Programmed Cell Death-1/Programmed Cell Death-Ligand 1 Pathway in High-Risk Lymphocytic Acute Myocarditis

Abstract

Background

The programmed cell death-1 (PD-1)/programmed cell death-ligand-1 (PD-L1) pathway is affected in immune checkpoint inhibitor (ICI)-associated acute myocarditis (AM), while this pathway in non-ICI-associated lymphocytic AM (LAM) is unexplored.

Objectives

This study aimed to assess cardiac levels of PD-L1/PD-1 in patients with high-risk LAM, compared with cardiac sarcoidosis (CS), non-inflammatory cardiomyopathy (non-Infl CMP), ICI-AM, and acute cellular rejection (ACR).

Methods

PD-L1 expression was measured with a visual semiquantitative scale from 0 (0%) to 5 (≥75% of cardiomyocytes), and PD-1 from 0 (0%) to 5 (≥75% of inflammatory cells) in 142 patients with cardiac histology. We compared patients with LAM (n = 39), CS (n = 26), non-InflCMP (n = 11). As an additional group, we included 8 patients with ICI-AM. Finally, the expression of PD-L1/PD-1 was also evaluated in heart transplant recipients with ACR of grade ≥2R (n = 28) or no ACR (n = 30).

Results

Patients with LAM had increased median levels of PD-L1 compared with CS or non-Infl-CMP: 2 (first to third quartile: 2-3) vs 1 (1-2) vs 0 (0-0), respectively; P < 0.0001. Likewise, PD-L1 expression was higher in those with ACR ≥2R vs no ACR. Conversely, PD-1 was blunted in LAM compared to ACR ≥2R. Patients with LAM had similar cardiac PD-L1 expression but reduced cardiac PD-1 expression compared with ICI-AM.

Conclusions

PD-L1 is overexpressed in cardiomyocytes of patients with LAM compared with non-Infl-CMP or focal non-lymphocytic myocarditis, such as CS, suggesting that cardiac PD-L1 expression is likely a response to T-cell-mediated injury. PD-1 expression is blunted in infiltrating inflammatory cells in LAM compared with ACR, suggesting a disruption of the PD-L1/PD-1 axis in LAM.

Central Illustration

Acute myocarditis (AM) is an inflammatory disease of the heart that primarily results from infections, immune activation, or exposure to certain drugs.1, 2, 3, 4, 5 Beyond basic characterization of immune infiltrates on histology and immunohistochemistry (ie, the presence of lymphocytes, eosinophils, giant cells, or granulomas), the specific mechanisms underlying AM remain poorly understood, with most evidence derived from murine models.³ One pertinent unresolved query concerns why approximately a quarter of patients experience a complicated clinical course, presenting with left ventricular systolic dysfunction, ventricular arrhythmias, or even cardiogenic shock akin to fulminant myocarditis (FM),⁶ while the majority experience a relatively uncomplicated course that spontaneously resolves. The progression of AM to severe forms characterized by acute heart failure due to large inflammatory infiltrates and cardiomyocyte injury and abrupt and persistent left ventricular systolic dysfunction suggests a disruption in mechanisms that regulate the immune response against cardiomyocytes.⁷ The new anticancer drugs, immune checkpoint inhibitors (ICI) targeting the host immune negative regulation receptors, such as programmed cell death-1 (PD-1) and programmed cell death-ligand 1 (PD-L1) (also termed CD274 or B7-H1), have emerged as a potential rare cause of lymphocytic AM (LAM) with a mortality risk up to 50%,¹ providing an unexpected human model of severe LAM.8, 9, 10, 11 Anti-PD-1 and PD-L1 therapies account for approximately 98% of ICI-associated myocarditis based on a registry including 748 patients.¹² PD-1 plays a crucial role in peripheral immunotolerance by interacting with PD-L1 to inhibit effector T cell activation.^2,4 For instance, PD-L1 is highly expressed physiologically by placental villous trophoblasts to suppress maternal rejection of the placenta by interacting with maternal T cells.¹³ The PD-1/PD-L1 pathway involved in T-cell inhibition is centrally affected in ICI-associated myocarditis.² ICI-related autoimmunity is favored by abrogating the PD-1-mediated negative feedback on T cells.² In human ICI-related myocarditis, the up-regulation of PD-L1 in the myocardium serves as a cardioprotective mechanism against dysregulated T-cell activation induced by ICIs.^11,14, 15, 16 This phenomenon is also observed in heart transplant recipients experiencing acute cellular rejection (ACR), supporting the hypothesis that the PD-1/PD-L1 pathway may play a role in several T-cell-mediated cardiac inflammatory processes. Since the role of cardiac PD-1/PD-L1 in human non-ICI-related LAM is unexplored, we speculated that the cardiac PD-1/PD-L1 pathway could be dysregulated in LAM. Preclinical evidence from murine models of T-cell-mediated myocarditis supports a role for the PD-1/PD-L1 pathway.^17,18 Thus, we aimed to explore and characterize the patterns of PD-1/PD-L1 expression in high-risk LAM, defined as patients hospitalized presenting with cardiac dysfunction, heart failure, or ventricular arrhythmias in whom an endomyocardial biopsy (EMB) was deemed necessary. We leveraged cardiac tissue obtained from patients with LAM and carried out a comparative analysis of PD-1/PD-L1 expression with cardiac sarcoidosis (CS) as a model of focal non-lymphocytic inflammatory cardiomyopathy, ICI-AM, and ACR as a model of alloantigen-triggered T-cell-mediated myocarditis, and non-inflammatory cardiac diseases.

Methods

Patient cohorts

We retrospectively identified patients with histologic/immunohistologic diagnoses of LAM, CS, non-inflammatory cardiomyopathy (non-Infl CMP), ICI-AM, ACR, or no ACR, reviewing diagnostic cases from 4 pathology services at the Niguarda Hospital in Milan, Italy; Tübingen University Hospital in Tübingen, Germany; Helsinki University Hospital in Helsinki, Finland; and the University of California, San Diego, in La Jolla, California. This study complied with the Declaration of Helsinki, and the Institutional Review Board in Milano, Italy (Milano Area C) approved the study under identifier 120 to 032016 and its extension under identifier 416 to 092018, approved on 7/8/2019. As a retrospective observational study of deidentified data, local ethics committees granted waivers for consent. Other details are provided in the Supplemental Methods.

