Novel insights into the histopathological characteristics of immune checkpoint inhibitor-related myocarditis Authors: Tatsuya Nishikawa, Tomoyuki Otani, Hironori Yamamoto, Taku Yasui, Motohiro Shingu, Yuto Osumi, Kenta Ishibashi, Yoshihiko Ikeda, Akiko Nonaka, Yuko Fukuda, Ryosuke Murai, Yutaka Furukawa, Shigeo Hara, Daigo Sawaki, Noritoshi Hiranuma, Yoshitaka Ohashi, Tomofumi Takaya, Yujiro Asada, Natsuko Ishii, Toru Kubo, Yoshiharu Kinugasa, Emi Date, Norishige Iizuka, Naka Sakamoto, Takeshi Kitai, Chisato Izumi, Kisaki Amemiya, Keiko Ohta-Ogo, Kensaku Nishihira, Mitsuo Kinugasa, Toshio Shimokawa, Koichi Tamita, Masashi Fujita, Kinta Hatakeyama Published: 2026 ABSTRACT AbstractBackground and AimsPathological examination is commonly used to diagnose immune checkpoint inhibitor (ICI)-related myocarditis. However, well-organized clinicopathological evidence remains limited. In this study, we aimed to investigate the pathological and clinical characteristics of myocarditis to enhance understanding.MethodsThis multicentre retrospective observational study included 38 patients diagnosed with or suspected of having ICI-related myocarditis. Patients were stratified using the pathological grading system proposed by Palaskas et al. Lymphocyte infiltration was assessed using both global and hot-spot (hotspot) methods. Immunohistochemistry was performed for lymphocytes, macrophages, and inflammatory markers, including human leucocyte antigen–DR isotype (HLA-DR), programmed death-ligand 1 (PD-L1), granzyme B, and tenascin C.ResultsAn increased infiltration of CD3, CD4, CD8-positive T cells, and CD68 and CD163-positive macrophages was observed with higher pathological grades. HLA-DR and tenascin C (clone 4F10) were sensitive markers of inflammation, whereas PD-L1 and tenascin C (clone 4C8) were more specific for high-grade inflammation. Global assessment more accurately predicted myocarditis-related mortality than hotspot assessment, although both methods similarly predicted fulminant myocarditis and major adverse cardiac events. Two patients presented with suspected ICI-mediated non-inflammatory left ventricular dysfunction (NILVD), with minimal myocardial inflammation.ConclusionsPathological grading using both global and hotspot assessment is valuable for the diagnosis and risk stratification of ICI-related myocarditis. As low-grade inflammation is not uncommon, supplementary markers such as HLA-DR, granzyme B, tenascin C, and PD-L1 may aid diagnosis. Furthermore, ICI-mediated NILVD may encompass cases of low-grade myocarditis.Keywords: Myocarditis, Immune checkpoint inhibitors, Ligands, Inflammation, HLA-DR antigens Introduction The progression of cancer treatment has been remarkable, particularly with the advent of immune checkpoint inhibitors (ICIs), which have revolutionized oncology care.1,2 Unlike conventional chemotherapy, ICIs are associated with a distinct spectrum of side effects known as immune-related adverse events (irAEs).2 IrAEs involve inflammation of specific tissues and can affect virtually any organ system in the body.3,4 Among these, ICI-related myocarditis is potentially fatal.5,6 Even when not life-threatening, it can lead to the discontinuation or suspension of cancer therapy.7 Therefore, cardio-oncologists have shown great interest in elucidating the pathophysiology of ICI-related myocarditis. Shared antigens between myocardial and tumour tissues may trigger an autoimmune response mediated by cytotoxic immune cells.8,9 Although numerous preclinical studies have been conducted, the underlying mechanisms of ICI-related myocarditis remain unclear.10 In clinical practice, this condition exhibits diverse phenotypes in terms of severity, clinical course, and prognosis. Myocardial pathological examination provides the most valuable insights into the disease mechanism and diagnosis. Although it is often utilized in diagnosing ICI-related myocarditis, only a limited number of well-organized pathological datasets have been reported to date.11,12 Therefore, in the present study, we aimed to investigate the pathological and clinical characteristics of ICI-related myocarditis to enhance our understanding of its pathogenesis. Methods Patients This was a retrospective, multicentre, observational study. The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Akashi Medical Center Institutional Review Board (approval number, 2024-27) and the institutional review boards of all participating institutions. We recruited 38 consecutive patients diagnosed with or suspected of having ICI-related myocarditis from nine institutes in Japan, and we consulted the pathology department at the National Cerebral and Cardiovascular Center between September 2018 and December 2024. Among 45 patients who underwent endomyocardial biopsy (EMB) or autopsy, 38 patients were consulted with the National Cerebral and Cardiovascular Center (Supplementary Figure S1) Myocardial tissue samples were obtained