Diabetes, hyperglycemia, and coronary heart disease

PREVENTIVE

MEDICINE

12,

Diabetes,

70-74 (1983)

Hyperglycemia,

and Coronary

GEOFFREY London

School of Tropical

Medicine,

Heart Disease’

ROSE

Kepprl

Street, London WCI, Englund

An international collaborative group examined the relation between asymptomatic hyperglycemia and coronary heart disease (CHD). After excluding clinically diagnosed diabetes and treated hypertensives, there were 48,912 men aged 35-64 in the study population. Post-load circulating glucose correlated with age, relative weight, and systolic blood pressure. The prevalence on ECG of S-T/T findings, but not of Q/QS items, was higher in the top quintile of glycemia. For CHD mortality there was no consistent, strong, or independent association with glycemia. In the Whitehall Study (the largest), it appeared that there was a threshold effect, CHD mortality being about twice as high above the 95th percentile of glycemia. Stroke mortality was increased to a similar extent.

INTRODUCTION

Diabetes mellitus greatly increases the risk of developing coronary heart disease (CHD), particularly in women (9). For many years it has been suspected that asymptomatic impairment of glucose tolerance represents a precursor or milder form of the same disease, and that it likewise may increase the risk of atherosclerosis. This suspicion has led some to advocate screening for asymptomatic hyperglycemia. THE INTERNATIONAL

COLLABORATIVE

STUDY

In previous longitudinal studies detailed analysis has been limited by the size of the study group, while differences in methods have prevented pooling or comparisons. This led Dr. Jeremiah Stamler to bring together investigators from 15 such studies to look for an association between glycemia and (a) CHD prevalence (electrocardiographic findings), or (b) CHD mortality, and to examine whether any such associations were independent of other risk factors. Study Populations

The 15 studies came from 11 countries, representing 48,912 men aged 35 to 64 years; there were five community samples, nine occupational groups, and a group of volunteers for screening (8). Men already under care for diabetes (2.0%) or hypertension (4.4%) were excluded, and men whose initial electrocardiogram showed evidence of myocardial infarction were excluded from the mortality analysis. Problems

of Comparability

Like the Pooling

Project

(5) but unlike

the Seven Countries

Study (3) the results

of different studies were brought together after they had been completed. We 1 Presented at the International Symposium on Epidemiology and Prevention of Atherosclerotic Disease, June 24-26, 1981, Anacapri, Italy. 70 0091-7435/83/010070-05$03.00/O Copyright All rights

@ 1983 by Academic Press, Inc. of reproduction in any form reserved.

SYMPOSIUM:

ATHEROSCLEROTIC

71

DISEASE

therefore made no attempt either to pool the findings, or to compare their absolute levels. Within each study we contrasted the normoglycemic and hyperglycemic groups, and then looked for any general consistency in the patterns that emerged. The most troublesome diversity of methods concerned the measurement of glycemia. In 13 studies measurements followed an oral load, but neither load nor interval was constant. For most analyses we simply compared the within-study linear correlation coefficients or the outcomes in high and low quintiles. But if relations are not linear, or if some absolute level of glycemia is critical, this might not be recognized in such an analysis. interrelations of Variables Table 1 presents a matrix for each study of simple correlations of glycemia. The correlation with age is consistent but not strong. With cholesterol it is smaller and sometimes absent. The strongest correlations are with relative weight and blood pressure, especially systolic pressure. This is surprising, in view of evidence that the association of relative weight with blood pressure is primarily with diastolic, not systolic. This finding, first reported in the Israeli Heart Study (2) and confirmed in the Whitehall Study, suggests that different mechanisms link blood pressure with relative weight and with glycemia. Prevalence of Positive ECG Findings The prevalence of findings known to be associated with risk of major CHD (Minnesota Code 1.1-3, 4.1-3, 5.1-3, 7.1) (7) increased in the top versus the lowest quintile of glycemia in 13 of the 15 studies. The specificity of such findings, however, is in some doubt, particularly during post-load glycemia (4). Analysis based on more specific criteria for myocardial infarction (Minnesota Code 1. 1- 2) did not show any association with glycemia, although numbers were small.

