Introduction
Anthracyclines (ANT) are the most used anti-cancer drugs in the world [1]. ANT are antibiotic antitumor drugs, a class that consists of drugs like Doxorubicin, Epirubicin, and Idarubicin, and are widely used against leukemia, lymphoma, uterine, ovarian, and breast malignancies [2]. ANT has a complex integrated mechanism of action and works as a cytotoxic drug, inducing apoptosis, necrosis, mitochondrial dysfunction, and other cytotoxic effects [3].
However, the use of ANT is associated with severe side effects mainly due to its mechanism of action. Cardiotoxicity is the long-term and most important side effect of ANT. Due to widespread use, efficacy, and cumulative dose-cardiotoxicity of ANT, several studies have discussed its mechanism, course, and prevention [4].
On retrospective analysis, as high as 4.7 %, 26 %, and 48 % incidence of Congestive Heart Failure(CHF) was found associated with doses of ANT 400, 500, and 700 mg/m2, respectively [5]. A meta-analysis reported a significant decline in clinical and subclinical Left Ventricular Ejection Fraction (LVEF) associated with ANT [6]. ANT-induced cardiotoxicity has a high incidence in young age, females, high doses of ANT, and fat consumption [4].
Several general cardioprotective strategies have been implicated to prevent ANT-induced cardiotoxicity. These include reducing risk factors, changing drug formulation and coating, and using pharmaceutical interventions like statins, antihypertensives, beta blockers, and dexrazoxane [7]. However, several studies have been conducted to compare the efficacy of beta blockers in this regard.
As a non-selective adrenergic blocker and a potent antioxidant, carvedilol acts as a better cardioprotective agent [8].
Many trials have been conducted, and meta-analyses have been done to shed light on its efficacy; however, there is a lack of comprehensive literature on this topic, and no single study has elaborated on the direct effect of carvedilol in preventing ANT-induced cardiotoxicity. This meta-analysis and systematic review aim to compare the efficacy of carvedilol to placebo and patients taking no drugs in the prevention of cardiotoxicity in patients treated with ANT.
Methodology
Protocol and registration
This systematic review and meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (Supplemental Table 1)[9] and prospectively registered on PROSPERO under the registration number CRD420251021778 [10]. As the data were synthesized from previously published studies, ethical approval was not required.
Eligibility Criteria
Observational studies, randomized and non-randomized studies that compared the efficacy and safety of carvedilol against placebo in patients with ANT-induced cardiotoxicity were eligible for inclusion. We included studies if they included human participants of all ages and genders, and reported at least one relevant outcome (e.g., LVEF, LVSD, LVsD, LVdD, mortality, biomarkers, echocardiographic parameters). Studies were excluded if they 1) were animal studies, 2) did not include carvedilol as the pharmacological intervention, or 3) if relevant outcome data were not reported.
Additionally, case reports, reviews, meta-analyses, animal studies, editorials, and conference abstracts were excluded.
Information sources and search strategy
A comprehensive search of literature was performed independently by two reviewers across electronic databases, including PubMed, Cochrane, Scopus, Embase, and Clinicaltrials.gov for relevant studies published up to 18th March 2025. There were no restrictions on Language or publication date. Our search strategy combined MeSH terms and free text keywords related to our PICO framework (Population, Intervention, Comparison, and Outcome).
Complete search strategies for each database are provided in the Supplemental Appendix(Supplemental Table 2). Additionally, reference lists of all included studies were screened manually for further eligible studies.
Study selection
All identified citations were imported into Rayyan.ai for blinded screening. Duplicate entries were identified and resolved. Two reviewers independently carried out screening by title and abstract, followed by full-text screening of potentially relevant studies.
Conflicts were resolved through discussion or adjudication by a third reviewer. The overall study selection process is illustrated using a PRISMA 2020 flow diagram. (Fig.
1).Fig. 1PRISMA 2020 Flow Diagram of Study Selection. Flow chart showing study selection process.
1598 records identified, 1469 screened after removal of 129 duplicates. 26 full texts were assessed and 14 studies were included. Key exclusion reasons: irrelevant outcomes, registry reports and missing data.
Abbreviation: PISMA= Preferred Reporting Items For Systematic Reviews and Meta-Analysis.
Data extraction
Data were extracted by two reviewers independently, onto an Excel spreadsheet with cross verification by a third reviewer. Extracted information included characteristics such as first author, year of publication, total sample size, study design, Chemotherapy dosage, intervention dosage, frequency, and follow-up. The reviewers also extracted multiple outcome measures such as Left Ventricular Ejection Fraction (LVEF), Left Ventricular Systolic Dysfunction (LVSD), Mortality, Echocardiographic parameters: Left ventricular systolic and diastolic diameters (LVsD and LVdD), left atrial diameter, and Pulmonary Artery Pressure (PAP), Right ventricular (RV) ejection fraction, RV fractional area change, fractional shortening, Strain echocardiography: global and segmental strain and strain rate, Cardiac biomarkers: B-type Natriuretic Peptide (BNP), Troponin-I and Doppler imaging such as mitral annular inflow velocity (Em).
