Methods

Studies were conducted in a quiet, temperature- and humidity-controlled environment. The local ethics committee approved all studies and written informed consent was obtained from all patients. All studies were conducted between 11.00 am and 2.00 pm. Subjects were in a fasting state for at least 6 h.

Radial artery diameter, blood flow, L-FMC, FMD and their composite end point were measured using an Acuson Sequoia 512 (Mountain View, USA) with a 15 MHz linear-array transducer and automatic analysis software as described in fig 1 and previously published.9 Briefly, a pneumatic cuff was positioned distal to the ultrasound probe in order to avoid ischaemia of the artery studied. Radial artery diameter was measured at rest, during inflation of the distal cuff to suprasystolic pressure (4.5 min) and for the 5 min following deflation. All files were coded by a nurse not involved in other study procedures and blinded, completely automatic analysis was performed using custom-developed software (SpLiNeS, Siena, Italy).11

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Representative plot of radial artery diameter measured continuously using offline software. A 15 MHz ultrasound probe is positioned on the radial artery and B-mode images are acquired continuously. After 1 min of resting diameter measurement, a pneumatic cuff (placed distal to the probe in order to avoid radial artery ischaemia) is inflated to suprasystolic pressure. The subsequent decrease in local blood flow causes a progressive decrease in the radial artery diameter until a plateau (L-FMC). Upon cuff deflation, the increased blood flow causes radial artery dilatation (FMD). L-FMC is calculated as the percentage decrease in arterial diameter in the last 30 s of cuff occlusion as compared with resting diameter. FMD is calculated as the maximum percentage increase in arterial diameter following cuff deflation. A composite end point is also calculated as the sum of the absolute values of FMD and L-FMC. Radial artery blood flows for each time point are presented. More details are given in Gori et al.9

Accuracy of L-FMC and FMD in detecting diagnosed cardiovascular disease

Forty healthy young volunteers (age range 25–35, 17 women), 24 healthy middle-age adults (age 40–65, 12 women), 24 hypertensive patients (age 40–75, eight women), 24 patients with coronary artery disease (CAD, age 49–71, one woman) and 24 patients with congestive heart failure (age 60–80, seven women) were studied. All patients were consecutively enrolled from those attending our outpatient clinic and were clinically stable at the time of enrolment. Some of these patients (12 hypertensive patients and 10 patients with CAD) participated in other studies from our group testing separate end points. Healthy subjects had normal (<180 mg/dl) cholesterol and triglyceride (<150 mg/dl) levels, no family history of premature cardiovascular disease and were lifelong non-smokers with blood pressure <130/80 mm Hg. Hypertensive subjects were enrolled at the time of first diagnosis, before initiation of treatment, and had 24 h blood pressure monitoring documenting average systolic values >140/90 mm Hg. None of them was receiving treatment with any drug, including supplemental vitamins, none had a history of diabetes and none had clinical evidence of coronary artery or peripheral artery disease. Patients with CAD had at least one stenosis >70% in one major coronary as shown by angiography. All were receiving aspirin. Patients with heart failure had a left ventricular ejection fraction ⩽35%. They presented New York Heart Association (NYHA) class II–III symptoms and were on a stable regimen including diuretic agents, angiotensin converting enzyme inhibitors and β blockers for at least 1 month before participation in the study.

Effect of exercise on FMD and L-FMC

Twelve healthy young volunteers (age 25–35, seven men) were enrolled in this crossover, investigator-blinded, randomised study. In the first of two visits, subjects were asked to lie supine for at least 10 min, and were then randomised to isometric exercise or no intervention. Isometric exercise consisted of 4 min (or until muscular exhaustion) of rhythmic handgrip contraction of the ipsilateral hand. No intervention consisted of 4 min of rest. Radial artery diameter, radial artery blood flow, as well as L-FMC and FMD, were measured immediately at the end of the exercise (or after 4 min of rest in the subjects randomised to no exercise). All subjects returned to the laboratory 1 week later (visit 2) to undergo the alternate protocol (ie, isometric exercise or 4 min of rest) before the same measurements were taken again.

