Background
Anthracycline-induced symptomatic heart failure is often irreversible, underlining the usefulness of pretreatment risk assessment. Global longitudinal strain (GLS) before or after chemotherapy is associated with a later decrease in left ventricular ejection fraction (LVEF); however, whether prechemotherapy GLS is associated with symptomatic heart failure and cardiac death in patients treated with anthracyclines is unknown.
Methods
Patients with hematologic cancers treated with anthracyclines who underwent prechemotherapy echocardiography between November 2006 and June 2011 were retrospectively recruited. Basic demographic data, end-diastolic and end-systolic left ventricular volumes, LVEF, and GLS were measured. Clinical cardiac events (CEs) were defined as cardiac death or symptomatic heart failure. The association of prechemotherapy parameters with CEs was analyzed using proportional hazard analysis.
Results
Over a median follow-up period of 1,593 days (range, 13–2,891 days) after the start of chemotherapy, 28 of 450 patients (6%) experienced CEs. Prechemotherapy LVEF and GLS were lower in patients with CEs compared with those without CEs (58 ± 10% vs 62 ± 7% [ P = .005] and −15.0 ± 2.8% vs −19.7 ± 2.7% [ P < .0001], respectively). Diabetes (hazard ratio [HR], 7.06; P < .0001), hypertension (HR, 2.22; P = .04), LVEF (HR, 0.93; P = .005), and GLS (HR, 1.47; P < .0001) were associated with CEs. After controlling for clinical variables, prechemotherapy GLS remained independently associated with CEs ( P < .0001). GLS less than the absolute value of −17.5% was found in 105 patients (23%) and was associated with a sixfold increase in CEs ( P < .0001).
Conclusions
Prechemotherapy GLS is an effective tool to stratify patients at high risk for CEs after anthracycline therapy and may help tailor treatments to decrease anthracycline-induced cardiotoxicity.
Highlights
- •
Six percent of patients with hematologic malignancies and anthracycline treatment developed CEs.
- •
Prechemotherapy LVEF and GLS were associated with subsequent CEs.
- •
GLS less than the absolute value of −17.5% was found in 23% of patients and associated with a sixfold increase in CEs.
- •
Baseline GLS improved on clinical variables and LVEF to identify later CEs.
More than 14 million children and adults with histories of cancer were alive in 2014 in the United States; of these, it is estimated that >2 million were at risk for cancer treatment–induced cardiotoxicity. A majority of these patients were exposed to anthracyclines, a mainstay of treatment for hematologic malignancies and breast cancer. Anthracycline treatment is associated with a markedly increased risk for both clinical and subclinical cardiotoxicity that increases with the cumulative dose of treatment. The overall occurrence of symptomatic heart failure in patients treated with anthracyclines is estimated to be between 2% and 5% and may be higher in patients with hematologic malignancies.
Once anthracycline-induced heart failure is symptomatic, its prognosis is poor, with survival of <50% at 1 year.
Left ventricular (LV) ejection fraction (LVEF) is routinely used to monitor cardiac function in patients treated with anthracyclines. Although LVEF predicts the occurrence of symptomatic heart failure, some studies suggest that LVEF has limited sensitivity for detecting anthracycline-induced cardiomyocyte damage and subtle abnormalities in cardiac function. Even when the impairment in LVEF is diagnosed immediately after anthracycline treatment, 36% of patients fail to recover normal LV function despite therapeutic intervention.
In a study by Tan et al ., therapy with anthracyclines, taxanes, and trastuzumab induced alterations in LV volumes and function, which, though asymptomatic, persisted >2 years after the cessation of trastuzumab.
Because the decrease in LVEF may be an irreversible event, more sensitive markers of subclinical LV dysfunction or myocardial injury have been investigated in the detection of chemotherapy-induced cardiotoxicity. Insidious remodeling with LV dilation usually precedes a decline in ejection, and small increases in LV size are associated with a notable, independent increase in mortality in patients with coronary artery disease, underlining the potential value of measuring LV volumes.
