Prediabetes and Risk for Cardiac Death Among Patients With Coronary Artery Disease: The ARTEMIS Study

Type 2 diabetes is an epidemic health issue in Western societies and is expanding worldwide as well (1). Cardiovascular diseases—coronary artery disease (CAD) in particular—are the most common comorbid states in people with type 2 diabetes (1). Importantly, type 2 diabetes has a substantial adverse effect on the prognosis of patients with CAD (2) and recent myocardial infarction (3,4).

Prediabetes is a state of natural progression from normoglycemia to type 2 diabetes. It is defined by the presence of either impaired glucose tolerance (IGT) during an oral glucose tolerance test (OGTT) or an impaired fasting glucose (IFG) level (5). The glucose value 2 h after a load in an OGTT associates particularly well with an increased risk for mortality in population-based samples, outperforming fasting glucose and glycated hemoglobin (HbA_1c) in risk analyses and thus suggesting greater mortality risk for those with IGT than those with IFG (6–8). However, fewer data are available about the prognostic importance of prediabetes in people with a previous diagnosis of stable CAD. High plasma glucose and HbA_1c identified upon hospital admission for acute myocardial infarction have been associated with a worse prognosis in patients with type 2 diabetes (9), as have HbA_1c and 2-h postload glucose in patients without known type 2 diabetes who are hospitalized for CAD (10–13). In the European Action on Secondary and Primary Prevention by Intervention to Reduce Events (EUROASPIRE) IV registry-based survey, short-term (2 years) prognostic value of glucose markers was assessed in a more stable phase of CAD, that is, 6–36 months after hospitalization for CAD (14). The 2-h postload glucose value provided significant prognostic value for cardiovascular events independently of fasting glucose and HbA_1c, whereas 2-h postload glucose and HbA_1c were important independent risk factors for new-onset type 2 diabetes. In a limited sample of patients with CAD, however, prediabetes was not associated with increased risk for cardiovascular events after revascularization (15). Therefore, confirmatory studies are warranted to establish the long-term prognosis of patients with CAD and prediabetes who have undergone revascularization (if it was indicated) or who are receiving contemporary medical therapy (if they were not revascularized).

This study was designed to assess the relationship between prediabetes—particularly IGT—and cardiac mortality in patients with CAD after coronary angiography and, if necessary, revascularization. We hypothesized that patients with prediabetes would have a higher risk for cardiac death than would normoglycemic patients with CAD but a lower risk than those with type 2 diabetes. Secondary aims were to assess the prognostic importance of prediabetes regarding major adverse cardiac events (MACEs) and all-cause mortality.

The prospective, observational ARTEMIS (Innovation to Reduce Cardiovascular Complications of Diabetes at the Intersection) study (clinical trial reg. no. NCT1426685, www.clinicaltrials.gov) recruited patients with CAD, regardless of the presence of type 2 diabetes, 3–6 months after coronary angiography and revascularization, if indicated, or after contemporary medical treatment if revascularization was not needed (2). Patients without type 2 diabetes or prediabetes were matched at a group level for age, sex, prior myocardial infarction (non–ST-elevation or ST-elevation myocardial infarction), and revascularization with others from among a consecutive series of patients with type 2 diabetes who had undergone coronary angiography at the Division of Cardiology at Oulu University Hospital, Oulu, Finland. The initial examinations were conducted and inclusion/exclusion criteria determined at least 3 months after coronary angiography or the last revascularization. Significant CAD was defined as stenosis >50% in at least one major epicardial artery, as identified through coronary angiography.

