C-Reactive Protein and the Risk of Developing Hypertension

Context Although it has been hypothesized that hypertension is in part an inflammatory disorder, clinical data linking inflammation with incident hypertension are scarce.

Objective To examine whether C-reactive protein levels, a marker of systemic inflammation, are associated with incident hypertension.

Design, Setting, and Participants A prospective cohort study that began in 1992 of 20 525 female US health professionals aged 45 years or older who provided baseline blood samples with initially normal levels of blood pressure (BP) (systolic BP <140 mm Hg and diastolic BP <90 mm Hg, and no history of hypertension or antihypertensive medications) and then followed up for a median of 7.8 years for the development of incident hypertension. Plasma C-reactive protein levels were measured and baseline coronary risk factors were collected.

Main Outcome Measure Incident hypertension, defined as either a new physician diagnosis, the initiation of antihypertensive treatment, or self-reported systolic BP of at least 140 mm Hg or a diastolic BP of at least 90 mm Hg.

Results During follow-up, 5365 women developed incident hypertension. In crude models, the relative risks (RRs) and 95% confidence intervals (CIs) of developing hypertension from the lowest (referent) to the highest levels of baseline C-reactive protein were 1.00, 1.25 (95% CI, 1.14-1.40), 1.51 (95% CI, 1.35-1.68), 1.90 (95% CI, 1.72-2.11), and 2.50 (95% CI, 2.27-2.75) (linear trend P<.001). In fully adjusted models for coronary risk factors, the RRs and 95% CIs were 1.00, 1.07 (95% CI, 0.95-1.20), 1.17 (95% CI, 1.04-1.31), 1.30 (95% CI, 1.17-1.45), and 1.52 (95% CI, 1.36-1.69) (linear trend P<.001). C-reactive protein was significantly associated with an increased risk of developing hypertension in all prespecified subgroups evaluated, including those with very low levels of baseline BP, as well as those with no traditional coronary risk factors. Similar results were found when treating C-reactive protein as a continuous variable and controlling for baseline BP.

Conclusion C-reactive protein levels are associated with future development of hypertension, which suggests that hypertension is in part an inflammatory disorder.

The C-reactive protein is a marker of systemic inflammation that has been associated with an increased risk of incident myocardial infarction and stroke.^1-5 Inflammation has also been hypothesized to play a role in the development of hypertension, and cross-sectional evidence demonstrates higher C-reactive protein levels among those individuals with elevated blood pressure (BP).^6-11 Higher levels of C-reactive protein may increase BP by reducing nitric oxide production in endothelial cells,^12,13 resulting in vasoconstriction and increased production of endothelin 1.^14,15 C-reactive protein may also function as a proatherosclerotic factor by up-regulating angiotensin type 1 receptor expression.¹⁶

Inflammation has been shown to correlate with endothelial dysfunction¹⁷ and relate to the renin-angiotensin system.¹⁸ As a result, it has been hypothesized that hypertension may be in part an inflammatory disorder. However, clinical data linking inflammation with incident hypertension are scarce.¹⁹ We therefore determined whether elevated C-reactive protein levels in normotensive individuals are associated with an increased risk of developing hypertension.

The Women's Health Study is an ongoing randomized, double-blind, placebo-controlled trial of low-dose aspirin and vitamin E in the primary prevention of cardiovascular disease and cancer, which began in 1992.²⁰ In 1992, a total of 39 876 female US health professionals aged 45 years or older without prior myocardial infarction, stroke, transient ischemic attack, and cancer (except nonmelanoma skin cancer) were enrolled and randomized into the study.

Before randomization, baseline blood samples were collected from 28 345 participants and stored in liquid nitrogen until analysis. Samples were then transferred to a core laboratory facility in which they were assayed for C-reactive protein with a validated high-sensitivity assay (Denka Seiken Co, Tokyo, Japan).⁵ Of the samples received, 27 939 were evaluated and assayed for C-reactive protein. The baseline population was then restricted to 20 525 women without hypertension, defined as having no self-reported past or current history of hypertension, no past or current history of antihypertensive treatment, a systolic BP of less than 140 mm Hg, and a diastolic BP of less than 90 mm Hg at study entry. Baseline BP was reported in 1 of 9 ordinal systolic BP categories ranging from less than 110 mm Hg to at least 180 mm Hg and 7 ordinal diastolic BP categories ranging from less than 65 mm Hg to at least 105 mm Hg. A single measurement of self-reported BP in health professionals has been shown to be highly correlated with measured systolic BP (r = 0.72) and diastolic BP (r = 0.60).²¹

