Screening for atherosclerosis in patients with rheumatoid arthritis: Comparison of two in vivo tests of vascular function




Inflammation appears to play a central role in atherosclerosis, and endothelial damage mediated by systemic inflammation may contribute to the increased cardiovascular mortality in rheumatoid arthritis (RA). Brachial artery flow-mediated dilatation (FMD) and pulse wave analysis (PWA) are measures of vascular function. The aim of this study was to determine if FMD and PWA are abnormal in patients with RA.


Twenty-five RA patients and 25 matched healthy controls were studied. All were free of traditional cardiovascular risk factors. FMD was measured in all subjects. PWA was performed in 18 RA patients and 18 controls, with results expressed as large and small artery compliance (C1 and C2). Modified Sharp scores were calculated in 13 RA patients.


Results (mean ± SD) in RA patients and controls, respectively, were as follows: FMD 107.6 ± 4.6% versus 108.5 ± 4.1% (P = 0.49), C1 14.8 ± 2.8 ml/mm Hg × 10 versus 17.9 ± 3.1 ml/mm Hg × 10 (P = 0.0033), C2 4.5 ± 2.3 ml/mm Hg × 100 versus 7.7 ± 3.7 ml/mm Hg × 100 (P = 0.0039). There was an inverse correlation between C2 and modified Sharp scores in the RA patients (Spearman's rho −0.69, P = 0.0085).


FMD was normal in these RA patients, whereas arterial compliance was markedly reduced. PWA appears to be a more sensitive measure of vascular dysfunction than FMD in RA and may be the preferred surrogate marker of vascular dysfunction in longitudinal studies of RA patients. The inverse correlation between C2 and the modified Sharp score, a measure that reflects disease activity over time, supports the notion that chronic inflammation plays a role in RA-associated atherosclerosis.

Rheumatoid arthritis (RA) is associated with excess cardiovascular mortality, which is not explained by systemic vasculitis or traditional cardiovascular risk factors (1, 2). The role of more recently identified risk factors such as homocysteine and lipoprotein(a) in RA-associated vascular disease is unclear, although there have been reports of these occurring at elevated levels in RA patients (3–6). There is growing evidence that inflammation plays a role in the initiation and progression of atherosclerosis, the conversion of a stable to an unstable plaque, and the superimposed thrombosis which usually precipitates acute vascular events (7). For example, prospective epidemiologic studies in the general population have demonstrated a relationship between high-sensitivity C-reactive protein (hsCRP) levels and future risk of cardiovascular events (8, 9). Elevated concentrations of serum amyloid A, interleukin-6, and soluble intercellular adhesion molecule 1 have also been identified as risk markers for future cardiovascular events (10–12). The similarity between rheumatoid synovitis and atherosclerotic plaque has been noted (13), and most patients with RA have evidence of systemic inflammation in addition to articular disease.

It is reasonable, therefore, to postulate that the systemic inflammation associated with RA, acting either in isolation or in synergy with traditional and/or novel risk factors (1), could contribute to accelerated atherosclerosis in RA patients. In order to conduct further research into the prevalence and determinants of atherosclerosis in RA and to perform large-scale intervention studies, a surrogate marker of vascular function is needed. Ideally, such a test would be noninvasive and able to detect vascular disease at an early stage to allow effective intervention.

Maintenance of vascular homeostasis is largely dependent on the endothelial lining of blood vessels. Endothelial cells release vasoactive mediators (e.g., nitric oxide, endothelin 1) and express cell surface molecules (e.g., leukocyte adhesion molecules) which influence vascular tone, leukocyte adherence, platelet activation, coagulation, and smooth muscle proliferation (14). Endothelial dysfunction, characterized by reduced nitric oxide bioavailability, is an early stage of atherosclerosis. Several noninvasive tests of endothelial function are available. Flow-mediated dilatation (FMD) is an ultrasound-based technique which measures vasodilatation of the brachial artery in response to increased vessel wall shear stress (15). This vasodilatation is mediated by endothelial cell release of nitric oxide (16), and thus FMD is considered to be a measure of endothelium-dependent vasodilatation. FMD results correlate with findings obtained by invasive testing of coronary endothelial function (17) and with the extent and severity of coronary artery disease (CAD) (18). Impaired FMD has been demonstrated in subjects with vascular risk factors such as diabetes mellitus, smoking, hypertension, and hypercholesterolemia (15). FMD is also abnormal in inflammatory states including acute vaccine-induced systemic inflammation, systemic necrotizing vasculitis, and systemic lupus erythematosus (19–21).

