubicin (DOX), a commonly known anthracycline, is standard therapy for breast cancer, lymphoma, and leukemia. The benefits of DOX, however, are limited by toxicities, most notably cardiotoxicity. ¹ Cardiac failure during DOX treatment is defined as “an absolute LVEF [left ventricular ejection fraction] below 45%, a 10% decline in LVEF to below the lower limit of normal, or an absolute decline of 20% at any level.” ¹ The incidence of cardiotoxicity varies with the type of drug, the dose administered each cycle, the administration schedule, the combination of adjuvant drugs and therapies, and the preexistence of heart disease. ²

It has been suggested that DOX contributes to the progression of cardiomyopathy by influencing apoptosis through activation of cytochrome c , which in turn activates caspase 3, possibly through the mitochondrial pathway. ³ After the use of a tunneling technique for comparing DOX with saline injected into rat primary-cultured cardiomyocytes, left ventricular (LV) fractional shortening was significantly decreased and LV right wall thickness (RWT) was significantly reduced at 16 weeks of age. The reduction in RWT was thought to result from the increase in wall stress due to the LV morphological changes and was associated with impaired cardiac function. ³ This finding was also supported by studies that determined that through the release of cytochrome c , DOX induces an oxidative stress resulting in mitochondrial-mediated apoptosis. ⁴ This cumulative dose-dependent form of cardiotoxicity is known as type I chemotherapy-related cardiac dysfunction, which arises from cell damage that results in cell death, an irreversible process that is important to detect early. ⁵

Recent findings of the superiority of cardiovascular magnetic resonance imaging (CMRI) over echocardiography (ECHO) and multigated acquisition scanning (MUGA) ⁶ have led us to hypothesize that CMRI can be used to better predict cardiac failure in patients receiving DOX. With improvements in spatial and temporal resolution, CMRI has become an important noninvasive method for assessing cardiac changes in LV systolic dysfunction. ⁶

The purpose of our study was to investigate which imaging modality (CMRI vs. ECHO) would be useful in detecting early systolic dysfunction in patients treated with DOX in patients with lymphoma and breast cancer. Our study focused on a selected cohort of patients who were studied in a prospective manner to prevent bias in the interpretation of results.

Methods

Participant Population

All patients had been referred for the treatment of breast cancer or lymphoma and were treated with DOX with curative intent dosing (240–300 mg/m ² body surface area [BSA]). To meet eligibility requirements, patients needed to be 18 years or older, have a baseline LV ejection fraction (LVEF) of 50% or more, and women were required to be postmenopausal, surgically sterile, and sexually abstinent or, if sexually active, using an effective method of birth control. Patients were excluded from the study if they had a contraindication to undergoing CMRI, could not provide written consent, had a life expectancy of less than 1 year, had concurrent cardiac failure or a history of nonchemotherapy-induced cardiac failure, or had a glomerular filtration rate of less than 30 mL/min. The protocol of the study was approved by the Mayo Clinic Institutional Review Board.

Analyses of Imaging

For all patients, blinded ECHO of LVEF was evaluated by one cardiologist (T.T.). Likewise, blinded CMRI was evaluated by one radiologist (S.A.M.). LVEF was measured with standard cardiac evaluation software with manual tracing applied on a set of contiguous short-axis views (section thickness, 8–10 mm) at end-diastole and end-systole.

ECHO Protocol

Standard transthoracic ECHO images were obtained with the ACUSON Sequoia C256 (Model GTIN 8246951, serial number 52133) (Siemens Medical Solutions USA, Inc., Mountain View, CA). Image acquisition was in accordance with the recommendations of the American Society of Echocardiography (ASE) with the 16-segment model. LVEF was calculated with the Simpson biplane method. Complete M-mode and Doppler ECHO measurements were performed according to the ASE guidelines. The ECHO examinations were done by two different sonographers. Fig. 1 shows the two-dimensional (2D) measurements for volume calculations with the biplane method.

Fig. 1.

Two-dimensional measurements for volume calculations using the biplane method of discs (modified Simpson rule) in the apical four-chamber (A4C) and apical two-chamber (A2C) views at end-diastole (LV EDD) and at end-systole (LV ESD). (Reproduced with permission fromLang, R M, Bierig, M. Recommendations for Chamber Quantification. Oxford University Press, 2006; 7(2): 79–108.)

