Anti-mitochondrial M2 antibody links to cardiac involvement and immune-mediated inflammatory myopathy-like pathology in myositis

Abstract

Objectives

Cardiac involvement in idiopathic inflammatory myopathies (IIM) is rare but potentially severe. Anti-mitochondrial M2 antibody (AMA-M2) has been implicated in cardiac involvement, but the association remains underexplored. This study aims to evaluate the clinical, pathological and prognostic features of AMA-M2 IIM.

Methods

This historic prospective cohort included IIM patients hospitalized at Peking Union Medical College Hospital between 2008 and 2020. Outcomes were prospectively collected through the Prospective Registry Of MyositIS (PROMIS) registry. Cox regression models were employed to identify risk factors of cardiac involvement and mortality.

Results

Among 987 IIM patients, 55 (6%) were AMA-M2 positive. These patients exhibited higher rates of PM (56% vs 23.5%, P < 0.001), and elevated baseline gamma-glutamyl transferase (78.0 vs 35.0, P < 0.001) and alkaline phosphatase (85.0 vs 64.0, P < 0.001). Throughout disease courses, AMA-M2-positive patients had significantly higher rates of cardiac involvement (60% vs 12.9%, P < 0.001), including arrhythmias (56%), heart failure (44%) and pulmonary hypertension (31%). Some of the muscle biopsies showed features consistent with immune-mediated necrotizing myopathy, cardiac biopsies demonstrating structural degeneration with minimal inflammation and liver biopsies confirming early-stage primary biliary cholangitis (PBC). Multivariate Cox analysis identified AMA-M2 positivity as an independent risk factor for cardiac involvement (hazard ratio 3.156, P < 0.001). Despite frequent cardiac manifestations, long-term survival did not differ between AMA-M2-positive and -negative patients (mean survival: 103.9 months vs 98.0 months, P = 0.86).

Conclusion

AMA-M2 positivity defines an IIM subgroup with significant cardiac involvement and an immune-mediated inflammatory muscle histology, but not necessarily worse long-term survival. These findings highlight the need for early recognition and tailored management of AMA-M2 IIM.

Rheumatology key messages.

• AMA-M2 positivity is independently associated with increased risk of cardiac involvement in myositis.

• AMA-M2-associated myositis displays an immune-mediated necrotizing myopathy–like muscle pathology pattern.

• AMA-M2-positive myositis patients show similar long-term survival to AMA-M2-negative patients despite frequent cardiac involvement.

Introduction

Idiopathic inflammatory myopathies (IIM) are a group of heterogeneous autoimmune diseases characterized by proximal muscle weakness, and are classified into several clinical subtypes, including DM, PM, immune-mediated necrotizing myositis (IMNM), etc. [1]. Although muscle weakness is the hallmark of IIM, patients often experience extra-muscular involvements, with cardiac involvement being a rare but serious condition. While anti-signal recognition particle (SRP) and anti-HMG-CoA reductase (HMGCR) antibodies have been associated with cardiac manifestations [2], their relatively low prevalence limit the utility in early and accurate cardiac involvement detection.

AMAs have emerged as important biomarkers in various autoimmune diseases. Among them, AMA-M2 targets the E2 subunit of the pyruvate dehydrogenase complex (PDC-E2), and is most notably associated with primary biliary cirrhosis (PBC), with a 90–95% positive rate among PBC patients [3]. Nevertheless, AMA-M2 is not specific for PBC, and is also detected in autoimmune conditions like SLE, SS and IIM.

Recent studies have suggested potential correlations between AMAs and cardiac involvement [4, 5] and specific pathological patterns in IIM [6]. However, these associations remain inconclusive due to small sample sizes and cross-sectional or case-series designs [7]. Furthermore, the treatment response and long-term prognosis in AMA-positive IIM remain unknown. To address these gaps, we conducted a single-centre historic prospective cohort to evaluate the clinical, laboratory and pathological features, as well as the prognosis of AMA-M2-positive IIM.

Methods

Study population

The study population were derived from the Prospective Registry Of MyositIS (PROMIS) cohort, a Chinese multicentre prospective registry for IIM initiated in 2017 [8]. IIM diagnoses were initially made based on the 1975 Bohan and Peter Criteria [9] and retrospectively confirmed with the 2017 EULAR/ACR Classification Criteria [1], and patients were categorized as DM and PM (including IMNM). Exclusion criteria included: (i) overlap with other autoimmune CTDs; (ii) incomplete medical records; and (iii) absence of AMA-M2 testing (Supplementary Fig. S1). The study protocol and the waiver of informed consent in retrospective studies using de-identified data was approved by the institutional ethics committee (JS-2038).

