- Email: [email protected]
A population at risk: Prevalence of high cholesterol levels in hypertensive patients in the framingham study
A Population at Risk Prevalence of High Cholesterol Levels in Hypertensive Patients in the Framingham Study
In the Framingham Study, coronary heart disease devetoped in every fffth man and every 17th woman by the age of sixty. The level of total cholesterol proved to be an excellent predictor of coronary heart disease in those aged less than 50 years. However, in those aged over 50 years, more accurate predictors of coronary heart disease risk were serum lipoprotein measurements, including low-density lipoproteins, very-low-density lipoproteins, very-low-density lipoprotein triglycerides, and high-density lipoproteins. Both lowdensity and very-low-density lipoproteins have a linear associatton wtth coronary heart disease. On multivariate analysis, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol are independently related to coronary heart disease risk in both sexes. In women, but not in men, very-low-density lipoprotein cholesterol or the triglyceride level is an independent risk factor on multivariate analysis. By f’ikelihood ratio analysis, high-den&y lipoprotein is shown to be the most powerful single factor for predicting coronary heart disease risk in both sexes relative to the lipid fractions. It appears that one of the most reliable profiles in this regard is the ratio of total cholesterol to high-density lipoprotein cholesterol. However, a special constellation of elevated triglycerides, low highdensity lipoprotein levels, and “normal” cholesterol values should no longer be overlooked in assessing coronary heart disease risk. Both systolic and diastolic blood pressures are also related to risk of coronary heart disease in a linear fashion: the higher the level of pressure, the greater the incidence of coronary heart disease. Blood pressure and serum cholesterol are correlated with an r factor of 0.12, suggesting that those with higher blood pressure values tend to have higher serum cholesterol levels. Most physicians agree that treatment is advisable in those with cholesterol levels above 300 mg/dl; some believe therapy is necessary in those with levels of more than 250 mgldl. Few realize that half of the patients in whom coronary heart disease will evehtually develop have cholesterol values under 250 mg/dl. The National Institutes of Health Consensus Development Conference on Lipid Lowering has recommended that cholesterol levels be reduced to 200 mg/dl in all persons. Practicing physicians argue that coronary heart disease does not develop in most patients with cholesterol levels between 200 and 250 mg/dl. The problem lies in deciding which patients with these cholesterol levels actually have a lipid abnormality. Measuring high-density lipoprotein levels may provide an answer. In the Framingham Heart Study, coronary heart disease developed with great consistency in patients with a ratio of total cholesterol to high-density lipoprotein
WILLIAM P. CASTELLI, M.D. KEAVEN ANDERSON, Ph.D. Framingham,
Massachusetts
From the Framingham Heart Study, Framingham, Massachusetts. Requests for reprints should be addressed to Dr. Will/am P. Castelli, Framingham Heart Study, 118 Lincoln Street, Framingham, Massachusetts 01701.
February
14,1936
The
AmerMn
Journal
of Mediclne
Volume
30
(suppl PA)
23
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
factors studied were introduced at later examinations. Subjects were evaluated systematically for development of coronary heart disease, stroke, peripheral vascular disease, congestive heart failure, and other vascular diseases, as well as for many other medical conditions, e.g., cancer and peptic ulcers. Also recorded were specific end points of coronary heart disease, including angina pectoris, coronary insufficiency, and myocardial infarction, or the mortal end points such as sudden death or nonsudden coronary heart disease death. Definitions of these conditions are described in greater detail elsewhere [3]. Briefly, angina was defined as chest discomfort brought on with exertion and relieved by rest within 15 minutes; two examiners had to agree it was angina. Coronary insufficiency was defined as prolonged chest discomfort accompanied by transient nonspecific changes on the electrocardiogram-usually ST-segment depression or inverted T waves-and not associated with enzyme changes. Myocardial infarction was defined as prolonged chest discomfort (more than 15 minutes) associated with pathognomonic Q wave changes and/or enzyme changes. Sudden death was defined as death within one hour in a person without any other illness or explanation. Coronary heart disease death (nonsudden) was defined as death in a person having an acute episode of angina, coronary insufficiency, or myocardial infarction who took more than one hour to die. Lipid analyses were performed using the protocols established by the Laboratory for Molecular Disease of the National Heart, Lung, and Blood Institute [4]. Total cholesterol was determined using the Abell-Kendall method (51, and triglycerides were determined by the Kessler-Lederer method [6]. Pure primary standards were made from cholesterol and triolein purchased, for the most part, from Sigma Chemical Corporation (St. Louis). Quality control specimens were used in each analytic run. Periodic checks of levels and accuracy were made by performing tests on samples from the Communicable Disease Center. (Atlanta, Georgia). In general, coefficients of variation were under 5 mg/dl for cholesterol and under 9 mgl dl for triglycerides. The evaluation comparing baseline characteristics with the subsequent development of coronary heart disease used a logistic function, parameters of which were estimated by the method of Walter-Duncan [7]. These were standardized by multiplying them by the standard deviation of the characteristic. Such standardized coefficients may be used for an approximate comparison of the strength of the risk factor for coronary heart disease. Likelihood ratios of the logistic functions indicate how well the risk functions fit the actual data.
cholesterol of more than 4.5. Half of the women and more than half of the men who present with treatable hypertenslon already have an abnormal lipid profile. Control of coronary heart disease should be multifactorial. Intervention should include therapy aimed at reducing blood pressure and lowering lipid levels, particularly low-density lipoprotein and very-lowdensity lipoprotein triglycerides. Since many antihypertensive agents have an adverse effect on lipids, considerable thought should be given to those drugs that represent a therapeutic alternative.
In general, today’s physicians respond promptly to counteract the potential dangers of elevated blood pressure, whereas their reaction to elevated cholesterol levels is much slower. Although there is a certain ambiguity concerning the actual levels of blood pressure that require treatment, the consensus is that a diastolic blood pressure of 90 mm Hg and a systolic blood pressure of 140 mm Hg are the accepted cutoff values separating normotensive from hypertensive patients although in some antihypertension trials, a diastolic blood pressure of under 90 mm Hg was the goal of therapy [1,2]. Unfortunately, no generally accepted “cutoff” numbers exist for cholesterol values; traditionally, cholesterol intervention trials have not quoted an actual level of cholesterol as a goal of therapy. Rather, the goal has been simply maintenance of a certain diet or therapy with a specific amount of a cholesterollowering agent. This article reviews the relationship between cholesterol and the subsequent development of coronary heart disease as it was demonstrated in the Framingham Heart Study. We consider not only total cholesterol but also low-density lipoprotein cholesterol, very-low-density lipoprotein cholesterols or triglycerides-a better measure of very-low-density lipoproteins-and high-density lipoprotein cholesterol. Not considered herein is the cholesterol found in chylomicrons or in intermediate-density lipoproteins that is usually negligible in most persons in freeliving populations who have fasted for 12 to 14 hours. We will try to define which levels and which types of cholesterol lead to coronary heart disease so that we can establish some easily recognizable parameters for treatment. Another objective of this article is to explore the interaction of cholesterol levels and hypertensive levels that need to be treated concomitantly.
