Cardiotoxicity As An Adverse Effect Of Immunomodulatory Drugs And Proteasome Inhibitors In Multiple Myeloma: A Network Meta-analysis Of Randomized Clinical Trials

Abstract

We aim to determine the cumulative and comparative risk of cardiovascular events associated with different Immunomodulatory Drugs (iMiDs) and Proteasome Inhibitor (PIs) in Multiple Myeloma (MM) patients through pairwise and network meta-analysis. Electronic searches were conducted using Ovid MEDLINE, EMBASE, CINAHL, Web of Science, and Clinical Trial Registry (ClinicalTrials.gov) up to May 2021. Phase 3 randomized clinical trials (RCTs) reporting cardiotoxicity in MM patients (newly diagnoses and/or relapsed) treated with iMiD and/or PI. Studies, where iMiD or PI was used alongside the chemotherapy vs. placebo or no additional drugs (control) in the other arm were included. The primary outcome was the presence of cardiotoxicity after follow-up. Pairwise meta-analysis and network meta-analysis were performed using the frequentist’s approach to estimate the odds ratio (OR). Twenty RCTs with 10,373 MM patients were included in this analysis. Eleven studies compared iMiDs with control, seven studies compared PIs with control, and two studies compared bortezomib against carfilzomib. CTACE high-grade (≥ grade 3) cardiotoxic events were increased with iMiDs compared to their control counterpart (OR 2.05; 95% CI 1.30–3.26). Similar high-grade cardiotoxicity was also noted more frequently with PI use when compared to the control group (OR 1.67; 95% CI 1.17–2.40). Among the PIs, carfilzomib was associated with a maximum risk of cardiotoxicity (OR 2.68; 95% CI 1.63–4.40). There was no evidence of publication bias among studies. iMiDs and PIs, particularly carfilzomib, appear to be associated with increased risk of high-grade cardiovascular events in MM patients.

Introduction

Targeted therapies are an emerging area in cancer treatment, leading to better survival outcomes in patients with certain types of cancer^1,2. However, long-term sequelae of treatment with targeted therapies have been reported with systemic adverse events, including cardiovascular events^3–7. The targeted therapeutic armamentarium in the treatment of multiple myeloma (MM) is expanding, with Immunomodulatory Drugs (iMiDs) and Proteasome Inhibitors (PIs) constituting its backbone⁸. Current guidelines recommend the use of a combination of iMiDs and PIs to treat patients with newly diagnosed MM as well as relapse MM patients⁹. Patients with MM are typically diagnosed with the disease at an advanced age when there is a pre-existing underlying burden of atherosclerotic cardiovascular disease (ASCVD)^10,11. Apart from the overlapping risk factors, disease progression in MM has an adverse cardiovascular profile as well. Hence, proper assessment of cardiotoxicity due to treatment reagents is necessary for optimal patient management. Adequate estimation of the cardiotoxic side-effect profile of iMiDs and PIs lacks in the existing literature¹². Additional cardiotoxic effects of the treatment have the potential to negate the benefits attained from the treatment itself. iMiDs have been documented to be associated with cardiac arrythmias^10,13. Evidence suggests a strong association between carfilzomib, a PI, and cardiotoxicity^14,15. However, whether the cardiotoxicity of carfilzomib is a class-specific adverse profile of PI remains poorly studied. In this pairwise meta-analysis and network meta-analysis, we explored direct cardiotoxicity with iMiDs and PIs in patients with MM. Also, we compared the cardiotoxic profile of individual PI against each other.

Methods

This systematic review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and was conducted following a priori established protocol¹⁶.

