A systematic review and meta-analysis of beta-blockers and renin–angiotensin system inhibitors for preventing left ventricular dysfunction due to anthracyclines or trastuzumab in patients with breast cancer

Abstract

Aims

Trastuzumab and anthracyclines, often used in the treatment of breast cancer, may impair myocardial function, and reduce left ventricular ejection fraction (LVEF), potentially causing heart failure. Randomized controlled trials (RCTs) have evaluated the effects of beta-blockers (BBs), angiotensin receptor blockers (ARBs), and angiotensin-converting enzyme inhibitors (ACEI) on trastuzumab- and anthracycline-associated cardiotoxicity. We report a meta-analysis of these RCTs in patients with breast cancer.

Methods and results

The primary analysis was on the effect of BBs and ACEI/ARBs on LVEF in patients treated with either trastuzumab or anthracyclines. A secondary analysis was done investigating the effect of BBs or ACEI/ARBs on LVEF in trastuzumab and anthracycline treatments. Only RCTs were included using the search term ‘ARBs, ACEIs, BBs, anthracyclines, trastuzumab, and breast cancer’ in PubMed, Embase, and CENTRAL up to 31 March 2021. A meta-analysis was conducted to estimate the mean difference (MD) in LVEF between intervention and placebo groups at follow-up. A total of nine RCTs (n = 1362) were included in the analysis. All patients were women. BBs and ACEI/ARBs were shown to attenuate the decline in LVEF during trastuzumab and anthracycline treatments [MD: 2.4; 95% confidence interval (CI): 0.3–4.2 and MD: 1.5; 95% CI: –0.6 to 3.7]. Compared with placebo, LVEF was significantly higher in patients assigned to BB or ACEI/ARB on trastuzumab (MD: 2.3; 95% CI: 0.0–4.6) but not on anthracyclines (MD: 1.9; 95% CI: –0.5 to 4.2).

Conclusion

Both BB and ACEI/ARB therapies were associated with the preservation of LVEF during trastuzumab and anthracycline-containing regimens as compared with placebo, suggesting both to be beneficial.

Structured Graphical Abstract

Breast cancer treatment trajectory and protective effects of angiotensin-converting enzyme inhibitor/angiotensin receptor blockers and beta-blockers. ACEI, angiotensin-converting enzyme inhibitor; ARBs, angiotensin receptor blockers; BB, beta-blocker; MD, mean difference.

Open in new tabDownload slide

See the editorial comment for this article ‘Cardioprotection in breast cancer patients: one size fits all?’, by Geeta Gulati, https://doi.org/10.1093/eurheartj/ehab810.

Introduction

Breast cancer is the most common form of cancer in women.¹ Apart from surgery and radiotherapy, systemic chemotherapy, and immunotherapy, including anthracyclines and trastuzumab, are recommended to try to eradicate residual tumour cells. Anthracyclines and trastuzumab can have off-target effects, including negative effects on cardiac function.² Patients with breast cancer expressing human epidermal growth factor receptor 2 (HER2) typically receive both anthracyclines and trastuzumab (a monoclonal antibody that blocks HER2 receptors) in a sequential manner. Detrimental effects of trastuzumab and anthracyclines on left ventricular systolic function, including a reduction in left ventricular ejection fraction (LVEF), have been found in clinical practice.³

To prevent cardiotoxicity, evidence-based treatments for heart failure (HF) with reduced ejection fraction (HFrEF) have been prescribed concomitantly with cancer treatments containing anthracyclines and trastuzumab.⁴ Thus far, published reviews of cardiotoxicity associated with anthracyclines and trastuzumab therapies for breast cancer patients have rarely included meta-analyses of existing randomized controlled trials (RCTs) evaluating distinct vs. indistinct blockade of the sympathetic and/or renin–angiotensin system despite the low numbers of participants in existing trials. Therefore, we performed a meta-analysis investigating the treatment effect of angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor blockers (ARBs), and beta-blockers (BBs) on LVEF during either anthracycline or trastuzumab regimens in patients with no pre-existing HF.

