Introduction

With the development of the Fontan operation [1] and its subsequent modifications [2–4], more and more patients with single ventricles are surviving into adulthood. Despite improvements in early outcomes, there is time-dependent attrition and increasing cardiovascular morbidity [5]. The factors that contribute to the hemodynamic compromise are, however, still incompletely understood. The Fontan circulation is necessarily abnormal, with a chronically low cardiac output state [6], ventricular diastolic dysfunction [7], abnormal peripheral hemodynamics [8], and with the circulation described as afterload impaired and preload challenged [9]. Despite these data concentrating on the systemic circulation, the pulmonary vascular bed may be a key component in the overall circulatory disturbance. The absence of a subpulmonary pumping chamber imposes an elevated central venous pressure that is transmitted to the abdominal organs and peripheral venous and lymphatic systems, which is thought to contribute to chronic venous insufficiency in the lower extremity [10] as well as to more serious conditions such as liver cirrhosis [11], protein-losing enteropathy, [12] and plastic bronchitis [13].

The effect of these central hemodynamic abnormalities on arterial and venous properties has come under scrutiny. Increased carotid intima-media thickness (IMT) [14], peripheral endothelial dysfunction, reduced venous compliance, increased microvascular filtration pressures and threshold for edema, and higher arterial resistance in the lower extremity have been reported in post-Fontan adolescents [8, 15, 16].

Despite these physiologic observations, the peripheral arterial and venous wall morphology in patients that have adapted to the Fontan circulation is largely unknown. We have recently validated a noninvasive very high resolution ultrasound (VHRU) method to assess the combined intima-media, adventitia, and total wall thicknesses of superficial elastic and muscular arteries using 25–55 MHz transducers [17], and we have reported on its use in adolescents with congenital and acquired cardiovascular conditions [18, 19]. In this study, we validated the VHRU method for the assessment of peripheral venous wall layers. This allowed us to study the arterial and venous wall morphology in detail in conjunction with functional vascular testing in adolescents that have adapted to the Fontan circulation in comparison to an age-matched control group. The hypothesis was that the Fontan circulation is associated with remodeling of the peripheral arteries and veins.

Methods

The study was conducted in accordance with the Declaration of Helsinki, approved by the Institutional Research Ethics Board and Health Canada, and participation required written informed consent.

Study design and populations

This was a prospective cross-sectional study. Twenty-eight out of 31 consecutive Fontan patients aged 12–17 (14.8 ± 1.3) years attending the echocardiography laboratory at the Hospital for Sick Children, Toronto, between November 2009 and early June 2010 consented to participate in the study. None of the patients had a known chromosomal abnormality, diabetes or a lipid disorder requiring medication, or admitted smoking. One patient had the VACTERL association with a previously operated trachea-esophageal fistula. All of the patients had undergone completion of the Fontan procedure with a fenestrated extracardiac conduit. The fenestration had been closed with a device in 23 out of 28 patients, and spontaneous closure was confirmed by echocardiography in the rest. None of the Fontans were in significant heart failure, and they were all hemodynamically compensated and in relatively good clinical condition. Peak workload, peak VO2 (oxygen consumption), and percent of predicted peak VO2 and VO2 at the anaerobic threshold of the patient’s most recent clinical exercise tests were, in addition to the use of medication, retrospectively reviewed from hospital records. Twenty-one patients had undergone bicycle ergometry and four patients treadmill exercise tests using the Bruce protocol within the preceding 2 years (median 0.5, range 0–2 years). The characteristics are shown in Table 1.

Table 1 Clinical characteristics of controls and Fontan patients
Full size table

The arterial measurements were compared with data from a healthy age-matched control group of 54 subjects without medications who had been recruited from nearby schools or referred to the hospital for the assessment of an innocent murmur. In addition, 32 age-matched (14.8 ± 3.1 years) subjects with no abnormal venous hemodynamics or diastolic dysfunction on standard echocardiography were recruited as controls for peripheral venous imaging.

All measurements were obtained in the supine position after at least 30 min of rest in the echo laboratory at the Hospital for Sick Children, Toronto. Height was measured with a stadiometer, weight with an electronic balance, oxygen saturation with a pulse oximeter, and blood pressure (BP) with an oscillometric device (Dinamap, Critikon Inc.) during simultaneous echocardiographic and vascular imaging [20]. Reliability assessments of the different methods were made on healthy personnel.

