Immune-checkpoint-inhibitor (ICI)–associated myotoxicity involves the heart (myocarditis) and skeletal muscles (myositis), which frequently occur concurrently and are highly fatal. We report the results of a strategy that included identification of individuals with severe ICI myocarditis by also screening for and managing concomitant respiratory muscle involvement with mechanical ventilation, as well as treatment with the CTLA4 fusion protein abatacept and the JAK inhibitor ruxolitinib. Forty cases with definite ICI myocarditis were included with pathologic confirmation of concomitant myositis in the majority of patients. In the first 10 patients, using recommended guidelines, myotoxicity-related fatality occurred in 60%, consistent with historical controls. In the subsequent 30 cases, we instituted systematic screening for respiratory muscle involvement coupled with active ventilation and treatment using ruxolitinib and abatacept. The abatacept dose was adjusted using CD86 receptor occupancy on circulating monocytes. The myotoxicity-related fatality rate was 3.4% (1/30) in these 30 patients versus 60% in the first quartile (P < 0.0001). These clinical results are hypothesis-generating and need further evaluation.

Significance:

Early management of respiratory muscle failure using mechanical ventilation and high-dose abatacept with CD86 receptor occupancy monitoring combined with ruxolitinib may be promising to mitigate high fatality rates in severe ICI myocarditis.

See related commentary by Dougan, p. 1040.

This article is highlighted in the In This Issue feature, p. 1027

Immune-checkpoint inhibitors (ICI) have revolutionized cancer treatment and have been approved in multiple cancer types, and include antibodies that target CTLA4, PD-1, the PD-1 ligand PD-L1, and LAG3 (1). ICIs restore T cell–mediated host immunity against tumors (1), but they may also induce immune-related adverse events (irAE; ref. 2) including ICI myocarditis, which occurs infrequently (∼1%) but carries a 30% to 50% mortality rate (3–7). Clinical suspicion for ICI myocarditis often arises after associated symptoms and increases in cardiac biomarkers (troponins) or electrocardiographic alterations (3, 8–10). Endomyocardial biopsy is the gold standard for establishing the diagnosis (10, 11), whereas cardiac magnetic resonance imaging (cMRI) and echocardiography may be misleadingly normal especially early in the clinical course (10, 12, 13). Other forms of myotoxicity often accompany myocarditis, and these include peripheral myositis, which can affect respiratory muscles or mimic myasthenia gravis–like syndrome (6, 10, 14). The pathophysiology of these myotoxicities is characterized by myocyte necrosis induced by lymphohistiocytic infiltrates secondary to activation of clonal autoreactive T cells against muscle antigens, notably α-myosin expressed in the heart and peripheral muscles (7, 15). Preliminary data suggest that myositis, particularly when affecting respiratory muscles, accompanying ICI myocarditis can negatively affect survival and requires an aggressive patient care strategy (5, 6, 10, 14). The main causes of death attributed to ICI myotoxicities are life-threatening ventricular arrhythmias, cardiogenic shock, or hypercapnic respiratory muscle failure (10).

Current treatment of ICI myocarditis includes empiric corticosteroids, which are not universally effective; second-line therapies including immunosuppressive therapies targeting T cells have been described (16–18). In preclinical models, CTLA4 signaling appears to play a critical role in the development of myocarditis, with CTLA4–immunoglobulin (Ig) fusion protein (abatacept) attenuating myocardial inflammation (19). Preliminary case reports have suggested that the CTLA4–Ig fusion protein abatacept may be a promising option (19–21). However, the optimal dosing scheme and the effect of combining CTLA4–Ig with other immunosuppressants are unknown. Abatacept (a CTLA4–Ig) acts as a decoy receptor for CD80/CD86 expressed on antigen-presenting cells (e.g., macrophages), interfering with T-cell costimulation through CD28. Therefore, abatacept leads to global T-cell anergy by reversing pathways activated by ICI (1, 19–21). When abatacept is used for approved indications such as rheumatoid arthritis, CD86 receptor occupancy on circulating monocytes (CD86 RO) is a pharmacodynamic biomarker of clinical activity, with CD86 RO ≥80% as a target for maximal efficacy (22, 23). Another consideration is the time of onset of abatacept, which is slow. Consistently, in a mouse model of ICI myocarditis, myocardial immune infiltration was attenuated after 10 weeks of abatacept treatment; 2-week duration of treatment with abatacept had minimal effect on immune infiltration (19). Preliminary reports suggest that abatacept clinical efficacy can be enhanced by addition of the JAK inhibitor ruxolitinib, which impairs T-cell activation via blockade of proinflammatory cytokines and is used to treat graft-versus-host disease, a condition in which donor T cells attack host tissue (21, 24). Ruxolitinib (a JAK1/JAK2 inhibitor) is thought to exert a rapid (within hours of start) synergistic effect with abatacept (slower time to onset of effect) by also decreasing CD86 expression on macrophages while acting downstream in the immunologic synapse, inactivating T cells (1, 25–27).

With these considerations in mind, we have developed a strategy in which we identify individuals with severe ICI myocarditis by also screening for and managing concomitant respiratory muscle involvement with mechanical ventilation and treating these individuals with high-dose CTLA4 fusion protein abatacept with CD86 RO monitoring and the JAK inhibitor ruxolitinib. We applied this strategy to 30 consecutive patients with ICI myocarditis. By comparison, we also describe a cohort of 10 patients seen previously in which patients underwent usual recommended care with the initiation of corticosteroids and the addition of a secondary immunosuppressive therapy.

Study Cohort

Of the 69 patients consecutively admitted for suspicion of ICI myocarditis, 40 were confirmed as definite cases and are included in this report (Fig. 1 for flow chart, Table 1 for demographics, and Supplementary Table S1 for diagnostic certainty criteria). These patients were analyzed as the first 10 (quartile 1, or Q1, admitted 05/2018–03/2020) treated as per current expert consensus guidelines, starting with high-dose corticosteroids boluses followed by second-line agents in case of corticosteroid resistance (16) versus the subsequent 30 (Q2–4, 03/2020–08/2021) treated with a mechanism-based approach. The mechanism-based approach included prompt use of high-dose abatacept with real-time CD86 RO monitoring, ruxolitinib, lower doses of corticosteroids, and systematic screening for and early management of concomitant respiratory muscle failure. The median age of the overall cohort was 72 [interquartile range (IQR) = 62, 79] years; 58% were male; and the most represented cancers were lung (9/40, 23%), skin (9/40, 23%), and genitourinary (10/40, 25%) cancers. A majority of patients (30/40, 75%) developed severe (grade ≥3) myotoxicity (see Table 1 and Fig. 2 for details concerning severity features of these patients and Supplementary Tables S2–S5 for detailed severity adjudication criteria). During the course of the disease, all patients had increased troponin-T (Fig. 2), most (34/40, 85%) developed abnormal electrocardiogram (ECG) features, and 9/40 (23%) patients developed severe systolic cardiac dysfunction (grade ≥3). A total of 13% (5/40) required inotropic agents or extracorporeal hemodynamic support. At least one cMRI examination [median/patient: 2 (1–3)] supported the diagnosis in 26/35 patients (74%; refer to Table 1 for results and Supplementary Table S1 for cMRI diagnostic criteria; 5 patients died prior to cMRI or refused it). Cardiac and muscular biopsies supported the ICI myotoxicity diagnosis in 24 of 32 (75%; 8 patients died prior to cardiac biopsy or refused it) and 36 of 40 (90%), respectively (see Table 1 for results and Supplementary Table S1 for pathology criteria). Cardiac and muscular histopathology findings consistently showed T-cell and macrophage infiltration with associated myocyte necrosis (Supplementary Fig. S1A–S1E). A total of 12 of 40 (30%) developed ventricular tachycardias (6/12, 50% sustained), and 9 of 40 (23%) developed severe conduction disorders (7 complete atrioventricular block and 2 sinus node dysfunction requiring pacemaker implantation). Respiratory muscle involvement was present in 27 of 38 cases (71%; 2 patients in Q1 died prior to evaluation), with specific evidence for diaphragmatic involvement in 22 of 38 cases (see Supplementary Table S3 for the detailed framework used to evaluate the global respiratory muscle and more specifically the diaphragm function). Respiratory muscle involvement was severe and required mechanical ventilation in over a third of patients (15/38; Table 1). Electromyogram showed a myopathic pattern in half (19/37; 3 died before evaluation) of patients with no associated neuromuscular junction disorder (0/28 abnormal repetitive nerve stimulation test) despite clinical features of myasthenia gravis (Table 1). Acetylcholine receptor antibodies were positive (>0.5 nmol/L) in 8/40 (20%), of which 5/5 (100%) were previously known as abnormal before initiation of ICI. Antimuscle-specific kinase antibodies were negative in all patients. In the 4 patients autopsied during the index hospitalization for ICI myotoxicity, myocarditis and diaphragmatic myositis were severe and associated with lesions in peripheral muscles (Supplementary Fig. S1A–S1E).

Figure 1.

Flow chart and evolution of the mortality in our prospective cohort of patients admitted for suspicion of ICI myocarditis in Pitié-Salpêtrière Hospital, Paris, France.

Figure 1.

Flow chart and evolution of the mortality in our prospective cohort of patients admitted for suspicion of ICI myocarditis in Pitié-Salpêtrière Hospital, Paris, France.

