Abstract
Cardiovascular risk assessment in patients with diabetes relies on traditional risk factors. However, numerous novel biomarkers have been found to be independent predictors of cardiovascular disease, which might significantly improve risk prediction in diabetic patients. We aimed to improve prediction of cardiovascular risk in diabetic patients by investigating 135 evolving biomarkers. Based on selected biomarkers a clinically applicable prediction algorithm for long-term cardiovascular mortality was designed. We prospectively enrolled 864 diabetic patients of the LUdwigshafen RIsk and Cardiovascular health (LURIC) study with a median follow-up of 9.6 years. Independent risk factors were selected using bootstrapping based on a Cox regression analysis. The following seven variables were selected for the final multivariate model: NT-proBNP, age, male sex, renin, diabetes duration, Lp-PLA2 and 25-OH vitamin D3. The risk score based on the aforementioned variables demonstrated an excellent discriminatory power for 10-year cardiovascular survival with a C-statistic of 0.76 (P < 0.001), which was significantly better than the established UKPDS risk engine (C-statistic = 0.64, P < 0.001). Net reclassification confirmed a significant improvement of individual risk prediction by 22% (95% confidence interval: 14–30%) compared to the UKPDS risk engine (P < 0.001). The VILDIA score based on traditional cardiovascular risk factors and reinforced with novel biomarkers outperforms previous risk algorithms.
Similar content being viewed by others
Introduction
There is broad evidence that diabetes mellitus is associated with elevated all-cause and cardiovascular mortality1,2,3. However, the increase in risk associated with diabetes mellitus varies considerably4. Based on novel biomarkers great efforts have been made to improve risk estimation in people with cardiovascular disease5, 6. Nevertheless, an up-to-date risk score specifically addressing people with diabetes is not available. Hence, the most widely used algorithm for people with diabetes mellitus is still the U.K. Prospective Diabetes Study (UKPDS) risk score7. This algorithm only includes two biomarkers, namely glycated hemoglobin and the total cholesterol to high-density lipoprotein cholesterol ratio7. Since the development of the UKPDS score numerous biomarkers have been found to be independent predictors of cardiovascular disease, for example C-reactive protein, Vitamin-D, and renin8.
Hence, the present analyses aimed to improve the prediction of cardiovascular death in diabetic patients using a variety of evolving biomarkers and to show superior prognostic accuracy compared with the UKPDS risk score and compared with single predictors. The initial choice of the biomarkers was based on the multifactorial pathogenesis of cardiovascular disease in patients with diabetes and the routine availability of lab assays to facilitate a rapid clinical translation. The choice of the final biomarker set was based on individual predictive power in the univariate screen. We studied participants of the LUdwigshafen Risk and Cardiovascular health (LURIC) study who had a known personal history of diabetes mellitus or were newly diagnosed with diabetes mellitus9. LURIC is a large prospective observational study which was designed to investigate clinical and biochemical cardiovascular risk factors9. Potential clinical implications of improved risk stratification are a tailored use of preventive strategies with a particular focus on high-risk patients.
Results
Baseline characteristics and univariate survival analysis
We prospectively enrolled 864 diabetic patients of the LURIC study into our study. Forty-three percent of all patients (n = 369) died during a median follow-up of 9.6 years (IQR: 5.5–10.4) corresponding to 6837 overall person-years of follow-up. Sixty-eight percent of deaths in patients were due to cardiovascular causes. Thirty-five percent of patients experienced myocardial infarction before study inclusion. Detailed baseline characteristics of the study population are displayed in Supplemental Table 1.
In the univariate analysis, the strongest adverse risk factors for long-term outcome in diabetic patients were NT-proBNP with a crude hazard ratio (HR) of 2.04 (95% confidence interval [CI] 1.82–2.28; P < 0.001) per 1-SD increase and age (crude HR 1.51; 95% CI 1.35–1.70; P < 0.001; Table 1). 25-OH vitamin D3 had a strong inverse effect on mortality with a HR of 0.68 per 1-SD increase (95% CI 0.61–0.75 P < 0.001; Table 1).
Bootstrapping results and multivariable survival analysis
Variables for the multivariate model were selected using a bootstrapping resampling procedure. NT-proBNP, age, male sex, renin, diabetes duration, Lp-PLA2 and 25-OH vitamin D3 were selected as significant mortality predictors for the final multivariate model (Fig. 1). In this multivariable model male sex and NT-pro-BNP were the strongest adverse risk factors and 25-OH vitamin D3 remained the strongest inverse risk factor (Table 1).
