Author + information
- Received July 5, 2016
- Revision received October 16, 2016
- Accepted October 18, 2016
- Published online January 23, 2017.
- Giuseppe Patti, MDa,∗ (, )
- Markus Lucerna, PhDb,
- Ilaria Cavallari, MDa,
- Elisabetta Ricottini, MDa,
- Giulia Renda, MD, PhDc,
- Ladislav Pecen, PhDd,
- Fabio Romeo, MDe,
- Jean-Yves Le Heuzey, MDf,
- Josè Luis Zamorano, MD, PhDg,
- Paulus Kirchhof, MDh,i and
- Raffaele De Caterina, MD, PhDc,j,∗∗ ()
- aDepartment of Cardiovascular Sciences, Campus Bio-Medico University of Rome, Rome, Italy
- bDaiichi-Sankyo Europe, Munich, Germany
- cG. d’Annunzio University of Chieti, Chieti, Italy
- dInstitute of Informatics, Academy of Sciences of Czech Republic, Prague, Czech Republic
- eDaiichi-Sankyo Italy, Rome, Italy
- fCardiology and Arrhythmology, Georges Pompidou Hospital, René Descartes University, Paris, France
- gDepartment of Cardiology, University Hospital Ramòn y Cajal, Madrid, Spain
- hInstitute of Cardiovascular Sciences, University of Birmingham and Sandwell and West Birmingham Hospitals and University Hospitals Birmingham NHS Trust, Birmingham, United Kingdom
- iHospital of the University of Münster, Münster, Germany
- jFondazione G. Monasterio, Pisa, Italy
- ↵∗Reprint requests and correspondence:
Dr. Giuseppe Patti, Department of Cardiovascular Sciences, Campus Bio-Medico University of Rome, Via Alvaro del Portillo, 200, 00128 Rome, Italy.
- ↵∗∗Dr. Raffaele De Caterina, University Cardiology Division, “G. d'Annunzio” University – Chieti, Ospedale SS. Annunziata, Via dei Vestini, I-66013 Chieti, Italy.
Background Diabetes is a known risk predictor for thromboembolic events in patients with atrial fibrillation (AF), but no study has explored the prognostic weight of insulin in this setting.
Objectives This study evaluated the differential role of insulin versus no insulin therapy on thromboembolic risk in patients with diabetes and AF.
Methods We accessed individual patient data from the prospective, real-world, multicenter, PREFER in AF (European Prevention of thromboembolic events-European Registry in Atrial Fibrillation). We compared the rates of stroke/systemic embolism at 1 year according to diabetes status (no diabetes, diabetes without insulin therapy, diabetes on insulin therapy).
Results In an overall population of 5,717 patients, 1,288 had diabetes, 22.4% of whom were on insulin. For patients with diabetes who were on insulin, there was a significantly increased risk of stroke/systemic embolism at 1 year versus either no diabetes (5.2% vs. 1.9%; hazard ratio: 2.89; 95% confidence interval: 1.67 to 5.02; p = 0.0002) or diabetes without insulin treatment (5.2% vs. 1.8%; hazard ratio: 2.96; 95% confidence interval: 1.49 to 5.87; p = 0.0019). Notably, rates of stroke/embolism were similar in patients with diabetes not receiving insulin versus patients without diabetes (hazard ratio: 0.97; 95% confidence interval: 0.58 to 1.61; p = 0.90). The selective predictive role of insulin-requiring diabetes was independent of potential confounders, including diabetes duration, and was maintained in various subpopulations, including the subgroup receiving anticoagulant therapy.
Conclusions In this cohort of anticoagulated patients with AF, the sole presence of diabetes not requiring insulin did not imply an increased thromboembolic risk. Conversely, insulin-requiring diabetes contributed most, if not exclusively, to the overall increase of thromboembolic risk in AF.
Estimation of thromboembolic risk is crucial in patients with atrial fibrillation (AF) in order to perform an accurate stratification of risk during follow-up and establish optimal therapeutic strategies. Diabetes mellitus has been considered an independent risk factor for thromboembolic events in patients with AF (1) and this has led to inclusion of diabetes in the CHADS2 (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack) score (2) and the more recent CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke, transient ischemic attack, or thromboembolism, vascular disease, age 65–74 years, sex category [female]) score (3). Patients with diabetes mellitus have a prothrombotic state because of changes in primary (platelet aggregation and vascular function) and secondary (coagulation and fibrinolysis) hemostasis; this is particularly enhanced in those with long-lasting disease receiving insulin therapy (4). Here, low-grade inflammation, increased levels of coagulation factors, impairment of fibrinolysis, oxidative stress, and reduced expression of protective endothelial factors have been indicated as responsible for these prothrombotic changes (4).