Search Strategy

Electronic searches were conducted of Ovid MEDLINE, EMBASE, CINAHL, Web of Science, and Clinical Trial Registry (Clinical Trials.gov) up to May 2021. The complete database search strategies are provided in Appendix 1. The searches were limited to studies published in English, and animal studies were excluded. All identified studies were combined in a single reference manager file (EndNote) and uploaded in an online software (Covidence). Duplicates were discarded, and the titles and abstracts were reviewed by two authors independently (AD and SDG) to exclude studies that did not report the adverse events due to treatment with iMiDs and PIs in MM patients. The full texts of the remaining articles were examined to determine whether they contained relevant information. Disagreements were harmonized by consensus, in conjunction with the senior investigator (BR and AG). Second, the reference lists from included original articles and recent reviews and meta-analyses on related topics were hand-searched to identify additional studies. When necessary, the original investigators were contacted to clarify data or provide other data. Figure 1 summarizes the study identification and selection process.

Figure. 1.

Study selection flow chart.

Study Selection

RCTs (Phase III) reporting incidence rate or percentage of cardiotoxic in MM patients who received either iMiDs or PIs alone or a combination of iMiD or PI along with standard therapy, and which adhered to Common Terminology Criteria for Adverse Events (CTCAE) for grading of cardiotoxicity were included in this study¹⁷. Cardiotoxic events were reported as cardiac toxicity, cardiac arrhythmia, conduction abnormality, cardiac failure, coronary artery stenosis, ischemic heart disease, myocardial infarction/angina, or a composite of any of the mentioned events in individual studies. Cardiotoxicity of grades 3–5 were defined as high-grade cardiotoxicity. A detailed account of the patient, intervention, comparison, and outcomes (PICOS) element of inclusion and exclusion is shown in Appendix Table 1. A detailed account of the baseline characteristics of the individual clinical trials is listed in Table 1A, B.

Table 1A.

Characteristics of individual clinical trial in iMiD

Trial	Year of Study	Place of Study	Men (%)	Female (%)	Disease characteristics	Intervention Arms	Age (years)	Follow-up (months)	Cardiotoxicity (CTACE)
Rajkumar	2005	USA	54.8	45.2	Newly Diagnosed Multiple Myeloma	Thalidomide-Dexamethasone vs Dexamethasone	65	NA	3–5
Palumbo	2006	Italy	NA	NA	Newly Diagnosed Multiple Myeloma	Melphalan-Prednisone-Thalidomide vs Melphalan-Prednisone	72	17·6	3–5
Attal	2006	Multicentered	55.7	44.3	Newly Diagnosed Multiple Myeloma	Lenalidomide vs Placebo	58.5	39	3–5
Facon	2007	Multicentered	43.9	56.1	Newly Diagnosed Multiple Myeloma	Melphalan-Prednisone-Thalidomide vs Melphalan-Prednisone	67	36·8	3–5
Rajkumar	2008	USA	50.6	49.4	Newly Diagnosed Multiple Myeloma	Thalidomide-Dexamethasone vs Dexamethasone	64.2	18	3–5
Spencer	2009	Australia	60.5	39.5	Newly Diagnosed Multiple Myeloma	Thalidomide-Prednisone vs Prednisone	57	36	All-Grade
Beksac	2010	Turkey	53.9	46.1	Newly Diagnosed Multiple Myeloma	Melphalan-Prednisone-Thalidomide vs Melphalan-Prednisone	70.5	23	3–5
Ludwig	2010	Multicentered	53.1	46.9	Newly Diagnosed Multiple Myeloma	Thalidomide-Interferon or Interferon	71.5	35	All-Grade
McCarthy	2012	USA	54.3	45.7	Previously Treated Multiple Myeloma	Lenalidomide vs Placebo	58.5	34	3–5
Palumbo	2012	Italy	51.1	48.9	Newly Diagnosed Multiple Myeloma	Melphalan-Prednisone-Lenalidomide vs Melphalan-Prednisone	71.5	30	3–5
Jackson	2019	UK	62	38	Newly Diagnosed Multiple Myeloma	Lenalidomide vs Observation	66	31	3–5

Table 1B.

