Author + information
- Binita Shah, MD, MS∗ (, )
- Joseph Burdowski, MD,
- Iryna Lobach, PhD,
- Eugenia Gianos, MD and
- Steven P. Sedlis, MD
- ↵∗Veterans Affairs New York Harbor Health Care System, New York University School of Medicine, 227 East 30th Street, 835, New York, New York 10016
Periprocedural hyperglycemia is a potentially reversible cause of poor outcomes in patients with diabetes mellitus (DM) undergoing percutaneous coronary intervention (PCI) (1,2). However, simple preventive strategies such as continuing long-acting glucose-lowering medications are not routinely implemented due to concern for hypoglycemia (3). We recently demonstrated the efficacy of continuing rather than holding long-acting glucose-lowering medications in patients with DM undergoing coronary angiography with possible PCI (4). In this study, we sought to develop a risk score to predict hyperglycemia at the time of coronary angiography in patients with DM to help target a high-risk group of patients for whom the benefits of treatment may outweigh the risks of hypoglycemia.
In the periprocedural glycemic control trial at the Manhattan Veterans Affairs Hospital, patients with DM were randomized to continue or hold long-acting glucose-lowering medications before coronary angiography (4). The hold group (n = 86) serves as the development cohort, and details regarding this cohort were previously described (4).
Patients undergoing PCI between November 2010 and February 2012 participated in a prospective registry at the New York University Langone Medical Center. Those with pharmacologically treated DM and recorded blood glucose at the time of the procedure (n = 168) served as the validation cohort. A nurse educator provided telephone instructions to hold glucose-lowering medications before the procedure during this time per routine care. Although variables of interest in the validation cohort were prospectively collected, fasting time and periprocedural glucose were collected by retrospective review of the electronic procedural charts.
Predictors of periprocedural hyperglycemia, defined as glucose ≥140 mg/dl, were determined in the development cohort using multivariable logistic regression. Each independent predictor was assigned a weighted integer based on an estimate of the log odds ratio. Summation of integers determined risk score. Trends in proportions of hyperglycemia across tertiles of risk score were examined using the Cochran Armitage trend test. Proportions of hyperglycemia in the development cohort versus validation cohort in each tertile risk score were examined using a 2-sample test for equality of proportions with continuity correction.
The proportion of patients with periprocedural hyperglycemia in the development cohort was 47.7% (n = 41). Although the patients in the development cohort were all men and 73.3% of white race with a median age of 66 years, patients in the validation cohort were 68.5% men and 46.4% of white race with a median age of 67 years. A little less than half of patients were taking sulfonylurea agents (45.3% development cohort, 40.5% validation cohort) and long-acting insulin (40.7% development cohort, 33.3% validation cohort), whereas more than half were receiving metformin therapy (61.6% development cohort, 67.3% validation cohort). A minority of patients was on thiazoladinediones (4.7% development cohort, 14.9% validation cohort) or sitagliptin (0% development cohort, 29.2% validation cohort). The median (interquartile range) random glucose and hemoglobin A1c was 145.5 mg/dl (120 to 186.3 mg/dl) and 7.3% (6.8% to 8%) in the development cohort, respectively, and 158.8 mg/dl (109 to 184.5 mg/dl) and 7.6% (6.5% to 8.4%) in the validation cohort, respectively. Finally, the median (interquartile range) fasting time was 17.3 h (14.4 h to 19.3 h) in the development cohort and 10.6 h (8.3 h to 12.8 h) in the validation cohort.
Independent predictors of hyperglycemia in the development cohort, with assigned integer, are shown in Table 1. Increasing risk score was associated with greater proportion of patients with hyperglycemia (development cohort risk score <6: 11%; score 6 to 9: 31.25%; score ≥9: 68%; p < 0.001; validation cohort risk score <6: 24.44%; score 6 to 10: 44.68%; score ≥10; 73.24%; p < 0.001). There was no significant difference between the proportion of patients with hyperglycemia in the development and validation cohorts in each risk tertile (first, p = 0.35; second, p = 0.52; third, p = 0.67). The risk score model displayed good discriminative power in the development cohort (C-statistic = 0.75; 95% confidence interval: 0.66 to 0.85) and the validation cohort (C-statistic = 0.74; 95% confidence interval: 0.66 to 0.81) with a similar effect in both cohorts (interaction p value = 0.23).
Hyperglycemia contributes to endothelial dysfunction and inflammation (5). Acute hyperglycemia increases platelet activation despite the use of antiplatelet agents, enhances thrombin formation, and impairs fibrinolysis (6). Among patients with DM undergoing PCI, periprocedural hyperglycemia is associated with target vessel revascularization, myocardial injury, and contrast-induced renal injury, making periprocedural hyperglycemia an important risk factor to investigate further (2). We have shown that continuing oral glucose-lowering drugs in patients undergoing coronary angiography with possible PCI reduces the incidence of hyperglycemia and normalizes platelet function (4).
Event rates are too small to identify predictors of hypoglycemia in patients with DM undergoing coronary angiography (4). However, the risk of hyperglycemia in this population can be evaluated using readily available clinical data and may be used to guide the intensity of periprocedural glycemic control as well as define a high-risk population that may be appropriate for future studies of periprocedural glycemic control.
The data analysis and statistical support were provided by New York University School of Medicine Cardiovascular Outcomes Group. The authors thank Yu Guo, MA, of the New York University School of Medicine for the statistical analyses presented here. They also thank Madeline Stecy, Princeton University, for data collection.
Please note: Dr. Shah was partially funded by a NIH/NHLBI grant (T32HL098129 in 2012; UL1 TR000038 in 2013). Dr. Shah has received research grants from Guerbet, Siemans, and Takeda. Dr. Sedlis has received a research grant from Takeda. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Marfella R.,
- Sasso F.C.,
- Siniscalchi M.,
- et al.
- Gresele P.,
- Guglielmini G.,
- De Angelis M.,
- et al.