by EMB from the right ventricular septum (n = 35), the left ventricle (n = 1), or from autopsy specimens (n = 2). The diagnosis of ICI-related myocarditis was made based on criteria established in the European cardio-oncology guideline published in 2022.13 Measurement We retrospectively collected clinical data from medical records at each institute. This included medical history, medications, echocardiography, cardiac magnetic resonance imaging (CMR), and histopathological findings. The modified Lake Louise Criteria for myocarditis were used for CMR analysis. Either B-type natriuretic peptide (BNP, n = 3) or N-terminal pro BNP (NT-proBNP, n = 35), and either cardiac troponin I (n = 31) or T (n = 7) were measured, depending on each institute. The troponin assay kit and manufacturer’s data are shown in Supplementary Table S1. BNP data were converted to NT-proBNP using a previously reported method.14 However, BNP could not be converted owing to outrange of the determined creatinine level. The definitions of ‘fulminant myocarditis’ and cancer therapy-related cardiovascular dysfunction (CTRCD) grading followed the European guideline on cardio-oncology.13 Fulminant myocarditis was defined as haemodynamic instability, heart failure requiring non-invasive or invasive ventilation, complete or high-grade heart block, and/or considerable ventricular arrhythmia.12 CTRCD grading is outlined in Supplementary Table S2. Immunohistochemical staining Tissue samples obtained from EMB or autopsy were processed into formalin-fixed, paraffin-embedded specimens and sectioned into 4 µm-thick sections. Following haematoxylin and eosin (HE) staining, immunostaining for lymphocytes and inflammation markers was performed using the BOND-III automated immunohistochemical staining system (Leica Microsystems K.K., Tokyo, Japan) and a BOND Polymer Refine Detection Kit (Leica Microsystems K.K.). The primary antibodies used included: CD3 (code IR503, polyclonal rabbit anti-human CD3, ready-to- use; Dako Denmark, Glostrup, Denmark), CD4 (NCL-L-CD4-368, clone 4B12, diluted 1:10; Novocastra, UK), CD8 (NCL-L-CD8-4B11, clone 4B11, diluted 1:50; Novocastra, UK), CD68 (#M0814, clone PG-M1, diluted 1:1000; Agilent Technologies Inc., Santa Clara, CA, USA), CD163 (#ab74604, clone 10D6, diluted 1:10; Abcam, Cambridge, UK), Tenascin C (TNC) (#10337, 4F10TT clone, diluted 1:1000; and #10335, 4C8MS clone, diluted 1:50; Immuno-Biological Laboratories, Gunma, Japan), human leucocyte antigen–DR isotype (HLA-DR) (code-M0746, clone TAL.1B5, Dako Denmark, Glostrup, Denmark), programmed death-ligand 1 (PD-L1) (#PA0832, clone 73-10, Leica Biosystems, Newcastle, UK), and granzyme B (code-M7235, clone GrB-7, diluted 1:25; Dako Denmark, Glostrup, Denmark). CD3, CD4, and CD8 were used to identify total T-lymphocytes, helper T cells, and cytotoxic T cells, respectively; CD68 and CD163 were used to identify macrophages. Immunohistochemically stained slides were scanned using a NanoZoomer S210 C13239 (Hamamatsu Photonics K.K., Hamamatsu, Japan) and saved as whole-slide images (WSIs). All the tissue samples obtained were mounted on one slide. WSIs were analysed using QuPath (ver. 0.5.0).15 For CD3, CD4, CD8, CD68, CD163, and granzyme B, the number of immunohistochemistry (IHC)-positive cells was counted both (i) within a high-power field (HPF) (0.24 mm2) hotspot and (ii) globally across the entire biopsy specimen (for autopsy samples, areas >10 mm2 were selected by pathologists). For TNC, the percentage of IHC-positive interstitial area was measured. All quantitative data were reviewed and confirmed by two board-certified cardiovascular pathologists (TO and KH). PD-L1 and HLA-DR expression were visually assessed by two pathologists and recorded as either positive or negative, based on previous reports.11,16 Grading of ICI-related myocarditis Pathological grading was based on the criteria proposed by Palaskas et al. in 2021.12 Briefly, Grade 0 was defined as an infiltrating inflammatory cell count of <10 cells/HPF (‘negative’); Grade 1 was defined as ‘multifocal inflammatory infiltrates without overt cardiomyocyte loss,’ and was subdivided into Grade 1A (10–20 inflammatory cells/HPF) and Grade 1B (>20 inflammatory cells/HPF); Grade 2 was defined as ‘multifocal inflammatory cell infiltrates (>40 inflammatory cells/HPF) with overt cardiomyocyte loss’ (‘definite myocarditis’). In the present study, we used the count of CD3-positive cells to represent the inflammatory cell count. Statistical analysis All statistical analyses were performed using JMP version 11.0 (SAS Inc., Tokyo, Japan), EZR version 1.68 (https://www.jichi.ac.jp/saitama-sct/SaitamaHP.files/statmedEN.html), and STATA version 16 (StataCorp, TX, USA). Graphs were generated using GraphPad Prism 9 (GraphPad Software, CA, USA). Continuous variables were presented as median (interquartile range, [IQR]). Categorical data were presented as percentages (%) and compared using the two-tailed