Data collection

Clinical data were derived from IIM patients hospitalized at Peking Union Medical College Hospital between January 2008 and November 2020, who were later enrolled in the PROMIS cohort and followed longitudinally at this single-centre subset. Demographic data included gender, age at symptom onset and diagnosis. Clinical manifestations, included IIM subtype, comorbidities, cutaneous symptoms (heliotrope rash, Gottron’s sign and papules, Shawl and V signs, and mechanic’s hands), digits vasculitis, ectopic calcinosis, muscle weakness, diaphragmatic muscle paralysis, oesophageal dysmotility and interstitial lung disease (ILD). Cardiac involvement was defined by the presence of at least one of the following: (i) arrhythmia requiring medications, pacemaker or defibrillator implantation, or radiofrequency ablation; (ii) pulmonary hypertension (PH) confirmed via echocardiogram; and (iii) heart failure meeting the Framingham criteria. Laboratory information included autoantibodies, maximum creatine kinase, and baseline gamma-glutamyl transferase (GGT) and ALP. AMA-M2 was measured by either chemiluminescence, indirect immunofluorescence or digital liquid chip method. Muscle, cardiac and liver biopsy findings were extracted from available pathology reports. Follow-up included the time of death, time of cardiac involvement and the last recorded medical visit.

Statistical analysis

Continuous variables were reported as mean (s.d.) or median (interquartile range). Categorical variables were expressed as frequencies and percentages, with percentages rounded to integers for groups comprising <100 individuals. Comparisons were made using Student’s t-test or Mann–Whitney U test for continuous variables, and χ² or Fisher’s exact test for categorical variables. Univariate and multivariate Cox proportional hazards models were used to identify factors associated with cardiac involvement and mortality. Kaplan–Meier survival curves were constructed and compared using the log-rank test, and median time-to event was reported. To minimize confounding factors, AMA-M2-positive patients were 1:4 matched with AMA-M2-negative patients using propensity score (nearest neighbour method with a 0.2 calliper width). All analyses were performed using SPSS v.27.0.1.0 (IBM Corp., Armonk, NY, USA) and RStudio (v.2024.04).

Results

Clinical and laboratory characteristics

A total of 987 IIM patients were identified through record screening, and 55 (6%) were AMA-M2 positive. As shown in Table 1, there were no significant differences between gender distribution, age at disease onset and at diagnosis between AMA-M2-positive and -negative patients. However, AMA-M2-positive patients had higher rates of comorbid coronary artery disease (22% vs 6.3%, P < 0.001) and hypertension (46% vs 23.5%, P < 0.001). This group of patients exhibited a higher proportion of PM subtype (56% vs 23.5%, P < 0.001), but fewer skin manifestations such as heliotrope rash (4% vs 33.6%, P < 0.001), Gottron’s sign and papules (15% vs 43.5%, P < 0.001), and Shawl and V signs (6% vs 37.3%, P < 0.001). Prevalence of digit vasculitis (0% vs 12.7%, P = 0.005), diaphragmatic muscle paralysis (15% vs 1.9%, P < 0.001) and ILD (38% vs 63.3%, P < 0.001) were also different between the two patient groups, while that of muscle weakness, oesophageal dysmotility and ectopic calcinosis remained comparable.

Table 1.

Demographical, clinical and laboratory characteristics of AMA-M2-positive and AMA-M2-negative patients.