Statistical analysis

Data are presented as mean (SD). For the analysis of the effect of exercise on L-FMC and FMD, a two-way analysis of variance (ANOVA) for repeated measurements was employed. For the comparison between healthy subjects and the different patient groups, one-way ANOVA was employed. Post hoc comparisons were done with the Fisher test. Statview version 5 (SAS Institute, Cary, North Carolina, USA) was used for ANOVA and correlation analysis. To compare the accuracy of L-FMC with that of FMD and of their composite end point in distinguishing patients with cardiovascular disease from healthy volunteers, receiver operating characteristic (ROC) curves were constructed12 13 14 and compared using Medcalc version 9.2.1.0 (Mariakerke, Belgium). The threshold for significance was set as p<0.05.

Results

Effect of isometric exercise on L-FMC and FMD

All data are presented in table 1 and fig 2. Resting arterial diameter and blood flow were similar in the two visits, at rest. Both parameters increased significantly in response to isometric exercise, so that there was a significant difference between visits in both end points immediately before the beginning of L-FMC and FMD procedures (p<0.05). During the inflation of the wrist cuff, both absolute blood flow and absolute radial artery diameter decreased significantly (p<0.001) and reached a nadir that was not different across visits. Similarly, both absolute blood flow and radial artery diameter increased after deflation of the blood cuff, and there was no difference across groups in the peak absolute values of either parameter (p>0.1). Despite similar nadir and peak diameters (measured in mm), L-FMC and FMD (expressed as percentage changes over values measured at the end of the rest or the exercise—that is, immediately before L-FMC and FMD procedures), were significantly different across visits. In detail, L-FMC and the percentage decrease in blood flow induced by cuff inflation were significantly larger after exercise as compared with after rest (p<0.05 between visits), while percentage reactive hyperaemia and FMD were significantly lower on the exercise visit as compared with the other visit (p<0.05). Importantly, the combined end point of endothelial function (ie, the sum of the absolute values of L-FMC and FMD) was not different across visits (p>0.8).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

The effect of exercise on radial artery diameter, blood flow, low-flow-mediated constriction (L-FMC) and flow-mediated vasodilatation (FMD). Left panel: both radial artery diameter (columns) and radial artery blood flow (dashed lines) were significantly higher immediately at the end of isometric exercise (blue columns and dashed lines). Right panel: As a consequence of this difference in baseline arterial diameter, L-FMC (negative columns) was also significantly higher after exercise (blue), while FMD was significantly blunted. *p<0.05 compared with before exercise and corresponding time point, no intervention.

L-FMC in healthy volunteers and patients with cardiovascular disease

All data are presented in table 2. As previously published,9 FMD, L-FMC and their composite end point were similar across healthy young and middle-age volunteers and were significantly greater than those measured in patients with hypertension, coronary artery disease and congestive heart failure (p<0.01 for all).

When studied with ROC analysis (fig 3, table 2), the area under the curve for the composite end point was significantly higher than that for FMD alone for each of the patients subgroups (fig 3). When all patients were combined, the area under the curve was 0.78 (confidence intervals 0.70 to 0.84) for FMD, 0.80 (0.73 to 0.87) for L-FMC (p = NS vs FMD) and 0.87 (0.80 to 0.92) for their composite end point (p<0.005 vs FMD, p<0.05 vs L-FMC).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Receiver operating characteristic curves (ROC) of percentage flow-mediated vasodilatation (FMD), percentage low-flow-mediated constriction (L-FMC), and of the composite end point in differentiating patients diagnosed with cardiovascular disease. The lower right panel presents a comparison of the three methods. L-FMC and FMD showed similar sensitivity and specificity for the best cut-off value and a similar area under the ROC curve. With the composite end point, the area under the ROC curve was significantly higher: p<0.05 composite end point versus L-FMC, p<0.005 versus FMD).

