Newly Discovered Abnormal Glucose Tolerance in Patients With Acute Myocardial Infarction and Cardiovascular Outcomes: A Meta-analysis

Type 2 diabetes is a cardiometabolic disease characterized by defects in insulin secretion and insulin sensitivity (1) and the development of both microvascular and macrovascular complications. Diabetic retinopathy, nephropathy, and neuropathy are strongly related to both the severity and the duration of hyperglycemia and are associated with considerable morbidity (2). However, cardiovascular (CV) disease is the leading cause of mortality. Individuals with type 2 diabetes are at high risk for atherosclerotic CV disease (ASCVD), including myocardial infarction (MI), stroke, and CV death, as well as heart failure (3). Hypertension, dyslipidemia, metabolic syndrome, obesity (especially visceral), endothelial dysfunction, procoagulant state, glycated hemoglobin (HbA_1c) level, and insulin resistance represent important risk factors for ASCVD in type 2 diabetes (4).

The worldwide prevalence of diabetes has increased progressively over the past 50 years. In 2017, 425 million people had type 2 diabetes (5). It is estimated that another 300 million are at increased risk for developing type 2 diabetes and CV diseases, including individuals with impaired fasting glucose (IFG), impaired glucose tolerance (IGT), and insulin resistance (5).

Although the incidence of MI has declined significantly over the past decade, from 1987 to 1996 and from 2003 to 2009, the incidence of coronary heart disease declined from 3.9 to 2.2 per 1,000 person-years in people without diabetes and from 11.1 to 5.4 per 1,000 person-years among those with diabetes, and CV disease remains the leading cause of death in the U.S. The overall prevalence for MI is 3.0% in U.S. adults ≥20 years of age. Approximately every 40 s, an American will have an MI. In 2019, coronary events were expected to occur in ∼1,055,000 individuals, including 720,000 new and 335,000 recurrent coronary events (6).

Prediabetes is a condition that reflects abnormal glucose metabolism and is associated with an increased risk of CV disease (7). Previously undiagnosed diabetes and impaired glucose tolerance are common in patients with acute MI. A 75-g oral glucose tolerance test (OGTT) performed within 1 week after the event reveals that as many as 50% have previously unknown abnormal glucose tolerance (AGT) on the basis of fasting plasma glucose (FPG), 2-h glucose during OGTT, or HbA_1c (8). However, the incidence of recurrent CV events in people with AGT is incompletely understood. A better appreciation of the incidence and significance of AGT in patients with acute MI without a history of abnormal glucose metabolism could enhance awareness about these associated conditions and lead to more systematic screening. Also, early detection of individuals with prediabetes/diabetes among those with acute coronary syndrome (ACS) and aggressive secondary risk reduction strategies could markedly reduce the rate of recurrent ASCVD complications (9). The present meta-analysis was undertaken to 1) examine the prevalence of AGT in individuals presenting with acute MI who were not previously known to have prediabetes/diabetes and 2) examine the impact of newly discovered AGT on subsequent major cardiovascular adverse events (MACE) (a composite of nonfatal MI, nonfatal stroke, and CV death), all-cause mortality, and hospitalization for heart failure.

We searched electronic databases (Medline, Embase, Cochrane Library, and Google Scholar) for articles published in English up to 30 November 2018 using a combined Medical Subject Heading and text search strategy with the following terms: “hyperglycemia,” “MACE,” “mortality,” “undiagnosed,” “newly diagnosed,” “diabetes,” “diabetes mellitus,” “myocardial infarction,” and “heart failure.” The search strategy aimed to include all articles concerning newly diagnosed AGT and CV outcomes. Three main search domains were combined with the Boolean operator AND. Search terms contained within each domain were combined with the Boolean operator OR. The first search domain included “hyperglycemia,” “abnormal glucose tolerance,” “diabetes,” “prediabetes,” “myocardial infarction,” and “acute coronary syndrome.” The second domain included “undiagnosed” and “newly diagnosed.” The third search domain included mortality,” “death,” “MACE,” “heart failure,” and “hospitalization.” We reviewed all abstracts obtained from our search. We also manually reviewed reference lists and review articles for additional citations and obtained the full text of all potentially relevant publications. Only full articles were considered for data extraction; conference abstracts were not included in the analysis. The present systematic review and meta-analysis was conducted and reported according to the recommendation of the Meta-analysis of Observational Studies in Epidemiology reporting guideline (Supplementary Table 5).

