Cardiovascular toxicity of immune therapies for cancer
BMJ 2024; 385 doi: https://doi.org/10.1136/bmj-2023-075859 (Published 15 May 2024) Cite this as: BMJ 2024;385:e075859
This article has a correction. Please see:
- Nicolas L Palaskas, associate professor1,
- Hyeon-Ju Ali, assistant professor1,
- Efstratios Koutroumpakis, assistant professor1,
- Sarju Ganatra, assistant professor2,
- Anita Deswal, professor1
- 1University of Texas MD Anderson Cancer Center, Houston, TX, USA
- 2Lahey Hospital and Medical Center, Burlington, MA 01805
- Correspondence to: A Deswal adeswal{at}mdanderson.org
ABSTRACT
In addition to conventional chemoradiation and targeted cancer therapy, the use of immune based therapies, specifically immune checkpoint inhibitors (ICIs) and chimeric antigen receptor T cell therapy (CAR-T), has increased exponentially across a wide spectrum of cancers. This has been paralleled by recognition of off-target immune related adverse events that can affect almost any organ system including the cardiovascular system. The use of ICIs has been associated with myocarditis, a less common but highly fatal adverse effect, pericarditis and pericardial effusions, vasculitis, thromboembolism, and potentially accelerated atherosclerosis. CAR-T resulting in a systemic cytokine release syndrome has been associated with myriad cardiovascular consequences including arrhythmias, myocardial infarction, and heart failure. This review summarizes the current state of knowledge regarding adverse cardiovascular effects associated with ICIs and CAR-T.
Introduction
In addition to conventional chemoradiation and targeted cancer therapy, the use of immune based therapies for patients with cancer has increased exponentially.1 This increase has been paralleled by off-target immune related adverse events (irAEs) that can affect almost any organ system including the cardiovascular system.23 Many of the cardiovascular toxicities associated with immune therapy have been characterized in post-approval, real world studies that include more patients, often with more frequent cardiovascular comorbidities, and longer follow-up compared with those originally reported in randomized clinical trials (RCTs). Recognizing, diagnosing, and treating cardiovascular irAEs is imperative, as the therapeutic spectrum and indications for use of immune therapies continue to expand. In this review, we summarize the current state of knowledge regarding adverse cardiovascular effects associated with two of the most established immune therapies, immune checkpoint inhibitors (ICIs) and chimeric antigen receptor T cell therapy (CAR-T).
Epidemiology
In 2018 nearly 44% of cancer patients in the United States were eligible for ICIs and 14% of patients received them. Use of ICIs has risen over time (22% in 2020).4 As a result of high cost, availability of ICIs is severely limited in low and middle income countries.567 Limited data from such countries suggest that a lower dose or a shorter course of therapy may have comparative efficacy in progression-free survival, which, if proven, may improve availability globally.67
In 2019 an estimated 450 000 cancer patients worldwide were eligible for CAR-T, and this number is expected to rise to two million by 2029, with the caveat that access to CAR-T may be limited, especially in low and middle income countries, owing to logistic challenges of current cell therapy manufacturing, cost, and administration only at select centers. The Nationwide Readmissions Database (2017-19), suggested that more than 90% of CAR-T was administered in large metropolitan areas, 49% of patients had private insurance, and 33% belonged to the highest income communities.89 CAR-T has been used for at least 27 000 US patients since approval by the Food and Drug Administration in 2017 and by more than 34 000 patients worldwide.1011
Background and mechanisms
Immune checkpoints are regulatory molecules that attenuate T cell activation, prevent unchecked immune system activation, and guard against autoimmunity. ICIs are monoclonal antibodies that block the inhibitory signals of T cell activation, thus enabling tumor reactive T cells to mount an effective antitumor response.112 Most ICIs target either the cytotoxic T lymphocyte antigen-4 (CTLA-4), programmed cell death protein-1 (PD-1) or the PD-ligand 1 (PD-L1) checkpoints present on the surface of CD4+ and CD8+ lymphocytes or multiple immune cell types, including T cells, B cells, and NK cells and tumor cells.
Since approval of the first CTLA-4 inhibitor, ipilimumab, for metastatic melanoma in 2011, nine additional ICIs have received approval across a spectrum of cancers including lung, genitourinary, head and neck, gastrointestinal, breast, and gynecologic cancers. They include another CTLA-4 inhibitor (tremelimumab), PD-1 inhibitors (nivolumab, pembrolizumab, cemiplimab, toripalimab), and PD-L1 inhibitors (atezolizumab, avelumab, durvalumab), and the latest one targets the lymphocyte activation gene-3 (LAG-3) immune checkpoint (relatlimab).13 Furthermore, ICIs are increasingly being used in combination—for example, CTLA-4/PD-1 inhibitors and PD-1/LAG-3 inhibitors—which carry increased risk of cardiovascular irAEs.
Adoptive cellular therapy consists of the intravenous transfer of either tumor resident immune cells (for example, tumor infiltrating lymphocytes) or peripheral blood modified immune cells (for example, chimeric antigen receptor (CAR) modified T cells or T cell receptor gene therapy) to patients with cancer to mediate anti-tumor functions.1415 CAR modified T cells are genetically modified T cells from patients that express recombinant receptors on the cell surface against a specific antigen expressed on the tumor cell surface. Binding of CARs to tumor antigens results in strong T cell activation which induces tumor cell apoptosis.1516 Since 2017, four CD19 targeted CAR-Ts (tisagenlecleucel, axicabtagene ciloleucel, brexucabtagene autoleucel, lisocabtagene maraleucel) have been approved for treatment of B cell malignancies including relapsed or recurrent acute lymphoblastic leukemias and certain B cell lymphomas. Recently, two B cell maturation antigen targeted CAR-Ts were approved for relapsed or recurrent multiple myeloma (idecabtagene vicleucel, ciltacabtagene autoleucel).17
Sources and selection criteria
We searched PubMed and Embase for articles, letters, and editorials published from 2010 to October 2023 using the following search terms restricted to the title/abstract: “immune checkpoint inhibitor” OR “chimeric antigen receptor” OR “CAR T” OR “immunotherapy” AND “myocarditis” OR “cardiotoxicity.” A total of 761 articles were retrieved from PubMed and 470 from Embase. After filtering to exclude books and other documents and removing articles that were not relevant, 446 articles remained from PubMed and 424 from Embase. We applied limited exclusion criteria for articles focused on cardiac conditions not related to immunotherapy toxicity and immunotherapy toxicity of non-cardiovascular organs. We included case reports and case series given the rare nature of ICI associated myocarditis (ICIMy); we included translational studies when they were focused on mechanistic insights into immunotherapy cardiotoxicity. We reviewed additional sources outside the selection criteria by cross referencing.
