C Reactive protein and its relation to cardiovascular risk factors: a population based cross sectional studyBMJ 1996; 312 doi: http://dx.doi.org/10.1136/bmj.312.7038.1061 (Published 27 April 1996) Cite this as: BMJ 1996;312:1061
- M A Mendall, senior lecturera,
- Praful Patel, research fellowa,
- Lydia Ballam, research assistanta,
- D Strachan, senior lecturerb,
- T C Northfield, professora
- a Division of Biochemical Medicine, St George's Medical School, London SW17 0RE
- b Public Health Sciences, St George's Medical School, London SW17 0RE
- Correspondence to: Dr Mendall.
- Accepted 13 February 1996
Objective: To test the hypothesis that minor chronic insults such as smoking, chronic bronchitis, and two persistent bacterial infections may be associated with increases in C reactive protein concentration within the normal range and that variations in the C reactive protein concentration in turn may be associated with levels of cardiovascular risk factors and chronic coronary heart disease.
Design: Population based cross sectional study.
Setting: General practices in Merton, Sutton, and Wandsworth.
Subjects: A random sample of 388 men aged 50-69 years from general practice registers. 612 men were invited to attend and 413 attended, of whom 25 non-white men were excluded. The first 303 of the remaining 388 men had full risk factor profiles determined.
Interventions: Measurements of serum C reactive protein concentrations by in house enzyme linked immunosorbent assay (ELISA); other determinations by standard methods. Coronary heart disease was sought by the Rose angina questionnaire and Minnesota coded electrocardiograms.
Main outcome measures: Serum C reactive protein concentrations, cardiovascular risk factor levels, and the presence of coronary heart disease.
Results: Increasing age, smoking, symptoms of chronic bronchitis, Helicobacter pylori and Chlamydia pneumoniae infections, and body mass index were all associated with raised concentrations of C reactive protein. C Reactive protein concentration was associated with raised serum fibrinogen, sialic acid, total cholesterol, triglyceride, glucose, and apolipoprotein B values. C Reactive protein concentration was negatively associated with high density lipoprotein cholesterol concentration. There was a weaker positive relation with low density lipoprotein cholesterol concentration and no relation with apolipoprotein A I value. C Reactive protein concentration was also strongly associated with coronary heart disease.
Conclusion: The body's response to inflammation may play an important part in influencing the progression of atherosclerosis. The association of C reactive protein concentration with coronary heart disease needs testing in prospective studies.
Factors that determine levels of inflammatory mediators in the normal general population have not been explored, nor has their relation to cardio- vascular risk factors
Among 50-69 year old men many environmental and lifestyle risk factors for cardiovascular disease are associated with raised serum concentrations of C reactive protein
Circulating concentrations of lipids, glucose, and clotting factors are also associated with serum C reactive protein concentrations
The body's response to inflammation may influence the development of atherosclerosis
The acute phase response is part of the body's reaction to injury or infection. It is associated with changes in lipid and glucose metabolism. Concentrations of high density lipoprotein cholesterol consistently fall1 and glucose and triglyceride concentrations rise.2 3 4 Inconsistent changes in total cholesterol and apolipoprotein B values have been observed, possibly due to the timing of sampling and the different pathological conditions studied.2 4 5 6 Apolipoprotein A I concentration does not change, and low density lipoprotein cholesterol concentration falls a little. The cellular and rheological properties of blood change, with an increase in white cell and platelet counts. There is an increase in synthesis of proteins such as fibrinogen by the liver which increase coagulability and the viscosity of the blood. All these changes have been shown to be associated with cardiovascular disease in prospective studies. Sialic acid is found in association with many acute phase proteins and in one prospective study was a powerful predictor of coronary heart disease.7
C Reactive protein is the major acute phase protein in humans. In unchallenged subjects concentrations are usually low, rising several hundredfold in acute illness.8 The causes of variation in C reactive protein concentrations in otherwise normal people have received little attention. Raised concentrations of C reactive protein have been associated with smoking9 and aging.10 By using a comparatively insensitive nephelometric method we have shown that two chronic bacterial infections—Helicobacter pylori (a persistent cause of gastric inflammation) and Chlamydia pneumoniae (a respiratory pathogen)—are associated with raised concentrations of C reactive protein and raised levels of inflammatory mediators within conventional normal ranges.11 Other chronic exposures which could be important include periodontal disease and chronic bronchial inflammation. All these exposures have been linked to coronary heart disease.12 13 14
The relation between cardiovascular risk factors and serum C reactive protein concentrations within conventional reference ranges in otherwise normal people has also received little attention. Associations between C reactive protein concentration and serum fibrinogen concentration have been observed in normal elderly people15 and between C reactive protein concentration and fasting serum insulin concentration in patients with chronic coronary heart disease.16 These findings suggest that low levels of inflammatory activity may produce qualitatively similar effects to those seen during acute illness or injury.
