Background
Left ventricular global longitudinal systolic strain (GLS) has been shown to be superior to ejection fraction in detecting subclinical dysfunction in patients with cancer and predicting mortality in patients with cardiovascular disease. Cancer-related fatigue is common in the later stages of neoplastic malignancies and may be indicative of nonovert heart failure. The aim of this study was to determine whether reduced strain by echocardiography was associated with all-cause mortality in a cancer cohort.
Methods
In this retrospective study, 120 patients with cancer undergoing or scheduled to undergo chemotherapy and with normal ejection fractions (>50%) underwent assessments of GLS. GLS was derived by averaging all speckle-tracking strain segments of the left ventricle.
Results
Over an average follow-up period of 21.6 ± 13.9 months, 57 of 120 patients died. Univariate predictors of all-cause mortality ( P < .10) were Eastern Cooperative Oncology Group performance status, male sex, hematologic malignancy, β-blocker use, and GLS. Multivariate analysis of all significant univariate variables showed that Eastern Cooperative Oncology Group performance status (hazard ratio, 2.12; 95% confidence interval, 1.54–2.92; P < .001), male sex (hazard ratio, 1.93; 95% confidence interval, 1.14–3.27; P = .014), and GLS (hazard ratio, 0.89; 95% confidence interval, 0.81–0.97; P = .012) were significantly and independently associated with mortality. Stepwise analysis of the multivariate associations showed an increase in the global χ 2 value after adding GLS ( P = .011) to significant clinical variables.
Conclusions
Eastern Cooperative Oncology Group performance status, male sex, and GLS were significantly associated with all-cause mortality in patients with cancer with normal ejection fractions receiving chemotherapy. Adding GLS to significant clinical variables provided incremental prognostic information.
Left ventricular ejection fraction (EF) has long been considered the standard for assessing cardiac function in most cardiac disorders and in monitoring chemotherapy-induced cardiotoxicity. However, EF rarely declines to a measurable degree from some chemotherapy agents (i.e., anthracyclines) until irreversible damage has been done. Strain echocardiography is an emerging technique that has been shown to be clinically useful because of its ability to detect myocardial dysfunction earlier and with greater sensitivity than EF or wall motion analysis. Left ventricular global longitudinal systolic strain (GLS) has been demonstrated to be a significant predictor of chemotherapy-induced cardiotoxicity in patients with preserved EFs. It has also been demonstrated to be a predictor of multiple outcomes, including all-cause mortality in patient populations with a strong prevalence of cardiac disease. This suggests that reduced strain is associated with death due to cardiac disease. However, strain echocardiography has not been investigated for its potential to predict death independently of underlying cardiac disease.
Because cancer-related fatigue (CRF) worsens as cancer progresses, it stands to reason that some factors responsible for CRF may correlate with mortality in patients with cancer. Schünemann et al . noted that CRF may be associated with subclinical cardiac dysfunction. Thus, we investigated whether mortality in patients with cancer without excess cardiac disease was associated with reduced cardiac function on strain echocardiography.
We compared the association of strain by speckle-tracking strain echocardiography with mortality against clinical performance status, which is a validated measure of determining prognosis across multiple cancer types.
Methods
Study Population
This study was approved by the Indiana University Institutional Review Board. We queried the echocardiography database (Fuji Synapse 4.0, Indianapolis, IN) for patients with cancer who underwent echocardiography with strain measurements in relation to chemotherapy or bone marrow transplantation. Per laboratory policy, if possible, strain was calculated at the time of echocardiography for all studies ordered either as standard cardiac monitoring for chemotherapy regimens or for workup in symptomatic patients with concern of chemotherapy-related cardiac toxicity. We identified 138 patients from 2009 to 2011 cared for at the Indiana University Simon Cancer Center who had both two-dimensional echocardiographic images and preexisting speckle-tracking polar plots calculated directly on the instrument consoles. We excluded patients with known systolic dysfunction (14 patients with EFs < 50% and one patient on inotropic therapy) and patients with overall poorly tracking strain studies ( n = 3). The study population thus comprised 120 retrospectively identified patients with histories of recent or imminent chemotherapy, normal EFs, and previously acquired strain images ( Figure 1 ). Nine patients underwent their studies before ever receiving chemotherapy for their active malignancies.