Immune checkpoint inhibitor associated myocarditis
ICI associated cardiovascular events are much less frequent than more common irAEs such as gastrointestinal (8-27%) and cutaneous (up to 72%) toxicities.18 Short to intermediate term cardiovascular complications include ICIMy, pericarditis and pericardial effusions, vasculitis, and thromboembolism, with the most impactful being ICIMy with mortality as high as 30-50%. Since the first report of ICIMy in 2016,19 awareness of ICIMy has increased. Meta-analysis of RCTs and pharmacovigilance data described a low incidence of ICIMy (0.27-0.67%), with a higher incidence (1.3%) in patients treated with combination ICIs.19202122 In 2016 a multicenter registry standardized case identification by biopsy and clinical criteria with a reported incidence of 1.14% for ICIMy.23 Smaller single centered prospective surveillance studies reported a higher incidence of 1.40-2.46%.242526 Initial reports likely recognized only the severe presentations of ICIMy, whereas contemporary studies include a wider severity spectrum, increasing the reported incidence. Nevertheless, the large number of patients treated with ICIs and the poor prognosis of ICIMy underscore its importance.
Risk factors
The one consistent risk factor for ICIMy is use of combination ICIs such as CTLA-4 and PD-1/PD-L1 inhibitors.19222728 In a retrospective pharmacovigilance study, the combination of nivolumab and ipilimumab conferred a 4.74-fold higher risk of developing myocarditis compared with nivolumab alone.19 A recent RCT reported an incidence rate of 0.6% for myocarditis for single agent ICI therapy (anti-PD-1) and 1.7% for the combination (anti-PD-1 and LAG3).29 Furthermore, ICIMy associated with combination therapy is often more severe and associated with overlap myositis and myasthenia gravis. Factors suggested as conferring risk for other irAEs remain unproven for ICIMy, including sex, underlying cardiovascular disease, other cardiotoxic therapies, previous autoimmune disease, or genetic factors.22303132 A study using pharmacovigilance data (VigiBase), single center data, and a meta-analysis of RCTs showed that ICIs for thymic tumors, especially thymomas, were more frequently associated with ICIMy, with more frequent life threatening arrhythmias and concurrent myositis leading to respiratory muscle failure and death, compared with other ICI treated cancers.33
Pathophysiology of ICI associated myocarditis
Myocarditis is defined by the presence of myocardial inflammatory infiltrate with associated myocyte death. The inflammatory infiltration observed in ICIMy is predominantly CD8+ T cells, often in a two to one ratio to CD4+ T cells. CD8+ cytotoxic T cells are the suspected primary drivers of toxicity. The role of accompanying infiltration with CD68+ monocytes/macrophages, is unclear.34 The proposed mechanism of ICIMy is through molecular mimicry in which CD8+ T cells recognize myocardial antigens that are the same as or similar to tumoral antigens, a mechanism supported by a two patient case series with ICIMy in which the same T cell clones were present in the tumor, skeletal muscle, and myocardium.19 The myocardial antigens targeted are uncertain, although α-myosin was identified as an autoantigen in a mouse model.35 Furthermore, activated T cells can produce inflammatory cytokines that contribute to myocardial damage.1 In addition, a retrospective case series of patients with ICIMy showed pericapillary C4d+ deposition, suggesting an autoantibody mediated component of disease.36 Clarification of the mechanisms of ICIMy is needed to target therapeutic interventions.
Clinical presentation of ICI associated myocarditis
Patients with ICIMy present with varying symptoms including dyspnea, palpitations, chest pain, fatigue, syncope, or signs of heart failure.3738 Patients with fulminant myocarditis present with cardiogenic shock, heart block, or incessant ventricular arrhythmias.39 However, some patients have no symptoms, and ICIMy is suspected owing to incidental elevation of biomarkers or abnormal cardiac imaging. In some cases, left ventricular dysfunction was found years after exposure to ICIs.40414243 Whether such presentations are secondary to early subclinical ICIMy,42 represent indolent long term inflammation,41 or are unrelated to ICIMy remains unclear.
Patients with ICIMy may have concomitant irAEs, particularly overlap myositis and myasthenia gravis,19 observed in 23-30% of patients with ICIMy.2123 Such patients present with myalgias, weakness, diplopia, and bulbar symptoms such as dysphagia and dysarthria. They are at higher risk of respiratory failure, cardiogenic shock, life threatening arrhythmias, and death.444546 Therefore, patients with any of these conditions should be evaluated for the other overlap irAEs.
The onset of ICIMy usually occurs early after initiation of ICI, although cases have been reported from days to years after initiation.20212338 A meta-analysis of 65 RCTs found onset times from 3.2 months to 32.8 months.20 VigiBase data showed a median onset at 30 (interquartile range 18-60) days,21 and a multicenter registry reported a median of 34 days, with 81% of cases within three months of ICI initiation.23
Outcomes of patients with ICIMy have varied across studies. Compared with other irAEs, ICIMy was associated with a higher fatality rate of up to 50% in initial studies.1947 Among 131 cases of ICIMy in an international ICIMy registry, major adverse cardiovascular events (MACE) occurred in 40% over a median follow-up of 148 days and included complete heart block (16%), cardiogenic shock (15%), cardiac arrest (15%), and cardiovascular death (17%).37 At the other end of the spectrum are patients who have no symptoms but have biomarker (troponin) elevation.48 As awareness of the condition continues to increase, ICIMy is diagnosed and treated earlier, and early cardiovascular mortality is expected to be lower than 30%.
Diagnosis
Diagnostic criteria
The diagnosis of ICIMy can be challenging in the setting of non-specific symptoms and biomarker and imaging abnormalities. Alternative diagnoses including acute coronary syndrome, type II myocardial infarction, or non-ICIMy cardiomyopathy/heart failure can have similar clinical presentations. In 2019 diagnostic certainty criteria proposed categories of definite, probable, and possible ICIMy based on a hierarchy of evidence ranging from histopathology and cardiovascular magnetic resonance (CMR) to echocardiography, electrocardiography, and biomarkers accompanying appropriate clinical presentations.24 The International Cardio-Oncology Society proposed a definition encompassing symptoms, histopathology, biomarkers, electrocardiographic findings, and imaging findings constituting “major” and “minor” criteria (fig 1).49 The definition also includes three additional modifiers —severity, refractoriness to steroids, and stage of recovery—which can further standardize the language and guide clinical decisions such as withholding ICI therapy, adding immunomodulators, or even re-challenging with ICIs. Individual testing modalities are discussed below.