This is the basis for hypothesizing a stronger predictive role of diabetes requiring insulin therapy compared with less severe forms of diabetes, usually not requiring insulin, on AF-related thromboembolic risk. To date, no study has explored the differential prognostic weight of diabetes on insulin therapy versus no insulin therapy on thromboembolic events in patients with AF. We explored this issue in a recent multicenter, European AF registry.
We accessed individual patient-level data from the PREFER in AF registry (Prevention of thromboembolic events-European Registry in Atrial Fibrillation) (5). This prospective, observational, real-world registry enrolled 7,228 patients with AF from 461 hospitals in 7 European countries (Austria, France, Germany, Italy, Spain, Switzerland, and the United Kingdom). The first patient was enrolled in January 2012 and the last follow-up visit occurred in January 2014. Inclusion criteria were age ≥18 years; written informed consent for study participation; and history of AF within the preceding 1 year, as demonstrated by an electrocardiogram or by an implanted pacemaker/defibrillator. Patients were included irrespective of the type of AF. To reduce selection bias, patients were consecutively enrolled at each site, with no explicit exclusion criteria.
The study design consisted of a baseline clinical evaluation at the time of enrollment and at 1-year follow-up. Demographic data, clinical characteristics, risk factors, and treatment modalities were collected at baseline; at this time, documentation related to previous AF episodes and use of AF-related antithrombotic therapy within 1 year was also inspected, if needed. The follow-up was performed by office visit at 12 ± 2 months. For the purpose of this study, we only included patients with a complete CHA2DS2-VASc score evaluation and both baseline and 1-year follow-up visits. Only documented stroke or systemic embolism were considered relevant efficacy endpoints, with the date of any event being after the baseline visit.
Individual data were entered into an electronic case report form including various plausibility checks for the considered variables. Furthermore, on-site verification of source data was performed in approximately 5% of the centers. The study management was overseen by a scientific Steering Committee; the registry was sponsored by Daiichi-Sankyo Europe GmbH (Munich, Germany) via a contract research organization (SSS International Clinical Research GmbH, Munich, Germany) coordinating various local/national contract research organizations.
Definitions and endpoints
For the purpose of this study, patients with diabetes were separately considered if they were or were not on insulin therapy (6). The primary study endpoint was the incidence of stroke/systemic embolism at 1-year follow-up according to diabetes status (no diabetes, noninsulin-requiring diabetes, or insulin-requiring diabetes). Stroke and systemic embolism were defined following the ENGAGE AF–TIMI 48 (Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis In Myocardial Infarction 48) definitions (7) as follows:
• Stroke: abrupt onset of a focal neurological deficit, generally distributed in the territory of a single brain artery (including the retinal artery), that is not attributable to an identifiable nonvascular cause (i.e., brain tumor or trauma). The deficit must either be characterized by symptoms lasting >24 h or cause death within 24 h of symptom onset. Stroke definition used in ENGAGE and in our study reflects the Statement for Healthcare Professionals From the American Heart Association/American Stroke Association (8) that incorporates the World Health Organization definition of stroke (9).
• Systemic embolic event: abrupt episode of arterial insufficiency with clinical or radiological documentation of arterial occlusion in the absence of other likely mechanisms (e.g., atherosclerosis, instrumentation); venous thromboembolism and pulmonary embolism were also included in this outcome measure. Arterial embolic events involving the central nervous system (including the eye) were not considered as systemic embolism.
For categorical variables, absolute and percentage frequencies (n, %) are presented. For continuous variables, mean ± SD is presented. For the analyses of the time-to-stroke/systemic embolism, the Cox proportional hazard regression model was used, with diabetes status as a fixed effect. The hazard ratio (HR), 95% confidence interval (CI), and corresponding p values are presented.