Characteristics of individual clinical trial in PI

Trial	Year of Study	Place of Study	Men (%)	Female (%)	Disease characteristics	Intervention Arms	Age (years)	Follow-up (months)	Cardiotoxicity (CTCAE)
Cavo	2010	Italy	57.6	42.4	Newly Diagnosed Multiple Myeloma	Bortezomib-Thalidomide-Dexamethasone vs Thalidomide-Dexamethasone	57	36	3–5
Garderet	2012	Multicentered	62.8	37.2	Previously Treated Multiple Myeloma	Bortezomib-Thalidomide-Dexamethasone vs Thalidomide-Dexamethasone	61.3	24	All-Grade
Stewart	2014	Multicentered	NA	NA	Previously Treated Multiple Myeloma	Carfilzomib-Lenalidomide-Dexamethasone vs Lenalidomide-Dexamethasone	64.5	24	All-Grade
Moreau	2016	Multicentered	56.6	43.4	Previously Treated Multiple Myeloma	Ixazomib-lenalidomide-dexamethasone vs lenalidomide-dexamethasone	66	14.7	All-Grade
Dimopoulos	2017	Multicentered	NA	NA	Previously Treated Multiple Myeloma	Carfilzomib-dexamethasone vs Bortezomib-dexamethasone	65	12	All-Grade
Hou	2017	China	68.7	31.3	Previously Treated Multiple Myeloma	Ixazomib-lenalidomide-dexamethasone vs lenalidomide-dexamethasone-placebo	61.25	23	All-Grade
Durie	2017	Multicentered	37.3	62.7	Newly Diagnosed Multiple Myeloma	Bortezomib-lenalidomide-dexamethasone vs lenalidomide-dexamethasone	NA	30	All-Grade
Facon	2019	Multicentered	50.5	49.5	Newly Diagnosed Multiple Myeloma	Carfilzomib-melphalan-prednisone vs Bortezomib-melphalan-prednisone	72	22	All-Grade
Dimopoulos	2019	Multicentered	63.1	36.9	Previously Treated Multiple Myeloma	Ixazomib vs Placebo	59	30·9	All-Grade

Data abstraction

Two reviewers (AD and SDG) independently abstracted data on the following study and patient-related characteristics onto a standardized form: (a) study characteristics – last name of the primary author, time period of study/year of publication, country/region of the population studied and type of study, (b) patient characteristics – the total number of participants and number of patients receiving intervention and number of patients receiving placebo/observation, gender of the patients, duration of follow-up of the studies, type of active interventions, phase of the study, all-grade cardiotoxicity in each arm and high-grade cardiotoxicity in each arm.

Outcomes assessed

Our primary outcome was comparing the cardiotoxicity with iMiD and PI treatment vs. control and comparing individual PI for cardiotoxic events in MM patients. A composite of cardiac toxicity, cardiac arrhythmia, conduction abnormality, cardiac failure, coronary artery stenosis, ischemic heart disease, myocardial infarction/angina or individual type of event were considered as cardiotoxic events in our analysis and outcomes. Attrition bias was taken into consideration and patients who completed the trials were taken in our analysis..

Quality assessment

The risk of bias in these individual clinical trials was assessed by three authors (AD, BR, and AG) independently using the Cochrane collaboration’s tool for assessing the risk of bias, which evaluates validity and bias in studies of prognostic factors across seven domains: random sequence generation bias, allocation concealment, blinding of participants/personnel, outcome assessor blinding, incomplete outcome data, selective reporting (reporting bias) and other bias¹⁸(Appendix Table 2A, 2B). Additionally, the contribution of direct and indirect evidence in the network meta-analysis was assessed using the GRADE framework, which characterizes the quality of a body of evidence incorporating study limitations inconsistency, imprecision, indirectness, and publication bias for the primary outcomes¹⁹.