In most studies of FMD, endothelium-independent vasodilatation (a measure of smooth muscle function) is also assessed by recording the vasodilator response to an exogenous nitric oxide donor such as sublingual glyceryl trinitrate (GTN) (22). Analysis of pooled data from 800 subjects who had participated in various studies of endothelial function demonstrated a significant relationship between GTN- and flow-mediated dilatation, independent of the effects of vessel size or traditional vascular risk factors (23). More recently, the dose-response curve for GTN-mediated vasodilatation has been examined in subjects with proven CAD and in healthy controls (24). Compared with controls, the CAD patients had significantly reduced GTN-mediated vasodilatation, with the greatest difference observed with lower doses of GTN (150–400 μg). This suggests that, in addition to endothelial dysfunction, atherosclerosis is associated with functional abnormalities of vascular smooth muscle cells (24).

Pulse wave analysis (PWA) is a technique in which large and small artery compliance (or elasticity) is estimated from analysis of the peripheral arterial waveform. The radial artery pulse wave is measured with a tonometer, and large and small artery compliance (C1 and C2, respectively) is calculated from the recorded waveform, using a modified Windkessel model (25). The modified Windkessel model is a mathematical depiction of the vasculature as a number of related components: capacitive compliance (C1 index), oscillatory or reflective compliance (C2 index), inductance, and resistance (systemic vascular resistance), measured during the diastolic decay portion of the cardiac cycle (25). This technique has been validated against invasive tests of arterial compliance (26), and reduced arterial compliance has been demonstrated in patients with hypertension, patients with diabetes, and smokers (27–29). Increased arterial stiffness as measured by pulse wave velocity has also been correlated with coronary atherosclerosis (30) and has been shown to predict coronary events and mortality in patients with hypertension (31, 32).

This study was undertaken to determine the relationship between RA-associated systemic inflammation and vascular function, using FMD and PWA. To reduce the confounding effects of known determinants of endothelial and vascular function, we studied young patients (ages ≤55 years) who did not have major cardiovascular risk factors. Brachial artery FMD was measured in 25 RA patients and 25 healthy controls, as well as in 10 individuals with multiple cardiovascular risk factors or proven cardiovascular disease. PWA was measured in a subgroup of 18 RA patients and 18 matched healthy controls.


Demographic characteristics of the study subjects.

We recruited 25 patients (6 men, 19 women; age range 26–55 years) with RA according to the criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (33) and 25 age- and sex-matched healthy controls. Exclusion criteria for both groups were known or suspected cardiovascular disease or the presence of 1 or more traditional cardiovascular risk factors. We defined traditional cardiovascular risk factors as cigarette smoking (in the previous 10 years), hypertension (systolic blood pressure >140 mm Hg or diastolic blood pressure >90 mm Hg), diabetes mellitus, hypercholesterolemia (total cholesterol >6.5 mmoles/liter or taking cholesterol-lowering medication), and a history of premature (in men, age ≤45 years; in women, age ≤55 years) CAD in a first-degree relative. We also recruited 10 subjects with either proven CAD or 2 or more traditional cardiovascular risk factors, to demonstrate the validity of FMD. The study was approved by the Institutional Ethics Committee, and written informed consent was obtained from all subjects. Demographic and clinical characteristics of the RA and healthy control groups are shown in Table 1.

Table 1. Demographic and clinical characteristics of the healthy controls and the RA patients*
 Healthy controls (n = 25)RA patients (n = 25)P
  • *

    Except where indicated otherwise, values are the mean ± SD. Normal ranges for fasting glucose, fasting insulin, lipoprotein(a), and homocysteine, respectively, are 3.0–6.0 mmoles/liter, <15 mIU/liter, <0.24 gm/liter, and 5.0–15.0 μmoles/liter. RA = rheumatoid arthritis; LDL = low-density lipoprotein; HDL = high-density lipoprotein. ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; hsCRP = high-sensitivity CRP; DMARDs = disease-modifying antirheumatic drugs; NSAID = nonsteroidal antiinflammatory drug.

  • Determined in 13 patients.

