Predictive factors in assessing cardiovascular risk: learning from ADVANCE

Andre Pascal KENGNE, MD, PhD
The University of Sydney and the Royal Prince Alfred Hospital
Sydney, New South Wales AUSTRALIA

The reductions in cardiovascular events observed in the Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation (ADVANCE) trial through routine blood pressure lowering with perindopril/indapamide and intensive glucose control with a Diamicron MR– based regimen have provided added incentive to develop tools for more accurate assessment of cardiovascular risk in patients with type 2 diabetes.Models based on earlier studies, including the Framingham and United Kingdom Prospective Diabetes Study risk equations, were based on populations treated in a much earlier therapeutic environment, before the availability of many of today’s protective cardiovascular drugs. Accordingly, we have used the opportunity afforded by the ADVANCE population with type 2 diabetes, a population that is both contemporary and representative of the broad cross-section of people with type 2 diabetes worldwide, to develop a new, improved risk engine for predicting the risk of cardiovascular events. The predictive baseline characteristics used to estimate cardiovascular risk through the ADVANCE risk engine are age at diagnosis of diabetes, known duration of diabetes, sex, pulse pressure, treated hypertension, atrial fibrillation, retinopathy, HbA1c, urinary albumin/creatinine ratio, and non-high-density–lipoprotein cholesterol. This model provides a considerable improvement over the older risk equations in predicting the risk of cardiovascular events. The new ADVANCE risk engine will shortly be available, through a specific ADVANCE Web site, to assist physicians around the world in calibrating their patients’ risk profile and in optimizing their therapeutic management to alleviate the global burden of cardiovascular disease in patients with type 2 diabetes.

Medicographia. 2011;33:47-51 (see French abstract on page 51)

As documented elsewhere in this issue, the epidemic of diabetes is evolving at alarming speed in both developed and developing regions of the world, with dramatic growth in the burden of cardiovascular disease (CVD). The most common cause of death among individuals with diabetes is coronary disease, but stroke, renal failure, and heart failure are also important contributors. The elevated risks of major vascular events have been well documented in a range of populations,1 and the risks are apparent in the young and the old, in the East and the West.1 As a result, there is an urgent need, not only for effective preventive and treatment programs that minimize the complications of diabetes, but also for tools to help physicians in assessing the risks of cardiovascular complications in individuals with type 2 diabetes.

Figure 1
Figure 1. 4-year predicted rates of major cardiovascular events in ADVANCE by the Framingham and UKPDS equations.

E = expected event rate; O = observed event rate; Vertical line at 1 is the line of perfect agreement between expected and observed event rates.
Abbreviations: ADVANCE, Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation; CHD, coronary heart disease; CVD, cardiovascular disease; UKPDS, United Kingdom Prospective Diabetes Study.
Modified from reference 9: Kengne et al; ADVANCE Collaborative Group. Diabetologia. 2010;53:821-831. © 2010, Springer-Verlag.

Indeed, many patients with diabetes live long lives with quite good health, while others are cut down prematurely! It is clearly of vital interest to these patients, their families, and their physicians to have an accurate assessment of the risk of serious complications of diabetes so that appropriate plans for prevention may be formulated. Since cardiovascular disease constitutes the major burden of ill health in type 2 diabetes, there is an urgent need for tools that help the physician to advise the patient about the risks of serious cardiovascular events, such as heart attacks and strokes, so that together they can plan and implement the lifestyle and therapeutic measures needed to reduce this risk and prevent these complications.

The need for accurate prediction of cardiovascular risk is much greater now that the Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation (ADVANCE) has clearly demonstrated the benefits of both routine blood pressure lowering with the fixed combination of perindopril and indapamide and more intensive glucose control with a Diamicron MR–based regimen, irrespective of baseline blood pressure or baseline glycated hemoglobin (HbA1c).2,3 In each case, the regimens used in ADVANCE reduced the composite primary outcome of major macrovascular and microvascular events by around 10% and renal events by around 20%, with additional significant reductions in mortality and coronary events in the perindopril/indapamide arm of the trial.