Risk score design and comparison with established risk scores
Furthermore, we generated the VIenna Ludwigshafen DIAbetes (VILDIA) risk score, a weighted risk score for long-term survival using the aforementioned variables obtained by the bootstrapping procedure (see supplemental method 1 for the formula). The C-statistic of the VILDIA score indicated an excellent discriminatory power for 10-year survival with a value of 0.75 (P < 0.001) for all-cause mortality and for 10-year survival free of cardiovascular death with a value of 0.76 (P < 0.001). The P values for the Grønnesby and Borgan statistics indicated a good calibration for the VILDIA score for all-cause and cardiovascular mortality models (P > 0.20). The discriminatory ability of our score was superior to the previously published UKPDS risk engine7 for 10-year survival free of cardiovascular death (C-statistic 0.64; P < 0.001 for comparison). Furthermore, the VILDIA score was significantly superior to the UKPDS risk engine with respect to improved risk classification with a NRI of 21% (95% CI: 12–30%; P < 0.001) for all-cause mortality (Table 2) and of 22% (95% CI: 14–30%; P < 0.001) for cardiovascular mortality (Table 3).
Kaplan Meier analysis confirmed the high discriminatory value when plotting tertiles of the new risk score for all-cause mortality (Fig. 2A) and cardiovascular mortality (Fig. 2B). The 10-year survival rates in the first, second and third tertile of the new score were 85%, 63%, and 25%, respectively (P < 0.001 between all tertiles). With regard to cardiovascular mortality, a comparable trend for the 10-year survival was observed, with survival rates of 91%, 73%, and 41% in the first, second and third tertile of the new risk score (P ≤ 0.001 between all tertiles).
Simplified score
Ultimately, we aimed to create a simplified VILDIA score for optimal clinical use. The design of this simplified score was based on the same variables as the weighted risk score. Continuous variables were dichotomized according to optimal cut-off levels, as described in Fig. 3. Two points were assigned to the variable Nt-proBNP with the strongest predictive value in the multivariate model while one point was allocated to the other variables. The simplified VILDIA score still demonstrated a strong discriminatory power with a C-statistic of 0.72 for all-cause mortality and 0.72 for cardiovascular mortality over 10 years. The discriminatory ability of the simplified VILDIA score was still superior to the UKPDS risk engine (P = 0.01). All-cause mortality ranged from 5.3% with zero points to 94.9% with six or more points (Fig. 3A). Cardiovascular mortality ranged from 5.3% to 56.4% (Fig. 3B). With regard to the use of pharmacologic treatment reducing cardiovascular mortality we found inverse associations between the simplified VILDIA score and aspirin use (for trend, p < 0.007, Fig. 3C), beta blocker use (for trend, p < 0.001, Fig. 3D) and statin use (for trend, p < 0.001, Fig. 3E) at discharge.
Conclusions
We developed the VILDIA score, a multimarker score for patients with diabetes, which provides an excellent discriminatory power for the prediction of 10-year survival with a C-statistic of 0.75. The final set of seven variables was selected using a bootstrapping resampling procedure from a large set of 135 clinical and biochemical markers encompassing various pathophysiologic pathways and reflecting different aspects of the vulnerable patient with diabetes. The clinical relevance of this new multimarker score was further emphasized by a significant improvement of net reclassification by 21%, which reflects the proportion of individuals, who may benefit from an improved risk prediction compared to the regularly used UKPDS risk engine. The new VILDIA score predicted long-term mortality irrespective of the presence of a previous MI.
In general, diabetes is associated with an approximately three times higher risk of cardiovascular events10. Accordingly, similar preventive measures such as stricter target values for cholesterol or blood pressure have been proposed11. However, an risk increase of diabetic patients equivalent to established coronary artery disease has been recently questioned12. Therefore, a refinement of cardiovascular risk estimation in patients with diabetes is warranted to manage the cardiovascular risk of patients with diabetes in an individualized form. So far the UKPDS risk engine has been the best available tool to predict the cardiovascular risk of patients with diabetes. This score included clinical characteristics including age, sex, ethnicity, smoking, HbA1c, systolic blood pressure and cholesterol7. However, the new VILDIA score outperformed this score by adding a set of selected biomarkers. As shown by a superior predictive value of the score compared to single risk markers this score may also reflect the multifactorial pathogenesis of cardiovascular events more accurately than a single marker. Robustness of the selection of markers (Fig. 1) and independency of their predictive value as shown in the multivariable model (Table 1) strengthened the overall predictive value of the score. Furthermore, selection from an initial set of more than 100 biomarkers strongly increases the probability of covering the majority of relevant pathogenetic factors. The predictive value of VILDIA was not influenced by previous MI and, even more importantly, the presence of a previous MI did not increase the predictive value of the VILDIA score (data not shown). Furthermore, as expected, the presence or absence of angiographic coronary artery disease or semi-quantitative measures of the atherosclerotic burden to the coronary arteries did not substantially contribute to risk stratification. This is due to the fact that angiography visualizes the vessel lumen only and does not provide information on the vulnerability of plaques nor on their actual size13, 14.