Close modal
Table 1.

Demographics, diagnostic workup, and severity criteria of patients included by time period of inclusion

Dates of inclusion in the consecutive quartiles (months/years)Overall 05/18–08/21Quartile 1 05/18–03/20Quartile 2 03/20–12/20Quartile 3 12/20–04/21Quartile 4 05/21–08/21P valuea
n/N (%), or median [interquartile range as appropriate]
Demographics 
Age, years 72 [62–79] 67 [61–77] 68 [51–79] 75 [68–84] 72 [64–75] 0.33 
Sex, male (%) 23 (58%) 6 (60%) 6 (60%) 6 (60%) 5 (50%) 0.67 
Body mass index, kg/m2 24 [22–27] 24 [19–26] 24 [21–29] 25 [20–26] 24 [22–29] 0.63 
Time to onset after first ICI dose 
 Days 33 [21–42] 32 [16–64] 34 [27–41] 28 [21–41] 34 [19–64] 0.97 
 Number of doses 2 [1–3] 2 [1–4] 2 [1–2] 2 [1–2] 2 [1–3] 0.92 
Past medical history (%) 
 Heart failure 2 (5%) 0 (0%) 2 (20%) 0 (0%) 0 (0%) 0.52 
  Hypertension 20 (50%) 4 (40%) 5 (50%) 5 (50%) 6 (60%) 0.40 
  Diabetes 7 (18%) 2 (20%) 1 (10%) 2 (20%) 2 (20%) 0.85 
  Dyslipidemia 8 (20%) 3 (30%) 2 (20%) 1 (10%) 2 (20%) 0.48 
  Autoimmune disease 7 (18%) 2 (20%) 2 (20%) 0 (0%) 3 (30%) 0.85 
Cancer status (%) 
 Metastatic 33 (83%) 8 (80%) 7 (70%) 9 (90%) 9 (90%) 0.91 
Type of ICI, anti–PD-(L)1 (%)      0.10 
 Monotherapy 30 (75%) 9 (90%) 8 (80%) 7 (70%) 6 (60%)  
 Combination with anti-CTLA4 10 (25%) 1 (10%) 2 (20%) 3 (30%) 4 (40%)  
Diagnostic workup 
Cardiac symptoms 
 Chest pain 11 (28%) 6 (60%) 3 (30%) 2 (20%) 0 (0%) 0.003 
  Syncope/faintness 7 (18%) 3 (30%) 1 (10%) 2 (20%) 1 (10%) 0.35 
  Dyspnea 24 (60%) 9 (90%) 5 (50%) 5 (50%) 5 (50%) 0.08 
  Palpitations 6 (15%) 2 (20%) 3 (30%) 1 (10%) 0 (0%) 0.11 
Muscular symptoms 
  Muscular pain or weakness 28 (70%) 8 (80%) 8 (80%) 8 (80%) 4 (40%) 0.06 
  Ptosis 16 (40%) 4 (40%) 5 (50%) 3 (30%) 4 (40%) 0.77 
  Diplopia 16 (40%) 5 (50%) 5 (50%) 2 (20%) 4 (40%) 0.39 
  Dysphagia 13 (33%) 5 (50%) 3 (30%) 2 (20%) 3 (30%) 0.29 
  Dysphonia 14 (35%) 4 (40%) 4 (40%) 2 (20%) 4 (40%) 0.77 
Electrocardiogram abnormal 34 (85%) 9 (90%) 10 (100%) 8 (80%) 7 (70%) 0.11 
Creatinine kinase abnormal 36 (90%) 9 (90%) 8 (80%) 9 (90%) 10 (100%) 0.35 
cMRI tissular analysisb      0.33 
 Definite 16/35 (46%) 5/8 (63%) 6/9 (67%) 2/10 (20%) 3/8 (38%)  
 Suggestive 10/35 (29%) 2/8 (25%) 2/9 (22%) 3/10 (30%) 3/8 (38%)  
 Normal 9/35 (26%) 1/8 (13%) 1/9 (11%) 5/10 (50%) 2/8 (25%)  
Endomyocardial pathologyb      0.03 
 Definite 3/32 (9%) 2/9 (22%) 0/8 (0%) 0/9 (0%) 1/6 (17%)  
 Suggestive 21/32 (66%) 3/9 (33%) 4/8 (50%) 9/9 (100%) 5/6 (83%)  
 Normal 8/32 (25%) 4/9 (44%) 4/8 (50%) 0/9 (0%) 0/6 (0%)  
Peripheral muscle pathologyb      0.74 
  Definite 32 (80%) 8 (80%) 8 (80%) 8 (80%) 8 (80%)  
  Suggestive 4 (10%) 1 (10%) 2 (20%) 0 (0%) 1 (10%)  
  Nonspecific lesions 2 (5%) 0 (0%) 0 (0%) 1 (10%) 1 (10%)  
  Normal 2 (5%) 1 (10%) 0 (0%) 1 (10%) 0 (0%)  
Severity grade (at presentation)c 
Cardiac dysfunction      0.65 
 Grade ≤2 39/40 (98%) 10/10(100%) 9/10 (90%) 10/10 (100%) 10/10 (100%)  
 Grade ≥3 (severe) 1/40 (2%) 0/10 (0%) 1/10 (10%) 0/10 (0%) 0/10 (0%)  
Cardiac arrhythmia      0.16 
 Grade ≤2 36/40 (90%) 8/10 (80%) 9/10 (90%) 9/10 (90%) 10/10 (100%)  
  Grade ≥3 (severe) 4/40 (10%) 2/10 (20%) 1/10 (10%) 1/10 (10%) 0/10 (0%)  
Respiratory muscle dysfunction      
 Grade ≤2 36/40 (90%) 9/10 (90%) 9/10 (90%) 9/10 (90%) 9/10 (90%)  
  Grade ≥3 (severe) 4/40 (10%) 1/10 (10%) 1/10 (10%) 1/10 (10%) 1/10 (10%)  
Severity maximal gradec 
Cardiac dysfunction      0.06 
  Grade ≤2 31/40 (78%) 5/10 (50%) 8/10 (80%) 10/0 (100%) 8/10 (80%)  
  Grade ≥3 (severe) 9/40 (22%) 5/10 (50%) 2/10 (20%) 0/10 (0%) 2/10 (20%)  
Cardiac arrhythmias      0.26 
  Grade ≤2 23/40 (58%) 4/10 (40%) 5/10 (50%) 7/10 (70%) 7/10 (70%)  
  Grade ≥3 (severe) 18/40 (45%) 6/10 (60%) 5/10 (50%) 3/10 (30%) 4/10 (40%)  
Respiratory muscle dysfunction      0.84 
 Grade ≤2 23/38 (61%) 3/8 (38%) 6/10 (60%) 9/10 (90%) 5/10 (50%)  
 Grade ≥3 (severe) 15/38 (39%) 5/8 (63%) 4/10 (40%) 1/10 (10%) 5/10 (50%)  
Myotoxicity (overall)      0.74 
  Grade ≤2 10/40 (25%) 2/10 (20%) 2/10 (20%) 4/10 (40%) 2/10 (20%)  
  Grade ≥3 (severe) 30/40 (75%) 8/10 (80%) 8/10 (80%) 6/10 (60%) 8/10 (80%)  
Other severity criteria 
Respiratory muscle involvementd      0.23 
 Definite 20/38 (53%) 4/8 (50%) 5/10 (50%) 8/10 (80%) 3/10 (30%)  
 Probable 7/38 (18%) 2/8 (25%) 1/10 (10%) 0/10 (0%) 4/10 (40%)  
 Absent 11/38 (29%) 2/8 (25%) 4/10 (40%) 2/10 (20%) 3/10 (30%)  
Troponin-T peak value (ratio vs. 99th upper reference limit) 111 [31–175] 245 [50–991] 96 [34–157] 64 [18–153] 115 [43–150] 0.18 
Dates of inclusion in the consecutive quartiles (months/years)Overall 05/18–08/21Quartile 1 05/18–03/20Quartile 2 03/20–12/20Quartile 3 12/20–04/21Quartile 4 05/21–08/21P valuea
n/N (%), or median [interquartile range as appropriate]
Demographics 
Age, years 72 [62–79] 67 [61–77] 68 [51–79] 75 [68–84] 72 [64–75] 0.33 
Sex, male (%) 23 (58%) 6 (60%) 6 (60%) 6 (60%) 5 (50%) 0.67 
Body mass index, kg/m2 24 [22–27] 24 [19–26] 24 [21–29] 25 [20–26] 24 [22–29] 0.63 
Time to onset after first ICI dose 
 Days 33 [21–42] 32 [16–64] 34 [27–41] 28 [21–41] 34 [19–64] 0.97 
 Number of doses 2 [1–3] 2 [1–4] 2 [1–2] 2 [1–2] 2 [1–3] 0.92 
Past medical history (%) 
 Heart failure 2 (5%) 0 (0%) 2 (20%) 0 (0%) 0 (0%) 0.52 
  Hypertension 20 (50%) 4 (40%) 5 (50%) 5 (50%) 6 (60%) 0.40 
  Diabetes 7 (18%) 2 (20%) 1 (10%) 2 (20%) 2 (20%) 0.85 
  Dyslipidemia 8 (20%) 3 (30%) 2 (20%) 1 (10%) 2 (20%) 0.48 
  Autoimmune disease 7 (18%) 2 (20%) 2 (20%) 0 (0%) 3 (30%) 0.85 
Cancer status (%) 
 Metastatic 33 (83%) 8 (80%) 7 (70%) 9 (90%) 9 (90%) 0.91 
Type of ICI, anti–PD-(L)1 (%)      0.10 
 Monotherapy 30 (75%) 9 (90%) 8 (80%) 7 (70%) 6 (60%)  
 Combination with anti-CTLA4 10 (25%) 1 (10%) 2 (20%) 3 (30%) 4 (40%)  
Diagnostic workup 
Cardiac symptoms 
 Chest pain 11 (28%) 6 (60%) 3 (30%) 2 (20%) 0 (0%) 0.003 
  Syncope/faintness 7 (18%) 3 (30%) 1 (10%) 2 (20%) 1 (10%) 0.35 
  Dyspnea 24 (60%) 9 (90%) 5 (50%) 5 (50%) 5 (50%) 0.08 
  Palpitations 6 (15%) 2 (20%) 3 (30%) 1 (10%) 0 (0%) 0.11 
Muscular symptoms 
  Muscular pain or weakness 28 (70%) 8 (80%) 8 (80%) 8 (80%) 4 (40%) 0.06 
  Ptosis 16 (40%) 4 (40%) 5 (50%) 3 (30%) 4 (40%) 0.77 
  Diplopia 16 (40%) 5 (50%) 5 (50%) 2 (20%) 4 (40%) 0.39 
  Dysphagia 13 (33%) 5 (50%) 3 (30%) 2 (20%) 3 (30%) 0.29 
  Dysphonia 14 (35%) 4 (40%) 4 (40%) 2 (20%) 4 (40%) 0.77 
Electrocardiogram abnormal 34 (85%) 9 (90%) 10 (100%) 8 (80%) 7 (70%) 0.11 
Creatinine kinase abnormal 36 (90%) 9 (90%) 8 (80%) 9 (90%) 10 (100%) 0.35 
cMRI tissular analysisb      0.33 
 Definite 16/35 (46%) 5/8 (63%) 6/9 (67%) 2/10 (20%) 3/8 (38%)  
 Suggestive 10/35 (29%) 2/8 (25%) 2/9 (22%) 3/10 (30%) 3/8 (38%)  
 Normal 9/35 (26%) 1/8 (13%) 1/9 (11%) 5/10 (50%) 2/8 (25%)  
Endomyocardial pathologyb      0.03 
 Definite 3/32 (9%) 2/9 (22%) 0/8 (0%) 0/9 (0%) 1/6 (17%)  
 Suggestive 21/32 (66%) 3/9 (33%) 4/8 (50%) 9/9 (100%) 5/6 (83%)  
 Normal 8/32 (25%) 4/9 (44%) 4/8 (50%) 0/9 (0%) 0/6 (0%)  
Peripheral muscle pathologyb      0.74 
  Definite 32 (80%) 8 (80%) 8 (80%) 8 (80%) 8 (80%)  
  Suggestive 4 (10%) 1 (10%) 2 (20%) 0 (0%) 1 (10%)  
  Nonspecific lesions 2 (5%) 0 (0%) 0 (0%) 1 (10%) 1 (10%)  
  Normal 2 (5%) 1 (10%) 0 (0%) 1 (10%) 0 (0%)  
Severity grade (at presentation)c 
Cardiac dysfunction      0.65 
 Grade ≤2 39/40 (98%) 10/10(100%) 9/10 (90%) 10/10 (100%) 10/10 (100%)  
 Grade ≥3 (severe) 1/40 (2%) 0/10 (0%) 1/10 (10%) 0/10 (0%) 0/10 (0%)  
Cardiac arrhythmia      0.16 
 Grade ≤2 36/40 (90%) 8/10 (80%) 9/10 (90%) 9/10 (90%) 10/10 (100%)  
  Grade ≥3 (severe) 4/40 (10%) 2/10 (20%) 1/10 (10%) 1/10 (10%) 0/10 (0%)  
Respiratory muscle dysfunction      
 Grade ≤2 36/40 (90%) 9/10 (90%) 9/10 (90%) 9/10 (90%) 9/10 (90%)  
  Grade ≥3 (severe) 4/40 (10%) 1/10 (10%) 1/10 (10%) 1/10 (10%) 1/10 (10%)  
Severity maximal gradec 
Cardiac dysfunction      0.06 
  Grade ≤2 31/40 (78%) 5/10 (50%) 8/10 (80%) 10/0 (100%) 8/10 (80%)  
  Grade ≥3 (severe) 9/40 (22%) 5/10 (50%) 2/10 (20%) 0/10 (0%) 2/10 (20%)  
Cardiac arrhythmias      0.26 
  Grade ≤2 23/40 (58%) 4/10 (40%) 5/10 (50%) 7/10 (70%) 7/10 (70%)  
  Grade ≥3 (severe) 18/40 (45%) 6/10 (60%) 5/10 (50%) 3/10 (30%) 4/10 (40%)  
Respiratory muscle dysfunction      0.84 
 Grade ≤2 23/38 (61%) 3/8 (38%) 6/10 (60%) 9/10 (90%) 5/10 (50%)  
 Grade ≥3 (severe) 15/38 (39%) 5/8 (63%) 4/10 (40%) 1/10 (10%) 5/10 (50%)  
Myotoxicity (overall)      0.74 
  Grade ≤2 10/40 (25%) 2/10 (20%) 2/10 (20%) 4/10 (40%) 2/10 (20%)  
  Grade ≥3 (severe) 30/40 (75%) 8/10 (80%) 8/10 (80%) 6/10 (60%) 8/10 (80%)  
Other severity criteria 
Respiratory muscle involvementd      0.23 
 Definite 20/38 (53%) 4/8 (50%) 5/10 (50%) 8/10 (80%) 3/10 (30%)  
 Probable 7/38 (18%) 2/8 (25%) 1/10 (10%) 0/10 (0%) 4/10 (40%)  
 Absent 11/38 (29%) 2/8 (25%) 4/10 (40%) 2/10 (20%) 3/10 (30%)  
Troponin-T peak value (ratio vs. 99th upper reference limit) 111 [31–175] 245 [50–991] 96 [34–157] 64 [18–153] 115 [43–150] 0.18 