Many studies have highlighted the importance of key individual risk factors, such as the level of HbA1c and presence or absence of albuminuria, in determining the level of risk in the individual patient. However, modern guidelines increasingly emphasize the importance of estimating the individual’s global cardiovascular risk as a more appropriate basis for risk factor management. Global cardiovascular risk is a quantitative estimate of an individual’s chances of experiencing a cardiovascular event within a given time period. This estimate depends on the combination and intensity of all risk factors rather than on the presence of any single risk factor. In this paper, we examine the performance of a variety of existing “clinical prediction models,” also referred to as “absolute risk equations,” in estimating cardiovascular risk, and we present preliminary information describing the new “ADVANCE risk engine,” based on a contemporary and representative population of individuals with type 2 diabetes around the world.

Performance of older cardiovascular risk equations in the contemporary ADVANCE population

Currently, the most widely used models in people with diabetes are the Framingham cardiovascular risk equations and the United Kingdom Prospective Diabetes Study (UKPDS) risk engines. The utility of both the Framingham and UKPDS equations in people with diabetes has been documented in historic cohorts,4 but not established in contemporary populations. The Framingham equations are less than ideal for predicting risk in people with diabetes, since they are not only based on populations from a previous era with a very differ- ent therapeutic environment, but are also based on a general population having only a small minority of individuals with diabetes.5,6 The UKPDS equations are also based in people with diabetes in the previous century, and they provide separate estimates or the risks of coronary heart disease or of stroke, but no estimate for the risk of combined major cardiovascular events.7,8

In order to assess the performance of a risk equation in a contemporary population, it is necessary to test and quantify how well a particular model, derived from a distinct and specific population, predicts risk in another quite different and independent modern population. The main criteria used to describe the performance of the model being tested are termed “calibration” and “discrimination.”We illustrate this below, in assessing the performance of the older Framingham and UKPDS equations in the contemporary ADVANCE cohort.5-8

“Calibration” quantifies how close the predictions are to the actual outcome. For instance, a 5-year estimated probability of cardiovascular disease of 20% for a patient means that, in a given group of patients with similar characteristics, 20% will experience a cardiovascular event within a 5-year period. Figure 1 shows that, in the cohort of ADVANCE patients who had no known history of cardiovascular disease at enrolment in the trial, the 4-year risk of cardiovascular events was largely overestimated by the Framingham-Anderson,5 Framingham- D’Agostino,6 and UKPDS equations.7,8 This overestimation was observed in men and women, whites and non-whites, and in the double-placebo cohort of ADVANCE (ie, those assigned to the placebo group in the blood pressure–lowering arm and the standard care group of the blood pressure– control arm).9

“Discrimination” describes the performance of a model in distinguishing between patients who go on to develop a cardiovascular event and those who remain event free. Discrimination using the Framingham and UKPDS equations in predicting CVD events in the ADVANCE patients was modest to acceptable for coronary heart disease and for total CVD, but poor for stroke.9

Recalibration is a method for improving an equation’s predictive capacity. It usually consists of adjusting the equation by replacing the average value of the risk factors and event rates in the equation (derived from the original population) by those in the test population. When applied to the Framingham and UKPDS equations, this approach substantially attenuated the overestimation of risk for the ADVANCE patients. However, discrimination was not improved, indicating the need for a new equation with improved discriminatory capability for people with diabetes, particularly those receiving many contemporary cardiovascular risk-reducing therapies.9

Table I
Table I. Baseline characteristics for the ADVANCE participants and other cross-sectional studies.

Abbreviations: ADVANCE, Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation; AUSDIAB, AUStralian DIABetes, obesity, and lifestyle study; DEPAC, Diabetes Experts Panel from Accessing Countries; ENTRED, Échantillon National Témoin Représentatif des Personnes Diabétiques [Nationalrepresentative sample of diabetics]; HbA1c , glycated hemoglobin.
Modified from reference 9: Kengne et al; ADVANCE Collaborative Group. Diabetologia. 2010;53:821-831. © 2010, Springer-Verlag.