	AMA-M2 positive (n = 55)	AMA-M2 negative (n = 932)	P-value
Demographic features
Age at onset, years [median (IQR)]	51.0 (15.0)	49.0 (20.0)	0.239
Age at diagnosisa, years [median (IQR)]	53.5 (16.0)	50.0 (19.0)	0.093
Gender, female (n, %)	36 (66)	633 (67.9)	0.704
Comorbidities at diagnosis (n, %)
Coronary artery disease	12 (22)	59 (6.3)	<0.001
Hypertension	25 (46)	219 (23.5)	<0.001
Stroke	3 (6)	29 (3.1)	0.340
Malignancies	3 (6)	91/931 (9.8)	0.289
Type 2 diabetes mellitus	9 (16)	183 (19.6)	0.551
Clinical manifestations (n, %)
PM subtype	31 (56)	219 (23.5)	<0.001
Skin involvementsb	16 (29)	618 (66.3)	<0.001
Digits vasculitis	0	118 (12.7)	0.005
Muscle weakness	39 (71)	625 (67.1)	0.554
Diaphragmatic muscle paralysis	8 (15)	18 (1.9)	<0.001
Oesophageal dysmotility	10 (18)	229 (24.6)	0.282
Interstitial lung disease	21 (38)	590 (63.3)	<0.001
Cardiac involvement	33 (60)	120 (12.9)	<0.001
Arrhythmia	31 (56)	48 (5.2)	<0.001
Sinus node dysfunction	5 (9)	/	/
Atrial/supraventricular	25 (46)	/	/
Ventricular	28 (51)	/	/
Conductional	8 (15)
Additional interventionsc	10 (18)	/	/
Heart failured	24 (44)	34 (3.6)	<0.001
HFrEF	12 (22)	/	/
HFmrEF	6 (11)	/	/
HFpEF	6 (11)	/	/
Pulmonary hypertensione	17 (31)	74 (7.9)	<0.001
Mild	8 (15)	/	/
Moderate	8 (15)	/	/
Severe	1 (1)	/	/
Laboratory results (n, %)
CK elevation	47 (87)	536 (58.3)	<0.001
GGT elevation	33/52 (64)	325/863 (37.3)	<0.001
ALP elevation	19/51 (37)	106/858 (12.4)	<0.001
Bile duct enzymes abnormalities	33 (60)	339 (36.4)	<0.001
Treatments (n, %)	(n = 54)	(n = 832)
Steroid pulse therapy	15 (28)	217 (26.1)	0.784
Immunomodulatory therapy	52 (95)	761 (91.4)	0.615
IVIG	18 (33)	292 (35.2)	0.783
MTX	29 (54)	318 (38.2)	0.024
CYC	26 (52)	450 (54.1)	0.750
MMF	6 (11)	73 (8.8)	0.559
AZA	3 (6)	39 (4.7)	0.737
CNIs	18 (33)	252 (30.3)	0.638
Biological agentsf	1 (2)	65 (7.8)	0.174

^aThe age at diagnosis represents the patient’s actual age at baseline hospitalization for idiopathic inflammatory myopathies diagnosis.

^bSkin involvements are defined as the presence of heliotrope rash, Gottron’s sign or papule, Shawl and V sign, or mechanic’s hands.

^cAdditional interventions include implantable cardioverter-defibrillator, pacemaker and radiofrequency ablation.

^dHFrEF/HFmrEF/HFpEF: heart failure with reduced/mildly reduced/preserved ejection fraction, defined by left ventricular ejection fraction ≤40%/41–49%/≥50% measured through echocardiogram.

^eThe degree of pulmonary hypertension was classified according to pulmonary artery systolic pressure (PASP) estimated by echocardiography: mild: 30 < PASP ≤ 50 mmHg; moderate: 50 < PASP ≤ 70 mmHg; severe: PSAP >70 mmHg.

^fRecorded biological agents include anti-CD20 antibodies, Janus kinase inhibitors and IL-6 receptor inhibitors. Bold text highlights significant results. CNI: calcineurin inhibitor; IQR: interquartile range; CK: creatinine kinase; GGT: glutamyl transferase; ALP: alkaline phosphatase.

In terms of laboratory results, there was no significant difference between the maximum creatine kinase (CK) level during disease course between AMA-M2-positive and -negative subgroup [957.5 (1041.3) vs 909.5 (4163.25), P = 0.850], albeit the rate of CK elevation in the former was higher (87% vs 58.3%, P < 0.001). AMA-positive patients also had higher baseline GGT [78.0 (165.0) vs 35.0 (57.0), P < 0.001] and ALP [85.0 (130.0) vs 64.0 (31.0), P < 0.001], together with a considerably higher abnormality rate in both bile duct enzymes (64% vs 37.3%, 37% vs 12.4%, P < 0.001, respectively).

AMA-M2-positive patients exhibited remarkably higher incidence of cardiac involvement throughout disease course (60% vs 12.9%, P < 0.001) (Table 1), among which 56% (31) had arrhythmia, 44% (24) had heart failure and 31% (17) had PH, while the equivalent proportions were 5.6%, 3.6% and 7.9%, respectively, in the AMA-M2-negative group. In the AMA-M2-positive group, the most common arrhythmia subtypes were atrial/supraventricular (46%) and ventricular abnormalities (51%), and 10 of them required further intervention including pacemakers, implantable cardioverter devices or radiofrequency ablation. PH was further classified into mild, moderate and severe according to the pulmonary artery systolic pressure (Table 1). Among these 17 patients, 6 (35%) had ILD, in contrast, ILD was seen in 55/74 (74.3%) AMA-M2-negative IIM-PH patients. In addition, according to the latest guideline [10], PBC was diagnosed in 33/55 (60%) AMA-M2-positive patients. Notably, AMA-M2-positive patients with comorbid PBC (i.e. with liver enzyme abnormalities) tend to have higher incidence of cardiac involvement compared with those without [27/33 (82%) vs 8/22 (36%), P < 0.001].