A non-invasive alternative to coronary angiography would be computed tomography of the coronary arteries to visualize and quantify coronary calcium. In a systematic review of current guidelines the authors found that 10 of 14 guidelines considered the CT calcium score as a test for improvement of coronary risk assessment, however only 4 guidelines concluded that there was sufficient evidence for the use of coronary calcium scoring and only 1 guideline recommended its use solely in an intermediate CAD risk population15. More importantly for low CAD risk persons and persons already known to be at high CAD risk, guidelines were unanimous in not advocating CT calcium scoring15. Yet, it remains to be determined whether coronary calcium scoring would indeed yield superior prognostic information above and beyond the VILDIA algorithm. Finally, an alternative strategy could be to perform cheaper biomarker tests in advance of calcium scoring in a step-wise fashion. It is however, unlikely, that calcium scoring will considerably improve prediction beyond the biomarkers.
Other risk scores to predict CAD risk in type 2 diabetes have been described but to the best of our knowledge the UKPDS risk score is the most powerful tool to assess CAD risk and that is why we focused on comparison to the UKPDS risk engine. The AD-ON score from the ADVANCE trial showed excellent discrimination for major renal events albeit only moderate discrimination for CHD (c-statistic = 0.67) compared to most risk scores16. However, the AD-ON score showed good calibration among different regions. This is specifically important as a risk score might not be generally applicable to all regions. For instance Yang and coworker showed that the UKPDS risk engine overestimated CHD risk among Chinese patients with type 2 diabetes17. Stable calibration among different geographical regions for the VILDIA score remains to be demonstrated. The look AHEAD research group showed the predictive value of a genetic risk score in obese patients with type 2 diabetes18.
The LURIC study participants represent a cohort at intermediate to high cardiovascular risk. At a first glance, our conclusions may therefore not be generalizable to a random population sample. However, global cardiovascular risk in our patients comes very close to that of patients with diabetes mellitus recruited in other settings. For instance, in a population-based study3 the proportion of diabetes patients with a history of myocardial infarction was similar to that in our cohort and, more importantly, the overall annual rate of death from cardiovascular causes in diabetes mellitus patients was approximately 2.9 percent while in LURIC it was 2.8 percent. In other population based-studies, very similar incidence rates of cardiovascular deaths have been seen19,20,21. It thus appears that our study population is indeed representative of patients with diabetes mellitus in general, and we speculate that the relatively high baseline cardiovascular risk of the study participants is sufficiently explained by the presence of diabetes mellitus rather than by other reasons such as referral bias. Finally, the comparatively high risk may represent an important strength of the study, because precisely targeted treatment is most likely to yield benefit in these patients and previous risk assessment in the high risk population has been exceptionally inconsistent22.
All (bio)markers included in the VILDIA score have been shown to be associated with cardiovascular risk. Duration of diabetes is associated with increased risk of major coronary heart disease events23 and has been repeatedly identified as a potent predictor of all-cause and cardiovascular mortality24,25,26. ProBNP and its N-terminal fragment NT-proBNP are one of the strongest and most robust predictors of mortality in the cardiovascular field, particularly in patients with heart failure27 and acute coronary syndrome28, 29. Natriuretic peptides may also be important predictors of cardiovascular and all-cause mortality in patients with diabetes30,31,32 {Hillis, 2014 #38}. In our diabetes cohort, Nt-proBNP has been the most robust parameter in the bootstrapping resampling procedure, even more robust than left ventricular function measured by echocardiography (Fig. 1). Renin, part of the renin-angiotensin-aldosterone system (RAAS), is not only involved in development of arterial hypertension33 but also in the progression of cardiovascular diseases and is associated with cardiovascular events34, 35. Conflicting data derived from the Framingham Heart Study where renin was associated with mortality but not with incidence of cardiovascular disease36, 37. Furthermore, direct inhibition of renin in addition to standard RAAS inhibition in patients with diabetes has not shown to be beneficial with regard to a composite endpoint including cardiovascular and renal events38. 1,25-OH vitamin D3, the active form of vitamin D, requires sunlight as well as conversion in the liver and the kidney. 25-OH vitamin D3, the major circulating form, may be preferred as biomarker. Apart from its role in skeletal pathologies vitamin D deficiency also affects extraskeletal diseases including cardiovascular diseases. Particularly, in patients with metabolic syndrome or diabetes, low levels of vitamin D have been associated with increased cardiovascular mortality39, 40. Finally, activity of Lp-PLA2, also described as platelet-activating factor acetylhydrolase, is closely correlated with cardiovascular risk factors suggesting an important role in atherogenesis41. Experimental data indicated that inhibition of Lp-PLA2 may prevent progression of atherosclerotic plaques42. Furthermore, Lp-PLA2 is associated with cardiovascular mortality43 and predicts cardiovascular outcome in diabetic patients44.
