Multiple myeloma (MM), despite rare, is the second most common hematologic malignancy worldwide, with estimated 159 985 new cases and 106 105 deaths in 2018 (1). MM remains uncurable but the survival of MM patients has substantially improved with the clinical development of novel therapeutics (2). Proteasome inhibitor (PI) has been widely used in treating MM and become the backbone therapy for MM both in the first- and second-line settings (3). Carfilzomib, authorized in 2012, is a second-generation PI that irreversibly inhibits the chymotryptic site of the proteasome (4). Importantly, carfilzomib is the first and only MM treatment that prolongs overall survival in relapsed settings over the current standard therapy (5).

Carfilzomib has a more favorable safety profile than other PI agents and is used with increasing frequency in clinical practice to treat MM sufferers. However, of the PIs used, carfilzomib is the most strongly associated with cardiotoxicity (6). Although most cardiovascular complications are reversible, in some cases they may lead to life-threatening and even fatal complications (7, 8). Whereas, evidence based on real-world data is sparse and the cardiovascular safety profile of carfilzomib remains incompletely characterized. Therefore, we conducted a pharmacovigilance study based on the FDA Adverse Events Reporting System (FAERS) database to explore the real-world pattern of cardiovascular diseases occurring secondary to carfilzomib in detail. Our study may update the currently described cardiovascular complications of carfilzomib, as well as bring important clues for further study.

Carfilzomib is an effective PI agent with an overall tolerable safety profile for the treatment of relapsed or refractory MM (20). The advent of this agent has contributed to significant survival improvements in this challenging population and it is now used with increasing frequency (21). However, there is growing evidence that carfilzomib use is linked with higher than the expected rate of cardiovascular risks. Since being approved in 2012, there has been a large body of articles reporting cardiovascular complications associated with this agent, encompassing cardiac failure, cardiomyopathy, hypertension, cardiac arrhythmias, ischemic heart disease, etc., (21–23). However, details of these cardiovascular events among the general population remain unclear. There is growing recognition of the need to provide a comprehensive understanding of the cardiovascular toxicity profile after carfilzomib administration from real-world evidence.

To our knowledge, the current study presents the most extensive and exhaustive characterization of carfilzomib-associated cardiovascular toxicities based on the FAERS database. It is a detailed and important examination of the currently available observational data, with the potential to improve management in affected patients. A large number of records were involved, enabling us to conduct a systematic analysis of the frequency, spectrum, clinical features, timing, and outcomes of carfilzomib-related cardiovascular toxicities. We have enumerated the notable and interesting findings as below:

First, cardiovascular records accounted for a proportion of 14.84% of all carfilzomib-associated records, which was similar to the incidence rate (18.1%) reported in an extensive meta-analysis (8). Importantly, our study suggested that carfilzomib use was associated with a significantly increased risk of cardiovascular AEs, which was evident by the higher disproportionality in contrast with the whole database. Since carfilzomib is mostly prescribed for MM and it has been reported that many patients with MM have cardiovascular comorbidities or relevant risk factors (24, 25). Therefore, we conducted an additional disproportionality analysis in this subpopulation. Results are shown in Table S9. In general, despite being undermined, the disproportionality association between cardiovascular events and carfilzomib administration remains significant (IC₀₂₅/ ROR₀₂₅ = 0.46/1.46). There are several published studies supporting that augmented cardiovascular risk might be associated with carfilzomib utilization. A large meta-analysis has reported that carfilzomib utilization significantly increased the risk of cardiotoxicity and arterial hypertension (26). Similarly, a study pooling data across phase 1–3 trials with 2000 patients exposed to carfilzomib demonstrated that patients receiving carfilzomib had an increased incidence of cardiovascular events (27). Subsequently, Waxman et al. performed the first meta-analysis investigating the potential association of carfilzomib with cardiovascular AEs, concluding that carfilzomib was associated with an augmented risk of high-grade and all-grade cardiovascular events compared with noncarfilzomib-receiving controls (8). Currently, the precise pathogenesis of cardiovascular events in carfilzomib receivers has not been completely understood, though several underlying mechanisms included endothelial effects, proteasome inhibition, and cardiorenal syndrome were proposed (28, 29). More mechanistic studies are needed in the future to unravel the mechanism of the carfilzomib-associated cardiovascular toxicity.

Second, we noticed that the eight broad entities of cardiovascular toxicity were all disproportionately associated with carfilzomib when data mining in the entire database, ranging from cardiac arrhythmias to cardiac failure. When further exploring these toxicities in the MM-related indication subset, only signal of embolic and thrombotic events disappeared, while others were detected as disproportionality signals in line with our main analysis (Table S9). Taken together, it indicated that carfilzomib-associated cardiovascular events were highly versatile, which was largely consistent with those being reported in the two extensive reviews (29, 30). Cardiomyopathy-related complications were the most common reported cardiovascular AEs in our study regardless of the use of the whole dataset or the MM-related indication subset, which is not in accordance with previous literature that suggested hypertension (31, 32) was the most prevalent. In our main analysis, cardiac failure occurred in 3.37% of all carfilzomib-associated records, approximately to 4.1% in Waxman et al.'s study (8), which is highlighted with the strongest disproportionality. Notably, a general increasing trend of the signal was captured across all categories over time using the entire database, which needs to be paid attention. With further data mining in different subpopulations, cardiac failure was over-reported in all subgroups. Data from a recent meta-analysis included four phase III RCTs also supports the elevated risk of cardiac failure secondary to carfilzomib usage (32).