Diagnostic and severity criteria of immune checkpoint inhibitor associated myocarditis suggested by International Society of Cardio-Oncology,49 Common Terminology Criteria for Adverse Events (CTCAE v.5),50 and National Comprehensive Cancer Network.18 CMR=cardiac magnetic resonance; ECG=electrocardiography; irAE=immune related adverse event
Biomarkers
Serum troponin concentrations are the primary cardiac biomarkers used for the diagnosis of ICIMy. Despite the prognostic significance of troponin T,23 cross reactivity with skeletal muscle in the setting of myositis is possible. Therefore, troponin I assays have been suggested given their higher myocardial specificity.24 However, a retrospective observational cohort study of 60 patients with ICIMy found that troponin T was elevated in 98% of patients compared with 88% with troponin I elevation (P=0.03) within 72 hours of hospital admission with ICIMy. Similar discrepancies were also observed in 87 cases in an international ICIMy registry (93% v 68%, respectively; P<0.001). Additionally, the highest troponin T concentrations within 72 hours of admission performed best when predicting major adverse cardiomyotoxic events.51 Furthermore, troponin T elevations above 1.5 ng/mL have been associated with a fourfold increase in MACE.23 However, troponin elevation lacks specificity for ICIMy, as more common conditions such as acute coronary syndrome or myocardial injury secondary to sepsis or exacerbation of heart failure also cause troponin elevations.52 Diagnostically, the magnitude of rise considered significant may vary depending on the specific assay used. An international retrospective cohort of ICIMy reported across assays and institutions in 13 different countries that the median ratio of troponin I and troponin T to their upper limits of normal at ICIMy diagnosis was 8 (interquartile range 2-34) and 33 (11-78), respectively. Furthermore, the best predictor of MACE was troponin T >32 times the upper limit of normal within the first 72 hours of diagnosis.51 An ongoing RCT for ICIMy (ClinicalTrials.com; NCT05335928) is using troponin five times the upper limit of normal as the diagnostic threshold.
Surveillance for ICIMy with serial troponin is debated. In a prospective observational study, 214 patients receiving ICI were monitored with troponin I every two to four weeks; 11% developed a rise above normal. However, ICIMy was diagnosed in only three (1.4%) patients.26 The benefit of troponin surveillance remains unclear, given the low incidence of ICIMy, lack of specificity of high sensitivity troponin assays, and potential for harm with further unnecessary testing and interruption of ICIs.53 However, consensus suggests baseline troponin measurement before starting ICIs to enable comparison for changes if clinically indicated.5354
By contrast, natriuretic peptides have not consistently correlated with the severity of ICIMy, given that not all patients with ICIMy develop heart failure.34 Other biomarkers such as serum creatine kinase, aspartate aminotransferase, alanine aminotransferase, and lactate dehydrogenase are often elevated in patients with ICIMy, as shown by a retrospective observational cohort of 2606 patients taking ICIs, of whom 1.0% developed ICIMy. However, of these non-troponin biomarkers, only elevations in creatine kinase were associated with incident myocarditis (hazard ratio 1.83, 95% confidence interval 1.59 to 2.10) and all cause mortality (1.10, 1.01 to 1.20). All these markers have low specificity owing to their additional non-cardiac origins.55 Given that ICIMy may present in overlap with other neuromuscular irAEs such as myositis and myasthenia gravis, significant elevations of non-troponin biomarkers such as creatine kinase may raise suspicion for overlap with muscular irAEs.
Electrocardiography
Various electrocardiographic changes have been described with ICIMy. Although non-specific, they occur frequently with ICIMy. In a retrospective cohort, compared with electrocardiographs before ICI initiation, those at ICIMy diagnosis showed higher heart rate, QRS and QTc prolongation, decreased voltage, conduction disorders, and repolarization abnormalities.39 Arrhythmias included supraventricular tachycardia (24%), complete heart block (17%), and sustained ventricular tachycardia (11%).39 The prognostic value of electrocardiography has been demonstrated in international registries—for example, a 10 ms increase in QRS duration was associated with 1.3-fold (95% confidence interval 1.07 to 1.61) increase in MACE and pathologic Q waves with a sixfold (2.8 to 12.79) increase in MACE.3956
Imaging
Echocardiography is widely available and used when ICIMy is suspected.244954 Reduced left ventricular ejection fraction (LVEF), global left ventricular hypokinesis, new wall motion abnormalities, increased wall thickness, pericardial effusion, and changes in global longitudinal strain have been described.2457 However, in a retrospective cohort of 35 patients with ICIMy, most (51%) maintained normal LVEF but with similar rates of MACE compared with those who had reduced LVEF.23 In another retrospective cohort, reduced global longitudinal strain was observed in patients with ICIMy compared with controls, regardless of LVEF. Every 1% reduction in global longitudinal strain was associated with a 1.5-fold and 4.4-fold increase in MACE in patients with depressed and normal LVEF, respectively.58 Similarly to biomarkers and electrocardiographic findings, echocardiographic findings are non-specific and serve as only supportive criteria for ICIMy diagnosis.