Age, years44.8 ± 8.646.4 ± 8.00.51
Female, no. (%)19 (76)19 (76)1.00
Systolic blood pressure, mm Hg123 ± 9.2125 ± 10.00.21
Diastolic blood pressure, mm Hg78 ± 7.980 ± 7.00.29
Mean arterial pressure, mm Hg93 ± 7.895 ± 7.70.28
Body mass index, kg/m225.2 ± 4.826.5 ± 4.80.32
Total cholesterol, mmoles/liter5.5 ± 0.85.3 ± 0.80.24
LDL cholesterol, mmoles/liter3.3 ± 1.22.9 ± 0.80.20
HDL cholesterol, mmoles/liter1.7 ± 0.61.7 ± 0.60.67
Triglycerides, mmoles/liter1.2 ± 0.41.5 ± 0.90.10
Fasting glucose, mmoles/liter5.2 ± 0.55.2 ± 0.60.60
Fasting insulin, mIU/liter, median (range)4.5 (1.6–18.5)7.4 (2.2–17.5)0.0008
Lipoprotein(a), gm/liter, median (range)0.07 (0.01–1.21)0.14 (0.01–0.88)0.44
Homocysteine, μmoles/liter, median (range)8.9 (5.4–12.4)9.0 (5.8–17.3)0.91
ESR, mm/hour, median (range)12 (1–48)19 (1–68)0.07
CRP, mg/liter, median (range)1 (1–12)4 (1–59)0.006
hsCRP, mg/liter, median (range)0.8 (0.18–14.4)3.43 (0.56–18.2)0.0002
Rheumatoid factor positive, no. (%)16 (64)
Erosions present, no. (%)22 (88)
Modified Sharp score, median (range)58 (0–364)
Duration of RA, years13.7 ± 9.0
No. of DMARDs ever used3.3 ± 1.9
Currently using DMARD, no. (%)24 (96)
Currently using prednisolone, no. (%)9 (36)
Currently using NSAID, no. (%)14 (56)
Disease Activity Score4.6 ± 1.6

Baseline evaluation.

Venous blood was obtained from all subjects after they had been fasting. Measurement of erythrocyte sedimentation rate (ESR), rheumatoid factor, lipid levels (total, high-density lipoprotein, and low-density lipoprotein cholesterol, and triglycerides), lipoprotein(a), insulin, glucose, and homocysteine was performed by the Royal Melbourne Hospital Pathology Service. Serum hsCRP was measured by immunonephelometry (BN-II nephelometer; Dade-Behring, Marburg, Germany) (intraassay coefficient of variation 2.7%). Resting 12-lead electrocardiography (EKG) was performed on all subjects. Radiography of the hands and feet of the RA patients was performed to detect bone erosions, unless the presence of erosions had been documented previously. Radiographs obtained during the study were scored according to the modified Sharp method (34). Disease activity in the RA patients was measured using the Disease Activity Score (DAS), a validated composite score incorporating tender and swollen joint counts, ESR, and a patient global assessment of disease activity (100-mm visual analog scale) (35). A DAS of ≤1.6 indicates remission, while a value of ≥4.3 suggests active disease (35).

Vascular function testing.

Brachial artery FMD was performed in all RA patients, healthy controls, and positive controls (60 subjects in total). Eighteen of the RA patients and their matched healthy controls were available for PWA testing.

Brachial artery ultrasonography was performed by a trained investigator using the standard technique (15). It was administered in the morning, after the subject had fasted overnight. Nonsteroidal antiinflammatory drugs (NSAIDs) were withheld for 24 hours prior to study. No subject was taking aspirin, cardiac medication, or a lipid-lowering agent. Subjects rested supine in a quiet, temperature-controlled (22–24°C) room. Continuous EKG monitoring and regular blood pressure measurement were performed. An inflatable cuff was placed around the upper forearm. Studies of the left brachial artery were performed using high-resolution ultrasound (ATL HDI Ultramark 9; generously provided by Philips Medical Systems, Seattle, WA) with a 7–10-MHz linear array transducer. The left brachial artery was scanned in longitudinal section 2–15 cm above the elbow, and the transducer was fixed in position by a magnetic clamp. Focus, depth, and gain settings were set to optimize images of the lumen/arterial wall interface and remained unchanged throughout the study. A baseline scan was recorded for 2 minutes, followed by inflation of the cuff to 250 mm Hg for 5 minutes. The FMD scan commenced 30 seconds prior to release of the cuff and continued for 2 minutes after cuff deflation. Ten minutes later a second baseline scan was performed. A tablet of GTN (600 μg) was administered sublingually and recording continued for a further 5 minutes (the GTN scan).