Fisher’s exact test. Trends across pathological grades were evaluated using the Cochran–Armitage trend test for categorical variables and the Jonckheere–Terpstra test for continuous variables. In graphs, continuous data are shown as median with IQR. Statistical significance was set at P < .05. Results Patient characteristics As summarized in Table 1, among the 38 study patients, 37% were female, and the median (IQR) age was 70 (65–75) years. The most common cancer types were lung (42%), followed by kidney and bladder (21%). Among the ICIs, pembrolizumab was used in 55% of patients. Dual ICI therapy was observed in 8% (ipilimumab and nivolumab, 5%; durvalumab and tremelimumab, 3%). Other non-cardiovascular irAEs were observed in 24 patients (63%). A history of percutaneous coronary intervention was noted in two patients (5%). In the present study, half of the patients, 19 (50%), were symptomatic, and the other half had asymptomatic myocarditis. Among the 38 patients, 11 did not meet the clinical criteria for ICI-related myocarditis according to the European guideline.13 Of these 11 cases, 2 were diagnosed with ICI-related myocarditis based on pathological criteria. Steroid therapy was initiated in 26 (68%) patients, and 13 (34%) received additional immunosuppressive therapy. Fulminant myocarditis occurred in eight patients (21%), and five patients (13%) died from myocarditis. In this cohort, the median time from ICI initiation to the diagnosis of myocarditis or suspected myocarditis was 134 days (IQR, 43–222). More specifically, in patients meeting the ESC guideline clinical or pathological criteria, the median time was 126 days (IQR, 58–183) (n = 27), whereas for others, the median time was 142 days (IQR, 58–245) (n = 11) (P = .986). Data from echocardiography, electrocardiography, and right heart catheterization are outlined in Supplementary Table S3. The median left ventricular ejection fraction was 46.7% (IQR, 37.0–61.8). Table 1Baseline characteristics of all patients N = 38Age [years]70[65–75]Co-occurring irAEs24(63)Female14(37)At the diagnosis of myocarditis21(55)BMI21[18.8–24.3]After diagnosis of myocarditis4(11)Hypertension16(42)Grades of CTRCDDiabetes9(24)Symptomatic-very severe7(18)Dyslipidaemia10(26)Symptomatic-severe9(24)SmokingSymptomatic-moderate2(5) Current smoker5(13)Symptomatic-mild1(3) Past smoker19(50)Asymptomatic-severe1(3)Chronic kidney disease12(32)Asymptomatic-moderate5(13)History of PCI2(5)Asymptomatic-mild13(34)History of cardiac surgery0(0)Fulminant myocarditis8(21)History of DOX treatment3(8)ECG abnormality22(58)Type of cancerCAVB6(16) Lung16(42)AF6(16) Kidney and bladder8(21)PVC4(11) Larynx and pharynx4(11)SSS3(8) Uterus4(11)VT, NSVT2(5) Hepatobiliary2(5)ST-T change4(11) Oesophagus2(5)PAC1(3) Breast1(3)Treatment for ICI Unknown1(3)Steroid26(68)Type of ICISteroid pulse treatment (high-dose)18(47) Pembrolizumab21(55)Immunosuppression therapy13(34) Atezolizumab5(13)MMF7(18) Nivolumab4(11)Tocilizumab2(5) Durvalumab5(13)IVIg2(5) Ipilimumab + Nivolumab2(5)Tacrolimus2(5) Durvalumab + Tremelimumab1(3)Troponin I, T [pg/ml]302[31–2717]Time from ICI initiation to myocarditis [day]134[43–222]NT-proBNP [pg/ml]2719[608–6119]Cardiac magnetic resonance imaging18(47)Death by myocarditis5(13)Abbreviations: PCI, percutaneous coronary intervention; DOX, doxorubicin; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event; CTRCD, cancer therapy-related cardiovascular dysfunction; ECG, electrocardiogram; CAVB, complete atrioventricular block; AF, atrial fibrillation; PVC, premature ventricular contraction; SSS, sick sinus syndrome; VT, ventricular tachycardia; NSVT, non-sustained ventricular tachycardia; MMF, mycophenolate mofetil; IVIg, intravenous immunoglobulin; MRI, magnetic resonance imaging; BMI, body mass index Baseline characteristics of all patients Abbreviations: PCI, percutaneous coronary intervention; DOX, doxorubicin; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event; CTRCD, cancer therapy-related cardiovascular dysfunction; ECG, electrocardiogram; CAVB, complete atrioventricular block; AF, atrial fibrillation; PVC, premature ventricular contraction; SSS, sick sinus syndrome; VT, ventricular tachycardia; NSVT, non-sustained ventricular tachycardia; MMF, mycophenolate mofetil; IVIg, intravenous immunoglobulin; MRI, magnetic resonance imaging; BMI, body mass index Clinical and histopathological characteristics of myocarditis In this study, histopathological samples from all patients were reviewed by pathologists. Grading was based on the system developed by Palaskas et al. in 2021. Two grading methods were used to count CD3-positive lymphocytes using the QuPath application: global grading, which involved counting all CD3-positive lymphocytes in the whole slide image and converting the count to HPF units, and hotspot grading, which involved counting CD3-positive cells in the most densely infiltrated HPF (‘hotspot’). Representative pathological images