Incident cases of hypertension were defined by meeting at least 1 of 4 criteria: self-reports of a new physician diagnosis on follow-up questionnaires at years 1 and 3, and all annual questionnaires thereafter; self-reports of newly initiated antihypertensive treatment at years 1, 3, and 4; self-reported systolic BP of at least 140 mm Hg; or self-reported diastolic BP of at least 90 mm Hg. Women reporting a new physician diagnosis of hypertension also provided the month and year of diagnosis. A missing date for a physician diagnosis or hypertension defined by another criterion was assigned a date of incident hypertension by randomly selecting a date between the current and previous annual questionnaire. Those individuals developing major concomitant diseases, the management of which may have impact on BP, at or after baseline but before the development of hypertension, including myocardial infarction, stroke, pulmonary embolism, and peripheral vascular disease, were censored at that date of diagnosis and not considered an incident case of hypertension. Based on this definition, 5365 cases of incident hypertension developed during a median follow-up of 7.8 years (range, 0.001-8.82 years).

Women were first compared according to level of C-reactive protein by using mean values or proportions of baseline risk factors to assess potential confounding. Because hormone therapy increases the level of C-reactive protein,^22,23 levels were based on the quintiles of C-reactive protein among the subset of women not taking hormone therapy. We used Cox proportional hazards regression model analyses to compute the relative risks (RRs) and 95% confidence intervals (CIs) of incident hypertension for increasing plasma C-reactive protein levels, with the lowest level as the referent. Models first included only C-reactive protein, then were adjusted for age and randomized treatment assignment. The final multivariable model added body mass index (calculated as weight in kilograms divided by the square of height in meters), smoking status, exercise, alcohol consumption, parental history of myocardial infarction before age 60 years, diabetes, postmenopausal hormone use, and hypercholesterolemia (either any baseline history of cholesterol-lowering medication use or a physician diagnosis of high cholesterol or a self-reported cholesterol of at least 240 mg/dL [≥6.22 mmol/L]). Linear trend tests across levels of plasma C-reactive protein levels were performed by using the median value for each level as an ordinal variable.

Both stratified and joint models incorporating C-reactive protein and either systolic BP or diastolic BP were also considered. Women in the upper half of the prehypertension BP classification according to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7)²⁴ (either systolic BP of 130-139 mm Hg or diastolic BP of 85-89 mm Hg) were excluded from these analyses out of concern for potential misclassification and the vast majority of these women became hypertensive during follow-up. Joint models of C-reactive protein and BP used clinically relevant cut points for C-reactive protein: less than 1, 1 to less than 3, and 3 or more mg/L.²⁵ A test for interaction between C-reactive protein, BP, and the risk of developing hypertension was assessed with C-reactive protein as a continuous variable and BP as an ordinal variable. All analyses were performed with SAS version 8 (SAS Institute Inc, Cary, NC), using a 2-tailed P = .05 for significance.

Baseline characteristics of the study population according to levels of C-reactive protein are shown in Table 1. As expected, traditional coronary risk factors were more prevalent among those participants with elevated levels of C-reactive protein. During a median follow-up of 7.8 years for the 20 525 women comprising the baseline population, 5365 women developed hypertension either on the basis of newly initiated treatment (n = 2295 [43%]) or having a systolic BP of at least 140 mm Hg or diastolic BP of at least 90 mm Hg (n = 3070 [57%]).

Overall, there was a positive association between increasing levels of C-reactive protein and risk of developing hypertension (Table 2). In crude models, the RRs of developing hypertension increased from the lowest (referent) to the highest level of C-reactive protein (linear trend P<.001), such that those participants with levels of C-reactive protein in excess of 3.5 mg/L had an RR of 2.50 (95% CI, 2.27-2.75). In risk factor–adjusted models as well as those that additionally adjusted for baseline BP, risks were attenuated but retained a positive association with risk of developing hypertension (linear trend P<.001). For example, in models additionally adjusting for baseline BP, women with increasing levels of C-reactive protein had RRs of hypertension of 1.04 (95% CI, 0.92-1.17), 1.11 (95% CI, 0.99-1.24), 1.19 (95% CI, 1.07-1.33), and 1.34 (95% CI, 1.20-1.50; linear trend P<.001). Analyses stratified by use or nonuse of hormone therapy showed essentially identical results. In log-transformed models that considered C-reactive protein as a continuous variable, similar strong effects were observed that were all statistically significant (for 1-SD increase of C-reactive protein, adjusted RR, 1.14; 95% CI, 1.11-1.17).

