Assessment of Cardiovascular Events Caused by New-Generation Androgen Receptor Pathway Inhibitors Used for Prostate Cancer: A Real-World Study in Japan

Prostate cancer is the most commonly diagnosed cancer among males. Globally, 1,414,259 new cases and 375,304 deaths are predicted for 2020 [1]. Intravenous chemotherapy has been the standard treatment for castration-resistant prostate cancer, the most advanced form of the disease. Recently, however, a new class of androgen receptor pathway inhibitors (ARPIs), including enzalutamide and abiraterone acetate [2, 3], has been offered as an alternative treatment. According to a recent analysis [4], enzalutamide and abiraterone acetate accounted for approximately USD 2.7 billion in Medicare Part D costs.

In Japan, enzalutamide was approved for castration-resistant prostate cancer in March 2014, and abiraterone was approved in July 2014. Abiraterone is an androgen biosynthesis pathway inhibitor that specifically and efficiently blocks androgen synthesis in testicular, adrenal, and prostate tissues. Enzalutamide is an androgen receptor signaling inhibitor that competitively inhibits androgen binding to androgen receptors in prostate cells [5]. Clinical trials report that both greatly increase patients’ overall survival and progression-free survival [6‒9].

In clinical trials, abiraterone acetate and enzalutamide have demonstrated a good safety profile [8‒10]; however, they can cause major adverse events related to their mode of action [11]. Reports indicate a higher incidence of cardiac problems with abiraterone and a higher incidence of hypertension with enzalutamide [12, 13]. However, some studies report no difference in the occurrence of cardiovascular events [14], and the difference in risk of cardiovascular events between enzalutamide and abiraterone remains unclear. Cardiovascular events have been reported as the leading cause of death other than cancer itself in patients with prostate cancer (15–30%) [15].

Although a recent analysis of ARPI and cardiovascular events using the US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) has been reported [16]. FAERS is one of the largest pharmacovigilance databases in the world and is used to assess the safety and efficacy of drugs in clinical practice by providing necessary safety signals. However, since the cases registered in FAERS are based on spontaneous reports, the total number of drug users is not available to analysts, the incidence rate cannot be calculated, and the detected signals are subject to various reporting biases [17]. Consequently, the detected signals are hypotheses that require validation. Using a large Japanese database of medical institutions, the Japanese Medical Data Center (JMDC) medical institution database (JMDC Inc., Tokyo, Japan), this study tested the hypothesis that the risk of cardiovascular events with enzalutamide is lower than with abiraterone acetate.

This study used the JMDC medical institution database, which consists of claims and DPC survey data for all patients who visited contracted medical institutions from more than 600 medical institutions contracted with JMDC. This database collects data regardless of the patient’s insurance coverage and includes a significant number of patients aged 65 years and older, and a population rarely included in previous claims databases [18]. The International Classification of Diseases, Tenth Edition (ICD-10) codes were used to record the subscriber’s data, including sex, date of birth, and diagnosis.

The database includes data from April 2014 to August 2022. To ensure at least 6 months of follow-up before and after drug use, the study covered the period from October 2014 to February 2022. Patients with new use of ARPIs between October 2014 and February 2022 were included. New use was defined as the initiation of any drug by a patient who had not used any ARPI previously. Exclusion criteria (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000540864) were patients with a major adverse cardiovascular event in the past, those who used both abiraterone acetate and enzalutamide during the follow-up period, and those who used the drugs for less than 1 month (Fig. 1). Monthly data were used for disease diagnosis. Patients who took the drug for less than 1 month and developed cardiovascular disease in the same month were excluded, as they may have developed the disease before starting the drug.

The potential confounding factor was controlled using the propensity scores (PS) method. Logistic regression analysis was used to estimate the PS, and a nearest neighbor matching technique was applied at a 1:1 ratio. The matching caliber was fixed at 0.2 times the standard deviation of the logit score. Covariates included demographic factors, comorbidities, and medications. Demographic factors included age. Comorbidities included acute coronary syndrome, ischemic heart disease, cardiomyopathy, valve disorders, cerebrovascular disease, atrial fibrillation, other arrhythmia, diabetic complications, chronic obstructive pulmonary disease, other lung diseases, venous thromboembolism, non-prostate cancer, liver disease, rheumatic disease, psychiatric disorders, fractures, arterial disease, and renal disease. Medications included drugs for diabetes, hypertension, nitrate, platelet inhibitors, anticoagulants, lipid-lowering agents, antidepressants, antipsychotics, anxiolytics, hypnotics or sedatives, beta-2 agonist inhalers, anticholinergic inhalers, glucocorticoid inhalers, nonsteroidal anti-inflammatory drugs, and opioids (online suppl. Table S2). Comorbidities were defined as illnesses diagnosed in the previous 6 months, and drugs were defined as those used in the past 6 months.