In patients treated with anthracyclines, myocardial deformation indices (strain and strain rate) are able to detect subtle, early changes in LV systolic dysfunction before a decrease in LVEF parameters. Importantly, these early decreases in deformation indices (in particular LV peak global longitudinal strain [GLS]), present when LVEF is still preserved, have been demonstrated to predict subsequent LVEF decreases in women with breast cancer treated with anthracyclines and trastuzumab. Recently, GLS measured after chemotherapy was found to be associated with all-cause mortality in patients with cancer. This interesting finding raises the possibility that some of the subclinical LV dysfunction observed in patients with cancer may reflect the cancer burden; alternatively, LV dysfunction may contribute to cancer mortality. However, the value of GLS in the prediction of symptomatic heart failure and cardiac death in patients treated with anthracyclines is still underinvestigated. The identification of patients at high risk for heart failure and cardiac death would be especially useful in patients at high risk for clinical events, such as patients with hematologic malignancies.
The purpose of the present study was to evaluate the value of LV volumes, LVEF, and LV peak GLS measured before chemotherapy in identifying the development of symptomatic heart failure and cardiac death in patients with hematologic malignancies treated with anthracyclines.
Methods
Study Subjects
Patients with hematologic cancers treated with anthracyclines at the Massachusetts General Hospital who underwent prechemotherapy echocardiography between November 2006 and June 2011 were retrospectively recruited. The institutional review board of Massachusetts General Hospital approved the study protocol.
Anthracycline Regimens
At our institution, patients with acute myeloid leukemia are usually treated with a “7 + 3” regimen that includes a 7-day infusion of cytarabine with 3 consecutive days of anthracyclines, generally 50 mg/m 2 doxorubicin equivalent dose per day for 3 days. This dose of 150 mg/m 2 over 3 days is much higher in a short period of time than the usual 60 mg/m 2 given to patients with breast cancer (50 mg/m 2 for non-Hodgkin’s lymphoma) every 3 to 4 weeks.
Echocardiography and Strain Analysis
Transthoracic echocardiography was performed using commercially available equipment (Vivid 7 or E9 [GE Medical Systems, Milwaukee, WI] or iE33 [Philips Medical Systems, Andover, MA]). Cine loops from three standard apical views (apical four-chamber, apical two-chamber, and apical three-chamber) were recorded using grayscale harmonic imaging and saved in Digital Imaging and Communications in Medicine format. LV volumes and LVEF were measured by a single observer blinded to the clinical outcome using the Simpson biplane method.
Endocardial longitudinal strain was quantified by an offline vendor-independent analysis program (2D Cardiac Performance Analysis; TomTec Imaging Systems, Munich, Germany) on the Digital Imaging and Communications in Medicine stored images (30 frames/sec). The endocardial borders were traced from the three apical views by an observer blinded to the clinical outcome. GLS and the average peak LV longitudinal systolic strain from the three apical windows were measured. GLS is a late systolic strain, calculated by measuring the entire endocardial line length at the end-diastole and end-systole ( L 1 − L 0 / L 0 × 100%) in each view and averaging the results from the three views. In contrast, the average peak longitudinal strain value is a peak systolic strain derived by averaging the peak systolic strain values from the 18 cardiac segments.
Outcomes
The end point of the study was the occurrence of cardiac events (CEs), defined as symptomatic heart failure or cardiac death. An additional analysis limiting the end point to heart failure was also done and is provided in the Supplemental Material . Heart failure diagnosis was confirmed from the patients’ records by a cardiologist blinded to the baseline echocardiographic analysis in agreement with the definition of cardiovascular events in clinical trials by the American College of Cardiology and American Heart Association. Severe anemia, pneumonia, septic shock, and multiple organ failure or death complicating septic shock and renal failure were not counted as CEs. The circumstances of death were carefully studied. All patients who were transferred to hospice or considered for palliative measures and administered morphine for pain relief and comfort and then died were considered as having noncardiac death. A patient with decompensated heart failure who subsequently died was considered as having a CE.