While scores are superior in risk prediction compared to single risk factors they are underutilized in clinical practice6, 45. An easy-to-use form such as the simplified VILDIA score using categorized variables based on optimized cut-offs is therefore of utmost importance for daily clinical use. The simplified VILDIA score shows a comparable predictive accuracy and is able to fulfill the most important aim of ranking patients46 into low 10-year fatal cardiovascular risk (<10%) with zero or one point, intermediate risk (20–30%) with two to three points and high risk (about 50% or more) with four or more points. Our results strongly support that patients with diabetes mellitus are very heterogeneous with regard to their cardiovascular risk. In accordance, it seems obsolete to regard diabetes mellitus simply as a cardiovascular disease risk equivalent. The VILDIA score may provide a novel tool for clinicians, to more easily estimate the risk of patients with diabetes mellitus on the basis of a few clinical and laboratory parameters. The refinement of cardiovascular risk estimation for the individual patients as evidenced by an improved NRI may lead to a more accurate treatment of the individual patient. A negative correlation between the VILDIA score and aspirin, statin and beta-blocker use suggests that there may be potential to optimize the therapy. E.g. less than 30% of patients with a simplified VILDIA score of ≥6 received statins. However, this high-risk group has a 10-year cardiovascular mortality of more than 50%. While there is an ongoing discussion about an increased incidence of diabetes in patients on statins they certainly reduce cardiovascular events in patients with known diabetes47. Likewise, aspirin and beta-blockers, both reducing cardiovascular risk, have been prescribed with the lowest frequency in patients with a high VILDIA score suggesting a benefit of an individualized risk management based on the VILDIA score. Nevertheless, little consensus exists concerning the question whether diabetic patients without overt cardiovascular disease should receive antiplatelet therapy48. Another open question is whether diabetic patients will benefit from screening for silent coronary artery disease4. Trying to answer these questions it seems essential to find those patients, who will most likely profit from expensive screening and intervention strategies. The novel VILDIA-score appears to provide this information at affordable costs.
The limited number of included patients and the lack of external validation is a relevant limitation of this study. However, assessment of more than 100 biomarkers in a larger or second cohort is hardly accomplishable. Replication of the score using selected variables only will be required in an external cohort before clinical use as varying population characteristics may lead to different risk estimates. Additionally, medication at time of study enrollment may have to some extent influenced the recorded biomarker levels. Further, a long-term follow-up implicates that patients have been selected about 10 years ago. Therefore, measures of secondary prevention may be used in a different way nowadays. Another hurdle to the clinical application may be significant costs due to measurement of biomarkers. However, the set has been limited to the most important markers and resulted in a superior predictive value. Cost savings following the implementation of the VILDIA score for efficient prevention of cardiovascular events are likely. However, cost-effectiveness analysis will be needed before implementation. While we anticipate that implementation of such a score would require further considerations of risk and benefits, we believe that a comprehensive cost-benefit simulation would go far beyond the scope of this manuscript. Furthermore, there were notable differences in age range among the present cohort and the UKPDS cohort. Partially this can be attributed to changes in the general population characteristic. The UKPDS however is -up to now- the most broadly used tool for risk stratification and the scores performed equally well within age groups in the present cohort. Importantly, comparison of the predictive value of the scores is also limited as the UKPDS risk engine was initially developed as a predictive model for CAD. Strengths of this study are the comprehensive set of biomarkers and a long-term follow-up of 10 years. Of course the current data are observational in nature and only a randomized trial would have the potential to demonstrate whether risk algorithm guided management of patients with diabetes mellitus improves long term prognosis above and beyond usual care.
In conclusion, the VILDIA score has confirmed the hypothesis that a multi-biomarker score may improve risk estimation in patients with diabetes. Assessment of a set of independent biomarkers may reflect the multi-factorial pathogenesis of cardiovascular events better than single risk factors and scores based on routine variables. A more individualized treatment based on the VILDIA score will provide the potential benefit of improving the therapy of high-risk patients as well as avoiding overtreatment in low-risk patients.
