Evaluation of time-to-onset and outcome of cardiac adverse events related to pembrolizumab using post-marketing surveillance in Japanese patients

Abstract

Background

Pembrolizumab has been widely used in patients since its release, but information on cardiac Adverse Events (AEs) related to pembrolizumab remains lacking, particularly in Japanese populations.

Objectives

This study aims to evaluate time to onset, incidence rates, and outcomes for pembrolizumab-induced cardiac AEs in patients with cancer using the Japanese Adverse Drug Event Report database.

Methods

We analysed data for the period from April 2004 to March 2022. Data on cardiac AEs were extracted and relative risks of AEs were estimated using the reporting odds ratio.

Results

We analysed 2,021,907 reports and identified 15,306 reports of AEs caused by pembrolizumab. Of these, 399 cardiac AEs were associated with pembrolizumab. Signals were detected for six cardiac AEs: myocarditis, immune-mediated myocarditis, pericardial effusion, cardiac tamponade, pericarditis, and pericarditis malignant. A histogram of median times to onset showed occurrence from 33 (21–97) days for immune-mediated myocarditis to 138 (67–168) days for pericarditis malignant, but some cases occurred even more than 1 year after the start of administration. Among these, myocarditis was the most frequently reported (27.1%), with fatal cases also reported.

Conclusion

This study focused on cardiac AEs caused by pembrolizumab as post-marketing AEs. Patients should be monitored not only at the time of administration, but also over time for signs of these AEs, especially myocarditis, as some patients may have serious outcomes.

Graphical abstract

Introduction

Pembrolizumab, a humanized monoclonal antibody used in cancer immunotherapy as a programmed cell death protein-1 (PD-1) inhibitor [1], blocks the PD-1/PD-L1 interaction, thereby releasing the brakes on the immune system. This enables activated T cells to recognize and effectively attack cancer cells [1]. In Japan, pembrolizumab has been approved for the treatment of various cancers, including malignant melanoma, non-small-cell lung cancer, Hodgkin lymphoma, and other advanced solid tumours. Additionally, it is indicated for advanced or recurrent solid tumours with high levels of oncogene mutation (high tumour mutational burden) that have advanced after cancer chemotherapy (only if refractory to standard therapy). However, there are limitations compared to FDA-approved indications [2].

Pembrolizumab has been used in combination with chemotherapy and has reportedly prolonged both overall and progression-free survival in clinical trials [3–7]. Despite beneficial effects in treating numerous types of tumours, pembrolizumab can cause a variety of adverse events (AEs) [3–13]. Among these, cardiac AEs resulting from pembrolizumab are rare but concerning AEs due to the high mortality rates [8, 10–14]. Pembrolizumab has been widely used in patients since its release, but information on cardiac AEs from pembrolizumab remains lacking, particularly in Japanese populations. Identification of safety signals for chemotherapeutic agents using spontaneous AE reporting databases could represent a useful method for generating hypotheses regarding possible drug relationships with unknown or potential AEs. AE data obtained from clinical trials before a drug is approved represent information obtained from a relatively small population under a variety of constraints. On the other hand, after approval, repeated use in patients with diverse profiles may lead to the discovery of new potential AE trends. Pharmacovigilance to monitor drug safety is therefore central to the proper use of all drugs [15–17]. We conducted the present study to evaluate times to onset, incidence trends, and outcomes for pembrolizumab-induced cardiac AEs in patients with cancer. Analyses in this study were based on information obtained from the spontaneous reporting system in the Japanese Adverse Drug Event Report database (JADER), a large pharmacovigilance database provided by the Pharmaceuticals and Medical Devices Agency (PMDA).

Method

Data source

Healthcare professionals and pharmaceutical companies submit AE reports to the PMDA. Information was obtained from the JADER database on the PMDA website [18], which includes AE cases. All data from the JADER were fully anonymised by the supervisory authority before we accessed them. All methods were thus performed in accordance with the relevant guidelines and regulations. We analysed AE reports recorded between April 2004 and March 2022. The data structure of the JADER comprises four datasets: patient demographic information (DEMO), drug information (DRUG), AEs (REAC), and medical history (HIST). AEs in the JADER database are coded according to the preferred terminology of the Medical Dictionary for Regulatory Activities/Japanese, version 25.1 [19].