At the initial enrollment visit, an OGTT was performed in all patients who had never received a diagnosis of type 2 diabetes in order to identify those subjects with undiagnosed type 2 diabetes, IFG, or IGT. Type 2 diabetes was defined according to the World Health Organization definition and diagnostic criteria for type 2 diabetes and prediabetes: fasting capillary plasma glucose ≥7.0 mmol/L, a 2-h postload value ≥12.2 mmol/L on the OGTT, or both. IGT was defined as fasting capillary plasma glucose <7.0 mmol/L and 2-h postload glucose ≥8.9 mmol/L but <12.2 mmol/L. IFG was defined as fasting capillary plasma glucose ≥6.1 mmol/L but <7.0 mmol/L and 2-h postload glucose <8.9 mmol/L. IGT and IFG were considered to be conditions preceding diabetes (5).

Subjects who fulfilled the guideline-based criteria for prophylactic implantation of a cardioverter/defibrillator (left ventricular [LV] ejection fraction <35%) were excluded from the study. Additionally, subjects with a life expectancy <1 year were excluded from the study. In some of them, revascularization procedures could not be performed. Both a specialist in endocrinology and diabetes care (O.H.U.) and a cardiologist (O.-P.P. or E.S.L.) optimized medical therapy for each patient during his or her qualifying visit. The study population included 834 subjects with type 2 diabetes (102 new diagnoses), 314 with IGT, 103 with IFG, and 695 with normal glucose metabolism (NG). Laboratory samples were obtained after a 12-h fast overnight through the use of standardized methods. HbA_1c is reported as millimoles per mole and percentages; we primarily use the former in the statistical analyses. Echocardiography was performed to quantify LV ejection fraction and mass, which was normalized to body surface area. Three experienced interventional cardiologists (including O.-P.P. and E.S.L) calculated a SYNTAX Score using the web-based calculator version 2.11 on the SYNTAX Score website (www.syntaxscore.com) after index angiography and revascularization had been performed. All patients included in the study gave informed consent, and the study was approved by the local ethics committee. The study complies with the tenets of the Declaration of Helsinki.

The mode of death was evaluated on the basis of death certificates, autopsy data, hospital records, and interviews with next of kin. Hospital records and data regarding resuscitation by paramedics were used to determine aborted cardiac arrest. Two independent investigators (M.J.J., H.V.H.) reviewed the cause and mode of death, and if necessary they determined these through consensus. The primary end point in this study was cardiac death or resuscitation from cardiac arrest, whichever occurred first. MACE (including cardiac death, hospitalization due to acute coronary syndrome [ACS], or hospitalization due to congestive heart failure [CHF]) and all-cause mortality were secondary end points. Study subjects were mailed a questionnaire and contacted by telephone at the 2- and 5-year follow-up to inquire about any hospitalizations and any new diagnosis of type 2 diabetes during the interim. Hospitalization due to ACS or CHF, and new type 2 diabetes, were ascertained from medical records.

One-way ANOVA or the Kruskal-Wallis test, followed by post hoc tests (the Bonferroni or the Mann-Whitney U test), were conducted for continuous variables between the study groups, depending on the distribution of the data (Gaussian if ∣skewness∣ <1); these were corrected for multiple comparisons. The χ² test was used for categorical variables. Univariate Cox regression was performed, followed by adjustment for predefined covariates (age, sex, BMI, resting systolic and diastolic blood pressure, Canadian Cardiovascular Society [CCS] grading for angina pectoris, SYNTAX Score, and LV ejection fraction) (2). Additional Cox regression was performed to assess risk for new type 2 diabetes among patients without type 2 diabetes at baseline; this regression was adjusted for age, sex, BMI, and waist-to-hip ratio. Using Cox regression, we studied interactions between sex and glycemic status in the risk evaluations. We used Kaplan-Meier analysis to illustrate survival curves for groups with type 2 diabetes, IGT, IFG, and NG. The data were analyzed using SPSS Statistics software (version 21; IBM Corp. ). A P value <0.05 was considered statistically significant.