Research Design and Methods
Study population
The Ludwigshafen Risk and Cardiovascular Health (LURIC) study comprises 3,316 Caucasian patients referred to the Cardiac Center Ludwigshafen (Germany) for coronary angiography between 1997 and 2000. The detailed study protocol has been previously published9. The study protocol complies with the Declaration of Helsinki and was approved by the Ethics Committee of the “Ärztekammer Rheinland-Pfalz”. Written informed consent was collected before study enrollment in all patients. For the present analysis, we chose 864 patients with diabetes. Patients with an acute coronary syndrome at hospital admission were excluded from our analysis.
Clinical definitions
Diagnosis of diabetes was based on the revised American Diabetes Association criteria49. Subjects with increased fasting (≥126 mg/dl) and/or post-challenge (2 h after the 75 g glucose load ≥200 mg/dl) glucose or with an HbA1c > 6.5% were diagnosed with diabetes50. People with a self-reported or documented history of diabetes or treatment with oral antidiabetics or insulin were considered diabetic (9). Diabetes duration was categorized into newly diagnosed diabetes, duration ≤5 years, duration >5 to ≤10 years, duration >10 years as previously published26. Clinical risk factors such as hypertension, current smoking, lipid disorders and previous myocardial infarction (MI) were diagnosed according to the respective guidelines as previously described9. Venous blood samples were drawn under standardized conditions after an overnight fasting period. NT-proBNP was measured by electrochemiluminescence on an Elecsys 2010 (Roche Diagnostics). Plasma renin concentration was determined by an immunoradiometric assay (Active Renin, Nichols Institute Diagnostics, San Juan, Capistrano, CA, USA). Lipoprotein associated phospholipase A2 (Lp-PLA2) activity was measured by use of the Azwell Auto PAF-AH reagent set (Azwell). Serum levels of 25-OH vitamin D3 were measured using a radioimmunoassay (DiaSorin SA, Antony, France). Intra and inter-assay coefficients of variation were <10% for these assays.
Study endpoints and follow-up
The primary study endpoints were all-cause mortality and cardiovascular mortality. Information on patient outcome was obtained by screening local registries. No patient was lost to follow-up. Death certificates were obtained in 97% of deceased participants and used to classify them into those who died of cardiovascular versus non-cardiovascular causes as previously described9. This classification was performed independently by two experienced clinicians who were blinded to the study participants, except for information that was required to classify the cause of death.
Statistical analysis
Discrete data was presented as counts and percentages, continuous data were presented as median and interquartile range. Cox proportional hazard regression analysis was used to evaluate the effect of the investigated variables on survival. Continuous variables were log-transformed before entering analysis to account for not normally distributed variables. For standardization purposes we decided to analyze all variables uniformly. A bootstrap resampling procedure was used to identify best-fitting predictors of all-cause mortality for the final multivariate Cox regression model. Before inclusion of variables into the bootstrapping procedure collinearity between continuous variables was assessed using Pearson correlation coefficients (data not shown). In case of close correlation only the strongest univariate predictor of a cluster (defined by the highest univariate hazard ratio for an increase of 1 standard deviation (SD)) entered the selection procedure. Samples with a size of 100% of the original cohort were chosen for repeats. 100 repeats with backward selection and 100 repeats with forward selection (with a P < 0.1 for selection) were executed stepwise. Variables selected in more than 80% of all repeats were included in the final multivariable model. Harrell’s C-statistic was used to estimate the discriminatory power of the designed risk score. Moreover, the calibration of the score was assessed using the Grønnesby and Borgan statistics, comparing the observed and model-based estimated expected number of events within four risk groups. For further analysis the cohort was stratified by tertiles of the biomarker risk score. An improvement of individual risk prediction was examined by the net reclassification improvement (NRI). Kaplan-Meier analysis (log-rank test) was applied to verify the time-dependent discriminative power of risk scores. Cut-off values for a simplified score using categorized variables were estimated using chi-square values derived from Martingale residuals of the Cox model for all possible cut-off values. For optimized clinical use these calculated cut-off values were rounded and extreme values were avoided. Chi square test for linear to linear association was used to assess the trend between medication and simplified risk score categories. Two-sided P-values < 0.05 were used to indicate statistical significance. The STATA software package (Stata Statistical Software: Release 11, StataCorp LP, USA) and SPSS 17.0 (IBM SPSS, USA) were used for all analysis.
References
Emerging Risk Factors, C. et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med 364, 829–841, doi:10.1056/NEJMoa1008862 (2011).
Schramm, T. K. et al. Diabetes patients requiring glucose-lowering therapy and nondiabetics with a prior myocardial infarction carry the same cardiovascular risk: a population study of 3.3 million people. Circulation 117, 1945–1954, doi:10.1161/CIRCULATIONAHA.107.720847CIRCULATIONAHA.107.720847 (2008).