These analyses were repeated for different subgroups of patients based on the demographic/clinical characteristics indicated in Table 1. In addition, these characteristics were added as single covariates to the model. Comparisons of all demographic/clinical characteristics for diabetes status were executed by means of a logistic regression model presenting the odds ratio, 95% CI, and corresponding p value. All analyses are not confirmatory, but purely descriptive/exploratory. All statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, North Carolina).
From the overall PREFER in AF population (N = 7,228), a total of 816 patients had no 1-year follow-up visit; therefore, the full analysis set consisted of 6,412 patients, 695 of whom were excluded because of lack of information on stroke/systemic embolic events and/or no availability of CHA2DS2-VASc scores and/or no information on diabetes status (Figure 1). Thus, a total of 5,717 patients were included in this subanalysis. Prevalence of thromboembolic risk factors and different antithrombotic therapies in patients included in this analysis was consistent with the overall PREFER in AF population (data not shown). Among these 5,717 patients, a total of 1,288 (22.5%) had diabetes mellitus. Furthermore, 288 of these 1,288 (22.4%) patients with diabetes were on insulin treatment. Patients with diabetes, irrespective of insulin therapy status, had an increased prevalence of systemic hypertension, congestive heart failure (CHF), prior transient ischemic attack/stroke/thromboembolism, vascular disease, chronic renal impairment, left atrial enlargement, chronic obstructive pulmonary disease (COPD), and body mass index >30 kg/m2 compared with patients without diabetes (Table 1). Patients receiving insulin showed higher percentages of CHF, vascular disease, chronic renal impairment, COPD, and body mass index >30 kg/m2 versus those with noninsulin-requiring diabetes. Of note, in our study, only 18 patients had type 1 diabetes and only 1 patient experienced a thromboembolic event during follow-up.
We also evaluated the prevalence of different antithrombotic strategies in the various subgroups (Table 1). Compared with patients without diabetes, those on insulin treatment had higher use of vitamin K antagonists (VKA) plus antiplatelet therapy (16.7% vs. 9.6%; p = 0.0002) at baseline and higher use of VKAs (71.5% vs. 62.9%; p = 0.0036) at 1 year. No antithrombotic therapy was less frequent in patients with diabetes on insulin both at baseline and at 1 year (2.4% vs. 6.4% in patients without diabetes, p = 0.0093, and 5.6% vs. 9.6%, p = 0.0238, respectively). Antithrombotic therapy was similar in patients with diabetes with and without insulin, with the exception of a higher prevalence of VKAs plus an antiplatelet agent at baseline in the former (16.7% vs. 11.1%; p = 0.012). In the comparison between patients without diabetes and patients with noninsulin-requiring diabetes, the latter more frequently were given VKAs only at baseline (69.8% vs. 66.2%; p = 0.03) and less frequently received antiplatelet treatment and no antithrombotic drug both at baseline and at 1 year.
In the overall population, the incidence of stroke/systemic embolism at 1 year was 2.0 per 100 patients/year. Insulin-requiring diabetes was associated with a higher risk of stroke/systemic embolism versus both no diabetes (5.2 per 100 patients/year vs. 1.9 per 100 patients/year; HR: 2.89; 95% CI: 1.67 to 5.02; p = 0.0002) and noninsulin-requiring diabetes (5.2 per 100 patients/year vs. 1.8 per 100 patients/year; HR: 2.96; 95% CI: 1.49 to 5.87; p = 0.0019) (Central Illustration). Rates of stroke/systemic embolism were not different in patients with diabetes not receiving insulin and in patients without diabetes (HR: 0.97; 95% CI: 0.58 to 1.61; p = 0.90). Adjustment for potential confounders provided similar results (Table 2). After the addition of various risk factors as covariates to the Cox proportional hazard regression model, the correlation between insulin-requiring diabetes and the higher occurrence of thromboembolic events remained always significant, with HRs ranging from 2.60 to 3.52 (Table 3).
In the comparison between insulin-requiring diabetes and noninsulin-requiring diabetes, of 15 tested covariates, 2 had statistically significant interactions with the group; in particular, the relative increase of thromboembolic events related to insulin therapy was higher in patients with CHF (vs. those without) and in patients receiving antithrombotic therapy at baseline. Conversely, the HR for the comparison of patients with no diabetes versus patients with noninsulin-requiring diabetes remained consistently nonsignificant (Table 3).