Data for each individual subject, along with the cut-off values of FMD and L-FMC that according to ROC analysis were associated with the best sensitivity and specificity, are presented in fig 4. As detailed in the figure caption, the addition of L-FMC substantially improved the accuracy of FMD in distinguishing patients with cardiovascular disease from age-matched controls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Graphic representation of the advantages of combining flow-mediated vasodilatation (FMD) and low-flow-mediated constriction (L-FMC) measurements. Each point in the graph represents one patient (triangles) or a healthy volunteer (circles). Cut-off values for normality, derived from receiver operating characteristic (ROC) curve analysis, are represented by the horizontal (L-FMC <−2.7%) and vertical (FMD ⩾3.8%) dashed lines. When studied with either of the two methods taken alone, endothelial function assessment results in a relatively high number of both false-positive and false-negative results. For instance, 22 of the subjects with cardiovascular disease had an FMD higher than 3.8% (right of the vertical dashed line). Similarly, 12 of the healthy volunteers had FMD values lower than this cut-off point. Use of L-FMC alone is associated with a similar proportion of false results. Combination of L-FMC and FMD results in a higher diagnostic accuracy. Furthermore, the graphic representation of individual results allows four different subsets of patients to be identified: those with normal L-FMC and FMD (“normal subjects”, lower right quadrant), those with both abnormal L-FMC and FMD (“endothelial dysfunction”, left upper quadrant), those with abnormal L-FMC and normal FMD (“blunted resting endothelial function”, upper right quadrant) and those with blunted FMD and normal L-FMC (“impaired endothelial reserve”, lower left quadrant). While the meaning of the two last cases will have to be studied further, concordance of L-FMC and FMD results (either normal or abnormal) was a very effective predictor of health or disease.

There was absolutely no correlation between FMD and L-FMC in any of the subgroups (CAD: R = 0.19, p>0.3; hypertension: R = 0.09, p>0.5; CHF: R = 0.03, p>0.8; healthy volunteers: R = 0.0, p = 0.98). Interestingly, while FMD was inversely correlated to resting radial artery diameter (R = -0.26, p<0.05), FMC showed no correlation with it (R = 0.03, p = 0.8, fig 5), and the composite end point showed a non-significant trend (R = −0.2, p>0.1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Correlation between flow-mediated vasodilatation (FMD), low-flow-mediated constriction (L-FMC) and resting arterial diameter in healthy volunteers. In agreement with previous evidence, FMD was inversely correlated with resting arterial diameter. In contrast, there was no correlation between L-FMC and resting diameter.

Discussion

FMD is a well-established technique for the measurement of endothelial function. Its strengths include simplicity, reproducibility and non-invasiveness, as well as the ability to predict an adverse prognosis in patients with coronary artery disease, hypertension and peripheral artery disease.2 15 However, the simple measurement of FMD may be an inadequate representation of endothelial function. For instance, FMD is a function of the hyperaemic blood flow pattern.16 17 Further, since it is expressed as a percentage increase over a baseline, FMD is strongly influenced by “baseline” arterial diameter.17 Interpretation of FMD is based on the assumption that these baseline data were obtained in a “true” resting condition that is the same for all subjects undergoing the measurement. Thus, a conceptual limitation of FMD is that while it quantifies how much the endothelium modifies its biosynthetic activity in response to reactive hyperaemia, it does not provide any information about the baseline activity of endothelial function. This latter parameter may or may not be in different states of activation, based on resting shear stress and other factors that can impact the state of endothelial “tone” (for instance, emotional status, use of drugs, genetic or environmental influences). As such, FMD measures endothelial “recruitability” or “stimulability” in response to an endothelium-specific stimulus (ie, a sudden increase in shear stress), but it does not measure “basal” endothelial function. Of note, many authors have reported that resting diameter tends to be larger in patients than in healthy volunteers,17 which might in certain cases reflect “basal” (pre-FMD) endothelial activation and obviously complicates FMD analysis.

It has been recently proposed that resting endothelial function, measured as a decrease in radial artery diameter induced by a local reduction in blood flow, can be assessed using magnetic resonance imaging.10 In a recent publication, we confirmed these data and proposed L-FMC as an index of endothelial function that is non-invasive, is simple and reproducible, and does not require further procedures as compared with “traditional” FMD.9 Of importance, L-FMC appears to be independent of any change in blood pressure and/or of autonomic stimuli, and previous research from our group and from Spieker et al shows that it allows assessment of the complex coordinate action of at least three (endothelial) mediators—that is, endothelin-1, endothelium-derived hyperpolarisation factor and a cyclo-oxygenase product.9 18 Thus, although L-FMC is less well established than FMD and further studies are necessary, a number of lines of evidence show that it can be effectively and reliably measured by at least two different methods, and that it gives unique information on parameters that are of interest for the study of endothelial function. Because it measures the vasoconstriction provoked by local decreases in blood flow, L-FMC is a measure of “resting” or “basal” shear-stress-dependent vascular tone. Therefore, the advantage of adding quantification of L-FMC to FMD measurements is that it provides information that does not overlap or substitute, and is complementary to “traditional” FMD. In this paper, we apply two different approaches to demonstrate the incremental nature of the information provided by L-FMC.