are shown in Figure 1 based on global grading. As presented in Supplementary Tables S4 and S5, hotspot grading yielded more Grade 2 cases and fewer Grade 0 cases than global grading. According to global grading, the number of cases was as follows: Grade 0 (n = 19), Grade 1A (n = 8), Grade 1B (n = 5), and Grade 2 (n = 6). By hotspot grading: Grade 0 (n = 11), Grade 1A (n = 4), Grade 1B (n = 9), and Grade 2 (n = 14). As shown in Figure 2 and Supplementary Table S6, CD3, CD4, CD8, CD68, and CD163-positive immune cells showed a consistent and significant trend with both global and hotspot grading. Furthermore, granzyme B, HLA-DR, PD-L1, and TNC (4C8) showed similar trends, corresponding to the extent of immune cell infiltration (Figure 3). However, HLA-DR and TNC (4F10) appeared to be more sensitive markers than others. Notably, 16% of Grade 0 cases (global) and 9% of Grade 0 cases (hotspot) were HLA-DR positive by immunostaining. In addition, TNC (4F10) demonstrated high sensitivity to inflammation, showing no significant trend across pathological grades and %positive area of 31% (IQR 12–53) in Grade 0 (global) cases and 31% (IQR 7–52) in Grade 0 (hotspot) cases. Meanwhile, TNC (4C8) demonstrated higher levels, specifically for more severe inflammation. Regarding hotspot grading, it was mostly positive in Grade 2 and a subset of Grade 1B cases. In the present study, we assessed the inter-observer agreement for qualitative evaluation of PD-L1 and HLA-DR. Kappa coefficient of PD-L1 was 0.74 (P < .001) and of HLA-DR was 0.48 (P = .003). These findings showed that the inter-observer agreement was sufficiently high for evaluation. Representative microscopic images of ICI-related inflammation for each pathological grade using grade (global). Immunostaining of tenascin C in Grades 0 and 1A, PD-L1 in Grades 0 and 1A, and granzyme B in Grade 0 were negative, as shown. Panel width = 275 µm. Abbreviations: HE, haematoxylin and eosin staining; ICI, immune checkpoint inhibitor; PD-L1, programmed death-ligand Trend of pathological count of inflammatory cells according to grade (global) and grade (hotspot). (A) CD3, (B) CD4, (C) CD8, (D) CD68, (E) CD163. P values were derived from the Jonckheere–Terpstra test for continuous variables. Graphs depict median and IQR for continuous variables and prevalence in percentage for categorical variables. Abbreviation: IQR, interquartile range Trend of immunohistological biomarkers according to grade (global) and grade (hotspot). (A) Granzyme B, (B) HLA-DR, (C) PD-L1, (D) Tenascin C (4C8), (E) Tenascin C (4F10). P values were derived from the Jonckheere–Terpstra test for continuous variables and the Cochran–Armitage trend test for categorical variables. Graphs depict median and IQR for continuous variables and prevalence in percentage for categorical variables Troponin I or T levels (expressed as fold upper normal limit [ULN]) showed a significant trend in both grading methods (global and hotspot), whereas NT-proBNP showed no significant trend in either method (Supplementary Figure S2). In the present study, more than half of the Grade 0 cases (either global or hotspot) received steroid therapy. All cases in Grade 2 (global) had steroid therapy, whereas approximately 80% of cases in Grade 2 (hotspot) received steroid therapy (Figure 4A). Only the Grade (global) showed a significant trend across grades (Cochran-Armitage trend test, P = .048). Whereas the high-dose steroid therapy has significant trends in both gradings (Cochran-Armitage trend test, P < .05) (Figure 4B). However, the prevalence of immunosuppressive therapy other than steroids showed no significant trend in either group (Cochran-Armitage trend test, P > .05) (Figure 4C). In addition, the severity of the CTRCD grade, as defined in the European guideline for onco-cardiology13 (Supplementary Table S2), was assessed for both grading methods (global and hotspot) (Figure 4D). In Grade 2 (global), no cases were asymptomatic myocarditis, whereas in Grade 2 (hotspot), approximately half of the cases were asymptomatic. The specific data for Figure 4 are shown in Supplementary Table S4. The time from ICI initiation to the diagnosis of myocarditis or suspected myocarditis, as well as echocardiography, electrocardiography, and right heart catheter data, did not differ significantly across the pathological grades (Supplementary Tables S4 and S5). Trends in clinical therapies and CTRCD grading according to pathological grade in ICI-related myocarditis. (A) Steroid therapy, (B) high-dose steroid therapy, (C) other immunosuppressive therapies, and (D) CTRCD grade according to global and hotspot pathological grades. P values were derived from the Cochran–Armitage trend test. Graphs depict prevalence in percentage for categorical variables. Abbreviations: CTRCD, cancer therapy-related cardiovascular dysfunction; ICI, immune checkpoint inhibitor. Prognosis based on the pathological grading of myocarditis In