Methods

We followed the PRISMA guidelines for reporting meta-analysis (see Supplementary material online).⁵ The target was RCTs assessing LVEF before and after treatment with either anthracyclines or trastuzumab, concomitantly with either mono- or combination therapy with an ARB or ACEI or BB vs. placebo.

Search methods

PubMed was queried through the search term ‘ACE inhibitors, angiotensin receptor blockers, beta-blockers, anthracyclines, trastuzumab and breast cancer’. Embase and CENTRAL were searched for additional RCTs. Studies were identified and included from database inception to 31 March 2021. All titles and abstracts were investigated for randomization to intervention vs. placebo. All potentially relevant articles were read in full. Moreover, we reviewed previous systematic reviews, guidelines, or statements from cardiovascular society meetings for references to any RCTs not found in our own search results. The searching process was restricted to two of the authors (C.L. and T.H.N.).

Study and data collection

Studies for inclusion were RCTs randomizing to ACEI or ARBs or BBs or both as compared with placebo. Furthermore, studies had to contain information on LVEF in per cent (%) before and after breast cancer treatment. Measurement of LVEF could have been performed by either echocardiography, magnetic resonance imaging (MRI), or multi-gated acquisition (MUGA) scan.

Non-randomized studies were excluded. Further exclusion criteria were previous HF.

Baseline characteristics including age, gender, type, and duration of medical breast cancer treatment, duration of BB and ACEI/ARB treatment, radiotherapy before, during, or after the trial, other medical breast cancer therapy prescribed before or after the study drug, LVEF at baseline, type of LVEF measurement, LVEF as a primary outcome, and observation time were gathered.

Study selection was made independently by two reviewers (C.L. and T.H.N.). In case of disagreement, consensus was reached by the reviewers.

Risk of bias assessment

The Cochrane collaboration tool was applied to assess internal validity of the selected studies through assessment of risk of bias.⁶ The categories were sequence generation, allocation concealment, blinding, incomplete outcome data, and selective outcome reporting. Evaluation of risk of bias was conducted by two independent reviewers (C.L. and T.H.N.) reaching consensus in case of disagreement. The risk of bias was categorized as ‘low’ or ‘high’; unclear was systematically classified as high risk.

Outcomes

Our primary outcomes were the distinct effects of BBs and ACEI/ARBs on LVEF during either trastuzumab or anthracycline treatments. Secondary outcomes were the effect of either BBs or ACEI/ARBs on LVEF during anthracycline and trastuzumab treatments.

Statistics

Continuous data were calculated as means [standard deviations (SDs)] where discrete data were presented as percentages. Means of LVEF after chemotherapy were applied in calculating functional differences between the intervention groups. Meta-analyses estimated the pooled mean difference (MD) of LVEF at the follow-up time and associated 95% confidence intervals (CIs) between the group randomized to either BB or ACEI/ARB treatment vs. a control group receiving placebo. Overall, a random effect model was undertaken translating moderate to high heterogeneity due to the double-sided nature of ACEI/ARBs, BBs, anthracyclines, and trastuzumab in the analysis. The effects sizes of the individual studies were pooled by inverse variance weighting through the DerSimonian–Laird (DL) estimate and displayed through forest plots. Test statistics and CIs were adjusted according to the Hartung and Knapp method.⁷ The 95% CIs were defined as the mean ± 1.96 × standard error (SE) in a normal distribution. The SD was the SE × √n, where n was number of participants. If follow-up LVEF was reported indirectly as a MD, we approximated follow-up LVEF and associated SD. LVEF was calculated as the sum of baseline LVEF and the MD. The SD of follow-up LVEF was substituted by the baseline SD of LVEF when missing (only the Guglin study). Heterogeneity was calculated as I  ².⁸

Meta-regressions were undertaken through univariate analysis with the above-described baseline characteristics from the studies. Heterogeneity was assumed at least moderate in the primary outcomes due to the double-sidedness of breast cancer and HF preventive treatments.