Baseline characteristics of the study groups are presented in Table 1. During the mean follow-up of 6.3 years (SD 1.6 years), cardiac mortality was 8.2% for those with type 2 diabetes, 3.8% for the IGT group, 2.9% for the IFG group, and 2.6% for the NG group. The IGT group had higher all-cause mortality than the NG group and lower incidence of MACEs than the patients with type 2 diabetes. The IFG group had lower cardiac and all-cause mortality and a lower incidence of MACEs than the patients with type 2 diabetes (Table 1).

In the Cox regression analyses, the risks for cardiac death, hospitalization due to ACS or CHF, and MACE did not differ between the IGT, IFG, and NG groups (Table 2 and Fig. 1). The IGT group had a higher univariate risk for all-cause mortality than did the NG group; this did not remain significant after adjustments. The IGT group had significantly lower univariate risks for cardiac death, hospitalization due to CHF, and MACE than did the type 2 diabetes group; the adjusted risk for MACE was significant, and adjusted risk for cardiac death (P = 0.069) and death (P = 0.076) approached significance (Table 2). The IFG group had lower risks for mortality and MACE than did the patients with type 2 diabetes. The mortality risk remained significant after adjustments, and we observed tendency in the adjusted risk for MACE (P = 0.070). When the IFG and IGT groups were pooled, the adjusted risks for cardiac death, MACE, and all-cause mortality did not differ between the IFG/IGT and NG groups. The IFG/IGT group had significantly lower adjusted risks for cardiac death (hazard ratio [HR] 0.44; 95% CI 0.22–0.89; P = 0.021), MACE (HR 0.63; 95% CI 0.46–0.86; P = 0.003), and all-cause mortality (HR 0.57; 95% CI 0.37–0.86; P = 0.008) than did patients with type 2 diabetes. We observed no significant sex∗glycemic status interactions. Except for an association between 2-h postload glucose and all-cause mortality (HR 1.18 per unit increase; 95% CI 1.03–1.35; P = 0.014), fasting glucose, 2-h postload glucose, and HbA_1c were not associated with these outcomes in the multivariate Cox regression, in which they were assessed as continuous variables in the patients without type 2 diabetes at baseline.

Among the patients with newly diagnosed type 2 diabetes at baseline (n = 102), we counted 6 cardiac deaths (5.9%), 12 hospitalizations for ACS (14.5%), 6 hospitalizations for CHF (7.1%), 21 MACEs (20.6%), and 13 deaths (12.7%). In the univariate analysis, newly diagnosed type 2 diabetes involved a 3.0-fold risk for CHF (95% CI 1.2–7.7; P = 0.024) and a 2.2-fold risk for all-cause mortality (95% CI 1.8–3.6; P = 0.014) compared to patients with NG, but this was not independent of covariates. The risks related to newly diagnosed type 2 diabetes did not differ from those for the patients with IFG, IGT, or previously diagnosed type 2 diabetes.

The incidence of new type 2 diabetes was higher in the IGT and IFG groups than in the NG group (P < 0.001 for both) (Table 1). The adjusted HRs were 2.78 (95% CI 1.50–5.16; P = 0.001) for the IGT group and 2.68 (95% CI 1.19–6.03; P = 0.017) for the IFG group; thus, these groups did not differ in this respect. In univariate Cox regression, new onset of type 2 diabetes involved a 2.4-fold higher risk for ACS (95% CI 1.4–4.1; P = 0.003; 109 events), a 4.6-fold higher risk for CHF (95% CI 1.5–13.9; P = 0.007; 18 events), and a 2.2-fold higher risk for MACE (95% CI 1.3–3.8; P = 0.003; 123 events) than in those who remained free of type 2 diabetes. Patients with new onset of type 2 diabetes tended to have a 2.9-fold higher univariate risk for all-cause mortality than those who did not develop type 2 diabetes (95% CI 1.0–8.6; P = 0.054; n = 22) but not for cardiac mortality (HR 1.9; 95% CI 0.2–15.5; P = 0.548). A limited number of cardiac deaths was observed among patients for whom follow-up data were available for the incidence of new type 2 diabetes (one event [1.9%] in patients with new onset of type 2 diabetes [n = 52] and seven events [0.9%] in patients without type 2 diabetes [n = 746]).