For outcomes not reported as mean ± standard deviation, we employed the Meta-Converter tool [11] to convert medians and interquartile ranges to means and SDs, and to calculate change from baseline. A coefficient value of 0.5 was applied, as recommended in the tool’s default settings.
Risk of bias assessment
Two independent reviewers assessed the risk of bias using the Cochrane Risk of Bias tool 2.0 (RoB 2) [12] covering five domains: the Randomization process, Deviations from intended interventions, Missing outcome data, Measurement of outcomes, and Selection of the reported result. Each study was classified as having “low risk”, “some concerns”, or “high risk”. (Supplemental Fig.
1) Discrepancies were resolved by discussion or by adjudication by a third reviewer.
Data synthesis and statistical analysis
We synthesized data for primary and secondary outcomes in our analysis. Primary outcomes comprised changes in LVEF, LVSD, Change in Left ventricular diastolic diameter (LVdD), Change in Left ventricular systolic diameter (LVsD), and mortality. Secondary outcomes included Doppler imaging parameters and echocardiographic parameters (change in left atrial diameter, change in fractional shortening, change in isovolumic relaxation time, change in early mitral inflow velocity, E/A ratio, and E/E’ ratio).
Additionally, outcomes documented in fewer than three studies were excluded from the meta-analysis due to inadequate data for pooling and heterogeneity assessment. However, we retained these outcomes in the systematic review and presented our findings in tabular format (Supplemental Table 3) to ensure a complete representation of all available evidence.
Our meta-analysis included a total of thirteen studies, and the analysis was performed using RevMan version 5.4. For Dichotomous outcomes, we calculated Risk Ratios (RRs) with corresponding 95 % confidence intervals, calculated using the Mantel-Haenszel method [13]. To account for potential variability across the studies, a random effects model was used.
We generated forest plots to visually represent the pooled estimate and confidence intervals. A p-value less than 0.05 was considered statistically significant. For continuous outcomes, we employed the generic inverse variance method, which involves weighing each study’s effect estimate by the inverse of its variance.
For studies evaluating more than one carvedilol dose, each dose arm was extracted and analyzed independently to investigate potential dose-dependent effects. For Change in LVEF, both standardized mean differences (SMD) and absolute mean differences (MD) were calculated to enhance clinical interpretability, with MD results presented in Supplemental Fig. 2.
LVSD definitions varied across studies (e.g., LVEF < 50 %, or ≥ 10 % decline from baseline). Each study’s reported definition was extracted, and outcome pooling was performed using study-defined thresholds. Because clinical definitions were heterogeneous, results were interpreted with caution, and sensitivity analyses were performed when feasible.
Leave-one-out, subgroup, and meta-regression analyses
Heterogeneity was assessed using the I2 statistic, considering values exceeding 50 % as evidence of substantial heterogeneity. In such cases, we performed leave-one-out sensitivity analyses to explore the influence of individual studies. To explore potential sources of heterogeneity, we conducted prespecified subgroup analyses based on cancer type (breast cancer vs. other cancer types) and carvedilol dosing regimen (less than 12.5 mg, 12.5 mg, greater than 12.5 mg).
For the change in LVEF, where sufficient data were available, we also performed meta-regression to evaluate whether carvedilol dosage was associated with treatment effect size. (Fig. 2) These analyses were also performed using the generic inverse-variance method under a random-effects model.Fig.
2Meta-regression bubble plot of carvedilol dosage and treatment effect on left ventricular ejection fraction (LVEF) Each bubble represents an included study, with the size indicating the inverse variance. (1/vi). The x axis displays carvedilol dosage in mg while the y-axis shows the standardized mean difference (SMD) for change in LVEF.
The regression line (blue) demonstrates the dose response relationship, revealing a slight negative slope, indicating a non-significant trend towards smaller benefit with increasing dosage. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Handling of Non-Parametric data and complete-case sensitivity analysis
For studies that reported medians, interquartile range, and confidence intervals instead of means and standard deviations, we converted these values using a validated transformation tool. This procedure follows the Cochrane Handbook guidance on imputing missing standard deviations when direct estimates are unavailable (Section 16.1.3). To assess whether these imputations influenced the robustness of the pooled estimates, we conducted a prespecified complete-case sensitivity analysis for outcomes in which at least one included study required conversion.
At the study level, three studies contained non-parametric summary statistics requiring conversion, affecting three outcomes (change in LVEF, LVdD, and LVsD). (Supplementary Fig. 3, 4, 5) In this analysis, all studies that required conversion of summary statistics were excluded, and the meta-analysis was repeated using only studies reporting complete parametric data (mean ± SD).
We then compared the direction, magnitude, and statistical significance of the complete-case pooled effect with the primary analysis to determine whether the overall findings were dependent on imputed data.
Assessment of publication bias
Funnel plots were generated for outcomes with an adequate number of contributing studies; Change in LVEF (Supplementary Fig. 6) and Echocardiographic Parameters (Supplementary Fig. 7), using the standard error of the SMD plotted against the effect size.
Because very few outcomes met the recommended threshold of ≥ 10 studies, interpretation focused on visual symmetry rather than formal asymmetry testing. Consistent with Cochrane recommendations, formal statistical tests for funnel plot asymmetry (Egger’s regression, Begg’s test) and trim-and-fill procedures were not performed for outcomes with fewer than ten studies, as such tests are considered unreliable and may produce misleading results under these conditions.