The primary outcomes were cardiovascular death and all cardiovascular events. Secondary outcomes included major cardiovascular events (defined as myocardial infarction [MI] and stroke) and individual endpoints of MI, heart failure (HF), and stroke. These outcomes were identified from the primary diagnosis at admission (online suppl. Table S3), as registered in the JMDC medical institution database. Cardiovascular death was defined as death with a diagnosis of a cardiovascular event.

This retrospective cohort study used a “new user” design, focusing on patients who initiated treatment with abiraterone acetate or enzalutamide. Patients were followed from the start of treatment to outcome events unless they switched to control drugs or the observation period ended. Cox proportional hazard regression was used to estimate the hazard ratio (HR) from the start of the treatment, with the time axis. EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria), was used for all statistical analyses in this study [19].

Overall, 3,033 new users of enzalutamide and 2,021 new users of abiraterone acetate met the study eligibility criteria (Fig. 1). Online suppl. Table S4 lists the baseline characteristics of the cohort before propensity score matching. After 1:1 matching, the cohort included 1,940 new enzalutamide users and 1,940 new abiraterone acetate users. Covariates between the two groups were well balanced (Table 1). The total follow-up period was 29,654 months (median: 11 months) for enzalutamide users and 26,505 months (median: 9 months) for abiraterone acetate users. The total follow-up period overall was 56,159 months (median: 10 months). The median follow-up period was longer for enzalutamide users than for abiraterone acetate users (11.0 months vs. 9.0 months, p < 0.001).

Figure 2 shows the cumulative incidence of cardiovascular death, all cardiovascular events, major cardiovascular events, MI, HI, and stroke. Compared with abiraterone acetate, enzalutamide was associated with a significantly lower risk of cardiovascular death (incidence rate: 0.49 events per 100 person-years vs. 0.54; HR: 0.30, 95% confidence interval [CI]: 0.10–0.93), all cardiovascular events (incidence rate: 6.80 events per 100 person-years vs. 9.16; HR: 0.79, 95% CI: 0.64–0.98), major cardiovascular events (incidence rate: 6.66 events per 100 person-years vs. 9.07; HR: 0.79, 95% CI: 0.64–0.97), and MI (incidence rate: 2.88 events per 100 person-years vs. 4.48; HR: 0.62, 95% CI: 0.46–0.84). There were no significant differences in HF (incidence rate: 0.12 events per 100 person-years vs. 0.09; HR: 0.95, 95% CI: 0.19–4.71), and stroke (incidence rate: 4.42 events per 100 person-years vs. 5.03; HR: 1.01, 95% CI: 0.78–1.32) (Table 2; Fig. 3).

Recently, a disproportionality analysis of ARPIs and cardiovascular events using the FAERS was reported [16]. FAERS is one of the largest pharmacovigilance databases in the world and is used to assess the safety and efficacy of drugs in clinical practice by providing the necessary safety signals. However, since the cases registered in FAERS are based on spontaneous reports, the total number of drug users is not available to the analysts, the incidence rate cannot be calculated, and the detected signals are subject to various reporting biases due to spontaneous reporting [18], the detected signals are hypotheses that require validation. However, evidence on the real-world incidence of cardiovascular events with abiraterone acetate and enzalutamide remains insufficient. Therefore, evaluating these drugs in clinical practice using case data is crucial to identifying differences in cardiovascular event incidence and risk. This study used individual patient-level data from a claims-based database to compare the incidence and risk of cardiovascular events among users of abiraterone acetate and enzalutamide in patients with prostate cancer. After propensity score matching, the mean patient age was 76.7 ± 7.92 years for all patients, 76.7 ± 7.91 years for enzalutamide users, and 76.7 ± 8.10 years for abiraterone acetate users. The HR and 95% CI for each adverse cardiovascular event were as follows: cardiovascular death (0.30, 95% CI: 0.10–0.93), all cardiovascular events (0.79, 95% CI: 0.64–0.98), major cardiovascular events (0.79, 95% CI: 0.64–0.97), MI (0.62, 95% CI: 0.46–0.84), HF (0.95, 95% CI: 0.19–4.71), and stroke (1.01, 95% CI: 0.78–1.32). These results indicate that enzalutamide is associated with a significantly lower cumulative incidence of cardiovascular death, all cardiovascular events, major cardiovascular events, and MI compared to abiraterone acetate.

Abiraterone acetate and enzalutamide have not been directly compared in a phase III randomized clinical trial; the only phase II trial addressing this comparison involved newly diagnosed metastatic castration-resistant prostate cancer [20]. In the AFFIRM and PREVAIL trials, enzalutamide did not significantly increase the incidence of high-grade cardiovascular events compared to placebo [8, 9]. Furthermore, while enzalutamide did not consistently raise the risk of cardiovascular events, abiraterone acetate was consistently associated with an increased risk compared to placebo in a meta-analysis of randomized controlled trials [12, 13].

A report by Kulkarni et al. [21] in their risk assessment of acute MI or stroke using the MarketScan claims databases (from January 1, 2013, to September 30, 2018) showed a 31% increased risk of MI or stroke with abiraterone acetate use compared to enzalutamide.