Our prespecified inclusion criteria were 1) cohort studies in patients ≥18 years of age with acute MI and without known previous history of diabetes or AGT; 2) studies that provided FPG, HbA_1c, or 2-h plasma glucose during OGTT at baseline; and 3) studies that provided the incidence of MACE, CV death, heart failure, or all-cause mortality in patients with newly discovered AGT compared with those with normal glucose tolerance (NGT). We excluded studies if they 1) had no original data, 2) did not include people with newly discovered AGT, 3) involved nonprospective (i.e., cross-sectional and retrospective case control) studies, or 4) had follow-up duration <6 months. When published studies were performed on the same sample, we extracted data from each report separately and then collated information from the data collection forms afterward. The study with more information and a longer follow-up period was selected as the main data source.

Two investigators independently reviewed each article that met selection criteria and extracted the data. Discrepancies were resolved by consensus. For each study that met our inclusion criteria, we extracted the number of participants in each group, geographic data of the study population, types of MI, duration of follow-up, method of measurement of AGT status, number of CV events, and adjusted effect estimates (odds ratio, relative risks, or relative hazards) for the association between CV risk and baseline glucose tolerance status.

Individuals with AGT were divided into the following groups: IFG, IGT, and newly diagnosed diabetes on the basis of American Diabetes Association 2010 criteria (10). SEs for the estimates were derived using the published data. If this information was not provided, we calculated the effect estimates and SEs from the raw data.

Publication bias was assessed by inspection of the funnel plots. The risk of bias assessment for each study was evaluated by the Newcastle-Ottawa Scale (11) and summarized in Supplementary Table 3. To be considered a high-quality study, three or four stars (score) in the selection domain and one or two stars in the comparability domain and two or three stars in the outcome domain were required.

All combined effect sizes of pooled prevalence and 95% CIs were estimated using inverse variance random-effects models (12). Incidence rate ratios were used to compare the incidence of mortality and MACE between NGT and AGT (prediabetes and newly discovered diabetes) groups. Incidence rate ratio, pooled prevalence of NGT, prediabetes, and diabetes were calculated using Meta-Essentials (13). Hazard ratios (HRs) for the outcomes, corresponding SEs, and 95% CIs were calculated using the Mantel-Haenszel approach (using Review Manager version 5.3; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark). Heterogeneity among studies was assessed with the Cochran Q test and the I² statistic (values of 25%, 50%, and 75% represent low, medium, and high heterogeneity, respectively). Even when heterogeneity was low or absent, we report the results from random-effects over fixed-effects models because the included studies differed to some extent both clinically and methodologically.

We conducted a sensitivity analysis for assessing unmeasured confounding by calculating an E value for each analysis (14). The lowest possible E value is 1, which means no unmeasured confounder to explain the observed association. The higher the E value is, the stronger the confounder to explain the risk ratio. The E value for the limit of CI is the effect of the confounder that would move the CI limit to 1.

Our initial search identified 3,711 articles. After screening titles and abstracts, 66 articles qualified for a full review. After applying all inclusion/exclusion criteria, 19 prospective cohort studies, comprising 41,509 individuals, were included for analysis (Fig. 1).

The studies were geographically heterogeneous, with sample sizes ranging from 160 to >10,000 subjects. All studies described basic inclusion/exclusion criteria for study participants. From 19 studies, 10 (n = 21,007) provided details regarding classification of acute MI. Of those, 58.8% were classified as ST-elevation MI and the others were classified as non–ST-elevation MI. The proportion of ST-elevation ACS and non–ST-elevation ACS were balanced between the NGT and AGT groups. From 19 included studies, 12 (n = 22,632) provided percutaneous coronary intervention data. Of those, 84.8% of the NGT group and 85.8% of the AGT group received primary percutaneous coronary intervention.