Vascular morphology imaging

Ultrasound recordings from peripheral muscular arteries were obtained using the Vevo 770 ultrasound system (Visualsonics, Toronto, Canada) with mechanical 25 MHz (RMV710B), 35 MHz (RMV712), and 55 MHz (RMV708) linear transducers, and stored for off-line analysis. Care was taken not to compress the vessels during image acquisition.

The radial and ulnar arteries were studied 1–2 cm proximal to the skin fold that separates the palma manus from the anterior antebrachial region, the anterior tibial artery proximal to the first metatarsal bone, the dorsal tibial artery at the medial malleolar level, the common carotid artery 1 cm proximal to the carotid bulb, the brachial artery 2 cm proximal to the cubital skin fold, and the common femoral artery at the inguinal skin fold. The cubital vein was imaged at the cubital skin fold and the dorsal tibial vein at the medial malleolar level. The largest vein in the region was chosen. The 35 MHz transducer was used to determine the lumen dimension (LD) of the uncompressed vein and the 55 MHz transducer applied to study the venous far-wall morphology. Both right and left vessels were imaged.

Offline measurements of coded high-quality vascular images showing a good distinction between structural interfaces were performed manually by a single investigator (TS) blinded to the clinical data with electronic calipers in a very high-resolution zoom window (Vevo 770 software, version 2.3.0). The combined end-diastolic IMT and intima-media-adventitia thickness (IMAT) were measured from the far wall with the leading edge technique, and the end-diastolic and peak-systolic (carotid only) LDs were measured with reference to the simultaneously recorded electrocardiogram. Adventitia thickness (AT) was calculated as the difference between IMT and IMAT. We have recently reported arterial method validation, accuracy, and precision [17]. The intra- and interobserver and the test–retest CVs for arterial IMT were in the range 7–9% (N = 44–113) for 25 MHz, in the range 9–19% (N = 25–65) for 35 MHz, and in the range 11–16% (N = 43–114) for 55 MHz. The intra- and interobserver and the test–retest CVs for arterial IMAT were in the range 5–8% (N = 40–112) for 25 MHz, in the range 8–11% (N = 24–65) for 35 MHz, and in the range 14–15% (N = 41–112) for 55 MHz. The intra- and interobserver and the test–retest CVs for the calculated arterial AT were in the range 15–19% (N = 38–111) for 25 MHz, in the range 20–25% (N = 23–65) for 35 MHz, and in the range 20–22% (N = 39–112) for 55 MHz.

Venous imaging accuracy

Transcutaneous VHRU has not been validated to examine veins. For accuracy, wall measurements on 1–2 cm venous specimens of the inferior vena caval, internal jugular, subclavian, renal, and femoral veins were, therefore, obtained within 2 h post mortem from 6 pigs (weight about 35 kg) to obtain a range of different vessel structures and sizes. The specimens were tied on catheters, fixed in buffered (1:10) formalin for at least 24 h, and imaged in long and short axis sweeps with 55 MHz. A negligible shrinkage of 1% was noted for IMAT (N = 8). Gain settings were optimized to minimize the amount of scatter and to produce a sharp distinction between the different layers. Venous IMT and IMAT were defined as the distance between the lumen–vessel interface and the external border of the echolucent zone and the distance between the lumen–vessel interface and the external vessel wall border, respectively [17, 21, 22]. Each specimen was cut in cross-section, embedded in paraffin, sectioned, and stained with elastic (Verhoeff’s) trichrome. Anatomical orientation of the vessel was matched to the corresponding long and short axis ultrasound image, and the thickness of IMT and IMAT was measured at least 1 week after the ultrasound measurements using a Leica EZ4D stereomicroscope (Spectronic Analytical Instruments) with 20× or 30× magnification.

Arterial stiffness and endothelial function assessments

Carotid artery stiffness was calculated using the formula: stiffness index β = ln(SBP/DBP)/((LDs – LDd)/LDd), where SBP is systolic BP, DBP is diastolic BP, LDs is systolic lumen dimension, and LDd diastolic lumen dimension [23]. Abdominal aortic stiffness was assessed in a similar manner to that previously described [24], but with M-mode using the Vivid 7 ultrasound system and EchoPAC (GE Medical Systems, Horten, Norway). The intra- and interobserver and the test–retest CVs for the carotid stiffness index were 15, 17, and 22%, respectively (N = 27). The intra- and interobserver and the test–retest CVs for the end-diastolic abdominal aortic lumen diameter were 4% (N = 19), 4% (N = 19), and 3% (N = 9), respectively, and for aortic stiffness index 25, 26, and 37%, respectively (N = 10).