First, we removed duplicate cases from the DRUG and REAC tables. We then used the identification number of each AE case to integrate the matching case data from the DRUG, REAC, and DEMO tables. Drugs that contributed to AEs were classified as "suspected drugs," "concomitant drugs," or "interactions." Only cases classified as "suspected drugs" were selected for this study. We examined reports of spontaneous AEs that were sent to the JADER database to assess the relationship between pembrolizumab and cardiac AEs. Cardiac AEs coded according to the terminology recommended by the Medical Dictionary for Regulatory Activities were collectively referred to as cardiac AEs in this study.

Ethics approval was not required for this study, given the database-related, observational design without direct participation of any research subjects. All results were obtained from data openly available online from the PMDA website [18]. All data from the JADER were fully anonymised by the relevant regulatory authority before we accessed them. Thus, all methods were achieved according to the relevant guidelines and regulations.

Statistical analyses

Data on cardiac AEs with more than five reported cases were obtained, and the relative risk of AEs was calculated using the reporting odds ratio (ROR). ROR is frequently used in spontaneous reporting databases as an indicator of the relative risk of AEs. We used the analysis data table and constructed 2 × 2 tables based on two classifications: presence or absence of “cardiac AEs”, and presence or absence of suspected pembrolizumab use.

The ROR was calculated by dividing the incidence of AEs reported in association with pembrolizumab by the incidence of the same AEs reported in association with all other drugs in the database. The signal of an AE was considered positive if the lower limit of the 95% confidence interval (95% CI) of the ROR was > 1 [20].The time to onset of AEs was calculated and the number of cases was counted for reports in which the date of AE onset, date of starting administration, and date of ending administration were described as year/month/day or year/month [18]. As a rule, onset time was calculated as "(date of AE onset)—(date of start of administration) + 0.5" [21]. The time to onset of AEs for analysis was restricted to 2 years (730 days). The Weibull distribution is indicated by scale parameter α and shape parameter β. Scale parameter α indicates the scale of the distribution function, as the quantile in which 63.2% of AEs develop [22]. A large value for this parameter indicates a wide distribution, while a small value indicates a narrow distribution. Shape parameter β represents the change in hazard over time in the absence of a reference population. Depending on the value of shape parameter β, an upper limit of the 95% CI for β < 1 indicates that the hazard increases initially, then decreases (early failure type). Conversely, a β value of 1 or almost 1 and a 95% CI including 1 indicates that the hazard remains constant throughout the exposure period (random failure type), and a lower limit of the 95% CI for β > 1 indicates that the hazard increases over time (wear-out failure type). All statistical analyses were performed using JMP® Pro version 16.2 (SAS Institute, Cary, NC, USA).

Results

Incidence of cardiac AEs related to pembrolizumab

We joined the three tables of DRUG (4,134,536 reports), REAC (1,280,060 reports), and DEMO (775,566 patients) by ID number. We removed duplicate data from the DRUG and REAC tables [20]. As noted above, causes of AEs fall into three categories: "suspected drugs," "concomitant drugs," or "interaction." Of these, all data included in the category of "suspected drugs," were extracted and used as the "data table" (2,021,907 reports).

We analysed this data table and obtained 15,306 reports of AEs caused by pembrolizumab. Of these, 399 cardiac AEs were reportedly associated with pembrolizumab (Fig. 1). Patient characteristics are shown in Table 1. Approximately 75.4% of patients were male. According to the age distribution of the study population, cardiac AEs occurred most frequently among individuals in their 70 s (41.4%). We found that the following numbers of cases were reported for concomitant use of anticancer drugs that may cause cardiac AEs, such as anthracyclines or molecular targeted therapeutics: only doxorubicin (3 cases), trastuzumab (4 cases), and bevacizumab (10 cases). Smoking history or comorbidities (cardiovascular disease) were not known due to lack of data.

Fig. 1.

Process of constructing a data analysis table

Table 1.