Certainty of evidence
Certainty of evidence for each outcome was assessed using the GRADE approach [14] (Grading of Recommendations, Assessment, Development, and Evaluations). This approach examines multiple quality dimensions, including risk of bias, inconsistency, imprecision, indirectness, and publication bias. Each outcome was rated as high, moderate, low, or very low, with ratings reported alongside effect sizes.
(Supplemental Table 4).
Results
Eligible studies
A total of 1,598 results were yielded after the initial database searches. After removing 129 duplicates, 1,469 results were screened based on title, abstract, and full-text review leading to the removal of 1,455 records. Finally, 14 studies were found to fit our PICO, out of which 13 were included in the meta-analysis while one study, Karvandi et al, 2023, was only included in the systematic review.
The PRISMA flow diagram, shown in Fig. 1 summarizes the literature search and selection process.
Study Characteristics
The total number of patients included in the pooled analysis was 1,245, out of which 566 belonged to the control group and 679 belonged to the carvedilol group. The control group was either given a placebo [15], [16], [17], [18], [19], [20], [21] or given no drug [22], [23], [24], [25], [26] while the intervention group received carvedilol in various doses ranging from 3.125 mg to 25 mg per day. All the patients included had a well-documented history of being treated with Anthracyclines (ANT) for their respective malignancies which mainly included breast and blood cancer.
The mean age of the intervention group was 42.84(16.45) while that of the control group was 42.86(19.64). All of the studies except one [23] comprised the adult population. The male/female ratio in the intervention group was 155:504, while that in the control group was 120:426, and sex distribution was available for all studies except Liu et al.
The larger female population was mainly due to the inclusion of breast cancer patients in the study.
Quality assessment and bias analysis
RoB-2:
We determined an overall moderate risk of bias. Eight of the included studies (Karvandi et al., Myunhun Lee et al., Jhorawat et al., Beheshti et al., Salehi et al., Elito et al., Kalay et al., El-shitany et al.) were judged to have “some concerns,” while five of the included studies (Abousa et al., Armenian et al., Nabati et al., Avila et al., Bosch et al.) were judged to have overall “low risk” as shown in Supplemental Fig. 1.
GRADE
Certainty of evidence for four out of seven outcomes (Mortality, Change in LVEF, Echocardiographic Parameters, change in mitral annular inflow velocity [Em]) was assessed to be “moderate,” while in three outcomes (LVSD, Change in LVdD, Change in LVsD), it was assessed to be “high,” as shown in Supplemental Table 4.
Primary outcomes
We focused on change in LVEF, left ventricular systolic dysfunction, mortality and Change in Left ventricular diastolic and systolic diameter as the primary outcomes. Two of our included studies reported data for more than one dose of carvedilol [15], [19].
Change in left ventricular ejection fraction
Eleven studies reported left ventricular ejection fraction (LVEF) before and after anthracycline therapy. Both standardized mean difference (SMD) (Fig. 3) and mean difference (MD) (Supplementary Fig.
2). analyses demonstrated significantly greater preservation of LVEF in the carvedilol group compared with control: SMD = 0.33 [95 % CI: 0.09–0.58, p = 0.008] and MD = 1.64 [95 % CI: 0.37–2.90, p = 0.01] These results indicate a modest but statistically significant cardioprotective effect of carvedilol.Fig. 3Forestplot of changes from baseline in Left Ventricular Ejection Fraction with subgroup analyses by cancer type and carvedilol dosing regimen. Forest plot of 11 studies showing carvedilol significantly preserved LVEF versus control (SMD = 0.34; 95 % CI: 0.19–0.49; p < 0.00001).
Subgroup analyses were performed by cancer type (other cancer types vs. breast cancer) and carvedilol dosing regimen (<12.5 mg, 12.5 mg, >12.5 mg daily). Black diamonds represent pooled effect sizes for each subgroup and overall; squares indicate individual studies with proportional weights. Substantial heterogeneity was observed overall (I2 = 77 %, Chi2 = 156.18, df = 36, p < 0.00001).
Test for overall effect: Z = 4.50 (p < 0.00001). Test for subgroup differences: Chi2 = 4.32, df = 5 (p = 0.50), I2 = 0 %. Abbreviations: CI = confidence interva SMD = standardized mean difference.
The heterogeneity was substantial (I2 = 77 %). Nabati et al. (2017) was identified as an outlier because it reported an unusually large effect size relative to all other included studies, shifting the pooled estimate disproportionately and contributing most to heterogeneity.
Exclusion of Nabati et al., 2017 reduced heterogeneity to I2 = 61 % (SMD = 0.24 [95 % CI: 0.05–0.44]), confirming that this study contributed most to variability (Supplementary Fig. 8).
Subgroup Analyses
To further explore potential sources of heterogeneity, subgroup analyses were performed based on (1) cancer type and (2) carvedilol dose regimen.