The prevalence of sustained (persistent or permanent) AF tended to be higher in patients on insulin treatment: 80% on insulin versus 76% in patients with noninsulin-requiring diabetes, compared with 67% of patients without diabetes. However, adjustment for the type of AF did not change the overall study results, in particular regarding stroke/systemic embolism: insulin-requiring diabetes mellitus versus no diabetes (HR: 2.83; 95% CI: 1.60 to 5.03; p = 0.0004); stroke/systemic embolism with insulin-requiring diabetes mellitus versus diabetes without insulin therapy (HR: 2.98; 95% CI: 1.48 to 6.02; p = 0.0023); and noninsulin-requiring diabetes mellitus versus no diabetes (HR: 0.98; 95% CI: 0.59 to 1.63; p = 0.94).
We collected additional patient-level data on diabetes duration, daily insulin dose, presence/absence of microvascular complications, and use of oral glucose-lowering agents in a subgroup of 344 patients with diabetes (i.e., 27% of the overall study population with diabetes). The risk profile of these 344 patients was similar to that of the remaining population of patients with diabetes: age 73.3 ± 9.2 years versus 72.7 ± 8.6 years (p = 0.31), female sex 36% versus 37% (p = 0.66), and mean CHA2DS2-VASc score 4.7 ± 1.6 versus 4.6 ± 1.6 (p = 0.42). The duration of diabetes was higher in patients with diabetes on insulin compared with those not receiving insulin (12.8 ± 8.2 years vs. 9.2 ± 6.7 years; p = 0.0003), but the HR of stroke/systemic embolism with insulin therapy, adjusted for duration of diabetes, remained significant (HR: 8.72; 95% CI: 2.89 to 26.33; p = 0.0001).
The total daily insulin dose was similar in patients with versus without stroke/systemic embolism (37.8 ± 9.9 IU vs. 38.5 ± 26.2 IU; p = 0.22), and no relationship between insulin dose and the occurrence of thromboembolic events was observed (HR: 1.00; 95% CI: 0.98 to 1.02; p = 0.94). We found a significantly higher risk of stroke/systemic embolism in patients with at least 1 microvascular complication of diabetes (retinopathy, neuropathy, or nephropathy): HR: 9.27; 95% CI: 2.07 to 41.41; p = 0.0036. We attempted an analysis of different therapies in noninsulin-requiring diabetes (e.g., diet vs. oral antidiabetic agents, or among various classes of oral antidiabetic drugs), but these analyses were precluded by the overall low rate of thromboembolic events observed in these subgroups.
We further evaluated the risk of stroke/systemic embolism in patients without diabetes, with diabetes not receiving insulin, and in those with insulin-requiring diabetes according to different subgroups, including presence or absence of female sex, age ≥75 years, CHF, systemic hypertension, previous transient ischemic attack/stroke, any vascular disease, COPD, chronic renal impairment, body mass index >30 kg/m2, CHA2DS2-VASc score >1, coronary artery disease, peripheral artery disease, or use of anticoagulant therapy. The highest incidence of thromboembolic events continued to be seen in patients with diabetes on insulin treatment. However, there was no significant difference in thromboembolic events comparing patients with diabetes without insulin treatment with those without diabetes and this absence of any difference for these 2 groups was maintained across various subpopulations (Figure 2). Of note, among patients with diabetes on insulin therapy, the rate of stroke/systemic embolism was even high in those receiving any anticoagulant therapy at baseline (5.1 per 100 patients/year vs. 6.1 per 100 patients/year in those without anticoagulation); the increased incidence of thromboembolic complications in patients with insulin-requiring diabetes was irrespective of the use of anticoagulant therapy (patients receiving any anticoagulant treatment: 5.1 per 100 patients/year in patients with insulin-requiring diabetes vs. 1.6 per 100 patients/year in patients with noninsulin-requiring diabetes and 1.8 per 100 patients/year in patients with diabetes; patients without anticoagulant therapy: 6.1 vs. 3.5 vs. 1.9 per 100 patients/year).
A total of 4,354 patients had no diabetes or noninsulin-requiring diabetes and a CHA2DS2-VASc score >1; the occurrence of stroke/systemic embolism at 1 year in such patients was 2.0%. All patients with diabetes on insulin therapy had a CHA2DS2-VASc score >1 and showed an annual stroke/embolism rate of 5.2 per 100 patients/year (p = 0.0005).