In patients with unstable angina and in chronic coronary heart disease C reactive protein concentration may be a powerful predictor of subsequent cardiac events.17 18 It is unknown, however, whether C reactive protein is a risk factor for chronic coronary heart disease in comparison with general population controls, as available tests have not been sensitive enough to detect low serum concentrations reliably.
We tested the hypothesis that chronic exposures causing low grade inflammation are associated with variations in C reactive protein concentration within the normal range in the general population and that these variations may be associated with differences in levels of various cardiovascular risk factors and the presence of coronary heart disease in middle aged men.
Subjects and methods
We recruited a random sample of men aged 50-69 years from the registers of general practices in the Merton, Sutton, and Wandsworth District Health Authority area, south London. A total of 612 men were invited and 413 (67%) attended. Of these, 25 were non-white and were excluded. Information was obtained on history and symptoms of coronary heart disease, lifestyle, and socioeconomic circumstances, as described previously. Cardiovascular risk factor profiles and serological tests for H pylori and C pneumoniae were also performed as described.11 Electrocardiograms were Minnesota coded. We took tracings to indicate coronary heart disease if they showed any of the following: Q waves, ST segment depression, left bundle branch block, or T wave inversion. Only the 303 men who had complete cardiovascular risk factor profiles were included in the study.
C Reactive protein concentration was measured by in house enzyme linked immunosorbent assay (ELISA). Rabbit antihuman C reactive protein (Dako) was used to coat the plates at a concentration of 10 mg/l. Test serum was then added at dilutions of between 1 in 100 and 1 in 1000. C Reactive protein was detected with peroxidase conjugated rabbit antihuman C reactive protein at a dilution of 1 in 6000 (Dako) and a standard detection method used. The assay was standardised against international standard 85/506 C reactive protein. The standard curve was linear up to 5 µg/l and logarithmic thereafter. The interassay and intra-assay coefficients of variation were 14% and 8%. The proportional recovery derived from spiking serum samples with standards was 94% (range 85-104%).
The distribution of C reactive protein concentrations was positively skewed and hence C reactive protein concentration was log transformed for all analyses. Log C reactive protein concentration was analysed as the outcome variable in relation to age, smoking, chronic infection, and body mass index and as an explanatory variable in relation to cardiovascular risk factors and prevalent cardiovascular disease.
The relation of log C reactive protein concentration to age, smoking, chronic infection, and body mass index was analysed by multiple regression in Stata (Stata Corporation, United States). All determinants were controlled for each other and for the subject's social class (registrar general's classification I, II, IIINM (non-manual), IIIM (manual), IV, V) and father's social class (I, II, IIINM, IIIM, IV, V, unclassified).
Regression models for the association of log C reactive protein with cardiovascular risk factors included as explanatory variables age as a continuous variable; smoking (never smoker, former smoker, current smoker); pack years of smoking; current daily cigarette consumption; subject's social class; and father's social class. High density lipoprotein cholesterol, triglyceride, glucose, and apolipoprotein A I and B concentrations; white cell count; ratio of total cholesterol to high density lipoprotein cholesterol; and ratio of apolipoprotein B to apolipoprotein A I were log transformed as they had a positively skewed distribution. The relation between C reactive protein concentration and cardiovascular diseases was analysed by logistic regression in Stata.
Figure 1 shows the distribution of C reactive protein concentrations in our population. The median value was 1.72 mg/l (interquartile range 0.69-3.95 mg/l). There was an approximate doubling in C reactive protein concentration between adjacent fifths of the distribution.
Relation of environmental factors and inflammation to serum C reactive protein concentration—Table 1 shows the prevalence of potential determinants of C reactive protein concentration by fifths of the distribution. Higher concentrations were associated with increasing age, smoking, history of chronic bronchitis, H pylori seropositivity, C pneumoniae seropositivity, and body mass index. Table 2 shows the effect of adjusting these factors for each other and for father's occupation and subject's current occupation. A strong independent relation was found with body mass index. Other variables independently associated with C reactive protein concentration were age and symptoms of chronic bronchitis, seropositivity to H pylori and C pneumoniae being rendered of borderline statistical significance. The most influential aspect of smoking related to pack years.
Relation of serum C reactive protein concentration to cardiovascular risk factors—Table 3 shows the relation between cardiovascular risk factors and C reactive protein concentration by fifths of the distribution. There were graded effects on most risk factors in a direction to increase risk except with low density lipoprotein cholesterol and apolipoprotein A I concentrations and diastolic blood pressure. After adjustment for age, social class, father's social class, smoking, and body mass index these relations persisted. Exceptions were the relation with blood pressure, which disappeared, and a relation with low density lipoprotein cholesterol, which appeared.