Development of the new ADVANCE risk engine for predicting risk and improving cardiovascular event prevention in type 2 diabetes

In developing a new model for risk prediction, it is important to address the limitations of the existing models. The inclusion in ADVANCE of participants from many countries has provided the opportunity to account for the substantial variation in the care of diabetes and cardiovascular disease around the world, whereas existing models have been derived from homogenous populations from the UK and the USA. The generalizability of the ADVANCE cohort to current contemporary populations of patients with type 2 diabetes around the world is shown in Table I, which compares the characteristics of patients participating in ADVANCE with those of individuals with type 2 diabetes at community level in a number of countries.10-13 The ADVANCE model also aims to predict total cardiovascular risk and therefore to capture the interrelation between components of cardiovascular disease, such as coronary heart disease and stroke, unlike other models, such as the UKPDS equations, that focus specifically and separately on these components. Moreover, the complexity of the relationship between chronic hyperglycemia and cardiovascular risk has not been as well addressed in existing models. In the ADVANCE model, further improvements have been achieved through the integration of risk factors to capture exposure to chronic hyperglycemia both before and after the clinical diagnosis of diabetes.

Several risk factors were considered for inclusion in the ADVANCE model, including the traditional risk factors for CVD, HbA1c, and some novel risk factors. The final ten predictors that emerged as independent predictors and were chosen for inclusion in the ADVANCE risk engine are shown in Table II.14 Age at diagnosis of diabetes and known duration of diabetes were preferred to age at baseline, in order to improve the applicability of the ADVANCE equation to other populations. Although cognitive function has been confirmed as an independent predictor of CVD in ADVANCE,15 this characteristic was not considered for inclusion in the model, given the difficulties in assessing cognitive function in a standardized way around the world. Surprisingly, smoking status was not a significant predictor when tested alone or together with other risk factors in the ADVANCE population, possibly reflecting the small proportion of current smokers in that cohort.

The capacity to predict the risk of events was tested against the actual recording of events that occurred during the postrandomization period of follow-up within the ADVANCE population. The calibration of the ADVANCE model was excellent and the discrimination acceptable with an area under the curve (AUC) of 0.7. Validation of the ADVANCE model in an external, independent population of people with type 2 diabetes is currently proceeding. To facilitate the uptake of the ADVANCE model in clinical practice, a risk scoring table is being developed. This will assign scores for various levels of each of the ten final risk factors selected, in each individual patient. Other tools from this model are also planned, including an online calculator.


The ADVANCE trial has provided a major impetus for development of accurate tools to predict the risk of cardiovascular events in individuals with type 2 diabetes. The routine administration of the fixed combination of perindopril and indapamide reduced all-cause mortality by 14%, cardiovascular mortality by 18%, major vascular events by 9%, and coronary and renal events by 14% and 21 %, respectively.1 Additionally, more intensive glucose control using a gliclazide MR–based regimen reduced the risk of major macro- or microvascular events by 10%, of major microvascular events by 14%, and of new or worsening nephropathy by 20%. Given these potential benefits, it is incumbent on physicians responsible for the care of patients with type 2 diabetes to ensure that they prescribe such treatments for all patients at high risk of experiencing cardiovascular events. In turn, this places a premium on developing the capacity to predict the risk of cardiovascular events much more accurately than was previously possible.

Table II
Table II. Predictors selected for the ADVANCE risk engine.

Abbreviations: ADVANCE, Action in Diabetes and Vascular disease: PreterAx and DiamicroN MR Controlled Evaluation; HbA1c , glycated hemoglobin; HDL,
high-density lipoprotein.
Modified from reference 14: Kengne et al. International Diabetes Federation
2009, Montreal. Abstract 0199 (p.71). © 2009, International Diabetes Federation.