The method for glucose tolerance assessment varied across studies. OGTT was used to classify glucose tolerance status in the majority of studies (58%). Most studies reported the incidence of IGT or combination of IFG and IGT. IFG and IGT were combined to comprise the prediabetes group (15–33). Table 1 summarizes characteristics and main results of all studies in our analysis. For assessment of CV outcomes, studies used standard methods, such as reviews of medical records or death certificates. Most studies reported crude event rates, so we calculated the effect size on the basis of crude data (Table 1). The pooled prevalence of newly discovered AGT in subjects with acute MI was 48.4% (95% CI 40.2–56.6). Pooled prevalence of prediabetes and newly diagnosed diabetes was 35.4% (28.9–42.0) and 19.7% (16.0–23.4%), respectively (Table 1 and Supplementary Fig. 1A–C). The pooled prevalence of AGT was very similar across the study period (Supplementary Table 1).

There was a strong association between AGT and subsequent all-cause mortality, recurrent MACE, CV death, and hospitalization for heart failure (Fig. 2A–D). For the association between all-cause mortality and newly discovered AGT, 16 studies provided data. The AGT group had a higher all-cause mortality risk compared with the NGT group (HR 1.51 [95% CI 1.34–1.70], P < 0.001). For the association between MACE and newly discovered AGT, 11 studies provided data. MACE risk was greater in patients with AGT than in those with NGT (1.44 [1.23–1.68], P < 0.001). For the association between CV death and newly discovered AGT, four studies provided data, but two studies had no events in either the AGT or the NGT groups, precluding calculation of a ratio; hence only two studies contributed to the risk ratio calculation. Risk for CV death was higher in the AGT group than in the NGT group (2.49 [1.46–4.24], P < 0.001). For the association between hospitalization for heart failure and newly discovered AGT, four studies provided data. Risk for hospitalization for heart failure was greater in patients with AGT than in those with NGT but not statistically significant (1.74 [0.87–3.46], P = 0.12). The relative hazards (95% CI) for all-cause mortality and MACE were very similar across the study period (Supplementary Table 2).

A strong association between prediabetes and all-cause mortality, recurrent MACE, CV death, and hospitalization for heart failure also was observed (Fig. 3A–D). For the association between all-cause mortality and newly discovered prediabetes, 10 studies provided data. The prediabetes group had a higher risk of mortality than the NGT group (HR 1.36 [95% CI 1.13–1.63], P = 0.001). Compared with the NGT group, the prediabetes group also had a higher risk of MACE (1.42 [1.20–1.68], P < 0.001), CV death (2.41 [1.36–4.26], P = 0.002), and hospitalization for heart failure (2.17 [1.09–4.34], P = 0.03).

Compared with those with NGT, individuals with acute MI and newly diagnosed diabetes had a higher risk of all-cause mortality (HR 1.74 [95% CI 1.48–2.05], P < 0.001), MACE (1.54 [1.23–1.93], P < 0.001), and CV death (2.63 [1.44–4.81], P = 0.002). Risk of hospitalization for heart failure tended to be higher, but this did not reach statistical significance (1.40 [0.75–2.63], P = 0.20) (Supplementary Fig. 2A–D).

When the prediabetes and newly diagnosed diabetes groups with acute MI were directly compared, using prediabetes group as a reference, no significant differences were observed for all-cause mortality (HR 0.82 [95% CI 0.63–1.07], P = 0.14), recurrent MACE (0.94 [0.79–1.12], P = 0.48), CV death (0.94 [0.59–1.51], P = 0.81), and hospitalization for heart failure (1.36 [0.71–2.58], P = 0.35) (Supplementary Fig. 3A–D).

Subgroup analysis comparing studies using OGTT and studies using non-OGTT (FPG, HbA_1c) showed similar effect sizes. There was no difference between the OGTT group and non-OGTT group in terms of risks of mortality and MACE. AGT status defined by OGTT and methods other than OGTT were associated with higher risk of mortality (HR 1.66 [95% CI 1.39–1.98], P < 0.001) and MACE (1.44 [1.22–1.70], P < 0.001), respectively. For MACE, AGT, detected by either OGTT or non-OGTT, was associated with higher risk compared with NGT for OGTT (1.42 [95% CI 1.16–1.73], P < 0.001) and for non-OGTT (1.44 [1.23–1.68], P < 0.001) (Supplementary Fig. 4A and B).