The ultrasound images were recorded on Super-VHS videotape for subsequent analysis. All scans were analyzed by an investigator who was blinded to subject identity and disease status. Arterial diameter was measured from the intima–lumen interface of the near and far walls of the artery using ultrasonic calipers, coincident with the R-wave peaks on EKG. A minimum of 12 cardiac cycles were analyzed during the baseline and GTN scans. During the FMD scan, measurements were obtained at 5-second intervals following cuff deflation. FMD and GTN-mediated dilatation are expressed as the percentage change in artery diameter from baseline.

PWA was performed by a trained investigator using the standard technique (26). It was administered in the morning, after the subject had fasted overnight. A radial artery pressure sensor (tonometer) was placed over the radial artery at the right wrist and fixed in position. The position of the sensor was adjusted to optimize the radial artery waveform recording and remained in the same position throughout the study. The waveform was calibrated with a blood pressure cuff on the opposite arm and an internal calibration system. Thirty-second tracings of the radial artery waveform were digitized and waveform data were analyzed using a computerized modified Windkessel model (HDI) to derive estimations of large and small artery compliance (C1 and C2). The mean of 3 separate measurements was used in data analysis.

Reliability and validity.

Test–retest reliability of brachial artery FMD was assessed prior to institution of the formal study, by measuring FMD and GTN-mediated dilatation on 2 occasions, up to a week apart, in 11 volunteers. Between-test differences (mean ± SD) in FMD and GTN-induced dilatation were 1.5 ± 1.3% (intraclass correlation 0.79) and 3.1 ± 1.5% (intraclass correlation 0.88), respectively. Validity was demonstrated by measuring FMD and GTN-mediated dilatation under blinded conditions in 10 positive control subjects with either known CAD or ≥2 cardiovascular risk factors. Test–retest reliability of PWA was assessed prior to institution of the formal study, by measuring C1 and C2 on 2 occasions, up to a week apart, in 11 volunteers. Between-test differences (mean ± SD) in C1 and C2 were 2.8 ± 2.3 ml/mm Hg × 10 (intraclass correlation 0.74) and 1.0 ± 0.6 ml/mm Hg × 100 (intraclass correlation 0.95), respectively.

Statistical analysis.

Sample size calculations were performed based on the FMD component of the study. In previous studies of healthy subjects, reported mean FMD values have ranged from 106% to 112% (SD 3–6.5%), while in subjects with vascular risk factors such as cigarette smoking, hypertension, or hypercholesterolemia, mean FMD ranged from 100% to 106.5% (20, 36, 37). For the purpose of sample size calculations we estimated a mean ± SD FMD of 108 ± 4% in our healthy controls. Detection of a minimum reduction in mean FMD of 4% in RA patients compared with healthy controls, with power = 0.9 and α = 0.05, required 22 subjects per group. Results are expressed as the mean ± SD unless indicated otherwise. The RA and control groups were compared using the 2-sample (independent) t-test (for normally distributed data) and Wilcoxon's rank sum test (for skewed data). P values less than 0.05 were considered significant.


The RA and healthy control groups were well matched in terms of age, sex, blood pressure, lipid levels, and body mass index (BMI) (Table 1). The majority of RA patients had seropositive, erosive disease which was longstanding and had been treated aggressively with disease-modifying antirheumatic drugs (DMARDs) (Table 1). The mean DAS was 4.6, consistent with active disease despite DMARD treatment. The RA patients had a significantly higher median CRP level (4 mg/liter versus 1 mg/liter; P = 0.006), and median hsCRP level, (3.4 mg/liter versus 0.8 mg/liter; P = 0.0002) and a higher median ESR (19 mm/hour versus 12 mm/hour; P = 0.07) than the healthy controls (Table 1), reflecting active systemic inflammation. Fasting insulin levels were higher in the RA patients; however, the difference was not significant after correction for prednisolone usage. Fasting glucose, homocysteine, and lipoprotein(a) levels were similar in the RA and healthy control groups. Five of the female RA patients and 4 of the female healthy control subjects were postmenopausal. The positive control subjects were older (mean ± SD 60.3 ± 9.2 years) and had higher systolic blood pressure (150 ± 14.8 mm Hg) and BMI (33.3 ± 9.5 kg/m2) compared with the other 2 groups.

Flow-mediated dilatation.