Statistical Analysis
Continuous data are presented as mean ± SD or as median and range and categorical variables as percentages. Differences in continuous data between the patients with or without CEs were compared using Student’s t test or the Wilcoxon rank comparison for nonparametric variables. Categorical variables were compared using the χ 2 test. Time to first CE was defined as the number of days between the start of anthracycline therapy and the date of first CE and was considered for survival analysis. Patients who had not experienced CEs as of their last visit date were censored at this date. Cox proportional hazard analysis was used to determine significant clinical and echocardiographic identifiers of time to CE. To perform a comparison of LVEF and GLS, we performed a proportional hazard analysis using these two terms while controlling for the clinical parameter with the highest hazard ratio (HR) (diabetes) in a multivariate analysis. The incremental value of LVEF and GLS compared with a clinical model (age, cancer type, coronary artery disease, congestive heart failure, hypertension, diabetes mellitus, β-blocker use, angiotensin-converting enzyme inhibitor use, and statin use) in identifying subsequent CEs was tested with χ 2 tests. Receiver operating characteristic curve analysis was used to determine the relationship of GLS value to the occurrence of CE. Survival without CE as a function of GLS was analyzed using Kaplan-Meier analysis. All analyses were performed using a standard statistical software program (IBM SPSS Statistics for Windows version 21.0; IBM, Armonk, NY). P values < .05 were considered to indicate statistical significance.
Results
Patient Characteristics and Follow-Up
Four hundred seventy-six patients with hematologic malignancies treated with anthracyclines were eligible. Of those, 26 patients (6%) were excluded because of suboptimal image quality. Over a median follow-up period of 1,593 days (4.2 years), 28 patients (6%) developed CEs. The median time to development of a CE was 523 days (range, 11–1,843 days). Of the 28 patients with CEs, 24 (86%) developed symptomatic heart failure and eight (25%) died of cardiac causes (four patients had both outcomes). Of the patients who died of cardiac-related causes, two patients had acute myocardial infarctions, and six patients had severely depressed LVEFs before death with ( n = 4) or without ( n = 2) heart failure symptoms. One hundred twelve patients died of noncardiac causes (malignancy or complications linked to malignancy).
The comparison of the clinical characteristics of patients with and those without CEs is presented in Table 1 . Patients who developed CEs were older and had more risk factors for, histories of, or treatment for cardiovascular disease than patients who did not develop CEs. Patients with leukemia developed more CEs than patients with lymphoma (13% vs 5%, P = .015), but their cardiac risk profile was similar to that of patients with lymphoma ( Supplemental Table 1 ). The additional analysis results limiting the end point to heart failure are provided in Supplemental Tables 2-4 .
| Variable | Total ( n = 450) | No CEs ( n = 422) | CEs ( n = 28) | P |
|---|---|---|---|---|
| Age (y) | 59 ± 18 | 58 ± 18 | 65 ± 18 | .038 |
| Men | 267 (59%) | 253 (60%) | 14 (50%) | .300 |
| Cancer type | ||||
| Leukemia | 80 (18%) | 70 (17%) | 10 (36%) | .010 |
| Lymphoma | 370 (88%) | 352 (87%) | 18 (64%) | |
| Follow-up period (d) | 1,593 (13–2,891) | 1,605 (13–2,891) | 1,337 (90–2,688) | .000 |
| Dose of anthracycline (mg/m 2 doxorubicin equivalent) | 247 (8–670) | 250 (8–670) | 213 (34–583) | .23 |
| Radiotherapy | 117 (26%) | 111 (26%) | 6 (21%) | .660 |
| Cardiac risk factors | ||||
| BP > 140/90 mm Hg | 90 (20%) | 80 (19%) | 10 (36%) | .032 |
| Hypercholesterolemia | 39 (9%) | 35 (8%) | 4 (14%) | .280 |
| Diabetes mellitus | 53 (12%) | 40 (10%) | 13 (46%) | .000 |
| Smoking | 16 (4%) | 15 (4%) | 1 (4%) | .990 |
| Prior coronary artery disease | 29 (6%) | 24 (6%) | 5 (18%) | .020 |
| Prior congestive heart failure | 17 (4%) | 13 (3%) | 4 (14%) | .016 |
| Cardiovascular therapy | ||||
| ACE inhibitors/ARBs | 17 (4%) | 13 (3%) | 4 (14%) | .016 |
| BBs | 57 (13%) | 50 (12%) | 7 (25%) | .050 |
| Statins | 44 (10%) | 36 (8%) | 8 (29%) | .003 |
Baseline Echocardiography
Baseline echocardiographic findings are listed in Table 2 . The mean LVEF of the patients before chemotherapy was 62 ± 7% in those who did not develop CEs and 58 ± 10% in those who did ( P = .005). Of 420 patients with normal LVEFs (>53% for women and >51% for men), 24 patients (6%) developed CEs, and of the 30 patients with impaired LVEFs, four (13%) had CEs ( P = .09).