RESULTS One of the first findings of the Framingham Heart Study was the incidence of coronary heart disease in this population. Figure 1 shows the rate of coronary heart disease in the first 14 years. The disease is more prevalent in men than in women and increases with age in both sexes. In practical terms, this graph shows that some form of coronary heart disease developed in every eighth man aged 40 to 44 years in the first 14 years of this study. Rates for a similar period of time were every sixth man aged 45 to 49 years, every fifth man aged 50 to 54 years, and every
SUBJECTS AND METHODS At the start of the Framingham study in 1949, a random sample of 50 percent of the population identified 5,127 men and women who had no evidence of cardiovascular disease (31. This group has been followed by medical examinations at two-year intervals for almost 36 years. Data in this report are derived from various stages of follow-up, since some of the
24
February
14,1986
The American
Journal
of Medlclne
Volume
80
(suppl 2A)
SYMPOSIUM
fourth man aged 55 years or more. In young women, coronary heart disease developed at a much lower rate than it did in men. Of the 1,600 premenopausal women, coronary heart disease developed in only six prior to menopause. In postmenopausal women, however, coronary heart disease developed at the same rate as in men who were 10 to 15 years younger. The most negative finding is that coronary heart disease had developed in approximately every fifth man and every 17th woman by age 60. Figure 2 shows the relationship of low-density lipoprotein and very-low-density lipoprotein to morbidity. At the time of participant entry into the Framingham Heart Study, these factors were measured in an analytic ultracentrifuge, courtesy of John Gofman and associates [8], at the Donner Laboratories of the University of California in Berkeley. The higher the low-density lipoprotein or verylow-density lipoprotein level, the greater the subsequent
ON INITIAL ANTIHYPERTENSIVE
Figure 2. Rate of coronary heart by lipoprotein levels at time (Framingham Heart Study, follow-up). LDL = low-density tein; VLDL = very-low-density tein.
and ANDERSON
300M *--*
Men Women
200-
Rate per l.ooO loo
-
-0
_ 7.
30-34
35-39
-
I 40-44
I I 45-49 50-54 Age at entry
I 55-59
. .
.
I 60-62 .,.
FourTeen-year mclaence of coronary nean a/sease (all clinical manifestations) according to age and sex: men and women aged 30 to 62 years at entry (Framingham Heart Study).
hgure
Lipoprotein Sf O-20 ma WL)
Morbidity ratio
THERAPY-CASTELLI
Lipoprotein Sf 20400 Pre-8&a (VLDL)
I
I
100
disease of entry 14-year lipoprolipopro-
HDL cholesterol
Men
Low:
40
Triglycerides 44
Medium: 40-50 Hiah: Xl
94-144 >144
100 CHD cases/ 1,ooo
low med high Triglycerides Figure 3. Incidence of coronary heart disease (CHD) by levels of high-density lipoprotein (HDL) cholesterol and Wiglycerides in men. Adapted from [9].
*Cell with 45
February
persons
14, 1999
HDC c&sterol at risk
The American
Journal
low med high Triglycerides Medium HDL cholesterol
of fdedlcine
Volume
low med high Triglycerides Hi h HDL ch ttee ted
99
(suppl 2A)
25
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
TABLE I
THERAPY-CASTELLI
and ANDERSON
Univariate and Multivariate Logistic Regression Coefficients for Coronary Heart Disease and Lipoproteins in Men and Women Aged 50 to 80 Years (Framingham Heart Study) Univariate
Lipopmteins
Men
High-density lipoprotein cholesterol Low-density lipoprotein cholesterol VetjAow-density lipoprotein cholesterol (triglycerides)
-0.506
Multivariate
Women -0.682’
Men
Women
-0.610
-0.650+
0.241’
0.266
0.333*
0.280’
0.032
0.325’
0.036
0.227*
‘p CO.05. +p
rate of coronary heart disease. These analyses were repeated using the Beta Quantification Scheme developed by the Laboratory for Molecular Diseases of the National Heart, Lung, and Blood Institute [4]. Results show (Table I) that in those aged over 50 years, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol are powerful and independent predictors of cqronary heart disease. Triglycerides were not strong predictors in men using the traditional multivariate analysis, but they were in women. Actually, there is probably a “symptom complex” in both sexes involving triglycerides that is strongly related to coronary heart disease. It is characterized by a normal level of cholesterol, a high level of triglycerides, and a low level of high-density lipoprotein. Figure 3 illustrates this special relationship in men [9]. At medium to high levels of high-density lipoprotein, there is no relationship of triglycerides to risk; such a relationship is seen only at very low
CHD rate/l ,000 5025 c25
25-34
4544
55-64
HbL cholesterol
3544
(mg/dL)
6574
75+
Figure 4. Incidence of coronary heart disease (CHD) by high-density lipoprotein cholesterol (HDL-C) level in men and women aged 50 to 80 years (Framingham Heart Study).
26
February
14,1966
The American
Journal
of Medicine
Volume
levels of high-density lipoprotein. Not only do persons with this complex have a higher incidence of coronary heart disease, they also have an incidence of diabetes mellitus that is twofold the rate in the general population and are usually overweight. Whether a high triglyceride level or obesity are truly independent risk factors in multivariate analysis in this setting may be academic. Clinicians should be aware of the constellation of findings: overweight, high triglycerides, average total cholesterol and low-density lipoprotein cholesterol, and low high-density lipoprotein all most likely adding to the risk of diabetes and coronary heart disease development. High-density lipoprotein cholesterol was found to be inversely related to the subsequent development of coronary heart disease, as shown in Figure 4. This relationship is independent on multivariate analysis, as was seen previously in Table I. Table II is a likelihood ratio study that not only shows the strength of high-density lipoprotein as a powerful predictor of coronary heart disease risk, but also demonstrates that the ratio of total cholesterol to high-density lipoprotein cholesterol is an extremely potent predictor by itself. Although the total cholesterol level was an important and independent predictor of coronary heart disease in those aged under 50 years, it no longer predicts the risk (Figure 5) in those aged over 50 years. Indeed, the relationship of the total cholesterol level to coronary heart disease risk is best seen in Figure 6, in which the solid line is the distribution of the cholesterol levels of all the men in Framingham in whom coronary heart disease did not develop in the first 16 years of our study. The non-solid line is the distribution of the cholesterol levels of all the men in whom coronary heart disease developed in the first 16 years. Many clinicians, particularly those with an expertise in lipids, are generally concerned about the total cholesterol level if it is 300 mg/dl or higher. In all likelihood, such patients have monogenic or polygenic hypercholesteroiemia and coronary heart disease will develop early in life. Often, many of these patients’ first-degree blood relatives will be similarly affected. Although this is an important population to identify and treat, those persons with cholesterol levels between 150 and 300 mg/dl should also be considered. About three quarters of the myocardial infarctions in our society occur in people whose cholesterol levels are between 150 and 300 mg/dl. Some experts suggest that 250 mg/di be used as the cutoff number between normal and abnormal values. Anyone with a cholesterol level over 250 mg/dl should be considered to have a “cholesterol problem.” The average man in Framingham who had a myocardial infarction had a cholesterol level of 244 mg/dl, and coronary heart disease developed in more persons with that level than with any other. Is there a cholesterol level threshold for coronary heart disease? is there a value below which myocardiai infarctions do not occur, regardless of the presence of other risk 60
(suppl2A)
JM ON INITIAL ANTIHYPERTENSIVE
TABLE II
factors? In Framingham, in 30 years, we have yet to report a myocardial infarction in any person whose cholesterol was 150 mg/dl or below. Of course, a simple inspection of Figure 6 will reveal that there is no one in Framingham with a cholesterol level under 150 mg/dl. However, three quarters of the people in the world will not have a myocardial infarction in their lifetime. These people live in Asia and Africa, and in South America outside of big cities; they are poor, rarely eat meat, and their cholesterol values are all near the 150 mg/dl level. In our own society, half of the myocardial infarctions occur in people with cholesterol levels below 250 mg/dl, a level considered “normal” by many physicians. Recently, the National Institutes of Health Consensus Development Conference on Lipid Lowering [lo] recommended that all Americans reduce their cholesterol levels to 200 mg/dl or below. Many physicians have argued that most of their patients with cholesterol levels between 200 and 250 mg/dl will not have a myocardial infarction. Although this may be valid, it is also true that half of the myocardial infarctions they see in their practices will occur in these same patients. Obviously, the total cholesterol value cannot accurately predict which patients have a lipid problem when the cholesterol levels are between 200 and
I
30-49 Years
THERAPY-CASTELLI
and ANDERSON
Likelihood Ratios for Various Lipid Profiles of Coronary Heart Disease (Framingham Heart Study, examination 11) lipid pmfile
High-density lipoprotein Low-density lipoprotein Triglyceride Total cholesterol High-density lipoprotein cholesterol Low-density lipoprotein triglyceride High-density lipoprotein triglyceride High-density lipoprotein lipoprotein cholesterol, High-density lipoprotein lipoprotein cholesterol High-density lipoprotein cholesterol, low-density cholesterol
Men
cholesterol cholesterol
Women
14.63’ 4.39t 0.51 1.98 17.11*
21.21* 4.53t 9.5P 2.26 20.41’
and total cholesterol,
8.28’
19.69’
and total cholesterol,
19.19
24.21’
and lowdenslty triglyceride and low-density
18.96
24.73
18.66
23.70’
cholesterol/total lipoprotein
17.16
26.77’
cholesterol/total
Note: This table estimates the relative power of a test or set of tests to predict coronary heart disease from asymptomatic subjects. The higher the number, the better the ability of the test(s) to predict.
ip
I
?
