The Relation Between HbA_1c and Cardiovascular Events in Patients With Type 2 Diabetes With and Without Vascular Disease Free

Strict glycemic control has been proposed as an important means to lower the risk of both microvascular and macrovascular complications of type 2 diabetes. A strong relation between glycemic control and microvascular complications (nephropathy, retinopathy, and neuropathy) and macrovascular complications is observed in patients with type 2 diabetes (1–4). However, most of the cohort studies (3,5) conducted with patients having type 2 diabetes have investigated the relation between glycemic control and cardiovascular disease in patients without vascular disease at baseline.

As prolonged exposure to hyperglycemia results in vascular damage, it seems feasible that strict glycemic control will be associated with a decrease in cardiovascular risk (6). Although cohort studies (7–9) have indeed found a relation between glycemic control and the incidence of cardiovascular disease, the cardiovascular risk in patients with type 2 diabetes does not seem to further decrease with intensive glycemic control beyond an HbA_1c level of 7%. Identifying those patients with type 2 diabetes who would benefit from intensive glycemic control might be an opportunity for improving treatment. Existing guidelines on the treatment of diabetes are based on this principle and thus stress the importance of identifying the characteristics for determining the optimal HbA_1c target in individual patients. Guidelines do address which patient groups are more likely to profit or suffer from strict glycemic control. Unfortunately, as studies (10,11) in specific patient groups are lacking, setting practical treatment goals in different patient groups is still difficult.

Post hoc analyses from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) (9); the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) (8); the UK Prospective Diabetes Study (UKPDS) (12); and the Veterans Affairs Diabetes Trial (VADT) (7) suggested that the presence of vascular disease is an important patient characteristic to determine an individual glycemic goal, since a different effect of glycemic control was found in patients with and without vascular disease (13). The goal of the current study is to investigate the relation between HbA_1c level and cardiovascular events and all-cause mortality in patients with type 2 diabetes with and without clinical manifestations of vascular disease.

For this study, data from 1,687 participants with type 2 diabetes who were enrolled in the Second Manifestations of Arterial Disease (SMART) study before 1 March 2013 were used. Diabetes was defined as a referral diagnosis of type 2 diabetes, self-reported type 2 diabetes, a fasting serum glucose concentration of ≥7.0 mmol/L at study inclusion with the initiation of glucose-lowering treatment within 1 year, or the use of oral anti-hyperglycemic agents or insulin at baseline. Participants with known type 1 diabetes were excluded for this analysis. The SMART study is an ongoing prospective, single-center, cohort study in patients with manifest vascular disease and/or cardiovascular risk factors. Starting from September 1996, consecutive patients, who were 18–80 years of age and had been referred to the University Medical Centre Utrecht (UMCU), the Netherlands, with manifest vascular disease or a cardiovascular risk factor underwent a vascular screening. Written informed consent was obtained from all participants. The study was approved by the Medical Ethics Committee of the UMCU.

Patients were biannually asked to fill out a questionnaire. Events of interest for the current study were the occurrence of vascular death, stroke, myocardial infarction, and the composite of these vascular events. In addition, we were interested in mortality and nonvascular death. Definitions have been described previously (14) and are included in Supplementary Table 1. When a possible event was recorded by the participant, hospital discharge letters and results of relevant laboratory and radiology examinations were collected. With this information, all events were audited by three members of the SMART study End Point Committee, comprising physicians from different departments.

HbA_1c level was measured at baseline in all patients who were enrolled in the SMART study after 2006. In patients who were enrolled in the SMART study before 2006, HbA_1c level was determined using available stored blood samples.

Missing data for HbA_1c (n = 128; 7.6%) were singly imputed by weighted probability matching on the basis of multivariable regression using covariate and outcome data.

Baseline data are presented as the mean ± SD or median with interquartile range in the case of a skewed distribution.

Cox proportional hazards analyses were performed to estimate hazard ratios (HRs) and 95% CIs for the relation between HbA_1c and the occurrence of cardiovascular events, defined as a composite of nonfatal and fatal myocardial infarction, nonfatal and fatal stroke, or vascular mortality. If a patient had multiple events, the first event was used in the analyses. The proportional hazards assumption was satisfied based on a Schoenfeld residual plot.

To estimate the relation between HbA_1c level and cardiovascular events and mortality, we built three models. First, the unadjusted relation between HbA_1c level and cardiovascular events was estimated. In model II, age and sex were added. In model III, the model was additionally adjusted for current smoking status, systolic blood pressure, diabetes duration, non–HDL cholesterol level, and modification of diet in renal disease (MDRD). The variables mentioned in the models were a set of previously chosen confounders of the relation between HbA_1c level and cardiovascular events and mortality (age, sex, current smoking, and diabetes duration), and a set of previously chosen traditional cardiovascular risk factors (systolic blood pressure, non–HDL cholesterol, and MDRD).