| Variable | Total ( n = 450) | No CEs ( n = 422) | CEs ( n = 28) | P |
|---|---|---|---|---|
| Baseline LVEF (%) | 62 ± 8 | 62 ± 7 | 58 ± 10 | .005 |
| LV EDV (mL) | 114 ± 32 | 114 ± 33 | 111 ± 32 | .640 |
| LV EDVI (mL/m 2 ) | 59 ± 14 | 59 ± 14 | 60 ± 12 | .890 |
| LV ESV (mL) | 45 ± 16 | 45 ± 16 | 48 ± 20 | .340 |
| LV ESVI (mL/m 2 ) | 24 ± 7 | 23 ± 7 | 26 ± 13 | .072 |
| GLS (%) | −19.4 ± 2.9 | −19.7 ± 2.7 | −15.0 ± 2.8 | .000 |
| Average segmental longitudinal strain (%) | −19.9 ± 3.1 | −20.3 ± 2.4 | −13.2 ± 2.3 | .000 |
Both GLS and average segmental strain were lower in patients who develop CEs compared with those who did not (−15.0 ± 2.8% vs −19.7 ± 2.7% and −13.2 ± 2.3% vs −20.3 ± 2.4%, respectively, P < .0001 for both).
Identifiers of CEs
The identifiers of CEs are detailed in Table 3 . Age, cancer type, previous coronary artery disease or cardiomyopathy, hypertension, diabetes mellitus, and cardiovascular treatment were associated with CEs. LVEF was associated with later CE occurrence (HR, 0.93; P = .005), but LV volumes or indexed volumes were not. Of note, LVEF was not associated with CEs in patients with normal LVEFs on echocardiography before chemotherapy (HR, 0.97; P = .54).
| Variable | HR | 95% CI | P |
|---|---|---|---|
| Age (y) | 1.03 | 1.006–1.056 | .013 |
| Gender (female) | 1.47 | 0.70–3.10 | .300 |
| Cancer type (leukemia) | 1.87 | 1.26–2.75 | .002 |
| Hypertension | 2.22 | 1.02–4.80 | .040 |
| Hyperlipidemia | 1.80 | 0.60–5.20 | .279 |
| Diabetes mellitus | 7.06 | 3.36–14.85 | .000 |
| Smoking | 0.87 | 0.12–6.44 | .890 |
| Prior coronary artery disease | 3.61 | 1.37–9.51 | .009 |
| Prior cardiomyopathy | 5.21 | 1.8–15.06 | .002 |
| Dose of anthracyclines | 0.99 | 0.99–1.002 | .230 |
| ACE inhibitors/ARBs | 5.42 | 1.88–15.66 | .002 |
| Statins | 5.45 | 2.39–12.46 | .000 |
| β-blockers | 2.80 | 1.19–6.63 | .018 |
| Baseline LVEF (%) | 0.93 | 0.89–0.98 | .005 |
| LV EDV | 0.99 | 0.98–1.01 | .580 |
| LV EDVI | 1.00 | 0.97–1.03 | .910 |
| LV ESV | 1.01 | 0.98–1.04 | .370 |
| LV ESVI | 1.05 | 0.99–0.11 | .058 |
| GLS | 1.47 | 1.35–1.59 | .000 |
| Average segmental longitudinal strain | 1.85 | 1.65–2.08 | .000 |
GLS and average longitudinal strain measured before the initiation of anthracycline treatment were both associated with the later development of cardiotoxicity (HR, 1.47 and 1.85, respectively; P < .0001 for both). GLS was also associated with CEs in patients with normal LVEFs (HR, 1.45; P < .0001).
As an example, on the basis of the receiver operating characteristic curve ( Figure 1 ), the area under the curve was 0.89 (95% CI, 0.84–0.95; P < .0001), and a cutoff GLS value of less than the absolute value of −17.5% before anthracycline treatment would have correctly identified 24 of 28 of the study patients (86%) who developed CEs and 341 of 422 patients (81%) who did not. The CE-free survival difference between patients with GLS < −17.5% or GLS ≥ −17.5% is shown in Figure 2 .
In multivariate analysis including diabetes mellitus, LVEF, and GLS, both diabetes mellitus (HR, 3.11; P < .011) and GLS (HR, 1.36; P < .0001) remained associated with CEs. The incremental value of LVEF and GLS to a model including all clinical variables significant by univariate analysis is presented in Figure 3 .