We also considered more restrictive definitions of hypertension to address potential misclassification. The multivariable RRs of hypertension restricted to women initiating treatment, having a physician diagnosis, having a systolic BP of at least 160 mm Hg, or having a diastolic BP of at least 95 mm Hg during follow-up (4982 of 5365 women developing hypertension) for increasing levels of C-reactive protein were 1.00, 1.06 (95% CI, 0.94-1.20), 1.18 (95% CI, 1.05-1.33), 1.30 (95% CI, 1.16-1.46), and 1.52 (95% CI, 1.36-1.70; linear trend P<.001). Alternatively, analyses of hypertension restricted to women initiating treatment or having a physician diagnosis during follow-up (2295 of 5365 women developing hypertension) resulted in RRs for increasing levels of C-reactive protein of 1.00, 1.03 (95% CI, 0.87-1.22), 1.02 (95% CI, 0.86-1.20), 1.14 (95% CI, 0.97-1.33), and 1.41 (95% CI, 1.20-1.65; linear trend P<.001).

We next examined the age-adjusted and multivariable-adjusted associations between C-reactive protein and the risk of developing hypertension stratified by either systolic BP or diastolic BP. In Table 3, increasing levels of C-reactive protein were associated with an increased risk of developing hypertension at all levels of baseline BP in fully-adjusted models, even among women with baseline systolic BP of less than 110 mm Hg and baseline diastolic BP of less than 65 mm Hg. Additional adjustment for diastolic BP in models stratified by systolic BP, or for systolic BP in models stratified by diastolic BP, did not change the RRs from our fully-adjusted models. The consideration of log-transformed C-reactive protein as a continuous variable also generated statistically significant results decreased to a systolic BP of 110 mm Hg and a diastolic BP of 65 mm Hg (for 1-SD increase of C-reactive protein among women with systolic BP 110-119 mm Hg, adjusted RR, 1.12; 95% CI, 1.10-1.18, and for 1-SD increase of C-reactive protein among women with diastolic BP 65-74 mm Hg, adjusted RR, 1.14; 95% CI, 1.08-1.20).

Overall, there were 7065, 6920, and 6540 women with C-reactive protein levels of less than 1, 1 to less than 3, and 3 or more mg/L, respectively, cut points recently set forth in clinical guidelines.²⁵ For these clinical cut points of C-reactive protein, the RRs of developing hypertension in crude analyses were 1.00, 1.46 (95% CI, 1.36-1.57), and 2.11 (95% CI, 1.97-2.26; linear trend P<.001) respectively; in risk factor–adjusted models, the RRs were 1.00, 1.16 (95% CI, 1.08-1.25), and 1.42 (95% CI, 1.31-1.54; linear trend P<.001) respectively. To evaluate the robustness of the findings among women without major coronary risk factors, we limited analyses to women who were nonsmokers, not using hormone therapy, had a body mass index of less than 30, and did not have diabetes and hypercholesterolemia (n = 6795; 1384 incident cases of hypertension). In this subgroup of women, the age-adjusted RRs were 1.00, 1.26 (95% CI, 1.11-1.42), and 1.69 (95% CI, 1.47-1.94; linear trend P<.001) for corresponding C-reactive protein levels of less than 1, 1 to less than 3, and 3 or more mg/L, respectively. In risk factor–adjusted models, the RRs were 1.00, 1.24 (95% CI, 1.10-1.40), and 1.64 (95% CI, 1.42-1.89; linear trend P<.001) respectively.