Statistical analysis

R software version 3.4.1 and package meta was used to perform statistical analysis reporting odds ratios (OR)²⁰. For studies reporting zero events in an arm, the classic half-integer correction was applied²¹. For the meta-analysis, the random-effects model was considered, and DerSimonian and Laird estimate was utilized²². Additionally, we conducted a network meta-analysis for PIs incorporating data from all studies in a random-effects model, similar to methods we have described previously²³. Our model accounts for heterogeneity in study effect across trials but assumes that the cardiotoxic effects due to PIs are not systematically different across trials²⁴. Hence, while direct and indirect estimates for a comparison between the cardiotoxicity between two PIs (A and B) may differ across studies due to heterogeneity, these differences are not reflective of the systematic differences as a function of trial design, i.e., the estimate for comparing the cardiotoxic profile of PIs from A and B comparison from two-arm trials comparing A and B are similar to those derived from three-arm trials (A-B-C). Statistical heterogenicity among studies included in the meta-analysis was assessed using Cochrane’s Q statistics, and inconsistency was quantified with the I² statistics to estimate the proportion of total variation across studies due to heterogeneity rather than chance. Quartiles of <30%, 30%–59%, 60%–75%, and >75% were suggestive of low, moderate, substantial, and considerable heterogeneity, respectively²⁵. Qualitative assessment of publication bias was conducted through visualization of funnel plot²⁶.

Data Availability Statement

All the data used in this paper were extracted from the referenced articles (Appendix 2), results were updated/clarified by corresponding trial results in clinicaltrials.gov, and they are publicly available. Moreover, the number of events in the iMiDs and PIs and control groups are described in the forest plots. Statistical codes used for this meta-analysis are publicly available through the R package meta.

Standard Protocol Approvals, Registrations, and Patient Consents

There were no direct human participants or any live vertebrates or higher invertebrates in this study which would require approval from the Institutional Review Board or the institutional animal experiment licensing committee. For a similar reason, no written informed consents were required. We have not used any recognizable personal patient information in this study. This study does not report any particular clinical trial.

Results

Overall, 4,473 citations were identified by the search, and 290 potentially eligible articles were retrieved in full texts. A total of 20 RCTs were identified and included for quantitative synthesis in our meta-analysis, which randomized 10,373 MM patients (Figure 1). Eleven studies compared iMiDs against control, seven studies compared PIs with control, and two studies compared bortezomib against carfilzomib (Appendix Table 2A, 2B). While eleven clinical studies reported high-grade cardiotoxicity (grade 3–5) as adverse events with iMiDs, seven clinical trials reported high-grade cardiotoxic events with PIs. The mean age of the patients in the treatment versus control group was 63.8±6.7 years vs. 67±7.6 years in the iMiD clinical trials (p=0.27 between groups) and 63.7±4.3 years vs. 63.6±4.2 years in the PI trials (p=0.07 between groups). The proportion of men in RCTs was similar between treatment and control groups in iMiD (52.7% vs. 54.7%) and PI trials (38.5% vs. 35.8%). The follow-up duration in the iMiD trials was 30.7 months and in the PI trials was 23.1 months.

Overall, the risk of bias was low to moderate for all clinical trials. A detailed account of the risk of bias assessment of individual studies is illustrated in Appendix Table 2A, 2B.

Cardiotoxicity with iMiDs

High-grade cardiotoxic events (CTCAE grade 3–5) occurred in 747 out of the 2733 MM patients receiving additional iMiDs in one armand 545 of 2727 MM patients not receiving any iMiDs in the control group (OR 2.05; 95% CI 1.30–3.26, low heterogeneity, I²= 10%, p-value =0.35) (Figure 2A). Any grade cardiotoxicity was also noted more frequently with PI use when compared to the no PI group (OR 1.47; 95% CI 1.19–1.82, low heterogeneity, I²= 0%, p-value =0.65) (Figure 2B). Similar difference was observed between high-grade cardiotoxic complication (CTCAE grade 3–5) between the PI and no-PI group (OR 1.67; 95% CI 1.17–2.40, low heterogeneity, I²= 0%, p-value =0.92) (Figure 2C).

