Cardiac Autoimmunity is Associated with Subclinical Myocardial Dysfunction in Patients with Type 1 Diabetes

The data and analytic methods will be/have been made available to other researchers through the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Central Repository, with biosamples subject to approval by the NIDDK.

The existence of a ‘diabetic cardiomyopathy’ has been postulated to explain the increased risk of heart failure (HF) in patients with diabetes; however, whether mechanisms specific to type 1 diabetes (T1D) affect the myocardium is unclear. We have recently shown that poor glycemic control in patients with T1D followed in the DCCT (Diabetes Control and Complications Trial) – but not in patients with type 2 diabetes (T2D) – is associated with positivity for multiple (≥2) cardiac autoantibodies (AAbs), similar to a HF cohort with Chagas cardiomyopathy,¹ raising the possibility of autoimmune-associated myocardial dysfunction in T1D.

To test this hypothesis, we measured AAbs to 5 myocardial antigens¹ at the time of cardiac magnetic resonance (CMR) during the post-DCCT EDIC (Epidemiology of Diabetes Interventions and Complications) study in 892 (of 950) subjects^{2, 3} without prior CVD events (acute MI, silent MI, stroke, confirmed angina, coronary revascularization, or HF), in whom samples were available. De-identified samples and datasets were obtained from the NIDDK Repository, and were deemed not to constitute human subjects research by the Committee on Human Studies at Joslin Diabetes Center.

The mean age was 49 years and T1D duration was 28 years (Table). Cardiac AAbs were detected in 146 (16%) participants, with 102 (11%) positive for 1 AAb, and 44 (5%) positive for ≥2 AAbs. Clinical characteristics of subjects with 1 AAb were similar to those with ≥2 AAbs and compared to those without AAbs, AAb-positive subjects were more likely to have been in the conventional glucose-lowering treatment group during DCCT, have microalbuminuria, macroalbuminuria, and higher DCCT/EDIC time-weighted mean HbA_1c and systolic blood pressure levels (Table), suggesting that cardiac AAbs are markers of long-term glycemic exposure.

Table.

Association between Cardiac Autoantibodies, Clinical Characteristics and CMR Indices of LV Structure and Function

	Cardiac Autoantibody (AAb) Status			P Value for Trend*
	No AAbs (n=746)	1 AAb (n=102)	≥2 AAbs (n=44)
Clinical characteristics
Male sex	380 (51)	56 (55)	27 (61)	0.17
Primary prevention cohort	385 (52)	59 (58)	27 (61)	0.10
Conventional DCCT treatment	347 (47)	62 (61)	24 (55)	0.008‡
Age at CMR exam, y	49±7	50±7	48±7	0.56
Body-mass index, kg/m2	28±5	27±4	28±4	0.30
Diabetes duration, y	28±5	28±5	29±5	0.08
Recent HbA1c, %	7.8±1.2	8.1±1.5	7.7±1.3	0.51
Smoking	87 (12)	11 (11)	5 (11)	0.82
Alcohol use	338 (45)	41 (40)	20 (45)	0.46
Treated hypertension	357 (48)	48 (47)	23 (52)	0.83
Treated hypercholesterolemia	463 (62)	64 (63)	31 (70)	0.45
Proliferative diabetic retinopathy	138 (19)	23 (23)	9 (20)	0.35
Cardiac autonomic neuropathy||	200 (28)	38 (40)	10 (25)	0.10
Diabetic nephropathy
Sustained microalbuminuria/ESRD	169 (23)	35 (34)	15 (34)	0.003‡
Macroalbuminuria/ESRD	58 (8)	15 (15)	6 (14)	0.01†
DCCT/EDIC time-weighted mean variables#
HbA1c, %	7.9±0.9	8.1±1.0	8.1±1.2	0.02‡
HDL cholesterol, mg/dl	55±13	53±13	54±11	0.23
LDL cholesterol, mg/dl	109±20	110±23	113±17	0.17
Triglycerides, mg/dl	82±39	86±47	88±40	0.17
Systolic blood pressure, mm Hg	117±8	119±8	119±8	0.02†
Diastolic blood pressure, mm Hg	74±5	75±5	75±5	0.05
CMR indices of LV structure and function
Minimally adjusted**
End-diastolic volume, mL	136±1	133±2	151±3	<0.001§
End-systolic volume, mL	52±0	50±1	69±2	<0.001§
Stroke volume, mL	84±1	83±1	82±2	0.59
Ejection fraction, %	62±0	62±1	56±1	<0.001§
Cardiac output, L/min	5.9±0.0	6.0±0.1	5.8±0.2	0.52
LV mass, gm	137±1	137±2	151±3	<0.001§
Fully adjusted***
End-diastolic volume, mL	136±1	133±2	151±3	<0.001§
End-systolic volume, mL	52±1	50±1	69±2	<0.001§
Stroke volume, mL	84±1	83±1	82±2	0.65
Ejection fraction, %	62±0	62±1	55±1	<0.001§
Cardiac output, L/min	5.9±0.0	6.0±0.1	5.8±0.2	0.63
LV mass, gm	137±1	135±2	149±3	<0.001§

Values are: n (%) or mean ± SD (clinical characteristics) or least square mean ± SE (CMR indices). CVD, Cardiovascular disease; CMR, cardiac magnetic resonance; HbA_1c, glycated hemoglobin; ESRD, end-stage renal disease; HDL, high-density lipoprotein; LDL, low-density lipoprotein; LV, left ventricular.