In terms of treatment strategies, both groups had comparable use of steroid pulse therapy, IVIG, CYC, MMF, AZA, calcineurin inhibitors and biological agents. However, MTX use was significantly more frequent among AMA-M2-positive patients (54% vs 38%, P = 0.024).

Pathological features

Among the 55 AMA-M2-positive patients, 30 (55%) had traceable skeletal muscle biopsy, 5 had cardiac biopsy and 8 had liver biopsy. Two of the 30 patients with skeletal muscle pathology were anti-Mi2 positive and 4 were positive for anti-synthetase antibodies (2 anti-Jo1, and 2 anti-PL12), while the rest tested negative for other myositis-specific antibodies. As shown in Table 2, According to the 2017 European NeuroMuscular Centre (ENMC) primary IMNM pathology features [2], more than half (15/25) of the biopsies reported scattered necrotic fibres, 12/26 reported scattered ongoing necrosis and regeneration and 8/25 reported macrophage predominant but paucilymphocytic infiltrates. Of the samples with available immunohistochemistry 10/14 had sarcolemmal MHC-I expression, 8/14 had sarcolemmal complement deposition patches, 4/10 were reported with endomysial fibrosis and proliferation, and enlarged capillaries were seen in 3/7 patients. In total, 16/30 muscle biopsies were considered as IMNM or possible IMNM by the neuromuscular pathology department.

Table 2.

Overview of pathological examination and skeletal muscle histology features of AMA-positive idiopathic inflammatory myopathies patients.

	Muscle biopsy (30/55) of AMA-M2-positive patients
Primary IMNM pathological features
Scattered necrotic fibers	15/25
Different stages of necrosis, myophagocytosis, and regeneration	12/26
Macrophage dominant, paucilymphocytic infiltrates	8/25
Additional consistent features
Sarcolemmal MHC-I expression	10/14
Sarcolemmal complement deposition	8/14
Endomysial fibrosis and proliferation	4/10
Enlarged capillaries	3/7
Muscle pathology pattern
IMNM or possible IMNM	16/30
Non-IMNM	14/30

IMNM: immune-mediated necrotizing myositis.

Cardiac pathology of all five patients (Supplementary Table S1) suggested varying degrees in cellular structural abnormalities or degeneration. This includes cardiomyocyte hypotrophy, deformation, malalignment and degenerative changes (e.g. lipofuscin deposition). Only one patient reported lymphocytic infiltration, while the rest denied the characteristic features of myocarditis. In terms of specific staining, most patients had positive Masson stain findings, some were positive for phosphotungstic acid–haematoxylin and none was positive for Congo red staining. Meanwhile, liver biopsies demonstrated typical features of PBC, primarily at stage I or II, including portal lymphocytic infiltration, bile duct lesions and parenchymal damage (Supplementary Table S2).

Associated factors for all-cause mortality and cardiac involvement in all IIM patients

Univariate and multivariate Cox regression models were employed to identify factors associated with all-cause mortality and cardiac involvement. To control confounding, the follow-up period was capped at 120 months, and variables with >10% missing data were excluded from regression analyses.

As presented in Table 3, the univariate Cox regression revealed that older age at onset, comorbid hypertension, coronary artery disease, stroke history, elevated bile duct enzymes, IVIG use and AMA-M2 positivity were associated with increased cardiac risk. In contrast, skin manifestations such as heliotrope rash, and Shawl and V signs appeared as protective factors. However, multivariate Cox regression identified only AMA-M2 positivity [hazard ratio (HR) 3.156 (1.909, 5.218), P < 0.001] and GGT elevation [HR 1.622 (1.061, 2.479), P = 0.026] were identified as independent risk factors for cardiac involvement.

Table 3.

Univariate and multivariate Cox regression for overall cardiac involvement in idiopathic inflammatory myopathies patients.

Cardiac involvement	Univariate		Multivariate
	HR (95% CI)	P-value	HR (95% CI)	P-value
Demographics and comorbidities
Age at onset	1.011 (0.999, 1.023)	0.061	1.005 (0.990, 1.020)	0.549
Male gender	1.237 (0.886, 1.728)	0.212	0.994 (0.676, 1.463)	0.976
Coronary artery disease	2.544 (1.644, 3.936)	<0.001	1.489 (0.845, 2.622)	0.168
Hypertension	1.345 (0.949, 1.906)	0.096	0.794 (0.510, 1.237)	0.308
Stroke	2.688 (1.455, 4.966)	0.002	1.823 (0.847, 3.922)	0.125
Antibodies
AMA-M2	4.723 (3.180, 7.015)	<0.001	3.156 (1.909, 5.218)	<0.001
Clinical manifestations
PM	1.646 (1.182, 2.291)	0.003
Heliotrope rash	0.512 (0.341, 0.768)	0.001	0.609 (0.367, 1.009)	0.054
Shawl and V signs	0.590 (0.407, 0.855)	0.005	0.849 (0.535, 1.348)	0.489
Diaphragmatic muscle paralysis	1.964 (0.919, 4.197)	0.082	1.300 (0.605, 2.792)	0.502
Laboratory results
CK elevation	1.436 (1.009, 2.045)	0.045	1.039 (0.687, 1.571)	0.857
GGT elevation	2.356 (1.689, 3.285)	<0.001	1.622 (1.061, 2.479)	0.026
ALP elevation	2.890 (2.018, 4.289)	<0.001	1.605 (0.989, 2.605)	0.056
Treatment
IVIG	1.357 (0.962, 1.916)	0.082	1.326 (0.903, 1.945)	0.150