Figure 2.

A. Forest Plot demonstrating the comparative effects of iMiDs in the risk of high-grade cardiotoxicity. B. Forest Plot demonstrating the comparative effects of PIs in the risk of all-grade cardiotoxicity. C. Forest Plot demonstrating the comparative effects of PIs in the risk of high-grade cardiotoxicity.

PIs compared against each other

The network of eligible comparisons for all-grade cardiotoxicity in MM patients treated with individual PI compared to control is shown in Figure 3A. Overall, in comparison to control, only carfilzomib (OR=2.68, 95% CI 1.63–4.40) showed increase rates of all-grade cardiotoxicity, while bortezomib (OR=1.18, 95% CI 0.73–1.92) and ixazomib (OR=1.56, 95% CI 0.84–2.90) did not have any significantly increased risk (Figure 3B). All comparisons for the all-grade cardiotoxic events among the PIs in the network meta-analysis are presented in the league table Figure 3C. The rank probability of cardiotoxicity for all PIs in MM patients is highest with carfilzomib, followed by ixazomib and bortezomib, respectively in the order of decreasing potential as shown in Appendix Table 3. Additionally, assessment of direct and indirect evidence in the network meta-analysis is conducted using the GRADE framework, as shown in Appendix Table 4. A network meta-analysis between different iMiDs could not be undertaken due to the absence of any head-to-head comparison of cardiotoxic events between any two iMiDs in clinical trials.

Figure 3.

A. Network meta-analysis for cardiotoxicity in with individual PI with control. B. Forest Plot demonstrating the comparative effects of the individual PI in the risk of cardiotoxic events. C. League table demonstrating the effects of the individual PI contributing to the risk of cardiotoxicity in MM patients.

Subgroup analysis

We conducted a meta-regression analysis on the duration of follow-up of the patients, age and sex of the patients in the RCTs. Although the duration of follow-up can be accounted for the cardiotoxicity noted in the iMiDs (p=0.01), no such difference was observed withPIs (p=0.46) (Appendix Table 5) between the treated and untreated groups. No difference was observed in cardiovascular events with age in iMiDs (p=0.58) and PIs (p=0.52). However, there was difference in cardiovascular outcomes with gender in iMiD group (p=0.025) and no such difference in PIs (p=0.78) based on gender.

Small-study effects

The number of studies with all-grade cardiotoxicity with PIs was 9 (<10), precluding us from performing meaningful small-study effects assessment. Quantitative measurement using the weighted linear regression of the treatment effect and there was no evidence of publication bias in clinical trials involving iMiDs (p = 0.02) and PIs (p = 0.19).

Discussion

Multiple myeloma patients are at a higher risk of developing CVDs due to common underlying risk factors^10,27. Current guidelines recommend the use of iMiDs and PIs as first-line drugs in the treatment of MM^8,9. Burgeoning evidence points towards the cardiotoxic effects of these two groups of drugs in MM patients, further contributing to the CVD risk in these patients^11,12,28. To the best of our knowledge, this is the first meta-analysis to demonstrate the cumulative cardiotoxic effects of iMiDs and perform a network meta-analysis to rank the PIs based on their cardiotoxicity.

Our results demonstrate that the odds of developing high-grade cardiotoxicity with the use of iMiDs in MM patients is almost two folds higher than their untreated counterparts, independent of their well-documented thromboembolic potential. Although PIs also increase the risk of high grade cardiotoxic events in MM patients, our results highlight that the effect might not be class-specific, with carfilzomib demonstrating the most significant cardiotoxicity amongst them. Carfilzomib has been shown to cause cardiac dysfunction in preclinical animal models through inhibition of the AMPKα/mTORC1 pathway¹⁵. Evidence accrued also shows the greater selectivity of carfilzomib for β5 chymotrypsin, the pivotal subunit of proteasome, compared to bortezomib²⁹. Further, the binding of carfilzomib to the proteasome is irreversible compared to reversible binding by bortezomib and its lipophilic nature due to its chemical structure allowing for easier permeability in organs compared to other PIs^30–32. These may partly explain the cardiotoxic profile of carfilzomib; however, further research is required to glean insight into the mechanism governing their cardiotoxicity.