By cancer type
Six studies included breast cancer patients, whereas three involved other malignancies. LVEF improvement favored carvedilol in both subgroups; breast cancer: SMD = 0.35 [95 % CI: –0.07, 0.78], I2 = 83 %; other cancers: SMD = 0.41 [95 % CI: –0.23, 1.05], I2 = 85 %, although wide confidence intervals and persistent heterogeneity indicate variation not fully explained by cancer type alone.
By carvedilol dose
Studies were stratified into three subgroups: less than 12.5 mg, = 12.5 mg, and greater than 12.5 mg daily dosing. LVEF preservation consistently favored carvedilol across all doses.
< 12.5 mg: SMD = 0.70 [95 % CI: 0.16–1.24], I2 = 80 %.
12.5 mg: SMD = 0.07 [95 % CI: –0.25–0.40], I2 = 67 %.
> 12.5 mg: SMD = 0.35 [95 % CI: 0.02–0.67], I2 = 59 %.
Although heterogeneity decreased modestly in the moderate- and high-dose subgroups, differences between doses were not statistically significant. Similar findings were observed when expressed as absolute mean differences (MD), suggesting dose-independent cardioprotection within the studied range. (Supplementary figure 2).
Meta-Regression
A meta-regression was performed to evaluate the relationship between carvedilol dose and effect size. The regression line showed a mild negative slope, indicating a non-significant trend toward smaller benefit with increasing dose. However, this association was not clinically or statistically meaningful; supporting the interpretation that carvedilol’s cardioprotective effect is largely dose-independent within the 3.125–25 mg/day range (Fig.
2).
Sensitivity Analysis
A complete-case sensitivity analysis was performed excluding studies that required conversion of median ± IQR to mean ± SD (Armenian et al., Myunhee et al., Nabati et al.). The pooled estimate remained robust (SMD = 0.38 [95 % CI: 0.10–0.66], I2 = 77 %), confirming that statistical conversions did not significantly influence the results (Supplementary Fig. 3).
Left ventricular systolic dysfunction
Four studies reported the incidence of left ventricular systolic dysfunction (LVSD). The data were pooled according to each study’s definition, which generally classified LVSD as either LVEF < 50 % or a > 10 % absolute decline in LVEF from baseline [6]. The pooled analysis demonstrated a significantly lower risk of LVSD in the carvedilol group compared with the control group (RR = 0.26 [95 % CI: 0.11–0.62], p = 0.002), as shown in Fig.
4. There was no observed heterogeneity among studies (I2 = 0 %), indicating consistent findings across trials.Fig. 4Forest plot of left ventricular systolic dysfunction (LVSD) with subgroup analysis by carvedilol dosing regimen.
Forest plot of 4 studies showing carvedilol significantly lowered the risk of LVSD compared to control (RR = 0.26; 95 % CI: 0.13–0.53; p = 0.0002). Subgroup analysis by carvedilol dosing regimen (all doses vs. 12.5 mg) is shown.
Black diamonds represent pooled effect sizes; squares indicate individual studies with proportional weights. No heterogeneity was observed (I2 = 0 %, Chi2 = 0.23, df = 6, p = 1.00). Test for overall effect: Z = 3.74 (p = 0.0002).
Total events: 9/225 (carvedilol) versus 35/225 (control). Test for subgroup differences: Chi2 = 0.00, df = 1 (p = 0.96), I2 = 0 %. Abbreviations: CI = confidence interval; RR = risk ratio; LVSD = left ventricular systolic dysfunction.
Subgroup Analyses
A subgroup analysis was performed based on carvedilol dose. In the 12.5 mg subgroup, carvedilol was associated with a significantly reduced risk of LVSD (RR = 0.25 [95 % CI: 0.07–0.87], I2 = 0 %). These results remained statistically significant, suggesting a protective effect of carvedilol against LV systolic dysfunction that is consistent across dosing regimens.
Change in left ventricular diastolic diameter
Seven studies reported the change in left ventricular diastolic diameter (LVdD) following anthracycline therapy. The pooled analysis slightly favored the carvedilol group (SMD = –0.19 [95 % CI: –0.38, –0.00] p = 0.05), indicating a modest reduction in LVdD compared with control (Fig. 5).
However, as the confidence interval approached the null value, this finding should be interpreted with caution. The overall heterogeneity was moderate (I2 = 48 %).Fig. 5Forestplot of Change in Left Ventricular Diastolic Diameter with subgroup analyses by cancer type and carvedilol dosing regimen.
Forest plot of 7 studies showing carvedilol significantly reduced LVdD compared to control (SMD = −0.19; 95 % CI: −0.31, −0.08; p = 0.0008). Subgroup analyses were performed by cancer type (breast cancer vs. other cancer types) and carvedilol dosing regimen (12.5 mg vs. > 12.5 mg daily). Black diamonds represent pooled effect sizes for each subgroup and overall; squares indicate individual studies with proportional weights.
Moderate heterogeneity was observed (I2 = 40 %, Chi2 = 36.42, df = 22, p = 0.03). Test for overall effect: Z = 3.37 (p = 0.0008). Test for subgroup differences: Chi2 = 0.56, df = 4 (p = 0.97), I2 = 0 %.
Abbreviations: CI = confidence interval; SMD = standardized mean difference; LVdD = left ventricular diastolic diameter.
