The effect of change in fasting glucose on the risk of myocardial infarction, stroke, and all-cause mortality: a nationwide cohort study

Diabetes is related to both microvascular and macrovascular complications as well as mortality [1, 2]. In particular, macrovascular complications such as myocardial infarction (MI) and stroke account for 80% of all deaths in patients with type 2 diabetes mellitus [3]. Therefore, preventing macrovascular complications by controlling hyperglycemia is imperative in reducing the risk of cardiovascular disease and mortality [4‐6].

Recent evidence suggests that a prediabetic status may also elevate the risk of cardiovascular disease and mortality. A prediabetic status referred to people who have impaired fasting glucose (IFG, fasting glucose of 100–125 mg/dL), impaired glucose tolerance (IGT, 2-h post glucose load of 140–199 mg/dL), or both, according to American Diabetes Association [5]. Some studies have shown that IGT is superior to IFG as a risk factor for cardiovascular disease [7‐10]. However, the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab) reported IFG as well as IGT to be an independent predictor of cardiovascular disease and all-cause mortality [11]. Moreover, meta-analyses also showed that not only IGT but IFG is related to risk of cardiovascular disease [12, 13].

Blood glucose levels in most prior studies, however, were assessed only one time, possibly resulting in misclassification of IFG or IGT. The effect of change in blood glucose levels on the risk of cardiovascular disease among people without diabetes are currently unclear. We, therefore, aimed to examine the association of change in fasting serum glucose with incident cardiovascular disease and all-cause mortality within a large population without diabetes using the Korean National Health Insurance Service-Nation Health Screening Cohort (NHIS-HEALS).

260,487 individuals without diabetes were followed up to an average of 8 years (standard deviation, SD 0.9), resulting in 2,042,960 person-years. During the follow-up, the number of incident MI, incident stroke, and all-cause mortalities were 1318 (0.5%), 8144 (3.13%), 10,065 (3.86%), respectively. The baseline characteristics of study participants are presented in Table 1. Those who belonged to the NFG group at the baseline were more likely to be younger women, with a lower BMI, never smokers, with a lower alcohol consumption, lower blood pressure, and lower total cholesterol than those who were in the IFG group at baseline.

Figure 2 shows adjusted cumulative hazard curves for 8-year MI, stroke, and all-cause mortality by changes in fasting glucose, after adjusting for age, socioeconomic status, smoking habit, alcohol consumption, physical activity, BMI, CCI, blood pressure, and fasting glucose at baseline. Compared to individuals with persistent NFG (i.e., NFG to NFG), individuals who shifted from NFG to DFG had an increased risk of stroke (HR [95% CI]: 1.19 [1.02–1.38]) and individuals who shifted from NFG to IFG or DFG had increased risks of all-cause mortality (HR [95% CI]: 1.08 [1.02–1.14] for NFG to IFG and 1.56 [1.39–1.75] for NFG to DFG). Compared to individuals with persistent IFG, individuals who shifted from IFG to DFG had an increased risk of MI and all-cause mortality (HR [95% CI]: 1.65 [1.20–2.27] and 1.16 [1.02–1.33], respectively).

Subgroup analyses using the multivariable model of MI, stroke, and all-cause mortality risks by changes in fasting glucose from NFG and IFG are shown in Tables 2, 3, 4. Individuals who were 65 years or older had an increased risk for MI. With regard to all-cause mortality, results of all subgroups stratified by age, sex, and BMI were similar to the main results. The magnitude of HRs of all-cause mortality was higher in patients who were more than 50 years old than in those less than 50 years, in men than in women, and those with BMI ≥ 25 kg/m² (obese based on Asian criteria) than with BMI < 25 kg/m².

The results of the sensitivity analyses based on the multivariable model of MI, stroke, and all-cause mortality risks, in relation to changes in fasting glucose, from NFG and IFG, are shown in Additional file 1: Tables S1 and S2. The results were almost similar to the results shown in Fig. 2.

In this general population-based, retrospective cohort study with more than 2,000,000 person-years of follow-up, we examined the association between 2-year change in fasting serum glucose and risk of cardiovascular disease and all-cause mortality after a median follow-up of 8 years, after adjusting for cardiovascular covariates. Among individuals with NFG at the first health examination, individuals who shifted to IFG were more, and a shift to DFG after 2 years was associated with much higher risks of stroke and all-cause mortality, compared to persistent NFG. Among individuals with IFG at the first health examination, compared to persistent IFG, individuals who shifted to DFG had higher risks of MI and all-cause mortality.

