In this nationwide population-based study, visit-to-visit variability of FPG was a remarkable predictor of PAD development in individuals without diabetes. These findings may be robust considering an independent association between greater glycemic variability and a higher risk of PAD incidence even after extensive adjustment for atherosclerotic risk parameters, including mean FPG levels. In addition, these results were consistently preserved in the sensitivity analysis by excluding crucial risk factors.
As proven in previous studies, glucose levels play an important role in developing arteriosclerosis in population without diabetes. In the Atherosclerosis Risk in Communities (ARIC) study, HbA1c was significantly associated with the risk of CVD and all-cause mortality in adults without diabetes [21]. Using a meta-analysis of 102 prospective studies, Sarwar et al. showed that the FPG levels were associated with atherosclerosis-related vascular disease in participants without diabetes [22]. A meta-analysis of 17 prospective cohort studies in the Asia Pacific region indicated that the FPG levels was a critical risk factor for cardiovascular disease, and that there were significant potential benefits to reducing fasting glucose levels to at least 4.9 mmol/l in participants with/without diabetes [23]
Additionally, reducing glycemic variability is an emerging glycemic target, and has been suggested as a critical predictor of cardiovascular complications. Several observational studies [7] and post-hoc analysis of trials, including the Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) trial [9], the Veterans Affairs Diabetes Trial (VADT) [24], the Trial Comparing Cardiovascular Safety of Insulin Degludec vs. Insulin Glargine in Patients with Type 2 Diabetes at High Risk of Cardiovascular Events (DEVOTE) [25], the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial [26], and the Empagliflozin Cardiovascular Outcome Event (EMPA-REG OUTCOME) trial [27] demonstrated a positive association between visit-to-visit glycemic variability and the risk of micro/macrovascular disease in participants with type 2 diabetes. Furthermore, long-term glycemic variability has also been suggested to affect cardiovascular disease and mortality even in individuals without diabetes. [16, 28].
Recently, several studies have investigated the effect of glycemic variability on PAD development in patients with diabetes. In an Italian multicenter study, Penno et al. showed that HbA1c-SD correlated with lower extremity PAD and severe PAD, such as ulceration/gangrene in patients with type 2 diabetes. [29]. A meta-analysis showed that increased HbA1c variability was associated with a higher risk of severe PAD in patients with type 2 diabetes [7]. In a population-based cohort in Taiwan Yang et al. demonstrated a significant association between FPG-CV and development of PAD, independent of HbA1c levels, in participants with type 2 diabetes [30]. However, various confounding factors, such as medications for diabetes and compliance by patients in their use as well as other diabetes-related complications, might affect their relationship in patients with diabetes. Sun et al. showed a positive relationship between FPG variability and the risk of PAD even in individuals without diabetes, using the ARIC study cohort from US database [31]. However, Sun et al.’s study was performed based on only three measurements of FPG levels. Furthermore, they simply adjusted for baseline FPG as a confounding factor along with other cardiovascular risk factors, and did not use the mean FPG levels. Because FPG is known as a critical risk factor for atherosclerotic cardiovascular disease, in addition to glycemic variability, the mean FPG levels during study period might also affect the final result.
In this study, the participants in the highest quartile of FPG variability presented a similar increased risk of PAD incidence as those in the lowest quartile, as observed using three different methods: FPG-CV, FPG-SD, and FPG-VIM. Moreover, the main findings were consistent in diverse sensitivity analysis, excluding populations with a history of stroke, CAD, CHF, CKD, smoking, or drinking, which may support the robustness of the findings in this study. In addition, the present study, that included an Asian population with a relatively lower risk of PAD than the Caucasian population, may broaden the insights into the relationship between glycemic variability and atherosclerosis.
The pathophysiological mechanisms that link the association between long-term FPG variability and PAD are still unknown; thus, further research is required. There are some possible explanations for why high glycemic variability could be a risk factor for PAD incidence. First, glycemic variability, regardless of persistent hyperglycemia, could play a crucial role in atherosclerosis by increasing chronic inflammation, oxidative stress, endothelial dysfunction, and insulin resistance [10–13]. Second, glycemic variation may lead to pancreatic beta cell dysfunction and apoptosis [32]. Furthermore, metabolic memory, and chromatin remodeling caused by repeated glycemic fluctuations, could induce atherosclerosis [33]. Third, glycemic variability may induce cardiovascular autonomic dysfunction with sympathetic activation that may contribute to atherosclerosis [34–36]. It has been reported that patients with PAD have decreased heart rate variability than those without PAD in type 2 diabetes, which implies cardiovascular autonomic dysfunction [34]. Lastly, individuals with high glycemic variability tended to have more traditional risk factors, such as advanced age, high BMI, smoking history, high blood pressure, and dyslipidemia. The present study attempted to alleviate their influence using various sensitivity analyses and extensive adjustment for diverse confounding factors.
Different effects of FPG variability on the risk of PAD were observed across subgroups of age, exercise, and income. In younger individuals (<60 years), regular exercisers, and those with higher income, FPG variability was associated with a higher risk of PAD than in those of older age (≥60 years), non-regular exercisers, and those with lower income. Although statistically insignificant, Sun et al. also showed that individuals aged <60 years had a higher risk of PAD using FPG-CV and FPG-SD than individuals aged ≥60 years among those without diabetes [31]. In epidemiologic studies, PAD is more common in older age [37], sedentary lifestyle [38], and low socioeconomic status [39, 40]. This study showed that increased FPG variability had a greater impact on the risk of PAD in relatively favorable conditions for cardiovascular risk.
This study has several limitations. Other assessments of glycemic levels, including HbA1c or oral glucose tolerance tests, were not used in the health checkup data by the NHIS. In addition, PAD was defined by physicians’ diagnostic codes of the ICD-10; thus, under- or over-diagnosis of PAD could be possible. In this study, the prevalence of PAD was 11%, similar to that (10.7%) of a large cohort study of individuals aged 40 or older in the United States [2]. Previous studies also defined PAD based on medical claim codes [2, 30]. We used the ICD-10 code number according to the American Heart Association [20]. Lastly, although we tried to adjust for diverse confounding variables that likely influenced the incidence of PAD, the probability of residual confounding variables cannot be excluded. However, this study has several noteworthy strengths. This study had a large sample size, extensive information about potential confounding variables, and sufficient duration of observation, using a standardized database validated by the Korean government. In particular, we showed coincident and consistent implications for PAD incidence according to long-term glycemic variability determined by three different indices, FPG-CV, FPG-SD, and FPG-VIM, even after adjusting for multiple confounding variables including average FPG. Additionally, we used samples taken from the participants representing the entire Korean population; thus, these findings may reflect the real-world situation.