Subgroup analyses
To further explore potential sources of heterogeneity, subgroup analyses were performed based on (1) the types of cancer in the population and (2) the doses of carvedilol given to the population
By cancer type
Two studies included patients with breast cancer, while two involved other malignancies. In the breast cancer subgroup, carvedilol was associated with a non-significant reduction in LVdD (SMD = −0.11 [95 % CI: −0.35, 0.13], I2 = 0 %). In the other cancers subgroup, the pooled analysis also favored carvedilol (SMD = –0.22 [95 % CI: −0.35, 0.13]), though heterogeneity increased to I2 = 50 % and the result remained statistically non-significant.
By carvedilol dose
When studies were stratified according to carvedilol dosage, six studies using 12.5 mg/day showed a trend favoring carvedilol (SMD = −0.24 [95 % CI: −0.54, 0.05], I2 = 59 %), whereas studies using > 12.5 mg/day also favored carvedilol (SMD = –0.18 [95 % CI: −0.50, 0.14], I2 = 46 %). While heterogeneity slightly decreased in the higher-dose subgroup, none of the subgroup effects reached statistical significance, indicating that carvedilol’s modest benefit in LVdD reduction is consistent but not dose-dependent.
Sensitivity analysis
A complete-case sensitivity analysis was conducted to assess the influence of data conversion on heterogeneity. Excluding Armenian et al. (which required median-to-mean conversion) yielded similar results (SMD = –0.16 [95 % CI: –0.37, 0.05], I2 = 48 %), confirming that unit conversion did not materially affect the pooled estimate (Supplementary Fig.
4).
Change in left ventricular systolic diameter
Seven studies reported changes in left ventricular systolic diameter (LVsD). The pooled analysis favored the carvedilol group ([SMD = –0.39, 95 % CI: –0.53 to –0.26], p < 0.00001) indicating a significantly greater reduction in LVsD compared to the control group (Fig. 6).
The heterogeneity was negligible (I2 = 0 %), and the results were statistically significant.Fig. 6Forest plot of change in left ventricular systolic diameter (LVsD) with subgroup analyses by cancer type and carvedilol dosing regimen. Forest plot of 7 studies showing carvedilol significantly reduced LVsD compared to control (SMD = −0.37; 95 % CI: −0.46, −0.29; p < 0.00001).
Subgroup analyses were performed by cancer type (breast cancer vs. other cancer types) and carvedilol dosing regimen (12.5 mg vs. > 12.5 mg daily). Black diamonds represent pooled effect sizes for each subgroup and overall; squares indicate individual studies with proportional weights. No heterogeneity was observed (I2 = 0 %, Chi2 = 17.88, df = 22, p = 0.71).
Test for overall effect: Z = 8.54 (p < 0.00001). Test for subgroup differences: Chi2 = 1.12, df = 4 (p = 0.89), I2 = 0 %. Abbreviations: CI = confidence interval; SMD = standardized mean difference; LVsD = left ventricular systolic diameter.
To further explore potential sources of heterogeneity, subgroup analyses were performed. Based on cancer type, the pooled estimate for the breast cancer subgroup favored the carvedilol group [SMD = –0.27, 95 % CI: –0.51 to –0.03], indicating a greater reduction in LVsD, with low heterogeneity (I2 = 0 %) and statistically significant results. Similarly, in the subgroup comprising other cancer types, the pooled analysis favored carvedilol [SMD = –0.37, 95 % CI: –0.64 to –0.09], again with no heterogeneity (I2 = 0 %) and statistically significant results.
Subgroup analysis based on carvedilol dosage also demonstrated consistent findings. For six studies using a 12.5 mg dose, the pooled analysis favored carvedilol [SMD = –0.43, 95 % CI: –0.67 to –0.20], with moderate heterogeneity (I2 = 35 %), and results remained statistically significant. In studies administering doses greater than 12.5 mg, the pooled estimate similarly favored carvedilol [SMD = –0.35, 95 % CI: –0.57 to –0.13], with no heterogeneity (I2 = 0 %) and statistically significant outcomes.
To assess the impact of unit conversion, a sensitivity (complete-case) analysis was conducted, excluding Armenian et al. The results [SMD = –0.40, 95 % CI: –0.56 to –0.25, I2 = 4 %] indicated no meaningful change in heterogeneity, suggesting that unit conversion had no significant influence on the overall findings. (Supplementary Fig.
5).
All other primary outcomes did not involve studies requiring summary-statistic conversion; thus, complete-case analysis was not applicable to those outcomes.
Mortality
Six studies reported mortality outcomes (Fig. 7). Overall, the pooled analysis favored the carvedilol group, with numerically lower mortality [RR = 0.92, 95 % CI (0.52, 1.63), p = 0.76]; however, this difference was not statistically significant.
Heterogeneity was low (I2 = 0 %).Fig. 7Forest plot of mortality with subgroup analyses by cancer type and carvedilol dosing regimen. Forest plot of 6 studies showing no significant difference in mortality between carvedilol and control groups (RR = 0.90; 95 % CI: 0.63, 1.28; p = 0.54).
Black diamonds represent pooled effect sizes; squares indicate individual studies with proportional weights. No heterogeneity was observed (I2 = 0 %). Total events: 56/909 (carvedilol) versus 65/902 (control).