The main finding in this study is that the risk of cardiac events does not differ between patients with CAD and prediabetes who underwent revascularization or optimal medical therapy and those with NG status; this risk is substantially lower among these groups than among patients with CAD and type 2 diabetes. This finding suggests that that prediabetes does not increase the risks for cardiac death and major cardiac morbidities in the current treatment era. New onset of type 2 diabetes during the 5-year follow-up period was, however, more likely in patients who had prediabetes at baseline. That development worsened the prognosis of patients with CAD without prior type 2 diabetes. Thus, the progression from prediabetes to type 2 diabetes warrants effective countermeasures that can improve the prognosis for patients with CAD.

While the detrimental effects of type 2 diabetes on prognosis are well established in people with stable CAD (2) and those with recent myocardial infarction (3,4), less is known about the prognostic value of prediabetes in patients with CAD who have been treated according to current guidelines, such as with revascularization (if it is deemed necessary). Plasma glucose and HbA_1c at admission and, to a lesser extent, 2-h postload glucose have been used as predictive markers and have shown associations with cardiac events (9–13,15). Notably, 2-h postload glucose—unlike fasting glucose—seems to provide additional prognostic value to the Global Registry of Acute Coronary Events (GRACE) score for cardiac events, an established risk model (13). Our findings do not support our primary hypothesis that, on the basis of previous studies of population-based samples and the EUROASPIRE IV survey (6–8,14), prediabetes—particularly IGT—is associated with higher cardiac mortality and MACE. A glucose metabolism disorder suggesting prediabetes may not accelerate cardiac disease–associated event expression when criteria for overt type 2 diabetes are not fulfilled. Our study population consisted of subjects with known CAD at baseline, which limits this conclusion in that it does not permit us to extrapolate the findings to a possible role for prediabetes as a risk marker for CAD development. Contradictory results between this study and the EUROASPIRE IV multicenter survey may not be fully explained. The strength of this study, however, is that it was a prospective, observational study with considerably longer follow-up than previous studies. Although revascularization characteristics of these populations seem to be comparable, patients with advanced cardiac disease (LV ejection fraction <35% or life expectancy <1 year) were excluded from this study, which may explain the differences. Also, modest differences in medical treatment may play a role.

Moreover, increased risk for cardiac events was observed among patients who received a diagnosis of type 2 diabetes during the follow-up. Because prediabetes is a phase in the progression from normoglycemia to type 2 diabetes, it is to be expected that the incidence of type 2 diabetes would be higher in patients with IFG or IGT than in those with NG. However, this did not occur to an extent that would worsen the prognosis of patients in the IFG and IGT groups. Even higher age, BMI, and prevalence of CAD symptoms (CCS class) for the patients with IGT did not imply greater risk for cardiac mortality. However, many patients with CAD with an initially normal glycemic status of prediabetes may have developed type 2 diabetes during the follow-up, but we did not formally assess OGTT and HbA_1c during that time. The prevalence of undiagnosed type 2 diabetes may be up to 30% (16). These aspects probably explain the lower 5-year incidence of new type 2 diabetes among patients with prediabetes (11.4%) in our study than in, for example, the Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research (NAVIGATOR) trial (33.9%) (17).

