Takotsubo syndrome in patients with cancer treated with immune checkpoint inhibitors: a new adverse cardiac complication.

Abstract

Takotsubo syndrome (TTS) is increasingly being reported in patients with cancer and in those undergoing chemotherapy. Up to 29% of patients admitted with a final diagnosis of stress TTS have had cancer.1 Several cases of TTS have been described in patients taking oncology drugs including fluoropyrimidines (5-fluorouracil, capecitabine), molecular-targeted agents (sunitinib, bevacizumab, trastuzumab), and cisplatin.1 By stimulating the immune system, immune checkpoint inhibitors (ICIs) may induce immune-related adverse events that could involve the skin, endocrine glands, liver, and lungs.2 Recent reports have also highlighted the risk of cardiotoxicity related to ICI administration, including atrial fibrillation, atrioventricular block, pericardial effusion, and myocarditis.2 We have recently described cases of takotsubo-like syndrome occurring in patients receiving ICIs.3 To identify whether this is clinically important, we performed a disproportionality analysis using the World Health Organization (WHO) global database of individual case safety reports (ICSRs) (VigiBase®), including adverse drug reaction (ADR) reports, to evaluate the reporting odds ratios (RORs) of stress cardiomyopathy (preferred term) associated with ICIs. The WHO database contains more than 17 million ICSRs received from 127 full member countries worldwide since 1968.4 ICSRs include administrative information (country, type of report, qualification of reporter), patient data (sex, age), date of onset of drug reaction, and outcome using Medical Dictionary for Regulatory Activities (MedDRA) 20.1 terms, WHO assessment of causality, and drug(s) involved (name, drug start and stop dates, time to onset, indication, dose, de-challenge, re-challenge). Data extraction was carried out using Vigilyze® (Uppsala Monitoring Centre) for records between 1 January 2008 (the date the term ‘stress cardiomyopathy’ was first introduced in the database) and 1 June 2018. We extracted all ICSRs associated with ‘stress cardiomyopathy’ as the preferred term. ICIs (ipilimumab, pembrolizumab, nivolumab, atezolizumab, avelumab, and durvalumab) were selected as suspect or concomitant drug. Case/non-case analyses were performed for ICIs compared with negative (paracetamol) and positive (adrenaline and venlafaxine) controls. This method allowed us to compare the proportions of specific ADRs reported for each drug or group of drugs. Disproportionality was estimated by calculating RORs and their 95% confidence intervals (CIs), as previously described.5 The cut-off for signal detection was defined as a lower 95% CI ≥ 1. An association between TTS and ICI could be suspected if the stress cardiomyopathy proportion was greater in patients exposed to ICI compared with negative and positive control drugs. This method allows the detection of a potential pharmacovigilance safety signal. All analyses were performed using R software, v3.4.4 for Windows (R Foundation for Statistical Computing). Statistical significance was defined as P < 0.05. Among the 13 188 920 ICSRs reported in VigiBase® during the study period, 1333 were cases of stress cardiomyopathy and 38 310 were cases of adverse events related to ICI drugs. We identified 13 reports of TTS in patients who received ICI drugs: ipilimumab, pembrolizumab, and nivolumab monotherapy (Table 1). No cases of stress cardiomyopathy were reported when ICIs were prescribed in combination. The absence of TTS reported with atezolizumab, avelumab, and durvalumab is probably due to the recent labelling of these ICIs. The cases included adults of both sexes with a median age of 79 years (Table 1). The most common cancer types (available in nine ICSRs) were lung cancer (n = 7) and melanoma (n = 2). Only three patients received concomitant cardiovascular or diabetes medications (this information was available in all 13 ICSRs). Data for the precise timing of onset from the initiation of ICI treatment were available in five ICSRs. Among these ICSRs, the median (interquartile range) time to TTS onset was 6 (4–96) days, the mean time to onset was 76 days, and three cases (60%) occurred during the first 6 weeks of treatment. Clinical evolution was available in eight ICSRs: one patient (13%) died, four (50%) recovered, and three (38%) did not recover. Overall, ICIs were associated with a higher proportion of TTS (13/38 310; ROR 3.39, 95% CI 1.96–5.86, P < 0.0001; Table 1). The RORs for adrenaline and venlafaxine (positive controls) were 74.25 (95% CI 62.11–88.76, P < 0.0001) and 5.02 (95% CI 3.65–6.90, P < 0.0001), respectively. Paracetamol (negative control) was not associated with TTS (ROR 1.21, 95% CI 0.88–1.67, P = 0.28). It must be noted that the ROR lower 95% CI for nivolumab was <1. However, this result could be explained by a Weber effect, as a large number of ADRs had been reported as nivolumab was labelled compared with other ICIs (Table 1). This analysis of the WHO pharmacovigilance database suggests that ICIs could provoke TTS. Clinicians should be aware of such a potentially severe cardiac condition, which should be cautiously characterized and evaluated. Editorial support was provided by Jenny Lloyd and Sophie Rushton-Smith (MedLink Healthcare Communications) and was funded by the authors. Conflict of interest: none declared.