Abbreviations: CI = confidence interval; RR = risk ratio.
Subgroup analyses
Subgroup analyses were performed to explore potential sources of variability. Based on cancer type, the pooled estimate for the breast cancer subgroup suggested higher mortality in the carvedilol group [RR = 1.34, 95 % CI (0.26, 7.01)], with negligible heterogeneity (I2 = 0 %) and non-significant results. In the subgroup comprising other cancer types, mortality was higher in the control group [RR = 0.71, 95 % CI (0.23, 2.22)], with moderate heterogeneity (I2 = 49 %), and results remained statistically non-significant.
Subgroup analysis based on carvedilol dosage demonstrated consistent findings. In the < 12.5 mg subgroup, mortality was slightly higher in the control group [RR = 0.89, 95 % CI (0.13, 5.93)], with no heterogeneity (I2 = 0 %). For the 12.5 mg subgroup, mortality was higher in the carvedilol group [RR = 1.05, 95 % CI (0.40, 2.70)], with low heterogeneity (I2 = 18 %).
In the > 12.5 mg subgroup, mortality was higher in the control group [RR = 0.81, 95 % CI (0.28, 2.31)], with moderate heterogeneity (I2 = 26 %). Across all dose subgroups, differences were statistically and clinically non-significant.
Secondary outcomes
Doppler imaging parameter [change in mitral annular inflow velocity- em (cm/s)]
Three studies reported change in mitral annular inflow velocity (Em). The pooled analysis showed no statistically significant difference in mitral annular inflow velocity (Em) between the carvedilol and control groups [SMD = 0.24, 95 % CI: –0.48 to 0.95; P = 0.52]. The overall heterogeneity (I2) was 81 % (Supplemental Fig.
6) that reduced to 3 % after leaving Bosch et. al 2013 out as shown in (Supplemental Fig. 7).
Echocardiographic parameters
All the included studies reported echocardiographic parameters however there was a lack of homogeneity in the selection of the final outcomes reported. Three studies reported the change in left atrial diameter, showing a non-significant reduction in the carvedilol group compared to control [SMD = −0.27, 95 % CI (−0.97, 0.42); p = 0.44]. The heterogeneity in the outcome was high (I2 = 89 %) but on removing Avila et al.
2018, it reduced to 72 %. The results were statistically insignificant.
Three studies from the pooled analysis reported the change in fractional shortening (FS%). The results showed a non-significant improvement in the carvedilol group compared to the control group [SMD = 0.54, 95 % CI (−0.57, 1.66); p = 0.34]. The heterogeneity was 94 % but it reduced to 8 % on removing El-shitany et al.
2012 from the analysis, though the results remained statistically insignificant.
Three studies reported a change in Isovolumic relaxation time (IVRT) with results showing no significant difference between the carvedilol and control groups [SMD = 0.03, 95 % CI (−0.68, 0.73); p = 0.93]. The heterogeneity in the result was 82 % that reduced to 55 % upon removing Elitok et al. 2014 from the analysis and the results were not statistically significant.
Three studies reported data on the change in early mitral inflow velocity (E). The results showed no statistically significant difference between the carvedilol and control groups [SMD = 0.18, 95 % CI (−0.07, 0.43); p = 0.16]. The heterogeneity was 0 %.
Seven studies reported data for the E/A ratio, defined as the rate of early (E) to late (A) mitral inflow velocity. The pooled analysis showed a reduction in the E/A ratio in the carvedilol group compared to the control group; however, the difference was not statistically significant [SMD = −0.46, 95 % CI (−1.03, 0.11); p = 0.12]. The heterogeneity was high (I2 = 93 %).
Upon leaving Salehi et al. 2011 from the analysis, I2 was reduced to 68 % and the results remained statistically insignificant.
Three of the included studies reported E/E’ ratio that is defined as the ratio of early mitral inflow velocity to early mitral annular motion velocity. The analysis showed no statistically significant difference between the carvedilol and control groups [SMD = −0.08, 95 % CI (−0.27, 0.12); p = 0.44], with low heterogeneity among studies (I2 = 0 %) (Supplemental Fig. 8 and 9).
Publication bias
Publication bias was assessed through visual inspection of funnel plots for both primary and secondary outcomes (Supplementary Figs.10 and 11). The funnel plot for LVEF demonstrated a roughly symmetrical distribution of studies around the pooled effect estimate, without a clear pattern of directional asymmetry. Although one study appeared right-shifted; consistent with its influence on heterogeneity, there was no indication of systematic small-study effects.
Similarly, the funnel plot for echocardiographic secondary outcomes did not show meaningful asymmetry. Because echocardiographic outcomes represented multiple distinct parameters, some dispersion was expected and does not indicate asymmetry.
Discussion
We conducted a systematic review and meta-analysis to evaluate the safety and efficacy of carvedilol in reducing the risk of AIC among patients compared to control/ placebo. Our results demonstrated significantly greater preservation of LVEF with carvedilol in preventing anthracycline-mediated decline in cardiac function. Subgroup analysis and meta-regression on LVEF confirmed the overall protective effect and suggested that this benefit is independent of the carvedilol dose administered within the studied range.