terms of risk assessment, grading (global) may serve as a better tool for stratifying potentially high-mortality groups than grading (hotspot) (Cochran–Armitage trend test, P = .028 vs. P = .355) (Figure 5A). However, for other outcomes, both grading had similarly significant trend by pathological grade. The prevalence of fulminant myocarditis was also assessed for both grading (global) (Cochran–Armitage trend test, P < .001) and grading (hotspot) (Cochran-Armitage trend test, P = .003) (Figure 5B). For further prognostic assessment, we evaluated two differently defined major cardiovascular adverse events (MACEs) (Figure 5C and D). Both MACEs showed a statistically significant trend, indicating that a higher pathological grade is associated with a higher incidence of MACEs. Additionally, the similarity of grade (global) and grade (hotspot) was assessed using Fleiss and Cohen’s weighted Kappa coefficient. Consequently, the Kappa coefficient was 0.669 (P < .001), indicating a significant clinical similarity between the two grading methods. As a sensitivity analysis, we assessed the prognostic events and immunosuppressive therapy used in cases without autopsy (n = 36). As shown in Supplementary Figure S3, fulminant myocarditis and MACEs showed trends consistent with the original analysis, whereas myocarditis-related death showed no significant trend. Trends in clinical outcomes of ICI-related myocarditis. Grade (global) may be a better marker for death due to myocarditis than grade (hotspot); however, for other outcomes such as fulminant myocarditis and MACEs, both grading systems showed similarly significant trends. P values were derived from the Cochran–Armitage trend test. Graphs depict prevalence in percentage for categorical variables Additionally, we have shown differences in proportions of cardiac death, fulminant myocarditis, and MACEs for each grade with grade 0 as a reference. (Supplementary Table S7). Furthermore, we compared our histopathological gradings with the recently reported clinical risk score developed by Power J et al.17 The area under the curve (AUC) for assessing MACEs was not significantly different between our grading (global) and Power’s risk score. However, grading (hotspot) was inferior to Power’s risk score (Supplementary Figure S4). In the present study, an atypical and noteworthy case classified as Grade (global) 0 and Grade (hotspot) 0, with only slight lymphocytic infiltration (<10 CD3 + lymphocyte cells/HPF) was identified. Despite this low pathological grade, the patient had a peak troponin I of 681 pg/ml and fulfilled the clinical diagnostic criteria of the European guideline on cardio-oncology. This case exhibited steroid resistance, resulting in death from myocarditis. We also assessed the utility of the diagnosis criteria defined in the European guideline on cardio-oncology.13 As shown in Figure 6, among the cases not diagnosed with ICI-related myocarditis based on these clinical or pathological criteria,13 none developed fulminant myocarditis, died from myocarditis, or experienced a MACE. This suggests that the criteria appropriately identify high-risk patients. Outcomes of myocarditis with or without diagnosis of ICI-related myocarditis based on the European guideline for cardio-oncology (13). Patients who did not meet the ESC guideline criteria for ICI-related myocarditis experienced no adverse outcomes, such as death due to myocarditis, fulminant myocarditis, or MACEs. Abbreviations: ESC, European Society of Cardiology; VT, ventricular tachycardia; CAVB, complete atrioventricular block; ADHF, acute decompensated heart failure ICI-mediated myocarditis, non-inflammatory left ventricular dysfunction, and low-grade myocarditis Two noteworthy cases were identified. As shown in Figure 7, case 1 had Grade (global) 1A and Grade (hotspot) 2, with myocardial loss, but troponin I below the sensitive threshold. Case 2 had Grade (global) 0 and Grade (hotspot) 1A, without myocardial loss and with troponin I also below the threshold. Both were measured using the Abbott troponin I kit with a ULN of 26 pg/ml. The ICI administered was durvalumab in case 1 and pembrolizumab in case 2. These cases did not meet the clinical diagnostic criteria for ICI-related myocarditis; however, they exhibited clear myocardial inflammation. Notably, case 1 fulfilled the pathological diagnostic criteria for ICI-related myocarditis.13 Specifically, case 1 had a left ventricular ejection fraction (LVEF) of 43%, NT-proBNP of 1338 pg/ml, with no other irAEs, arrhythmia, or heart failure symptoms. The patient died of cancer within a few weeks; therefore, CMR could not be performed. Case 2 had an LVEF of 45%, NT-proBNP of 48 pg/ml, with no other irAEs, arrhythmia, or symptoms of heart failure. CMR showed a suggestive increase in T2 values but normal T1 values and no late gadolinium enhancement. Microscopic images of two cases of low-grade myocarditis potentially diagnosed as