Clinical Data Methods
The clinical history was obtained via chart review in which the type of malignancy, age at time of last visit or death, sex, β-blocker use, Eastern Cooperative Oncology Group performance status (ECOGPS), and various comorbidities were documented from approximately the time the echocardiographic study was performed. Any past or imminent exposures to anthracyclines were recorded. ECOGPS and Karnofsky performance status are widely validated tools for measuring performance status and are often used to predict mortality for multiple cancer types. ECOGPS was taken from oncology clinic and consult notes. See Table 1 for clinical descriptions of the different ECOGPS grades.
| ECOGPS grade ∗ | Description |
|---|---|
| 0 | Fully active, able to carry on all predisease performance without restriction |
| 1 | Restricted in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature (e.g., light housework, office work) |
| 2 | Ambulatory and capable of all self-care but unable to carry out any work activities; up and about >50% of waking hours |
| 3 | Capable of only limited self-care, confined to bed or chair >50% of waking hours |
| 4 | Completely disabled; cannot carry on any self-care; totally confined to bed or chair |
| 5 | Dead |
∗ Reproduced courtesy of Robert Comis, MD, group chair, Eastern Cooperative Oncology Group.
For cases in which only Karnofsky performance status was available, the value was converted to an ECOGPS grade as in equation 1 . Four patients without distinct documentation of performance status had notes that heavily implied no functional limitations and were counted as an ECOGPS of 0.
Rounddown ( 5 − KPS 20 ) = ECOGPS
To determine if cancer was a likely cause of death, all patients who died were analyzed via chart review for the presence of advanced (i.e., potentially rapidly fatal) cancer and for any clinical indications of subsequent cardiotoxicity, defined as an absolute drop in EF of >10% to a level <50% at the time of the last echocardiographic study not readily explained by another acute process.
Echocardiographic Acquisition
Two-dimensional echocardiographic measurements were performed using a Vivid 7 or Vivid Q echograph (GE Medical Systems, Milwaukee, WI). The echocardiographic examinations were performed according to American Society of Echocardiography guidelines. The images were recorded digitally and analyzed offline (Fuji Synapse 4.0).
Strain Acquisition
The strain measurements were obtained by experienced sonographers using a Vivid 7 or Vivid Q echograph at the same time that the standard two-dimensional echocardiographic examinations were performed. To make the speckle-tracking strain calculations, apical four-, three-, and two-chamber views were stored in a cine loop format. Images were obtained between 50 and 70 frames/sec. The analysis was made at the conclusion of the examinations using a propriety software package on the instrument. The software used requires the operator to identify three points on each view (the two mitral annular attachments to the myocardium and also the apex). The software then automatically determines the strain measurements in the six segments of each view. The calculated peak strain values for the entire myocardium can then be displayed as a 17- or 18-segment polar plot, or “bull’s-eye” diagram. The software calculates an average strain value for each view individually then a global average incorporating all three ( Figure 2 ).
GLS measurements were taken directly from the polar plot screen without adjustment. Those echocardiograms in which polar plots could not be constructed and those with grossly poorly tracking data were not considered.
Reproducibility
Because our institution does not have the capability to retrospectively collect strain reproducibility data using the native GE software, we used data we had prospectively collected from a general cohort of patients who underwent echocardiography at roughly the same time as the study patients. Interobserver and intraobserver variability was measured using different cohorts of patients ( n = 36 for both) with various pathologies and with EFs > 50%. These studies had repeat strain analyses performed by the same sonographer or by a different sonographer for the purposes of determining reproducibility.
Outcomes
All-cause mortality was the primary end point. Follow-up was obtained by using the patient’s hospital and clinic charts, online obituary records, or the Social Security Death Index.
Statistical Analysis
Analysis was carried out using SPSS version 16 (SPSS, Inc, Chicago, IL). To simplify analysis and communication, all strain values were multiplied by −1, such that greater positive strain values refer to greater myocardial contraction and vice versa. The χ 2 test was used to determine significant differences between two groups of categorical or continuous variables.
Univariate binary logistic regression was performed to determine if variables that might affect strain, such as age, sex, time from chemotherapy to echocardiography, prior thoracic radiation, prior anthracycline use, hypertension, coronary artery disease, and diabetes mellitus, were independently associated with abnormal GLS (defined as <18).