aP values are those calculated for the comparison between quartiles, using Kruskal–Wallis and χ2 tests for trend when applied to quantitative and qualitative variables, respectively.

bSee Supplementary Table S1 for the detailed features used for this assessment.

cSee Supplementary Tables S4 and S5 for the detailed features used for this assessment.

dSee Supplementary Table S3 for the detailed features used for this assessment.

Figure 2.

Troponin-T circulating levels as a function of ICI myotoxicity severity. Evolution of circulating troponin-T levels (cTnT, ng/L) as a surrogate for myotoxicity severity grade in the 40 ICI myocarditis patients included in our study. All patients had increased troponin-T levels on admission. Peak values of troponin-T levels per patient are represented by * and track the severity grading of these patients. Five patients had grade 1 asymptomatic myocarditis, 4 of whom were not treated with any immunosuppressant (but ICI withhold) and evolved favorably. LVEF, left ventricular ejection fraction.

Figure 2.

Troponin-T circulating levels as a function of ICI myotoxicity severity. Evolution of circulating troponin-T levels (cTnT, ng/L) as a surrogate for myotoxicity severity grade in the 40 ICI myocarditis patients included in our study. All patients had increased troponin-T levels on admission. Peak values of troponin-T levels per patient are represented by * and track the severity grading of these patients. Five patients had grade 1 asymptomatic myocarditis, 4 of whom were not treated with any immunosuppressant (but ICI withhold) and evolved favorably. LVEF, left ventricular ejection fraction.

Close modal

Cardiomyotoxicities Outcomes with Evolving Management Protocol

The overall clinical demographic and diagnostic characteristics of patients in Q1 versus Q2–4 were similar with no difference in peak cardiac troponin-T reached and in the different myotoxicity severity features studied (arrhythmia, cardiac, and respiratory muscle dysfunction) upon presentation or during follow-up (Table 1). In Q1, all patients were treated with corticosteroids [≥500 mg/day intravenous methylprednisolone pulse ≥2 days in 9/10 (90%)] started within 2 (0–4) days of presentation. Eight of 10 patients did not respond to corticosteroids (“cortico-resistant”) and subsequently received plasmapheresis (1–5 cycles, n = 8), low-dose abatacept (n = 7), or mycophenolate mofetil (n = 4; time to start from presentation and dose are detailed in Supplementary Table S6 and Supplementary Fig. S2). ICI myotoxicity–related mortality was 60% (6/10) in Q1, with deaths due to cardiogenic shock or respiratory muscle failure (n = 3 each). Among the first 10 patients, the diagnosis of respiratory muscle involvement was confirmed on chart review in 6 of 8 (75%). From Q1, we also identified that plasmapheresis partially cleared ICI blood levels (Supplementary Fig. S3), but identification of PD-(L)1 expression on T cells was markedly delayed to a few months after pheresis (Fig. 3). We also inferred that plasmapheresis would clear abatacept (an example provided in Supplementary Fig. S4A), so we delayed using the drug after plasmapheresis (1). Moreover, in the first 10 patients, we found that the initial abatacept dosing strategy did not improve outcomes (60% of ICI-related mortality in Q1). Indeed, even with higher abatacept doses used subsequently in Q2–4, CD86 RO peak levels were often below the objective of CD86 RO ≥80% and we identified that CD86 RO varied dose dependently after abatacept injections (Fig. 4A and B).