CMRI Protocol

CMRI was performed with a 1.5-T magnetic resonance scanner with a high-definition cardiac coil and standard cardiac software equipment. Cardiac function was measured in two- and four-chamber views with Fast Imaging Employing Steady-state Acquisition-Cine sequence (ECHO time [TE] 1.4 milliseconds, repetition time [TR] 3.3 milliseconds, matrix 192 × 192, fractional anisotropy [FA] flip angle 50 degree, FOV (field of view) 360 × 360 mm, 20 phases per heart cycle, section thickness 8 mm). Contiguous short-axis planes (section thickness 8–10 mm, gap 0–2 mm) were used to cover the ventricle. Fig. 2 shows a sample CMRI images used for the calculation of EF.

Fig. 2.

Cardiac magnetic resonance imagings: ( A ) four-chamber view, ( B ) two-chamber view, ( C ) three-chamber view, and ( D ) short-axis view.

A T1-weighted fast-gradient ECHO sequence was used (TR 4.7 milliseconds, TE 1.3 milliseconds, inversion time [TI] 175–250 milliseconds flip angle 30°, section thickness 8–10 mm, section spacing 0–2 mm, matrix 192 × 160, FOV 360 × 360 mm). Gadolinium-diethylenetriamine-penta-acetic acid was injected into an intravascular peripheral catheter and flushed with 10 mL saline. Postcontrast images were acquired 10 minutes after the injection while patients were asked to hold their breath. All diagnostic testing (both CMRI and ECHO) was performed within 1 week of the last cycle of chemotherapy.

Statistical Analysis

Data were analyzed with the Student's paired t test for evaluating the significance of results between the two modalities (CMRI and ECHO) and the two time frames (baseline and posttreatment). A p- value of less than 0.05 was considered statistically significant. Statistical analysis was performed with SPSS software (IBM, Armonk, NY) by a co-author (C.M.J.).

Results

A total of 28 eligible patients were enrolled into the study; 26 underwent baseline ECHO and CMRI; 18 completed the posttreatment studies after four cycles of DOX (15 women and 3 men, aged 39–78 years; mean age, 55.9 years); and 8 patients declined to undergo posttreatment studies. No patients withdrew for clinical reasons. Of the 18 patients who completed the study, 14 were being treated for breast cancer and 4 were being treated for lymphoma. All the patients were chemotherapy-naïve before starting the study, and no patient received cardioprotective substances before or during treatment. The mean (±standard deviation [SD]) DOX dose for the full group was 459.3 (92.5) mg/m ² BSA.

For the entire study population, the mean (± SD) baseline LVEF by ECHO was 61.5% (6.8%) and by CMRI, 61.8% (7.1%) with a mean difference (MD) of −0.311% (4.0%) ( t  =  − 0.164, df = 17, p  = 0.44). Posttreatment mean LVEF by ECHO was 60.9% (5.1%) and by CMRI, 57.1% (7.3%). This posttreatment LVEF difference was significant (MD =  − 3.75% [2.9%], t  = 2.753, df = 17, p  = 0.03). On CMRI, LVEF decreased significantly from baseline to posttreatment (MD =  − 4.47% [3.3%], t  = 2.596, df = 17, p =0.009). ECHO showed a very small decrease in LVEF from baseline to posttreatment, which was not significant (MD = 0.611% [3.3%], t  = 0.393, df = 17, p =0.35) ( Fig. 3 ).

Fig. 3.

Full-group paired differences for left ventricular ejection fraction (LVEF) between modalities and time frames. Mean and standard error (error bars) are shown. CMRI indicates cardiac magnetic resonance imaging; ECHO, echocardiography; ns, not significant; PostTx, posttreatment.