The LAM cohort included 39 patients with the onset of acute cardiac symptoms and a histologic diagnosis of lymphocytic myocarditis, based on the presence of infiltrating inflammatory mononuclear cells and cardiomyocyte necrosis, satisfying the Dallas criteria for active myocarditis and the European Society of Cardiology criteria.^19,20 The 39 patients included in this analysis were high-risk cases in whom an EMB was deemed necessary to guide clinical management according to current guidelines.^5,20 Of them, 27 were classified as FM (patients requiring inotropes or temporary mechanical circulatory support for hemodynamic instability during the acute phase),¹ and 12 as non-FM (patients without hemodynamic instability). We studied the LAM subgroups (FM vs non-FM) to assess whether cardiac PD-L1/PD-1 expression differs by LAM severity. None of the patients with myocarditis received prior ICI therapy. In one case, the histology was derived from an explanted heart after heart transplantation, another from a postmortem examination, and the other 37 cases from EMB. The CS cohort included 26 patients with a histologic diagnosis of sarcoidosis based on the presence of noncaseating cardiac granulomas on EMB. The CS group served as a control group for focal non-lymphocytic inflammatory cardiomyopathy compared with LAM. The non-Infl CMP cohort served as an additional control group, comprising 11 patients with nonischemic cardiomyopathy without evidence of myocardial inflammation. As an additional group, we included 8 patients with ICI-AM to characterize the cardiac PD-L1/PD-1 expression. The heart transplantation cohort included 58 patients and was subgrouped into patients with or without ACR. The severity of ACR was based on the revised International Society for Heart and Lung Transplantation consensus classification for cardiac allograft rejection, ranging from 0R (no rejection, the group termed no ACR, including 30 patients in our study), 1R (mild rejection), to 2R (moderate rejection), and 3R (severe rejection),²¹ where generally ACR≥2R prompts immunosuppressive treatment. The ACR was used as an additional comparator for LAM, serving as a human model of T-cell-mediated myocarditis against alloantigens. Figure 1 presents a flowchart of the study design, illustrating the comparisons among groups. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline for cohort studies (the STROBE checklist is presented in the Supplemental Material).

Figure 1.

Flowchart of the Study

The flowchart illustrates the comparisons made among groups assessing cardiac PD-1 and PD-L1 expression. ACR = acute cellular rejection; AM = acute myocarditis; CS = cardiac sarcoidosis; ICI = immune checkpoint inhibitor; LAM = lymphocytic acute myocarditis; Infl-CMP = inflammatory cardiomyopathy; PD-1 = programmed cell death-1; PD-L1 = programmed cell death-ligand 1.

Cardiac samples and histology

Tissues from previously performed EMB, explanted heart, or autoptic heart, which were taken for diagnostic reasons, were fixed in formalin and embedded in paraffin. The heart tissue sections were stained with hematoxylin and eosin. Monoclonal antibodies against CD3 and CD68 were used for immunohistological detection of cardiac immune cells. In 29 (74.4%) patients with LAM, polymerase chain reaction was performed to detect viral genomes, yielding positive results in 11 (40.7%) patients, with parvovirus B19 identified most often (8/11; 72.7%), all at low copy numbers. All 11 patients with non-Infl CMP underwent viral genome search by polymerase chain reaction, and all were negative.

PD-1 and PD-L1 stains and assessment

All histologic specimens were analyzed at 3 centers (Niguarda Hospital in Milan, Italy; University Hospital Tübingen in Tübingen, Germany; and Helsinki University Hospital in Helsinki, Finland) by 5 local pathologists (D.P., E.B., M.I.M., T.M., and K.K.). Specimens from human tonsils were used as PD-1 and PD-L1 positive controls, as most germinal center-associated T cells express PD-1, and the epithelial cells of tonsillar crypts express PD-L1 (Figure 2A to 2E). Initial staining to assess PD-L1 and PD-1 levels on EMB was performed on slides, including human tonsils as a positive control (Figure 2F). Cardiac tissue from a patient in the non-Infl CMP group revealed no expression for PD-1 or PD-L1, serving as a negative control (Figure 2G to 2I). Using a shared positive control for PD-1 and PD-L1 across the 3 pathology services, we sought to reduce variability in results. The staining methods used in this study were validated for diagnostic use in non-small cell lung cancer and were referred to across the 3 centers as the “non-small cell lung cancer-method.” PD-1 and PD-L1 antibodies were used across the 3 centers. For PD-1 staining, the clone NAT105 produced by Roche in Tubingen and by Abcam in Milano and Helsinki was used. For PD-L1 staining, the clone SP142 was used in Tubingen (produced by Ventana, Roche) and Helsinki (produced by Abcam), while in Milano the clone 22C3 was used (produced by Agilent, Dako). The local choice of antibodies for PD-1 and PD-L1 staining was based on their routine clinical use.

Figure 2.

Positive and Negative Tissue Controls for Programmed Cell Death-1 and Programmed Cell Death Ligand 1 Staining

(A to C) Staining for PD-1 of a human tonsil at different magnifications (from 4× to 20×, indicating magnification of 40- to 200-fold, respectively) marks germinal center-associated T cells that express PD-1 (arrows). In contrast, few T cells in the interfollicular T-cell zone express PD-1. (D and E) Staining for PD-L1 of a human tonsil at different magnifications (from 4× to 10×, indicating 40- to 100-fold magnification) marks the epithelium of the tonsillar crypts highly expressing PDL-1 (arrows). In the tonsil germinal centers, macrophages show low-intensity PD-L1 positivity, whereas T cells are negative for PD-L1 (red rectangle in D). (F) Staining of PD-L1 in endomyocardial biopsies (EMB) and a human tonsil on the same slide as a positive control. (G to I) Staining of an EMB from a patient admitted with heart failure served as a negative control, showing no cardiac inflammatory infiltrates on hematoxylin and eosin (H&E) stain (G), and no expression for PD-1 (H) or PD-L1 (I) (10×, indicating 100-fold magnification). Abbreviations as in Figure 1.

The study pathologists estimated the percentages of PD-1- and PD-L1-stained cells using a semiquantitative visual scale. PD-L1 expression levels were graded based on the percentage of positive cardiomyocytes with the following grading score: 0 = none, 1 = dim signal/junctional signal, 2 = <25%, 3 = 25%-49%, 4 = 50%-74%, and 5 = ≥75%. PD-1 expression levels were graded based on the percentage of positive inflammatory mononuclear cells in the myocardium with the following grading score: 0 = none, 1 = 1-10%, 2 = 11%-25%, 3 = 26-50%, 4 = 51%-75%, and 5 = >75%. To establish and compare the staining, a few initial samples were stained in parallel in Milan and Tübingen (see a representative patient stained for PD-L1 in Supplemental Figure 1). All cases were read in Tübingen and Milano by the same pathologists, who shared the images. In the event of disagreement, a third pathologist in Milano reached an agreement. Cases stained in Finland (Helsinki) were read in Tübingen and Milano.

Statistical analysis

Data were presented as median and first to third quartiles (Q1-Q3). Categorical variables were summarized as absolute frequency (percentage). Associations between categorical variables were tested using Fisher's exact test. Group differences for continuous variables, including PD-1 and PD-L1 expression levels, were tested using the Mann-Whitney test when 2 groups were compared or the Kruskal-Wallis test with Dunn’s multiple comparisons test when 3 groups were compared. A P value <0.05 was considered significant. All statistical analyses were performed using GraphPad Prism (version 10.2.0, GraphPad Software Inc).

Results

Baseline characteristics of the study cohorts

Table 1 reports the summary baseline characteristics of LAM patients and comparators (CS and non-infl-CMP). LAM patients were younger (median age 41 years vs 52 and 58 years in CS and non-Infl-CMP, P < 0.01), while the sex distribution (53.8% vs 57.6% vs 27.2% in LAM, CS, and non-Infl-CMP, P = 0.23) and degree of left ventricular systolic dysfunction (median left ventricular ejection fraction [LVEF] of 40% vs 39% vs 40% in LAM, CS, and non-Infl-CMP, P = 0.43) were similar among all groups. Patients with non-infl-CMP underwent EMB significantly later than the other 2 groups, relative to symptom onset (median of 120 days vs 3 and 5 days in LAM and CS, P < 0.01). A fulminant presentation most often occurred in patients with LAM (69.2% vs 26.9% and 0% in CS and non-Infl-CMP, P < 0.01).⁷

Table 1.

