Introduction
Recent advancements in cancer treatment have remarkably improved the overall survival of patients with cancer [
1]. However, chemotherapy-induced cardiotoxicity has been related to unfavorable prognosis in cancer survivors [
2,
3], which has put oncologists into a therapeutic dilemma to maintain a balance between the antitumor efficacy and cardiac injury.
Heterogeneous factors contributing to chemotherapy-induced cardiotoxicity are cumulative anthracycline doses [
4], adjunct therapy with trastuzumab [
5,
6], prior radiotherapy [
7], preexisting cardiovascular risks (e.g., diabetes mellitus, hypertension, and dyslipidemia) [
8,
9], and age [
10]. Multiple mechanisms are associated with anthracycline-induced cardiotoxicity; however, the most extensively accepted mechanism seems to be the generation of oxygen free radicals, which cause cardiomyocyte damage and apoptosis [
11].
Lately, various efforts have been made to attenuate chemotherapy-induced cardiotoxicity, including early detection of subclinical toxicity, development of derivatives [
12], and the prophylactic use of cardioprotectants. The effects of cardioprotective beta-blockers have been tested in multiple animal experiments and clinically randomized controlled trials (RCTs). The primary mechanism associated with beta-blockers is reduction of oxidative stress and apoptosis [
13,
14]. Carvedilol is a nonselective β and α1 adrenergic antagonist and has been proven to be cardioprotective not only through the intrinsic β-blocking effect (attenuating the sympathetic activity and cardiac remodeling) [
15] but also the additional antioxidant and anti-inflammation properties compared with other beta-blockers [
16,
17]. In particular, carvedilol can protect the myocardium without interfering with the antineoplastic efficacy of anthracyclines [
18]. Although several beta-blockers have been tested through clinical RCTs, the number of enrolled patients was limited, and the results remained debatable. This study aims to evaluate the prophylactic effects of beta-blockers, especially carvedilol, on chemotherapy-induced cardiotoxicity by systematic review and meta-analysis.
Methods
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [
19]. This study was prospectively registered with the PROSPERO database of systematic reviews (registration number: CRD42018086747).
Search strategy
We systematically searched MEDLINE (PubMed), Embase (Ovid SP), Cochrane CENTRAL (Ovid SP), and the World Health Organization International Clinical Trials Registry Platform, as well as two Chinese medical databases, CNKI and WANFANG, until December 2017. In addition, a follow-up search was conducted for ongoing trials. The search strategy comprised a combination of Medical Subject Headings terms and free-text terms, primarily including “carvedilol,” “beta-blockers,” “anthracycline,” “chemotherapy,” “cardiotoxicity,” “cardiomyopathy,” and “heart failure.” Furthermore, we reviewed the reference lists of relevant studies and review articles. Of note, we limited the search to human beings without any language restriction.
Inclusion and exclusion criteria
The inclusion criteria were as follows: (a) patients with cancer aged > 18 years due to receive chemotherapy; (b) beta-blockers used against the cardiotoxic effect of chemotherapy in a prophylactic setting (started before chemotherapy); (c) prospective RCT study design; and (d) left ventricular ejection fraction (LVEF) assessed both at the baseline and post-chemotherapy. We excluded studies in which beta-blockers were used concomitantly with other cardioprotective agents, such as angiotensin-converting enzyme inhibitors.
Data extraction and study quality assessment
Data extraction and study quality assessment were performed per the PRISMA statement [
19]. For systematic review, the following data and information were extracted: study design, age, gender, sample size, type of cancer, cumulative anthracycline doses, beta-blockers, follow-up time, baseline LVEF, and outcomes. The primary outcome was LVEF after chemotherapy, and secondary outcomes were all-cause mortality, clinically overt cardiotoxicity, and other cardiac measurements (e.g., E/A ratio and LV chamber size).
The quality of studies was assessed per the Cochrane risk of bias tool [
20]. The adequacy of blinding was ascertained by whether cardiac measurements were assessed by a third person blinded to patients’ information. Any disagreement was resolved by consensus. For key information that was not reported in the articles, we contacted the authors by e-mails.