Figure 3.

Effects of plasmapheresis on ICI circulating levels and immune checkpoint expression on peripheral blood mononuclear cells. Although ICI concentrations were decreased by plasmapheresis, there was a surge a few days after each plasmapheresis with ICI blood levels staying well above the limit of detection for months (<1 μg/mL for nivolumab; <2 μg/mL for durvalumab)—that is, well above their expected IC50 for their PD-(L)1 targets (∼0.3 μg/mL for nivolumab and ∼0.02 μg/mL for durvalumab; ref. 1). This phenomenon may explain why PD-1 and PD-L1 receptors on T cells (CD3+) and B cells (CD19+) are blocked by ICI for months after plasmapheresis. BID, bis in die, twice a day; MMF, mycophenolate mofetil.

Figure 3.

Effects of plasmapheresis on ICI circulating levels and immune checkpoint expression on peripheral blood mononuclear cells. Although ICI concentrations were decreased by plasmapheresis, there was a surge a few days after each plasmapheresis with ICI blood levels staying well above the limit of detection for months (<1 μg/mL for nivolumab; <2 μg/mL for durvalumab)—that is, well above their expected IC50 for their PD-(L)1 targets (∼0.3 μg/mL for nivolumab and ∼0.02 μg/mL for durvalumab; ref. 1). This phenomenon may explain why PD-1 and PD-L1 receptors on T cells (CD3+) and B cells (CD19+) are blocked by ICI for months after plasmapheresis. BID, bis in die, twice a day; MMF, mycophenolate mofetil.

Close modal
Figure 4.

Abatacept binds its target, CD86, on circulating monocytes dose dependently. A, Evolution of CD86 RO on circulating monocytes after abatacept injections as a function of the various dose injected in the patients with severe myotoxicity treated with abatacept in Q2–4 (n = 22). Data after the seventh injection are not shown (n = 2 patients). B, Correlation (Spearman) between delta CD86 RO (maximal value within 72 hours after each abatacept injection compared with the baseline just before each injection) and doses of abatacept injected.

Figure 4.

Abatacept binds its target, CD86, on circulating monocytes dose dependently. A, Evolution of CD86 RO on circulating monocytes after abatacept injections as a function of the various dose injected in the patients with severe myotoxicity treated with abatacept in Q2–4 (n = 22). Data after the seventh injection are not shown (n = 2 patients). B, Correlation (Spearman) between delta CD86 RO (maximal value within 72 hours after each abatacept injection compared with the baseline just before each injection) and doses of abatacept injected.

Close modal

With the above information from Q1 patients, we modified the treatment and monitoring strategy for patients in Q2–4. We stopped performing plasmapheresis in severe cases, instead opting for initiation of higher dose abatacept if grade ≥3 (severe) myotoxicity criteria were met. The abatacept starting dose was increased to ∼20 mg/kg on days 0, 5, and 14, and the total dose administered in the first 2 weeks was 60 (53–65) mg/kg in Q2–4 versus 41 (24–60) mg/kg in Q1 (P = 0.01 for Q1 vs. Q2–4; Supplementary Table S6 for the evolution of abatacept dose used per specific quartile). In severe cases in Q2–4 (22/30 patients), we added ruxolitinib to abatacept and corticosteroids based on preclinical and translational studies outlined below, as well as given the rapid onset of action of ruxolitinib (in contrast to abatacept, which has delayed onset of action). Ruxolitinib was used in 0 of 8 of severe cases in Q1 versus 17/22, 77% in Q2–4 (P = 0.0002). In Q2–4, abatacept was used in all 22 severe cases. In Q2–4, we used lower concomitant doses of corticosteroids, seeking less related adverse events versus Q1 [intravenous methylprednisolone equivalent pulse ≥500 mg/day in 13/30, 43% in Q2–4 vs. 9/10, 90% in Q1, P = 0.01; 155 (122–223) mg of mean daily dose in the first month of treatment in Q1 vs. 96 (21–156) mg/day in Q2–4, P = 0.02; Supplementary Table S6 for evolution of treatment modalities per specific quartile]. The median time between presentation, appearance of grade ≥3 (severe) ICI myotoxicity criteria, and use of immunosuppressants are shown in Supplementary Fig. S2 and Supplementary Table S6. Frequencies of side effects observed on immunosuppressants, including infections, are detailed in Supplementary Table S6. In Q2–4, all patients were screened for respiratory muscle involvement for a decision concerning ventilator indication and management. Although the proportion of severe cases (grade ≥3) was comparable in Q1 (8/10, 80%) and Q2–4 (22/30, 73%, P = 0.67), the combined pharmacotherapeutic and ventilatory strategy was associated with a decrease in ICI-related myotoxicity death from 6/10 (60%) in Q1 to 1/30 (3%) in Q2–4 (P < 0.0001). When we restricted the analysis to severe (grade ≥3) cases, results were similar with 6/8 (80%) ICI-related myotoxicity death in Q1 versus 1/22 (5%) in Q2–4 (P < 0.0001). With the use of screening for respiratory muscle failure in Q2–Q4, 10/30 (30%) were found eligible for elective ventilation, and this was started in all but one who declined intubation (and who was the only death in Q2–Q4), and one with hepatic cancer who died of decompensated cirrhosis before ventilation was started. In these 8 ventilated patients (details on the modality of ventilation in Supplementary Fig. S2), recovery was complete in 6 of 8 [median weaning time = 17 (2–111) days] and 2 of 8 were recovering (weaned from diurnal ventilation) to being on residual nocturnal ventilation at 6-month follow-up.

There were deaths due to other causes (Fig. 1), including COVID-19 (in Q2–4) and sepsis (Q1). All-cause mortality at 3 months was 6 of 10 (60%) in Q1 and dropped to 7 of 30 (23%, P = 0.03) in Q2–4; the corresponding figures at 6 months were 7 of 10 (70%) versus 9 of 30 (30%, P = 0.03). Restricting analysis of all-cause mortality to severe cases showed similar results at 3 months (6/8, 75% in Q1 vs. 5/22, 23%; P = 0.009) and 6 months (7/8, 88% in Q1 vs. 7/22, 32%; P = 0.007). Median overall survival 6 months after presentation for ICI myocarditis was not reached in ICI myotoxicity survivors (n = 33) versus 0.7 months in those who died from ICI myotoxicity (n = 7, P < 0.001; Fig. 5A and B). Among patients surviving ICI-related myotoxicity, 82% had stable disease or a partial response, and the median progression-free survival (PFS) was 5.7 months (Fig. 5C and D). Best change from baseline in tumor burden as a function of cancer type and immunosuppressants received are shown in Fig. 5E and F, respectively. Pacemakers were placed in 8 patients and in 5 of 8 (63%), the conduction disorder resolved after 8 (2–22) days. Three patients were still pacemaker-dependent at 1-year follow-up: two were alive and the other died of COVID-19 with complete destruction of the sinus node at autopsy (Supplementary Fig. S1F).

Figure 5.

Overall survival and cancer-related outcomes in our cohort. Overall survival since ICI start (A) or ICI myocarditis presentation (B). Cancer PFS since ICI start (C) or ICI myocarditis presentation (D) in patients surviving ICI myotoxicity. Best change from baseline in tumor burden in patients surviving myotoxicity as a function of their cancer type (E) and immunosuppressants already received (F). For the latter analysis, 4 patients receiving ICI in an adjuvant setting were excluded and 1 patient died before evaluation. Among the 28 remaining patients, 8 (8/28, 29%) had partial response, 15 (15/28, 54%) had stable disease, and 5 (5/28, 18%) had progressive disease using RECIST 1.1 criteria (48). CI, confidence interval; ENT, ear, nose, and throat; mOS, median overall survival; mPFS, median progression-free survival; NR, not reached.

Figure 5.

Overall survival and cancer-related outcomes in our cohort. Overall survival since ICI start (A) or ICI myocarditis presentation (B). Cancer PFS since ICI start (C) or ICI myocarditis presentation (D) in patients surviving ICI myotoxicity. Best change from baseline in tumor burden in patients surviving myotoxicity as a function of their cancer type (E) and immunosuppressants already received (F). For the latter analysis, 4 patients receiving ICI in an adjuvant setting were excluded and 1 patient died before evaluation. Among the 28 remaining patients, 8 (8/28, 29%) had partial response, 15 (15/28, 54%) had stable disease, and 5 (5/28, 18%) had progressive disease using RECIST 1.1 criteria (48). CI, confidence interval; ENT, ear, nose, and throat; mOS, median overall survival; mPFS, median progression-free survival; NR, not reached.

Close modal

The Q2–4 group included 4 patients with asymptomatic myocarditis with grade 1 ICI myotoxicity criteria, diagnosed by abnormal troponin-T but with no concomitant symptomatic irAE; there was a mild increase of peak troponin-T of 5 (3–8) fold change above upper reference limit versus 116 (60–223) in the other 36 (P = 0.0002; Fig. 2). In these 4 patients, immunosuppressant drugs were not used, but ICIs were withheld. One such patient was rechallenged with ICI 2 months after the first episode but developed severe hepatitis (16) leading to permanent ICI discontinuation.