To investigate the possible modifying effect of the presence of vascular disease at baseline, we stratified the population accordingly and performed separate analyses in the different strata. In addition, we performed standard multiplicative interaction analyses by adding the cross-product to the Cox proportional hazards models. Using a similar methodology, we investigated whether vascular disease duration was an effect modifier in the relation between plasma HbA_1c level and new cardiovascular events, since a differential effect with vascular disease of longer duration might be plausible from a pathophysiological perspective. Also, the relation between HbA_1c level and all-cause mortality, nonvascular mortality, and separate end points (vascular mortality, myocardial infarction, and stroke) was studied in the strata of patients with and without previous vascular disease. Finally, the relation between HbA_1c level and cardiovascular events and mortality was assessed in strata of HbA_1c tertiles, and in a continuous way using plots of restricted cubic splines. The P values of the nonlinear effect of baseline HbA_1c level on cardiovascular events and mortality was based on the χ² statistic.

Sensitivity analyses were performed after excluding patients with the 1% highest and 1% lowest HbA_1c levels to eliminate the effect of outliers. As it is possible that the relation of interest differed in the number of years of follow-up, for instance because of better risk management over time, the year of study inclusion was added to the Cox models. As differences in the use of thrombocyte aggregation inhibitors and anticoagulants were expected between patients with and without vascular disease, it was investigated in the same manner whether addition of the use of these medications changed the direction or magnitude of the relation. Also, the use of glucose-lowering medication, blood pressure–lowering treatment, and statins was included in the models as a sensitivity analysis. Finally, sensitivity analyses were performed investigating whether the relation was similar when analyses were performed separately in patients with cerebrovascular, peripheral artery, coronary artery, or vascular disease on various locations at baseline. The level of significance was set at P < 0.05 for all analyses. As the analyses were prespecified, no correction for multiple comparisons was performed. All statistical analyses were performed using SPSS version 21 and R version 3.1.0.

Baseline characteristics are presented in Table 1. The mean age was 60.2 years (SD 10.2 years), and 30% of participants were female. During a median follow-up time of 6.1 years (range 3.1–9.5 years), 293 patients experienced a new cardiovascular event (event rate 17.3%). Of those 293 patients, 189 died of a vascular cause, while 340 patients died of all causes (event rate 20.1%). In total, 6.9% of the patients were lost to follow-up (8.5% in patients without vascular disease and 6.2% in patients with vascular disease).

In all patients with type 2 diabetes, higher levels of HbA_1c were nonsignificantly related to cardiovascular events (HR 1.06 [95% CI 0.97–1.17]). In patients with type 2 diabetes and vascular disease at baseline, no relation between HbA_1c level and new cardiovascular events was found (HR 1.03 [95% CI 0.93–1.15]) (Table 2). Results were similar when performed separately in patients with cerebrovascular, peripheral artery, coronary artery, or vascular disease on various locations at baseline (Supplementary Table 2). On the other hand, in patients without vascular disease, a strong relation between HbA_1c level and cardiovascular events was observed (HR 1.27 [95% CI 1.06–1.51]). Additional sensitivity analyses (the exclusion of patients with the 1% highest and lowest levels of HbA_1c; and adjustment for the year of study inclusion or use of platelet inhibitors, anticoagulants, glucose-lowering medication, blood pressure–lowering treatment, and statins) did not change the direction and magnitude of the relation (data not shown). Analyses in tertiles of HbA_1c (data not shown) and cubic splines describing the relation of HbA_1c level with new cardiovascular events and mortality did not indicate the presence of nonlinearity in patients both with and without vascular disease (Supplementary Fig. 1). The P value for the cross-product of HbA_1c level and the presence of vascular disease was 0.195, indicating no significant interaction.

The relation between HbA_1c and separate events (vascular mortality and the occurrence of coronary ischemic disease or ischemic stroke) in patients with and without vascular disease is shown in Table 3. Though the numbers of events are small, HbA_1c level was significantly associated with the occurrence of ischemic stroke in patients without vascular disease (HR 1.40 [95% CI 1.01–1.94]), while no relation was found in patients with vascular disease (HR 1.03 [95% CI 0.81–1.31]).

The P value for the cross-product of vascular disease duration and new cardiovascular events was 0.490, indicating no statistical interaction by the duration of vascular disease. In all different groups of patients based on the duration of vascular disease, no significant relations were found between HbA_1c level and new cardiovascular events (Table 4).