To further explore the joint effects of C-reactive protein and BP, we considered clinical cut points of C-reactive protein and its association with hypertension in combination with baseline BP (Figure 1). For systolic BP, a significant statistical interaction with C-reactive protein was observed (P<.001). C-reactive protein levels of 1 to 3 and more than 3 mg/L conferred an additional risk of developing hypertension in women in which systolic BP levels were classified as normal (<120 mm Hg) or prehypertension (<130 mm Hg) by the JNC 7 guidelines, with significant RRs compared with women with a systolic BP of less than 110 mm Hg and C-reactive protein level of less than 1 mg/L (all P<.001). For diastolic BP, a similar additive effect was observed, although the interaction did not reach statistical significance (P = .11).

This study provides evidence that baseline levels of C-reactive protein are modestly but independently associated with an increased risk of incident hypertension, even among those with very low initial systolic BP and diastolic BP as classified by the JNC 7.²⁴ This finding for C-reactive protein was independent of baseline levels of systolic BP and diastolic BP. Similar effects were observed among those participants without baseline coronary risk factors and in analyses where C-reactive protein was considered as a continuous variable. These data suggest that inflammation may have a potentially important role in the development of hypertension.

Despite the fact that up to 50 million US individuals are affected,²⁴ the etiology of hypertension often remains unclear. These data from the Women's Health Study represent the first major prospective analysis of the association between C-reactive protein and incident hypertension and are consistent with the findings for other inflammation-sensitive plasma proteins, including fibrinogen, α₁-antitrypsin, haptoglobin, orosomucoid, and ceruloplasmin.¹⁹ Previous evidence of an association between C-reactive protein and BP has been derived solely from cross-sectional studies in which no causal link can be established.^6-9 In those studies, C-reactive protein has been more strongly associated with systolic BP than diastolic BP, which is consistent with the emerging importance of systolic BP as a means of cardiovascular risk prediction.²⁶

Although the current data provide evidence for a critical role of inflammation in the development of hypertension, the mechanisms of this effect are uncertain and require further evaluation. C-reactive protein has been reported to decrease production of nitric oxide by endothelial cells^12,13 and thus might indirectly promote vasoconstriction, leukocyte adherence, platelet activation, oxidation, and thrombosis.^14,15,27 C-reactive protein also has been reported to have proatherosclerotic properties by upregulating angiotensin type-1 receptor expression,¹⁶ affecting the renin-angiotensin system and contributing to the pathogenesis of hypertension. These changes are all indicative of progressive atherosclerosis and endothelial dysfunction,²⁸ with structural and functional changes in the endothelium ultimately leading to the development of hypertension.²⁹ These findings are supported by cross-sectional associations not only for C-reactive protein but also IL-6, intercellular adhesion molecule 1, and tumor necrosis factor α with either BP or hypertension.^8,9,30

Higher levels of C-reactive protein also play an important role in the induction of plasminogen activator inhibitor 1 (PAI-1), a marker of impaired fibrinolysis and atherothrombosis.^15,31 The recent novel finding that C-reactive protein increases PAI-1 expression and activity in human aortic endothelial cells¹⁵ supports a possible mechanism by which the association between C-reactive protein and the development of hypertension is mediated. Patients with hypertension have markedly higher levels of PAI-1 than normotensive patients,³² and both systolic and diastolic BP were significantly (P<.001) and positively associated with PAI-1 levels in the Framingham offspring cohort.³³

Potential limitations of our study warrant discussion. As in any epidemiologic study, residual confounding is of concern. After controlling for other known risk factors for both hypertension and atherothrombosis, as well as baseline BP, our results remained significant. We observed similar results in subgroup analyses limited to those participants without baseline coronary risk factors. We relied on a single baseline measurement of C-reactive protein. However, C-reactive protein has been shown to be stable over long periods of follow-up, with little or no diurnal variation.^34,35

In our study, incident hypertension was based on self-reported BP, treatment, physician diagnosis, or all 3 combined. To address this potential limitation, we performed a separate study in which an 86% validation rate for self-reported hypertension was observed, consistent with other studies.^36,37 For example, in a comparable population in the Nurses' Health Study, 99% of women who reported high BP confirmed their diagnosis based on medical records.³⁷ Sensitivity analyses considering various definitions of hypertension also yielded consistent and significant linear trends across increasing levels of C-reactive protein. Any impact of misclassification would likely be random, biasing our observed RRs toward the null hypothesis.

Although hypertension awareness, treatment, and control rates have increased during the past 3 decades, the identification of individuals at risk for hypertension remains a high priority.²⁴ These data provide evidence that inflammation may be an important mechanism through which hypertension develops.