The Main Characteristics of Patients With Lymphocytic Acute Myocarditis, Cardiac Sarcoidosis, and Non-Inflammatory Cardiomyopathy

	Lymphocytic Acute Myocarditis (n = 39)	Cardiac Sarcoidosis (n = 26)	Noninflammatory CMP (n = 11)	P Value
Demographics
Age, y, median (Q1-Q3)	41 (20-51)	52 (42-59)	58 (51-78)	<0.01
Female, n (%)	21 (53.8)	15 (57.6)	3 (27.2)	0.23
Days from onset of symptoms to histology, median (Q1-Q3)	3 (1-7)	5 (1-16)	120 (17-361)	<0.01
Echocardiography
LVEF on admission, %, median (Q1-Q3)	40 (25-45)	39 (30-50)	40 (27-52)	0.43
Fulminant presentation, n (%)	27 (69.2)	7 (26.9)	0 (0.0)	<0.01
Temporary mechanical circulatory support	19 (48.7)	4 (15.3)	0 (0.0)	<0.01

CMP = cardiomyopathy; LVEF = left ventricular ejection fraction.

LAM

Overall, 39 patients with LAM were included. Table 2 reports baseline characteristics and in-hospital course for patients with LAM stratified by fulminant presentation (FM vs non-FM). The median age was 41 (Q1-Q3: 20-51) years, and 21 (53.8%) were female. Of the 38 patients with available data, 37 (97.4%) had elevated cardiac troponin on admission, with a median LVEF of 40% (Q1-Q3: 25%-45%). Patients underwent EMB after a median of 3 (1-7) days since admission. Only 6/37 (16.2%) patients received immunosuppressive therapy before EMB, whereas in 22/37 (59.5%) patients, an immunosuppressive agent was initiated after a histologic diagnosis of lymphocytic myocarditis. Among 27 patients with fulminant presentation, 19 (70.3%) required temporary mechanical circulatory support, and 22 received any immunosuppressive treatment. Nine (23.1%) patients died in-hospital, 2 (5.1%) patients underwent urgent heart transplantation, and 28 (71.8%) patients were discharged alive without heart transplantation. Patients with FM had higher in-hospital cardiac mortality compared with non-FMs (33.3% vs 0.0%, respectively, P = 0.04).

Table 2.

Clinical Characteristics of Patients With Lymphocytic Acute Myocarditis (LAM) Stratified by Fulminant vs Non-Fulminant Presentation

	Available Data	Lymphocytic Acute Myocarditis (N = 39)	Non-Fulminant Myocarditis (n = 12)	Fulminant Myocarditis (n = 27)	P Value
Demographics	39 (100)
Age, y, median (Q1-Q3)		41 (20-51)	39 (29-45)	42 (18-58)	0.49
Female, n (%)		21 (53.8)	4 (33.3)	17 (62.9)	0.16
Main presenting symptoms, n (%)	39 (100)
Dyspnea		13 (33.3)	4 (33.3)	8 (29.6)	0.81
Chest pain		17 (43.5)	7 (58.3)	11 (40.7)	0.31
Abdominal pain		5 (12.8)	0 (0.0)	5 (18.5)	0.66
Palpitation		4 (10.2)	1 (8.3)	3 (11.1)	0.79
Any prodromal symptomsa, n (%)	35 (89.7)	26 (74.3)	7 (63.6)	19 (79.1)	0.41
Fever		20 (57.1)	7 (63.6)	13 (54.1)	0.56
Gastrointestinal symptoms		11 (31.4)	2 (18.1)	9 (37.5)	0.29
Respiratory symptoms		8 (22.8)	3 (27.2)	5 (20.8)	0.64
Arthralgia		6 (17.1)	0 (0.0)	6 (25.0)	0.30
Onset of symptoms, median (Q1-Q3)	38 (97.4)	3 (1-7)	4 (2-14)	3 (1-6)	0.22
Autoimmune disorders, n (%)	39 (100)	4 (10.3)	2 (16.6)	2 (7.4)	0.57
Elevated cardiac troponin on admission, n (%)	38 (97.4)	37 (97.4)	11 (100.0)	26 (96.3)	1.00
ECG at admission, n (%)	35 (89.7)				1.00
Normal		4 (11.4)	1 (10.0)	3 (12.0)	0.79
ST-segment elevation		12 (34.3)	3 (30.0)	9 (36.0)	0.60
Other ST-T abnormalities		19 (54.3)	6 (60.0)	13 (52.0)	0.91
QRS >120 msec	36 (92.3)	11 (30.5)	1 (9.1)	10 (38.4)	0.10
Echocardiography at admission	37 (94.8)
LVEF, %, median (Q1-Q3)		40 (25-45)	40 (35-43)	35 (20-48)	0.42
Left ventricular-end diastolic diameter at admission, mm, median (Q1-Q3)		46 (41-51)	47 (44-51)	45 (41-50)	0.50
Coronary angiogram, n (%)	36 (92.3)	26 (72.2)	7 (63.6)	19 (76.0)	0.45
Endomyocardial biopsy performed, n (%)	39 (100)	37 (94.9)	12 (100.0)	25 (92.6)	1.00
Right ventricle, n (%)		32 (86.5)	10 (83.3)	22 (88.0)	1.00
Days from admission, median (Q1-Q3)	38 (97.4)	3 (1-7)	2 (0-9)	3 (1-7)	0.73
Immunosuppressive therapy, n (%)	37 (94.8)	28 (75.6)	6 (54.5)	22 (84.6)	0.09
Before endomyocardial biopsy, n (%)		6 (16.2)	0 (0.0)	6 (23.1)	0.30
Steroids		26 (70.2)	6 (54.5)	21 (80.7)	0.08
Intravenous Ig		10 (27.0)	0 (0.0)	10 (38.4)	0.07
Other		4 (10.8)	0 (0.0)	4 (15.3)	0.576
Mechanical circulatory support use, n (%)	39 (100)	19 (48.7)	0 (0.0)	19 (70.3)	<0.001
Cardiac magnetic resonance imaging during hospitalization, n (%)	38 (97.4)	27 (71.0)	10 (90.9)	17 (62.9)	0.12
STIR positive		22 (57.8)	10 (90.9)	12 (70.5)	0.65
Late gadolinium enhancement		24 (63.1)	10 (90.9)	14 (82.3)	0.93
In-hospital outcome	39 (100)
Cardiac death, n (%)		9 (23.1)	0 (0.0)	9 (33.3)	0.04
Heart transplantation, n (%)		2 (5.1)	1 (8.3)	1 (3.7)	0.55
LVEF at discharge among survivors, %, median (Q1-Q3)	28 (100)	54 (50-63)	53 (48-65)	54 (52-62)	0.58

ECG = electrocardiogram; STIR: short tau inversion recovery; other abbreviations as in Table 1.

^aA patient can report more than 1 prodromal symptom.