Bold text highlights significant results. CK: creatine kinase; GGT: glutamyl transferase; ALP: alkaline phosphatase; HR: hazard ratio.

Risk factors for all-cause mortality were similarly analysed and listed in Table 4. All the examined comorbidities (coronary artery disease, hypertension, stroke, malignancies and diabetes) contributed to increased mortality in IIM. Other associated factors included older age at onset, male gender, ILD, Gottron’s sign/papules, elevation in bile duct enzymes, steroid pulse therapy and IVIG use. Conversely, PM subtype, muscle weakness, CK elevation and MTX use appeared as the protective factors. Furthermore, multivariate Cox regression revealed the independent associated factors, which include older age at onset [HR 1.044, 95% CI (1.027, 1.061), P < 0.001], male gender [HR 1.726 (1.226, 2.430), P = 0.002], comorbid stroke [HR 2.281 (1.146, 4.539), P = 0.019] and malignancies [HR 3.637 (2.398, 5.515), P < 0.001], GGT elevation [HR 1.955 (1.337, 2.857), P < 0.001], used of steroid pulse therapy [HR 1.626 (1.094, 2.417), P = 0.016), IVIG [HR 1.541 (1.078, 2.205), P = 0.018] and MTX [HR 0.506 (0.335, 0.764), P < 0.001]. Notably, AMA-M2 positivity was not significantly associated with mortality in multivariate analysis.

Table 4.

Univariate and multivariate Cox regression for all-cause mortality in idiopathic inflammatory myopathies patients.

All-cause mortality
	Univariate		Multivariate
	HR (95% CI)	P-value	HR (95% CI)	P-value
Demographics and comorbidities
Age at onset	1.052 (1.039, 1.065)	<0.001	1.044 (1.027, 1.061)	<0.001
Male gender	1.647 (1.223, 2.220)	0.001	1.726 (1.226, 2.430)	0.002
Coronary artery disease	1.674 (1.070, 2.617)	0.024	0.799 (0.449, 1.421)	0.445
Hypertension	1.790 (1.318, 2.433)	<0.001	1.110 (0.765, 1.609)	0.582
Stroke	2.862 (1.658, 4.940)	<0.001	2.281 (1.146, 4.539)	0.019
Malignancies	3.464 (2.462, 4.875)	<0.001	3.637 (2.398, 5.515)	<0.001
Diabetes	1.713 (1.239, 2.367)	<0.001	0.819 (0.552, 1.216)	0.322
Antibodies
AMA-M2	0.950 (0.539, 1.676)	0.860
Ro52	1.167 (0.869, 1.567)	0.306
Jo-1	0.897 (0.563, 1.429)	0.647
PMScl	0.348 (0.086, 1.401)	0.137
Clinical manifestations
PM	0.676 (0.472, 0.969)	0.033
Gottron’s sign/papules	1.147 (0.472, 0.969)	0.033	1.056 (0.730, 1.528)	0.771
Muscle weakness	0.730 (0.538, 0.990)	0.043	0.910 (0.612, 1.353)	0.640
ILD	1.586 (1.145, 2.195)	0.005	1.154 (0.764, 1.742)	0.497
Cardiac involvement	1.346 (0.943, 1.922)	0.102	1.149 (0.747, 1.768)	0.528
Laboratory
CK elevation	0.763 (0.564, 1.031)	0.078	0.893 (0.606, 1.315)	0.566
GGT elevation	2.081 (1.528, 2.835)	<0.001	1.955 (1.337, 2.857)	0.001
ALP elevation	1.802 (1.223, 2.656)	0.003	1.435 (0.899, 2.289)	0.130
Treatment
Steroid pulse therapy	1.670 (1.208, 2.309)	0.002	1.626 (1.094, 2.417)	0.016
MTX	0.423 (0.295, 0.607)	<0.001	0.506 (0.335, 0.764)	0.001
IVIG	2.026 (1.487, 2.760)	<0.001	1.541 (1.078, 2.205)	0.018

Bold text highlights significant results. CK: creatine kinase; GGT: glutamyl transferase; ALP: alkaline phosphatase; HR: hazard ratio; ILD: interstitial lung disease.