One study examined the effects of change in fasting glucose over time, which was in line with our results. Fasting glucose variabilities are associated with subsequent risks of MI in non-diabetic patients [19]. Other previous studies were also consistent with our findings, although they did not report a change in fasting glucose but rather one-time fasting glucose level. AusDiab study reported that all-cause mortality was greater for IFG than for NFG among 10,428 participants after a median follow-up period of 5.2 years [11]. Additionally, two meta-analyses showed that elevated fasting glucose was associated with the risk of cardiovascular disease in people without diabetes [12, 13]. A prospective study of a large cohort in Korea reported that IFG is associated with the risk of stroke and coronary heart disease [20]. Another prospective national survey in Israel reported that a linear association between admission blood glucose and 10-year mortality among heart failure patients without diabetes [21].

By contrast, some have reported that IFG is less likely to be a risk factor for cardiovascular disease compared to IGT [7‐10]. The differences between the findings of these prior studies and the present study could originate from differences in sample size or from the use of different methods of fasting glucose analysis. Data from several different centers in different countries demonstrate that there is no uniform method for examining fasting glucose [8‐10], which can result in the possibility of misclassification of fasting glucose status. However, most single large cohorts such as that mentioned above used a uniform analysis method for evaluating fasting glucose, which led to the indication that an IFG could be a significant predictor for cardiovascular disease risks like our findings.

Compared to persistent IFG, shift to DFG from IFG during 2 years (i.e. more rapid change in fasting glucose compare to shift to persistent IFG) was associated with a higher risk of cardiovascular and all-cause mortality, which could imply that glycemic control could slow or halt the progression of macrovascular complications and all-cause mortality, in line with previous studies [21‐24]. In Korea, when participants revealed an IFG status in the national health screening program, they received an advisory opinion via a letter from a doctor, including suggestions for lifestyle modifications and a recommendation for follow-up 3–6 months later in nearby clinics. According to this advisory opinion, people with prediabetes status intentionally aim to elicit a reduction in hyperglycemia through lifestyle modification such as healthy diets, exercise, quitting smoking, or abstemious in drinking. Accordingly, early detection of IFG via screening of glycemic status could be one of the strategy to prevent mortality. However, the shift to NFG from IFG is not significantly associated with the risk of cardiovascular disease, which may be due to minor events.

The Cardiovascular Heart Study in the U.S. reported no evidence that prediabetes is associated with subsequent 13-year incident cardiovascular events or mortality in community-dwelling adults aged more than 65 years [25]. In our subgroup analysis of participants older than 65 years, the shift to IFG or DFG from NFG was not also significantly associated with risk of MI or stroke; however, it was associated with all-cause mortality. Other previous studies were consistent with our results. Among hospitalized patients with heart failure without pre-existing diabetes, there was a linear relationship between admission glucose level and 10-year mortality [21]. Although heart failure could be the direct cause of death, higher fasting glucose may be an additional contributor to mortality. The difference between the present study and Cardiovascular Heart Study in U.S. is the presence of an independent variable (i.e. change in fasting glucose vs. one-time measurement of fasting glucose). Accordingly, although a prediabetic status itself was not significant among the elderly, increasing fasting glucose was associated with all-cause mortality.

There are some possible mechanisms reported. Abnormal glucose status disrupts normal endothelial function by oxidative stress [26], protein kinase C activation, and advanced glycated end product receptor activation [27], thereby accelerates atherosclerotic plaque formation as well as increasing arterial stiffness [28]. IFG status has also been associated with arterial endothelial dysfunction and intima-media thickening [29], which linked to incident MI and stroke [30, 31]. Furthermore, a series of experimental studies demonstrated that variability in blood glucose may be more prejudiced to increase cardiovascular disease than constantly high blood glucose [31‐34].

There are limitations to our study that need to be noted. First, the development of MI and stroke confirmed by hospitalization for 2 days or more for the relevant disease based on ICD-10 codes, which are conceivable to underestimate the actual number of cardiovascular disease cases. However, a previous study showed that identifying cardiovascular disease events via the ICD-10 code has an accuracy higher than 80% [35]. Second, future development of diabetes during follow-up period is hard to consider, despite the fact that the risk of major adverse cardiovascular events is significantly greater in diabetic patients who have a longer illness duration [36]. However, we investigated the 2-year change in fasting glucose, which could distinguish the progression to DFG from NFG or IFG status. Third, participants in this study were older than the middle-aged, and the elderly may already have subclinical cardiovascular disease. However, among patients with heart failure, high glucose level was associated with mortality [21]. In addition, some studies reported that target organ damage precedes the clinical appearance of diabetes [37]. Lastly, fasting glucose was assessed through serum, not plasma as was recommended [5], which induced an error in the serum analyses of 1.15% compared to plasma analyses [38]. This error may generate when the sample is stored at room temperature after drawing blood; however, the error is very low because the NHIS recommends refrigeration of the sample at health screening centers. Despite these limitations, these findings show representative results based on data from a nationwide NHIS database [15].