CS

The clinical characteristics of 26 patients with biopsy-proven CS are detailed in Table 3. The median age was 52 (Q1-Q3: 42-59) years, and 15 (57.6%) were female. The median LVEF was 39% (Q1-Q3: 30%- 50%), and dyspnea (57.6%) was the most common symptom, with fulminant presentation in 7 (26.9%). Patients underwent EMB after a median of 5 (1-16) days since admission. No patient received immunosuppressive treatment before EMB, whereas in 96.1% of patients, immunosuppressive agents were started following CS diagnosis. On admission or during hospitalization, high-degree atrioventricular block occurred in 13 (50.0%) patients, and ventricular arrhythmias in 14 patients (53.8%). One (3.8%) patient died in the hospital, 1 (3.8%) patient underwent heart transplantation, and 24 (92.3%) were discharged alive without heart transplantation.

Table 3.

Clinical Characteristics of Patients Admitted With Histologically Proven Active Cardiac Sarcoidosis

	Available Data	Cardiac Sarcoidosis (n = 26)
Demographics
Age, y, median (Q1-Q3)	26 (100)	52 (42-59)
Female, n (%)	26 (100)	15 (57.6)
Main presenting symptoms, n (%)	26 (100)
Dyspnea		15 (57.6)
Chest pain		2 (7.6)
Palpitation		3 (11.5)
Syncope		4 (15.3)
Other		2 (7.6)
EMB performed, n (%)	26 (100)	26 (100)
Right ventricle, n (%)		26 (100)
Days from onset of symptoms, median (Q1-Q3)		5 (1-16)
Diagnosis of sarcoidosis before EMB, n (%)	26 (100)	0 (0.0)
Elevated cardiac troponin on admission, n (%)	26 (100)	26 (100)
ECG at admission, n (%)	26 (100)
Normal		1 (3.8)
ST-segment elevation		13 (50.0)
Other ST-T abnormalities		3 (11.5)
QRS >120 msec		19 (73.0)
Arrhythmias during the acute phase, n (%)	26 (100)
High-grade atrioventricular block		13 (50.0)
VT/VF		14 (53.8)
Echocardiography
LVEF pre-EMB, %, median (Q1-Q3)	26 (100)	39 (30-50)
LVEF at hospital discharge, %, median (Q1-Q3)	26 (100)	40 (35-50)
Immunosuppressive therapy, n (%)	26 (100)
Before EMB		0 (0.0)
After EMB		25 (96.1)
Steroids		25 (96.1)
Cyclosporine		4 (15.3)
Azathioprine		4 (15.3)
Infliximab		1 (3.8)
Methotrexate		1 (3.8)
Fulminant presentation, n (%)	26 (100)	7 (26.9)
MCS use, n (%)		4 (15.4)
CMRI during hospitalization, n (%)	26 (100)
STIR positive		26 (100)
LGE		26 (100)
PET-CT positive during hospitalization, n (%)	26 (100)	26 (100)
In-hospital outcome	26 (100)
Cardiac death, n (%)		1 (3.8)
Heart transplantation, n (%)		1 (3.8)

CMRI = cardiac magnetic resonance imaging; EMB = endomyocardial biopsy; LGE = late gadolinium enhancement; MCS = mechanical circulatory support; PET-CT = positron emission tomography-computed tomography; VT/VF = ventricular tachycardia/ventricular fibrillation; other abbreviations as in Table 1.

Non-infl CMP cohort

The clinical characteristics of 11 patients undergoing EMB for heart failure diagnostic work-up are detailed in Table 1 and Supplemental Table 2. The median age was 58 years (Q1-Q3: 51-78), and 3 (27.2%) were female, with a median LVEF of 40% (Q1-Q3: 27%-52%). Histology revealed no cardiac inflammatory infiltrates. Immunohistochemistry for CD3 and CD68 did not show increased numbers of T cells and macrophages in the heart, respectively. A final diagnosis of hypertensive cardiomyopathy was reached in all these patients. The main reasons for performing an EMB included new evidence of hypokinetic dilated left ventricle in 3 (27.3%), hypertrophic left ventricle in 3 (27.3%), suspected inflammatory cardiomyopathy in 2 (18.2%), hypokinetic non-dilated left ventricle in 1 (9.1%), suspected amyloidosis in 1 (9.1%), and new onset of atrial fibrillation after a recent systemic infective episode in 1 (9.1%). Patients in this group had a longer duration of cardiac symptoms and were referred to EMB after a median of 120 (17-361) days since the symptoms’ onset.

ICI-AM cohort

Eight patients with ICI-AM with available EMB samples were included; details of this population are reported in Supplemental Table 3. The median age was 66 (Q1-Q3: 62-72) years, and 5 (62.5%) were female, and the most frequently used ICI was pembrolizumab (6/8; 75.0%). They had a median LVEF of 43% (Q1-Q3: 20%-54%). Patients underwent EMB after a median of 11 (4-35) days since onset of symptoms. Of them, 5 (62.5%) patients received immunosuppressive therapy before EMB, and in 7/8 (87.5%) patients, an immunosuppressive agent was started or continued following histologic diagnosis of lymphocytic myocarditis. Of them, 5 (62.5%) had a fulminant presentation, and 1 (12.5%) required a temporary mechanical circulatory support. Two (25.0) patients died in hospital.

Heart transplantation cohort

Overall, 58 adult heart transplant recipients with available EMB samples were included; 28 (48.3%) had ACR ≥2R and 30 (51.7%) had no ACR. Clinical characteristics are reported in Table 4. The median age was 44 (Q1-Q3: 19-56) years, and 17 (29.3%) were female. Forty-one patients (70.7%) underwent EMB for clinical surveillance, while 17 (29.3%) had a clinical suspicion of rejection. Further details are presented in Supplemental Appendix.

Table 4.

Characteristics of Heart Transplant Recipients Stratified by Those With Acute Cellular Rejection (ACR) ≥2R vs No ACR

	Available Data (n = 58, 100.0%)	Overall (n = 58, 100.0%)	No ACR (n = 30, 51.7%)	ACR ≥2R (n = 28, 48.3%)	P Value
Age, y, median (Q1-Q3)	58 (100)	44 (19-56)	46 (28-58)	25 (17-51)	0.02
Female, n (%)	58 (100)	17 (29.3)	11 (36.6)	6 (21.4)	0.26
Days post-transplant, d, median (Q1-Q3)	56 (96.5)	373 (165-1,642)	412 (215-3,501)	312 (80-1,116)	0.06
Reason for performing EMB, n (%)	58 (100)				0.11
HF symptoms with reduced LVEF		7 (12.1)	2 (6.6)	5 (17.8)	0.19
HF symptoms with preserved LVEF		2 (3.4)	0 (0.0)	2 (7.1)	0.51
Ventricular arrhythmias		2 (3.4)	0 (0.0)	2 (7.1)	0.51
Post-transplant surveillance		41 (70.6)	25 (83.3)	16 (57.1)	0.03
Other		6 (10.3)	3 (10.0)	3 (10.7)	0.93
Raised troponin, n (%)	43 (74.1)	12 (27.9)	1 (5.0)	11 (47.8)	<0.01
Induction therapy, n (%)	47 (81.0)				0.09
None		10 (21.2)	6 (25.0)	4 (17.4)	0.56
Anti-thymocyte globulin		20 (42.5)	11 (45.8)	9 (39.1)	0.72
Basiliximab		8 (17.0)	2 (8.3)	6 (26.0)	0.10
Daclizumab		4 (8.5)	4 (16.6)	0 (0.0)	0.19
Other		5 (10.6)	1 (4.1)	4 (17.4)	0.14
Maintenance immunosuppressive regimen, n (%)	52 (89.6)		27 (100)	25 (100)	0.13
Cyclosporine		8 (15.3)	4 (14.8)	4 (16.0)	0.92
Tacrolimus		38 (73.1)	22 (81.4)	16 (64.0)	0.19
Mycophenolate mofetil/mycophenolic acid		29 (55.7)	14 (51.8)	15 (60.0)	0.60
Everolimus/sirolimus		26 (50.0)	13 (48.1)	13 (52.0)	0.81
Azathioprine		2 (3.8)	1 (3.7)	1 (4.0)	0.96
Prednisone		28 (53.8)	17 (62.9)	11 (44.0)	0.19
LVEF pre-EMB, %, median (Q1-Q3)	53 (91.4)	60 (49-64)	60 (58-63)	60 (35-65)	0.19
Previous rejection, d, median (Q1-Q3)	50 (86.2)	0 (0-2)	0 (0-0)	1 (0-2)	<0.01
Concurrent CMV infection, n (%)	51 (87.9)	9 (17.6)	4 (15.3)	5 (20.0)	0.47
Rejection treatment, n (%)				25 (89.2)	-
Pulsed steroids			-	16 (57.1)
Pulsed steroids + other			-	4 (14.3)
Immunosuppressive therapy modifications			-	5 (17.8)