Role of AMA-M2 in mortality and cardiac involvement of IIM patients

To further investigate the impact of AMA-M2 on mortality and cardiac involvement in IIM patients, survival analyses were performed in AMA-M2-positive and -negative subgroups. Regarding cardiac involvement, propensity score matching (PSM) was performed to adjust for baseline imbalances as described in the Methods section. Patients were 1:4 matched by the presence of comorbid coronary artery disease, hypertension and PM subtype, yielding 55 AMA-M2-positive and 202 AMA-M2-negative cases. Covariates were well-balanced after PSM adjustment (Supplementary Fig. S2 and Table S3).

As shown in Fig. 1A and B, AMA-M2 positivity was significantly associated with increased risk of cardiac involvement over time [HR 4.94 (3.36, 7.27), log-rank P < 0.001]. In the AMA-M2-positive group, the median time to cardiac involvement was 29.0 months (22 120 months), whereas the median time was not reached in the AMA-M2-negative group, indicating fewer than half developed cardiac affectations during follow-up. This association remained significant after PSM adjustment. Further stratified analyses demonstrated that AMA-M2-positive patients had remarkably worse cardiac involvement-free survival in specific cardiac complications, including arrythmia (Fig. 1C), heart failure (Fig. 1D) and PH (Fig. 1E), all with P < 0.001.

Figure 1.

Survival analyses for cardiac involvement in AMA-M2-positive and -negative patients. (A) Time to cardiac involvement in the overall cohort stratified by AMA-M2 status (log-rank P < 0.0001). (B) Cardiac involvement-free survival after 1:4 propensity score matching by comorbid coronary artery disease, hypertension and PM subtype, showing persistent statistical significance (log-rank P < 0.0001). (C–E) Subgroup Kaplan–Meier curves for specific cardiac complications including arrhythmia (C), heart failure (D) and pulmonary hypertension (E), all of which showed significantly higher cumulative incidence in the AMA-M2-positive group (log-rank P < 0.0001 for all). Shaded areas indicate 95% CIs

In contrast, AMA-M2 status did not significantly affect long-term mortality. As shown in Fig. 2A, the mean survival was 103.9 months (95% CI 95.1–112.8) for AMA-M2-positive patients and 98.0 months (95% CI 95.0–101.0) for AMA-M2-negative patients, with no statistical difference observed (P = 0.86). The 10-year overall survival was 65.1% and 71.5%, respectively. To explore survival outcomes across different myositis-specific antibody subgroups, we analysed 425 patients with complete myositis antibody profiles (Fig. 2B). Anti-MDA5-positive patients had the worst prognosis compared with those with anti-synthetase antibodies, TIF1-γ, AMA-M2 or other antibodies (P < 0.001). A detailed antibody profile is summarized in Supplementary Table S4 and Fig. S3.

Figure 2.

Survival analyses for all-cause mortality. (A) Comparison of overall survival between AMA-M2-positive and -negative IIM patients. No significant difference was observed between the two groups (log-rank P = 0.86). (B) Subgroup survival analysis based on myositis-specific and associated autoantibodies in patients with complete antibody profiles (n = 425), showing significantly poorer prognosis in anti-MDA5 positive patients compared with other subgroups, including those with AMA-M2, anti-ARS, TIF1-γ or other antibodies (log-rank P = 0.00055). Shaded areas represent 95% CIs

Discussion

To the best of our knowledge, this is the first cohort study to provide long-term prognostic information in AMA-M2-positive IIM patients. While their survival was comparable to that of AMA-M2-negative patients, AMA-M2-positive individuals were more likely to develop clinically significant cardiac involvement, with arrhythmia being the most common. Importantly, AMA-M2 positivity is identified as an independent risk factor for cardiac involvement in IIM, and might also be related to an IMNM-like pathological pattern. The prevalence of AMA-M2 was 5.6% in our cohort, consistent with previously reported range from 2.5% to 11.3% [11]. Particularly, while earlier studies assessed general AMAs, our study specifically examined the M2 subtype and reached similar results, which likely accounts for the majority of AMAs observed in IIM.

Our study confirms and extends prior reports linking AMA-M2 with cardiac affectations in IIM. While heart has previously been described in AMA-IIM [12], it was mostly evaluated as a cross-sectional feature. Our survival analysis, however, revealed cardiac involvement as a progressive complication, with a median onset of 29 months. Unlike previous studies suggesting that most manifestations were subclinical [6], we observed clinically significant events requiring interventions. Consistent with a prior multicentre Chinese study [13], AMA-M2 positivity and elevated GGT were identified as independent risk factors of cardiac involvement. Notably, patients presenting cardiac involvement as the initial manifestation were excluded due to the nature of survival analysis, suggesting that the actual incidence of cardiac involvement might still be underestimated.