A reference analysis was performed as a sensitivity assessment including all the selected studies without distinguishing between medical breast cancer or preventive HF treatment. This was pre-specified to investigate linear associations between the MD of LVEF and anthracycline, trastuzumab, BBs, and ACEI/ARBs. Finally, meta-analyses estimating the standardized mean differences (SMDs) were undertaken to adjust for potential measurements differences between echocardiography, MUGA, and MRI of the heart. The package Meta in R statistics was applied in the data synthesis.⁹

Results

Studies included vs. excluded

A total of 237 studies were found in our preliminary search in PubMed, Embase, and CENTRAL. Among these, a total of nine fulfilled our inclusion criteria (Table 1 and Supplementary material online).^10–18 We did not find supplementary studies from our search in guidelines and statements from cardiovascular society meetings.

Characteristics

A total of 1362 patients were included in the analysis, and all were women. Mean age varied from 40.8 to 53.6 years between the studies. Overall, the RCT populations were without previous cardiovascular disease since it was part of their exclusion criteria. The baseline LVEF means varied between 59.5% and 66.0%. Three studies had a factorial design and randomized the experimental group to either ACEI/ARBs, BBs, or both at the same time.^10,11,15 Five studies randomized solely to BBs vs. placebo,^{13,14,16–18} and only one RCT allocated patients to ACEI/ARBs and a placebo group, respectively.¹² A total of three studies examined the preventive effect of HF treatments with trastuzumab^10–12 vs. six with anthracycline treatments^13–18.

The duration of chemotherapy ranged between 63 and 365 days. In addition to these, between 0% and 65.1% of patients received radiotherapy as part of their post-surgical treatments and all patients, except for one trial, received other biological and chemotherapeutic agents in addition to trastuzumab or anthracyclines.¹⁸

Primary and secondary outcomes

ACEI/ARB therapy irrespective of concomitant anthracycline or trastuzumab therapy resulted in a non-significantly higher LVEF compared with placebo (MD: 1.5; 95% CI: –0.6–3.7; P = 0.11; I  ² = 52%) (Figure 1A). In contrast, BB therapy preserved LVEF significantly better compared with placebo (MD: 2.4; 95% CI: 0.3–4.5; P = 0.033; I  ² = 82%) (Figure 1B).

As a secondary outcome, ACEI/ARB or BB therapy preserved LVEF compared with placebo (MD: 1.9; 95% CI: –0.5 to 4.2; P = 0.096; I  ² = 77%) although non-significantly during anthracycline therapy (Figure 2A). Furthermore, during trastuzumab therapy, we also found preservation of LVEF with ACEI/ARBs or BBs (MD: 2.3, 95% CI: 0.0–4.6; P = 0.047; I  ² = 72%) as compared with placebo (Figure 2B).

Sensitivity analysis

Irrespective of cancer therapies, those randomized to active treatment had better preservation of LVEF compared with those assigned to placebo (MD: 2.0; 95% CI: 0.7–3.4; P = 0.007; I  ² = 75%) (Supplementary material online). The decline in LVEF was similar whether trastuzumab or anthracycline was the background breast cancer treatment (P = 0.38) (Table 2). Potential LVEF protection was similar for ACEI/ARBs and BBs (P = 0.57). Whether LVEF was a primary outcome or not did not influence the MD (P = 0.65), nor did the imaging modality. Meta-analyses using SMDs were similar to the MD models (Supplementary material online).

Risk of bias

A total of six out of nine studies were assessed to have high risk of bias in the randomization process and in their blinding of both participants and investigators (Supplementary material online).^{12–14,16–18} The overall trend among these assessments was lack of clear description of methods. A single study was stated to be patient-blinded only.¹⁴ Attrition bias was high in four of the studies due to substantial numbers of patients not completing the study protocol without known impact on the study outcome. In general, other biases were assessed as high when incomplete description of the methodology existed, and attrition bias was suspicious.