Survival was assessed using Kaplan-Meier analysis, and significant differences in survival were assessed with the log-rank test. Cox proportional hazards analysis was used to determine significant univariate factors associated with all-cause mortality. Two different types of multivariate analyses were performed. The initial analysis was conducted with a forward selection method, including all clinical and echocardiographic variables with P values < .10 on univariate analysis. Variables with P values < .05 on multivariate analysis were considered significantly and independently associated with mortality. A second stepwise analysis was performed using the clinical variables that were independently associated with mortality in the first step and strain echocardiographic variables in the second step to determine if strain variables added incremental prognostic value. Reproducibility was reported as the average percentage measurement difference/mean.
Results
Table 2 contains the clinical and echocardiographic information for the population. Table 3 provides a breakdown of the malignancies. There were 122 malignancies, as two patients had two simultaneous primary cancers. Only four patients (3%) had significant coronary disease. One of the surviving patients had minimal records available after leaving the country and thus was the only survivor with <14 months of follow-up. Aside from one patient with mild to moderate valvular disease, all patients had no more than mild valvular disease.
| Clinical/echocardiographic variable | Average or prevalence |
|---|---|
| Age (y) | 53.1 ± 12.4 |
| Months from echocardiography to follow-up/death | 21.6 ± 14.0 |
| Months from chemotherapy to echocardiography | 29.3 ± 39.3 |
| Male sex | 43 (36%) |
| ECOGPS | 0.48 ± 0.72 |
| Hematologic malignancy | 52 (43%) |
| History of anthracycline use | 69 (58%) |
| History of thoracic radiation | 31 (26%) |
| β-blocker use | 24 (20%) |
| ACE inhibitor use | 26 (22%) |
| Statin use | 20 (17%) |
| Salicylate use | 15 (13%) |
| Diabetes mellitus | 18 (15%) |
| Coronary artery disease | 4 (3%) |
| Hypertension | 42 (35%) |
| Tobacco use | 37 (37%) |
| Hyperlipidemia | 29 (24%) |
| Ejection fraction | 61.1 ± 5.0 |
| LVEDV (mL) | 127.3 ± 41.9 |
| IVS diastolic thickness (cm) | 1.04 ± 0.15 |
| LVPW diastolic thickness (cm) | 1.03 ± 0.12 |
| Left atrial volume index (mL/m 2 ) | 17.8 ± 6.7 |
| Mitral E/septal e′ ∗ | 9.39 ± 2.76 |
∗ Mitral E/septal e′ data were interpretable in 108 of 120 patients.
| Cancer type | Died (%) |
|---|---|
| ALL | 5/8 (63) |
| AML | 14/19 (74) |
| CLL | 1/2 (50) |
| CML | 0/1 (0) |
| Hodgkin’s lymphoma | 0/5 (0) |
| Non-Hodgkin’s lymphoma | 8/12 (67) |
| Myeloma ∗ | 2/6 (33) |
| Breast cancer | 8/42 (19) |
| Carcinoid cancer | 3/3 (100) |
| Colorectal cancer | 2/4 (50) |
| Endometrial cancer | 0/1 (0) |
| Esophageal cancer | 2/2 (100) |
| Head and neck cancers | 3/3 (100) |
| Lung cancer | 2/2 (100) |
| Ovarian cancer | 1/3 (33) |
| Pancreatic cancer | 3/3 (100) |
| Renal cancer | 5/6 (83) |
| All cancer | 59/122 (48) |
Over 21.6 ± 13.9 months of follow-up, there were 57 deaths (48%). One patient died of heart failure thought to be due primarily to chemotherapy-induced cardiotoxicity, and one patient died abruptly without signs of cardiotoxicity or an active malignancy at last follow-up. The remaining 55 patients had advanced cancer at last follow-up before death.
Binary logistic regression showed that among the risk factors of age, sex, time from chemotherapy to echocardiography, prior thoracic radiation, prior anthracycline use, hypertension, coronary artery disease, and diabetes mellitus, only hypertension ( P = .064) and coronary artery disease ( P = 0.009) had P values < .10 associated with decreased GLS.
Table 4 shows the results of univariate and multivariate analyses of all clinical and echocardiographic variables. In this set of variables, the significant univariate factors associated with mortality ( P < .10) were male sex, ECOGPS, hematologic malignancy, β-blocker use, and GLS. Neither prior anthracycline use nor prior thoracic radiation was associated with mortality. On multivariate analysis of the significant univariate associated factors, male sex, ECOGPS, and GLS were independently associated with mortality.