The antiandrogenic effect of abiraterone acetate results from the selective and irreversible inhibition of 17α-hydroxylase/17,20-lyase (CYP17A1), leading to decreased androgen synthesis. Inhibition of CYP17A1 decreases cortisol levels and increases aldosterone synthesis due to negative feedback on adrenocorticotropic hormone secretion. Increased aldosterone levels can cause hypokalemia, fluid retention, and edema, which may predispose patients to HF [22]. Current treatment guidelines recommend prescribing abiraterone acetate with steroids to reduce mineralocorticoid toxicity. However, although steroids can reduce cardiovascular risk, they may not completely counteract the effects of mineralocorticoid upregulation. Enzalutamide, in contrast, does not inhibit androgen synthesis or cause mineralocorticoid excess, making it less likely to induce hypokalemia, which predisposes to HF. However, no significant differences in cardiovascular risk were found in this study. In addition, serum testosterone levels, which are increased by enzalutamide [23], play an important role in sexual development, behavior, and body composition. Studies suggest that lower testosterone levels are associated with an increased risk of cardiovascular events and all-cause mortality [24‒26]. Therefore, increased serum testosterone levels with enzalutamide might reduce the incidence of cardiovascular events. Overall, the findings regarding cardiovascular risk with ARPIs in Japan are consistent with the results of previous studies and are pharmacologically valid.

The JMDC medical institution database used in this study consists of claims and DPC survey data for patients who visited contracted medical institutions. Unlike conventional claims databases, the JMDC medical institution database has the advantage that it can be collected regardless of the insurance that patients join; it covers elderly persons aged 65 and over in abundance. Conventional insurance claims databases have limited data for individuals aged 65 and older and no data for those aged 75 and older due to insurance [18]. Conversely, the JMDC medical institution database used in this study, although limited to contracted medical institutions, provided abundant coverage of individuals over age 65 and can be collected regardless of the patient’s insurance coverage. In addition, because the JMDC medical institution data includes non-DPC hospitals, the number of hospitals by hospital bed scale is close to the national average, and the data represent the entire country. This study focused on new users of abiraterone acetate or enzalutamide. Propensity score matching allowed for an adjusted comparison of baseline characteristics between males receiving abiraterone acetate and those receiving enzalutamide. Although concomitant steroid use is recommended with abiraterone, it is not typically mentioned for enzalutamide. In this study, 395 patients (20.4%) in the enzalutamide group and 1,933 patients (99.6%) in the abiraterone group used steroids concomitantly, which precluded their inclusion in the propensity score matching. In addition, patients using abiraterone were more likely to have underlying ischemic heart disease, potentially influencing clinicians’ treatment preferences beyond steroid use.

However, the effect is likely minimal as we adjusted for various diseases and drugs in the propensity score matching. A history of cancer was present in 67% of the matched patients (online suppl. Table S4), but this type of cancer was not considered as patients were matched only based on the presence or absence of cancer. Since both enzalutamide and abiraterone are indicated exclusively for prostate cancer, it is clear that patients in this study were treated for prostate cancer. Two significant confounding factors – body mass index and smoking status – could not be considered due to limitations in the database. Data on Eastern Cooperative Oncology Group performance status and frailty were also unavailable. Patients were assigned to a drug group based on their initial treatment unless they changed drugs; those who discontinued treatment were censored. Therefore, dosage variations, such as dose reductions or increases, were not considered. Furthermore, the accuracy of database records could impact internal validity due to the secondary use of claims data, and diagnostic accuracy may vary. The relationship between treatment response and cardiovascular events remains unclear due to a lack of data on cancer stage and clinical outcomes.

Overall, males newly treated with enzalutamide had a decreased incidence of adverse cardiovascular events than those treated with abiraterone acetate in this national prostate cancer sample. Based on these findings, healthcare professionals should be aware of the hazards associated with abiraterone acetate and enzalutamide, even as prostate cancer management typically focuses on cancer outcomes.

The Gifu Pharmaceutical University’s Ethical Committee approved this study (4–38) and complied with the Declaration of Helsinki. The requirement for informed consent was waived because all data in the JMDC medical institution database were deidentified.

The Endowed Course receives financial support from CHUBU YAKUHIN CO., LTD. The other authors have no conflicts of interest.

Yoshihiro Noguchi was supported by JSPS KAKENHI Grant No. 22K12890 in this study.

All authors wrote the manuscript. Yoshihiro Noguchi, Kazuhiro Iguchi, and Tomoaki Yoshimura designed the research. Rikuto Masuda, Kazuhiro Iguchi, Haruka Aizawa, Shunsuke Yoshizawa, and Tomoaki Yoshimura performed the research. Rikuto Masuda, Haruka Aizawa, and Shunsuke Yoshizawa analyzed the data. Yuki Nomura and Mitsuru Saguchi contributed analytical tools.

Use of the data supporting the results of this study is under license from JMDC Inc. Therefore, restrictions apply, and the data are not publicly available. To obtain access to the dataset utilized in this investigation, please reach out to JMDC Inc., https://www.jmdc.co.jp.