It is in this context that the new ADVANCE risk engine presents a new and valuable tool. In an effort to overcome some of the limitations of the existing models for estimating cardiovascular risk in people with diabetes, the new model is founded on some of the unique features of the ADVANCE cohort. The ADVANCE model is based on parameters that are easily assessable and widely available in routine clinical care. When tested, the performance of this new model was highly acceptable. Inclusion of participants from developing countries in the ADVANCE cohort highlights the potential of the ADVANCE risk engine for assisting cardiovascular risk stratification efforts in many settings around the world.

We are currently actively working to develop a specific ADVANCE risk engine Web site that will assist physicians all around the world to determine the risk of cardiovascular disease in their own individual patients with type 2 diabetes, and we invite all physicians to use this tool as soon as it becomes available. _

1. Asia Pacific Cohort Studies Collaboration. The effects of diabetes on the risks of major cardiovascular diseases and death in the Asia-Pacific region. Diabetes Care. 2003;26:360-366.
2. ADVANCE Collaborative Group. Effects of a fixed combination of perindopril and indapamide on macrovascular and microvascular outcomes in patients with type 2 diabetes mellitus (the ADVANCE Trial): a randomised controlled trial. Lancet. 2007;370:829-840.
3. ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl JMed. 2008;358:2560-2572.
4. Chamnan P, Simmons RK, Sharp SJ, et al. Cardiovascular risk assessment scores for people with diabetes: a systematic review. Diabetologia. 2009;52: 2001-2014.
5. Anderson KM, Odell PM, Wilson PW, Kannel WB. Cardiovascular disease risk profiles. Am Heart J. 1991;121:293-298.
6. D’Agostino RB Sr, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008; 117:743-753.
7. Stevens RJ, Kothari V, Adler AI, Stratton IM. The UKPDS risk engine: a model for the risk of coronary heart disease in type II diabetes (UKPDS 56). Clin Sci (Lond). 2001;101:671-679.
8. Kothari V, Stevens RJ, Adler AI, et al. UKPDS 60: risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine. Stroke. 2002;33:1776-1781.
9. Kengne AP, Patel A, Colagiuri S, et al; ADVANCE Collaborative Group. The Framingham and UKPDS risk equations do not reliably estimate the probability of cardiovascular events in a large ethnically diverse sample of patients with diabetes: the Action in Diabetes and Vascular Disease: Preterax and Diamicron-MR Controlled Evaluation (ADVANCE) Study. Diabetologia. 2010;53:821-831.
10. Marant C, Romon I, Fosse S, et al. French medical practice in type 2 diabetes: the need for better control of cardiovascular risk factors. Diabetes Metab. 2008;34:38-45.
11. Berthold HK, Gouni-Berthold I, Bestehorn KP, et al. Physician gender is associated with the quality of type 2 diabetes care. J Intern Med. 2008;264:340-350.
12. Barr EL, Zimmet PZ, Welborn TA, et al. Risk of cardiovascular and all-cause mortality in individuals with diabetes mellitus, impaired fasting glucose, and impaired glucose tolerance: the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Circulation. 2007;116:151-157.
13. Andel M, Grzeszczak W, Michalek J, et al; DEPAC Group. A multinational, multicentre, observational, cross-sectional survey assessing diabetes secondary care in Central and Eastern Europe (DEPAC Survey). Diabet Med. 2008;25: 1195-1203.
14. Kengne AP, Patel A, Colagiuri S, et al. Derivation of the ADVANCE models for predicting the risk of major cardiovascular disease in people with diabetes. International Diabetes Federation 2009, Montreal. Abstract 0199 (p.71).
15. De Galan BE, Zoungas S, Chalmers J, et al; ADVANCE Collaborative Group. Cognitive function and risks of cardiovascular disease and hypoglycaemia in people with type 2 diabetes: the ADVANCE trial. Diabetelogia. 2009;52:2027- 2036.

Keywords: perindopril/indapamide; Preterax; Diamicron MR; type 2 diabetes; blood pressure lowering; intensive glucose control; cardiovascular risk; risk prediction; calibration; discrimination