FMD was significantly impaired in the positive controls compared with the healthy controls (mean ± SD 101.1 ± 3.9% and 108.5 ± 4.1%, respectively; P < 0.0001) (Figure 1 and Table 2). This result was expected and demonstrates the discriminative capability of this test under the experimental conditions used in this study. In contrast, FMD was normal in the RA patients (107.6 ± 4.6%), with no significant difference in percentage dilatation compared with the healthy controls (P = 0.49). GTN-mediated dilatation and baseline brachial artery diameter were similar in the RA and healthy control groups (Table 2). There was no significant correlation between FMD and ESR, CRP, DAS, disease duration, or Sharp score (data not shown).

Figure 1.

Flow-mediated dilatation in rheumatoid arthritis patients versus healthy controls and subjects with either proven coronary artery disease or ≥2 traditional cardiovascular risk factors (positive controls).

Table 2. Brachial artery vasodilatation in control subjects and RA patients*
 Positive controls (n = 10)Healthy controls (n = 25)RA patients (n = 25)
  • *

    Positive controls had either proven coronary artery disease or ≥2 traditional cardiovascular risk factors. Values are the mean ± SD. RA = rheumatoid arthritis; GTN = glyceryl trinitrate.

  • P < 0.0001 versus healthy controls.

Brachial artery diameter, mm3.4 ± 0.93.0 ± 0.52.9 ± 0.5
Flow-mediated dilatation, %101.1 ± 3.9108.5 ± 4.1107.6 ± 4.6
GTN-mediated dilatation, %120.5 ± 13.3126.5 ± 7.3129.2 ± 10.7

Pulse wave analysis.

Both C1 and C2 were significantly reduced in the RA patients compared with the healthy controls. C1 was 14.8 ± 2.8 ml/mm Hg × 10 in RA patients versus 17.9 ± 3.1 ml/mm Hg × 10 in healthy controls (P = 0.0033) (Figure 2 and Table 3), and C2 was 4.5 ± 2.3 ml/mm Hg × 100 in RA patients versus 7.7 ± 3.7 ml/mm Hg × 100 in healthy controls (P = 0.0039) (Figure 3 and Table 3). At the time of PWA testing the diastolic blood pressure was normal (≤90 mm Hg) in all subjects; however, mean diastolic blood pressure and mean arterial pressure were greater in the RA patients than in the healthy controls (78 mm Hg versus 73 mm Hg and 94 mm Hg versus 89 mm Hg, respectively; P = 0.042 and P = 0.053). The resting pulse rate was also greater in the RA patients (72 beats per minute versus 62 beats per minute; P = 0.0042). A modified Sharp score was available for 13 of the 18 RA patients who had PWA. In these patients there was a significant inverse correlation between C2 and Sharp score (Spearman's rho −0.69, P = 0.0085) (Figure 4). There was no significant correlation between PWA and ESR, CRP, hsCRP, DAS, disease duration, or FMD (data not shown).

Figure 2.

Large artery compliance (C1) in rheumatoid arthritis patients versus healthy controls.

Table 3. Pulse wave analysis in healthy controls and RA patients*
 Healthy controls (n = 18)RA patients (n = 18)P
  • *

    Values are the mean ± SD. RA = rheumatoid arthritis.

Systolic blood pressure, mm Hg121 ± 11126 ± 100.1
Diastolic blood pressure, mm Hg73 ± 578 ± 80.042
Mean arterial pressure, mm Hg89 ± 6.694 ± 8.70.053
Pulse rate, beats per minute62 ± 9.972 ± 9.30.0042
Height, cm169 ± 8.5166 ± 8.60.22
Large artery compliance (C1), ml/mm Hg × 1017.9 ± 3.114.8 ± 2.80.0033
Small artery compliance (C2), ml/mm Hg × 1007.7 ± 3.74.5 ± 2.30.0039
Figure 3.

Small artery compliance (C2) in rheumatoid arthritis patients versus healthy controls.

Figure 4.

Correlation between small artery compliance (C2) and modified Sharp score in rheumatoid arthritis patients.