Rationale for JAK–STAT Pathway Inhibition to Treat ICI Myotoxicity

To elucidate underlying signaling mechanisms of ICI myocarditis, we have utilized a genetic mouse model (Ctla4+/−Pd-1−/−) of ICI myocarditis, which recapitulates the clinical and pathologic features seen in patients and was initially used to demonstrate the biological plausibility and efficacy of abatacept (19). Despite efficacy, abatacept onset was slow, with minimal clearance of cardiac immune infiltrates after 2 weeks of treatment with abatacept in mice (19). We, therefore, felt that abatacept would need to be combined with an immunosuppressive therapy with a shorter onset of action. In parallel, we performed bulk RNA sequencing (RNA-seq) assessing transcriptomics and pathways affected in our mouse model. We compared these identified pathways in mice to human tissue after performing bulk RNA-seq from endomyocardial biopsies of patients with ICI myocarditis compared with ICI-treated patients with no myocarditis (28). In ICI myocarditis mice (Ctla4+/−Pd-1−/−), bulk RNA-seq showed very distinct affected pathways in ICI myocarditis mice compared with unaffected control mice (Ctla4+/+Pd-1−/−; Fig. 6A). JAK–STAT signaling was especially upregulated, with Jak2 being the most significantly upregulated gene among the Jak family, and Stat1 and Stat4 were the most significantly upregulated genes among the Stat family (Fig. 6B). IFNγ/JAK2/STAT1 signaling, IL12/JAK2/STAT4 signaling, and IL6/JAK2/STAT3 signaling were the most significantly upregulated signaling pathways in ICI myocarditis mice compared with unaffected control mice (Fig. 6C). Bulk RNA-seq from patients also indicated that JAK2 was the only significantly upregulated gene in the JAK family in patients with ICI myocarditis (n = 9) compared with ICI-treated patients without evidence of myocarditis (n = 4; Fig. 6D). These findings from both mouse and human data support the potential beneficial effect of ruxolitinib (a JAK1/JAK2 inhibitor) in ICI myocarditis.

Figure 6.

The JAK–STAT pathway, particularly JAK2 (i.e., ruxolitinib target), is upregulated in mice and patients with ICI myocarditis. A, Heat map of differentially regulated genes in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). B, Gene expression of JAK and STAT family in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). ns, not significant. C, Expression of the genes in each JAK–STAT signaling pathway in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). D, Gene expression of JAK and STAT family in patients with ICI myocarditis (n = 9), in comparison with ICI-treated patients without myocarditis (ICI control, n = 4).

Figure 6.

The JAK–STAT pathway, particularly JAK2 (i.e., ruxolitinib target), is upregulated in mice and patients with ICI myocarditis. A, Heat map of differentially regulated genes in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). B, Gene expression of JAK and STAT family in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). ns, not significant. C, Expression of the genes in each JAK–STAT signaling pathway in ICI myocarditis (Ctla4+/−Pd-1−/− mice, n = 3) in comparison with unaffected control (Ctla4+/+Pd-1−/− mice, n = 4). D, Gene expression of JAK and STAT family in patients with ICI myocarditis (n = 9), in comparison with ICI-treated patients without myocarditis (ICI control, n = 4).

Close modal

In this study, we showed that prompt high-dose abatacept with real-time assessment of CD86 RO immunomonitoring combined with ruxolitinib, corticosteroids, and active management of respiratory muscle involvement was associated with a decrease in ICI myotoxicity–related death among patients with ICI myocarditis. With this strategy, ICI myotoxicity–related mortality decreased to ∼3% from historical controls (20%–60% in other published cohorts worldwide with ICI myocarditis) as well as the first 10 cases in our patient cohort (60% in our Q1 cases) who did not undergo screening and treatment with this strategy (2–4, 8, 29, 30). Although our study is not a clinical trial, our results may provide a promising platform by which to best treat the growing population that presents with fulminant ICI myocarditis with a mechanism-based and personalized assessment of drug receptor occupancy. The concept of abatacept use for ICI myocarditis initially came from successful treatment in a genetic mouse model in addition to clinical case reports (19, 20). Our study presents a comprehensive approach to treatment of these patients with attention to extracardiac myotoxicity associated with disease and attention to abatacept pharmacodynamics associated with ruxolitinib use.

One key element for improved care of ICI myocarditis has been our recognition that skeletal muscle involvement is near universal (95% abnormal muscular biopsy). With a very high frequency of respiratory muscle involvement comparable with that found in other types of systemic myositis (31), screening for (serial arterial blood gases) and management of respiratory muscle involvement appeared critical in avoiding life-threatening hypoventilation. As T cells and macrophages are critical in the development of ICI myocarditis and ICI myositis, optimizing immunosuppressant therapy was also a critical advance in our approach (1, 19, 20). Abatacept antagonizes the activation of ICI pathways through blockade of T-cell activation by antigen-presenting cells (i.e., blocking the interaction between CD86 on monocytes and CD28 on T cells; refs. 1, 24–27, 32). Moreover, using translational approaches in mice and humans, we were able to show that JAK2 signaling pathways were specifically activated in the heart of ICI myocarditis mice and patients (namely, IFNγ/JAK2/STAT1, IL12/JAK2/STAT4, and IL6/JAK2/STAT3; Fig. 6AD). These findings supported the specific use of ruxolitinib (a JAK1/JAK2 inhibitor notably blocking IL6 and IFNγ cytokine pathways) among available JAK inhibitors previously proposed in the treatment of other types of irAE such as ICI colitis (33–36). More research is needed to further decipher the full mechanistic spectrum of the potential beneficial effects of ruxolitinib to treat ICI myotoxicity, including its potential to decrease the antigenic presentation by muscle cells (making them harder targets) or to block some of the proapoptotic functions of IFNγ (37, 38).

All 8 patients in whom we implemented early mechanical ventilation for respiratory muscle failure had near-complete recovery to the point of full weaning in 6 cases and weaning during the day in 2 cases. This contrasts with the experience presented in a very similar cohort of severe ICI myotoxicity (14), in which 8 of 12 (67%) patients who developed respiratory failure attributable to respiratory muscle myositis died despite use of very high-dose corticosteroids, plasmapheresis, intravenous Ig, and rituximab (14). Notably, the time to recovery of respiratory muscle failure in our cohort varied from a few days to several months, apparently depending on the severity of initial presentation. Very slow recovery of respiratory muscle function after control of the causative insult has been well described in other neuromyopathies affecting respiratory muscles (39, 40). In the case of ICI myotoxicities, the concomitant use of high-dose corticosteroids favoring sarcopenia may also interfere deleteriously with muscular recovery (41, 42).

Several limitations of our analysis need to be recognized. As any nonrandomized observational study, there is a risk of unaccounted bias between compared groups. In our study, we did not find any difference between quartiles in known ICI myotoxicity severity criteria, including peak troponin levels, proportion of patients treated with combination ICI therapy, or proportion of patients presenting with severe arrhythmias, heart, and respiratory muscle failure (Table 1) but other unknown severity criteria (to date) may have been missed and different between quartiles (10). Our population was too small to perform ancillary analysis better adjusted to the dose of corticosteroids received or any other subtlety related to the type of anticancer treatment or immunosuppressant regimen received. Indeed, in Q2–4, both high-dose abatacept and addition of ruxolitinib were started precluding comparison of each treatment modality individually. Abatacept dose used within the 2 first weeks of treatment start in Q2–4 was higher versus Q1 (∼60 vs. 40 mg/kg, respectively), but it has to be acknowledged that both doses are much higher as compared with the intravenous loading dose used in abatacept's approved indication (10 mg/kg every 2 weeks in rheumatoid arthritis; ref. 1). Indeed, these doses are closer to the belatacept loading dose equivalent given in graft rejection prophylaxis (1).

A major challenge to consider while treating patients with severe irAEs is how to mitigate symptoms and mortality risk while preserving antitumor beneficial effects. In our cohort, abatacept and ruxolitinib were used only over a short period (∼1–2 months) of active treatment to achieve resolution of severity criteria, and this also limited the corticosteroid dose. To date, it is unclear to what extent the approach we used to treat ICI myotoxicities altered ICI anticancer efficacy or would be efficient in other life-threatening irAEs (2). It is also important to highlight that most irAE toxicities are not life-threatening and can even be detected while being infraclinical by active screening strategies (e.g., by systematic troponin surveillance; ref. 10). These low-grade irAEs would probably require no or a minimal immunosuppressive strategy contrasting with life-threatening irAE (16). The PFS estimate in our cohort (∼6 months) is unique, since most previous ICI myocarditis cohorts focused on describing the deadly cardiac outcomes and did not report cancer response. The variety of cancers in our cohort (mainly metastatic melanoma, lung, and kidney cancers) precluded us from performing reliable comparison with expected PFS in historical cohorts. However, median PFS in ICI clinical trials focusing on these latter cancers was consistently below 6 months (3–5 months; ref. 1). Altogether, our results are promising and should guide further research assessing the question of the optimal drug mix, dosage, and duration to be used to preserve ICI therapeutic effect while treating a severe irAE, the subject of further ongoing clinical study (NCT05195645). Comparison with other alternative pharmacologic approaches also targeting T cells (e.g., anti-thymoglobulin, alemtuzumab, anti-calcineurin drugs) and proposed in current treatment guidelines for irAEs also deserves further assessment (16). Lastly, the experimental part of the study in mice was done on a small number of animals. The latter data should therefore be considered preliminary and call for more detailed studies, particularly those at the protein level that would confirm what have been suggested by RNA. Meanwhile, the findings reported here justify a prompt referral to expert centers, able to evaluate the severity of patients notably by providing a systematic assessment and very close monitoring of respiratory muscle function in patients with ICI-myocarditis, to avoid missing lifesaving indications for ventilatory assistance.

