Risk of Coronary Heart Disease in Different Criterion of Impaired Fasting Glucose

Supplemental Digital Content is available in the text


INTRODUCTION
T he term prediabetes is used to define individuals with intermediate states of abnormal dysglicemia between normoglycemia and overt type 2 diabetes mellitus (T2DM), including those with impaired fasting glucose (IFG) and those with impaired glucose tolerance (IGT). 1 Subjects with IFG or IGT are at high risk for developing T2DM. 1 It has also been reported that IGT is associated with increased risk of cardiovascular disease (CVD). 2,3 However, the association of IFG and risk of CVD is far more unclear. 4 Furthermore, the 2003 American Diabetes Association (ADA) guideline lowered the fasting plasma glucose (FPG) cut-point for diagnosing IFG from 110-125 to 100-125 mg/dL, in order to better identify subjects with future T2DM risk. 5 Although more than a decade has passed, this change is still contentious and not adopted by the World Health Organization (WHO) Expert Group 6 or other international guidelines. 7,8 One of the main arguments against the cut-point of IFG proposed by 2003 ADA is that it greatly increases the number of subjects labeled with IFG, while without clear evidence of association with clinical complications. 9 A recently published meta-analysis reported that the risk of stroke was increased in people with IFG defined as FPG 110 to 125 mg/dL (IFG 110) but not in those with IFG defined as FPG 100 to 125 mg/dL (IFG 100). 10 However, another metaanalysis showed that the risks for CVD are similar in subjects with IFG 110 and IFG 100. 11 These inconsistencies may be caused by the differences in inclusion criteria and endpoint assessment.
Considering these inconsistent results, we aimed to evaluate the association between different definitions of IFG and risk of coronary heart disease (CHD).

Ethics Statement
This study does not involve patients, so ethical approval was not required.

Search Strategy and Selection Criteria
The search strategy was performed in accordance with the recommendations of the Meta-analysis of Observational Studies in Epidemiology (MOOSE) Group. 12 Electronic databases (PubMed and EMBASE) were searched for prospective cohort studies to May 31, 2015, using a combined text and MeSH heading search strategy with the terms ''blood glucose,'' ''impaired fasting glucose,'' ''hyperglycaemia,'' or ''borderline diabetes'' and ''cardiovascular events,'' ''cardiovascular disease,'' ''ischemic heart disease,'' ''coronary heart disease,'' ''coronary artery disease,'' ''myocardial ischemia,'' ''myocardial infarction,'' ''angina'' and ''risk,'' or ''risk factors.'' We restricted the search to human studies, but there were no language or publication form restrictions. The reference lists of published articles and reviews on this topic were also checked to identify other eligible studies. The detailed search strategy used for PubMed is presented in online supplementary Table S1, http://links.lww.com/MD/A453. The strategy for the EMBASE database was similar, but was adapted where necessary.
The inclusion criteria of studies for analysis were: prospective cohort studies involving adult participants (aged !18 years) with assessment of risk of CHD; blood glucose and other cardiovascular risk factors were evaluated at baseline; and adjusted relative risk (RR) and 95% confidence intervals (CIs) reported for events associated with IFG relative to normal fasting glucose (NFG). IFG defined as FPG of 100 to 125 mg/dL (IFG 100) or 110 to 125 mg/dL (IFG 110). 5,6 Corresponding NFG comparator was defined as FPG < 100 or < 110 mg/ dL, respectively.
Studies were excluded if: data were collected from patients with a particular condition (eg, previous history of hypertension, acute myocardial infarction, and kidney disease) but not general population; not accessed the risk of CHD in people with IFG compared with NFG; the risk of CHD in IFG was unadjusted for other risk factors; or reports were derived from the same cohort. If duplicate publications were identified as from the same cohort, only data from the most recent publication were used for analysis. 9,13 Data Extraction and Quality Assessment of Included Studies Two authors (TY and WL) independently conducted independent literature searches, reviewed the potentially articles, and abstracted data from eligible studies. The quality assessment was evaluated according to the Newcastle-Ottawa Quality Assessment Scale (NOS) for assessing the quality of nonrandomized studies in meta-analyses, 14 which based on the assessment of bias for selection, comparability, and exposure/outcome, with a total score up to 9. In this metaanalysis, included studies were graded as good quality if they with a score !7, fair if they had less than 7 score. 13 We also evaluated whether the studies were adequate adjusted for potential confounders (at least 6 of 8 factors: age, sex, blood pressure or antihypertensive treatment, body mass index or other measure of overweight/obesity, physical activity, cholesterol concentration or lipid-lowering medication use, history of CVD or exclusion of CVD at baseline, and smoking).