ICI-related non-inflammatory left ventricular dysfunction. In Case 1, the arrow in HE staining indicates myocardial loss, whereas the arrows in granzyme B immunostaining indicate positively stained cells. Case 1 showed negative tenascin C (clone 4C8), positive tenascin C (clone 4F10), positive HLA-DR, weakly positive granzyme B, and negative PD-L1. Case 2 showed negative tenascin C (4C8), positive tenascin C (4F10), positive HLA-DR, negative granzyme B, and negative PD-L1. LVEF was 43% in Case 1 and 45% in Case 2. Panel width = 275 µm. Abbreviations: HLA-DR, human leukocyte antigen—DR isotype; LVEF, left ventricular ejection fraction Furthermore, in the present study, 11 of the 38 cases did not fulfil the clinical diagnostic criteria for ICI-related myocarditis. Among them, five cases had Grade (global) 0 but higher Grade (hotspot) (1A; n = 3, 1B; n = 2). All met the immunohistochemical criteria for abnormal inflammatory infiltrates defined by the European Society of Cardiology Working Group on Myocardial and Pericardial Diseases.18 None was fulminant myocarditis or resulted in cardiovascular death; however, they could be defined as low-grade myocarditis. In addition, 6 out of 11 cases had a slight increase in peak troponin (13.0–37.7 pg/ml), varying LVEF (38–73%), and Grade (global) 0 and Grade (hotspot) 0, potentially indicating ICI-mediated non-inflammatory left ventricular dysfunction (NILVD). Finally, in the present study, 2 of the 38 patients underwent ICI rechallenge. One patient had an LVEF of 31%, troponin I level of 1.2× ULN, and NT-proBNP of 4932 pg/ml; the other had an LVEF of 42%, troponin I of 0.92× ULN, and NT-proBNP of 138 pg/ml. Neither patient had arrhythmias or other irAEs, and both met the ESC guideline clinical criteria for ICI-related myocarditis. No cardiac toxicity was observed after ICI rechallenge. Discussion In this study, several clinically important aspects of ICI-related myocarditis were revealed. First, pathological grading was useful for assessing both the diagnosis and prognosis of myocarditis (or myocardial inflammation). Although hotspot analysis is potentially useful for diagnostic purposes, global assessment is superior for assessing overall mortality, as suggested by the results. Second, the number of infiltrating lymphocytes and immunohistochemical assessment of granzyme B, HLA-DR, PD-L1, and TNC could provide supportive evidence of myocardial inflammation. Third, the recently proposed ‘ICI-mediated NILVD’ may, in real-world clinical settings, include cases of low-grade myocarditis. Fourth, we evaluated and showed that the diagnostic criteria for ICI-related myocarditis established in the first guideline of cardio-oncology worldwide, published by the ESC and other related organizations,13 showed great usefulness in diagnosing clinically significant myocarditis. To the best of our knowledge, this is the largest pathological study on the topic reported to date. Low-grade myocarditis typically exhibits multifocal lymphocyte infiltration in the myocardium, whereas high-grade myocarditis demonstrates diffuse, high-density lymphocyte infiltration. Therefore, diagnosing and grading low-grade myocarditis can be ambiguous, particularly in terms of how lymphocytes are counted. To address this, we employed two grading methods: one assessing global infiltration and the other focused on hotspots. As shown in Figure 2, all lymphocyte markers (CD3, CD4, and CD8) and macrophage markers (CD68 and CD163) were markedly elevated in Grade 2 of both global and hotspot grading. Owing to the nature of the grading methodology, the hotspot approach is more sensitive but reflects lower global myocardial inflammation. Therefore, global tissue grading may better identify more severe myocarditis than hotspot grading. This trend is evident in the rates of death due to myocarditis, fulminant myocarditis, and MACE (Figure 5). Notably, grading by hotspot still showed statistically significant trends, except for death from myocarditis. The mortality rate was as low as 13% in the total cohort, and the small number of total patients and death events might have affected the negative results of grading by hotspot. In clinical and pathological practice, there remains considerable uncertainty regarding how many HPF should be evaluated when counting infiltrating inflammatory cells. Even among pathological research studies, each employs its original methodology or criteria. Therefore, the dual approach we propose—using either global or hotspot grading—may offer greater consistency. In particular, hotspot assessment may be sufficiently useful for clinical application, as both grading methods demonstrated comparable statistical trends in clinical outcomes. Additionally, the similarity of the two methods was demonstrated by Fleiss and Cohen’s weighted Kappa coefficient. Grade 2 myocarditis is defined as ‘definite myocarditis’ with high-grade inflammation, whereas Grade 1 refers to low-grade