Cardiovascular risk evaluation

It has long been recognized that elevated blood pressure is a risk factor for future cardiovascular events and that lowering it may save lives, but deciding when to initiate antihypertensive drug therapy for the individual patient remains a daunting task. Over the past four decades, the value of drug treatment has been evaluated in a large number of controlled randomized trials, the outcomes of which have become the basis of establishing treatment benefit. Similarly, physicians increasingly have had at their disposal powerful new drugs to lower blood pressure, and were thus faced with challenging decisions as to when and how to use them. Over this time period, there has been a gradual evolution from the use of relative risk to absolute risk as the basis of deciding who and when to treat 1,2. Relative risk quantifies the likelihood of future cardiovascular events in a ‘hypertensive’ population compared to a ‘normotensive’ reference population. In contrast, absolute risk imparts information as to the expected absolute incidence of cardiovascular deaths or morbid events. In calculating absolute risk, not only hypertension should be included as a risk factor, but also age, gender and all other major cardiovascular risk factors that frequently cluster with hypertension 3. Indeed, over the past decade, there has been a plethora of recent national and international guidelines for antihypertensive therapy which have used absolute risk in varying degrees to assist physicians in decision making 4–8. Based on the results of considerable outcome data, there is general agreement that antihypertensive therapy should be used together with lifestyle treatment measures in high-risk patients 4,5. These would include individuals with persistent grade 2 or grade 3 hypertension (≥ 160/100 mmHg), those with complicated grade 1 hypertension (≥ 140–159/90–99 mmHg or those with complicated high normal blood pressure (≥ 130–139/85–89 mmHg). The latter two blood pressure categories would encompass those individuals with: (i) hypertensive target organ involvement (left ventricular enlargement, micro or macro albuminuria); (ii) vascular complications (coronary heart disease, heart failure, stroke, renal disease, peripheral vascular disease); or (iii) diabetes with or without vascular complications. On the other hand, there is considerable disagreement on when to supplement lifestyle treatment measures with antihypertensive therapy in individuals with uncomplicated grade 1 hypertension (≥ 140–159/90–99 mmHg). To date, there have been no trials to directly measure the degree of benefit of treatment in this group, which represents the largest single category within the hypertensive population. Ogden et al. 9 found that in the NHANES III USA adult hypertensive population, 20.7% had complicated hypertension, 31.3% had uncomplicated but grade 2 or greater hypertension, 5% had grade 1 hypertension without additional risk factors and 43% had uncomplicated grade 1 hypertension with at least one or more additional risk factors. It is this last group of uncomplicated grade 1 hypertensive individuals with widely varying absolute risk profiles that constitutes a dilemma in therapeutic-decision making. In this issue of the journal, Yikona et al. 10 address this problem by comparing the accuracy of five widely available risk-assessment guidelines: Sheffield UK (Sheffield table) 6, The Joint British Society (JBS chart) 7, New Zealand (NZ chart) 8, World Health Organization/International Society of Hypertension (WHO/ISH table) 5 and the Sixth United States Joint National Committee report (JNC-VI table) 4. These risk-assessment guidelines were not used in a prospective trial to predict future cardiovascular events; instead, 1991 Framingham-based risk functions 11 were used as a gold standard to rank these five risk- assessment methods regarding sensitivity, specificity, positive predictive value and negative predictive value for estimating coronary heart disease (CHD) events at 15% over 10 years and cardiovascular disease (CVD) events at 20% over 10 years; the latter represent CHD plus stroke and peripheral vascular disease events. Yikona et al. 10 concluded that assessment of CHD and CVD risk in uncomplicated stage 1 hypertension was grossly inferior for WHO/ISH and the JNC-VI tables 4,5 compared to the other three assessment methods: WHO/ISH because they only counted risk factors without weighting them and JNC-VI because they failed to both count (one or more risk factors placed subjects in the intermediate risk group B) and weight risk factors. In addition, both WHO/ISH and JNC-VI used a binary total cholesterol value rather than the superior continuous total cholesterol/high-density lipoprotein (HDL) ratio method in assessing dyslipidemia and they categorized age by only two groupings compared to four or more used by the other three assessment methods. Yikona et al. 10 correctly concluded that these short-comings in risk-assessment explain some of deficiencies in the two widely promulgated guidelines. Overall, specificity was sacrificed for sensitivity (i.e. specificity was limited to 49% for WHO/ISH and 9% for JNC-VI, while sensitivity was 95 and 100%, respectively). There were small differences between the Sheffield table 6, JBS chart 7 and NZ chart methods 8 which were of marginal significance, in part because of the small sample size of 202 test subjects that diminished the precision of the risk estimates. In summary, Yikona et al. 10 clearly showed that the qualitative and quantitative assessment of cardiovascular risk factors permits a better differentiation between those at high and low