TABLE 1 DISTRIBUTIONSOFSIMPLE CORRELATIONSBETWEENBLOODSUGARAND OTHERRISK FACTORS Correlation with glycemia >0.2.5 0.20-0.24 0.15-0.19 0.10-O. 14 0.05-0.09 0.00-0.04 -co.00

No. of studies Age”

3 3 6 1

n Two studies of single years of age.

Cholesterol

BMI

1

2 3 4 3 3

2 4 5 3

SBP 4 4 3 1 3

DBP 4 4 2 2 3

GEOFFREY ROSE

72 CHD Mortality

If hyperglycemia influences case fatality from CHD, prevalence may not reflect incidence. Therefore, age-adjusted cause-specific mortality has been analyzed in 11 studies with adequate follow-up (4 to 15 years); 608 men died of CHD (Table 2). In three studies, with 113 CHD deaths, the mortality ratio (top versus lowest quintiles) was >2.0; and in four studies, with 398 deaths, it was ~1.0. Considered alone, glucose tolerance did not predict death from CHD in any general or consistent way. This conclusion was generally supported by multivariate analysis, adjusting for age, relative weight, systolic blood pressure, cholesterol, and smoking. There was a highly significant result from one study (Chicago Peoples Gas), but taken as a whole, the conclusion is negative. It seems from the Collaborative Study that there is no consistent, strong, or graded association of CHD with asymptomatic hyperglycemia, although some association at the highest levels could not be excluded. THE WHITEHALL

STUDY

The Whitehall Study of London civil servants was the largest contributor to the Collaborative Study. Since that report was prepared, follow-up results have been extended, with CHD deaths increasing from 171 to 425. This gives more scope for detailed analysis, particularly in the higher part of the blood sugar distribution (1). CHD and Blood Sugar above the 95th Percentile

In 5% of men studied, 2 hr after a 50-g load glycemia was 96 mddl(5.3 mmole/ liter) or more. Below this level the age-adjusted lo-year CHD mortality rate was 2.3%; above, it was 4.9%. Over the lower 95% of the range there was-as in the Collaborative Study-no relation at all between blood sugar and risk (Fig. 1). Above, there was no evident further increase (although, with only 52 CHD deaths in this group, some continuing gradient cannot be excluded). Table 3 examines the predictive independence of other major risk factors in normoglycemic and hyperglycemic (i.e., above the 95th percentile) men. The effects of ECG abnormality and of blood pressure in hyperglycemics are similar to those in normoglycemics, while effects of cholesterol and smoking are perhaps reduced. Overweight does not reach significance in either group. Multiple logistic TABLE 2 DISTRIBUTIONOF AGE-ADJUSTEDCHD MORTALITY RATIOS, TOP:BOTTOMQUINTILES OF GLYCEMIA( 11 STUDIES) CHD mortality ratio (Ql:Q5)

Study

Cl.0 1.0-1.4 1.5-1.9 >2.0

England, Switzerland, USA (W. Electric and Chicago Heart) Denmark (50 yr), Finland (Sot. Ins.), Scotland Australia Finland (Police), France, USA (Peoples Gas)

SYMPOSIUM:

ATHEROSCLEROTIC

73

DISEASE

60 I c 50 & I

$40

ZO(O%)

66(20%)

72(40%) 78(60%1 2 hour blood sugar (mg/dll

85(80%)

g 2 00 E3” gx

FIG. 1. Age-adjusted 7!&year CHD mortality according to glycemia, Whitehall above columns are deaths; percentages below are percentile points of glycemia.