Morbidity ratio 100
----
Men Women
219
239
2s
279
299
c1So1So-20@220-240-260-2Scb300+ 190
219
239
259
279
299
Level of serum cholesterol at initial exam (mg/dL)
fgure 5. Risk of coronary heart disease by cholesterol ham Heart Study, 14-year follow-up).
level at time of entry in men and woman aged 30 to 62 years (Framing-
February 14,1986
Tbe Am&can
Journal of Medicine
Volume 80 (auppI 2A)
27
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
40-
-
and ANDERSON
THERAPY-CASTELLI
Non-CHD
No.
Mean
SD
1,370
219
41
193
244
51
30-
Figure 6. Distribution of total cholesterol levels in men in whom coronary heart disease developed and in men in whom there was no evidence of coronary heart disease (Framingham Heart Study, 1b-year follow-up).
0 150
200
250
300
350
Level of serum cholesterol at initial exarninatbn (mg/dL)
250 mg/dl, or even between 150 and 250 mg/dl. It is in these situations that the high-density lipoprotein cholesterol level becomes an important factor. As Table III shows, for every 5 mg/dl that the high-density lipoprotein cholesterol level decreases below 45 mg/dl, the risk of coronary heart disease increases by 25 percent. Of course, the reverse is also true: for every 5 mg/dl that high-density lipoprotein cholesterol increases above 45 mg/dl, risk drops by about 12.5 percent. We believe that the best way’to predict risk is to calculate a ratio of the total cholesterol to high-density lipoprotein cholesterol (Figure 7), recognizing that the higher the number, the greater the risk [ll]. We recommend keeping the ratio below 4.5. The distribution of our cases ranged from approximately 3.5 and above. However, if we recommended
TABLE Ill
treating everyone with a ratio of 3.5 or higher, we would be treating approximately 90 percent of all Americans. Instead, we have chosen 4.5 as a first-step goal because we already have a subgroup of women under observation in Framingham in whom coronary heart disease developed and whose average ratio was 4.6; this is also a more manageable group. Half of the women and two thirds of the men seen by physicians in America have ratios above this value; more than half of the women with hypertension and more than two thirds of the men with hypertension also have ratios above this value. In Framingham, we found that blood pressure increases with age, and by the time our population reaches their late 50s and early 6Os, almost a third of the men and women have hypertension (Figure 8). Blood pressure (to which we have assigned arbitrary cutoffs) is related to cardiovascular disease in a linear fashion. As seen in Figure 9, hypertensive patients have a twofold to threefold greater incidence of coronary heart disease and a sevenfold greater incidence of stroke; 75 percent of the congestive heart failure occurred in people with uncontrolled hypertension. Even borderline hypertension (diastolic blood pressure of 90 to 94 mm Hg; systolic blood pressure of 140 to 159 mm Hg) carries higher risks. We found a 50 percent increase in myocardial infarction rates and a 300 percent increase in the rate of stroke in borderline hypertensive patients compared with normotensive subjects. Elevated blood pressure is not tolerated well with age, and the relative and absolute impact of blood pressure on cardiovascular diseases worsens as age increases. There is not a strong correlation between blood pressure and serum cholesterol levels. As Table IV shows, the strongest correlates approach 0.20; blood pressure increases as cholesterol increases, but not dramatically. Nonetheless, it should be remembered that more than half of the patients with hypertension, or those requiring treat-
Multlpller for High-Demslty Llpoproteln Cholesterol Level
High-Density Lipopmteln Cholesterol (WW 30 35 40 45 50 55 60 65 70 75
Men
Women
1.82 1.49 1.22 1 .oo 0.62 0.67 0.55 0.45 -“Longevity
1.94 1.55 1.25 1.00 0.60 0.64 0.52 syndrome”-
Note: This table shows how risk increases or decreases in a relative sense in men and women. Since men have an average high-density lipoprotein cholesterol level of 45 mg/dl, no change in a risk profile occurs for men with this value. However, as highdensity lipoprotein cholesterol levels decline, one would multiply the risk profile value (e.g., calculated from Table V) by its appropriate number.
26
February
14,1966
The American
Journal
of hkdclne
Volume
80 (suppl
PA)
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
Ratio is determinsd by dividing totatchotesterolinbbodbyths amount of beneficial HDL Half the Average Risk of Heart Disease Boston Marathon Runners (Physicians) Average Disease
3.4
m
Women
Risk of Heart in Americans
Average Victim of Heart Disease
Goal 4.5
Twice Average Risk Triple Average
Figure 7. high-density
Ratio of total cholesterol lipoprotein
9.6
Risk
to
.4 0
4
a
12
16
20
24
cholesterol.
I
I
I
I
I
I
30
35
40
45
50
55
Figure 8. Prevalence of hypettension (Framingham Heart Study).
1 65
Age at entry
0
Normotensive
a
q
Borderline
(< 140/90)
300
kerage Annual Incidence per 10,000
Coronary heart disease
Intermittent
(>160/95) claudication
100
100 lverage Annual Incidence per 10,000
Hypertensive
200
0
Figure 9. Age-adjusted risk of cardiovascular morbidity according to hypertensive status at each biennial examination in men and women aged 45 to 74 years (Framingham Heart Study cohort, 18-year follow-up).
I 60
Wcirnen Men Congestive failure
Women Men Atherothrombotic brain interaction
a7
50
0
Women Men Women Men All trends statistically significant at P < 0.01
February
14, 1996
The American
Journal
of Medicine
Volume
60 (suppl
2A)
29
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
TABLE IV
THERAPY-CASTELLI
and ANDERSON
Correlation of Serum Total Cholesterol with Systolic Blood Pressure in Men and Women (Framingham Heart Study) CormlationCoelficieni Ap&?
Men Systolic Diastolic Women Systolic Diastolic
30-34
35-39
40-44
45-49
SO-54
55-59
60-M
blood pressure blood pressure blood pressure blood pressure
20 260* 240*
15 4-YW
incidence per 100
200*
10
5
I
I
I
I
I
I
1
I
I
Systolic 100 110 120 130 140 150 160 170 180 190 200 Diastolic 54 62 70 78 86 94 102 110 118 126 134 Blood pressure (mm Hg)
ment of hypertension, also have a lipid abnormality. Furthermore, the more the pressure rises, the greater the likelihood of a lipid problem. This is most easily discerned by calculating the simple ratio of total cholesterol to highdensity lipoprotein cholesterol, and, if the number is 4.5 or higher, treatment for lipid lowering should be initiated. Two of the most important risk fqctors for coronary heart disease are elevated levels of cholesterol and blood pressure. As Figure 10 shows, there is a powerful interaction between these two factors to produce coronary heart disease. Treating one factor is doing only half of the job, but if the therapy for decreasing blood pressure increases the cholesterol level as well, the beneficial effects of antihypertensive therapy will be offset or negated by the adverse lipid effects on coronary heart disease. Careful monitoring of cholesterol levels and blood pressure is necessary to assure that good control of both is achieved. 30
February 14,1986
The Amerlcan Journal of Madicina
Figure 10. Four-year incidence of coronary heart disease according to systolic blood pressure at specified levels of serum of men aged 55 years (Framingham Heart Study).