Patients with type 2 diabetes and manifest vascular disease had a 16% higher risk of all-cause mortality per 1 percentage point increase in HbA_1c (HR 1.16 [95% CI 1.06–1.28]), while a similar, albeit nonsignificant, relation was found in patients without vascular disease (HR 1.13 [95% CI: 0.97–1.31]). The P value for the cross-product of HbA_1c level and the presence of vascular disease in the relation with all-cause mortality was 0.749. The relation between HbA_1c level and all-cause mortality in patients with vascular disease was found specifically in patients with coronary disease or cerebrovascular disease at baseline (Supplementary Table 2).

When investigating the relation between HbA_1c and all-cause mortality in tertiles of vascular disease duration, differential relations were found between the groups (P value of cross-product = 0.044). Interestingly, a significant relation of HbA_1c level and all-cause mortality (HR 1.25 [95% CI 1.07–1.46]) was found in patients with the longest vascular disease duration (6–51 years). No significant relations were found between HbA_1c level and nonvascular mortality in all subgroups. The effect of baseline HbA_1c level on all-cause mortality was of similar magnitude in patients with and without vascular disease, although the relation was not statistically significant in patients without vascular disease.

The current study shows that in patients with type 2 diabetes there is a modest, but not statistically significant, relation between HbA_1c level and cardiovascular events, and as there was no statistically significant interaction, this relation was not different for patients with or without clinical manifestation of vascular disease. The effect of baseline HbA_1c level on all-cause mortality was of similar magnitude in patients with and without vascular disease, although the relation was not statistically significant in patients without vascular disease.

To the best of our knowledge, this is the first prospective cohort study to investigate the relation between HbA_1c level and new cardiovascular events in patients with and without vascular disease. Interestingly, a different relation between glycemic control and cardiovascular outcomes was suggested in a meta-analysis (13) of randomized controlled trials that investigated the effect of intensive versus standard glycemic control (interaction P = 0.04). Although the point estimates and CIs of the HRs are different in direction between the groups, no significant interaction P value (P = 0.195) for multiplicative interaction was found. Since we observed no statistical interaction, we cannot conclude that the relation between HbA_1c level and cardiovascular events is really different depending on the presence or absence of vascular disease. However, the results of this study are mainly important for hypothesis generation. Further research is warranted to specifically investigate this potential difference in effect.

Our findings in patients with type 2 diabetes (with and without vascular disease) are in line with those of other cohort studies (3,5) that studied the relation between HbA_1c level and macrovascular complications in patients with type 2 diabetes. Some cohort studies (15–17) have suggested the presence of a U-shaped relation between HbA_1c level and cardiovascular events. In this study, we did not find a U-shaped curve between HbA_1c level and macrovascular complications in patients with type 2 diabetes, and thus proceeded to analyze the data in a linear fashion. The explanation for the difference between our findings and those of earlier studies is probably to be found in the difference in study population, comprising younger patients and patients with diverse types of vascular disease in the current study versus only patients with coronary artery disease in most other studies.

The findings of cohort studies (18–22) investigating the relation between HbA_1c level and new cardiovascular events in patients with and without type 2 diabetes after cardiac interventions for coronary artery disease are in line with our findings. In these studies (19–21), no relation was found between HbA_1c level and cardiovascular events in patients with established type 2 diabetes after cardiac interventions. In patients without established type 2 diabetes, this relation did exist (18,21,22). In the current study, we expand on these findings by showing a consistent relation across different types of vascular disease.

Several explanations can be given for the different relation between HbA_1c level and new cardiovascular events in patients with type 2 diabetes with and without vascular disease. Although we could not support this hypothesis in this study, it is possible that the relation between HbA_1c level and cardiovascular events is more U shaped in patients with vascular disease compared with the relation in patients without vascular disease. Such a U-shaped relation between HbA_1c level and vascular events and mortality in the total diabetic population was indeed suggested in several cohort studies (15–17). As hypoglycemia is associated with severe cardiovascular events and arrhythmia (23), the left arm of this U shape might be caused by the occurrence of hypoglycemia. Analyses of the data from the ADVANCE trial (8) suggested that the increased all-cause mortality in the intensive treatment group could be associated with the occurrence of hypoglycemia, although the causality of this observation is uncertain. An explanation for the different relation in patients with and without vascular disease could thus be that patients with established vascular disease might be more susceptible to the detrimental effects of hypoglycemia. However, analyses in the current study did not indicate the presence of a U-shaped relation between HbA_1c level and new cardiovascular events. Our findings probably differ from earlier cohort studies because of the inclusion of patients without vascular disease and with different types of vascular disease besides coronary artery disease.