Study Cohort

This prospective, single-center cohort included 40 consecutive patients with definite ICI myocarditis admitted to our cardio-oncology unit (Pitié-Salpêtrière, Paris, France) among 69 evaluated for suspected ICI-related myocarditis between October 5, 2018, and August 18, 2021 (flow chart in Fig. 1). ICI-related myotoxicity was confirmed by endomyocardial or muscle biopsies in all patients. Diagnostic certainty for ICI myocarditis was determined based on modified criteria by Bonaca and colleagues (Supplementary Table S1; ref. 11). The general diagnostic workup strategy upon ICI myocarditis suspicion is detailed in Supplementary Table S2. A diagnosis of respiratory muscle involvement was established using the 2019 European Respiratory Society statement (43). All these patients were included after written informed consent in the MASC (Myositis, DNA, Serum, Cells: Clinical Database and Biobank of Patients With Inflammatory Myopathies) prospective cohort (ethical approval CPP#2013 Ile de France VI; NCT04637672). Patients were treated after providing their written consent for compassionate use for abatacept and ruxolitinib. This study was conducted in accordance with the Declaration of Helsinki ethical guidelines. Follow-up exceeded 6 months. Data will be shared with other researchers upon reasonable request.

Adjudication of Events

Patients’ charts were reviewed by two respiratory physicians to evaluate respiratory muscle dysfunction and its severity using a multiparametric approach detailed in Supplementary Table S3 (43). Cardiac and skeletal muscle involvement (Supplementary Tables S4 and S5) and oncologic status (PFS) were prospectively assessed and graded (adapted from current guidelines) by two cardio-oncologists, a medical oncologist, and a radiologist (16, 44). Cardiac involvement was evaluated by cardiac MRI, endomyocardial biopsy, coronary angiography, echocardiography, electrocardiography, and troponin-T (ultrasensitive assay, Elecsys Roche Diagnostics) monitoring. Skeletal muscle involvement was evaluated by peripheral muscle and diaphragmatic MRI, muscular biopsy, electromyogram, diaphragmatic echography, and pulmonary functional test specifically seeking for respiratory muscle dysfunction (Supplementary Table S3). All causes of death were evaluated by three independent investigators. Autopsies were performed to determine the exact cause of death if family consented. ICI-related myotoxicities were considered severe (grade ≥3) if any of the following was present: severe arrhythmias (defined as appearance of ventricular tachyarrhythmias or high-degree atrioventricular block or sinus node dysfunction), heart failure (defined as heart failure symptoms requiring intravenous diuretics or inotropes or hemodynamic support), respiratory muscle failure leading to hypoventilation, or deterioration of bioclinical status despite corticosteroids (Supplementary Tables S4 and S5; refs. 16, 44).

Therapeutic Strategy

The first 10 patients (starting in 2018 until March 2020) were all treated with high-dose boluses of corticosteroids regardless of symptoms and severity grade (16, 44). In severe cases (grade ≥3, defined above), plasmapheresis, mycophenolate mofetil, and abatacept (∼10 mg/kg per injection approximately every 2 weeks) were used according to current guidelines (16, 44). After the first 10 patients (March 2020 to August 2021), our strategy was modified to include in these ICI myocarditis patients systematic screening for and management of concomitant respiratory muscle failure (myositis) including serial and repeated arterial blood gases to identify alveolar hypoventilation—increase in carbon dioxide partial pressure (PaCO2; Supplementary Table S3)—and in severe cases: (i) prompt initiation of higher-dose abatacept (≈20 mg/kg, 3 doses within the first 2 weeks of treatment start) with dose adjustment based on real-time assessment of CD86 RO targeting peak value of CD86 RO ≥80% within 72 hours of abatacept and residual CD86 RO ≥50% (Supplementary Figs. S2 and S4A for examples) until resolution of myotoxicity severity to grade ≤2 (Fig. 2; Supplementary Table S5 for details concerning severity grading features); (ii) addition of ruxolitinib used with abatacept; and (iii) decrease in corticosteroid dose used to avoid associated side effects (Supplementary Table S6 for details concerning reporting of side effects; refs. 16, 21, 44). Also, from March 2020 until August 2021, asymptomatic cases detected by screening (grade 1, Fig. 2; Supplementary Table S5) were monitored after withholding ICI with no systematic immunossupressant treatment (16, 45, 46). A summary of our general therapeutic management starting March 2020 depending on the grade of severity of ICI myotoxicity is given in Supplementary Table S5.

Immune Monitoring

Details concerning the methodology for profiling of immune-checkpoint proteins on peripheral blood mononuclear cells (CD86 RO on monocytes, PD-(L)1 expression on T cells) and monitoring of abatacept or ICI plasma levels are detailed in Supplementary Fig. S4B and Appendix Methods (47).

Statistical Analysis

Quantitative and qualitative data are expressed as median (IQR) and n (%). Their comparison was performed by Mann–Whitney or Kruskal–Wallis tests (when comparing quantitative variables between two groups or more, respectively) and χ2 tests for comparison of qualitative variables (including trends by quartiles when appropriate). Correlation between quantitative variables was assessed by the Spearman correlation. P < 0.05 was deemed significant.

JAK–STAT Pathway Translational Studies

RNA Isolation and Sequencing Using Mouse Cardiac Tissues.

Mice with ICI myocarditis (n = 3, ctla4+/− pd-1−/−) were compared with unaffected control mice (n = 4, ctla4+/+ pd-1−/−; ref. 19). Flash-frozen mouse tissue (cardiac ventricle) was homogenized using a TissueLyser II (Qiagen) for 2 minutes at 30 Hz. RNA was collected from dissociated tissue using the Qiagen RNA Kit (catalog #74136). Libraries were prepared using the Illumina TruSeq Stranded Total RNA Kit (Illumina). Sequencing was performed as paired-end sequencing, with a read length of 150 bp on the Illumina NovaSeq 6000 platform.

Bioinformatical Analysis Using Human Cardiac Tissues.

We used previously published bulk RNA-seq data from endomyocardial biopsies of patients with ICI myocarditis (n = 9) compared with ICI-treated patients without myocarditis (n = 4) included at Heidelberg (28). These data were obtained from the EBI ArrayExpress Database, with accession number E-MTAB-8867. The raw data files were aligned to the mouse or human genome using RNA STAR software (Version 2.7.3). Reads (reads per kilobase million) were extracted from the mapped files using the Rsubread Bioconductor package (Version 3.1). Differential gene expression was calculated using DESeq2 algorithms (European Molecular Biology Laboratory) with an FDR <0.05. For visualizations, we used the ggplot2 and heatmap3 R packages.

Data Availability

The mouse RNA-seq data have been assigned Gene Expression Omnibus accession number GSE225099. Data will be shared with other researchers upon reasonable request to the corresponding author.

J.-E. Salem reports personal fees and nonfinancial support from Bristol Myers Squibb during the conduct of the study; personal fees from BeiGene, AstraZeneca, and Banook, and grants and personal fees from Novartis outside the submitted work; and a patent for PCT/EP2021/071319 pending and a patent for PCT/EP2020/052554 issued. B. Abbar reports personal fees from Novartis, Astellas, Sanofi, AstraZeneca, and Bristol Myers Squibb outside the submitted work. S. Ederhy reports other support from Bristol Myers Squibb, Celgene, Pierre Fabre, Siemens, General Electric, Pfizer, and Amgen during the conduct of the study, as well as other support from Bayer, Pfizer, and Mylan outside the submitted work. P. Gougis reports personal fees from Bristol Myers Squibb outside the submitted work. M. Dres reports grants and personal fees from Lungpacer outside the submitted work. A. Demoule reports grants from the French Ministry of Health and AP-HP, grants, personal fees, and nonfinancial support from Lungpacer, grants and personal fees from Respinor, personal fees and nonfinancial support from Fisher & Paykel, and personal fees from Tribunal Administratif, Getinge, Astra, Gilead, and Mindray outside the submitted work. C. Straus reports grants from Centre Hospitalier Universitaire de Lille, France, and other support from Carmat outside the submitted work. J. Gonzalez-Bermejo reports personal fees from the Breas Company and Lowenstein outside the submitted work. N. Weiss reports personal fees from Owkin outside the submitted work. M. Kerneis reports personal fees from Kiniksa, Eligo, Bayer, and Sanofi outside the submitted work, as well as a patent for abatacept for the treatment of ICI-induced myocarditis pending. N. Hammoudi reports personal fees from Abbott, Novartis, Bayer, Bristol Myers Squibb, and Boehringer Ingelheim and nonfinancial support from Philips outside the submitted work. L. Lehmann reports personal fees from Daiichi Sankyo, Servier, AstraZeneca, Novartis, and MSD and other support from Senaca outside the submitted work. J.J. Moslehi reports personal fees from Bristol Myers Squibb, Cytokinetics, AstraZeneca, BeiGene, Kiniksa, Kurome, Lapcorp, Prelude, and Takeda during the conduct of the study. Y. Allenbach reports grants from Sanofi, and personal fees from Lilly and Bristol Myers Squibb outside the submitted work, as well as a patent for WO2022023490 pending and a patent for WO2020161045 pending. No disclosures were reported by the other authors.