Haffner, S. M., Lehto, S., Ronnemaa, T., Pyorala, K. & Laakso, M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 339, 229–234, doi:10.1056/NEJM199807233390404 (1998).
Bloomgarden, Z. T. Diabetes and cardiovascular disease. Diabetes Care 34, e24–30, doi:10.2337/dc11-000734/3/e24 (2011).
Morrow, D. A. Cardiovascular risk prediction in patients with stable and unstable coronary heart disease. Circulation 121, 2681–2691, doi:10.1161/CIRCULATIONAHA.109.852749121/24/2681 (2010).
Goliasch, G. et al. Routinely available biomarkers improve prediction of long-term mortality in stable coronary artery disease: the Vienna and Ludwigshafen Coronary Artery Disease (VILCAD) risk score. Eur Heart J 33, 2282–2289, doi:10.1093/eurheartj/ehs164ehs164 (2012).
Stevens, R. J., Kothari, V., Adler, A. I. & Stratton, I. M., United Kingdom Prospective Diabetes Study, G. The UKPDS risk engine: a model for the risk of coronary heart disease in Type II diabetes (UKPDS 56). Clin Sci (Lond) 101, 671–679 (2001).
Roos, C. J., Quax, P. H. & Jukema, J. W. Cardiovascular metabolic syndrome: mediators involved in the pathophysiology from obesity to coronary heart disease. Biomark Med 6, 35–52, doi:10.2217/bmm.11.105 (2012).
Winkelmann, B. R. et al. Rationale and design of the LURIC study–a resource for functional genomics, pharmacogenomics and long-term prognosis of cardiovascular disease. Pharmacogenomics 2, S1–73 (2001).
Stamler, J., Vaccaro, O., Neaton, J. D. & Wentworth, D. Diabetes, other risk factors, and 12-yr cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care 16, 434–444 (1993).
Cho, E., Rimm, E. B., Stampfer, M. J., Willett, W. C. & Hu, F. B. The impact of diabetes mellitus and prior myocardial infarction on mortality from all causes and from coronary heart disease in men. J Am Coll Cardiol 40, 954–960, doi:S0735109702020442 (2002).
Bulugahapitiya, U., Siyambalapitiya, S., Sithole, J. & Idris, I. Is diabetes a coronary risk equivalent? Systematic review and meta-analysis. Diabet Med 26, 142–148, doi:10.1111/j.1464-5491.2008.02640.xDME2640 (2009).
Glagov, S., Weisenberg, E., Zarins, C. K., Stankunavicius, R. & Kolettis, G. J. Compensatory enlargement of human atherosclerotic coronary arteries. N Engl J Med 316, 1371–1375, doi:10.1056/NEJM198705283162204 (1987).
Libby, P. Mechanisms of acute coronary syndromes and their implications for therapy. N Engl J Med 368, 2004–2013, doi:10.1056/NEJMra1216063 (2013).
Ferket, B. S. et al. Systematic review of guidelines on imaging of asymptomatic coronary artery disease. J Am Coll Cardiol 57, 1591–1600, doi:10.1016/j.jacc.2010.10.055S0735-1097(11)00410-4 (2011).
Woodward, M. et al. Prediction of 10-year vascular risk in patients with diabetes: the AD-ON risk score. Diabetes Obes Metab 18, 289–294, doi:10.1111/dom.12614 (2016).
Yang, X. et al. Development and validation of a total coronary heart disease risk score in type 2 diabetes mellitus. Am J Cardiol 101, 596–601, doi:10.1016/j.amjcard.2007.10.019 (2008).
Look, A. R. G. Prospective association of a genetic risk score and lifestyle intervention with cardiovascular morbidity and mortality among individuals with type 2 diabetes: the Look AHEAD randomised controlled trial. Diabetologia 58, 1803–1813, doi:10.1007/s00125-015-3610-z (2015).
Juutilainen, A., Lehto, S., Ronnemaa, T., Pyorala, K. & Laakso, M. Type 2 diabetes as a “coronary heart disease equivalent”: an 18-year prospective population-based study in Finnish subjects. Diabetes Care 28, 2901–2907, doi:28/12/2901 (2005).
Fox, C. S. et al. Trends in cardiovascular complications of diabetes. JAMA 292, 2495–2499, doi:292/20/249510.1001/jama.292.20.2495 (2004).
Booth, G. L., Kapral, M. K., Fung, K. & Tu, J. V. Relation between age and cardiovascular disease in men and women with diabetes compared with non-diabetic people: a population-based retrospective cohort study. Lancet 368, 29–36, doi:S0140-6736(06)68967-810.1016/S0140-6736(06)68967-8 (2006).