CMV = cytomegalovirus; HF = heart failure; other abbreviations as in Tables 1 and 2.

Cardiac PD-L1 expression in LAM compared to CS and non-infl-CMP

The CS and non-Infl-CMP groups serve as controls to assess whether PD-L1 cardiomyocyte expression is mechanism-specific in high-risk LAM, rather than a generic response to inflammation, as in CS, or to low LVEF, as in non-Infl-CMP (Figure 1).

Patients with LAM had significantly higher PD-L1 expression on cardiomyocytes than patients with active CS or non-Infl-CMP. The PD-L1 level of expression in cardiomyocytes in LAM, based on a semiquantitative grading score, was 2 (Q1-Q3: 2-3); in CS, it was 1 (Q1-Q3: 1-2), and in non-Infl-CMP, it was 0 (Q1-Q3: 0-0; overall P < 0.001) (Figure 3A to 3G). Among LAM patients, those presenting with FM had higher levels of PD-L1 expression compared with those without fulminant presentation: in patients with FM, the grade of PD-L1 expression was 3 (Q1-Q3: 2-3) vs 2 (Q1-Q3: 1-2) in those with non-FM (P = 0.01) (Figure 3H to 3L). No difference in PD-L1 expression in cardiomyocytes was observed between those with evidence of viral genomes (n = 11) vs those with negative polymerase chain reaction results (n = 18), with a median score of 2 (Q1-Q3: 1-3) vs 2 (Q1-Q3: 2-3), respectively (P = 0.81, not shown).

Figure 3.

Cardiac PD-L1 Expression in Patients With Lymphocytic Acute Myocarditis Compared With Patients With Cardiac Sarcoidosis and Non-Inflammatory Cardiomyopathy, and in Patients With Fulminant Myocarditis vs Non-Fulminant Myocarditis

(A to F) Endomyocardial biopsies (EMB) from 3 representative patients with LAM, CS, and non-Infl-CMP stained with hematoxylin and eosin (H&E) and PD-L1. (A and B) In the patient with LAM (case AM24, a 30-year-old man), the H&E stain shows cardiac necrosis and inflammatory infiltrates, consistent with active myocarditis; the PD-L1 stain reveals high PD-L1 expression on cardiomyocytes (grade 4, indicating 50% to 74% of cardiomyocytes were positive for PD-L1). Note that in this case, the positive PD-L1 staining of the plasma membrane is also in non-gap-junction regions of cardiomyocytes, while in low-intensity cases, PD-L1 is mainly or exclusively seen in gap-junction regions. (C and D) In the patient with CS (case CS25, 53-year-old man), H&E shows myocardium without (w/o) granulomas and inflammatory infiltrates associated with dim-junctional PD-L1 expression on cardiomyocytes (grade 1 of the visual semiquantitative scale). (E and F) In the patient with non-Infl-CMP (case NM1, 51-year-old man), H&E shows an area of myocardium without inflammatory infiltrates, with no PD-L1 expression on cardiomyocytes (grade 0 on the visual semiquantitative scale). (G) PD-L1 expression level on cardiomyocytes in patients with LAM, CS, and non-Infl-CMP. Kruskal-Wallis test with Dunn’s multiple comparisons test was used. Arrows indicate PD-L1 expression in the 3 representative patients. (H) Among patients with LAM, PD-L1 expression on cardiomyocytes was significantly higher in patients with FM than in NFM. Mann-Whitney test was used. (I-L) EMB from 2 representative patients with FM (I and J, case AM27, 43-year-old man) and NFM (K and L, case AM3, 44-year-old man) stained with H&E and showing PD-L1. FM = fulminant myocarditis; NFM = non-fulminant myocarditis; other abbreviations as in Figure 1.

In patients with CS, PD-L1 expression was mainly localized within the non-caseating granuloma, likely expressed on macrophages/histiocytes (Figure 4, Supplemental Figure 2). In contrast, the myocardium unaffected by granulomas had a generally low PD-L1 expression in cardiomyocytes, with a median semiquantitative grading score of 1. To describe the PD-L1 expression in the non-caseating granuloma, we used an arbitrary visual semiquantitative grading scale of mild (value of 1), moderate (value of 2, as shown in the case CS25 reported in Figure 4), or intense (value of 3, as shown in the case CS19 reported in Supplemental Figure 2). Among the 26 patients with CS, the median PD-L1 expression in granuloma was 3 (Q1-Q3: 2-3). Unlike in patients with LAM, the severity of the clinical presentation in patients with CS was not associated with PD-L1 expression on cardiomyocytes or at the granuloma level. If we compared patients with CS and hemodynamic instability requiring inotropes or temporary mechanical circulatory support (n = 7) vs patients with CS without hemodynamic instability (n = 19), the median PD-L1 expression in cardiomyocytes was 1 (Q1-Q3: 1-2) vs 1 (Q1-Q3: 0-2), respectively (P = 0.44). Considering the median PD-L1 expression in granuloma, it was 3 (Q1-Q3: 1-3) in patients with CS and hemodynamic instability vs 2 (Q1-Q3: 2-3) in patients with stable hemodynamics (P > 0.99).

Figure 4.

Expression of PD-1 and PD-L1 in a Representative Patient With Cardiac Sarcoidosis

(A) H&E stain of an endomyocardial biopsy (EMB) showing an area of normal myocardium and a non-caseating granuloma (in the inset, the granuloma at higher magnification reveals a giant cell). The sample was from a 53-year-old man presenting with dyspnea, high-degree atrioventricular block, and an LVEF of 35% (case CS25). (B) At the top, PD-L1 stain shows dim/junctional PD-L1 expression in cardiomyocytes unaffected by granuloma (grade 1). In contrast, in the granuloma (inset), PD-L1 expression is moderate in inflammatory cells, likely macrophages/histiocytes, based on a visual semiquantitative scale (grade 2 expression on inflammatory cells). On the bottom, the PD-1 stain shows no expression of PD-1 in the myocardium unaffected by the granuloma. In contrast, in the granuloma (inset), there is grade 2 PD-1 expression on inflammatory cells, likely T cells (grade 2 on the visual semiquantitative scale represents 11% to 25% expression in inflammatory cells). H&E = hematoxylin and eosin; LVEF = left ventricular ejection fraction; other abbreviations as in Figure 1.

