Reply to S. Rossi et al and P. Palassin et al

We thank both Rossi et al¹ and Palassin et al² for critically evaluating our recently published results³ and providing additional insights on immune checkpoint inhibitor (ICI)–associated major adverse cardiac events (MACEs).

Rossi et al¹ described an interesting observation regarding the chronological relationship of neurological symptoms preceding myocarditis in a small series of four patients with overlapping myocarditis, myositis, and signs of myasthenia gravis (MG). Three of these patients presented with serological testing positive for anti-acetylcholine receptor (AChR) and anti-titin antibodies, with the latter potentially representing a valuable biomarker for these patients per the authors.

In a review of the data reported in our article,³ we observed myasthenia-like symptoms, ie, dysphagia, muscle weakness resulting in shortness of breath, ptosis, or ophthalmoplegia in 4 of 18 patients with myocarditis. Two patients had positive AChR-binding antibodies while the two others were negative. One of these patients with positive antibodies had a normal electromyogram. No electromyogram documentation was available for the other three patients. Three of these four patients also had grade 3 or higher creatine phosphokinase elevation. Only one of the four patients presented with diplopia, which was noted 2 days before hospitalization. For the other three patients, clinical signs or symptoms suspicious of myasthenia were identified during their hospitalization. Given these limited data, it is difficult to conclude that neuromuscular symptoms precede cardiac involvement in any predictable fashion. To date, limited published data have demonstrated that one third of patients with immune-mediated MG (iMG) may not have serological testing positive for AChR antibodies.^4,5 Thus, in addition to serological testing, a comprehensive clinical neurological diagnostic approach is highly encouraged. Although the triad of myositis/myocarditis/iMG appears to be uncommon,^6,7 we agree that a high index of suspicion for diagnosis and an aggressive treatment approach (eg, intravenous immune globulin or another immunosuppressive agent) are warranted with these syndromic presentations. Additionally, in patients with negative serology and nondiagnostic neurological testing but elevated creatine phosphokinase, the role of muscle biopsies in differentiating myositis from iMG could be helpful.⁸

We thank Palassin et al² for highlighting the role of the pharmacovigilance databases, specifically VigiBase⁹ in expanding on real-world immune toxicities, especially for MACE. These events are infrequent and thus such large population-level real-world databases can help in understanding toxicity trends and outcomes.¹⁰ Palassin et al² ask if our review would suggest an under-reporting of these events. To that end, we wanted to highlight our rigorous methodology. Our data source relied on reporting requirements enforced by clinical trial participation in National Cancer Institute (NCI)–sponsored trials with clinical trial-level auditing and monitoring of primary source medical records in the NCI-funded network of clinical trial sites. We then evaluated all cardiac-related adverse events, as defined by the Common Terminology Criteria for Adverse Events and other consensus definitions as proposed by Bonaca et al,¹¹ especially for myocarditis. Two physicians independently evaluated these adverse events to assess the likelihood of MACE from ICI-based therapies. Our cohort was reduced by > 50% by eliminating cardiac adverse events that had obvious confounders or a history of cardiac disease that could cause an alternate explanation for the adverse events in question.

As to the question regarding the estimation of MACE prevalence, we would like to highlight two potential limitations when interpreting such data from VigiBase. First, although real-world pharmacovigilance databases (such as VigiBase) can be helpful in looking for signals of drug toxicities, these real-world databases have considerable missing information compared with clinical trial data sets (eg, they often lack access to primary source medical records including possible confounders related to underlying comorbidities). As a result, cardiac adverse events collated through these databases may be overrepresented. Second, several pharmacovigilance databases lack the total number (denominator) of treated patients in the context of a specific therapy, leading to reporting bias. VigiBase captures a breadth of toxicity data, but without a denominator, reporting the approximate incidence of a particular adverse event in relation to therapy, including MACE, is impossible. VigiBase can help with signal detection, as was reported by Pallasin et al, but the strength of that signal in a given population of treated patients cannot be determined.

Nevertheless, postmarketing pharmacovigilance is essential to assess the risk of ICI-related adverse events especially MACE in the general population and serves as an important means to improve patient care. To the authors' point, myocarditis was perhaps an underrecognized entity from 2015 through 2019. In our reporting, this was mitigated by adjudication of all cardiac events, enabling appropriate classification of potentially unrecognized events. Nevertheless, low-grade or smoldering myocarditis could have been missed. Further efforts are certainly needed in characterizing both myocarditis and nonmyocarditis MACE both from clinical trials and pharmacovigilance databases.

Overall, we agree with both the author groups regarding the importance of patient and physician education on the various presentations of MACE. Real-world pharmacovigilance databases and clinical trial data sets provide complementary approaches for understanding the potential risks associated with ICIs. More data for these cardiac immune-related adverse events could facilitate early interventions and develop better algorithms to prevent the morbidity associated with these events.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Reply to S. Rossi et al and P. Palassin et al

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).