Connective tissue diseases such as RA and systemic lupus erythematosus appear to be associated with accelerated atherosclerosis (1, 38). The mechanisms involved are not clear, but may include endothelial dysfunction mediated by systemic inflammation. Elucidation of these mechanisms may provide insight into the role of inflammation in atherosclerosis. The onset of RA generally occurs in early adulthood, and it is therefore especially important to explore potential markers for the early stages of atherosclerosis in this population. In the present study we explored 2 measures of vascular function, FMD and PWA, in young RA patients who did not have traditional cardiovascular risk factors. We first confirmed the reproducibility of FMD and PWA and demonstrated that FMD measurement performed as expected in patients with known cardiovascular disease. We then measured FMD and PWA in RA patients and age- and sex-matched controls who were carefully chosen to exclude the confounding influence of traditional cardiovascular risk factors. We found that both large and small artery compliance was reduced in the RA patients compared with the healthy controls. In contrast, FMD was normal in these RA patients. We believe this latter finding is accurate because we have demonstrated our ability to detect an abnormality in FMD if one is present, and our sample size calculation suggests that the study had adequate power to detect any meaningful difference.

A number of issues need to be considered when interpreting these results. The effect of treatment with antirheumatic medications on vascular function is unknown. In particular, NSAIDs could have a significant impact on endothelial function. The nonselective NSAIDs inhibit production of both prostacyclin (a vasodilator) and thromboxane A2 (a vasoconstrictor), whereas the cyclooxygenase-2 (COX-2)–specific inhibitors selectively inhibit prostacyclin alone (39). Fifty-six percent of the RA patients in this study were taking various selective and nonselective NSAIDs, the plasma half-lives of which range from 1.5 to 17 hours. Although NSAID treatment was discontinued in all patients at least 24 hours prior to vascular function testing, it is possible that washout of the drug was incomplete in some patients. In a study of healthy volunteers, however, Verma et al (40) demonstrated that 7 days of treatment with either naproxen (a nonselective NSAID) or rofecoxib (a selective COX-2 NSAID) did not alter endothelial function. Furthermore, we noted no difference in FMD or PWA between NSAID-treated and non–NSAID-treated patients. Hence, we believe it is unlikely that recent use of NSAIDs impacted significantly on our results.

The HDI PWA system used in this study, which derives arterial compliance from diastolic pulse contour analysis using the modified Windkessel model, is one of a number of different techniques to measure arterial compliance. The validity of the physiologic assumptions used in the Windkessel model has recently been debated (41). This issue is not fully resolved; however, the HDI system has been validated against invasive tests of arterial compliance (26), and reduced arterial compliance has been demonstrated in subjects with a variety of cardiovascular risk factors (27–29). Therefore, C1 and C2 may still be clinically useful as biomarkers of arterial dysfunction or disease (41).

In this study of vascular function in RA patients, FMD was preserved whereas large and small artery compliance was significantly reduced. How can these apparently discordant results obtained using 2 different measures of vascular function be explained? Not all subjects who participated in the FMD study were able to participate in PWA testing, and this could potentially have given rise to selection bias. However, the 18 RA patients who had PWA performed had lower mean blood pressure and cholesterol levels and higher mean FMD than the RA patients who did not undergo PWA. Although these differences were not statistically significant, the selected group presumably had better vascular function than the untested subjects, and hence any selection bias would favor the null hypothesis.

FMD and PWA are both dependent on nitric oxide bioavailability (16, 42) and therefore reflect endothelial cell function; however, the techniques have important differences. FMD is a stress test of dilatation in response to increased flow, whereas PWA measures resting arterial compliance and tone. FMD is mediated predominantly by endothelial release of nitric oxide, since the vasodilator response is almost completely blocked by treatment with nitric oxide synthase inhibitors (16). Therefore, the FMD response can be considered to be almost entirely dependent on endothelial cell function. McVeigh et al (43) reported that systemic infusion of L-NG-nitroarginine methyl ester, an inhibitor of nitric oxide synthesis, reduced small artery compliance in healthy male volunteers. Using a different method of pulse wave analysis, Wilkinson et al (42) demonstrated that infusion of L-NG-monomethyl arginine, an inhibitor of nitric oxide synthase, led to a dose-dependent increase in systemic arterial stiffness in healthy men. These data indicate that arterial stiffness is, in part, regulated by nitric oxide availability, and therefore PWA, at least in part, measures endothelial cell function. However the relative contribution of other factors, such as smooth muscle and connective tissue content of the artery wall, to PWA is unknown. The relationship between FMD and PWA has not been examined extensively, although correlation between the 2 measurements has been reported in a population of growth hormone–deficient subjects with abnormal endothelial function (44).