Cardiac PD-L1 expression in heart transplant recipients

The ACR ≥2R group vs no ACR group served as confirmatory controls to assess whether PD-L1 cardiomyocyte expression is a T-cell-mediated injury mechanism (Figure 1).

PD-L1 expression in cardiomyocytes was compared between patients who received heart transplantation with ACR ≥2R and those with no ACR, as illustrated in Figure 5A-5F. In heart transplant recipients with ACR ≥2, the PD-L1 expression level on cardiomyocytes was significantly higher than in recipients with no ACR, with a median PD-L1 expression of 1 (Q1-Q3: 1-2) compared to 0 (Q1-Q3: 0-0), respectively (P < 0.001) (Figure 5G). A significant difference was observed in PD-L1 expression in cardiomyocytes between patients with ACR 3R (n = 4) and ACR 2R (n = 24), with a median of 4 (Q1-Q3: 2-5) vs 1 (Q1-Q3: 1-2), respectively (P < 0.01, not shown).

Figure 5.

Cardiac PD-L1 Expression in Heart Transplant Recipients With Acute Cellular Rejection or Without Acute Cellular Rejection

(A to F) Endomyocardial biopsy (EMB) from 2 representative patients with ACR of grade ≥2R and 1 patient with no ACR stained with H&E and showing PD-L1. At the top, the H&E stain shows different grades of ACR: 3R (case HT15) in A and 2R (case HT55) in B, with no ACR (case HT56) in C. On the bottom, high PD-L1 expression on cardiomyocytes in the patient with ACR of grade 3 (D), grade 2 PD-L1 expression in the patient with ACR of grade 2 (E), and no PD-L1 expression on cardiomyocytes of the patient without ACR (F). (G) In heart transplant recipients with ACR ≥2R, the PD-L1 expression level on cardiomyocytes was significantly higher than in recipients with no ACR. Abbreviations as in Figure 1.

Cardiac PD-L1 and PD-1 expression in inflammatory mononuclear cells in patients with LAM vs ACR ≥2R

The LAM vs ACR ≥2 R group served as controls to assess whether a disruption of the PD-1/PD-L1 pathway can be detected in patients with LAM, potentially due to a disrupted-brake signature due to the blunt of PD-1 expression in LAM (Figure 1).

As ACR can be considered a model of T-cell-mediated myocarditis against alloantigens in the transplanted heart, we compared PD-L1 expression in cardiomyocytes in LAM and ACR. The expression of PD-L1 in cardiomyocytes from patients with LAM was significantly higher than in heart transplant recipients with ACR≥2R, with a median grading score of 2 (Q1-Q3: 2-3) vs 1 (Q1-Q3: 1-2), respectively (P < 0.01) (Figure 6A to 6E). Conversely, while PD-1 was expressed by inflammatory mononuclear cells of heart transplant recipients with ACR ≥2R with a median grading score of 2 (Q1-Q3: 1-2), the PD-1 expression in patients with LAM was significantly lower with a median value of 1 (Q1-Q3: 0-1; P < 0.001) (Figure 6F to 6J). In patients with LAM, PD-1 expression in cardiac inflammatory cells was significantly lower than in patients with FM, with a median grading score of 0 (Q1-Q3: 0-1) compared with 1 (Q1-Q3: 1-2) in non-FM (P = 0.01) (Supplemental Figure 3). The staining of a representative patient with FM is shown in Supplemental Figure 4 and a representative patient with ACR ≥2R is shown in Supplemental Figure 5.

Figure 6.

Cardiac PD-1 and PD-L1 Expression in Patients With Lymphocytic Acute Myocarditis Compared With Heart Transplant Recipients With Acute Cellular Rejection

(A and B) In a representative patient with LAM (case AM15, a 56-year-old woman), staining for PD-1 shows no PD-1 expression on inflammatory cells, whereas staining for PD-L1 shows PD-L1 expression in cardiomyocytes. (C and D) In a representative patient with ACR of grade 3R (ACR ≥2R; case HT15, 48-year-old woman), staining for PD-1 shows grade 2 PD-1 expression on inflammatory cells, and staining for PD-L1 shows PD-L1 expression in cardiomyocytes. (E) In patients with LAM (n = 39), PD-L1 expression on cardiomyocytes was significantly higher than in heart transplant recipients with ACR ≥2R. Arrows indicate PD-L1 expression in the 2 representative patients (B and D). (F) In patients with LAM, the PD-1 expression level on inflammatory cells infiltrating the myocardium was significantly lower than in ACR ≥2R. Arrows indicate the PD-1 expression of the 4 representative patients (A, C, H, and J). Mann-Whitney test was used. (G and H) In a representative patient with LAM (case AM11, 51-year-old woman), H&E stain reveals myocardial necrosis and diffuse inflammatory infiltrates, in the absence of PD-1 expression on inflammatory cells. (I and J) In a representative patient with ACR ≥2R (case HT53, 57-year-old man), H&E stain reveals ACR grade 2R, and PD-1 stain shows grade 2 expression of PD-1 in inflammatory cells. Abbreviations as in Figures 1 and 4.

Cardiac PD-L1 and PD-1 expression in patients with LAM vs ICI-AM

The expression of cardiac PD-L1 and PD-1 in 2 patients with ICI-AM treated with pembrolizumab is shown in Figure 7A to 7D and Supplemental Figure 6. Notably, PD-L1 expression in cardiomyocytes was similarly elevated in both LAM and ICI-AM, with median grading scores of 2 (Q1-Q3: 2-3) vs 3 (Q1-Q3: 1-3), respectively (P = 0.87) (Figure 7E). PD-1 was more expressed by inflammatory mononuclear cells of patients with ICI-AM despite being treated with antibodies to target this receptor, with a median grading score of 3 (Q1-Q3: 2-3) compared with patients with LAM who had a median value of 1 (Q1-Q3: 0-1; P < 0.01) (Figure 7F).

Figure 7.

Cardiac PD-1 and PD-L1 Expression in Patients With Immune Checkpoint Inhibitor Acute Myocarditis vs Lymphocytic Acute Myocarditis

(A to D) Two representative patients with ICI-AM, stained for PD-1 and PD-L1 showing grade 3 PD-L1 expression in cardiomyocytes and grade 3 PD-1 expression on inflammatory mononuclear cells. (E) In patients with LAM (n = 39), PD-L1 expression on cardiomyocytes was similar than in patients with ICI-AM. Arrows indicate PD-L1 expression in the 2 representative patients (A and C). (F) In patients with LAM, the PD-1 expression level on inflammatory mononuclear cells infiltrating the myocardium was significantly lower than in ICI-AM. Arrows indicate the PD-1 expression in the 2 representative patients (B and D). Mann-Whitney test was used. Abbreviations as in Figure 1.