In this analysis of individual patient data from the prospective PREFER in AF registry, patients with diabetes on insulin therapy had a significantly higher risk of stroke/systemic embolism at 1 year versus both patients without diabetes and patients with noninsulin-requiring diabetes; yet, for people with diabetes not treated with insulin, there was no significantly increased risk.
The proportion of patients with diabetes in our population was 22.5%, of whom 22.4% were insulin-treated; this prevalence is similar to that observed in other contemporary registries of patients with AF (10). Of note, a 40% relative increase in the risk of AF development and progression has been demonstrated in patients with versus those without diabetes (11), and has been related to electrical and structural atrial remodeling, changes in the autonomic response, atrial inflammation, and oxidative stress (12).
A wide range (from 3.6% to 8.6%) of annual incidence of thromboembolic events has been reported in patients with diabetes and AF (11,13); this large variability reflects differences in study designs, definitions of outcome measures, patients’ baseline risk profile, concomitant therapies, and types of populations included. Previous large studies have found that patients with AF with coexisting diabetes present a significantly higher risk of thromboembolic events compared with those without. In a previous meta-analysis on the topic, including 7 studies and >12,000 patients, a 70% relative increase in risk has been observed in patients with diabetes (13). To date, however, no study had separately and independently quantified the annual rates of AF-related thromboembolic events in patients with diabetes according to insulin treatment.
The surprising finding of our study was the strikingly similar incidence of thromboembolic events at 1 year in patients with diabetes but without insulin treatment compared with patients without diabetes. The result was consistent for various analyses performed, even after adjusting for both clinical confounders and concomitant antithrombotic therapy. Of note, the event rate was similar in patients without diabetes and in patients with diabetes not receiving insulin despite the latter having a higher thromboembolic risk profile (i.e., older age, higher prevalence of hypertension, CHF, previous cerebrovascular events, vascular disease, and chronic renal failure). Thus, according to our data, the sole presence of diabetes does not imply an increased thromboembolic risk in patients with AF.
Conversely, patients with insulin-requiring diabetes had an approximately 2.5-fold higher risk of stroke or systemic embolism at 1 year compared with patients without diabetes and with patients with noninsulin-requiring diabetes. Of note, this higher risk was more pronounced between 6 months and 1 year of follow-up. A clustering of risk factors likely contributes to this heightened risk, because patients with diabetes on insulin treatment had a longer diabetes duration, and a higher prevalence of cardiovascular risk factors, CHF, COPD, and renal impairment than patients without diabetes or those with diabetes not requiring insulin. The association between insulin-requiring diabetes and thromboembolic events was independent of the type of AF and of other possible confounding factors; this association was also maintained in various subgroups, including those patients receiving anticoagulant therapy.
We observed no relationship of daily insulin dose and thromboembolic risk. We cannot exclude a type II error in these results, and it is possible that the daily doses of insulin (marking a diabetes of particular severity) could be related to outcomes in larger cohorts or with a longer follow-up. Of note, patients with diabetes with microvascular complications (retinopathy, neuropathy, or nephropathy) experienced a significantly increased incidence of thromboembolic events. Importantly, the selectively increased thromboembolic risk of patients receiving insulin (with no apparent increase in risk in the other set of patients with diabetes) was independent of all potential confounders from parameters collected in the PREFER in AF Registry, including duration of diabetes (14).
Similar data supporting a differential prognostic role of diabetes with versus without insulin therapy have been described in at least 1 other setting; in particular, an analysis from SHIFT (Systolic Heart Failure treatment with the If inhibitor ivabradine Trial) demonstrated that patients with chronic systolic heart failure (15) showed no increased incidence of cardiovascular death or hospitalization for worsening heart failure in patients with noninsulin-requiring diabetes compared with patients without diabetes, and a significant 33% higher risk of this outcome measure in patients with diabetes on insulin compared with those not on insulin.
Therefore, according to our data, it is the need for insulin therapy, rather than the presence of diabetes per se, that seems to be an independent factor affecting the occurrence of AF-related stroke/systemic embolism during follow-up. Results of this study might expand and strengthen observational data from certain investigations suggesting no overall increase of thromboembolic risk in patients with diabetes (16–20); a different prevalence of patients receiving insulin (generally not reported in most studies) may at least in part explain the important variability in the reported annual rates of thromboembolic events among patients with diabetes and the variable degree of increase in the thromboembolic risk by diabetes mellitus in the various studies.