^*P values for trend were calculated using the Jonckheere-Terpstra trend test and corrected for multiple testing using the post hoc Tukey-Kramer method or the Benjamini-Hochberg procedure.

The most significant pairwise comparisons were:

^†P<0.05 for AAb-positive (1 AAb or ≥ 2 AAbs) vs. no AAbs

^‡P<0.01 for AAb-positive vs. no AAbs

^§P<0.001 for ≥2 AAbs vs. 1 AAb or no AAbs

^||n=723, 96, and 40 for no AAbs, 1 AAb, and ≥2 AAbs, respectively.

^#The term “time-weighted mean” refers to variables that were measured from DCCT baseline through the EDIC year prior to CMR exams, and were calculated as previously described.²

^**Minimal adjustment includes: age at CMR exam, sex, body surface area (BSA), and machine type.

^***Full adjustment includes the variables in the minimally adjusted model along with the DCCT cohort (primary prevention/secondary intervention), smoking, alcohol use, macroalbuminuria/ESDR, and DCCT/EDIC time-weighted mean systolic blood pressure, HDL, LDL, and HbA_1c.² Cardiac autonomic neuropathy, microalbuminuria, macroalbuminuria, and ESDR are as previously defined.^2,3

In minimally adjusted models, subjects with 1 AAb had normal CMR indices, similar to those without AAbs (Table). However, subjects with ≥2 AAbs had markedly greater left ventricular (LV) end-diastolic volume (EDV), end-systolic volume (ESV), LV mass, and lower LV ejection fraction (EF). In multivariable models adjusting for the 8 additional covariates shown to affect CMR parameters in EDIC participants (Table),² the ≥2 AAb group continued to show greater LV volumes, greater LV mass, and lower EF, with minimal or no attenuation after adjustment (P<0.001 for all parameters). Adding positivity for ≥2 AAbs to the fully adjusted model was associated with increased EDV (+15.8 mL, 95% CI, 9.3–22.4), ESV (+17.4 mL, 95% CI, 13.5–21.3), LV mass (+12.4 gm, 95% CI, 6.3–18.5), and lower EF (−6.6%, 95% CI, −4.8 to −8.4). In contrast, HbA_1c was associated with minimally decreased EDV (−2.5 mL, 95% CI, −0.9 to −4.2), with no significant changes in ESV, EF, or LV mass. Cardiac AAbs were not associated with myocardial scar, with 28% (5/18) of subjects with scar having AAbs compared with 16% (105/638) without scar (P=0.20).

This study demonstrates that cardiac autoimmunity, as evidenced by ≥2 cardiac AAbs, is associated with increased LV chamber dimensions, increased LV mass, and lower EF. This dilated (“eccentric”) LV remodeling phenotype is different from the concentric remodeling previously reported in this cohort in association with HbA_1c and macroalbuminuria,^{2, 3} that is also observed in T2D, but is consistent with the type of remodeling found in other conditions associated with cardiac AAbs.^{1, 4, 5}

In previous DCCT/EDIC studies, the relationship between HbA_1c and CMR indices was also modest,^{2, 3} calling into question the extent to which T1D affects the myocardium.³ However, stratifying subjects by cardiac AAb status, we found evidence of a dilated remodeling phenotype suggesting a potentially novel pathophysiological process associated with cardiac autoimmunity in T1D. Consistent with this notion, although lower EF was not associated with traditional CVD risk factors in the present or previous studies,^{2, 3} we found a significant association with cardiac autoimmunity.

We believe these findings are unlikely due to unrecognized MI,⁵ given the rigorous ascertainment of MI during DCCT/EDIC^{2, 3} and lack of association between cardiac AAbs and scar. However, it is possible that preexisting damage to the myocardium may induce cardiac AAbs; in addition, eccentric remodeling is seen in athletic/trained hearts. Regardless, studies in animal models⁴ and patients with myocarditis suggest that ≥2 AAbs are markers of a disease process that is primarily T cell-mediated.^{1, 5} Indeed, if cardiac AAbs were pathogenic, subjects with 1 AAb should have had ‘intermediate’ levels of CMR abnormalities, but this was not observed.

Limitations of this study include the small number of first HF events (n=4), which precluded assessment of HF risk associated with cardiac AAbs, and lack of biomarkers relevant to HF. Furthermore, the cross-sectional study design limits interpretation of a causal relationship between cardiac autoimmunity and CMR abnormalities.

In conclusion, in this large cohort without prior CVD events, cardiac autoimmunity was associated with subclinical myocardial dysfunction independent of traditional CVD risk factors, suggesting a novel process linked to long-term glycemic exposure in T1D. Whether these subjects are at higher risk of future HF will require longer follow-up.