Of all the cardiac involvement subtypes in IIM, PH is relatively infrequent [14]. IIM-related PH can be arterial (group 1) [15], secondary to left heart diseases (group 2) or secondary to hypo-oxygenation (group 3). However, right-heart catheterization was not routinely performed due to its invasive nature, and therefore the haemodynamic distinction between pre- and post-capillary PH was not feasible. However, the mild extent of ILD in AMA-M2-positive patients and co-existing cardiac abnormalities have led clinicians to attribute the PH primarily to left-heart dysfunction. To partially address the confounding effect of ILD, we investigated the interaction of AMA-M2 and ILD in the development of cardiac involvement, specifically PH (Supplementary Fig. S4), which confirmed that the prognostic impact of AMA-M2 on PH was consistent regardless of ILD status.

Other established manifestations of AMA-IIM, including axial muscle impairment (e.g. respiratory muscle failure, dysphagia) and digits vasculitis, were controversial across studies [6, 16, 17]. In terms of axial muscle involvements, our study found that AMA-M2 patients had more frequent diaphragm muscle paralysis but not significantly increased dysphagia. These patients also displayed a lower proportion of digital vasculitis, conflicting the belief that they are prone to vasomotor instability [12]. Hepatic involvement is also a controversial manifestation in this group of patients. Our AMA-M2-positive patients showed significantly higher proportion of elevated baseline bile duct enzymes elevation, and thus can be diagnosed as PBC according to the guideline [10]. Contrary to a previous study suggesting that liver and cardiac impairment were mutually exclusive in AMA-IIM [6], our study finds that cardiac and hepatic abnormalities tend to overlap (Table 1). These clinical discrepancies may be explained by differences in ethnicity (our cohort being primarily Asian), focus on the AMA-M2 subtype rather than general AMAs and variations in classification criteria or data collection methods.

Importantly, the observed overlap between hepatic and cardiac involvement in AMA-M2 IIM raises the possibility of shared pathogenic mechanisms. A hypothesis is that skeletal muscle fibres, cardiomyocytes and hepatocytes are in high energy demands, and are thus susceptible to mitochondrial dysfunction. The interplay between mitochondria dysfunction, oxidative stress and respiratory chain defects in IIM had been comprehensively discussed in reviews, albeit the focus was primarily IBM [18]. Another elevated cell-free mitochondrial DNA was reported in AMA-IIM [16], it lacks specificity in distinguishing IIM subtypes or correlation with disease activity. Further functional experimentation is still required in the field.

Histological findings from our study and previous reports raise the question of whether AMA-M2 represents an IMNM autoantibody or defines a distinct IIM subtype. In previous studies, necrotizing myopathy features were frequently mentioned, and most patients were directly classified as IMNM [4–6]. In our cohort, we assessed and reported each pathological entry of IMNM separately. Our results align with previous studies showing that skeletal muscle biopsies in AMA-positive IIM predominantly exhibited an IMNM-like pattern. Across traceable reports, the positive rate of primary and additional IMNM entries ranged from 32% to 71.4%, and expert evaluation suggested certain or possible IMNM diagnosis in over half of the cases. Notably, none of these patients tested positive for anti-SRP or anti-HMGCR antibodies. However, before proposing that AMA-M2 could be the third autoantibody indicative of IMNM, it is important to notice the limitations of pathological investigations, as diagnostic variability can arise from differences in staining panels, interpretation criteria and reporting formats. A recent Japanese study re-examined 201 AMA-M2 muscle samples and established a detailed scoring system [17]. The results confirmed similarities to IMNM but also key differences, including the absence of C1q deposits and perifascicular or perimysial changes. Principal component analysis further distinguished AMA-M2 IIM from IMNM in higher incidence of cardiac involvement and respiratory failure. Moreover, typical IMNM features, including statin exposure, extremely elevated CK, and anti-HMGCR or anti-SRP antibodies, were uncommon in AMA-M2-positive patients. Herein, these observations suggest that AMA-M2 IIM may involve a distinct pathogenic process that mimics IMNM histologically, and requires further investigation. Additionally, we reported for the first time the cardiac and liver biopsy features of AMA-IIM. Cardiac biopsies exhibited predominantly features of degeneration and fibrosis yet with minimal inflammation, while liver biopsies confirmed early stage PBC. Considering that cardiac and liver biopsies are not routine in IIM diagnosis, their relatively frequent use in our cohort reflected the diagnostic challenges in AMA-IIM presenting with atypical symptoms at early disease stages.