Similarly, Carvedilol was also found to cause a significant reduction in the risk of LVSD, and a significantly greater reduction in LVdD and LVsD compared to the control group. However, we did not observe a significant decrease in mortality due to AIC in the carvedilol group. Secondary outcomes included Doppler imaging parameters and echocardiographic parameters (change in left atrial diameter, change in fractional shortening, change in isovolumic relaxation time, change in early mitral inflow velocity, E/A ratio, and E/E’ ratio), all of which yielded statistically non-significant results (Fig.
8).Fig. 8Central illustration: The efficacy of Carvedilol in improving Cardiac function and Survival in Patients with Anthracycline- Induced Cardiotoxicity. This visualization summarizes our meta-analysis findings from 14 studies (1,245 patients) highlighting carvedilol’s significant cardioprotective benefits; preserved LVEF, reduced risk of LVSD, improved ventricular dimensions, with comparable mortality outcomes versus control.
These findings support prophylactic carvedilol in anthracycline-treated patients. Abbreviations: LVEF = Left Ventricular Ejection Fraction; LVSD = Left ventricular systolic dysfunction.
Understanding the mechanistic basis for carvedilol cardio-protection is important. Carvedilol, a non-selective beta-blocker with potent antioxidant properties, counters anthracycline-induced cardiotoxicity (AIC) primarily by reducing reactive oxygen species (ROS) formation and preventing mitochondrial dysfunction. This mechanistic advantage aligns closely with our clinical findings [19].
The significant reductions in LVdD and LVsD, together with improved LVEF and a decreased incidence of LVSD, collectively highlight the preservation of cardiac structure as well as the maintenance of cardiac function with prophylactic carvedilol use. These results underscore carvedilol’s ability to mitigate early myocyte injury and counteract the structural remodeling typically induced by anthracycline exposure. [20], [28].
AIC has many cardiovascular manifestations and is mainly characterized by an asymptomatic decline in the LVEF [29] . For this reason, we categorized LVEF as one of our primary outcomes. Similar to our findings, previously published literature has also supported the claim of carvedilol in preventing ANT-mediated decline in LVEF [21], [26], [15] .
While carvedilol has demonstrated efficacious reductions in LVEF, other pharmacological agents like enalapril, candesartan, and statins also exhibit cardio-protective effects. Bosch et al. found that the combination of carvedilol and enalapril effectively prevented a decline in left ventricular ejection fraction (LVEF) and the development of left ventricular systolic dysfunction (LVSD) in patients with malignant hemopathies, compared to the control group, which experienced a significant reduction in LVEF [24].
Lee et al. not only found both candesartan and carvedilol cohorts to have significantly smaller changes in LVEF, but also reported the candesartan group to have a significantly lower incidence of doxorubicin-induced cardio-toxicity, findings that are consistent with evidence showing carvedilol's active role in maintaining structural and functional cardiac tissue integrity, thereby countering mitochondrial dysfunction [25], [27]. The STOP-CA trial reported that atorvastatin use was associated with a significantly lower incidence of cancer therapy-related cardiovascular toxicity and effectively prevented a reduction in left ventricular ejection fraction (LVEF) compared to the control group [30].
Such results should direct us towards the exploration of advanced combined therapy treatment options, which could yield more efficacious results. Combination therapies potentially involving carvedilol, enalapril, candesartan, and statins could help build on carvedilol’s proven cardioprotective effects, thereby blunting insult due to ROS. If combination therapies were to demonstrate greater preservation of cardiac function without added toxicity, it could establish a new standard of care for cardiovascular prophylaxis in oncology patients—and open the door to even more sophisticated multi-agent cardio-protective strategies.
Studies included in our systematic review and meta-analysis showed a wide variability in dosing regimens of carvedilol (3.125 mg/day to 25 mg/day). To further explore the impact of this variability, we performed meta-regression analysis, which showed that the overall effect of carvedilol on LVEF appears to be dose-independent across the dose range evaluated, showing only a minimal, non-significant decrease in effect size with increasing dosage. While our subgroup analysis also showed statistically significant results in the “less than 12.5 mg” and “more than 12.5 mg” subgroup, the reliability of these results was masked by the high heterogeneity.
Therefore, our results support a dose-independent relationship as the meta-regression result remains the more statistically reliable test of a continuous dose–response relationship.
Clinical trials report mild differences in observed results when comparing doses. Salehi et al. found 25 mg/day of carvedilol to be protective against both systolic and diastolic disorders, compared to 12.5 mg/day, which only protected against diastolic disorder [19]. Abuosa et al. recorded no significant decline in LVEF for any of the three groups assigned to carvedilol [6.25 mg/day (p = 0.059), 12.5 mg/day (p = 0.100), or 25 mg/day (p = 0.073)], despite trends favoring preservation of systolic function [33].
This discordance between clinical findings and our meta-regression highlights the lack of consensus on an optimal dose for anthracycline prophylaxis. Moreover, nearly all existing evidence links the “prophylactic” use of carvedilol to being efficacious in protecting against AIC [31].