COMMENTS Coronary heart disease is a multifactorial problem, and there are numerous handbooks available for calculating exact individual patient risk. One method is shown in Table V. In clinical practice, however, it is usually more effective to approach each problem separately and establish specific therapeutic goals (Table VI) for each. One of the dilemmas of lipid management is that there is no established goal of therapy similar to that used in blood pressure management. For example, regardless of the initial diastolic blood pressure or systolic blood pressure that prompted the physician to begin therapy, the goal of that therapy will probably be to reduce blood pressure to 140/90 mm Hg. We believe that patients’ lipid profiles should also be treated with a specific goal in mind: a ratio of total cholesterol to high-density lipoprotein cholesterol of less than 4.5. Similar strategies should be used to conVolume 60 (suppi 2A)
SYMPOSIUM
TABLE
V
ON INITIAL ANTIHYPERTENSIVE
Probability (per 1,000) of Cardiovascular Disease Developing Characteristics (Framingham Heart Study, la-year follow-up
in Eight Years According in 50-year-old men)
Do not smokecigarettes SBP LVH-ECG
Glucose intolerance absent
Glucose intolerance present
LVH-ECG
Glucose intolerance absent
Glucose intolerance present
negative Cholesterol 185 210 235 260 265 310 335 Cholesterol 185 210 235 260 285 310 335 posttlve Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335
THERAPY-CASTELLI
and ANDERSON
to Specified
Smokecigarettes
105
120
135
150
165
180
195
37 44 53 83 76 91 108
46 55 66 79 94 112 133
57 69 82 98 116 138 162
71 85 102 121 143 168 197
89 106 125 148 174 204 237
110 130 153 180 211 244 282
135 159 187 218 252 290 332
65 78 93 110 130 154 181
81 96 114 135 160 187 218
100 119 140
165 194 226 261
123 146 171 200 233 269 309
151 177 207 241 278 318 361
184 214 249 286 327 371 417
222 257 295 337 381 428 475
98
117 138 163 191 222 257
121 143 169 197 230 266 305
148 175 204 237 274 314 357
181 211 245 282 323 367 412
218 253 291 332 377 423 470
261 300 342 387 433 481 529
309 352 397 444 492 540 587
166 194 226 262 301 343 387
201 234 270 310 352 398 445
241 279 319 362 408 455 503
287 328 372 418 466 514 562
338 382 429 476 524 572 619
392 439 487 535 583 629 672
449 498 546 593 638 682 722
SBP
Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310
105
120
135
150
165
180
195
62 75 89 106 126 148 174
78 92 110 130 154 181 211
96 114 135 159 187 218 253
119 140 165 194 225 261 300
146 171 200 233 269 309 351
177 207 241 278 318 361 407
214 249 286 327 371 417 465
108 128 151 178 208 241 278
133 157 184 215 249 287 328
163 191 222 257 296 337 382
197 230 265 305 347 392 439
237 274 314 357 402 449 497
282 323 367 412 460 508 556
332 376 423 470 519 566 613
160 188 219 253 292 333 377
194 226 262 300 343 387 434
234 270 309 352 397 444 492
278 319 362 408 455 503 551
328 372 418 465 514 561 608
382 428 476 524 572 618 863
439 487 535 582 628 672 713
258 296 338 383 429 477
305 348 393 440 487 536 583
358 403 450 498 546 593 639
413 461 509 557 604 649 691
471 519 567 614 658 700 739
530 578 624 668 709 747 782
588 634 877 718 755 789 819
‘Framingham men aged 50 years have an average systolic blood pressure of 133 mm Hg and an average serum cholesterol Sixty-two percent smoke cigarettes, 1.5 percent have definite lefl ventricular hypertrophy by electrocardiography, and 6.5 percent intolerance. At these average values, the probability of cardiovascular disease developing in eight years is 115/l ,000. SBP = systolic blood pressure; LVH = left ventricular hypertrophy; ECG = electrocardiography.
TABLE
troi blood sugar, weight, smoking, uric acid, lack of physical activity, and possibly stress. Since many hypertensive therapies [12,13] adversely interfere with the lipid profile by lowering high-density iipoprotein, raising triglycerides, and to a lesser extent, raising the total cholesterol level, some investigators [14] believe that such changes may explain why the decline in coronary heart disease incidence was so small in most of the hypertension intervention trials. Two alternatives seem obvious: we must make optimal use of antihypertensive agents that do not adversely affect lipids or, in those cases in which patients are receiving drugs that do disrupt the lipid profile, we should apply lipid-lowering therapies to counteract such effects. We hope this will help to achieve an optimal blood pressure level and have a favorable impact on all the other risk factors as well. February
14,19BB
Vi
Cutoff Values as Goals of Therapy for an Optimal Risk Profile (Framingham Heart Study)
Risk Factor
Level
Total/high-density lipoprotein cholesterol Blood pressure Cigarette smoking Cigar and pipe smoking Diabetes mellitus Weight Electrocardiographic finding Physical activity Stress
The
American
of 236 mg/dl. have glucose
Journal
4.5 or less (unless total cholesterol < 150 mg/dl) Under 140190 mm Hg Must stop Must stop only if inhaled Strict control Keep under body mass index of 22 Any abnormality except first-degree A-V block or rare ectopic beat Regular, 3 to 4 days/week Relax
of Medicine
Volume
80
(suppl 2A)
31
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
REFERENCES 1.
2.
3.
4.
5.
8.
7.
32
Multiple Risk Factor Intervention Trial Research Group: Multiple Risk Factor Intervention Trial. Risk factor changes and mortality results. JAMA 1982; 248: 1465-1477. Hypertension Detection and Follow-Up Program Cooperative Group: I. Reduction in mortality of persons with high blood pressure, including mild hypertension. JAMA 1979; 242: 2582-2571. Dawber TR, Meadors GF, Moore FE: Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health 1951; 41: 279-288. Fredrickson DS, Levy RI, Lees RS: Fat transport and lipoproteins-an integrated approach to mechanisms and disorders. N Engl J Med.1967; 276: 32,94,148,273. Abell LL. Lew BB. Brodie BB. et al: A simolified method for the estimation of total cholesterol in serum and demonstration of its specificity. J Biol Chem 1952; 195: 357-365. Kessler B, Lederer H: Fluorometric measurements of triglycerides. In: Skeggs LT Jr, ed. Technicon symposia automation in analytical chemistry. New York: Medicad Inc., 1965; 341. Walker SH, Duncan DB: Estimation of the probability of an event as a function of several independent variables. Biometrika 1967; 54: 167.
February
14,1936
The American
Journal
of Medlclne
8. 9.
10.
11.
12.
13.
14.
Volume
Gofman JW, Young W, Tandy R: lschemic heart disease, atherosclerosis, and longevity. Circulation 1966; 34: 697. Gordon T, Castelli WP, Hjortland MC, et al: High-density lipoprotein as a protective factor against coronary heart disease. The Framingham Study. Am J Med 1977; 82: 707-714.’ NIH Consensus Panel on Blood Cholesterol: Lowering blood cholesterol to prevent heart disease. JAMA 1985; 253: 20802086. Brody J: New York Times, after Castelli WP: Categorical issues in therapy for coronary heart disease. Cardiol Pratt 1985; January/February: 267-273. Rouffy J, Jaillard J: Effects of two antihypertensive agents on lipids, lipoproteins, and apoproteins A and B. Comparison of prazosin and atenolol. Am J Med 1986; 80 (suppl 2A): lOO103. Grimm RH, Leon AS, Hunninghake DB, et al: Effects of thiazide diuretics on plasma lipids and lipoproteins in mildly hypertensive patients. Ann Intern Med 1981; 94: 7-11. Leren P, Helgeland A, Hjermann I, Holme I: The Oslo study: CHD risk factors, socioeconomic influences, and intervention. Am Heart J 1983; 106: 1200-1206.