Despite frequent and severe cardiac involvement, AMA-M2 positivity does not adversely affect overall survival. This could be attributed to two factors. Firstly, the AMA-M2-negative group is largely heterogeneous and included subgroups with poorer prognosis, such as the anti-MDA5-positive patients with associated ILD, which may led to higher mortality as discussed in the subgroup analysis in Fig. 2B. Secondly, proactive management of cardiac complications, such as the use of heart failure medications, the implantation of pacemakers and implantable cardioverter defibrillators, can potentially mitigate the severity of cardiac impairment, thereby improving patient survival. However, treatment strategies of AMA-M2 IIM still lack research evidence. In our study, MTX was used more frequently in AMA-M2-positive patients, while the usage of other immunosuppressants and biologics was comparable. We identified the use of steroid pulse therapy and IVIG as independent factors with mortality. However, this result does not necessarily reflect the causality due to indication bias, where aggressive treatments are more frequently administered in severe disease. Meanwhile, cardiac involvement remains a challenging complication to manage in IIM, and can hinder successful CS tapering or DAS recovery [6]. Prior experiences of successful treatment are often limited to scattered case studies. Nonetheless, it is evident that glucocorticoid monotherapy is insufficient to address this complication, as in our centre over 90% patients received at least one additional immunomodulatory therapy throughout the disease course. Another single-centre study of 15 AMA-positive patients showed that CS alone failed to halt cardiac function deterioration, as evidenced by the changes in left ventricular ejection fraction [19]. A recent case-based review also showed that despite CS and immunosuppressive therapy, over 40% of AMA-IIM patients deteriorated in cardiac function and required electronic devices [20]. Altogether, these findings highlight the need for early detection and aggressive immunosuppressive therapy in AMA-IIM to control cardiac lesions, and larger, prospective studies are needed for further evidence.

Several limitations should be addressed in this study. Since we retrospectively verify the initial diagnosis with the 2017 EULAR/ACR criteria, in which IMNM was included in PM due to the constraints of patient recruitment number while establishing the criteria, the comparison between AMA-M2 IIM and classic IMNM was limited. To address the limitation, we analysed the patients during hospitalization and follow-up for classic IMNM antibodies and muscle histology. Specifically, we acknowledge that the prevalence of cardiac involvement in AMA-M2 patients may have been underestimated since not all patients underwent advanced assessments like cardiac magnetic resonance or 24 h ECG (Holter) which could detect potential subclinical cardiac abnormalities [21]. Also, cardiac troponin I/T data were unavailable for all patients due to non-routine testing at baseline. Another limitation of the study is the inability to separate the inflammatory or atherosclerotic origin of cardiac involvement. Although some previous studies have reported elevated coronary artery risks in IIM patients [22], incidence of ischaemic lesions in IIM-related cardiac involvement is relatively low [21]. Hence, our study documented the presence of past coronary artery disease as a comorbid condition, and coronary angiography or cardiac magnetic resonance were not routinely performed. To mitigate this limitation, propensity score matching was used to balance these baseline comorbidities to reduce the likelihood of systemic bias. Finally, as we only documented the presence or absence of a certain medication in the whole disease course, we could not evaluate treatment duration or regimen changes over time.

Our study confirmed the association between AMA-M2 positivity and long-term cardiac involvement in IIM. AMA-M2 positivity and elevated GGT levels were identified as the only independent risk factors for developing cardiac involvement in myositis patients. In terms of long-term prognosis, AMA-M2-positive IIM patients did not show a significantly worse survival compared with AMA-M2-negative counterparts. Pathological examination revealed predominantly an IMNM-like pattern in skeletal biopsies, and for the first time we reported features of cardiac and liver biopsies in AMA-M2-positive patients. These findings suggest that AMA-M2 positivity may help identify a distinct subgroup of IIM patients characterized by significant cardiac and hepatic involvement, and require further investigations into the pathogenic mechanisms and possible treatment strategies.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data underlying this article cannot be shared publicly as they contain identifiable patient information and are subject to the hospital privacy and ethical restrictions. The data will be shared on reasonable request to the corresponding author.

Funding

This study was supported by the Chinese National Key Technology R&D Program, Ministry of Science and Technology (2024YFC2510304, 2022YFC2504603), CAMS Innovation Fund for Medical Sciences (CIFMS) (2023-I2M-2–005, 2021-I2M-1–005, 2023-I2M-C&T-B-035) and National High Level Hospital Clinical Research Funding (2022-PUMCH-C-020, 2022-PUMCH-A-107).

Disclosure statement: All the authors have declared no conflicts of interest.