Statistical analysis
We performed the statistical analyses using Review Manager (V.5.3, Cochrane) and Stata 13.0 (Stata Corp LP, College Station, TX). Heterogeneity among studies was assessed using the
Q test and
I2 statistics (
I2 > 50% suggested substantial heterogeneity). In the absence of considerable heterogeneity, the fixed effects model was selected to obtain the mean difference (MD) and 95% confidence interval (CI); else, the random effects model was used. For comparing the risk of adverse events and mortality in patients and control subjects, we used Peto’s one-step odds ratio method, which is the least biased and most potent method when event rates were low, per the Cochrane handbook [
21]. We considered
P < 0.05 as statistically significant. In addition, the Galbraith plot and sensitivity analysis were conducted to determine the primary source of heterogeneity and ascertain the stability of the statistical results. Publication bias was assessed by Egger’s test. Although we planned to perform subgroup analysis, trials were insufficient.
Discussion
We systematically reviewed ten prospective RCTs for four beta-blockers, carvedilol (six studies; 545 participants), nebivolol (one study; 45 participants), metoprolol (two studies; 146 participants), and bisoprolol (one study; 61 participants), and used meta-analysis to quantitatively assess, at least, three studies for the following outcomes: LVEF, LV end-diastolic diameter, E/A ratio, clinically overt cardiac dysfunction, and all-cause mortality. This review could update the information on the effects of the prophylactic use of carvedilol for cardioprotection in the setting of chemotherapy-mediated cardiotoxicity and help with a comprehensive therapy regimen design for cancer survivors.
Chemotherapy-induced cardiotoxicity incorporates multiple cardiovascular manifestations and is mostly characterized by an asymptomatic decline in the LVEF [
30]. Our initial outcome corroborated this and exhibited a considerable small LVEF decrease in the placebo group than the carvedilol group. Per the Galbraith plot, the substantial heterogeneity was primarily attributed to two studies, N. K. et al. and M. N. et al. By excluding the study by N. K. et al., the heterogeneity declined to 67%, but the outcome had been reversed (MD, 1.74; 95% CI, − 0.18 to 3.66;
P = 0.08). Several things could be accountable for this discrepancy. First, patients in this study received relatively higher doses of anthracyclines and, thus, were supposed to exhibit a higher incidence of cardiotoxicity considering the dose-dependent feature of anthracycline [
10]. And this could make the effect of using a cardioprotective agent more tangible. However, the contemporary chemotherapy protocol had been reformed, and the anthracycline dosage was much lower in the latest studies. Second, the SDs of the LVEF were not provided initially but extracted from another review [
31], which could have affected the results. Third, the sample size was small (25 patients/group). The pooled result of the remaining studies demonstrated that carvedilol exerted no impact on the early asymptomatic LVEF reduction (MD, 1.74; 95% CI, − 0.18 to 3.66;
P = 0.08;
I2 = 67%). Regarding the heterogeneity from the study by M. N. et al., one possible reason could be that, opposed to most included trials, the average daily dose of carvedilol was much lower (6.7 vs 12.5 mg, q.d.). Although the LVEF was similar at the baseline in the two groups in this study, troponin I was considerably higher in the control group than the carvedilol group, suggesting that the control group could have more impaired myocardium than the carvedilol group at the baseline. To the best of our knowledge, the study by Avila [
28] was the latest RCT on this subject with the largest sample size and the most credible results thus far. And our results were more in agreement with this study. In the trial by Elitok [
26], although the LVEF remained unchanged in both groups, carvedilol did exhibit a protective effect by preventing the decrease in LV strains, as strain imaging is more sensitive than the LVEF when detecting the subclinical cardiac damage [
32].