risk. However, despite these improvements in risk-assessment, there remain several controversial issues that impact treatment decisions in a significant percentage of individuals, regardless of which risk assessment method is used. Failure to consider the full risk of the ‘metabolic syndrome’ in current guidelines There is general agreement that people with the metabolic syndrome are at increased risk of developing cardiovascular disease and diabetes 12. Indeed, the cluster of abnormalities associated with the metabolic syndrome are probably more important than hypercholesterolemia in predisposing to CHD. According to the Third Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (ATP III) 11, the metabolic syndrome is defined as consisting of more than three of the following abnormalities: waist circumference > 102 cm in men and > 88 cm in women (abdominal obesity); serum triglycerides level ≥ 150 mg/dl (1.69 mmol/l); high-density lipoprotein cholesterol level < 40 mg/dl (1.04 mmol/l) in men and < 50 mg/dl (1.29 mmol/l) in women; blood pressure ≥ 130/85 mmHg; and impaired glucose tolerance (IGT) i.e. serum glucose level between 110 mg/dl (6.1 mmol/l) and 125 mg/dl (6.4 mmol/l). Of the five risk assessment methods, only the WHO/ISH considered obesity and IGT as risk factors. Although the Framingham equations 13 partially account for the metabolic syndrome (hypertension, low HDL cholesterol), they do not evaluate abdominal obesity, elevated triglycerides or IGT. It should be noted that a recent Framingham study 14 in subjects aged 65 years or older with high-normal blood pressure demonstrated that the 10-year absolute rates of major cardiovascular disease exceeded 20% in men and approached the same value in women. Undoubtedly, many of those patients who sustained cardiovascular events displayed elements of the metabolic syndrome. However, whether adding these metabolic syndrome components to the current Framingham risk equations would further improve risk prediction remains to be determined. Confirmation of the frequent occurrence of the metabolic syndrome came from the United States NHANES III adult population survey 15, which showed an age-adjusted prevalence of 23.7%. Moreover, with obesity at epidemic proportions in both the developed and developing world, the projected incidence of the metabolic syndrome will be increasing and undoubtedly appearing more frequently in younger age-groups 16. Thus, there is a sizable group of individuals with the metabolic syndrome, some of whom may have an estimated 10-year risk of 20% or more, constituting a CHD or CVD risk-equivalent by ATP III standards. Many of these individuals may benefit from treatment but, in the analysis of Yikona et al. 10, a number of these subjects might not be considered for therapy by current risk-assessment standards. Furthermore, individuals with high normal blood pressure and an absolute risk profile consistent with a CHD or CVD risk equivalent would qualify for treatment by Sheffield table and JBS chart but not necessarily by the other three risk-assessment guidelines. Finally, after concluding from risk-evaluation that many individuals with the metabolic syndrome warrant therapy, there is no consensus that treatment should consist of statins, insulin sensitizing drugs, antihypertensive medication, or a combination of these agents. At present, we await outcome data to answer these questions. Failure to appreciate the total benefit of antihypertensive therapy The benefit of antihypertensive therapy has largely been based on intervention studies of relatively short duration, which have shown that a reduction in diastolic blood pressure of 5–6 mmHg or a reduction in systolic blood pressure of 10–12 mmHg results in a decrease of 35 to 40% in stroke incidence and a decrease of 14 to 18% in CHD incidence 17. In general, trials tend to underestimate benefit because of (i) short duration; (ii) usual elimination of high-risk patients during recruitment; (iii) extensive cross-over from placebo to active therapy during the treatment phase; (iv) failure of some randomized patients to comply with active therapy; and (v) failure to correct for regression dilution. Moreover, there are stringent eligibility criteria which result in an enrolled population that may not be representative of the general hypertensive population. Suggestive, but not definitive, support for the underestimation of the true value of long-term sustained antihypertensive treatment was noted not in a clinical trial, but in a secular trend study from Framingham 18. Three successive cohorts between the ages of 50–59 years in the years 1950, 1960 and 1970 were identified and compared for cardiovascular incidence and mortality. After adjusting for risk factors in a Cox regression analysis of the combined cohorts, there was as much as a 60% reduction in cardiovascular mortality in those receiving antihypertensive therapy over a duration of 20 years compared to their untreated counterparts. Furthermore, there are additional antihypertensive benefits from therapy that escape detection in short-term studies focusing primarily on CHD and stroke events. Effective treatment of even mildly elevated systolic blood pressure in middle-aged and older individuals has been shown to prevent progression from grade 1 to more severe grades of hypertension 19,20; to retard arterial stiffness and the development of high-risk isolated systolic hypertension with wide pulse pressure 20; and to prevent or even regress left ventricular hypertrophy (LVH) and heart failure 21. In addition, antihypertensive therapy benefits may be greater in certain ethnic minority groups, such as African-Americans