In our sample population, CMRI showed a significant reduction in LVEF (MD = –6.0% [1.5%], t  = 3.092, df = 14, p  = 0.008) before and after treatment for women and a slight increase for men (MD = 2.0% [4.4%], t  = 0.795, df   = 2, p  = 0.51). On CMRI, participants who were 55 years of age or younger had a significant decrease in LVEF (MD =  − 7.2% [5.9%], t  = 3.635, df = 8, p  = 0.007) but participants older than 55 years did not (MD =  − 2.2% [8.6%], t  = 0.75, df = 8, p  = 0.48). On CMRI, patients treated for breast cancer had a significant decrease in LVEF (MD =  − 6.1% [5.7%], t  = 2.955, df = 13, p  = 0.01), but lymphoma patients had a slight increase in LVEF after treatment (MD = 0.5% [4.7%], t  = 0.213, df = 3, p  = 0.84) ( Table 1 ).

Table 1. Cohort characteristics and clinical data.

Cohort variable	DOX dose, mean (SD), mg/m 2	CMRI MD (SD)		LVEF, %, MD (SD)	p -Value a
		Baseline	PostTx
Gender, n (%)
Female 15 (83.3)	441.9 (80.9)	62.5 (7.2)	56.5 (7.1)	−6.0 (1.5)	0.008
Male 3 (16.7)	546.3 (115.6)	60.3 (9.1)	58.3 (6.6)	2.0 (4.4)	0.51
Age, n (%), mean (SD)
≤ 55 years 9 (50), 48.8 (6.3)	499.3 (89.2)	60.4 (7.9)	53.2 (7.9)	−7.2 (5.9)	0.007
> 55 years 9 (50), 63.1 (7.1)	419.3 (81.5)	63.2 (6.4)	61.0 (7.3)	−2.2 (8.6)	0.48
Cancer type, n (%)
Breast 14 (77.8)	443.5 (83.7)	62.4 (7.5)	56.3 (7.3)	−6.1 (5.7)	0.01
Lymphoma 4 (22.2)	514.8 (113.6)	59.8 (6.1)	60.3 (7.4)	0.5 (4.7)	0.84

Abbreviations: CMRI, cardiac magnetic resonance imaging; DOX, doxorubicin; LVEF, left ventricular ejection fraction; MD, mean difference; PostTx, post-treatment; SD, standard deviation.

^a p -Value (2-tail) significance set at < 0.05.

Discussion

Cardiac toxicity has been directly correlated with DOX in less than 1% of patients receiving cumulative doses of 450 mg/m ² BSA, but the proportion increases to 7% if the cumulative dose is increased to 100 mg/m ² BSA (overall ending dose, 550 mg/m ² BSA). ⁷ Although cumulative doses of 700 mg/m ² BSA have been documented after prolonged infusion, the cardiotoxicity rate of 18% inhibits the use of such doses. Oncology treatment guidelines also emphasize the need for extreme caution in applying those doses because cardiac failure has occurred when patients received only one DOX cycle. ¹ Based on this history, standard curative treatment regimens for adjunct breast cancer and large cell lymphoma use a DOX dose of 240 to 300 mg/m ² BSA. This common practice makes 450 mg/m ² BSA a ceiling dose that we are reluctant to exceed. In our study, the average DOX dose for the entire group was 459.3 (± SD 92.5 mg/m ² BSA).

Our data showed that CMRI and ECHO seem to provide comparable results for LVEF. However, CMRI appears to be more sensitive than ECHO in detecting a decrease in LVEF among patients treated with DOX. It is likely that this difference is attributed to the superior spatial resolution of CMRI, which increases its accuracy for measuring LVEF. The results of this study support our hypothesis that CMRI is better suited than ECHO for predicting cardiotoxicity in patients receiving DOX. Similarly, in a small trial of 22 patients in post-anthracycline follow-up ⁷ , CMRI was significantly more useful for detecting early myocardial changes as slightly deteriorating cardiac function with an increase in relative enhancement by more than 5 in predicting an associated loss in LVEF of 16%.