Characteristics of patients exhibiting cardiac adverse events related to pembrolizumab

Variable	Value (%)
Number of patients	399
Sex
Male	301 (75.4)
Female	90 (22.6)
Unknown	8 (2.0)
Age
30 s	6 (1.5)
40 s	20 (5.0)
50 s	45 (11.3)
60 s	89 (22.3)
70 s	165 (41.4)
80 s	59 (14.8)
90 s	2 (0.5)
Unknown	13 (3.3)

RORs with a lower limit of the 95% CI > 1 comprised myocarditis (5.62, 95% CI 4.40 – 7.18, P < 0.001), immune-mediated myocarditis (49.90, 95% CI 34.82 –71.52, P < 0.001), pericardial effusion (4.39, 95% CI 3.18–6.07, P < 0.001), cardiac tamponade (5.51, 95% CI 3.69 – 8.22, P < 0.001), pericarditis (2.11, 95% CI 1.19 – 3.73, P = 0.018), and pericarditis malignant (39.34, 95% CI 15.80 – 97.99, P < 0.0001). Signals were thus detected for these six cardiac toxicities with five or more cases reported (Table 2).

Table 2.

Numbers of reports and RORs of cardiac adverse events related to pembrolizumab

Variable	Pembrolizumab		Other drugs		Rate (%)	ROR	95% CI	P-value
	Cases (n)	Non-cases (n)	Cases (n)	Non-cases (n)
Myocarditis	67	15,239	1,568	2,005,033	0.44	5.62	4.40–7.18	< 0.001
Cardiac failure	50	15,256	11,593	1,995,008	0.33	0.56	0.43–0.74	< 0.001
Immune-mediated myocarditis	41	15,265	108	2,006,493	0.27	49.90	34.82–71.52	< 0.001
Pericardial effusion	38	15,268	1,136	2,005,465	0.25	4.39	3.18–6.07	< 0.001
Cardiac tamponade	25	15,281	596	2,006,005	0.16	5.51	3.69–8.22	< 0.001
Atrial fibrillation	25	15,281	3,195	2,003,406	0.16	1.03	0.70–1.52	0.839
Cardiac arrest	17	15,289	3,451	2,003,150	0.11	0.65	0.40–1.04	0.077
Myocardial infarction	14	15,292	2,790	2,003,811	0.09	0.66	0.39–1.11	0.126
Hypertension	12	15,294	5,280	2,001,321	0.08	0.30	0.17–0.52	< 0.001
Pericarditis	12	15,294	747	2,005,854	0.08	2.11	1.19–3.73	0.018
Cardio-respiratory arrest	11	15,295	3,491	2,003,110	0.07	0.41	0.23–0.75	0.001
Cardiomyopathy	10	15,296	1,378	2,005,223	0.07	0.95	0.51–1.77	1.000
Arrhythmia	10	15,296	2,539	2,004,062	0.07	0.52	0.28–0.96	0.029
Acute myocardial infarction	9	15,297	2,251	2,004,350	0.06	0.52	0.27–1.01	0.051
Stress cardiomyopathy	9	15,297	849	2,005,752	0.06	1.39	0.72–2.68	0.318
Ventricular tachycardia	8	15,298	2,242	2,004,359	0.05	0.47	0.23–0.94	0.027
Cardiac failure congestive	8	15,298	2,898	2,003,703	0.05	0.36	0.18–0.72	0.001
Blood pressure decreased	8	15,298	12,746	1,993,855	0.05	0.082	0.04–0.16	< 0.001
Cardiac failure acute	7	15,299	2,138	2,004,463	0.05	0.43	0.20–0.90	0.017
Hypotension	7	15,299	4,233	2,002,368	0.05	0.22	0.10–0.45	< 0.001
Pericarditis malignant	6	15,300	20	2,006,581	0.04	39.34	15.80–97.99	< 0.001
Angina pectoris	5	15,301	2,059	2,004,542	0.03	0.32	0.13–0.77	0.003

“Cases” indicates the number of reported cases of cardiac adverse events. ROR: reporting odds ratio; 95%CI: 95% confidence interval. Italicized P-values represent statistically significant results. We obtained data for each type of cardiac adverse event showing more than five reports. All analysed data were obtained from the Japanese Adverse Drug Event Report database

Time to onset of cardiac AEs related to pembrolizumab

A histogram of the median times to onset for these six detected cardiac AE signals showed that AEs occurred from 33 to 138 days after pembrolizumab administration (Fig. 2). Median time to onset was 36 days (interquartile range [IQR] 20–74 days) for myocarditis, 33 days (IQR 21–97 days) for immune-mediated myocarditis, 45 days (IQR 19–100 days) for pericardial effusion, 71 days (IQR 36–154 days) for cardiac tamponade, 71 days (IQR 29–296 days) for pericarditis, and 138 days (IQR 67–168 days) for pericarditis malignant caused by pembrolizumab. The Weibull distribution of histograms for time to onset showed that the range of 95% CIs for shape parameter β of myocarditis, immune-mediated myocarditis, pericarditis, cardiac tamponade, pericardial effusion, and pericarditis malignant were all β = 1 (Table 3).