Perhaps PWA is a more sensitive measure of endothelial dysfunction than is FMD, detecting subtle reduction of vascular compliance before impaired endothelium-dependent vasodilatation becomes apparent. Alternatively, RA may cause specific vascular abnormalities, such as structural change in vessel walls or smooth muscle dysfunction, that are more readily detected by PWA than FMD. In this case one might expect GTN-mediated vasodilatation to be reduced in RA patients. We did not demonstrate a difference in GTN-mediated vasodilatation between RA patients and healthy subjects; however, the dose of 600 μg of GTN used in this study (chosen to obtain the maximal vasodilator response) may be too high to allow detection of this. Ideally, a dose-response curve for GTN-mediated dilatation should be performed to fully evaluate endothelium-independent vasodilatation.

Blood pressure, pulse rate, and height are known determinants of measured arterial stiffness (45). At the time of PWA measurement the RA patients had a higher mean diastolic blood pressure, mean arterial pressure, and resting pulse rate than the healthy controls, although the values remained within the normal range. Although the between-group differences in C1 and C2 persisted after correction for blood pressure (in an analysis of covariance model, in which mean arterial pressure, pulse rate, and height were included as covariants, a significant difference between groups was still evident [C1: F = 5.7, P = 0.023; C2: F = 5.9, P = 0.021]), this relative hypertension in the RA patients may be clinically important. Hypertension is an established and powerful risk factor for cardiovascular disease (46). Meta-analysis of multiple observational studies involving more than 420,000 untreated individuals projected that a 5–6 mm Hg average decrease in diastolic blood pressure would lead to a 25% reduction in coronary mortality (46). Many of the medications used to treat RA elevate blood pressure, including NSAIDs, glucocorticoids, and cyclosporine. Perhaps the excess cardiovascular risk in RA can be partially attributed to a slightly higher blood pressure in this population, even though it may remain within the normal range.

Increased resting heart rate has also been identified as an independent risk factor for cardiovascular and all-cause mortality in prospective population studies (47–50). The mechanism by which this operates is not known, but it may reflect physical fitness. Exercise capacity and activity status are established predictors of cardiovascular and overall mortality (51, 52), and RA may be associated with reduced exercise capacity due to joint inflammation and fatigue. The higher resting pulse rate in these RA patients may reflect reduced physical fitness, which in turn is associated with endothelial dysfunction.

If inflammation does indeed contribute to vascular disease, it could be due to the cumulative effect of chronic, low-grade inflammation, acute inflammation precipitating rupture of an unstable plaque, or a combination of the two. We found a significant inverse correlation between C2 and modified Sharp scores in the RA patients, whereas there was no correlation between arterial compliance and ESR, CRP, hsCRP, disease duration, or current DAS. These latter measures reflect RA disease activity at a single point in time, whereas the Sharp score reflects the degree of joint inflammation that has occurred over time (53). This relationship between cumulative disease activity and severity of arterial stiffness supports the notion that chronic inflammation plays a role in RA-associated atherosclerosis.

This is the first reported study of FMD and PWA in RA patients who were carefully screened to exclude an alternative explanation for abnormal vascular function. In our RA patients with moderate-to-severe arthritis and many years of systemic inflammation, FMD was preserved whereas large and small artery compliance was significantly reduced. Greater understanding of the determinants of arterial compliance may provide insights into the underlying mechanism(s) of accelerated atherosclerosis in RA. Blood pressure is a major determinant of arterial stiffness, and regular screening for and rigorous treatment of hypertension in patients with RA may be especially important. It may even be that the “safe” level of blood pressure in RA patients is lower than that for the general population. Reduced physical fitness may be a major contributor to the excess mortality in RA. Exercise training has been shown to be safe and beneficial in RA patients (54), and pursuit of regular exercise should be strongly encouraged in this group. The reduction in C2 in the RA patients correlated with the Sharp score, a measure that is known to reflect disease severity and activity over time (53). Further research is needed to determine whether optimal disease control with DMARDs preserves vascular function in RA. PWA appears superior to FMD in identifying asymptomatic vascular disease in RA patients, and investigators conducting intervention studies aimed at reducing vascular disease in RA should consider including PWA as an end point.

Longitudinal followup of a cohort of RA patients is needed to ascertain the extent to which increased arterial stiffness correlates with future cardiovascular events. In the meantime, our results highlight blood pressure, physical inactivity, and disease activity as potentially correctable causes of the increased cardiovascular burden in RA.