All included studies were considered to be of high quality (combined score of ≥7) (Supplementary Table 3) according to Newcastle-Ottawa criteria (34). Funnel plot inspection showed possibilities of publication bias in some outcomes. Also, some evidence of incomplete outcome reporting was seen in these cohort studies. The E-value point estimate for all-cause mortality and MACE is 2.06–2.39, indicating that the risk of having unmeasured confounding is small. The E value for the risk of having an unmeasured confounder that would move the CI limit to 1 is 2.38–3.46, which also is small (Supplementary Table 4).

The present meta-analysis documents the high prevalence of newly discovered AGT in individuals with acute MI and not known to have any prior disturbance in glucose homeostasis. Of note, individuals with newly diagnosed prediabetes with acute MI had an increased risk of recurrent MACE and mortality that was as high as that observed in patients with overt diabetes. The discovery of AGT, whether prediabetes or diabetes, in patients who present with an acute MI is a strong predictor for mortality and recurrent MACE.

Prior studies have demonstrated an increased prevalence of CV risk factors long before the onset of diabetes (35). The relationship between ASCVD risk and 2-h plasma glucose concentration is a continuum, and the increased CV risk begins at a range below the threshold for diabetes diagnosis (24). However, glycemia is not the only factor that influences coronary artery disease risk (36,37). Insulin resistance and multiple CV risk factors associated with the insulin resistance (metabolic) syndrome, as well as duration of diabetes, also play important roles in the development of CV disease. People with prediabetes are maximally/near-maximally insulin resistant (1), manifest multiple CV risk factors (22–26), and are at high risk for CV disease (7,38,39).

On the basis of the preceding discussion, it is clear that individuals with prediabetes have multiple CV risk factors and are at high risk for CV events. Nonetheless, individuals with prediabetes, even those who present with acute MI, often receive less aggressive treatment than those with diabetes because of physician unawareness of their high CV risk. Although lifestyle intervention and pharmacologic intervention can reduce progression of IGT to type 2 diabetes, they have failed to decrease CV events in patients with IGT with acute MI (40–42) and in individuals with type 2 diabetes (43). In such individuals, antidiabetic agents with proven CV benefit should be considered, including thiazolidinediones (which have been shown to reduce recurrent ASCVD events in individuals with insulin resistance and no diabetes); other classes of agents that have been found to be beneficial in reducing ASCVD risk in patients with established type 2 diabetes, such as glucagon-like peptide 1 receptor agonists (GLP-1RAs) and sodium–glucose cotransporter 2 (SGLT2) inhibitors, may also play a role, although they have not yet been systematically evaluated in patients with prediabetes to date.

The insulin sensitizing agent pioglitazone has been shown to reduce recurrent stroke and MI, as well as progression to diabetes, in subjects with insulin resistance but without diabetes with recent stroke or transient ischemic attack (44). Pioglitazone also has been shown to slow progression of carotid intimal media thickness and reduce CV events in individuals with prediabetes independently of improvement in hyperglycemia, dyslipidemia, or blood pressure, suggesting a direct vascular benefit of the thiazolidinedione (45,46). In subjects with insulin resistance but without diabetes with recent stroke or transient ischemic attack, pioglitazone markedly reduced (HR 0.76 [95% CI 0.62–0.93], P < 0.001) recurrent stroke and CV events (44). GLP-1RAs (47–50) and SGLT2 inhibitors (51–56) also have been demonstrated to decrease three-point MACE, hospitalization for heart failure, and for some agents in the class, CV death in individuals with type 2 diabetes with established ASCVD or diabetic kidney disease (57). Although these agents have not been formally evaluated in CV outcome trials that included patients with prediabetes because the CV benefit of these antidiabetic agents in patients with type 2 diabetes is independent of the reduction in plasma glucose concentration (48,49,53), they have the potential to reduce recurrent MACE and CV mortality in patients with acute MI who are found to have AGT. Use of these potentially cardioprotective therapies in patients with prediabetes and recent MI should be further studied in dedicated CV outcome trials given the high residual risk of recurrent events in this patient group.