Fig. 2.

Histogram of cardiac adverse events for: 1) myocarditis, 2) immune-mediated myocarditis, 3) pericardial effusion, 4) cardiac tamponade, 5) pericarditis, and 6) pericarditis malignant. For each case, the date of onset of AEs is plotted on the horizontal axis. Dots indicate cases

Table 3.

Medians and Weibull parameters of cardiac adverse events

Adverse effects	Cases (n)	Median (day) (25–75%)	Scale parameter	Shape parameter
			α (95%CI)	β (95%CI)
Myocarditis	46	36 (20–74)	58.82 (41.28–82.71)	0.89 (0.71–1.10)
Immune-mediated myocarditis	32	33 (21–97)	63.76 (41.56–95.97)	0.90 (0.68–1.16)
Pericardial effusion	24	45 (19–100)	66.73 (44.46–97.88)	1.12 (0.79–1.51)
Cardiac tamponade	16	71 (36–154)	118.76 (69.20–197.28)	1.05 (0.70–1.49)
Pericarditis	11	71 (29–296)	117.35 (48.56–268.95)	0.81 (0.47–1.27)
Pericarditis malignant	5	138 (67–168)	188.25 (84.57–402.52)	1.45 (0.66–2.58)

“Cases” indicate the number of reported cases of cardiac adverse events. 95%CI: 95% confidence interval. Detected cardiac adverse event signals were analysed to determine the time to onset

Outcomes after occurrence of cardiac AEs related to pembrolizumab

The percentages of outcomes (recovered, remission, not recovered, with sequelae, death, or unclear) after onset of the six AEs are shown in Fig. 3. Fatal outcomes were observed for myocarditis, immune-mediated myocarditis, pericardial effusion, cardiac tamponade, and pericarditis.

Fig. 3.

Percentage of six cardiac adverse events associated with pembrolizumab by outcome

Discussion

This study focused on post-marketing pembrolizumab-induced cardiac AEs and suggested that those AEs for which signals were detected were myocarditis, immune-mediated myocarditis, pericardial effusion, cardiac tamponade, pericarditis, and pericarditis malignant.

Among these, myocarditis (including immune-mediated) was the most frequently reported (27.1%, 108/399), with fatal cases also reported. Myocarditis (including immune-mediated) caused by pembrolizumab has been reported in clinical trials [23, 24], and the present results are consistent with clinical findings. The Weibull distribution showed that both myocarditis and immune-mediated myocarditis developed at a constant rate after pembrolizumab administration. However, median times to onset for myocarditis and immune-mediated myocarditis in this study were 36 days (IQR, 20–74 days) and 33 days (IQR, 21–97 days), respectively. This represented an earlier onset similar to previous studies (after 1–2 cycles of therapy) [10, 25–31]. Clinicians should thus pay close attention to the onset of myocarditis (including immune-mediated) right after starting pembrolizumab administration. Although the incidence of myocarditis did not increase in a dose-dependent manner in this study, continuous monitoring is recommended throughout the entire treatment period, as several cases were observed after almost 2 years of treatment. In routine clinical practice, myocarditis due to immune checkpoint inhibitors in particular can sometimes develop asymptomatically and follow a rapid course of deterioration. That is, AEs that develop due to pembrolizumab are not always suspected solely on the basis of symptoms or vital signs, but may also be detected during routine examination. In particular, myocarditis can be asymptomatic and can lead to death [30–32]. Identifying cardiac AEs related to pembrolizumab as a system is therefore important. In particular, because myocarditis can be asymptomatic and fatal, periodic checks of baseline cardiac function (e.g., electrocardiogram, cardiac troponin) in addition to the usual observation of patient symptoms (palpitations, shortness of breath/dyspnoea, chest tightness/chest pain, etc.) are recommended whenever feasible. In any case, careful follow-up by a cardiologist is considered necessary.