Cardiac PD-1 expression in patients with CS and non-infl-CMP

As expected, in patients with non-Infl-CMP, PD-1 was not expressed in the absence of immune cell infiltrates. In patients with CS, inflammatory mononuclear cells were localized within non-caseating granuloma, and the median PD-1 expression grading score was 2 (Q1-Q3: 1-3) (Figure 4, Supplemental Figure 2).

Discussion

Cardiomyocytes from patients with LAM were found to express PD-L1, whereas those from patients with heart failure of non-inflammatory etiology (non-Infl-CMP patients) did not. PD-L1 levels are associated with higher inflammatory infiltration, as in fulminant forms. Furthermore, PD-L1 levels in cardiomyocytes from patients with LAM were significantly higher than in those from patients with other cardiac inflammatory forms, such as active CS or heart transplantation recipients, with and without ACR. PD-L1 levels in cardiomyocytes from patients with LAM and ICI-AM were similarly elevated.

Finally, a distinct PD-1/PD-L1 signature was observed in patients with LAM compared to ICI-AM and ACR. The expression of PD-1 on mononuclear inflammatory cells was blunted in patients with LAM compared with patients with ICI-AM and heart transplantation recipients with ACR (Central Illustration).

Central Illustration.

Cardiac PD-1 and PD-L1 Expression in High-Risk Lymphocytic Acute Myocarditis Compared With Other Myocardial Inflammatory Forms

PD-L1 is overexpressed in cardiomyocytes of patients with lymphocytic acute myocarditis (LAM). Other forms of myocardial inflammation, such as acute cellular rejection and immune checkpoint inhibitor (ICI)-associated acute myocarditis, driven by T-cell-mediated injury, also express PD-L1 on cardiomyocytes, which is generally not expressed by cardiomyocytes. Cardiac PD-1 expression in inflammatory mononuclear cells is blunted in LAM compared with ACR and ICI-AM, suggesting a potential disruption of the PD-L1/PD-1 axis in LAM that can favor T-cell infiltration in severe forms of myocarditis. Abbreviations as in Figure 1.

These findings, in the context of prior literature, could suggest that PD-L1 expression represents a cardioprotective mechanism to counter myocardial T-cell-mediated injury, even if the cross-sectional design of the study can only demonstrate the association and not causality. The association between PD-L1 levels and the severity of T-cell infiltration in the myocardium has been previously reported in the context of ACR- and ICI-associated myocarditis.^15,22 Potential clinical applications of our findings are related to diagnosis and monitoring of patients with LAM. In fact, a new tracer, ^99mTc-labeled anti-PD-L1 single-domain antibody, was used to noninvasively quantify PD-L1 cardiac expression with single-photon emission tomography/computed tomography.²³ This anti-PD-L1 tracer could be used to investigate cardiac PD-L1 expression, which we found to be specifically overexpressed in cardiomyocytes in patients with LAM, ICI-AM, and ACR, as well as in granulomas in patients with CS. Furthermore, soluble PD-L1 in serum has been proposed as a noninvasive marker of disease severity and treatment response in patients receiving ICI.²⁴ ACR can be seen as a human model of T-cell-mediated myocarditis against alloantigens expressed by the transplanted heart. PD-L1 expression is increased in ACR and LAM compared with no ACR, with even higher levels in LAM. On the other hand, PD-1 expression on infiltrating inflammatory cells in the myocardium of patients with LAM is reduced compared with that in heart transplant recipients with ACR. Compared with ICI-AM, patients with LAM appear to have lower PD-1 expression on inflammatory mononuclear cells. It could be that in ICI-AM, PD-1 receptors are expressed but functionally inactivated by anti-PD-1 antibodies (all 8 patients in our ICI-AM cohort received an anti-PD-1 antibody). The relatively dampened PD-1 expression we observed in LAM compared with ACR and ICI-AM may have distinct pathogenetic significance, representing a primary disruptive mechanism of LAM. Further studies should assess whether PD-1 dim expression is a phenotype acquired by T cells during AM, favoring the transition to the fulminant form, or whether some individuals are temporarily at risk of developing autoimmunity, including LAM, due to a transient reduction in PD-1 expression on cardiac tissue-infiltrating T cells. Disruption of the PD-1 gene causes autoimmune dilated cardiomyopathy, producing high-titer autoantibodies against troponin I in a mouse model.^25,26 Similarly, in a model of experimental autoimmune myocarditis, PD-1 knockout mice develop more severe disease than wild-type mice, and PD-1-deficient T lymphocytes demonstrate enhanced cardiac cytotoxic activity.¹⁷ In addition, a specific mouse strain (Murphy Roths large) with PD-1 deficiency (Pdcd^−/−) has recently been used as a spontaneous model of myocarditis to investigate the immune cell subsets relevant to ICI-associated myocarditis.²⁷ Furthermore, in humans, LAM occurrence has been repeatedly reported among patients with cancer treated with anti-PD-1 antibodies such as nivolumab and pembrolizumab.^2,9, 10, 11 Case reports of patients with LAM secondary to nivolumab or pembrolizumab with available cardiac histology showed consistent overexpression of PD-L1 in cardiomyocytes in areas with infiltrating lymphocytes,¹⁵ whereas a different range of PD-1 expression was observed in infiltrating T cells.9, 10, 11^,28, 29, 30 The degree of PD-L1 expression in cardiomyocytes has been shown to be induced by interferon-α, -β, and -γ in inflamed tissues in a mouse model.³¹ Furthermore, in a murine model of myocarditis caused by coxsackievirus B3, PD-L1 was overexpressed on cardiomyocytes, and an anti-PD-1 blocking monoclonal antibody increased myocardial inflammation, whereas an anti-PD-1-stimulating antibody decreased it.³² Increased cardiomyocytic PD-L1 expression has recently been reported in a case series of 16 patients with fulminant LAM, confirming that human cardiomyocytes express PD-L1 in response to T-cell infiltration.³³

Of potential clinical relevance, high-risk LAM mediated by the disruption of the PD-1/PD-L1 pathway could be treated with prompt administration of high-dose immunosuppression, including steroids or drugs specifically directed against T cells.³⁴ In line with this hypothesis, an ongoing phase 3 trial is assessing the efficacy and safety of pulsed high-dose intravenous methylprednisolone in patients with high-risk LAM (MYTHS (MYocarditis THerapy with Steroids) trial; NCT05150704).³⁵ In addition, targeted therapies could be considered, as those that are currently used in patients with ICI-myocarditis.³⁶ A phase 3 trial (ATRIUM trial; NCT05335928) is ongoing, investigating the use of abatacept, a cytotoxic T-lymphocyte antigen 4 (CTLA-4) immunoglobulin fusion protein that leads to global T-cell anergy by reversing ICI-activated pathways, in patients with ICI-associated myocarditis.⁴ Another phase 2, dose-finding trial of abatacept (ACHLYS trial; NCT05195645) has completed enrollment of patients with ICI-myocarditis.³⁷ In the event of positive trials with abatacept, our finding could provide a rationale for assessing its use in high-risk non-ICI-related LAM.

Another interesting finding is that there is no difference in PD-L1 expression on cardiomyocytes between patients with myocarditis with evidence of viral genome in the myocardium and those without detection of viral genomes. The low-copy parvovirus B19 (detected in 72.7% of cases) found in most cases with viral genome in the myocardium can be either a bystander or a virus that triggers an immune response against the myocardium.^7,38 This reinforces the notion that PD-L1 cardiomyocyte expression can be a final common response to T-cell-mediated injury in both non-infectious myocarditis and virus-triggered immune-mediated myocarditis.