Allan, G. M. et al. Agreement among cardiovascular disease risk calculators. Circulation 127, 1948–1956, doi:10.1161/CIRCULATIONAHA.112.000412CIRCULATIONAHA.112.000412 (2013).
Wannamethee, S. G., Shaper, A. G., Whincup, P. H., Lennon, L. & Sattar, N. Impact of diabetes on cardiovascular disease risk and all-cause mortality in older men: influence of age at onset, diabetes duration, and established and novel risk factors. Arch Intern Med 171, 404–410, doi:10.1001/archinternmed.2011.2171/5/404 (2011).
Brun, E. et al. Diabetes duration and cause-specific mortality in the Verona Diabetes Study. Diabetes Care 23, 1119–1123 (2000).
Nelson, R. G. et al. Low incidence of fatal coronary heart disease in Pima Indians despite high prevalence of non-insulin-dependent diabetes. Circulation 81, 987–995 (1990).
Silbernagel, G. et al. Duration of type 2 diabetes strongly predicts all-cause and cardiovascular mortality in people referred for coronary angiography. Atherosclerosis 221, 551–557, doi:10.1016/j.atherosclerosis.2012.01.011S0021-9150(12)00044-5 (2012).
Doust, J. A., Glasziou, P. P., Pietrzak, E. & Dobson, A. J. A systematic review of the diagnostic accuracy of natriuretic peptides for heart failure. Arch Intern Med 164, 1978–1984, doi:10.1001/archinte.164.18.1978 (2004).
Omland, T. et al. N-terminal pro-B-type natriuretic peptide and long-term mortality in acute coronary syndromes. Circulation 106, 2913–2918 (2002).
de Lemos, J. A. et al. The prognostic value of B-type natriuretic peptide in patients with acute coronary syndromes. N Engl J Med 345, 1014–1021, doi:10.1056/NEJMoa011053 (2001).
van Hateren, K. J. et al. The Midregional Fragment of Pro-A-Type Natriuretic Peptide, Blood Pressure, and Mortality in a Prospective Cohort Study of Patients With Type 2 Diabetes (ZODIAC-25). Diabetes Care, doi:dc12-042810.2337/dc12-0428 (2012).
Bhalla, M. A. et al. Prognostic role of B-type natriuretic peptide levels in patients with type 2 diabetes mellitus. J Am Coll Cardiol 44, 1047–1052, doi:10.1016/j.jacc.2004.05.071S0735-1097(04)01220-3 (2004).
Tarnow, L., Gall, M. A., Hansen, B. V., Hovind, P. & Parving, H. H. Plasma N-terminal pro-B-type natriuretic peptide and mortality in type 2 diabetes. Diabetologia 49, 2256–2262, doi:10.1007/s00125-006-0359-4 (2006).
Atlas, S. A. The renin-angiotensin aldosterone system: pathophysiological role and pharmacologic inhibition. Journal of managed care pharmacy: JMCP 13, 9–20 (2007).
Tomaschitz, A. et al. Associations of plasma renin with 10-year cardiovascular mortality, sudden cardiac death, and death due to heart failure. Eur Heart J 32, 2642–2649, doi:10.1093/eurheartj/ehr150ehr150 (2011).
Verma, S. et al. Plasma renin activity predicts cardiovascular mortality in the Heart Outcomes Prevention Evaluation (HOPE) study. Eur Heart J 32, 2135–2142, doi:10.1093/eurheartj/ehr066ehr066 (2011).
Parikh, N. I. et al. Plasma renin and risk of cardiovascular disease and mortality: the Framingham Heart Study. Eur Heart J 28, 2644–2652, doi:ehm39910.1093/eurheartj/ehm399 (2007).
Wang, T. J. et al. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med 355, 2631–2639 (2006). doi:355/25/263110.1056/NEJMoa055373.
Parving, H. H. et al. Cardiorenal end points in a trial of aliskiren for type 2 diabetes. N Engl J Med 367, 2204–2213, doi:10.1056/NEJMoa1208799 (2012).
Joergensen, C. et al. Vitamin D levels and asymptomatic coronary artery disease in type 2 diabetic patients with elevated urinary albumin excretion rate. Diabetes Care 35, 168–172, doi:10.2337/dc11-1372dc11-1372 (2012).
Thomas, G. N. et al. Vitamin D levels predict all-cause and cardiovascular disease mortality in subjects with the metabolic syndrome: the Ludwigshafen Risk and Cardiovascular Health (LURIC) Study. Diabetes Care 35, 1158–1164, doi:10.2337/dc11-1714dc11-1714 (2012).
Winkler, K. et al. Platelet-activating factor acetylhydrolase activity indicates angiographic coronary artery disease independently of systemic inflammation and other risk factors: the Ludwigshafen Risk and Cardiovascular Health Study. Circulation 111, 980–987, doi:01.CIR.0000156457.35971.C810.1161/01.CIR.0000156457.35971.C8 (2005).