J.-E. Salem: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft. M. Bretagne: Data curation, formal analysis, validation, investigation, visualization, writing–review and editing. B. Abbar: Data curation, formal analysis, investigation, visualization, writing–review and editing. S. Leonard-Louis: Resources, data curation, formal analysis, investigation, visualization, methodology, writing–review and editing. S. Ederhy: Data curation, investigation, writing–review and editing. A. Redheuil: Data curation, investigation, writing–review and editing. S. Boussouar: Data curation, investigation, writing–review and editing. L.S. Nguyen: Data curation, investigation, writing–review and editing. A. Procureur: Data curation, investigation, visualization, writing–review and editing. F. Stein: Data curation, investigation, writing–review and editing. C. Fenioux: Data curation, investigation, writing–review and editing. P. Devos: Data curation, investigation, writing–review and editing. P. Gougis: Software, formal analysis, visualization, writing–review and editing. M. Dres: Data curation, investigation, writing–review and editing. A. Demoule: Investigation, writing–review and editing. D. Psimaras: Data curation, investigation, writing–review and editing. T. Lenglet: Data curation, investigation, writing–review and editing. T. Maisonobe: Data curation, investigation, writing–review and editing. M. Pineton De Chambrun: Data curation, investigation, writing–review and editing. G. Hekimian: Data curation, investigation, writing–review and editing. C. Straus: Data curation, investigation, writing–review and editing. J. Gonzalez-Bermejo: Data curation, investigation, writing–review and editing. D. Klatzmann: Data curation, investigation, writing–review and editing. A. Rigolet: Data curation, investigation, writing–review and editing. P. Guillaume-Jugnot: Data curation, investigation, writing–review and editing. N. Champtiaux: Data curation, investigation, writing–review and editing. O. Benveniste: Data curation, investigation, writing–review and editing. N. Weiss: Data curation, investigation, writing–review and editing. S. Saheb: Data curation, investigation, writing–review and editing. P. Rouvier: Data curation, formal analysis, investigation, visualization, methodology, writing–review and editing. I. Plu: Data curation, investigation, writing–review and editing. E. Gandjbakhch: Data curation, investigation, writing–review and editing. M. Kerneis: Data curation, investigation, writing–review and editing. N. Hammoudi: Data curation, investigation, writing–review and editing. N. Zahr: Resources, data curation, software, formal analysis, investigation, writing–review and editing. C. Llontop: Data curation, investigation, writing–review and editing. C. Morelot-Panzini: Data curation, investigation, writing–review and editing. L. Lehmann: Data curation, formal analysis, writing–review and editing. J. Qin: Data curation, formal analysis, writing–review and editing. J.J. Moslehi: Supervision, visualization, methodology, writing–original draft. M. Rosenzwajg: Resources, data curation, software, formal analysis, validation, investigation, writing–review and editing. T. Similowski: Conceptualization, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft. Y. Allenbach: Resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing–review and editing.

This study was funded by CIC-1901. We thank Pr. Dan M. Roden (Vanderbilt University Medical Center, Nashville, TN) for his critical review of the manuscript.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note:Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).