Our findings are significant in the context of personalized medicine. Since these two drugs are employed to treat MM patients, the cardiotoxic effects may be additive. A proper clinical evaluation to risk-stratify patients based on pre-existing cardiovascular risk factors may aid in identifying patients who will benefit the maximum from the drugs. Since carfilzomib has a more potent cardiotoxic profile compared to the other PIs, suitable combination of drugs should be administered in MM patients with pre-existing cardiovascular risk factors to decrease the additional risk of developing cardiovascular events.

These two classes of drugs act by targeting the ubiquitin-proteasome system (UPS) protein degradation machinery of the cells through different mechanisms³³. While iMiDs facilitate the degradation of protein Cereblon by targeting E3 Ubiquitin Ligase (a component of UPS), PIs inhibit the proteasome activity itself to reduce the degradation of cellular proteins^34,35. Hence, the convergent cardiotoxic effects of iMiDs and PIs, especially carfilzomib, suggest proteotoxic stress as the consequence of targeting the components of the UPS pathway. Although the oxidative stress in myocardium and vasculature along with anti-angiogenic effect have been proposed as plausible causes in preclinical models^36,37, proper mechanistic understanding remains lacking. Future studies aimed at gleaning the underlying pathophysiology of cardiotoxicity with iMiDs and PIs, especially carfilzomib, can aid in developing therapies aimed at treating these adverse events. Furthermore, it may also aid in developing novel iMiDs and PIs with minimal to no adverse cardiovascular toxicity profiles without compromising its efficacy for primary cancers.

Despite the size of this meta-analysis, our study has certain limitations. First, this meta-analysis was performed at the individual study level but not on individual patient data. Therefore, confounding variables at the patient level such as dose of chemotherapy, the dose of PIs could not be incorporated into the analysis. Similarly, the impact of other risk factors for the development of cardiotoxicity such as diabetes, smoking status, obesity could not be ruled out. However, since we only included randomized clinical trials and compared exposure to no exposure, confounding would be lower than if observational studies were included. Secondly, data regarding the type and pattern of cardiotoxicity (infarction, arrhythmia, myocarditis) was limited. Whether an individual class of drugs has a predisposition towards specific types of cardiovascular events cannot be analyzed. Thirdly, participants with major adverse cardiovascular risk profiles are not included in most clinical trials; hence our analysis and results may not be generalizable in the general population of MM undergoing treatment. Further, proportion of males differ in between PI and iMiD arms of the trials, as a result of the variation in the participants in clinical trials. So potential gender bias cannot be ruled out in our study and the results should be interpreted with caution. This is especially true with iMiDs as there was dependence of cardiovascular outcomes to the gender of the treated population. Hence our results should be interpreted cautiously. Additionally, all the studies included in iMiDs trials had newly-diagnosed MM, while a majority of PI trials had patients with relapsed or treatment-refractory MM. Since the treatment failure patients received PI therapy on the backdrop of previous treatment with iMiDs, whether the cardiovascular adverse effects are a direct effect of PI or the long-term cardiotoxic effects of iMiDs remains obscure. Future prospective studies require to be undertaken to address this issue. Finally, our literature search was limited to published data and articles published in English only, creating some selection bias.

In conclusion, our results demonstrate that adverse cardiovascular profile in MM patients due to the treatment with iMiDs. It also highlights the increased scrutiny in the choice of PI in their treatment regimen to decrease the risk of the development of cardiotoxicity. Physicians also need to be aware and vigilant about the risk of cardiovascular events in MM patients treated with iMiDs and carfilzomib.