Our pooled results revealed a lower incidence of overt cardiotoxicity in patients receiving carvedilol than those receiving placebo, strengthening the conclusion that carvedilol could help prevent the deterioration of the cardiac function when previous studies only reported the numerical difference. Admittedly, different cutoff values were adopted among the included studies, but considering that the clinical endpoints of chemotherapy-induced cardiotoxicity remained debatable and the patient population is highly heterogeneous in the practical oncological setting, a safe conclusion could be that the prophylactic use of carvedilol could decrease the incidence of overt cardiac dysfunction in patients with cancer. However, the ultimate effect awaits confirmation by extensive, well-designed clinical trials.
Our analysis did not observe a statistically significant difference between the two groups regarding the 6-month all-cause mortality. Conversely, another meta-analysis [
33], which pooled data from five RCTs with 317 patients, reported that beta-blockers correlated with the reduction of all-cause mortality (Peto OR, 0.33; 95% CI, 0.12–0.92;
P = 0.03) in patients undergoing chemotherapy. However, as the full text of this meta-analysis was unavailable, more detailed documents would be needed to ascertain whether this discrepancy arose because of the types of beta-blockers or the included population. As these studies did not offer specific reasons for deaths, cardiovascular mortality could not be separately assessed in both our reviews.
Pooling LV end-diastolic diameters revealed that carvedilol use could inhibit the enlargement of the LV chamber, suggesting that carvedilol could affect LV remodeling in this setting. The PRADA trial [
34] tested whether candesartan and/or metoprolol could prevent anthracycline therapy–associated interstitial fibrosis by T1 mapping and ECV, which was proved to correlate favorably with myocardial biopsy measurements [
35].
Regarding the diastolic function, the study by Avila reported a lower incidence of diastolic dysfunction in the carvedilol group than the placebo group (37.2% vs 28.5%; P = 0.039). In fact, most studies reported that the E/A ratio reflects the diastolic function in patients. Our aggregated outcome of the E/A ratio did not exhibit a statistical difference in the two groups.
Pooling of the results was impossible for the other three beta-blockers (nebivolol, metoprolol, and bisoprolol) because of the insufficient number of studies. The study by Kaya et al. [
36] using nebivolol and that by Pituskin [
22] using bisoprolol reported a cardioprotective effect, but not for the two studies using metoprolol [
23,
37]; results were summarized in Table
3. The power of the evidence was considered relatively low either because of the small sample sizes or because of inadequate reporting. The results about the effect of metoprolol were consistent in the two studies; both invalidated the use of metoprolol in this scenario. An earlier meta-analysis [
38], which pooled results of the three different beta-blockers (nebivolol, metoprolol, and bisoprolol), reported an insignificant difference in the LVEF change (WMD, 3.05; 95% CI, − 7.22 to 1.12;
P = 0.15) between groups, but a lower heart failure incidence (OR, 0.33; 95% CI, 0.14–0.80;
P = 0.01) in the beta-blocker group, which was to some extent consistent with our outcomes.
Table 3
Characteristics of studies using other beta-blocker
| Nebivolol | BC | 27 (100%) | 18 (100%) | 51.4 (9.4) | 50.5 (11.1) | EPI 361 ADM 257 | EPI 348 ADM 235 | 65 (4.8) | 66.6 (5.5) | 63.8 (3.9) | 57.5 (5.6) | 6 m | No HF, hospitalization or death† |
| Bisoprolol | BC | 31 (100%) | 30 (100%) | 53 (10) | 51 (7) | 592 | 643 | 62 (4) | 61 (5) | 61 (4) | 56 (4) | 52 w | CTRCD (1 vs 6); interrupted TRZ (3 vs 9)* |
| Metoprolol | Lymphoma | 42 (48%) | 40 (47%) | 51.0 (18) | 49.1 (19.4) | 387.5 | 386.4 | 67.7 (5.0) | 67.6 (7.1) | 63.3 (7.4) | 66.6 (6.7) | 12 m | No death or interruption of CT† |
| Metoprolol | BC | 32 (100%) | 32 (100%) | 50.5 (9.1) | 50.8 (9.2) | NA | NA | 63.5 (5.0) | 63.6 (4.1) | 60.8 (4.7) | 63.6 (4.1) | 10–61 w | No symptomatic HF† |