Fatal outcomes were observed for 5 AEs (all except pericarditis malignant) for which signals were detected. Among the reported cases, a histogram of time to onset showed that these AEs occurred at a median time of 33 days (IQR, 21–97 days) for immune-mediated myocarditis to 71 days (IQR, 29–296 days) for pericarditis after the first pembrolizumab administration. The present results were consistent with the findings of previous studies that fatal AEs tend to occur in the early stages of administration [25–32]. The immediate diagnosis and treatment of immune-related cardiac AEs is thus crucial for achieving good outcomes. However, pericardial effusion can still arise more than 1 year after first administration (at a constant rate). Among the 5 AEs for which fatal outcomes were observed, pericarditis was also associated with a high rate of death, along with myocarditis. Median time to onset in this study was 71 days (IQR, 29 – 296 days) for pericarditis, so early detection and continuous monitoring are necessary. Since cardiac AEs caused by pembrolizumab are frequently associated with fatal outcomes, clinicians should pay close attention to clues to their development over an extended period and not just at the start of administration. It is also important to educate patients thoroughly on the signs and symptoms of cardiac AEs and encourage them to report any unusual symptoms promptly. Maintaining vigilance in monitoring and managing potential AEs is crucial for ensuring patient safety [8, 9, 14, 33, 34].

The mechanisms underlying pericardial effusion, cardiac tamponade, and pericarditis are poorly understood and pseudo-progression cannot be ruled out [35, 36]. Pembrolizumab is used in various cancers. The cardiovascular system is among the systems least frequently impacted by immunotoxicity, so very few cases of immune- related pericardial effusion have been reported with anti PD-1 immunotherapy. Therefore, despite the low incidence of cardiac AEs, it is crucial to focus on the prevention and management of immunotherapy with immune checkpoint inhibitors [9], which include PD-1 inhibitors like pembrolizumab and nivolumab, PD-L1 inhibitor atezolizumab, and CTLA-4 inhibitor ipilimumab [11, 14, 37]. Despite the availability of general guidelines in dealing with cardiovascular immune-related AEs, further research establishing more specific guidelines is necessary to better deal with this complex entity.

In this study, cardiac failure, hypertension, cardio-respiratory arrest, arrhythmia, ventricular tachycardia, cardiac failure congestive, blood pressure decreased, cardiac failure acute, hypotension and angina pectoris were not detected as signals (ROR < 1), suggesting that these AEs were not associated with cardiac AEs related to pembrolizumab. However, Mascolo et al. reported in their study using spontaneous reports in Eudravigilance that not only myocarditis and pericardial effusion, but also heart failure, atrial fibrillation, and myocardial infarction are major immune-related cardiac AEs [37]. This point needs to be further verified, but may be attributable to ethnic differences between Japanese and European populations.

Limitations

The present results have to be considered in light of some limitations. First, the JADER is based on self-reported data, which may contribute to reporting biases such as over- or underreporting. That is, serious AEs may have been reported preferentially. Second, the lack of comprehensive medical records and medication histories limits the scope of analyses, as dosages, durations, clinical laboratory data, severity of AEs, and more information on concomitant medications of pembrolizumab use were unavailable. Third, potential confounding, selection, and information biases cannot be fully excluded from this study. Indeed, the majority of cases involved men (75.4%) and the most common age group affected was the 70 s (41.4%). Fourth, we could not exclude the possibility that AEs may have been caused by concomitant use of anticancer drugs. However, the results of this study were based on extracted data in which pembrolizumab was judged as the suspect drug by a reporter (physician or pharmacist) who knew the details of the clinical course in the individual case. Our report provides useful information for monitoring cardiac AEs associated with pembrolizumab.

Conclusion

We focused on cardiac AEs caused by pembrolizumab as post-marketing AEs. Myocarditis could potentially result in serious outcomes after starting pembrolizumab. In particular, myocarditis was suggested to show earlier onset in post-marketing surveillance compared to clinical trial data. Patients should be monitored for signs of onset of these AEs, using regular examinations not only at the start of administration, but also over the long term. Additionally, clinicians must maintain vigilance for potential cardiac AEs and collaborate closely with cardiologists and other specialists in managing these rare AEs [14].

Author contributions

Y.K.: Data curation; Writing – original draft; Writing – review and editing. E.T.: Data curation; Writing – review and editing. T.S. and M.U.: Conceptualization; Supervision; Writing – review and editing.

Funding

No specific funding was received.

Data availability

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

All authors declare no conflict of interest.