Several pathophysiological mechanisms might explain these findings. In patients with diabetes mellitus there is a hypercoagulable state, and this is particularly evident in those with long-lasting disease receiving insulin therapy. In the latter, an increase in platelet reactivity and platelet turnover has been described, with consequently more pronounced platelet activation (21). Moreover, a high inflammatory status and oxidative stress cause endothelial dysfunction, with higher expression of adhesion molecules, reduced release of nitric oxide/prostacyclin, and increased production of endothelin-1 (4,22–24). Patients with diabetes on insulin treatment also showed increased levels and/or activity of various coagulation factors, including tissue factor, factor VII, von Willebrand factor, and fibrinogen, as well as enhanced thrombin generation (21,25,26). Finally, lower tissue-plasminogen activator activity, higher levels of type 1 plasminogen activator inhibitor (27,28), and higher levels of incorporation of the C3 complement component in the clot (29) have been demonstrated in such patients, leading to impaired fibrin clot lysis. The presence of insulin treatment is therefore certainly a marker for more advanced disease. However, insulin by itself might trigger some of the disease features, including atherosclerosis (30). Although the precise mechanisms triggering changes in coagulation in patients with diabetes receiving insulin therapy are not completely known, possible mechanisms include chronic exposure to high glucose levels, increased levels of advanced glycosylation end-products, and direct effects of exogenous insulin, providing pathologically high levels of insulin in the setting of insulin resistance, as occurs in all patients with type 2 diabetes receiving insulin (30,31).
This work has strengths in being a prospective analysis of patients with AF who received a complete baseline assessment and underwent a planned follow-up visit at 1 year with accurate evaluation of outcome measures. Limitations are that we could not establish the thromboembolic risk of the untreated population included or the risk associated with specific antithrombotic therapies. However, the crude increase in thromboembolic risk that occurred in the presence of insulin-requiring diabetes is probably even higher than that detected in our investigation. In fact, patients on insulin had a higher prevalence of VKA use and less frequently received no antithrombotic drug than those without diabetes; similarly, they more often were given VKAs plus antiplatelet drugs than those with diabetes without insulin. Thus, it is unlikely that we overestimated the risk of patients with insulin-requiring diabetes in our study.
Furthermore, residual confounding cannot be excluded and, because of the size of the population, we could not stratify the thromboembolic risk of patients with diabetes on insulin therapy according to different CHA2DS2-VASc scores (1 vs. >1). The issue of whether the relationship between the type of diabetes and thromboembolic risk was irrespective of the duration of diabetes was evaluated in approximately one-fourth of the population with diabetes within PREFER in AF (n = 344 or 27%); in this subset, the HR of stroke/systemic embolism in patients with insulin therapy, compared with patients not on insulin, remained significant even after adjusting for the duration of diabetes.
Therefore, main results of this sensitivity analysis continued to support 1 main conclusion of the paper, that insulin-requiring diabetes is a much worse condition than noninsulin-requiring diabetes. Importantly, the risk profile of those 344 patients providing additional data on diabetes duration was similar to that of the remaining population of patients with diabetes. We can therefore reasonably assume that the results of this further analysis were not affected by the selection of patients, and no bias was introduced in this secondary analysis.
Finally, a nonuniform definition of diabetes mellitus might have been used in the study population according to local practices, and more important, we have no data on the specific criteria for initiating insulin therapy and on glycemic control during follow-up. However, we consider it unlikely that use of nonuniform definitions of diabetes and criteria for initiating insulin therapy might have affected the study results, inasmuch as the physicians in the Western European countries participating in PREFER in AF are generally accustomed to contemporary, international guidelines for defining diabetes and initiating insulin treatment. Any such limitations should not, however, affect the main finding of our study, which is not only the higher risk of the insulin-requiring diabetes, likely clustering with a higher severity of diabetes, but rather the very low risk of noninsulin-requiring diabetes. This indicated for the first time a quite dichotomous behavior of the population with diabetes and AF as to thromboembolic risk according to the use or lack of use of insulin. Of note, results of our investigation apply essentially to patients with type 2 diabetes, who represented 98.6% of the population with diabetes studied, and it may be that insulin provision in type 1 diabetes, in the absence of insulin resistance, is not associated with increased thromboembolic risk. We cannot completely exclude the possibility that patients with noninsulin-requiring diabetes with AF are at somewhat higher thromboembolic risk than patients without diabetes, and that we could not demonstrate this difference because of a type II statistical error. However, the absence of any directional trend toward higher risk in patients with noninsulin-requiring diabetes argued against such possibility, but certainly further confirmatory research should address this important issue in studies with a larger sample size.