Cardiac magnetic resonance imaging is the preferred imaging modality for the diagnosis of non-ICIMy, given its ability to characterize tissue.5960 The modified Lake Louise Criteria provide a high degree of sensitivity and specificity for the diagnosis of non-ICIMy. These criteria include evidence of myocardial edema (with T1 mapping, T2 weighted imaging) and non-ischemic myocardial injury (T2 mapping extracellular volume quantification, late gadolinium enhancement). When the modified Lake Louise Criteria were applied to a retrospective cohort of 136 patients with ICIMy, 95% met the non-ischemic myocardial injury criteria and 53% met the myocardial edema criteria; 100% met at least one of the two main criteria. Native T1 values but not T2 values were independently associated with subsequent MACE.42 Among the patients who developed MACE, all had abnormal native T1 values compared with none of those without MACE. These findings contrast with an earlier retrospective study, which showed a lower than expected positivity rate for late gadolinium (48%) and T2 weighted imaging (28%) in patients with preserved LVEF, as well as a lack of association with MACE.37 However, if CMR was performed later (>4 days after admission), the presence of late gadolinium enhancement increased from 21% to 72%. Therefore, sensitivity of CMR diagnosis of ICIMy is higher if performed later but creates difficulty with the early diagnosis and early management of ICIMy. The findings suggest that complete CMR characterization (including T1 and T2 mapping and gadolinium enhancement) may be critical for diagnosis and prognosis. Positron emission tomography imaging with novel agents is in the early stages of evaluation as a non-invasive diagnostic tool for ICIMy.6162
Endomyocardial biopsy
Endomyocardial biopsy remains the gold standard for diagnosis of non-ICIMy, with histopathologic evidence by the Dallas criteria (1986). These include inflammatory infiltration and myocardial necrosis.63 However, several problems are observed with use of the Dallas criteria in general.63 Firstly, sampling error exists with only 25% of myocarditis cases confirmed by positive endomyocardial biopsy when a single sample is obtained. This improves to ~75% sensitivity with five or more biopsy samples per patient.6465 Secondly, histopathologic interpretation is subject to significant interobserver variability.66 Similar challenges in ICIMy may exist, given that the imaging findings show patchy involvement. Lastly, endomyocardial biopsy is an invasive procedure and large single center studies of biopsy for non-ICIMy patients have described complication rates of up to 6%, including access site bleeding, arrhythmias, and cardiac perforation leading to tamponade (<1%).67 Despite these limitations, several studies have described the immunohistochemical findings of ICIMy with predominantly CD8+ T cell infiltration in addition to CD4+ T cells, accompanied by lymphohistiocytic infiltration that includes CD68+ cells (fig 2). The histopathologic appearance is similar to that of acute cellular rejection in transplanted hearts.346869
Histopathology: endomyocardial biopsy findings in immune checkpoint inhibitor associated myocarditis. (A) Hematoxylin and eosin stain showing dense inflammatory infiltrate and myocyte loss. (B) Immunohistochemical staining showing CD8+ T cell infiltration. (C) Immunohistochemical staining showing CD68+ infiltration. (D) Immunohistochemical programmed death ligand 1 (PDL-1) staining
A review of endomyocardial biopsy samples from 28 suspected cases found a wide spectrum of biopsy findings in ICIMy,34 of which 11 had inflammatory infiltration without myonecrosis, thus not fulfilling the Dallas criteria. Of the 11 cases, four patients had continued ICI treatment without any immunosuppressive drugs and without MACE or decline in cardiac function at a median follow-up of 409 days.34 Similarly, another retrospective analysis evaluating the histopathology of 10 patients with ICIMy created classifications for high grade versus low grade myocarditis based on the density of the inflammatory infiltrate. All patients with high grade myocarditis died whereas all patients with low grade myocarditis survived.68 Whether the ICIMy grade by biopsy may better determine prognosis, guide tailored treatment, and potentially identify patients with ICIMy at low risk who could be re-challenged with ICIs remains to be confirmed.
Diagnostic algorithm
Figure 3 outlines a practical guide for establishing the diagnosis of ICIMy.49 Patients taking ICIs who present with a clinical syndrome consistent with myocarditis should have biomarkers (troponin I and/or T), electrocardiography, and echocardiography performed at presentation. If concerns exist about more common alternative diagnoses, appropriate investigation should be performed to exclude those possibilities. Concomitant coronary artery disease and ICIMy can occur. Therefore, just the presence of significant obstructive coronary artery disease on catheterization should not rule out the possibility of ICIMy in appropriate clinical scenarios, as this may result in therapeutic delays and poor outcomes.70 The subsequent steps in the investigation, endomyocardial biopsy and/or CMR, should take into consideration the patient’s hemodynamic stability, other clinical diagnostic criteria, and the center’s expertise and risk of complications.
Clinical algorithm for investigation and treatment of suspected immune checkpoint inhibitor associated myocarditis (ICIMy). CK=creatine kinase; CKMB=creatine kinase myocardial band; ICI=immune checkpoint inhibitor; irAE=immune related adverse event; NP=natriuretic peptides; TSH=thyroid stimulating hormone. *In addition to anti-acetylcholine receptor (anti-AChR) antibodies, other antibodies for diagnosis of myasthenia gravis can be measured if needed such as anti-muscle specific kinase and anti-low density lipoprotein receptor protein 4
Management
Corticosteroids are first line treatment as recommended by guidelines (table 1).254717274 In a retrospective international ICIMy registry, the use of high dose corticosteroids (501-1000 mg/day) was associated with decreased MACE compared with intermediate dose (60-500 mg/day) and low dose (<60 mg/day) (22.0% v 54.6% v 61.9%, respectively; P<0.01).37 In addition, early corticosteroid administration within 24 hours of presentation was associated with decreased persistence of troponin elevation. Therefore, guidelines recommend rapid diagnosis of ICIMy with early initiation of high dose corticosteroids between 500 and 1000 mg/day for three to five days followed by a steroid taper over four to six weeks (most guidelines) or up to 12 weeks (European Society of Cardiology guideline). During steroid tapering, close monitoring of symptoms and electrocardiographic parameters paired with serial troponin are needed for individualized patient management. Also, current guidelines do not differentiate steroid dosing on the basis of severity of ICIMy. Whether mild or asymptomatic cases need high dose steroids and discontinuation of ICI is unknown. This remains an area of active research.
Societal guidelines for diagnosis and treatment of immune checkpoint inhibitor associated myocarditis and chimeric antigen receptor T cell therapy associated cytokine release syndrome
For patients who do not respond clinically or have persistent troponin elevation, use of immunomodulators in addition to steroids is recommended (fig 4). Several case reports and case series describe the use of T cell mediated therapies (mycophenolate, tacrolimus, antithymocyte globulin), cytokine therapies (infliximab, ruxolitinib), antibody mediated therapies (plasmapheresis, intravenous immunoglobin), and other monoclonal antibody therapies such as alemtuzumab (CD52 target) and abatacept (CTLA-4 agonist).417879808182838485868788 Data from RCTs are unavailable at this time for their relative efficacy in ICIMy.