The exercise protocol serves to emphasise the complementary nature of L-FMC and FMD data in detecting endothelial dysfunction. McGowan et al recently reported an exercise-induced reduction of FMD, which they postulated as being evidence of endothelial dysfunction.19 While our data confirm that FMD is blunted immediately after exercise, we also show that exercise causes an increase in baseline blood flow and arterial diameter, and that these changes are associated with a larger L-FMC. In our hypothesis, rather than a toxic effect of exercise, the blunting in FMD seen here might simply be expression of the inability of a stimulated endothelium (as reflected by the higher resting diameter and L-FMC) to respond to further stimuli. Thus, we propose that, when observing a blunted FMD, one should be careful in taking this information as definitive evidence of “endothelial dysfunction”. While the measurement of glyceryl trinitrate-induced vasodilatation serves to effectively exclude endothelium-independent vasodilator dysfunction, our exercise data suggest that FMD is influenced also by resting endothelial function. Similar changes might occur in a number of other conditions—for instance, in the setting of active treatments or specific diseases, or for the effect of genetic variability. In principle, in the absence of data on the “basal” level of endothelial activity, one cannot a priori exclude the possibility that an observed blunted FMD results from “endothelial dysfunction” rather than from “basal” endothelial activation (ie, “lack of further recruitability”). Thus, the results of the exercise protocol serve to emphasise that interpretation of FMD data might be more complicated than normally believed. In this regard, besides providing information on a variable that might be of clinical relevance, the addition of L-FMC also serves to ensure that FMD measurements are performed in a “basal” or “normal” state of endothelial activation.

Of note, studies are also necessary to test whether the opposite condition may occur (ie, a condition or an intervention that blunts L-FMC and “artificially” increases FMD). Of importance, the existence of patients showing such “mixed results” (ie, blunted FMD and (supra)normal L-FMC or blunted L-FMC and (supra)normal FMD) is demonstrated in the second protocol presented here, and its importance should not be underestimated.

Data from the second protocol provide further evidence of the important role of L-FMC alone and help in interpreting FMD data. Ideally, FMD could be used as a large-scale test to detect (early) cardiovascular disease. These data suggest that integrated analysis of L-FMC and FMD might allow identification of a subgroup of subjects (those in whom FMD and L-FMC agree) in which the “diagnostic” error is limited as compared with L-FMC or FMD measurements alone (fig 5). Further, ROC curve analysis shows that the combination of L-FMC and FMD (“composite end point”) significantly improves the detection of patients who have been diagnosed with hypertension, congestive heart failure and coronary artery disease from healthy volunteers. Collectively, these data are a stimulus for further studies investigating the diagnostic and prognostic power of isolated “basal” and “recruitable” endothelial function, and of their combination, in larger prospective cohorts. Of note, since distinct biological mechanisms are involved in determining L-FMC and FMD, their simple mathematical combination in a combined end point may miss isolated abnormalities in one of the two. Although the patients were retrospectively enrolled in this study, which has to be acknowledged as a limitation, the concepts presented here encourage us to study whether L-FMC, combined with FMD, might have potential clinical applications. Future studies will have to test whether L-FMC and FMD measurements can provide clinically valuable information on endothelial and vascular pathophysiology.

In conclusion, the present data emphasise that the interpretation of forearm endothelial function might not be as straightforward as generally believed. The data from our exercise study underline the concept that endothelial “recruitability” data cannot be interpreted in the assumption that “resting” endothelial function does not vary (over time, between subjects or in response to external stimuli) and that it has no influence on FMD. Data from the second protocol confirm that this complex relationship between “resting” and “recruitable” endothelial function actually exists in the presence of disease and treatment, and suggest that L-FMC data may help in the detection of cardiovascular disease. The two observations presented collectively demonstrate that in studies enrolling patients, but probably also in a controlled experimental environment, one should not assume a priori that the endothelium is in a quiescent state at the time of FMD measurements. Use of L-FMC and the composite end point increases the spectrum of information gathered from the non-invasive measurement of endothelial function, and provides more insight into the complexity of its interpretation.