or South Asians that have higher absolute cardiovascular risks compared to Caucasian populations 4,5. Therefore, the higher the absolute cardiovascular risk, the greater the cardiovascular protection achieved by effective antihypertensive therapy. With the benefit of antihypertensive therapy being undervalued, the standard risk threshold for starting therapy (10-year risk of 15–20%) may be excessively high. Similarly, the development of new antihypertensive drugs with minimal side-effects, such as the angiotensin II receptor blocker, and the use of mini-dose diuretics has reduced therapeutic risk and improved the benefit-to-risk ratio of therapeutic interventions. In summary, if the overall benefit of antihypertensive therapy is under- represented, the absolute risk required for high risk status will be over represented in current treatment algorithms. Excessive weighting of advanced age in the assessment of cardiovascular risk The Sheffield table 6, JBS chart 7 and the NZ chart 8 use graded increased risk with ageing, which rightly emphasized the increased absolute risk in elderly subjects with uncomplicated hypertension. However, this emphasis on age may have minimized the risk in young and middle-aged subjects with the same level of hypertension and the same number and severity of associated risk factors that were present in elderly individuals. Must these younger individuals wait for the development of higher blood pressure, the worsening of existing risk factors, the passage of time to reach an older age, or the development of LVH to reach the threshold level of risk that would justify beginning antihypertensive therapy? Is it possible that the therapeutic blood pressure treatment goal would be more difficult to achieve and the reduction in cardiovascular risk be compromised by waiting too long before beginning therapy? This conundrum has been extensively discussed by Simpson 22, Zanchetti 23 and others 24, with the suggestion that absolute risk be ‘tempered’ by the inclusion of relative risk assessment. This would require using actuarial data to calculate reduction in life expectancy. It has been shown that, for the same elevation in blood pressure, life expectancy is more markedly reduced the earlier the age at which hypertension develops. The result could be an age-related sliding scale approach. The Second Joint Task Force of European and other Societies on Coronary Prevention 25 has recommended that the threshold of risk requiring antihypertensive therapy in uncomplicated mild hypertension be equal to a 10-year CHD risk exceeding 20%, or will exceed 20% if projected to age 60 years. This would appear to be an acceptable compromise for dealing with age–risk weighting. How accurate is current risk assessment for uncomplicated mild hypertension? Yikona et al. 10 concluded that the percentage of patients with uncomplicated grade 1 hypertension who were identified for treatment of CHD and CVD risk varied markedly, depending on the risk-assessment method: 100% for JNC-VI table, approximately 65% for WHO/ISH table (note that management strategy permits instituting drug therapy after a period of observation in selected patients in both medium- and low-risk groups) and approximately 22–28% for the other three more strict absolute risk-assessment methods. The sensitivity of the risk-assessment methods in general and the WHO/ISH risk-assessment method in particular has been challenged recently by the Assessment of Prognostic Risk Observational Survey (APROS study) 26, which reported on 1074 untreated hypertensive subjects attending 44 outpatient hypertensive clinics across Italy. They found that 53% of patients previously classified as mild or medium risk by the WHO/ISH classification were reclassified as high risk (10-year risk = 20%) after finding LVH on echocardiography and/or finding significant increased diffuse or focal carotid intima–media thickening by ultrasound examination. The large number of participating clinics and wide geographical representation of this study would tend to render the results representative of a general hypertensive population. Note that the WHO/ISH table assessment by Yikona et al. 10 showed a high sensitivity of 95% which, in retrospect, was falsely high, as judged by the APROS study. At present, the use of echocardiography and/or carotid ultrasound in patients with uncomplicated grade 1 hypertension have not been advocated as routine procedures, largely on the basis of cost–benefit analysis 27. However, the results of this study indicate that current risk-assessment methods, based on Framingham risk equations 13,28, may grossly under-represent uncomplicated grade 1 hypertensive patients who are at high risk from unsuspected LVH and from subclinical disease, such as carotid artery intima–media thickening. Conclusions Although the absolute risk-assessment methods reviewed by Yikona et al. 10 may lack sufficient sensitivity, they still represent an improvement over what was previously available in the era when it was believed that only the level of blood pressure and prior cardiovascular disease were relevant to therapeutic-decision making. However, in the final analysis, current treatment decisions for uncomplicated grade 1 hypertension represent a composite of underestimated absolute risk calculations, of actuarial life expectancy assessment in younger subjects, and of what the informed patient is willing to accept in terms of therapy. Indeed, to date, cardiovascular risk evaluation is an inexact science. Acknowledgements The authors wish to thank Drs. Daniel Levy and Marvin Moser for their expert scientific and editorial review of the manuscript.