Pulse wave velocity (PWV) was assessed by sequentially recording ECG-gated right carotid and right radial artery and right carotid and right femoral artery waveforms with a high-fidelity micromanometer (SphygmoCor, AtCor Medical Systems Inc., Sydney, Australia) [25–27]. The intra- and interobserver and the test–retest CVs were 4% (N = 20), 14% (N = 10), and 14% (N = 10) for right carotid to right radial PWV, and 4% (N = 20), 17% (N = 10), and 9% (N = 10) for right carotid to right femoral PWV.

The flow-mediated dilatation (FMD) assessment was performed at the end of the study schedule following at least 4 h of fasting and more than 1 h of resting in the supine position [28]. FMD was assessed with the Vivid 7 system equipped with a 12 MHz transducer attached to a stereotactic probe-holding device using an automatic edge-detection algorithm (Vascular Tools, Medical Imaging Applications, Coralville, IA, USA) to acquire longitudinal, ECG-gated, end-diastolic images of the right brachial artery 3–5 cm above the antecubital fossa every 3 s. Images were recorded for 1 min baseline, during 5 min pressure cuff inflation to >200 mmHg around the forearm, and for 5 min after cuff deflation. We were able to obtain adequate imaging in 49/54 controls and 21/28 Fontans due to lack of cooperation or consent in some study subjects. The intra- and interobserver and the test–retest CVs for FMD were 26% (N = 19), 38% (N = 19), and 43% (N = 9).

Data analysis

Descriptive statistics are reported as mean ± SD or median (range) for continuous variables and as percentages for categorical variables. BMI z scores were derived based on Centre for Disease Control (CDC) growth charts [29]. The relationship between histology- and VHRU-derived venous measurements was assessed with the Pearson correlation coefficient and the degree of agreement was calculated [30]. The average of left and right vessel measurements is reported (no difference). Student and paired t tests were used as appropriate. Univariate and multiple linear regression analyses were performed, with diagnosis of Fontan, male gender, and use of medications entered as nominal (0 or 1) and all other measures as continuous variables (PASW Statistics 18, SPSS Inc., Chicago, IL, USA).

Results

Reliability and accuracy of venous layer thickness measurements

The venous IMT was identified by light microscopy in 40 of the 45 samples and by VHRU in 21 of the 45 samples (due to limited VHRU resolution at <0.05 mm), with a good correlation between histologically and 55 MHz-derived IMT (r = 0.87, p < 0.001, range 0.01–0.21 mm) (Fig. 1). Agreement analysis of these 21 samples showed a mean difference of 0.004 mm, equal to 4% of the mean (NS for bias), 95% limits of agreement −0.058 to 0.065, and CV 38%. The venous IMAT was identified by light microscopy and by VHRU in all 45 samples, with a good correlation between histology and 55 MHz (r = 0.84, p < 0.001, range 0.05–0.35 mm). Agreement analysis for these 45 samples with IMAT showed a mean difference of 0.030 mm, equal to 17% of the mean (NS for bias), 95% limits of agreement −0.067 to 0.133, and CV 24%. The intra- and interobserver and the test–retest CVs for venous IMT were 12 and 13, 12 and 15, and 26 and 27% with the 35 and 55 MHz probes, respectively (N = 39). The intra- and interobserver and the test–retest CVs for venous IMAT were 16 and 16, 17 and 12, and 20 and 25% with the 35 and 55 MHz probes, respectively (N = 44).

Fig. 1
figure 1

Histology versus VHRU for venous wall morphology. The figure shows the histology of a pig inferior caval vein specimen compared with the VHRU image obtained with the 55 MHz probe. The vein displays a pattern consistent with the echolucent muscular media (M) and the relatively thick and echodence adventitia (A) containing mainly connective tissue. Histological staining for muscle in red, elastin in black, and collagen in green. The scale is in millimeters (color figure online)

Full size image

Arterial lumen dimensions and wall layer thickness

The characteristics of the controls and Fontans are shown in Table 1. The right arm systolic BP was lower among Fontans compared with controls. The common carotid, brachial, radial, and femoral arterial LDs were reduced in Fontans. Carotid, brachial, radial, ulnar, femoral, and dorsal tibial artery IMTs were all thinner in Fontans (Table 2; Fig. 2). Thinner AT was observed in brachial, ulnar, and femoral arteries of Fontans.