80 (suppl
PA)
In the Framingham Study, coronary heart disease devetoped in every fffth man and every 17th woman by the age of sixty. The level of total cholesterol proved to be an excellent predictor of coronary heart disease in those aged less than 50 years. However, in those aged over 50 years, more accurate predictors of coronary heart disease risk were serum lipoprotein measurements, including low-density lipoproteins, very-low-density lipoproteins, very-low-density lipoprotein triglycerides, and high-density lipoproteins. Both lowdensity and very-low-density lipoproteins have a linear associatton wtth coronary heart disease. On multivariate analysis, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol are independently related to coronary heart disease risk in both sexes. In women, but not in men, very-low-density lipoprotein cholesterol or the triglyceride level is an independent risk factor on multivariate analysis. By f’ikelihood ratio analysis, high-den&y lipoprotein is shown to be the most powerful single factor for predicting coronary heart disease risk in both sexes relative to the lipid fractions. It appears that one of the most reliable profiles in this regard is the ratio of total cholesterol to high-density lipoprotein cholesterol. However, a special constellation of elevated triglycerides, low highdensity lipoprotein levels, and “normal” cholesterol values should no longer be overlooked in assessing coronary heart disease risk. Both systolic and diastolic blood pressures are also related to risk of coronary heart disease in a linear fashion: the higher the level of pressure, the greater the incidence of coronary heart disease. Blood pressure and serum cholesterol are correlated with an r factor of 0.12, suggesting that those with higher blood pressure values tend to have higher serum cholesterol levels. Most physicians agree that treatment is advisable in those with cholesterol levels above 300 mg/dl; some believe therapy is necessary in those with levels of more than 250 mgldl. Few realize that half of the patients in whom coronary heart disease will evehtually develop have cholesterol values under 250 mg/dl. The National Institutes of Health Consensus Development Conference on Lipid Lowering has recommended that cholesterol levels be reduced to 200 mg/dl in all persons. Practicing physicians argue that coronary heart disease does not develop in most patients with cholesterol levels between 200 and 250 mg/dl. The problem lies in deciding which patients with these cholesterol levels actually have a lipid abnormality. Measuring high-density lipoprotein levels may provide an answer. In the Framingham Heart Study, coronary heart disease developed with great consistency in patients with a ratio of total cholesterol to high-density lipoprotein
WILLIAM P. CASTELLI, M.D. KEAVEN ANDERSON, Ph.D. Framingham,
Massachusetts
From the Framingham Heart Study, Framingham, Massachusetts. Requests for reprints should be addressed to Dr. Will/am P. Castelli, Framingham Heart Study, 118 Lincoln Street, Framingham, Massachusetts 01701.
February
14,1936
The
AmerMn
Journal
of Mediclne
Volume
30
(suppl PA)
23
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
factors studied were introduced at later examinations. Subjects were evaluated systematically for development of coronary heart disease, stroke, peripheral vascular disease, congestive heart failure, and other vascular diseases, as well as for many other medical conditions, e.g., cancer and peptic ulcers. Also recorded were specific end points of coronary heart disease, including angina pectoris, coronary insufficiency, and myocardial infarction, or the mortal end points such as sudden death or nonsudden coronary heart disease death. Definitions of these conditions are described in greater detail elsewhere [3]. Briefly, angina was defined as chest discomfort brought on with exertion and relieved by rest within 15 minutes; two examiners had to agree it was angina. Coronary insufficiency was defined as prolonged chest discomfort accompanied by transient nonspecific changes on the electrocardiogram-usually ST-segment depression or inverted T waves-and not associated with enzyme changes. Myocardial infarction was defined as prolonged chest discomfort (more than 15 minutes) associated with pathognomonic Q wave changes and/or enzyme changes. Sudden death was defined as death within one hour in a person without any other illness or explanation. Coronary heart disease death (nonsudden) was defined as death in a person having an acute episode of angina, coronary insufficiency, or myocardial infarction who took more than one hour to die. Lipid analyses were performed using the protocols established by the Laboratory for Molecular Disease of the National Heart, Lung, and Blood Institute [4]. Total cholesterol was determined using the Abell-Kendall method (51, and triglycerides were determined by the Kessler-Lederer method [6]. Pure primary standards were made from cholesterol and triolein purchased, for the most part, from Sigma Chemical Corporation (St. Louis). Quality control specimens were used in each analytic run. Periodic checks of levels and accuracy were made by performing tests on samples from the Communicable Disease Center. (Atlanta, Georgia). In general, coefficients of variation were under 5 mg/dl for cholesterol and under 9 mgl dl for triglycerides. The evaluation comparing baseline characteristics with the subsequent development of coronary heart disease used a logistic function, parameters of which were estimated by the method of Walter-Duncan [7]. These were standardized by multiplying them by the standard deviation of the characteristic. Such standardized coefficients may be used for an approximate comparison of the strength of the risk factor for coronary heart disease. Likelihood ratios of the logistic functions indicate how well the risk functions fit the actual data.
cholesterol of more than 4.5. Half of the women and more than half of the men who present with treatable hypertenslon already have an abnormal lipid profile. Control of coronary heart disease should be multifactorial. Intervention should include therapy aimed at reducing blood pressure and lowering lipid levels, particularly low-density lipoprotein and very-lowdensity lipoprotein triglycerides. Since many antihypertensive agents have an adverse effect on lipids, considerable thought should be given to those drugs that represent a therapeutic alternative.
In general, today’s physicians respond promptly to counteract the potential dangers of elevated blood pressure, whereas their reaction to elevated cholesterol levels is much slower. Although there is a certain ambiguity concerning the actual levels of blood pressure that require treatment, the consensus is that a diastolic blood pressure of 90 mm Hg and a systolic blood pressure of 140 mm Hg are the accepted cutoff values separating normotensive from hypertensive patients although in some antihypertension trials, a diastolic blood pressure of under 90 mm Hg was the goal of therapy [1,2]. Unfortunately, no generally accepted “cutoff” numbers exist for cholesterol values; traditionally, cholesterol intervention trials have not quoted an actual level of cholesterol as a goal of therapy. Rather, the goal has been simply maintenance of a certain diet or therapy with a specific amount of a cholesterollowering agent. This article reviews the relationship between cholesterol and the subsequent development of coronary heart disease as it was demonstrated in the Framingham Heart Study. We consider not only total cholesterol but also low-density lipoprotein cholesterol, very-low-density lipoprotein cholesterols or triglycerides-a better measure of very-low-density lipoproteins-and high-density lipoprotein cholesterol. Not considered herein is the cholesterol found in chylomicrons or in intermediate-density lipoproteins that is usually negligible in most persons in freeliving populations who have fasted for 12 to 14 hours. We will try to define which levels and which types of cholesterol lead to coronary heart disease so that we can establish some easily recognizable parameters for treatment. Another objective of this article is to explore the interaction of cholesterol levels and hypertensive levels that need to be treated concomitantly.
RESULTS One of the first findings of the Framingham Heart Study was the incidence of coronary heart disease in this population. Figure 1 shows the rate of coronary heart disease in the first 14 years. The disease is more prevalent in men than in women and increases with age in both sexes. In practical terms, this graph shows that some form of coronary heart disease developed in every eighth man aged 40 to 44 years in the first 14 years of this study. Rates for a similar period of time were every sixth man aged 45 to 49 years, every fifth man aged 50 to 54 years, and every
SUBJECTS AND METHODS At the start of the Framingham study in 1949, a random sample of 50 percent of the population identified 5,127 men and women who had no evidence of cardiovascular disease (31. This group has been followed by medical examinations at two-year intervals for almost 36 years. Data in this report are derived from various stages of follow-up, since some of the
24
February
14,1986
The American
Journal
of Medlclne
Volume
80
(suppl 2A)
SYMPOSIUM
fourth man aged 55 years or more. In young women, coronary heart disease developed at a much lower rate than it did in men. Of the 1,600 premenopausal women, coronary heart disease developed in only six prior to menopause. In postmenopausal women, however, coronary heart disease developed at the same rate as in men who were 10 to 15 years younger. The most negative finding is that coronary heart disease had developed in approximately every fifth man and every 17th woman by age 60. Figure 2 shows the relationship of low-density lipoprotein and very-low-density lipoprotein to morbidity. At the time of participant entry into the Framingham Heart Study, these factors were measured in an analytic ultracentrifuge, courtesy of John Gofman and associates [8], at the Donner Laboratories of the University of California in Berkeley. The higher the low-density lipoprotein or verylow-density lipoprotein level, the greater the subsequent
ON INITIAL ANTIHYPERTENSIVE
Figure 2. Rate of coronary heart by lipoprotein levels at time (Framingham Heart Study, follow-up). LDL = low-density tein; VLDL = very-low-density tein.
and ANDERSON
300M *--*
Men Women
200-
Rate per l.ooO loo
-
-0
_ 7.