Data Synthesis and Analysis
We analyzed the RR of CHD in individuals with different definition of IFG. Subgroup analyses were conducted according to sex (women vs men), ethnicity (Asian vs non-Asian), specific end points (fatal vs fatal plus nonfatal CHD), participant's age (average <50 vs !50 years), follow-up duration (<10 vs !10 years), possibility of enrolling patients with diabetes (yes vs no), and adjustment of risk factors (adequate vs un-adequate).
We extracted the most adjusted RRs and 95% CIs from each included studies and logarithmically transformed these values, calculated the corresponding standard errors (SEs) to stabilize the variance and normalize the distribution. 15,16 The inverse variance method was used to combine the log RRs and SEs using random effects models. The I 2 statistic was used to estimate between-study heterogeneity. Values of I 2 > 50% were considered to indicate significant heterogeneity. The estimated RRs were calculated using random-effects models. The test for subgroup differences was calculated by Chi-square statistics.
Publication bias was assessed by inspecting funnel plots for each outcome in which the natural log of RR was plotted against its SE. Sensitivity analyses were conducted by omitting one study at a time and recalculating the estimated RRs and CIs. P values were 2-tailed, and the statistical significance was set at 0.05. All analyses were performed with RevMan software (version 5.3 for Windows; The Cochrane Collaboration, Copenhagen, Denmark).
All of the included studies were derived from the general population. The characteristics of the 17 studies are presented in Table 1. Nine of the studies were from the US and Europe 17,19-21,24-26,30,31 and 8 were from Asia. 18,22,23,[27][28][29]32,33 One study only enrolled men 26 while all of the others included both men and women for analysis. The follow-up duration ranged from 4 to 20 years.
Sensitivity analyses confirmed that the risk of CHD in people with IFG 100 or IFG 110 were not influenced by the use of random-effects models compared with fixed-effects models, or recalculating the RRs by omitting one study at a time.

Subgroup Analyses
The results of subgroup analyses are presented in Table 2. In individuals with IFG 100, there were no significantly differences among subgroups conducted according to sex, ethnicity, participant's age, specific end points, and follow-up duration. However, the risk of CHD was significantly increased in studies with possibility of enrolling patients with increased 2-h PG (RR 1.19, 95% CI 1.06-1.33), but not in studies excluded participants with increased 2-h PG (RR 0.98, 95% CI 0.86-1.11). Furthermore, the risk of CHD was increased in studies with inadequate adjustment (RR 1.27, 95% CI 1.12-1.45), but not in those with adequate adjustment of other risk factors (RR 1.05, 95% CI 0.96-1.15). The differences of CHD risk between these subgroups comparison were both significant (both P ¼ 0.02).
In subgroups analysis of IFG 110, the risk of CHD was also increased in studies with possibility of enrolling patients with increased 2-h PG (RR 1.20, 95% CI 1.10-1.31), not in studies excluded participants with increased 2-h PG (RR 1.09, 95% CI 0.88-1.35), in studies with inadequate adjustment (RR 1.20, 95% CI 1.08-1.32), but not in those with adequate adjustment of other risk factors (RR 1.12, 95% CI 0.93-1.35). However, there were no significant differences among all subgroups comparison (all P > 0.1, I 2 ¼ 0%). Ã For heterogeneity among subgroups. y Adequate adjustment denoted adjustment of at least 6 of 8 factors: age, sex, blood pressure or antihypertensive treatment, body mass index or other measure of overweight/obesity, physical activity, cholesterol concentration or lipid-lowering medication use, history of CVD or exclusion of CVD at baseline, and smoking.