Our findings robustly indicated that insulin-requiring diabetes, essentially type 2 diabetes, largely contributed to the overall increase of thromboembolic risk in AF, but the mere presence of diabetes without insulin treatment did not apparently convey a negative prognostic value. Our investigation still supported that early diabetes has lower thromboembolic risk than later diabetes and, in our population, the reduced thromboembolic risk in patients without insulin treatment may be caused by the shorter duration of the disease.
Our results may have implications in the assessment of thromboembolic risk in the AF population with diabetes and might have therapeutic implications that need to be explored in further dedicated intervention studies.
COMPETENCY IN MEDICAL KNOWLEDGE: Although a history of diabetes mellitus has been identified as a risk factor for thromboembolism in patients with AF, in a cohort of 1,288 patients with AF and diabetes, only those who required insulin for glycemic control exhibited an increased risk of stroke, whereas the risk in those not receiving insulin was similar to a nondiabetic control group.
TRANSLATIONAL OUTLOOK: Further studies are needed to examine revised stroke risk stratification schemes for patients with AF and to explore other therapeutic implications of insulin dependency in patients with AF.
This analysis of the PREFER in AF registry was initiated by the Thrombosis Exchange Meeting in AF, TEAM in AF, funded and sponsored by Daiichi-Sankyo Europe.
Dr. Patti is on the speaker/consultant/advisory board for Bayer, Boehringer Ingelheim, Bristol-Myers Squibb/Pfizer, Daiichi-Sankyo, Eli Lilly, AstraZeneca, and Merck Sharp & Dohme. Dr. Lucerna is currently an employee of Daiichi-Sankyo Europe. Dr. Renda is on the speaker/consultant/advisory board for Boehringer Ingelheim, Daiichi-Sankyo, and Bayer. Dr. Romeo is currently an employee of Daiichi-Sankyo Italy. Dr. Le Heuzey is on the consultant/conferences/advisory board for Sanofi, Bristol-Myers Squibb/Pfizer, Meda, Boehringer Ingelheim, Merck Sharp & Dohme, Bayer, Servier, AstraZeneca, Novartis, and Daiichi-Sankyo. Dr. Zamorano has received speaker honoraria from Sanofi, Servier, and Daiichi-Sankyo. Dr. Kirchhof has received consulting fees/honoraria from AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi-Sankyo, 3M Medica, MEDA Pharma, Medtronic, Merck, Otsuka, Pfizer, Sanofi, Servier, Siemens, Takeda, Forschungsunterstützung von Cardiovascular Therapeutics, European Union, Fondation LeDucq, German Federal Ministry for Education and Research, German Research Foundation, OMRON, and St. Jude Medical. Dr. de Caterina has received fees, honoraria, and research funding from Sanofi, Boehringer Ingelheim, Bayer, Bristol-Myers Squibb/Pfizer, Daiichi-Sankyo, Novartis, and Merck Sharp & Dohme. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- congestive heart failure
- confidence interval
- chronic obstructive pulmonary disease
- hazard ratio
- vitamin K antagonist
- Received July 5, 2016.
- Revision received October 16, 2016.
- Accepted October 18, 2016.
- 2017 American College of Cardiology Foundation
- Camm A.J.,
- Lip G.Y.,
- De Caterina R.,
- et al.
- Kirchhof P.,
- Ammentorp B.,
- Darius H.,
- et al.
- The American Diabetes Association Expert Committee
- Sacco R.L.,
- Kasner S.E.,
- Broderick J.P.,
- et al.
- Berisso M.Z.,
- Fabbri G.,
- Gonzini L.,
- et al.,
- ATA-AF Investigators
- Ashburner J.M.,
- Go A.S.,
- Chang Y.,
- et al.
- el Khawand C.,
- Jamart J.,
- Donckier J.,
- et al.
- Rathcke C.N.,
- Persson F.,
- Tarnow L.,
- Rossing P.,
- Vestergaard H.