$ 0 $2 B a

Study. Numbers

analysis does not alter these conclusions; moreover the predictive effect of hyperglycemia persists even when adjusted for each of the other variables. Hyperglycemia

and Stroke

CHD did not account for the whole excess associated with hyperglycemia, and on further analysis a relation appeared with fatal stroke. This persisted after adjusting for age and blood pressure (Table 4). It seems not to have been reported before. CONCLUSIONS

For most individuals in the populations we have studied, post-load glycemia does not predict the occurrence of CHD, and therefore it presumably does not TABLE FOR VARIOUS LEVELS,

3

CORONARY RISK FACTORS, THE RATIO OF CHD MORTALITY AT HIGH AND Low SEPARATELY FOR THE GROUPS BELOW AND ABOVE THE 9%~ PERCENTILE OF BLOOD SUGAR (WHITEHALL STUDY)

7%Year Normoglycemics PI. cholesterol (Q5:Ql-4) Systolic BP (Q5:Ql-4) BMI (QYQl-4) Cigarettes (yes:no) ECG (positive:negative)

1.8 1.8 1.1 2.2 3.8

mortality

ratio (age-adj.) Hyperglycemics 1.2 2.1 1.4 1.5 3.4

GEOFFREY ROSE

74

TABLE 4 TEN YEARSTROKE MORTALITY(ADJUSTED FORAGEANDSYSTOLIC BLOODPRESSURE) ACCORDINGTOGLYCEMIA(WHITEHALL STUDY)

Percentile class of glycemia

IO-Year stroke deaths per 1000” (No. of deaths)

<90 90s95

4.6 (69) 6.9 (7) 8.6 (13)

New diabetics

9.1 (3)

0 Adjusted for age and systolic BP.

cause it. This is surprising, in view of the much increased CHD risk in diabetics. The threshold effect which emerged in the largest of our collaborative groups is quite different from the graded relation which exists for smoking, blood pressure, and cholesterol, suggesting prehaps that the hyperglycemia is not itself causing arterial disease. Further evidence from the two Finnish studies (6) suggests that the real underlying factor could be the level of plasma insulin. ACKNOWLEDGMENT The coordination and analysis of the collaborative group were undertaken by Mrs. Rose Stamler and Dr. Jeremiah Stamler, and those results are presented on behalf of the collaborative group.

REFERENCES 1. Fuller, J. H., Shipley, M. J., Rose, Cl., Jarrett, R. J., and Keen, H. Coronary heart-disease risk and impaired glucose tolerance. Lancet 2, 1373- 1376 (1980). 2. Goldbourt, V., Medalie, J. H., and Neufeld, H. N. Clinical myocardial infarction over 5-year period. III. A multivariate analysis of incidence. The Israeli Ischaemic Heart Disease Study. J. Chronic Dis. 28, 217-237 (1975). 3. Keys, A. “Seven Countries. A Multivariate Analysis of Death and Coronary Heart Disease.” Harvard Univ. Press, Cambridge, Mass., 1980. 4. Ostrander, L. D., and Weinstein, B. J. Electrocardiographic changes after glucose ingestion. Circulation 30, 67-76 (1964). 5. Pooling Project Research Group. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: Final report of the Pooling Project. J. Chronic Dis. 31, 201-306 (1978). 6. Pyorala, K. Relationship of glucose tolerance and plasma insulin to the incidence of coronary heart disease: Results from two population studies in Finland. Diabetes Care 2, 131- 141(1979). 7. Rose, G., Baxter, P. J., Reid, D. D., and McCartney, P. Prevalence and prognosis of electrocardiographic findings in middle-aged men. Bit. Heart J. 40, 636-643 (1978). 8. Stamler, R., and Stamler, J. (Eds.) Asymptomatic hyperglycaemia and coronary heart disease. A series of papers by the International Collaborative Group, based on studies in fifteen populations. J. Chronic Dis. 32, 683-837 (1979). 9. West, K. M. “Epidemiology of Diabetes and its Vascular Lesions,” p. 159. Elsevier, New York, 1978.