inflammation. However, HLA-DR, TNC, and PD-L1 were positive even in Grades 0 and 1. HLA-DR has been recognized as a useful myocarditis marker in international guidelines.19 Furthermore, PD-L1 and TNC are well-known pathological markers for ICI-related myocarditis.11,12,20–23 TNC relates to myocarditis, doxorubicin-induced myocarditis,24 and rheumatic heart disease.25 As shown in Figure 4D, the trend of CTRCD grading showed that the rate of symptomatic heart failure is not low, even in Grade 0 and Grade 1A. The definite diagnosis of myocarditis is difficult in low-grade inflammation, as myocardial loss is often difficult to determine, even by a pathologist. In addition, EMB always carries a risk of sampling error. HLA-DR and Tenascin C were expressed in some grade 0 cases. Previous studies have reported that these markers might be elevated owing to systemic inflammation, with no direct effect of infiltrated inflammatory cells.16,24 Therefore, these supportive pathological markers are useful for decision-making in the treatment of ICI-related myocarditis. In terms of the clinical course in the present study, steroid treatment with or without high-dose pulse therapy was frequently introduced even in Grades 0 and 1. Furthermore, additional immunosuppressive therapies were often used regardless of pathological grade. In these cases, the severity of concomitant irAEs may influence treatment decisions. Recently, cardiovascular irAEs have been classified into two categories: ‘ICI-mediated myocarditis’ and ‘ICI-mediated NILVD’, according to a statement by the Heart Failure Association of the European Society of Cardiology (ESC) and the ESC Council of Cardio-Oncology.10 As stated, we sometimes encounter cases of heart failure without obvious signs of inflammation. However, the mechanisms underlying non-inflammatory cardiotoxicity remain unclear. Previous studies have proposed several potential causes, including antibody-related mechanisms, PD-L1 expression on myocardium, tumour necrosis factor-alpha involvement, glucose and lipid metabolism alterations, and mitochondrial dysfunction.10,26–28 In the present study, two cases with normal troponin levels and abnormal inflammatory cell infiltration were identified. Notably, one of them was Grade (hotspot) 1B with myocardial loss, which satisfied the pathological criteria for ICI-related myocarditis. However, these could have been diagnosed as NILVD without EMB. Moreover, seven cases did not satisfy the clinical diagnostic criteria for ICI-related myocarditis according to the European guidelines, but all showed abnormal inflammatory cell infiltration as defined by the ESC 2013 position statement on myocarditis.18 Specifically, five of the seven cases had pathological Grade (hotspot) 1A (n = 3) or 1B (n = 2). Based on this evidence, these cases of NILVD may include low-grade myocarditis. These findings could provide novel insights into the pathology of ICI-related myocarditis, changing the paradigm in how we define truly non-inflammatory disease. After the precise diagnostic criteria were developed by the European guidelines, most cases of ICI-related myocarditis have recently been diagnosed based on the clinical criteria, which heavily rely on troponin levels and CMR. Consequently, the number of EMBs appears to have decreased in most countries. However, certain patients are unable to undergo CMR. For instance, those with a cylinder pump for venous injection, mechanical ventilation, or devices, those who are unconscious, or those with unstable vital signs. In such cases, EMB would be necessary. Accordingly, the latest expert consensus on myocarditis, published by the American College of Cardiology in 2024, recommended EMB for stage B ICI-related myocarditis.19 Pathological samples are highly valuable for diagnosis and for understanding the mechanisms underlying ICI-related myocarditis. Nonetheless, to further elucidate disease mechanisms, larger datasets with prospective studies are needed. This study has some limitations worth noting. First, this is a retrospective study with a small number of available cases; therefore, the statistical analysis lacks sufficient power to yield definitive conclusions. Second, pathological analysis always carries the potential bias of sampling error. In addition, this study included two autopsy samples, whereas all other samples were obtained via catheter-based EMB. Third, inter-institutional differences in the types of biomarkers used were present, such as troponin I vs. troponin T, and BNP vs. NT-proBNP. The local reference ranges also varied. However, as discussed, these differences did not substantially impact the overall results of the present study. In conclusion, pathological grading for ICI-related myocarditis was useful for diagnosis and risk stratification. Grading could involve either global or hotspot assessments. Given that low-grade inflammation is not uncommon pathologically, additional