Discussion

Our meta-analysis of nine RCTs showed a protective effect of ACEI/ARBs and BBs on LVEF during breast cancer treatments with anthracyclines and trastuzumab as compared with placebo (Graphical Abstract).

Issues in our analysis were the great variety among the RCTs of supplementary breast cancer treatments, such as radiotherapy and cardiotoxic drugs before or after anthracycline and trastuzumab regimens. Moreover, the haemodynamic side-effects of ACEI/ARBs and BBs leading to hypotension, tiredness, and dizziness will be discussed in the following.

The beneficial effects of ACEI/ARBs and BBs on LVEF decrease were demonstrated in our analysis. These results are equally reproduced in observational data with LVEF preservative effects of ACEI/ARBs and BBs in anthracycline- and trastuzumab-induced cardiotoxicity. Up to 10% of patients treated with anthracyclines and trastuzumab succumb to cardiotoxicity.¹⁹ Experimental studies with animal models have also demonstrated efficacy of ACEI/ARBs and BBs on preservation of LVEF during exposure to potentially cardiotoxic breast cancer treatments.²⁰ A meta-analysis investigating neurohormonal therapies in cancer patients receiving chemotherapy found similar significant beneficial effects favouring ACEI/ARBs and BBs in sensitivity analyses of breast cancer and anthracycline treatment.²¹ Preventive therapy including ACEI/ARBs and BBs to preserve the LVEF during both chemotherapeutic and biological breast cancer treatment therefore seems feasible.

The secondary outcome distinguishing between anthracycline and trastuzumab treatment in women with breast cancer but not the type of preventive drug established similar LVEF preservative effects as compared with the primary outcomes.

Combinatorial results from factorial designs were not included in the analysis. If both intervention drugs were effective independently, this suggests that ACEI/ARBs and BBs may be prescribed concomitantly, although the latter is based on HFrEF according to the HF guidelines.²² The factorial designed studies supported this as none of them showed interaction.

Although ACEI/ARBs and BBs may be prescribed together with anthracycline and trastuzumab therapy, their optimal dose remains unknown. Preventive drug doses in the included RCTs were often lower than target doses for HFrEF treatment indicating that up-titration was not achieved. This suggests that lower drug doses than recommended in HFrEF guidelines preserved LVEF in breast cancer patients. The optimal combination of ACEI/ARBs and BBs remains unknown.

Heterogeneity was moderate to high overall. Factors explaining the differences between the RCTs were not established through meta-regressions. In contrast to real-life studies suggesting a larger cardiotoxic profile of anthracyclines as compared with trastuzumab, the latter predominantly causing temporary LVEF decrease, meta-regressions were not able to show LVEF differences between anthracycline and trastuzumab treatments²³ (Table 2). By the same token, there were no difference in efficacy comparisons of ACEI/ARBs and BBs in the meta-regressions. This is despite their different mechanism of action on the heart through afterload reduction from vasodilation and reduced oxidative stress from sympathetic nerve stimulation, respectively.²⁴ The overall small MDs in the LVEF over the follow-up time could be calculated as an absolute LVEF difference up to 2.4% between intervention and placebo (Figure 1B). On the long-term perspective, a gradual decline in heart function without protective LVEF drugs is suspected. Clinical cases of severe LVEF deterioration many years after anthracycline and trastuzumab administrations support a ‘gradual decline’ theory.²⁵ From our meta-regression results, no priority of ACEI/ARBs as compared with BBs was found in the prevention of LVEF decrease in patients undergoing breast cancer treatment.

