Author + information
- Rafael S. Cires, MD,
- Harsh Patel, MD,
- Leonardo J. Tamariz, MD, MPH,
- Joshua M. Hare, MD and
- Juan P. Zambrano, MD⁎ ()
- ↵⁎University of Miami Miller School of Medicine, 1400 NW 12th Avenue, Suite 1179, University of Miami Hospital, Miami, Florida 33136
To the Editor:
An important side effect of several widely used chemotherapeutic agents is cardiotoxicity and heart failure. Animal models demonstrate that granulocyte-colony stimulating factor (G-CSF) may offer cardioprotection against doxorubicin-induced cardiomyopathy (1–3), but human data substantiating this are lacking.
To address this important issue, we conducted a retrospective cohort study investigating the impact of G-CSF on left ventricular ejection fraction (EF) in oncology patients between 2007 and 2008 at the University of Miami Miller School of Medicine. We evaluated patients who received either cardiotoxic or noncardiotoxic chemotherapy and G-CSF to prevent chemotherapy-induced neutropenia. Cardiotoxic drugs included anthracyclines and trastuzumab. G-CSF was administered daily 5 times after each chemotherapy cycle. We included patients who: 1) received G-CSF as a part of the treatment; and 2) had an echocardiogram before and after receiving at least 1 dose of G-CSF. For a control group, we evaluated patients who did not receive G-CSF but who had 2 echocardiograms. Only transthoracic echocardiograms were included. For this analysis, we selected the lowest reported EF determined by visual estimation. In addition, other echocardiographic parameters, notably systolic and diastolic dimensions, were collected.
The primary outcome was the effect of G-CSF on change in left ventricular EF between the baseline and follow-up echocardiogram. Secondary outcomes included the impact of G-CSF on left ventricular EF in patients receiving cardiotoxic chemotherapy compared with those who did not receive cardiotoxic drugs.
Baseline characteristics between the group that received G-CSF and the control group were analyzed using a t test for continuous variables and chi-square for categorical variables. For univariate analysis, we used a paired t test to compare the baseline and follow-up EF between the G-CSF and control groups. Finally, we performed a multivariate linear regression analysis using the change in EF as a dependent variable to examine the impact of G-CSF, adjusting for demographics, use of beta-adrenergic antagonists or angiotensin-converting enzyme inhibitors (ACEI), and cardiotoxic drugs.
We included an interaction term to evaluate the association between G-CSF and the use of cardiotoxic drugs. Because a significant interaction was found, we stratified the analysis by the use of cardiotoxic drugs. Analyses were performed using STATA version 10 (Stata Corp., College Station, Texas), and all significance tests were 2-tailed.
Of a total of 597 patients admitted, 107 patients had 2 echocardiograms. Of these, 70 were in the G-CSF group, and 37 were in the control group. In the G-CSF group, 28 (40%) patients received cardiotoxic drugs, whereas in the control group, 9 (24%) did so. Patients did not differ in age, sex, EF, use of beta-adrenergic antagonists or ACEI, days between echocardiograms, and baseline EF (Table 1).
Comparison of EF between baseline and follow-up echocardiograms revealed that in the G-CSF group, EF was preserved, having a baseline EF of 56.2 ± 12.3%, and a follow-up EF of 55.8 ± 10.1% (p = 0.83). In contrast, in the control group, EF decreased from 59.7 ± 6.4% at baseline to 55.2 ± 9.6% at follow-up (p = 0.02) (Fig. 1A).
Importantly, for patients who received cardiotoxic drugs, the effect of G-CSF was more pronounced. The G-CSF group had a baseline EF of 59 ± 10.5% and a follow-up EF of 57.6 ± 4.9% (p = 0.55); in contrast, in the control group, baseline mean EF was 58.3 ± 6.6%, which decreased to 48.8 ± 11.6% in follow-up (p = 0.04) (Fig. 1B). The EF did not decline in patients not exposed to cardiotoxic drugs.
End-systolic and end-diastolic volumes were not significantly changed in patients receiving G-CSF and cardiotoxic drugs (p > 0.05). The analysis, adjusted for independent variables, including age, sex, use of beta-blockers, ACEI, and cardiotoxic drugs, found a significant beta coefficient of change in EF (beta: 5.22; 95% confidence interval [CI]: 0.56 to 9.8, p = 0.02).
There are several mechanisms underlying a cardioprotective effect of G-CSF in patients receiving chemotherapy. First, G-CSF mobilizes bone marrow stem cells, at least in part by inducing down-regulation of soluble derived factor (SDF)-1 and up-regulation of its receptor, CXCR4 (4). Second, G-CSF may be directly cardioprotective in anthracycline models, reducing myocyte apoptosis by down-regulating the Fas protein (1) or by restoring GATA-4 expression, which in turn enhances myocardial expression of MHC and troponin I. G-CSF receptor signaling can also activate the Extracellular signal Regulated protein Kinase (ERK)/MAPK that is reduced in doxorubicin-induced cardiomyopathy and necessary for the activation of GATA-4 (2). Miyata et al. (3) suggested that the mechanism of doxorubicin-induced cardiomyopathy is fibrosis and autophagy, and G-CSF caused hypertrophy of the cardiomyocytes via Jak/STAT signals, and decreased the fibrosis of the heart by overexpression of metalloproteinase-2 and -9. Finally, Hamamoto et al. (5) demonstrated that G-CSF enhanced proliferation of cardiomyocytes by activating the G-CSF receptors present on the cardiomyocytes.
Currently, there are no randomized controlled trials that examine the efficacy of G-CSF to prevent doxorubicin-induced cardiomyopathy. Our study suggests that the patients who received G-CSF with cardiotoxic chemotherapy experienced a preservation of EF, unlike a control group in which EF fell. This effect cannot be attributed to the chemotherapy, as a similar group not receiving cardiotoxic chemotherapy also experienced a preserved EF during treatment.
This study is limited as it employed echocardiographic evaluation of left ventricular EF as opposed to the more accurate techniques such as multigated acquisition (MUGA) scans or magnetic resonance imaging. Furthermore, the interval between echocardiograms was not uniform across all the patients. Doses of G-CSF were not the same in all patients, and the mobilization of stem cells was not documented.
We conclude that patients receiving cardiotoxic chemotherapy might benefit from the use of G-CSF in order to reduce the chance of developing chemotherapy-induced cardiomyopathy. The present findings strongly support the conduct of randomized controlled trials to confirm this hypothesis.
Please note: Supported by University of Miami Miller School of Medicine. Drs. Hare and Zambrano are supported by National Institutes of Health grants U54 HL-HL081028 (Specialized Center for Cell-based Therapy) and HL094849.
- American College of Cardiology Foundation