Table 2 Arterial morphology and function in controls and Fontan patients
Full size table
Fig. 2
figure 2

In vivo VHRU in a Fontan patient and controls. In vivo transcutaneous VHRU of the common carotid artery (25 MHz), brachial artery (35 MHz), and dorsal tibial vein (55 MHz) in a 15-year old Fontan patient and control. The scale is in millimeters. The arterial walls (IMT and AT) are thin and the venous wall shows thickening of the echolucent media in the Fontan patient

Full size image

Regional systolic BP, BMI, age, and gender were all significantly and positively associated with most arterial site IMT and LD measurements (see the Electronic supplementary material, ESM, Tables 4a, 4b, and 4c). After adjustments, the diagnosis of Fontan remained significant for carotid LD, brachial LD and IMT, radial IMT, ulnar IMT, and femoral IMT and LD. Regional BP remained significant for carotid IMT, femoral AT, and dorsal tibial LD. BMI remained significant for carotid LD, brachial LD, radial LD, and anterior tibial LD. Age remained significant for brachial IMT and LD, ulnar LD, and femoral IMT and LD. Male gender remained significant for carotid LD, brachial IMT and LD, radial IMT and LD, ulnar IMT, and dorsal tibial IMT.

A subanalysis including Fontans only (N = 28) was performed regarding the long-term use of ACE inhibition or anticoagulation (N = 14 and 13, respectively), resting oxygen saturation, and time since Fontan completion. Use of ACE inhibitors was associated with a lower systolic (108 ± 9 vs. 98 ± 13 mmHg, p < 0.05) and diastolic (60 ± 7 vs. 51 ± 7 mmHg, p < 0.01) BP. We were unable to detect any significant associations between any of the arterial layer thicknesses or LDs and the use of ACE inhibitor, anticoagulation, resting oxygen saturation, or exercise test parameters. Time since Fontan completion was negatively associated with carotid (r = −0.36, p < 0.05) and radial (r = −0.46, p < 0.05) artery LDs, but not with other layer thickness or LDs, and these associations remained significant when adjusting for age, BMI, and gender with multiple regression analyses.

Arterial stiffness and endothelial function

No difference was observed in abdominal aortic and carotid artery stiffness indices and in carotid to femoral PWV between Fontans and controls (Table 2). The right carotid to right radial PWV was significantly higher in Fontans. This association remained significant when adjusting for age, BMI, gender, regional BP or carotid, and brachial or radial artery LD. Brachial artery IMT, AT, or LD was not associated with carotid to radial PWV. FMD and FMD time to peak dilatation were not different in Fontans (Table 2). No significant association between right carotid to right radial PWV or FMD and the use of ACE inhibitors, anticoagulation, resting oxygen saturation, peak VO2 or percent of predicted peak VO2, or time since Fontan completion was observed among Fontans.

Venous lumen dimensions and wall layer thickness

Venous LDs were not different between Fontans and controls. In contrast, both the upper arm cubital vein and lower leg dorsal tibial vein IMTs were increased in Fontans (Table 3; Fig. 2). The cubital vein IMT was thicker than the dorsal tibial vein IMT in both Fontans (p < 0.001) and controls (p < 0.001). BMI was positively associated with dorsal tibial vein IMT and IMAT. The differences in venous IMT and IMAT between the Fontans and controls remained significant when adjusting for age, BMI, and gender (ESM Table 4d).

Table 3 Venous lumen dimension and layer thickness in controls and Fontan patients
Full size table

In a subanalysis including Fontans only (N = 28), long-term use of ACE inhibition (N = 14) was associated with a decreased dorsal tibial vein IMAT (0.216 ± 0.039 vs. 0.256 ± 0.057 mm, p < 0.05), but no associations with other venous measurements were found. A similar decrease was observed regarding long-term use of anticoagulation (N = 13) for dorsal tibial vein IMAT (0.209 ± 0.032 vs. 0.259 ± 0.056 mm, p < 0.001) and cubital vein IMT (0.079 ± 0.016 vs. 0.096 ± 0.019 mm, p < 0.05), with no associations with other venous measurements. These remained significant when adjusting for age, BMI, and gender. No significant association between resting oxygen saturation, time since Fontan completion, or exercise test parameters and venous measurements was found.