The outcomes show benefits of ACEI/ARBs and BBs although the antecedents were non-significant. Implications could be administration of both preventive HF drugs from the beginning in breast cancer patients receiving anthracycline and trastuzumab. Nevertheless, caution should be exercised when combined with anthracyclines and trastuzumab. For example, nausea and fever induced by breast cancer treatments may be exacerbated by hypotension from the preventive drugs. This could even lead to renal impairment. To avoid such haemodynamic side-effects, a general recommendation is to pause ACEI/ARBs and BBs during a sick day.²⁶ As an isolated pharmacological mediator, ACEI/ARBs and BBs should not cause haemodynamic imbalance even in patients without cardiovascular disease.²⁷ The lower therapeutic dose from the RCTs compared with conventional HFrEF treatments does also reduce the risk of side-effects. The 2016 ESC position paper on cancer treatments and cardiotoxicities indicates that preventive HF therapy should be restricted to ACEI/ARBs and BBs; further research is required to define the feasibility of employing other HF drugs.²⁸ This is currently being investigated in the PRADA II study.²⁹

One virtue of well conducted meta-analyses is they may justify Class A evidence in guidelines depending on pre-defined endpoints. Moreover, our results were limited to patients treated for breast cancer, which eliminated gender differences and reduced the variation in study protocols. The latter was although substantial in the selected studies.

It could be a downside that the outcomes of ACEI/ARBs and anthracyclines were not statistically significant. Broader CIs were expected due to the Hartung–Knapp approach and fully qualified due to the uncertain heterogeneity present among the RCTs.⁷ However, in analyses with five or fewer studies, the amount of uncertainty may be overestimated. Supporting the lack of statistical significance is that none of the included RCTs were based on a mechanistic model. This means that neither ACEI/ARBs nor BBs were targeted directly to influence cardiotoxic metabolites from anthracyclines or trastuzumab. Instead, they were targeted to alter the physiology of the heart to a more favourable environment against cardiotoxicity. Therefore, significant outcomes can be hard to achieve. The statistical trend towards LVEF protection from ACEI/ARBs and BBs was highlighted by that all analyses were statistically significant when the DL estimate was unadjusted by the Hartung–Knapp approach (not reported).

A real caveat of our study was the great variation between RCT results and the lack of baseline characteristics to explain this in our meta-regressions. We presume this variation may largely have been due to differences in study protocols with respect to additional breast cancer drugs administered, beside anthracyclines and trastuzumab. Inconsistent reporting of left vs. right-sided thoracic radiotherapy could have influenced our effect sizes, but presumably in a consistent way. The double sidedness of two HF preventive and two breast cancer treatments in our analysis could be a drawback and may have presumably contributed to the moderate to high statistical heterogeneity in our analysis due to distinct mechanism of action of the four drugs. The significant heterogeneity among the studies made funnel plots inappropriate to assess publication bias.³⁰ We do not interpret the substantial heterogeneity of the included RCTs to be linked to the fact that they did not all apply the same left ventricular measurement at follow-up as their primary outcome. It is an issue in published medical science that data manipulations lead to false conclusions, but in our case, we think this was not existing. This is due to the integration of LVEF and additional functional measurements during the same imaging sessions of the heart. A priori, the mode of LVEF measurement could cause measurement error, but the comparison between modes was neutral in the meta-regression. It was equally a weakness that the follow-up periods of the RCTs were not long-term based since diagnosis of HF in some patients first appears decades after their breast cancer treatments.

Conclusion

BBs but not ACEI/ARBs were found to preserve LVEF significantly and more effectively than placebo in breast cancer patients receiving adjuvant trastuzumab and/or anthracyclines. Despite heterogeneity in effect sizes, only positive associations between LVEF and the interventional drugs were found. Our results therefore may suggest prescription of either ACEI/ARBs or BBs during trastuzumab and/or anthracycline therapy in patients undergoing breast cancer treatments. It remains to be seen what the modest mean preservative LVEF effects (max 2.4%) from ACEI/ARBs and BBs mean in the long term.

Supplementary material

Supplementary material is available at European Heart Journal online.

Conflict of interest: C.L., T.H.N., L.R.E., C.L., J.M.B., M.M.L., L.K., and F.B. declared no conflicts of interests. J.G.F.C. declared Grants from Vifor Pharma, Pharmacosmos, Novartis, and Amgen and personal fees from Vifor Pharma, Servier, Pharmacosmos, and Astra Zeneca.

References