30-34
35-39
-
I 40-44
I I 45-49 50-54 Age at entry
I 55-59
. .
.
I 60-62 .,.
FourTeen-year mclaence of coronary nean a/sease (all clinical manifestations) according to age and sex: men and women aged 30 to 62 years at entry (Framingham Heart Study).
hgure
Lipoprotein Sf O-20 ma WL)
Morbidity ratio
THERAPY-CASTELLI
Lipoprotein Sf 20400 Pre-8&a (VLDL)
I
I
100
disease of entry 14-year lipoprolipopro-
HDL cholesterol
Men
Low:
40
Triglycerides 44
Medium: 40-50 Hiah: Xl
94-144 >144
100 CHD cases/ 1,ooo
low med high Triglycerides Figure 3. Incidence of coronary heart disease (CHD) by levels of high-density lipoprotein (HDL) cholesterol and Wiglycerides in men. Adapted from [9].
*Cell with 45
February
persons
14, 1999
HDC c&sterol at risk
The American
Journal
low med high Triglycerides Medium HDL cholesterol
of fdedlcine
Volume
low med high Triglycerides Hi h HDL ch ttee ted
99
(suppl 2A)
25
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
TABLE I
THERAPY-CASTELLI
and ANDERSON
Univariate and Multivariate Logistic Regression Coefficients for Coronary Heart Disease and Lipoproteins in Men and Women Aged 50 to 80 Years (Framingham Heart Study) Univariate
Lipopmteins
Men
High-density lipoprotein cholesterol Low-density lipoprotein cholesterol VetjAow-density lipoprotein cholesterol (triglycerides)
-0.506
Multivariate
Women -0.682’
Men
Women
-0.610
-0.650+
0.241’
0.266
0.333*
0.280’
0.032
0.325’
0.036
0.227*
‘p CO.05. +p
rate of coronary heart disease. These analyses were repeated using the Beta Quantification Scheme developed by the Laboratory for Molecular Diseases of the National Heart, Lung, and Blood Institute [4]. Results show (Table I) that in those aged over 50 years, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol are powerful and independent predictors of cqronary heart disease. Triglycerides were not strong predictors in men using the traditional multivariate analysis, but they were in women. Actually, there is probably a “symptom complex” in both sexes involving triglycerides that is strongly related to coronary heart disease. It is characterized by a normal level of cholesterol, a high level of triglycerides, and a low level of high-density lipoprotein. Figure 3 illustrates this special relationship in men [9]. At medium to high levels of high-density lipoprotein, there is no relationship of triglycerides to risk; such a relationship is seen only at very low
CHD rate/l ,000 5025 c25
25-34
4544
55-64
HbL cholesterol
3544
(mg/dL)
6574
75+
Figure 4. Incidence of coronary heart disease (CHD) by high-density lipoprotein cholesterol (HDL-C) level in men and women aged 50 to 80 years (Framingham Heart Study).
26
February
14,1966
The American
Journal
of Medicine
Volume
levels of high-density lipoprotein. Not only do persons with this complex have a higher incidence of coronary heart disease, they also have an incidence of diabetes mellitus that is twofold the rate in the general population and are usually overweight. Whether a high triglyceride level or obesity are truly independent risk factors in multivariate analysis in this setting may be academic. Clinicians should be aware of the constellation of findings: overweight, high triglycerides, average total cholesterol and low-density lipoprotein cholesterol, and low high-density lipoprotein all most likely adding to the risk of diabetes and coronary heart disease development. High-density lipoprotein cholesterol was found to be inversely related to the subsequent development of coronary heart disease, as shown in Figure 4. This relationship is independent on multivariate analysis, as was seen previously in Table I. Table II is a likelihood ratio study that not only shows the strength of high-density lipoprotein as a powerful predictor of coronary heart disease risk, but also demonstrates that the ratio of total cholesterol to high-density lipoprotein cholesterol is an extremely potent predictor by itself. Although the total cholesterol level was an important and independent predictor of coronary heart disease in those aged under 50 years, it no longer predicts the risk (Figure 5) in those aged over 50 years. Indeed, the relationship of the total cholesterol level to coronary heart disease risk is best seen in Figure 6, in which the solid line is the distribution of the cholesterol levels of all the men in Framingham in whom coronary heart disease did not develop in the first 16 years of our study. The non-solid line is the distribution of the cholesterol levels of all the men in whom coronary heart disease developed in the first 16 years. Many clinicians, particularly those with an expertise in lipids, are generally concerned about the total cholesterol level if it is 300 mg/dl or higher. In all likelihood, such patients have monogenic or polygenic hypercholesteroiemia and coronary heart disease will develop early in life. Often, many of these patients’ first-degree blood relatives will be similarly affected. Although this is an important population to identify and treat, those persons with cholesterol levels between 150 and 300 mg/dl should also be considered. About three quarters of the myocardial infarctions in our society occur in people whose cholesterol levels are between 150 and 300 mg/dl. Some experts suggest that 250 mg/di be used as the cutoff number between normal and abnormal values. Anyone with a cholesterol level over 250 mg/dl should be considered to have a “cholesterol problem.” The average man in Framingham who had a myocardial infarction had a cholesterol level of 244 mg/dl, and coronary heart disease developed in more persons with that level than with any other. Is there a cholesterol level threshold for coronary heart disease? is there a value below which myocardiai infarctions do not occur, regardless of the presence of other risk 60
(suppl2A)
JM ON INITIAL ANTIHYPERTENSIVE
TABLE II
factors? In Framingham, in 30 years, we have yet to report a myocardial infarction in any person whose cholesterol was 150 mg/dl or below. Of course, a simple inspection of Figure 6 will reveal that there is no one in Framingham with a cholesterol level under 150 mg/dl. However, three quarters of the people in the world will not have a myocardial infarction in their lifetime. These people live in Asia and Africa, and in South America outside of big cities; they are poor, rarely eat meat, and their cholesterol values are all near the 150 mg/dl level. In our own society, half of the myocardial infarctions occur in people with cholesterol levels below 250 mg/dl, a level considered “normal” by many physicians. Recently, the National Institutes of Health Consensus Development Conference on Lipid Lowering [lo] recommended that all Americans reduce their cholesterol levels to 200 mg/dl or below. Many physicians have argued that most of their patients with cholesterol levels between 200 and 250 mg/dl will not have a myocardial infarction. Although this may be valid, it is also true that half of the myocardial infarctions they see in their practices will occur in these same patients. Obviously, the total cholesterol value cannot accurately predict which patients have a lipid problem when the cholesterol levels are between 200 and
I
30-49 Years
THERAPY-CASTELLI
and ANDERSON
Likelihood Ratios for Various Lipid Profiles of Coronary Heart Disease (Framingham Heart Study, examination 11) lipid pmfile
High-density lipoprotein Low-density lipoprotein Triglyceride Total cholesterol High-density lipoprotein cholesterol Low-density lipoprotein triglyceride High-density lipoprotein triglyceride High-density lipoprotein lipoprotein cholesterol, High-density lipoprotein lipoprotein cholesterol High-density lipoprotein cholesterol, low-density cholesterol
Men
cholesterol cholesterol
Women
14.63’ 4.39t 0.51 1.98 17.11*
21.21* 4.53t 9.5P 2.26 20.41’
and total cholesterol,
8.28’
19.69’
and total cholesterol,
19.19
24.21’
and lowdenslty triglyceride and low-density
18.96
24.73
18.66
23.70’
cholesterol/total lipoprotein
17.16
26.77’
cholesterol/total
Note: This table estimates the relative power of a test or set of tests to predict coronary heart disease from asymptomatic subjects. The higher the number, the better the ability of the test(s) to predict.
ip
I
?