Treatment options for immune checkpoint inhibitor associated myocarditis
Emerging treatments
An ongoing multicenter, placebo controlled RCT is comparing the immune checkpoint CTLA-4 agonist (CTLA-4-immunoglobulin fusion protein) abatacept and steroids versus placebo and steroids (NCT05335928). The first case of abatacept used to effectively treat refractory ICIMy was reported in 2019.85 Given that CTLA-4 receptor activation is upstream of PD-1, it is postulated that abatacept therapy will be effective for patients with ICIMy who received either CTLA-4 inhibitors or PD-1/PD-L1 inhibitors. In a genetic mouse model of ICIMy, abatacept led to significant improvement in survival.89 Recently, a retrospective study compared the use of a combination of abatacept and ruxolitinib (a Janus kinase inhibitor that impairs T cell activation via blockade of pro-inflammatory cytokines) in 30 patients with severe ICIMy with concomitant respiratory muscle involvement against 10 historical controls and found significantly lower mortality (3.4% v 60%; P<0.01).78 Results of prospective trials exploring strategies of upfront use of immunomodulators as early steroid sparing therapy are awaited.
ICI re-challenge after myocarditis
Data regarding the safety of ICI re-challenge after ICIMy are insufficient. Nine case reports including 16 patients have shown mixed results. After ICI re-challenge, four patients developed recurrent ICIMy and two patients died owing to cancer progression within one year. None of these patients had severe ICIMy at initial presentation.90 Similarly, as discussed in the endomyocardial biopsy section, some patients with inflammatory infiltrate but no myocyte loss (low grade myocardial inflammation) may be able to tolerate ICI re-challenge from a cardiovascular standpoint.34 Therefore, several factors should be considered for ICI re-challenge, including other cancer therapeutic options and severity of initial ICIMy, with very close symptom and troponin monitoring if re-challenged after discussion of potential risks with the patient.
Other adverse cardiovascular events associated with ICIs
Pericarditis/pericardial effusion
Initial case reports described pericarditis and pericardial effusions in patients taking ICIs that were not associated with ICIMy.9192 A retrospective cohort study reported an incidence rate of 1.57 events per 100 person years of pericarditis and/or new moderate to large pericardial effusions among 2842 patients treated with ICIs. Compared with matched controls, ICIs increased risk for pericardial disease (hazard ratio 4.37, 2.09 to 9.14; P<0.001). Furthermore, pericardial disease was associated with increased all cause mortality (hazard ratio 1.53, 0.99 to 2.36; P=0.05).93 However, the need for pericardiocentesis was rare (0.38%) in another retrospective cohort of 3966 patients.94 In the absence of concomitant myocarditis, initial treatment for ICI associated pericarditis is similar to that for pericarditis in general, with non-steroidal anti-inflammatory drugs and colchicine.
Vasculitis
Vasculitides associated with ICIs, including giant cell arteritis and small and medium vessel vasculitis, have been reported.95969798 The pharmacovigilance adverse event reporting database reported an increased risk of vasculitis with ICIs compared with other cancer therapies (relative odds ratio 1.56, 95% confidence interval 1.25 to 1.96), with temporal arteritis being the most common.21 Overall, vasculitis is a rare complication of ICIs. Whether ICIs trigger de novo vasculitis or unmask pre-existing subclinical vasculitis is unclear.
ICI associated thromboembolic disease
Retrospective studies have suggested that ICIs increase the risk of venous and arterial thromboembolic disease, and these are recognized as irAEs.7299100101102103104 The proposed mechanism is T cell activation leading to release of interferon-γ, causing an increase in tissue factor expression by monocytes/macrophages, which initiates the coagulation cascade.105
Retrospective studies have provided mixed estimates of prevalence, onset, and prognosis of ICI associated thromboembolism. Among a cohort of 956 patients, venous and arterial thromboembolic disease occurred in 8.5% of patients with lung cancer after an average of five completed ICI cycles and in 5.8% of patients with melanoma after an average of eight ICI cycles. Patients with venous and arterial thromboembolic disease had significantly worse overall survival.106 Some single center studies reported up to 16% incidence, with incidence of venous thromboembolism rising fourfold after ICI initiation.107108109
Risk factors for ICI associated venous thromboembolism are inconsistent. In patients with lung cancer, risk of venous thromboembolism increased in those with anemia and previous history of thrombosis; elevated lactate dehydrogenase concentrations (>198 U/L) were a risk factor in melanoma, and a higher neutrophil-lymphocyte ratio was a significant risk factor in both cancers. An elevated Khorana risk score, a venous thromboembolism risk calculator for patients with cancer, which includes certain high risk cancer types, thrombocytosis, anemia, leukocytosis, and obesity, was also associated with higher ICI associated venous thromboembolism.101107
Current guidelines for the management of ICI associated thromboembolic disease are based on expert opinion, including continuation of ICI unless other irAEs are present and treatment of venous thromboembolism with anticoagulation.72
ICI associated atherosclerosis
Inflammation and immune involvement in development of atherosclerosis is well established. Checkpoint receptors have also been shown to be involved in atherosclerotic regulation. Thus, their inhibition may lead to increased T cell activation and resultant accelerated atherosclerosis.110 However, clinical data on atherosclerosis progression are mixed.111 A pharmacovigilance study compared events in patients receiving ICI versus other therapies and found lower reporting of myocardial infarction (0.53% v 1.00%) and arterial ischemia (0.65% v 1.32%).21 By contrast, a retrospective study of 2842 patients taking ICI therapy compared with 2842 matched controls showed that patients taking ICI therapy had a three times higher risk of atherosclerotic events (hazard ratio 3.3, 2.0 to 5.5; P<0.001). In the same study, an analysis of 2842 patients compared atherosclerotic events in the two years before ICI therapy with the two years following ICI initiation and found an increase in events after ICI initiation (1.37 v 6.55 events per 100 person years; adjusted hazard ratio 4.8, 3.5 to 6.5; P<0.001).112 The largest meta-analysis included 63 RCTs (32 518 patients) with at least one ICI arm and observed an increased risk of myocardial infarction (odds ratio 1.51, 1.01 to 1.26) and cerebral ischemia (1.56, 1.10 to 2.20) with ICIs.20 A case-control study evaluating aortic plaque progression found that patients receiving ICI had seven times greater annual non-calcified plaque progression compared with controls when both groups had similar plaque volume at baseline.113 Although further study is needed, awareness of possible ICI associated accelerated atherosclerosis underscores the importance of early identification and optimization of traditional cardiovascular risk factors. Coronary artery calcium scoring on computed tomography scans performed for diagnosis and staging of cancer is one potential strategy to risk stratify patients for cardiovascular risk modification.114