1.
Geraud
A
,
Gougis
P
,
Vozy
A
,
Anquetil
C
,
Allenbach
Y
,
Romano
E
, et al
.
Clinical pharmacology and interplay of immune checkpoint agents: a yin-yang balance
.
Annu Rev Pharmacol Toxicol
2021
;
61
:
85
112
.
2.
Wang
DY
,
Salem
JE
,
Cohen
JV
,
Chandra
S
,
Menzer
C
,
Ye
F
, et al
.
Fatal toxic effects associated with immune checkpoint inhibitors: a systematic review and meta-analysis
.
JAMA Oncol
2018
;
4
:
1721
8
.
3.
Salem
JE
,
Manouchehri
A
,
Moey
M
,
Lebrun-Vignes
B
,
Bastarache
L
,
Pariente
A
, et al
.
Cardiovascular toxicities associated with immune checkpoint inhibitors: an observational, retrospective, pharmacovigilance study
.
Lancet Oncol
2018
;
19
:
1579
89
.
4.
Nguyen
LS
,
Cooper
LT
,
Kerneis
M
,
Funck-Brentano
C
,
Silvain
J
,
Brechot
N
, et al
.
Systematic analysis of drug-associated myocarditis reported in the World Health Organization pharmacovigilance database
.
Nat Commun
2022
;
13
:
25
.
5.
Allenbach
Y
,
Anquetil
C
,
Manouchehri
A
,
Benveniste
O
,
Lambotte
O
,
Lebrun-Vignes
B
, et al
.
Immune checkpoint inhibitor-induced myositis, the earliest and most lethal complication among rheumatic and musculoskeletal toxicities
.
Autoimmun Rev
2020
;
19
:
102586
.
6.
Anquetil
C
,
Salem
JE
,
Lebrun-Vignes
B
,
Johnson
DB
,
Mammen
AL
,
Stenzel
W
, et al
.
Immune checkpoint inhibitor-associated myositis: expanding the spectrum of cardiac complications of the immunotherapy revolution
.
Circulation
2018
;
138
:
743
5
.
7.
Johnson
DB
,
Balko
JM
,
Compton
ML
,
Chalkias
S
,
Gorham
J
,
Xu
Y
, et al
.
Fulminant myocarditis with combination immune checkpoint blockade
.
N Engl J Med
2016
;
375
:
1749
55
.
8.
Power
JR
,
Alexandre
J
,
Choudhary
A
,
Ozbay
B
,
Hayek
S
,
Asnani
A
, et al
.
Electrocardiographic manifestations of immune checkpoint inhibitor myocarditis
.
Circulation
2021
;
144
:
1521
3
.
9.
Power
JR
,
Alexandre
J
,
Choudhary
A
,
Ozbay
B
,
Hayek
SS
,
Asnani
A
, et al
.
Association of early electrical changes with cardiovascular outcomes in immune checkpoint inhibitor myocarditis
.
Arch Cardiovasc Dis
2022
;
115
:
315
30
.
10.
Lehmann
LH
,
Cautela
J
,
Palaskas
N
,
Baik
AH
,
Meijers
WC
,
Allenbach
Y
, et al
.
Clinical strategy for the diagnosis and treatment of immune checkpoint inhibitor-associated myocarditis: a narrative review
.
JAMA Cardiol
2021
;
6
:
1329
37
.
11.
Bonaca
MP
,
Olenchock
BA
,
Salem
JE
,
Wiviott
SD
,
Ederhy
S
,
Cohen
A
, et al
.
Myocarditis in the setting of cancer therapeutics: proposed case definitions for emerging clinical syndromes in cardio-oncology
.
Circulation
2019
;
140
:
80
91
.
12.
Zhang
L
,
Awadalla
M
,
Mahmood
SS
,
Nohria
A
,
Hassan
MZO
,
Thuny
F
, et al
.
Cardiovascular magnetic resonance in immune checkpoint inhibitor-associated myocarditis
.
Eur Heart J
2020
;
41
:
1733
43
.
13.
Ederhy
S
,
Salem
JE
,
Dercle
L
,
Hasan
AS
,
Chauvet-Droit
M
,
Nhan
P
, et al
.
Role of cardiac imaging in the diagnosis of immune checkpoints inhibitors related myocarditis
.
Front Oncol
2021
;
11
:
640985
.
14.
Aldrich
J
,
Pundole
X
,
Tummala
S
,
Palaskas
N
,
Andersen
CR
,
Shoukier
M
, et al
.
Inflammatory myositis in cancer patients receiving immune checkpoint inhibitors
.
Arthritis Rheumatol
2021
;
73
:
866
74
.
15.
Axelrod
ML
,
Meijers
WC
,
Screever
EM
,
Qin
J
,
Carroll
MG
,
Sun
X
, et al
.
T cells specific for alpha-myosin drive immunotherapy-related myocarditis
.
Nature
2022
;
611
:
818
26
.
16.
Schneider
BJ
,
Naidoo
J
,
Santomasso
BD
,
Lacchetti
C
,
Adkins
S
,
Anadkat
M
, et al
.
Management of immune-related adverse events in patients treated with immune checkpoint inhibitor therapy: ASCO guideline update
.
J Clin Oncol
2021
;
39
:
4073
126
.
17.
Wang
C
,
Lin
J
,
Wang
Y
,
Hsi
DH
,
Chen
J
,
Liu
T
, et al
.
Case series of steroid-resistant immune checkpoint inhibitor associated myocarditis: a comparative analysis of corticosteroid and tofacitinib treatment
.
Front Pharmacol
2021
;
12
:
770631
.
18.
Power
J
,
Meijers
W
,
Fenioux
C
,
Tamura
Y
,
Asnani
A
,
Alexandre
J
, et al
.
Predictors of steroid-refractory immune checkpoint inhibitor associated myocarditis
.
Eur Heart J
2020
;41 Suppl 2:ehaa946.3272.
19.
Wei
SC
,
Meijers
WC
,
Axelrod
ML
,
Anang
NAS
,
Screever
EM
,
Wescott
EC
, et al
.
A genetic mouse model recapitulates immune checkpoint inhibitor-associated myocarditis and supports a mechanism-based therapeutic intervention
.
Cancer Discov
2021
;
11
:
614
25
.
20.
Salem
JE
,
Allenbach
Y
,
Vozy
A
,
Brechot
N
,
Johnson
DB
,
Moslehi
JJ
, et al
.
Abatacept for severe immune checkpoint inhibitor-associated myocarditis
.
N Engl J Med
2019
;
380
:
2377
9
.
21.
Nguyen
LS
,
Bretagne
M
,
Arrondeau
J
,
Zahr
N
,
Ederhy
S
,
Abbar
B
, et al
.
Reversal of immune-checkpoint inhibitor fulminant myocarditis using personalized-dose-adjusted abatacept and ruxolitinib: proof of concept
.
J Immunother Cancer
2022
;
10
:
e004699
.
22.
Latek
R
,
Fleener
C
,
Lamian
V
,
Kulbokas
E
3rd
,
Davis
PM
,
Suchard
SJ
, et al
.
Assessment of belatacept-mediated costimulation blockade through evaluation of CD80/86-receptor saturation
.
Transplantation
2009
;
87
:
926
33
.
23.
Zhou
Z
,
Shen
J
,
Hong
Y
,
Kaul
S
,
Pfister
M
,
Roy
A
.
Time-varying belatacept exposure and its relationship to efficacy/safety responses in kidney-transplant recipients
.
Clin Pharmacol Ther
2012
;
92
:
251
7
.
24.
Zeiser
R
,
von Bubnoff
N
,
Butler
J
,
Mohty
M
,
Niederwieser
D
,
Or
R
, et al
.
Ruxolitinib for glucocorticoid-refractory acute graft-versus-host disease
.
N Engl J Med
2020
;
382
:
1800
10
.
25.
Lescoat
A
,
Lelong
M
,
Jeljeli
M
,
Piquet-Pellorce
C
,
Morzadec
C
,
Ballerie
A
, et al
.
Combined anti-fibrotic and anti-inflammatory properties of JAK-inhibitors on macrophages in vitro and in vivo: perspectives for scleroderma-associated interstitial lung disease
.
Biochem Pharmacol
2020
;
178
:
114103
.
26.
Deszo
EL
,
Brake
DK
,
Kelley
KW
,
Freund
GG
.
IL-4-dependent CD86 expression requires JAK/STAT6 activation and is negatively regulated by PKCδ
.
Cell Signalling
2004
;
16
:
271
80
.
27.
Kubo
S
,
Yamaoka
K
,
Kondo
M
,
Yamagata
K
,
Zhao
J
,
Iwata
S
, et al
.
The JAK inhibitor, tofacitinib, reduces the T cell stimulatory capacity of human monocyte-derived dendritic cells
.
Ann Rheum Dis
2014
;
73
:
2192
8
.
28.
Finke
D
,
Heckmann
MB
,
Salatzki
J
,
Riffel
J
,
Herpel
E
,
Heinzerling
LM
, et al
.
Comparative transcriptomics of immune checkpoint inhibitor myocarditis identifies guanylate binding protein 5 and 6 dysregulation
.
Cancers (Basel)
2021
;
13
:
2498
.
29.
Mahmood
SS
,
Fradley
MG
,
Cohen
JV
,
Nohria
A
,
Reynolds
KL
,
Heinzerling
LM
, et al
.
Myocarditis in patients treated with immune checkpoint inhibitors
.
J Am Coll Cardiol
2018
;
71
:
1755
64
.
30.
Nowatzke
J
,
Guedeney
P
,
Palaskas
N
,
Lehmann
L
,
Ederhy
S
,
Zhu
H
, et al
.
Coronary artery disease and revascularization associated with immune checkpoint blocker myocarditis: report from an international registry
.
Eur J Cancer
2022
;
177
:
197
205
.
31.
Teixeira
A
,
Cherin
P
,
Demoule
A
,
Levy-Soussan
M
,
Straus
C
,
Verin
E
, et al
.
Diaphragmatic dysfunction in patients with idiopathic inflammatory myopathies
.
Neuromuscul Disord
2005
;
15
:
32
9
.
32.
Chen
CY
,
Chiu
CF
,
Bai
LY
.
Treatment of pembrolizumab-induced cutaneous lesions with ruxolitinib
.
Eur J Cancer
2019
;
113
:
69
71
.
33.
Bishu
S
,
Melia
J
,
Sharfman
W
,
Lao
CD
,
Fecher
LA
,
Higgins
PDR
.
Efficacy and outcome of tofacitinib in immune checkpoint inhibitor colitis
.
Gastroenterology
2021
;
160
:
932
4
.
34.
Sasson
SC
,
Slevin
SM
,
Cheung
VTF
,
Nassiri
I
,
Olsson-Brown
A
,
Fryer
E
, et al
.
Interferon-gamma-producing CD8(+) tissue resident memory T cells are a targetable hallmark of immune checkpoint inhibitor-colitis
.
Gastroenterology
2021
;
161
:
1229
44
.
35.
Luoma
AM
,
Suo
S
,
Williams
HL
,
Sharova
T
,
Sullivan
K
,
Manos
M
, et al
.
Molecular pathways of colon inflammation induced by cancer immunotherapy
.
Cell
2020
;
182
:
655
71
.
36.
Yeleswaram
S
,
Smith
P
,
Burn
T
,
Covington
M
,
Juvekar
A
,
Li
Y
, et al
.
Inhibition of cytokine signaling by ruxolitinib and implications for COVID-19 treatment
.
Clin Immunol
2020
;
218
:
108517
.
37.
Heine
A
,
Held
SA
,
Daecke
SN
,
Wallner
S
,
Yajnanarayana
SP
,
Kurts
C
, et al
.
The JAK-inhibitor ruxolitinib impairs dendritic cell function in vitro and in vivo
.
Blood
2013
;
122
:
1192
202
.
38.
Albeituni
S
,
Verbist
KC
,
Tedrick
PE
,
Tillman
H
,
Picarsic
J
,
Bassett
R
, et al
.
Mechanisms of action of ruxolitinib in murine models of hemophagocytic lymphohistiocytosis
.
Blood
2019
;
134
:
147
59
.
39.
Hughes
PD
,
Polkey
MI
,
Moxham
J
,
Green
M
.
Long-term recovery of diaphragm strength in neuralgic amyotrophy
.
Eur Respir J
1999
;
13
:
379
84
.
40.
Rice
BL
,
Ashton
RW
,
Wang
XF
,
Shook
SJ
,
Mireles-Cabodevila
E
,
Aboussouan
LS
.
Modeling of lung function recovery in neuralgic amyotrophy with diaphragm impairment
.
Respir Care
2017
;
62
:
1269
76
.
41.
Ferguson
GT
,
Irvin
CG
,
Cherniack
RM
.
Effect of corticosteroids on respiratory muscle histopathology
.
Am Rev Respir Dis
1990
;
142
:
1047
52
.
42.
Weiner
P
,
Azgad
Y
,
Weiner
M
.
The effect of corticosteroids on inspiratory muscle performance in humans
.
Chest
1993
;
104
:
1788
91
.
43.
Laveneziana
P
,
Albuquerque
A
,
Aliverti
A
,
Babb
T
,
Barreiro
E
,
Dres
M
, et al
.
ERS statement on respiratory muscle testing at rest and during exercise
.
Eur Respir J
2019
;
53
:
1801214
.
44.
Brahmer
JR
,
Lacchetti
C
,
Schneider
BJ
,
Atkins
MB
,
Brassil
KJ
,
Caterino
JM
, et al
.
Management of immune-related adverse events in patients treated with immune checkpoint inhibitor therapy: American Society of Clinical Oncology clinical practice guideline
.
J Clin Oncol
2018
;
36
:
1714
68
.
45.
Palaskas
NL
,
Segura
A
,
Lelenwa
L
,
Siddiqui
BA
,
Subudhi
SK
,
Lopez-Mattei
J
, et al
.
Immune checkpoint inhibitor myocarditis: elucidating the spectrum of disease through endomyocardial biopsy
.
Eur J Heart Fail
2021
;
23
:
1725
35
.
46.
Ederhy
S
,
Devos
P
,
Pinna
B
,
Funck-Brentano
E
,
Abbar
B
,
Fenioux
C
, et al
.
(18)F-fluorodeoxyglucose positron emission tomography/computed tomography imaging for the diagnosis of immune checkpoint inhibitor-associated myocarditis
.
Arch Cardiovasc Dis
2022
;
115
:
114
6
.
47.
Pitoiset
F
,
Cassard
L
,
El Soufi
K
,
Boselli
L
,
Grivel
J
,
Roux
A
, et al
.
Deep phenotyping of immune cell populations by optimized and standardized flow cytometry analyses
.
Cytometry A
2018
;
93
:
793
802
.
48.
Eisenhauer
EA
,
Therasse
P
,
Bogaerts
J
,
Schwartz
LH
,
Sargent
D
,
Ford
R
, et al
.
New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1)
.
Eur J Cancer
2009
;
45
:
228
47
.

Supplementary data