DISCUSSION
In this meta-analysis, we found that in the general population, IFG was significantly associated with future risk of CHD. The risk of CHD was increased when FPG was as low as 100 mg/dL according to the lower cut-point of IFG by the ADA.
The 2003 ADA criterion of IFG had been criticized as it significantly increased the prevalence of IFG while without improvement of prediction for risk of CVD. 34 In this study, there was sufficient power to show that the presence of IFG, defined by the WHO or ADA criterion, was associated with increased risk of CHD. These findings support the lower IFG cut-point proposed by the ADA and highlight the importance of early management of mild hyperglycemia for the prevention of CHD. Our results were different with a prior meta-analysis, which showed that the risk of stroke was increased in people with IFG defined by the WHO but not in those defined by the ADA. 10 These inconsistent findings may be caused by differences in the events assessed. Furthermore, in the prior metaanalysis, they combined studies from general population, as well as studies from patients with coronary artery disease for analysis. 10 However, we only used studies from general population for analysis. Our more stringent inclusion criteria are important for avoiding between-study heterogeneity and reaching more reliable conclusion. In our study, the risk of CHD associated with IFG was significantly increased in studies with possibility of enrolling patients with increased 2-h PG, but not in studies excluded participants with increased 2-h PG. These results showed that the risk of CHD in people with FPG maybe confounded by the undetected increased 2-h PG (IGT or T2DM defined by 2-h PG). Many studies have shown that IGT was a stronger predictor of cardiovascular events than IFG. 17,35 However, routine detection of IGT had been questioned due to the inconvenient use of OGTT and the results are not highly reproducible. Our results highlight the notion that OGTT could be required for further diagnosing individuals with IFG. 36 It has been estimated that, by the year of 2025, the number of people with prediabetes will be 472 millions. 37 Successful interventions in this large population could have a major public health impact. It had been proved that lifestyle is a fundamental management approach that can effectively prevent the progression from prediabetes to diabetes. 38 Furthermore, recently data showed that lifestyle intervention in IGT can reduce incidence of cardiovascular and all-cause mortality. 39 However, the evidence regarding CVD prevention in people with IFG is still absent.
The main strengths of our study are the very large sample size with general population included from prospective cohort  studies. Detailed subgroup analyses also found interesting results that the risk of CHD associated with FPG may be confounded by the undetected increased 2-h PG and other cardiovascular risk factors. However, our study also has some limitations. First, individuals with IFG are more likely to progress to DM than those with normoglycemia, 1 but most of the included studies did not adjust for subsequent blood glucose levels. So, the long-term risk of CHD in people with IFG was caused by the mild elevation of blood glucose or the future progression of DM remains unknown. However, it had been indicated that coronary atherosclerosis detected by intravascular imaging modalities is already ongoing in prediabetic status. 40 Second, the adjusted confounders in the included studies were inconsistent and may be a potential source of bias in our study. However, it is interesting that, in both IFG 100 and IFG 110 subgroup analysis, the risk of CHD was increased in studies with inadequate adjustment, but not in those with adequate adjustment of other cardiovascular risk factors. These results reinforce the importance of detection of other cardiovascular risk factors in risk stratification of people with IFG. 41 In conclusion, this meta-analysis showed that IFG was associated with an increased risk of CHD. The risk increased in people with FPG as low as 100 mg/dL. These results reaffirm the importance of screening for prediabetes using the ADA criteria. Furthermore, detection of 2-h PG and other cardiovascular risk factors are important for risk stratification in people with IFG. These informations are important for the prevention of DM and CVD.