Adverse cardiovascular events associated with CAR T cell therapy (CAR-T)
Pivotal clinical trials of CAR-T documented low occurrence of cardiotoxicity, potentially stemming from the careful selection of patients.115 However, subsequent retrospective analyses have identified a significant but varying incidence of MACE in 10-20% of patients receiving CAR-T.116 The events include cardiomyopathy, heart failure, arrhythmias, myocardial infarction, shock, cardiac arrest, and cardiovascular death (table 2). One of the most common toxicities of CAR-T is cytokine release syndrome (CRS), a systemic inflammatory response caused by cytokine release early after CAR-T administration. Although available evidence concerning the susceptibility to CRS and cardiotoxicity primarily stems from studies of CD19 targeted CAR-T,115 early data suggest that the incidence of CRS, which is the major driver of cardiovascular events, is likely comparable for CAR-T directed toward B cell maturation antigen.115
Major studies reporting chimeric antigen receptor T cell therapy (CAR-T) associated cardiovascular events (adapted from Ganatra et al, JACC CardioOncol 2022117)
Pathophysiology
Cardiovascular toxicity associated with CAR-T has three proposed mechanisms. Firstly, on-target, on-tumor effects lead to CRS mediated by supraphysiologic concentrations of inflammatory cytokines released by the activated CAR T cells.132133134 Most CAR-T associated cardiovascular events have been reported in the context of CRS and are correlated to severity of CRS.131 However, whether cardiotoxicity is merely an epiphenomenon of CRS or also has a shared mechanism similar to CRS remains to be confirmed.117131 Secondly, with on-target, off-tumor effects cardiovascular toxicity may occur through direct T cell mediated injury caused by shared target antigens with the tumor.132133134 Lastly, with off-target, off-tumor effects the T cells unexpectedly attack an antigen other than the intended tumor antigen or shared antigens.132133134 Notably, the off-target, off-tumor effects are proposed on the basis of only two cases of CAR-T directed against melanoma associated antigen-3, which showed cross reactivity against titin, a striated muscle protein in the heart, and led to fulminant myocarditis.135
The infused CAR-T, other activated immune cells, and therapy induced tumor cell lysis can lead to stimulation of the immune system, resulting in a surge of cytokines and chemokines such as interleukin-2, interferon-α, interleukin-6, tumor necrosis factor (TNF)-α and granulocyte-macrophage colony stimulating factor.136 This leads to CRS, which is marked by fever possibly accompanied by tachycardia, hypotension, hypoxia, and organ toxicity.136137 CRS has been observed in 70-90% of patients receiving CAR-T.137 Its severity ranges from mild with fever (grade 1) to systemic effects (hypotension and/or hypoxia and other organ toxicity; grades 2-4) (table 3) and contributes to cardiovascular complications such as cardiomyopathy, heart failure, arrhythmias, myocardial infarction, vascular leak syndrome leading to circulatory collapse, and multiorgan failure.77115
ASTCT grading for cytokine release syndrome (adapted from Lee et al, Biol Blood Marrow Transplant 201977)
Interleukin-6 is implicated as a central mediator of the CRS inflammatory cascade, including the pathophysiology of cardiotoxicity.115 Additionally, TNF-α may contribute to immune related cardiac dysfunction and heart failure.116138 Furthermore, microvascular obstruction can occur in the context of CRS attributed to microvascular dysfunction and increased permeability that triggers a myocardial inflammatory bystander reaction with pro-coagulant factors.116138 Such microvascular obstruction can contribute to myocardial ischemia and cardiomyopathy.116138 Atrial and ventricular arrhythmias likely have multifactorial etiologies related to the inflammatory milieu, electrolyte disturbances, hypotension, and cardiovascular stress rather than a specific pro-arrhythmic effect of CAR-T.116139 Although varied estimates of cardiovascular events associated with CAR-T have been reported (table 2), recent contemporary reports suggest a lower incidence of serious cardiovascular events, likely related to more frequent use of interleukin-6 antagonists earlier in the course of CRS.131
Cardiomyopathy and heart failure
The accurate estimation of incident cardiomyopathy and stress (Takotsubo) cardiomyopathy after CAR-T is limited by lack of systematic echocardiographic pre-therapy and post-therapy screening.115 Reported incidence of cardiomyopathy/heart failure from prospective trials, retrospective studies, and institutional registries ranges from 1% to 15%.115117126127131 One of the few studies with systematic echocardiographic surveillance protocols from two institutions for all patients with higher grade CRS after CAR-T reported cardiomyopathy in 10% of patients with a median decline in LVEF from 58% to 37% after 12.5 (range 2-24) days from CAR-T infusion.127 This aligns with studies in the pediatric population.140 Although other studies have reported an incidence of less than 5%,117 repeat echocardiograms in those studies were performed at the discretion of the treating clinicians, potentially resulting in an underestimation of cardiomyopathy. Patients who develop cardiomyopathy are reported to need greater supportive care such as mechanical ventilation and vasopressors.116 Although LVEF may recover in many patients, up to 50% may have some persistent cardiac dysfunction.127 Whether earlier treatment of CRS with interleukin-6 blockade will result in fewer cases of cardiomyopathy and higher recovery awaits confirmation.