Wilensky, R. L. et al. Inhibition of lipoprotein-associated phospholipase A2 reduces complex coronary atherosclerotic plaque development. Nat Med 14, 1059–1066, doi:10.1038/nm.1870nm.1870 (2008).
Lp, P. L. A. S. C. et al. Lipoprotein-associated phospholipase A(2) and risk of coronary disease, stroke, and mortality: collaborative analysis of 32 prospective studies. Lancet 375, 1536–1544, doi:10.1016/S0140-6736(10)60319-4S0140-6736(10)60319-4 (2010).
Hatoum, I. J., Hu, F. B., Nelson, J. J. & Rimm, E. B. Lipoprotein-associated phospholipase A2 activity and incident coronary heart disease among men and women with type 2 diabetes. Diabetes 59, 1239–1243, doi:10.2337/db09-0730db09-0730 (2010).
Wang, T. J. Multiple biomarkers for predicting cardiovascular events: lessons learned. J Am Coll Cardiol 55, 2092–2095, doi:10.1016/j.jacc.2010.02.019S0735-1097(10)00864-8 (2010).
Chamnan, P., Simmons, R. K., Sharp, S. J., Griffin, S. J. & Wareham, N. J. Cardiovascular risk assessment scores for people with diabetes: a systematic review. Diabetologia 52, 2001–2014, doi:10.1007/s00125-009-1454-0 (2009).
Wang, K. L. et al. Statins, risk of diabetes, and implications on outcomes in the general population. J Am Coll Cardiol 60, 1231–1238, doi:10.1016/j.jacc.2012.05.019S0735-1097(12)02069-4 (2012).
Nicolucci, A. & Standl, E. Antiplatelet therapy for every diabetic person? Diabetes Care 34(Suppl 2), S150–154, doi:10.2337/dc11-s21034/Supplement_2/S150 (2011).
Sacks, D. B. et al. Guidelines and recommendations for laboratory analysis in the diagnosis and management of diabetes mellitus. Diabetes Care 34, e61–99, doi:34/6/e6110.2337/dc11-9998 (2011).
Silbernagel, G. et al. Isolated post-challenge hyperglycaemia predicts increased cardiovascular mortality. Atherosclerosis 225, 194–199, doi:10.1016/j.atherosclerosis.2012.08.008S0021-9150(12)00555-2 (2012).
Acknowledgements
Georg Goliasch was funded by an Erwin Schrödinger Fellowship of the Austrian Science Fund (FWF J 3319-B13). LURIC was supported by the 6th Framework Program (integrated project Bloodomics, grant LSHM-CT-2004-503485) and 7th of Framework Program (integrated project Atheroremo, Grant Agreement number 201668) of the European Union. B.O.B. received grants from the DFG GrK 1041 and the Centre of Excellence “Metabolic Diseases” Baden-Wuerttemberg.
Author information
Authors and Affiliations
Contributions
G.G. performed the statistical analysis, contributed to the interpretation of the results and wrote the manuscript. M.E.K. researched the data and contributed to the interpretation of the results and edited the manuscript. T.B.G., S.P., A.T., G.M., W.K. contributed to the interpretation of the results and reviewed and edited the manuscript. G.S. contributed to the interpretation of the results and contributed to the introduction and discussion. A.N. contributed to the interpretation of the results and wrote the manuscript. P.E.B. contributed to the interpretation of the results, the statistical analysis of the revised version and edited and wrote the manuscript. W.M. designed the study, reviewed and edited the manuscript and is the guarantor of this work.
Corresponding author
Ethics declarations
Competing Interests
The authors declare that they have no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Goliasch, G., Silbernagel, G., Kleber, M.E. et al. Refining Long-Term Prediction of Cardiovascular Risk in Diabetes – The VILDIA Score. Sci Rep 7, 4700 (2017). https://doi.org/10.1038/s41598-017-04935-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-017-04935-8
- Springer Nature Limited
This article is cited by
-
Focus on Diagnosis and Prognosis to Guide Timing of Intervention in Valvular Heart Disease
Current Cardiology Reports (2022)
-
Ameliorative effects of Carica papaya extracts against type II diabetes-induced myocardial pathology and dysfunction in albino rats.
Environmental Science and Pollution Research (2021)
-
Prognostic nomogram of hypoxia-related genes predicting overall survival of colorectal cancer–Analysis of TCGA database
Scientific Reports (2019)
-
The Evolving Cardiovascular Disease Risk Scores for Persons with Diabetes Mellitus
Current Cardiology Reports (2018)