When glucose metabolism markers were analyzed as continuous variables in patients without type 2 diabetes, this study partially confirmed the prognostic value of 2-h postload glucose, as reported from the EUROASPIRE IV trial (14): 2-h postload glucose showed an independent association with all-cause mortality. Also, we found univariate associations of fasting and 2-h postload glucose values with cardiac death, and of fasting glucose and HbA_1c, with hospitalization due to CHF. However, these observations did not remain significant after we adjusted for potential confounders. Notably, none of glucose metabolism markers were associated with MACE or ACS (the main MACE). Therefore, it seems that variation of glucose metabolism markers for type 2 diabetes within the subclinical range may not be an important risk factor for ACS in patients with stable CAD. This observation contradicts data from a meta-analysis by Liu et al. (10), who reported the prognostic importance of HbA_1c, particularly among patients with CAD without type 2 diabetes, though not among patients with CAD and type 2 diabetes. The univariate association of glucose metabolism markers with hospitalization due to CHF and cardiac death could suggest that they might contribute to progression of CHF, which is a potent association for increased mortality in patients with CAD and type 2 diabetes (2,18). Nonetheless, the lack of an independent association of glucose metabolism markers with the cardiac outcomes underscores the importance of the clinical variables we included in the multivariate analyses.

These results suggest that prediabetes in established CAD is not associated with increased risk for cardiac events and does not have a predictive value similar to that of type 2 diabetes in patients with CAD. In the present study, this can particularly be seen in the difference in risk for interim heart failure between patients with prediabetes and those with type 2 diabetes. The results are reassuring because increasing numbers of patients with CAD also have prediabetes. Preventive efforts should be made to impede the progression of prediabetes to type 2 diabetes. Exercise and dietary interventions after a diagnosis of CAD are potentially relevant (1,19).

This study is partly limited by the small number of end points, particularly regarding cardiac death among patients without type 2 diabetes. We recently reported that in patients with CAD but without type 2 diabetes, cardiac mortality—specifically incidence of sudden cardiac death—is almost equal to that observed at the population level (2). Therefore, we conducted substantial analyses of secondary fatal and nonfatal end points to verify the observations from this study regarding cardiac mortality. Although follow-up data for nonfatal end points are not available for all patients at this point, they are considerable enough to complement the main findings. At the time of this study, patients did not commonly use sodium–glucose cotransporter 2 or dipeptidyl peptidase-4 inhibitors and glucagon-like peptide 1 analogs; thus, we presume that the newer medications have little effect on these findings. However, these results may not fully represent the current differences in prognosis between patients with IFG or IGT and those with type 2 diabetes, whereas the results may correspond well to current clinical practice when treating the IFG/IGT and the NG groups. Finally, this study is limited by the lack of formal and consistent OGTT during follow-up for new onset of type 2 diabetes, IFG, or IGT and the lack of information about the history of IFG and IGT.

In conclusion, the risk of cardiac events among patients with CAD and prediabetes is comparable to that among patients with CAD and NG status and is lower than that in those with type 2 diabetes when CAD is treated by revascularization, optimal medical therapy, or both.

Acknowledgments. The authors gratefully acknowledge the assistance of registered nurses Pirkko Huikuri, Päivi Koski, Päivi Kastell, and Sari Kaarlenkaski at the Research Unit of Internal Medicine, Oulu University Hospital and University of Oulu.

Funding. This study was funded by the Finnish Technology Development Center (Tekes), Helsinki, Finland; the Academy of Finland (Research Council for Health), Helsinki, Finland (grant no. 267435); the Finnish Foundation for Cardiovascular Research, Helsinki, Finland; and the Paulo Foundation, Espoo, Finland. The authors appreciate the financial support received from the ARTEMIS consortium partners (Polar Electro, Kempele, Finland, and Hur Oy, Kokkola, Finland). R.J.M. is supported in part by the American Heart Association Chair in Cardiovascular Research at the University of Miami.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.M.K., E.S.L., O.-P.P., T.V.K., R.J.M., and M.J.J. researched data or contributed to the analysis. A.M.K., M.P.T., J.S.P., O.H.U., M.J.J., and H.V.H. designed the study. A.M.K. and M.J.J. wrote the manuscript. A.M.K., E.S.L., M.P.T., O.-P.P., T.V.K., J.S.P., O.H.U., R.J.M., M.J.J., and H.V.H. reviewed and edited the manuscript and contributed to the discussion. A.M.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.