Myocardial infarction
Clinically significant myocardial infarction is reported in 1-7% of patients.117 Potential mechanisms include plaque rupture due to systematic inflammation, type 2 myocardial infarction in the context of supply-demand mismatch from hemodynamic changes with CRS, and microvascular dysfunction.140141 A retrospective study reported clinical myocardial infarction in only 1.8% of 165 patients with CAR-T.131 By contrast, troponin elevation has been reported in up to 54% of patients overall and up to 71% with high grade (≥grade 2) CRS.137 Most of these troponin elevations seem to be related to supply-demand imbalance rather than acute coronary syndrome.142
Arrhythmias
The overall incidence of arrhythmias, typically atrial fibrillation, but also other supraventricular tachycardias and non-sustained ventricular tachycardia, ranges between 5% and 12%. Arrhythmias account for up to two thirds of MACE observed in patients within the first 30 days.131137 In many instances, arrhythmias are transient without longlasting impact, but some require acute treatment.131
Shock, cardiac arrest, and cardiovascular death
Hypotension requiring vasopressor support occurs in up to 25% of patients.137 Most often, it is due to distributive shock in the context of CRS.143 Although shock is reported in 40-50% of patients who experience MACE, true cardiogenic shock occurs in only a minority. Cardiac arrest and cardiovascular deaths are fortunately infrequent; overall incidence is reported to be ≤1.6%,126127131 with only one study reporting 4.6%.125
Risk factors for CAR-T associated cardiovascular events
Several risk factors for cardiovascular events after CAR-T have been reported, including traditional cardiovascular risk factors, grade of CRS, type of CAR-T, and previous cardiotoxic cancer therapy. One retrospective study reported that patients older than 60 years had a fourfold increased risk of MACE after adjustment for other risk factors (odds ratio 3.98, 1.47 to 10.78).131 However, another retrospective study did not find age to be an independent risk factor.126 Similarly, a history of traditional cardiovascular risk factors or established cardiovascular disease has not been consistently associated with higher rates of MACE. One retrospective study observed that each 1 mg/dL increase in baseline serum creatinine was associated with a 15-fold increased risk of MACE (hazard ratio 15.5, 3.7 to 65.9; P<0.001).126 These discrepancies may be related to differences in study populations, small sample sizes of the cohorts, or multi-collinearity with other factors such as age, high grade CRS, and renal dysfunction, which often coexist with cardiovascular disease. However, a graded association has consistently been noted between CRS grade and cardiovascular toxicity.125126131 For example, in one study the adjusted risk of MACE was eight times higher (hazard ratio 8.42, 3.48 to 20.40; P<0.001) in patients with grade ≥3 CRS and almost 30 times higher in those who developed grade 4 CRS (29.86, 9.80 to 90.94; P<0.001).126
Although no head-to-head comparison between different types of CAR-T is available, a pharmacovigilance study reported no significant association of CAR-T type with overall cardiovascular events.144 Finally, previous exposure to cardiotoxic antineoplastic therapy has been an area of interest given that most patients receiving CAR-T for hematologic malignancies received previous cardiotoxic therapies such as anthracyclines, cyclophosphamide, chest radiation therapy, and stem cell transplantation.115 However, such an association is difficult to tease out given that almost all patients receiving CAR-T have had such exposures.
Management
Cardiovascular monitoring during CAR-T and diagnosis of cardiovascular toxicity
Given that most cardiovascular events occur in the context of CRS, prompt recognition of CRS is important.116 Patients are usually monitored with continuous telemetry and pulse oximetry. In addition to CRS triggering cardiovascular events, both conditions have overlapping symptoms and signs such as dyspnea, hypoxia, and hypotension.136 The biomarkers of inflammation (ferritin, interleukin-6, C reactive protein) and cardiac biomarkers (troponin and natriuretic peptides) are often elevated with high grade CRS and associated with MACE and higher mortality rates.131 Despite their value in diagnosis and prognostication, recommendations and practice regarding their routine use varies significantly, given that they are sensitive but lack specificity.145 Although routine surveillance in patients without symptoms is of limited utility, cardiac biomarkers, electrocardiography, and echocardiography may be considered for patients with signs and/or symptoms of CRS grade 2 or 3 or suspicion for cardiovascular toxicity in order to guide management.73116136146
Treatment
Given that most MACE occur in the context of CRS, supportive care for the underlying cytokine storm is a cornerstone of management, including fluid resuscitation and, if appropriate, vasopressors and management of hypoxia with supplemental oxygen or mechanical ventilation (fig 5).116 Invasive hemodynamic monitoring may be necessary in acutely ill patients when the contribution of heart failure to the clinical presentation is unclear in order to tailor fluid management and/or pressor and/or inotrope management.116
CAR T cell therapy associated cardiovascular toxicity. BNP=B-type natriuretic peptide; CAR=chimeric antigen receptor; CV=cardiovascular; CRP=C reactive protein; CRS=cytokine release syndrome; ECG=electrocardiography; echo=echocardiography; HF=heart failure; IL=interleukin; INF=interferon; MAGE-A3=melanoma associated antigen-3; TCR=T cell receptor; TNF=tumor necrosis factor
Anti-interleukin-6 therapy is the primary pharmacologic therapy for CRS given the central role of interleukin-6 in CAR-T associated CRS (table 1).116 Tocilizumab, a monoclonal antibody that binds to interleukin-6 receptors and inhibits interleukin-6 signaling, is recommended as first line for treatment of CRS.116 A longer duration from onset of CRS to tocilizumab administration has been associated with increased cardiovascular events (odds ratio 1.22, 1.01 to 1.53; P=0.022).131 Although concerns existed about the association of anti-interleukin-6 agents with reduction in the anti-cancer efficacy of CAR-T, the aggregate evidence supports the prompt use of anti-interleukin-6 therapy with grade ≥3 CRS.73 The suggested dosing is 8 mg/kg intravenously, repeated up to three times, given eight hours apart as needed.131 Unlike tocilizumab, siltuximab is a monoclonal antibody that blocks interleukin-6 signaling by binding to interleukin-6 itself, which prevents it from activating immune effector cells.115 This has been an agent of investigation, given concerns arising from preclinical studies showing a paradoxical increase in interleukin-6 concentrations after tocilizumab therapy, potentially leading to a worsening of neurotoxicity.116147 Additionally, a severe shortage of tocilizumab occurred in 2021, given its use in covid-19, leading to the exploration of siltuximab as a first line agent.148 A retrospective cohort of 135 patients treated with CAR-T compared those receiving siltuximab versus tocilizumab and did not find any differences in outcomes, including need for corticosteroids (25% v 24%; P=1.00), intensive care unit transfer (5% v 13%; P=0.23), and length of hospital stay (10.5 v 14 days; P=0.09). The findings suggest that siltuximab is a valid alternative for CAR-T associated CRS.148 Corticosteroids are typically used as second line agents for CRS.149 For patients with concomitant neurotoxicity, corticosteroids are used as the frontline agents given that tocilizumab does not cross the blood-brain barrier.150 Other immunomodulatory agents such as interleukin-1 inhibitors remain investigational.150
The management of individual cardiovascular adverse events is usually directed at the clinical presentation.115 Cardiovascular medications and management often have to be considered in the context of accompanying CRS severity, as well as risk of bleeding given the higher prevalence of thrombocytopenia.115 Multidisciplinary cardiology-oncology discussion of risks and benefits of available therapeutic strategies is key.115
Prognostic impact of CAR-T associated cardiovascular events
The long term impact of CAR-T on cardiovascular health and outcomes of patients who experience cardiovascular toxicity is not well studied. A retrospective cohort study showed that 50% of patients who developed cardiomyopathy/heart failure achieved full recovery of systolic function, whereas 25% experienced partial recovery, and 25% had persistent declines in systolic function.127 The impact of cardiovascular events after CAR-T on overall survival and cancer outcomes is conflicting. A multicenter retrospective cohort study showed an increased adjusted risk of all cause mortality (hazard ratio 2.8, 1.6 to 4.7) and non-relapse mortality (3.5, 1.4 to 8.8) for patients who experienced severe cardiovascular events following CAR-T.141 However, another retrospective study observed similar 12 month progression-free survival (38% v 42%; P=0.55) and 12 month overall survival (58% v 62%; P=0.52) irrespective of the occurrence of MACE.131 Longitudinal follow-up and multi-institutional collaboration are needed to better characterize the long term impact of CAR-T associated cardiovascular events.