Morbidity ratio 100
----
Men Women
219
239
2s
279
299
c1So1So-20@220-240-260-2Scb300+ 190
219
239
259
279
299
Level of serum cholesterol at initial exam (mg/dL)
fgure 5. Risk of coronary heart disease by cholesterol ham Heart Study, 14-year follow-up).
level at time of entry in men and woman aged 30 to 62 years (Framing-
February 14,1986
Tbe Am&can
Journal of Medicine
Volume 80 (auppI 2A)
27
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
40-
-
and ANDERSON
THERAPY-CASTELLI
Non-CHD
No.
Mean
SD
1,370
219
41
193
244
51
30-
Figure 6. Distribution of total cholesterol levels in men in whom coronary heart disease developed and in men in whom there was no evidence of coronary heart disease (Framingham Heart Study, 1b-year follow-up).
0 150
200
250
300
350
Level of serum cholesterol at initial exarninatbn (mg/dL)
250 mg/dl, or even between 150 and 250 mg/dl. It is in these situations that the high-density lipoprotein cholesterol level becomes an important factor. As Table III shows, for every 5 mg/dl that the high-density lipoprotein cholesterol level decreases below 45 mg/dl, the risk of coronary heart disease increases by 25 percent. Of course, the reverse is also true: for every 5 mg/dl that high-density lipoprotein cholesterol increases above 45 mg/dl, risk drops by about 12.5 percent. We believe that the best way’to predict risk is to calculate a ratio of the total cholesterol to high-density lipoprotein cholesterol (Figure 7), recognizing that the higher the number, the greater the risk [ll]. We recommend keeping the ratio below 4.5. The distribution of our cases ranged from approximately 3.5 and above. However, if we recommended
TABLE Ill
treating everyone with a ratio of 3.5 or higher, we would be treating approximately 90 percent of all Americans. Instead, we have chosen 4.5 as a first-step goal because we already have a subgroup of women under observation in Framingham in whom coronary heart disease developed and whose average ratio was 4.6; this is also a more manageable group. Half of the women and two thirds of the men seen by physicians in America have ratios above this value; more than half of the women with hypertension and more than two thirds of the men with hypertension also have ratios above this value. In Framingham, we found that blood pressure increases with age, and by the time our population reaches their late 50s and early 6Os, almost a third of the men and women have hypertension (Figure 8). Blood pressure (to which we have assigned arbitrary cutoffs) is related to cardiovascular disease in a linear fashion. As seen in Figure 9, hypertensive patients have a twofold to threefold greater incidence of coronary heart disease and a sevenfold greater incidence of stroke; 75 percent of the congestive heart failure occurred in people with uncontrolled hypertension. Even borderline hypertension (diastolic blood pressure of 90 to 94 mm Hg; systolic blood pressure of 140 to 159 mm Hg) carries higher risks. We found a 50 percent increase in myocardial infarction rates and a 300 percent increase in the rate of stroke in borderline hypertensive patients compared with normotensive subjects. Elevated blood pressure is not tolerated well with age, and the relative and absolute impact of blood pressure on cardiovascular diseases worsens as age increases. There is not a strong correlation between blood pressure and serum cholesterol levels. As Table IV shows, the strongest correlates approach 0.20; blood pressure increases as cholesterol increases, but not dramatically. Nonetheless, it should be remembered that more than half of the patients with hypertension, or those requiring treat-
Multlpller for High-Demslty Llpoproteln Cholesterol Level
High-Density Lipopmteln Cholesterol (WW 30 35 40 45 50 55 60 65 70 75
Men
Women
1.82 1.49 1.22 1 .oo 0.62 0.67 0.55 0.45 -“Longevity
1.94 1.55 1.25 1.00 0.60 0.64 0.52 syndrome”-
Note: This table shows how risk increases or decreases in a relative sense in men and women. Since men have an average high-density lipoprotein cholesterol level of 45 mg/dl, no change in a risk profile occurs for men with this value. However, as highdensity lipoprotein cholesterol levels decline, one would multiply the risk profile value (e.g., calculated from Table V) by its appropriate number.
26
February
14,1966
The American
Journal
of hkdclne
Volume
80 (suppl
PA)
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
Ratio is determinsd by dividing totatchotesterolinbbodbyths amount of beneficial HDL Half the Average Risk of Heart Disease Boston Marathon Runners (Physicians) Average Disease
3.4
m
Women
Risk of Heart in Americans
Average Victim of Heart Disease
Goal 4.5
Twice Average Risk Triple Average
Figure 7. high-density
Ratio of total cholesterol lipoprotein
9.6
Risk
to
.4 0
4
a
12
16
20
24
cholesterol.
I
I
I
I
I
I
30
35
40
45
50
55
Figure 8. Prevalence of hypettension (Framingham Heart Study).
1 65
Age at entry
0
Normotensive
a
q
Borderline
(< 140/90)
300
kerage Annual Incidence per 10,000
Coronary heart disease
Intermittent
(>160/95) claudication
100
100 lverage Annual Incidence per 10,000
Hypertensive
200
0
Figure 9. Age-adjusted risk of cardiovascular morbidity according to hypertensive status at each biennial examination in men and women aged 45 to 74 years (Framingham Heart Study cohort, 18-year follow-up).
I 60
Wcirnen Men Congestive failure
Women Men Atherothrombotic brain interaction
a7
50
0
Women Men Women Men All trends statistically significant at P < 0.01
February
14, 1996
The American
Journal
of Medicine
Volume
60 (suppl
2A)
29
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
TABLE IV
THERAPY-CASTELLI
and ANDERSON
Correlation of Serum Total Cholesterol with Systolic Blood Pressure in Men and Women (Framingham Heart Study) CormlationCoelficieni Ap&?
Men Systolic Diastolic Women Systolic Diastolic
30-34
35-39
40-44
45-49
SO-54
55-59
60-M
blood pressure blood pressure blood pressure blood pressure
20 260* 240*
15 4-YW
incidence per 100
200*
10
5
I
I
I
I
I
I
1
I
I
Systolic 100 110 120 130 140 150 160 170 180 190 200 Diastolic 54 62 70 78 86 94 102 110 118 126 134 Blood pressure (mm Hg)
ment of hypertension, also have a lipid abnormality. Furthermore, the more the pressure rises, the greater the likelihood of a lipid problem. This is most easily discerned by calculating the simple ratio of total cholesterol to highdensity lipoprotein cholesterol, and, if the number is 4.5 or higher, treatment for lipid lowering should be initiated. Two of the most important risk fqctors for coronary heart disease are elevated levels of cholesterol and blood pressure. As Figure 10 shows, there is a powerful interaction between these two factors to produce coronary heart disease. Treating one factor is doing only half of the job, but if the therapy for decreasing blood pressure increases the cholesterol level as well, the beneficial effects of antihypertensive therapy will be offset or negated by the adverse lipid effects on coronary heart disease. Careful monitoring of cholesterol levels and blood pressure is necessary to assure that good control of both is achieved. 30
February 14,1986
The Amerlcan Journal of Madicina
Figure 10. Four-year incidence of coronary heart disease according to systolic blood pressure at specified levels of serum of men aged 55 years (Framingham Heart Study).