With the establishment of the direct correlative properties of cardiotoxicity in patients treated with DOX, standard American Society of Clinical Oncology guidelines dictate careful monitoring of cardiac LVEF during and after treatment. Risk factors associated with increased risk of toxicity with Anthracycline use include female gender, advanced age (> 65 years), systemic hypertension, cumulative anthracycline dose, underlying cardiovascular disease, and prior chest irradiation. Routine diagnostic measurement tools for cardiac function in the oncology patient consist of ECHO and MUGA. ¹ Although transthoracic 2D ECHO is the most widely used method for noninvasive analysis of the heart, its geometrical requirements of the ventricle for quantitative estimates result in a suboptimal determination of LV volume and LVEF. ⁸ Recently, three-dimensional (3D) ECHO has been proved to be superior to 2D ECHO and comparable to CMRI for the measurement of LVEF. ⁹ However, the impracticality of the operator process inhibits the use of 3D ECHO in the clinical setting. ¹⁰

Advantages of using ECHO include its wide availability, easy repeatability, versatility, and lack or radiation exposure. ¹¹ ECHO may be more suitable for serial scans because of its cost efficiency. According to standard ASE guidelines, the most accurate calculations of LVEF (with ECHO) should be done with the best method possible, preferably 3D ECHO because it has lower temporal variability than 2D ECHO. ¹¹ Although 3D ECHO has better reproducibility than 2D ECHO, it has limited availability and accuracy for measuring LVEF because it relies on image quality and operator experience. ¹² If 2D ECHO is used, the modified biplane Simpson technique is recommended for the measurement of LVEF. ¹¹ However, ECHO does not detect subclinical myocardial damage (possibly indicative of cardiotoxicity) or small changes in LVEF due to the following factors: LV geometric assumptions, lack of consideration for subtle regional wall motion abnormalities, and variability between measurements. ¹³ Our study shows that CMRI can be used to detect subclinical changes indicative of cardiotoxicity. The inter observer variability for 2D ECHO is 10%. According to the definition of cardiotoxicity (a decrease in LVEF of 10%, or 5% with symptoms of heart failure), the accuracy of detection is compromised. ¹⁴ CMRI is not based on these geometrical assumptions, which may hinder the accuracy of calculated LV volumes, mass, and function.

CMRI is considered to be the gold standard for measuring LVEF, LV volume, and mass. ¹⁵ CMRI should be used for patients with comorbidities or cardiac risk factors or if there are indications for changing treatment regimens. CMRI can also be used reliably to detect subtle areas of ischemia and to evaluate function, viability, and the heart valves. There are various strategies for using CMRI to identify indications of cardiotoxicity. Late gadolinium enhancement can help to detect myocardial fibrosis, cine imaging is used to evaluate cardiac structure and morphology, phase-contrast imaging is used to assess valvular function, and stress-perfusion is used to evaluate for cardiac ischemia. ¹⁶ The use of CMRI for monitoring and evaluating cardiotoxicity in patients treated with DOX allows a single test for a complete heart evaluation, which may lead to long-term cost savings. The use of T1- and T2-weighted images coupled with gadolinium for perfusion at rest and stress and myocardial and pericardial enhancement make CMRI, as recognized by the American College of Cardiology, an appropriate tool to test for chemotherapy-induced cardiotoxicity. ¹⁴ Furthermore, after simultaneously evaluating the LVEF by M-mode ECHO, 2D ECHO, radionuclide ventriculography, and CMRI, Bellenger et al ¹⁷ concluded that EF measurements by various techniques are not interchangeable, as there are variances in volumes and EF between imaging modalities.

Over time, cardiotoxicity has been classified into three categories: acute, early-onset chronic, and late-onset chronic. The most common is early-onset chronic, which develops within 1 year and typically presents as dilated cardiomyopathy, which may lead to overt heart failure. ¹⁸ In a study by Avelar et al, ¹⁹ CMRI was used to evaluate LV remodeling in response to adjuvant chemotherapy in women with low cardiovascular risk and newly diagnosed nonmetastatic breast cancer. Results showed an increase in end-systolic volume and end-diastolic volume that paralleled a declining LVEF and a decrease in RWT with no return to baseline. This eccentric remodeling can cause an increase in global LV wall tension; without compensatory hypertrophy, it may lead to overt heart failure. ²⁰ The use of CMRI to monitor for cardiotoxicity in patients undergoing DOX treatment may allow detection of signs of early-onset chronic heart failure because CMRI can be used to detect signs of eccentric remodeling, myocardial inflammation, fibrosis, and edema. ^{14 21}