Because of the high prevalence of prediabetes and overt but unrecognized diabetes in individuals who present with an acute MI, we recommend that all hospitalized patients with acute MI be screened with at least one of the following measurements: FPG concentration, HbA_1c, or OGTT. Although the OGTT is a useful and sensitive tool to detect AGT, this method might be confounded by stress from the illness, and measurements may not be reliable until at least 5–7 days after MI. In many cases, AGT diagnosed soon after an MI may be transient, and only some of the diagnosed patients will remain prediabetic or diabetic. However, the outcome of an OGTT performed in patients with acute MI at the time of hospital discharge still provides information about the long-term glucometabolic state (58,59). Therefore, we recommend using HbA_1c for the assessment of glucose tolerance status at the time of hospitalization followed by an OGTT after ≥1 weeks. HbA_1c measurements are particularly useful because they are not affected by the stress of an acute MI.

Finally, some comment is warranted concerning the most recent American Diabetes Association guidelines (60) about the use of antihyperglycemic agents in newly diagnosed patients with type 2 diabetes with established ASCVD. The guidelines still promote metformin as first-line therapy in such individuals, although GLP-1RAs and SGLT2 inhibitors are recommended irrespective of the HbA_1c. On the basis of the high recurrent rate of CV events in these patients, we believe that agents with established cardioprotective benefit (i.e., GLP-1RAs, SGLT2 inhibitors, pioglitazone) should be used before or concomitantly with metformin. Consistent with the European Society of Cardiology 2019 and the European Association for the Study of Diabetes, GLP-1RAs and SGLT2 inhibitors are recommended as first-line therapy in individuals with established CV disease.

This study has some limitations. First, the method to determine AGT status varied among studies. The majority used the OGTT to detect AGT, which is more time consuming, but the OGTT has higher sensitivity to detect AGT. Consequently, studies using FPG or HbA_1c alone might lead to an underestimate of AGT prevalence in the cohorts. Nonetheless, the subgroup analysis showed similar risk ratios for recurrent CV events in the prediabetic and diabetic groups. Since all the included studies were conducted during the period in which effective revascularization strategies and pharmacologic treatments (antiplatelets, statins, ACE inhibitors/angiotensin receptor blockers) were widely available and since all the studies showed a positive association with the risk of mortality and recurrent MACE, we do not believe that the time period when the studies were carried out influenced the results. We did not examine each component of MACE separately because not all studies reported data about individual MACE components. Furthermore, some studies used slightly different definitions of MACE, and this is not an individual patient meta-analysis. Thus, time-to-event analysis and covariate-adjusted analysis could not be conducted to examine heterogeneity reliably. Finally, the number of CV death and heart failure events are too few to yield reliable inference.

In conclusion, AGT, including prediabetes and diabetes, is common in patients who present with acute MI and are previously unknown to have any disturbance in glucose homeostasis. These individuals are at increased risk for recurrent MACE, CV mortality, and all-cause mortality compared with individuals with NGT who present with acute MI. Early and aggressive intervention, both lifestyle and pharmacologic, is important for the prevention of prediabetes progression to diabetes and recurrent CV events. Antidiabetic agents with proven CV risk benefit among patients with established diabetes should be extended to high-risk patients with prediabetes, given their high residual risk.

Acknowledgments. The authors thank Lorrie Albarado (UT Health San Antonio, San Antonio, TX) for expert assistance in preparation of the manuscript.

Duality of Interest. M.K. is a member of the following consultant/advisory boards: Amarin, Amgen, Applied Therapeutics, AstraZeneca, Bayer, Boehringer Ingelheim, Eisai, GlaxoSmithKline, Glytec, Intarcia, Janssen, Merck (Diabetes), Novartis, Novo Nordisk, and Sanofi. M.K. has research grants from AstraZeneca and Boehringer Ingelheim. R.A.D. is a member of the advisory boards of AstraZeneca, Novo Nordisk, Boehringer Ingelheim, Intarcia, and Janssen and is a member of the speakers’ bureau of AstraZeneca and Novo Nordisk. R.A.D. also has grant support from AstraZeneca, Janssen, Boehringer Ingelheim, and Merck. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. N.L. researched and analyzed data and wrote the manuscript. M.A.-G., M.K., and S.J.K. contributed to the discussion and reviewed/edited the manuscript. W.W.P. researched data. R.A.D. researched data, wrote the manuscript, and reviewed/edited the manuscript. N.L. 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.