Cardiovascular evaluation before CAR-T
Given the potential for significant hemodynamic changes associated with CAR-T and CRS, patients with pre-existing cardiovascular disease are at risk of further complications.146 As the indications for and use of CAR-T expand, it is likely to be offered to a broader population, including those with cardiovascular disease.116151 Before CAR-T, all patients should have a cardiovascular assessment that includes electrocardiography, troponin, natriuretic peptide, and echocardiography.116146 Those with significant cardiovascular risk factors, concerning cardiovascular symptoms, or exercise intolerance may benefit from an ischemic evaluation to rule out significant obstructive coronary artery disease.116 Optimization of cardiovascular medications including antihypertensives, antiplatelet agents, and anticoagulants is crucial in anticipating risks associated with hypotension and severe thrombocytopenia due to CRS or bone marrow suppression, respectively.115116 In general, in collaboration with cardiology, either reducing the dose of such agents or discontinuing them temporarily is advisable.115
Guidelines
The 2022 European Society of Cardiology Guidelines on Cardio-Oncology are dedicated to diagnosis and management of cardiovascular disease related to the diverse spectrum of cancer therapeutics.54 They contain comprehensive recommendations regarding ICIMy but with limited recommendations for other cardiovascular events related to ICIs and CAR-T cardiovascular toxicities. Other societies have also released recommendations for diagnosis, and some for management, of ICIMy and CRS with subtle differences (table 1). These recommendations will continue to evolve as data are generated for these relatively newer cancer therapies that have become available and increasingly used over the past decade.
Conclusions
The expanding use of immune based therapies with improved outcomes across a spectrum of cancers has led to rapidly growing numbers of patients exposed to these therapies and, therefore, at risk of cardiovascular toxicities. Whereas the cardiovascular adverse effects of ICIs are predominantly related to uninhibited direct immune effects on the cardiovascular system, those due to CAR-T are associated mostly with CRS. With ICIs, early recognition of cardiovascular irAEs, specifically ICIMy, resulting in earlier interruption of ICIs and initiation of immunosuppression, are key to improving outcomes. Conversely, cardiovascular optimization before CAR-T and treatment of CRS early with interleukin-6 blockers along with supportive treatment are germane to prevention and mitigation of cardiovascular events. Future studies are needed to identify more specific non-invasive strategies for diagnosis of ICIMy, clarify the role of more targeted immunosuppression (for example, CTLA-4 agonists and ruxolitinib) to treat ICIMy, and identify which patients can be safely restarted on ICI therapy after ICIMy. Furthermore, understanding the role of ICIs in atherosclerotic progression is critical given the increasing survivorship after ICI treatment. As the field of adoptive cellular therapies expands beyond CAR-T, understanding similarities and differences in cardiovascular effects between newer strategies and CAR-T will play an important role in patient selection, cardiovascular optimization and monitoring, and mitigation of cardiovascular adverse events. Collaboration between oncologists and cardio-oncologists (that is, cardiologists who specialize in the prevention and treatment of cancer therapy related cardiovascular toxicities and coexisting cardiovascular disease in patients with cancer) is needed, now more than ever.
Glossary of abbreviations
CAR—chimeric antigen receptor
CAR-T—chimeric antigen receptor T cell therapy
CMR—cardiovascular magnetic resonance
CTLA-4—cytotoxic T lymphocyte antigen-4
ICIMy—ICI associated myocarditis
ICI—immune checkpoint inhibitor
irAE—immune related adverse event
LAG-3—lymphocyte activation gene-3
LVEF—left ventricular ejection fraction
MACE—major adverse cardiovascular events
PD-1—programmed cell death protein-1
PD-L1—PD-ligand 1
RCT—randomized clinical trial
TNF—tumor necrosis factor
Questions for future research
How can we better diagnose immune checkpoint inhibitor associated myocarditis (ICIMy) by using non-invasive tests?
What is the role of upfront, more targeted and steroid sparing immunosuppression (eg, cytotoxic T lymphocyte antigen agonists, ruxolitinib) in the treatment of ICIMy?
Which patients can be safely restarted on lifesaving immune checkpoint inhibitor therapy after ICIMy?
Do immune checkpoint inhibitors accelerate atherosclerosis and increase the incidence of clinical atherosclerotic cardiovascular events?
Will newer adoptive cellular therapies be associated with similar cardiovascular adverse events to current chimeric antigen receptor T cell therapies?
Footnotes
Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors
Contributors: NLP did the initial literature search. AD and NLP conceptualized and designed the work. All authors provided drafts of assigned sections, which were compiled, and the manuscript was revised by all authors. NLP and AD made the final editorial changes and are the guarantors.
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: NLP is partly supported by the Cancer Prevention Research Institute of Texas (CPRIT) RP200670, NIH/NCI 1P01CA261669-01, and has served as a consultant for Kiniksa Pharmaceuticals and Replimmune; EK is supported in part by NIH/NCI 1RO1HL157273 and CPRIT RP200381; AD is supported in part by the Ting Tsung and Wei Fong Distinguished Chair, NIH/NCI 1RO1HL157273, and CPRIT RP200381 and has served as a consultant for Bayer.
Patient involvement: No patients were asked for input in the creation of this article.
Provenance and peer review: Commissioned; externally peer reviewed.