Although our study did not focus on patients with ACR and CS, we showed that the expression of cardiac PD-L1 is not associated with a nonspecific injury associated with left ventricular systolic dysfunction. In fact, there was no expression of PD-L1 in patients with non-Infl-CMP with a similar LVEF on admission compared with those with LAM. Additionally, in focal inflammatory cardiomyopathy such as CS, where inflammation is primarily non-T-cell-mediated, PD-L1 was not expressed on cardiomyocytes. This suggests that PD-L1 expression is not associated with an indiscriminate response by cardiomyocytes to an inflammatory environment. In addition, the PD-L1/PD-1 pathway is activated in the noncaseating granuloma of patients with CS, with high PD-L1 expression, likely by macrophages within the granuloma. The pathogenic role of this pathway in granuloma formation should be further assessed.

Finally, the expression of PD-L1 in cardiomyocytes from patients with severe ACR was confirmed,²² while the diagnostic and prognostic roles of PD-L1 should be assessed. Our findings suggest that PD-L1 is more highly expressed in more severe forms of ACR, with significantly higher levels in ACR 3R vs ACR 2 R patients.

Study limitations

Beyond the lack of mechanistic insights into the regulation of PD-1 and PD-L1 markers due to the cross-sectional design of the study, other relevant limitations concern patient selection, staining, and assessment of PD-1/PD-L1 expression in inflammatory cells and cardiomyocytes, respectively. Nonconsecutive patients identified across 4 centers were included retrospectively based on specific diagnoses (ie, LAM, CS), which may introduce selection bias. Overall, patients with LAM in this study must be seen as high-risk patients, in whom EMB was performed, and not representative of the average AM, where EMB is generally not performed and a definite diagnosis is reached using cardiac magnetic resonance.⁵ Furthermore, staining for PD-1 and PD-L1 was performed by 3 pathology services, which analyzed the samples. Nevertheless, techniques for assessing PD-1 and PD-L1 markers have reached high-quality standards due to their extensive use in clinical oncology. Another limitation was the visual semiquantitative assessment of PD-1/PD-L1 rather than a quantitative assessment, although this approach can, again, be more easily incorporated into clinical pathology reports. We did not perform double staining of CD3 on T cells and PD-1 to precisely assess the proportion of CD3⁺T cells expressing PD-1. Nevertheless, from a morphologic standpoint, we believe the PD-1-expressing mononuclear cells in the heart were primarily T cells, as shown by representative cases where consecutive tissue sections were stained by CD and PD-1 (Supplemental Figures 5 and 6). Our study focused on the PD-1/PD-L1 axis, which is more frequently associated with ICI-related myocarditis (approximately 98% of cases) when targeted by inhibitor agents. Nevertheless, other immune checkpoints could play a role in myocarditis but were not investigated in our study, such as CTLA-4.⁴ In fact, anti-CTLA-4 monotherapy accounts for only 1% of ICI-associated myocarditis cases.¹² Given that this is a retrospective study using samples from residual biopsies, we focused our analyses on the more likely relevant pathway, PD1/PD-L1. In addition, our study lacks systematic genotyping, As pathogenic or likely pathogenic variants occur in 8% to 16% of patients with AM,³⁹ and DSP hot-phase episodes resemble lymphocytic myocarditis,⁴⁰ a next step could be to integrate the genotypes of patients with AM with the PD-L1/PD-1 phenotype. Finally, some subgroup analyses must be seen as exploratory, because they are not adjusted for multiple testing. In addition, we did not adjust for clinical variables such as age, sex, or time since disease onset, which could be potential confounders.

Conclusions

While PD-L1 is highly expressed in cardiomyocytes in patients with LAM and is associated with clinical severity, PD-1 expression is dampened in the corresponding cardiac infiltrating inflammatory cells compared with patients with ACR and ICI-AM. Blunted PD-1 expression and PD-L1 overexpression in the hearts of patients with LAM could suggest a disruption of a negative feedback mechanism.

Perspectives.

COMPETENCY IN MEDICAL KNOWLEDGE: Cardiomyocytes of patients with lymphocytic acute myocarditis express PD-L1, which is generally not expressed by hearts without T-cell infiltration. PD-L1 expression on cardiomyocytes in lymphocytic acute myocarditis is likely a response to T-cell-mediated injury not observed in other forms of inflammatory cardiomyopathies like cardiac sarcoidosis. PD-1 expression is blunted in infiltrating inflammatory cells in lymphocytic acute myocarditis compared with acute cellular rejection in transplant or immune checkpoint inhibitor-associated myocarditis. Overall, these findings suggest a disruption of the PD-L1/PD-1 axis in lymphocytic acute myocarditis, even if the cross-sectional design of the study can only demonstrate association and not causality.

TRANSLATIONAL OUTLOOK: PD-L1 expression on cardiomyocytes could represent a cardioprotective mechanism to counter myocardial T-cell-mediated injury in lymphocytic acute myocarditis and acute cellular rejection. PD-L1 expression has been associated with higher inflammatory infiltration in the heart, as in fulminant forms. Dampened PD-1 expression in infiltrating inflammatory cells of patients with lymphocytic acute myocarditis could be a primary disruptive mechanism and a potential target of future therapies.

Funding support and author disclosures

Dr Ammirati has received a grant from the Italian Ministry of Health and NextGenerationEU (PNRR-MAD-2022-12376225) and is a consultant for Lexeo and ICON and was a consultant for Hotgen Health Inc. Dr Adler is a consultant for Lexeo Therapeutics and has equity in Lexeo Therapeutics and Rocket Therapeutics. Dr Brambatti is currently employed by Merck Sharp & Dohme (MSD). Dr Mäyränpää has received lecture fees or served on advisory boards for Bayer, Boehringer Ingelheim, Bristol Myers Squibb, Merck Sharp & Dohme (MSD), Amgen, Aiforia, and FILHA. Dr Heidecker is funded by the German Heart Center Foundation and the Deutsche Herzstiftung; and is inventor on patents that use RNA for diagnosis of myocarditis. Patent protection is in process for MCG for diagnosis and measurement of therapy response in inflammatory cardiomyopathy. Patent protection is in process for cytokines for targeted therapy in inflammatory cardiomyopathy and heart failure. Dr Moslehi is supported by the NIH (P01 HL141084, R01HL141466, R01HL155990, R01HL160688, R01HL156021) and has served as a consultant or served on advisory boards for Novartis, Bristol Myers Squibb, Takeda, Daiichi Sankyo, AstraZeneca, Arcus, Incyte, Bayer, ImmunoCore, Innovent Bio, BeiGene, Kiniksa, and Shattuck Laboratories. Dr Klingel has received speaker fees from Pfizer and Alnylam and declares speaker honoraria from Alnylam and Pfizer. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. The Fondazione Centro Cardiologia e Cardiochirurgia A. De Gasperis, Niguarda Hospital, Milan, Italy, and the European Union, NextGenerationEU—Mission 6/Component 2/Investment 2.1/PNRR—CUP H43C21000150006 (grant identifier PNRR-MAD-2022-12376225) funded this study.