COMMENTS Coronary heart disease is a multifactorial problem, and there are numerous handbooks available for calculating exact individual patient risk. One method is shown in Table V. In clinical practice, however, it is usually more effective to approach each problem separately and establish specific therapeutic goals (Table VI) for each. One of the dilemmas of lipid management is that there is no established goal of therapy similar to that used in blood pressure management. For example, regardless of the initial diastolic blood pressure or systolic blood pressure that prompted the physician to begin therapy, the goal of that therapy will probably be to reduce blood pressure to 140/90 mm Hg. We believe that patients’ lipid profiles should also be treated with a specific goal in mind: a ratio of total cholesterol to high-density lipoprotein cholesterol of less than 4.5. Similar strategies should be used to conVolume 60 (suppi 2A)
SYMPOSIUM
TABLE
V
ON INITIAL ANTIHYPERTENSIVE
Probability (per 1,000) of Cardiovascular Disease Developing Characteristics (Framingham Heart Study, la-year follow-up
in Eight Years According in 50-year-old men)
Do not smokecigarettes SBP LVH-ECG
Glucose intolerance absent
Glucose intolerance present
LVH-ECG
Glucose intolerance absent
Glucose intolerance present
negative Cholesterol 185 210 235 260 265 310 335 Cholesterol 185 210 235 260 285 310 335 posttlve Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335
THERAPY-CASTELLI
and ANDERSON
to Specified
Smokecigarettes
105
120
135
150
165
180
195
37 44 53 83 76 91 108
46 55 66 79 94 112 133
57 69 82 98 116 138 162
71 85 102 121 143 168 197
89 106 125 148 174 204 237
110 130 153 180 211 244 282
135 159 187 218 252 290 332
65 78 93 110 130 154 181
81 96 114 135 160 187 218
100 119 140
165 194 226 261
123 146 171 200 233 269 309
151 177 207 241 278 318 361
184 214 249 286 327 371 417
222 257 295 337 381 428 475
98
117 138 163 191 222 257
121 143 169 197 230 266 305
148 175 204 237 274 314 357
181 211 245 282 323 367 412
218 253 291 332 377 423 470
261 300 342 387 433 481 529
309 352 397 444 492 540 587
166 194 226 262 301 343 387
201 234 270 310 352 398 445
241 279 319 362 408 455 503
287 328 372 418 466 514 562
338 382 429 476 524 572 619
392 439 487 535 583 629 672
449 498 546 593 638 682 722
SBP
Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310 335 Cholesterol 185 210 235 260 285 310
105
120
135
150
165
180
195
62 75 89 106 126 148 174
78 92 110 130 154 181 211
96 114 135 159 187 218 253
119 140 165 194 225 261 300
146 171 200 233 269 309 351
177 207 241 278 318 361 407
214 249 286 327 371 417 465
108 128 151 178 208 241 278
133 157 184 215 249 287 328
163 191 222 257 296 337 382
197 230 265 305 347 392 439
237 274 314 357 402 449 497
282 323 367 412 460 508 556
332 376 423 470 519 566 613
160 188 219 253 292 333 377
194 226 262 300 343 387 434
234 270 309 352 397 444 492
278 319 362 408 455 503 551
328 372 418 465 514 561 608
382 428 476 524 572 618 863
439 487 535 582 628 672 713
258 296 338 383 429 477
305 348 393 440 487 536 583
358 403 450 498 546 593 639
413 461 509 557 604 649 691
471 519 567 614 658 700 739
530 578 624 668 709 747 782
588 634 877 718 755 789 819
‘Framingham men aged 50 years have an average systolic blood pressure of 133 mm Hg and an average serum cholesterol Sixty-two percent smoke cigarettes, 1.5 percent have definite lefl ventricular hypertrophy by electrocardiography, and 6.5 percent intolerance. At these average values, the probability of cardiovascular disease developing in eight years is 115/l ,000. SBP = systolic blood pressure; LVH = left ventricular hypertrophy; ECG = electrocardiography.
TABLE
troi blood sugar, weight, smoking, uric acid, lack of physical activity, and possibly stress. Since many hypertensive therapies [12,13] adversely interfere with the lipid profile by lowering high-density iipoprotein, raising triglycerides, and to a lesser extent, raising the total cholesterol level, some investigators [14] believe that such changes may explain why the decline in coronary heart disease incidence was so small in most of the hypertension intervention trials. Two alternatives seem obvious: we must make optimal use of antihypertensive agents that do not adversely affect lipids or, in those cases in which patients are receiving drugs that do disrupt the lipid profile, we should apply lipid-lowering therapies to counteract such effects. We hope this will help to achieve an optimal blood pressure level and have a favorable impact on all the other risk factors as well. February
14,19BB
Vi
Cutoff Values as Goals of Therapy for an Optimal Risk Profile (Framingham Heart Study)
Risk Factor
Level
Total/high-density lipoprotein cholesterol Blood pressure Cigarette smoking Cigar and pipe smoking Diabetes mellitus Weight Electrocardiographic finding Physical activity Stress
The
American
of 236 mg/dl. have glucose
Journal
4.5 or less (unless total cholesterol < 150 mg/dl) Under 140190 mm Hg Must stop Must stop only if inhaled Strict control Keep under body mass index of 22 Any abnormality except first-degree A-V block or rare ectopic beat Regular, 3 to 4 days/week Relax
of Medicine
Volume
80
(suppl 2A)
31
SYMPOSIUM
ON INITIAL ANTIHYPERTENSIVE
THERAPY-CASTELLI
and ANDERSON
REFERENCES 1.
2.
3.
4.
5.
8.
7.
32
Multiple Risk Factor Intervention Trial Research Group: Multiple Risk Factor Intervention Trial. Risk factor changes and mortality results. JAMA 1982; 248: 1465-1477. Hypertension Detection and Follow-Up Program Cooperative Group: I. Reduction in mortality of persons with high blood pressure, including mild hypertension. JAMA 1979; 242: 2582-2571. Dawber TR, Meadors GF, Moore FE: Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health 1951; 41: 279-288. Fredrickson DS, Levy RI, Lees RS: Fat transport and lipoproteins-an integrated approach to mechanisms and disorders. N Engl J Med.1967; 276: 32,94,148,273. Abell LL. Lew BB. Brodie BB. et al: A simolified method for the estimation of total cholesterol in serum and demonstration of its specificity. J Biol Chem 1952; 195: 357-365. Kessler B, Lederer H: Fluorometric measurements of triglycerides. In: Skeggs LT Jr, ed. Technicon symposia automation in analytical chemistry. New York: Medicad Inc., 1965; 341. Walker SH, Duncan DB: Estimation of the probability of an event as a function of several independent variables. Biometrika 1967; 54: 167.
February
14,1936
The American
Journal
of Medlclne
8. 9.
10.
11.
12.
13.
14.
Volume
Gofman JW, Young W, Tandy R: lschemic heart disease, atherosclerosis, and longevity. Circulation 1966; 34: 697. Gordon T, Castelli WP, Hjortland MC, et al: High-density lipoprotein as a protective factor against coronary heart disease. The Framingham Study. Am J Med 1977; 82: 707-714.’ NIH Consensus Panel on Blood Cholesterol: Lowering blood cholesterol to prevent heart disease. JAMA 1985; 253: 20802086. Brody J: New York Times, after Castelli WP: Categorical issues in therapy for coronary heart disease. Cardiol Pratt 1985; January/February: 267-273. Rouffy J, Jaillard J: Effects of two antihypertensive agents on lipids, lipoproteins, and apoproteins A and B. Comparison of prazosin and atenolol. Am J Med 1986; 80 (suppl 2A): lOO103. Grimm RH, Leon AS, Hunninghake DB, et al: Effects of thiazide diuretics on plasma lipids and lipoproteins in mildly hypertensive patients. Ann Intern Med 1981; 94: 7-11. Leren P, Helgeland A, Hjermann I, Holme I: The Oslo study: CHD risk factors, socioeconomic influences, and intervention. Am Heart J 1983; 106: 1200-1206.
80 (suppl
PA)