Association of Silent Myocardial Infarction With Major Cardiovascular Events in Diabetes: The ACCORD Trial

The epidemic of diabetes continues to grow, with recent prevalence estimates of 12% among U.S. adults. Though type 2 diabetes is associated with a two- to fourfold heightened risk of cardiovascular disease, this risk is heterogeneously distributed (1). Thus, markers of cardiovascular risk beyond traditional risk factors are needed. Prior efforts to incorporate novel blood biomarkers in risk prediction have found only modest additive value in risk discrimination, despite significant economic cost (2,3). The routine 12-lead electrocardiogram (ECG) may represent a cost-effective measure to refine atherosclerotic cardiovascular disease (ASCVD) risk. Silent myocardial infarction (SMI) is more common among those with diabetes due to impaired nociception. The risk of ASCVD among those with prior SMI and diabetes has not been adequately studied.

We used patient-level data from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study and the ACCORDION follow-up study to assess the additive utility of SMI for discrimination of risk. For this analysis, eligible participants were free of prevalent cardiovascular disease (defined as history of myocardial infarction, stroke, coronary revascularization, carotid or peripheral revascularization, or positive stress test) at study enrollment and had adequate baseline ECG data. SMI was defined as the presence of a major Q-wave abnormality or minor Q/QS waves in the setting of major ST-T abnormalities in the absence of history of clinical cardiovascular disease. Cox proportional hazards modeling was used to explore the association between cardiovascular events and baseline SMI. The improvement in discrimination with inclusion of SMI in risk models was assessed by comparing area under the receiver operating characteristic curves and the net reclassification improvement (NRI).

Among 5,539 eligible participants (mean age 62.8 ± 5.8 years, 45% women, 61% white), 5,246 (94.7%) did not have SMI at baseline, while 293 (5.3%) had SMI. ACCORD participants were enrolled between 2001 and 2005 and followed for a median of 3.5 years as part of the ACCORD study and then followed as part of the ACCORDION follow-up study for a grand total of 9.3 ± 2.2 years. After a combined 51,654 person-years of follow-up, there were 1,902 events (Table 1). In a multivariable-adjusted Cox proportional hazards model, the presence of SMI at baseline was associated with increased risks of all-cause mortality, cardiovascular mortality, congestive heart failure, and major coronary heart disease events. There was no evidence of effect modification by age, sex, or race. When models with and without SMI were compared by area under the receiver operating characteristic curves and with the NRI, inclusion of SMI yielded significant improvements in discrimination.

We found that SMI was associated with cardiovascular events and that inclusion of SMI in risk models improved discrimination. In the context of the literature to date on SMI in diabetes, we found a lower prevalence of SMI (5.3%) than many prior studies. This may be explained by the heightened awareness of the symptoms of cardiac ischemia that has come from public health education campaigns, such that a greater fraction of all patients who experience ischemic pain present to medical attention now than did previously, leading to both increased recognition of myocardial infarction and more opportunities for early treatment. With regard to risk discrimination, the largest prior study found no improvement in discrimination when SMI was added to the variables already included in the UK Prospective Diabetes Study (UKPDS) Risk Engine (4). In contrast, we found improved discrimination with the addition of SMI. This may be explained by our extended follow-up duration and a sufficiently large sample size (5,539 in ACCORD vs. 1,967 in the UKPDS SMI study) to support multivariable adjustment, allowing us to better characterize the risks attributable to SMI.

These findings suggest that an incidental electrocardiographic finding of a Q-wave may have value in refining risk, which has important treatment implications. For instance, the 2019 American College of Cardiology/American Heart Association guidelines recommend moderate-intensity statin therapy in those with diabetes but also recommend risk estimation to guide consideration of high-intensity statin (5). If SMI is validated as a marker of heightened cardiovascular risk, it may merit inclusion in risk estimation engines specific to type 2 diabetes. In addition, with increased use of glucagon-like peptide 1 agonists and sodium–glucose cotransporter 2 inhibitors for residual risk reduction among patients already on statins, reclassifying a patient from primary prevention to secondary prevention confers eligibility for more of these drugs, so considering SMI a risk equivalent for prevalent cardiovascular disease may have profound implications on treatment.

In summary, we found that SMI is associated with cardiovascular events and improves risk discrimination over and beyond traditional risk factors in diabetes. Inclusion of SMI in risk prediction models may allow identification of individuals most likely to benefit from aggressive treatment for the prevention of ASCVD. A serendipitous finding of a pathological Q-wave on ECG may serve as a marker that suggests a higher-risk patient population, which may benefit from consideration of therapies to mitigate risk.