Third, we observed that the spectra of cardiovascular toxicities differed in subpopulations. It is worthwhile to mention that several carfilzomib-associated cardiovascular complications tend to occur in specific populations. We found that women were disproportionately affected by embolic and thrombotic events, regardless of the age; by contrast, older patients were disproportionately affected by ischemic heart disease, regardless of sex. These findings suggested that age and sex might be important biological variables and deserve more attention. Continued efforts are needed concerning the identification and quantification of the sex and age differences, as well as to explore the underlying mechanisms and to monitor and manage the affected patients more efficiently.

Fourth, despite several studies have noted that carfilzomib-associated toxicities occurred early (usually within two cycles), the time-to-onset of these events after administrating carfilzomib is still unclear (3, 6, 33). Our study partially fill the knowledge gap and provide useful information for future studies. Note worthily, we found that cardiovascular toxicities tended to occur early after giving carfilzomib, with a median onset time of 41 days and majority of events occurring within 3 months. Importantly, we described the differential spectra of time-to-onset according to different types of cardiovascular AEs. We observed that median time to onset occurred fairly early for ischemic heart disease (16 days, Q1–Q3: 4–85 days), and was the most delayed for embolic thrombotic events (68 days, Q1–Q3: 21–139 days). It is worthwhile to recognize the difference of time-to-onset between different types of cardiovascular toxicities; furthermore, our findings underscore a need for early and closely follow-up after carfilzomib treatment, particularly in the first 3 months, to ensure an early intervention in the affected population.

Fifth, we noticed that death accounted for 16.35% of all associated records, implying the significant impact that cardiovascular complications imposed on the patient mortality. Additionally, we evaluated and compared outcomes according to different types of cardiovascular events. It turned out that the severity of these events varied in categories. Given the potential hyperacute onset of these diseases, early and intensified monitoring is particularly necessary. It has been recommended to hold carfilzomib until toxicity reducing to grade <2 when cardiovascular AEs occur, and if the decision is made to re-challenge the patients, a dose reduction may be considered (33). Animal model studies have noted several new agents to potentially attenuate and counteract carfilzomib-induced cardiovascular toxicity (34, 35); nevertheless, whether these agents are useful in clinical practice needs further investigation.

Last but not least, in our study, Torsade de pointes/QT prolongation was identified as a new signal linked to carfilzomib administration (IC₀₂₅/ROR₀₂₅ = 0.33/1.29) and was particularly notable with the highest death proportion (44.11%). Torsades de pointes is a specific form of ventricular arrhythmia associated with prolongation of the corrected QT interval and very frequently due to drug administration (36). It is rare but constitute a significant problem because of sudden cardiac death. Despite there have been several published studies regarding the cardiovascular toxicities after carfilzomib, this agent is not known to cause Torsade de pointes/QT prolongation according to previous studies. We postulated that this may be related to the rarity and high mortality of this toxicity. Compared with existing studies, the strength of enormous records at a national level supports our study to explore several rare cardiovascular events and quantify all possible signals. Our study might supplement existing literatures by revealing this novel potential cardiovascular toxicity following carfilzomib use. We hope our work can inspire future studies to gain a deeper insight into this newly identified signal.

The major strength of the presented study is that enormous records involved at a national level from the real-world setting, which enables a better evaluation of risk AEs and broader generalization of our findings. However, several limitations in the study should also be acknowledged. First, there is an inevitable and unquantifiable bias intrinsic to the use of spontaneous data, such as over-reporting, under-reporting, and incomplete information. Second, FAERS does not provide detail clinical information regarding the severity of AEs which might contribute to a better comprehensive analysis. Third, most of the reporting data in FAERS database come from America and Europe countries. Apart from Japan, other Asian countries have very few records in the database compared with America and Europe countries, which would lead to geographic bias of the results. Forth, the exact number of patients exposed to carfilzomib in the FAERS cannot be evaluated, therefore, we are unable to determine AE incidence. Fifth, changing awareness of toxicities over time could affect reporting behaviors (17). Sixth, despite disproportionality methods are efficient in signal detection, their shortcomings in dealing with confounding like masking effect and co-prescription should also be recognized (37). It is important to note that the significant disproportionality in this study did not support a causal relationship; instead, it indicates a potential safety signal which should be followed up and validated by prospective studies.

The exhaustive pharmacovigilance analysis contributes to the cumulative knowledge about the cardiovascular safety of carfilzomib. Several toxicities are highlighted and deserve further attention, such as cardiac failure with the strongest disproportionality association, ischemic heart disease with fairly early onset time and the highest proportion of life-threatening conditions, and Torsade de pointes/QT prolongation with the highest death proportion.

Publicly available datasets were analyzed in this study. This data can be found here: https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html.

YZ, XY, and FH: conception and design. JH: administrative support. JX, XG, and ZL: collection and assembly of data. YZ and FH: data analysis and interpretation. YZ: manuscript writing. XY, FH, YC, ZG, and XZ: manuscript reviewing and revising. All authors approved the final manuscript.

This study was supported by the National Nature Science Foundation of China (No. 82073671), the Leading Talents of Public Health in Shanghai (No. GWV-10.2-XD22), the Shanghai Municipal Commission of Health and Family Planning Fund for Excellent Young Scholars (No. 2018YQ47), and the Excellent Young Scholars of public health in Shanghai (No. GWV-10.2-YQ33), three-year Action Program of Shanghai Municipality for Strengthening the Construction of Public Health System (GWV-10.1-XK05) Big Data and Artificial Intelligence Application, and Military Key Discipline Construction Project (Health Service–Naval Health Service Organization and Command) (No.03).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.