supportive markers such as HLA-DR, granzyme B, TNC, and PD-L1 may be useful. Furthermore, we proposed that ICI-mediated NILVD might include cases of low-grade myocarditis. Supplementary Material Acknowledgements None. Contributor Information Tatsuya Nishikawa, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan; Department of Onco-Cardiology, Osaka International Cancer Institute, 3-1-69, Otemae, Chuo-ku, Osaka City, Osaka 541-8567, Japan. Tomoyuki Otani, Department of Pathology, National Cerebral and Cardiovascular Center, 6-1, Kishibe-Shimmachi, Suita, Osaka 564-8565, Japan. Hironori Yamamoto, Department of Onco-Cardiology, Osaka International Cancer Institute, 3-1-69, Otemae, Chuo-ku, Osaka City, Osaka 541-8567, Japan. Taku Yasui, Department of Onco-Cardiology, Osaka International Cancer Institute, 3-1-69, Otemae, Chuo-ku, Osaka City, Osaka 541-8567, Japan. Motohiro Shingu, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan. Yuto Osumi, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan. Kenta Ishibashi, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan. Yoshihiko Ikeda, Department of Pathology, National Cerebral and Cardiovascular Center, 6-1, Kishibe-Shimmachi, Suita, Osaka 564-8565, Japan. Akiko Nonaka, Department of Oncocardiology, Hyogo Cancer Center, Akashi, Japan. Yuko Fukuda, Department of Oncocardiology, Hyogo Cancer Center, Akashi, Japan. Ryosuke Murai, Department of Cardiovascular Medicine, Kobe City Medical Center General Hospital, Kobe, Japan. Yutaka Furukawa, Department of Cardiovascular Medicine, Kobe City Medical Center General Hospital, Kobe, Japan. Shigeo Hara, Department of Pathology, Kobe City Medical Center General Hospital, Kobe, Japan. Daigo Sawaki, Department of Cardiology, Jichi Medical University Hospital, Tochigi, Japan. Noritoshi Hiranuma, Department of Cardiology, Ako City Hospital, Ako, Japan. Yoshitaka Ohashi, Department of Cardiology, Ako City Hospital, Ako, Japan. Tomofumi Takaya, Division of Cardiovascular Medicine, Hyogo Prefectural Harima-Himeji General Medical Center, Himeji, Japan. Yujiro Asada, Department of Diagnostic Pathology, Miyazaki Medical Association Hospital, Miyazaki, Japan. Natsuko Ishii, Department of Cardiology and Geriatrics, Kochi Medical School, Kochi University, Kochi, Japan. Toru Kubo, Department of Cardiology and Geriatrics, Kochi Medical School, Kochi University, Kochi, Japan. Yoshiharu Kinugasa, Department of Cardiovascular Medicine and Endocrinology and Metabolism, Faculty of Medicine, Tottori University, Yonago, Japan. Emi Date, Department of Diagnostic Pathology, Kishiwada City Hospital, Kishiwada, Japan; Division of Molecular Pathology, Graduate School of Medicine, Kindai University, Sakai, Japan. Norishige Iizuka, Department of Diagnostic Pathology, Kishiwada City Hospital, Kishiwada, Japan. Naka Sakamoto, Division of Cardiology and Nephrology, Department of Internal Medicine, Asahikawa Medical University, Asahikawa, Japan. Takeshi Kitai, Department of Heart Failure and Transplantation, National Cerebral and Cardiovascular Center, Suita, Japan. Chisato Izumi, Department of Heart Failure and Transplantation, National Cerebral and Cardiovascular Center, Suita, Japan. Kisaki Amemiya, Department of Pathology, National Cerebral and Cardiovascular Center, 6-1, Kishibe-Shimmachi, Suita, Osaka 564-8565, Japan. Keiko Ohta-Ogo, Department of Pathology, National Cerebral and Cardiovascular Center, 6-1, Kishibe-Shimmachi, Suita, Osaka 564-8565, Japan. Kensaku Nishihira, Department of Cardiology, Miyazaki Medical Association Hospital, Miyazaki, Japan. Mitsuo Kinugasa, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan. Toshio Shimokawa, Department of Biostatistics, School of Medicine, Wakayama Medical University, Wakayama, Japan. Koichi Tamita, Department of Cardiovascular Medicine, Akashi Medical Center, Akashi, Japan. Masashi Fujita, Department of Onco-Cardiology, Osaka International Cancer Institute, 3-1-69, Otemae, Chuo-ku, Osaka City, Osaka 541-8567, Japan. Kinta Hatakeyama, Department of Pathology, National Cerebral and Cardiovascular Center, 6-1, Kishibe-Shimmachi, Suita, Osaka 564-8565, Japan. Supplementary data Supplementary data are available at ESC Heart Failure online. Declarations Disclosure of Interest None declared. Data Availability The data that support the findings of this study are available from the corresponding authors upon reasonable request. Funding This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number 24K10118. Ethical Approval This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Akashi Medical Center Institutional Review Board (approval number, 2024-27) and the institutional review boards of all participating institutions. Pre-registered Clinical Trial Number None supplied. References Associated Data Supplementary Materials Data Availability Statement The data that support the findings of this study are available from the corresponding authors upon reasonable request.