Another explanation for the different relation between HbA_1c level and new cardiovascular events in patients with type 2 diabetes with and without vascular disease could be that in patients with established vascular disease, hyperglycemia is not the key factor for progressive vascular damage. Factors such as hypertension (24,25) and dyslipidemia (26,27) have been shown to be strongly related to new cardiovascular events in patients with type 2 diabetes and in patients with already established vascular disease. Thus, if hypertension and dyslipidemia are important modifiable risk factors contributing to the pathogenesis of cardiovascular events in patients with type 2 diabetes in general, these risk factors may be even more important in patients with type 2 diabetes in whom a cardiovascular event developed before inclusion in our study. In patients with diabetes, the pathogenesis of vascular disease is at least in part intrinsically different from that of vascular disease in patients without diabetes, as medial vascular calcification or Mönckeberg medial sclerosis is often found in patients with diabetes (28). This difference in pathogenesis could therefore translate into a difference in the most important risk factors for new cardiovascular disease.

The interpretation of the strong relation between HbA_1c level and all-cause mortality in patients with type 2 diabetes with long-term vascular disease and the absence of this relation in patients with short-term vascular disease is not obvious. An explanation might be that HbA_1c, as a marker of glycemic regulation, is a proxy of overall condition or frailty and therefore singles out the patients with the poorest health status, especially in those patients with a longer duration of vascular disease.

The chief strengths of this study include the prospective design and large number of participants, both with and without cardiovascular disease. Because of the substantial follow-up period and large cohort size, there was a relatively high number of events. Furthermore, the risk of bias in this study was reduced because of the completeness of the data.

Several limitations of this study need to be addressed. As the SMART study is a single-center cohort study in an academic hospital, it may be questioned whether this cohort is a representation of the total population of patients with type 2 diabetes. It should be noted that the current cohort contains a broad scope of patients with type 2 diabetes with and without vascular disease representing clinical practice. Despite the relatively large number of participants in this cohort, a small number of end points in some groups could have resulted in insufficient power. This is most likely the case in the patients with type 2 diabetes without vascular disease at baseline because of the small number (n = 531) in this group. Furthermore, only baseline HbA_1c level was used for the analyses in the current study, while the median duration of follow-up in this study was 6 years. While a possible variation in HbA_1c level during follow-up could theoretically change the relations, this is not taken into account in the current study. The lack of statistically significant interaction of the presence of vascular disease at baseline needs to be taken into account when interpreting these results. Although the point estimates are clearly different between the two groups, no significant interaction was observed. The difference in point estimates may, of course, be due to chance, but we think that, in light of the relatively wide CIs, the absence of statistically significant interaction could very well be due to insufficient power in this study. Further studies are needed to support our hypothesis of a differential effect of HbA_1c level on cardiovascular events between patients with and without vascular disease.

As we only studied patients with type 2 diabetes, our findings cannot be extrapolated to patients with type 1 diabetes, a population in which strict control has been shown to be associated with fewer cardiovascular events (2,29). Last, our study only takes into account macrovascular complications of type 2 diabetes. As HbA_1c level has been shown to be strongly related to microvascular complications (2,12,29,30), the importance of strict glycemic control should not be devaluated completely in patients with type 2 diabetes and vascular disease. Microvascular complications should be taken into account when setting individualized therapeutic targets in patients with type 2 diabetes, including those with vascular disease. Nevertheless, our findings are important for generating hypotheses that may eventually lead to more tailored treatment in this very high-risk population. Further studies are therefore needed to establish whether the presence of vascular disease influences the relation between HbA_1c level and cardiovascular events.

HbA_1c level is related to cardiovascular events, and no interaction of the presence or absence of vascular disease on this relation was observed. The effect of baseline HbA_1c level on all-cause mortality was of similar magnitude in patients with and without vascular disease, although the relation was not statistically significant in patients without vascular disease.

Acknowledgments. For their contribution, the authors thank the SMART research nurses; R. van Petersen (data manager); H. Pijl (vascular manager); and the following members of the SMART Study Group: P.A. Doevendans, Department of Cardiology; A. Algra, Y. van der Graaf, D.E. Grobbee, and G.E.H.M. Rutten, Julius Center for Health Sciences and Primary Care; L.J. Kappelle, Department of Neurology; T. Leiner, Department of Radiology; F.L. Moll, Department of Vascular Surgery; and F.L.J. Visseren, Department of Vascular Medicine.

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

Author Contributions. G.K. researched the data and wrote the manuscript. Y.v.d.G., J.v.d.L., and F.L.J.V. contributed to the methodology and reviewed and edited the manuscript. H.M.W.N., G.J.d.B., and L.J.K. reviewed and edited the manuscript. J.W. wrote, reviewed, and edited the manuscript. J.W. 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.