Relation between C reactive protein concentration and electrocardiographic abnormalities, symptomatic heart disease, and claudication—Forty one men had an abnormal electrocardiogram, 27 had a positive Rose angina questionnaire result or history of myocardial infarction alone, and 24 had symptoms of intermittent claudication. After adjustment for age, smoking, current and childhood social class, and body mass index the odds ratio per doubling of C reactive protein concentration was 1.36 (95% confidence interval 1.08 to 1.72) for electrocardiographic abnormalities, 1.42 (1.13 to 1.78) for subjects with a positive Rose angina questionnaire result or history of myocardial infarction (irrespective of electrocardiographic findings), 1.55 (1.25 to 1.92) for subjects with either (that is, all prevalent heart disease), and 1.83 (1.29 to 2.58) for subjects with claudication.
So far as we know this is the first study to examine in detail the potential determinants of serum C reactive protein concentrations within the conventional reference range and to examine the relation of these concentrations to risk factors for cardiovascular disease. This is also the first population based study to show a relation between C reactive protein concentration and electrocardiographic abnormalities indicative of past or present coronary heart disease and symptoms of angina and claudication.
Serum C reactive protein concentration within conventional reference ranges is probably a reflection of the general level of inflammatory activity within the body. Three exposures which were related to the concentration of C reactive protein—namely, persistent phlegm, H pylori infection, and C pneumoniae infection—are associated with inflammation. The mechanism whereby smoking is related to C reactive protein concentration is unclear, but it may in part be mediated through bronchial injury or inflammation.
The production of C reactive protein is regulated by cytokines, principally interleukin 6,19 whose effects are modified by other cytokines and growth factors20 as well as by hormones such as cortisol and insulin.21 Production of cytokines and other stress hormones may be altered in conditions other than inflammation or injury. Adipocytes from obese humans have been shown to overproduce tumour necrosis factor (alpha) messenger RNA.22 Tumour necrosis factor (alpha) is a potent inducer of interleukin 6 production by various cells. This may explain why a high body mass index was associated with increased serum concentrations of C reactive protein.
Throughout the range of C reactive protein concentrations in this general population sample there was a strong graded relation with white cell count and total cholesterol, high density lipoprotein cholesterol, triglyceride, glucose, apolipoprotein B, fibrinogen, and sialic acid concentrations. The ratio of total cholesterol to high density lipoprotein cholesterol and of apolipoprotein B to apolipoprotein A I, which may be a better measure of cardiovascular risk,23 also displayed strong graded relations. There were weak relations with systolic blood pressure and low density lipoprotein cholesterol concentration and none with apolipoprotein A I. Controlling for body mass index weakened the relations with lipid values only modestly and strengthened the relation with low density lipoprotein cholesterol concentration but had no effect on the relation of C reactive protein value with sialic acid and fibrinogen concentrations and white cell count. These relations between C reactive protein concentrations within the normal range and cardiovascular risk factors were qualitatively similar to those seen in better defined acute illness or injury, or in animal models,6 supporting the notion that the acute phase response is a continuum and not an all or none response.
ROLE OF CYTOKINES
The association of C reactive protein concentration with cardiovascular risk factors could be explained by the actions of cytokines and other hormones which alter the concentrations of both. Interleukin 6 can increase hepatic synthesis of clotting factors and can also increase hepatic gluconeogenesis and triglyceride synthesis.24 It also stimulates general haemopoiesis.25 Tumour necrosis factor (alpha) has been strongly implicated in the pathogenesis of insulin resistance.22 26 This underlies syndrome X, which is characterised by alterations in blood glucose and serum lipid concentrations in the same pattern as we observed in association with raised serum C reactive protein concentrations.27 Tumour necrosis factor (alpha) also inhibits lipoprotein lipase activity and stimulates hepatic lipogenesis in rats.24 The decrease in high density lipoprotein cholesterol concentration has been postulated to result from the redistribution of lipids in this fraction to other fractions, with apolipoprotein A I being displaced as the major apolipoprotein by serum amyloid A protein.6 The cause of the rise in apolipoprotein B concentration in our subjects was uncertain.
The strong relation of C reactive protein concentration to claudication and electrocardiographic abnormalities may merely reflect tissue damage resulting from myocardial infarction. Our finding of a strong relation between C reactive protein concentration and chronic coronary heart disease of unspecified onset provides circumstantial evidence that C reactive protein concentration may relate to the underlying pathogenesis. This is supported by a reported relation between C reactive protein concentration and prognosis in patients with chronic stable angina.18
Serum C reactive protein concentration could be related to the pathogenesis of atherosclerosis via the effects of inflammation on conventional risk factors, or the raised C reactive protein concentration may result from inflammation in the arterial wall associated with the atherosclerosis itself. A further possibility is that the cytokine and cellular mediators of the acute phase response originating at a distance to the coronary arteries are directly involved in the pathogenesis of atherosclerosis.
What we need now are prospective studies to evaluate the association of C reactive protein concentrations with subsequent cardiac events and the extension of these observations to other age groups and to women. The relation between short term fluctuations in C reactive protein concentration and cardiovascular risk factors also requires evaluation. Further research could usefully explore the regulation of the acute phase response with respect to cytokines and their role as independent risk factors for coronary heart disease.
Funding British Heart Foundation.
Conflict of interest None.