To our knowledge, no large trial to date has tested the potential effects of initiating carvedilol therapy only after early cardiotoxicity is detected. This advocates for the need for further studies that employ dose-finding and time-response designs to identify the minimal effective dose and optimal initiation window for carvedilol in AIC prophylaxis, thereby balancing cardio-protection with tolerability and paving the way toward personalized, precision cardio-oncology protocols.
We did not observe a significant reduction in AIC-related mortality with carvedilol, which may reflect the relatively short follow-up durations of the included studies (6 to 24 months). This is in coherence with the findings of meta-analyses, which showed no difference in all-cause mortality between the carvedilol and control groups [32], [33]. Moreover, it is important to understand that anthracycline-induced cardiomyopathy can appear years later, especially with cumulative dosing.
Nearly all included randomized controlled trials (RCTs) focused on changes in LVEF or the incidence of LVSD as their primary outcomes. Since mortality is a rare event in the short-term prophylactic setting, it may be safe to assume that the included studies were not powered to detect differences in this outcome. Therefore, this non-significant finding should be interpreted as a limitation of statistical power and short follow-up duration rather than conclusive evidence of no long-term survival benefit.
Future trials with substantially larger sample sizes and follow-up periods need to be conducted to better analyze mortality focused outcomes to feasibly assess the true clinical impact on prognosis.
Strengths and limitations
Our systematic review and meta-analysis is the first paper to comprehensively synthesize and analyze all available data to ensure accurate reporting of the efficacy of carvedilol in the treatment of AIC. Two previously published meta-analyses [31], [32] have evaluated the role of carvedilol in preventing AIC. However, these two studies have not included certain RCTs which have been included by us El.Shitany et al, Bosch et al [23], [24] and Salehi et al [19].
Additionally, we have included four new RCTs [15], [16], [34], [25] which have contributed to a greater sample size and more rigorous analysis of all available data. This also contributed to the lowest heterogeneity observed in our meta-analysis for one of the primary outcomes (LVEF), in contrast to the consistently higher heterogeneity reported in previous studies, thereby supporting the greater reliability of our results. Moreover, we conducted a quality assessment of the included studies using both the Cochrane RoB 2 tool and the GRADE approach, which further strengthens the credibility and robustness of our findings.
Despite the strengths of our systematic review and meta-analysis, it is important to note that the majority of studies included mainly evaluated patients with breast cancer. This demographic imbalance significantly restricts the generalizability of our positive findings to patients with other anthracycline-treated malignancies, such as lymphoma, leukemia, or gynecological cancers. Therefore, we strongly emphasize the need for future prospective, randomized controlled trials to rigorously evaluate the effect of carvedilol in preventing cardiotoxicity across a more diverse patient population, which is essential before establishing broad clinical guidelines for AIC prophylaxis.
Conclusion
In conclusion, our systematic review and meta-analysis demonstrates that prophylactic use of carvedilol significantly reduces anthracycline-induced declines in left ventricular ejection fraction and the risk of systolic dysfunction, while favorably modulating ventricular dimensions. Although we did not observe a statistically significant impact on short-term mortality, these findings underscore carvedilol’s cardio-protective potential in cancer patients receiving anthracycline therapy. Given the predominance of breast cancer patients in included studies and relatively brief follow-up periods, we recommend future large-scale, long-term randomized trials across diverse malignancies to establish optimal dosing strategies, evaluate late-onset cardiotoxicity prevention, and confirm survival benefits.
Authors' contributions:
Conceptualization: R.B, S.M.A, A.R.
Literature Search:N.M, S.A.K.
Study Selection: N.A.R, R.N, S.M.A.
Protocol Registration: R.B.
Data Extraction: N.M, S.A.K, A.A.B.
Certainty of evidence and Quality Assessment: R.N, A.R, N.M.
Data Analysis: S.M.A, R.B, K.SManuscript writing - original draft:
Tables: S.M.
Central illustration: A.A.B.
PRISMA Flowchart: S.M.
Abstract: A.A.B.
Introduction: A.R.
Methods: N.M.
Results: A.R.
Discussion: R.N, R.A, K.S.
Supplementary File: N.M.
Final formatting and compilation: N.M.
Final review and supervision: R.A, K.S.
Sabahat Ul Ain Munir Abbasi and Riya Bhagwan contributed equally to this manuscript as first authors; Najaf Ahmed Rajpar and Rayyan Nabi contributed equally to this manuscript as fifth authors.
Data Availability Statement
All data generated or analyzed during this study are included in this published article and its supplementary information files.
[27].
CRediT authorship contribution statement
Sabahat Ul Ain Munir Abbasi: Writing – review & editing, Project administration, Formal analysis, Conceptualization. Riya Bhagwan: Writing – review & editing, Supervision, Project administration, Formal analysis, Data curation, Conceptualization. Aamna Rehman: Writing – original draft, Investigation, Data curation.
Neha Malik: Writing – original draft. Sanya Ashraf Khaskheli: Writing – original draft. Najaf Ahmed Rajpar: Writing – original draft.
Rayyan Nabi: Writing – original draft. Ayesha Amir Basra: Writing – original draft. Shehdev Meghwar: Writing – original draft.
Raheel Ahmed: Supervision. Kalpana Singh: Formal analysis, Supervision, Writing – review & editing.
Funding
No funding or grants were received for the completion of this study.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.