Late-onset chronic cardiotoxicity can develop years or decades after the completion of treatment. ¹⁸ CMRI can be used to detect late-onset cardiotoxicity if it was missed with conventional imaging methods. ECHO is suitable for detecting late-stage cardiotoxicity but lacks early prognostic markers that can be detected with CMRI. A relatively large decrease in LVEF detected with ECHO indicates late-stage cardiac disease that may not improve with meaningful intervention and cardio protective measures. However, the use of CMRI allows the detection of signs of subclinical LV dysfunction and early-onset chronic heart failure when interventions may be most beneficial. The early detection of DOX-induced cardiotoxicity is of great interest because it may allow for modification of the dose, administration rate, or schedule; the use of anthracycline analogues of comparable efficacy and fewer cardio toxic properties; or alteration of treatment combinations to reduce cardiotoxicity without stopping lifesaving cancer treatment. ¹⁹

Reverse remodeling therapy in response to DOX-induced cardiotoxicity may stabilize the progression of heart failure, but the underlying damage is irreversible. ¹² In a study by Cardinale et al, ¹⁸ 82% of patients (185 of 226) recovered from cardiotoxicity (which developed from anthracycline treatment of breast cancer or non-Hodgkin lymphoma) within a mean (SD) of 8 (5) months after standard heart failure therapy was initiated (even in asymptomatic patients). This may be due to heart failure therapy being initiated shortly after detection of LVEF reduction. The Prevention of Cardiac Dysfunction During Adjuvant Breast Cancer Therapy trial showed that in patients treated for early breast cancer with adjuvant anthracycline regimens, concomitant candesartan provided protection against early changes in LV function, but metoprolol did not. ¹⁶ Further investigation is needed to determine whether patients benefit from heart failure therapies initiated upon detection of LV dysfunction.

Limitations

Several limitations are associated with our study, which was completed 6 years ago. Since then, the advances in technology and the refinement of machinery for medical diagnostic purposes have been tremendous. Strain imaging is now an essential feature in ECHO and ultrasound technology for detection of abnormalities in cardiac function. One limitation of our study is that it lacks long-term follow-up of patients who underwent DOX treatment. In our study, posttreatment imaging was performed 1 week after the last dose and not thereafter, thus limiting the long-term results. Standard anthracycline therapy is associated with decreases in LVEF 3 weeks and more after completion of the fourth cycle of DOX. ²² Ideally, another series of diagnostic tests should have been performed more than 1 week posttreatment to check for late-onset cardiotoxicity. Another limitation of our study was its homogeneity in gender and diagnosis (83.3% of our patients were women; 77.8% had a diagnosis of breast cancer). This may limit the generalizability of our findings to the entire cancer population. A third limitation is that strain imaging is now an integral part of evaluating early changes in myocardial contractility. However, none of our patients underwent that evaluation at the time this study protocol was completed. A fourth limitation is that our machine lacked 3D technology at that time, so that LVEF could not be measured by that methodology. Further studies should investigate whether CMRI-detected subclinical myocardial changes warrant reducing the DOX dose or instituting cardio protective measures early in the treatment.

Conclusion

Compared with ECHO, CMRI appeared to better predict subclinical cardiac abnormalities suggestive of metabolic/functional changes. CMRI appears to be superior to ECHO as it allows a complete cardiac examination with imaging. With the increasing number of congestive heart failure patients undergoing DOX treatment, it is important to know which imaging modalities are more accurate in detecting early cardiotoxicity in regard to management of patients. Even though our study had a limited number of patients and 3D ECHO strain imaging was not used, our results showed that CMRI is more sensitive in detecting early changes in myocardial function in patients treated with DOX.

It is to be hoped that future studies will focus on the comparative utility and accuracy of available imaging modalities in evaluating which imaging modality is superior and more accurate in detecting early myocardial changes. This in turn could help in adopting countermeasures in limiting cardiotoxicity and management of patients undergoing DOX treatment for various malignancies.

Funding Statement

Funding Mayo Clinic Health System—Franciscan Healthcare Foundation, Grant/Award Number: “09–032”, “13–04”.