Discussion

This is the first study to use VHRU to examine the peripheral vascular properties of Fontans. We show that VHRU is able to assess the venous wall morphology in almost microscopical detail. Indeed, the results of the current study show important and hitherto undescribed evidence of both arterial and venous remodeling as a consequence of the Fontan circulation. In short, Fontans displayed small and thin peripheral arteries as well as thick peripheral veins.

The key feature of the modern iterations of the Fontan circulation is the rerouting of the systemic venous return directly to the pulmonary arteries without a subpulmonary pumping chamber. This necessarily imposes major alterations on the central venous pressure and the flow dynamics in the pulmonary vascular bed; as a consequence, it is usually associated with reduced preload to the single ventricle. The venous IMT thickening observed in our Fontans is well in line with the general principles of the Fontan circulation and is likely related to the thickening of the muscular media layer in response to increased venous pressure and wall stress, as our previous and present validations of the IMT show that the intima thickness of the intima-media complex represents <5% in nonatherogenic vessels [17]. The difference between upper and lower extremity venous IMT is also consistent with the difference in the peripheral venous pressure related to the weight of the column of blood between the heart and the ankle, albeit exaggerated in Fontans. These results are in line with recent studies showing venous stiffening in adolescent Fontans [8], as well data documenting a high frequency of chronic venous insufficiency in Fontans during young adulthood [10].

Our finding that chronic ACE inhibition (and anticoagulation) modulated venous remodeling in the dorsal tibial vein (but no other venous or arterial sites) should be interpreted with caution. Although not designed to address the role of ACE inhibition, half of our Fontans received this therapy, primarily based on physician preference. This is in line with the frequency of use described in the cross-sectional study recently reported by the Pediatric Heart Network [31]. That data set showed no difference between those receiving and not receiving ACE inhibition, and overall we were unable to show any significant difference as a result of chronic ACE inhibition, perhaps suggesting that the arterial and venous changes are secondary phenomena, and not primary determinants of abnormal cardiovascular performance.

The consistently small and thin carotid, femoral, and upper arm peripheral conduit arteries are in contrast to previous studies showing an increased carotid artery IMT in Fontans [14], and are consistent with a decreased wall stress due to decreased volume and pressure loading of these vessels long term. The observation of a negative correlation between time since Fontan completion and LDs is also in line with the hypothesis. That is not to say that systemic vascular resistance is lower in Fontans, as it has been consistently shown to be elevated in previous studies [8, 9]. Instead, we suggest that the arterial remodeling we describe is secondary to decreased flow, and that the findings of increased resistance may therefore, at least in part, be related to a decreased total cross-sectional area of the vascular tree, consistent with our finding of a widespread and consistent reduction in arterial luminal diameter. Interestingly, these differences remained significant when adjusting for regional BP, body size, age and gender, suggesting that the arterial changes are indeed related to the chronically decreased volume load, flow, and cardiac output described in Fontans [6, 9].

The novel finding of increased vascular stiffness in the upper right arm, but not in the central aorta or carotid artery, of the Fontans is in contrast with recent vascular findings in tetralogy of Fallot [32], and perhaps counterintuitive given the thinner arterial wall. However, the Moens–Korteweg equation for deriving PWV relates PWV directly to vessel wall thickness as well as to the reciprocal of vessel radius. This suggests that the changes in vessel diameter predominate over those of the arterial wall, at least in terms of PWV, although this is difficult to prove in the absence of direct measurement of the physical properties.

Finally, we were unable to demonstrate any difference in right brachial FMD. This is in contrast to previous reports showing diminished FMD responses [15, 16]. The FMD in our controls is, however, similar to that previously reported for healthy teenagers [33], and while the response among our Fontans was slightly lower, it was not statistically significant. Furthermore, none of the vascular parameters correlated with exercise performance in our cohort. The lack of a difference in FMD may be due to differences in the Fontan completion strategies applied, patient age, our relatively small sample size, and the influence of technical variability or preceding BP measurements [34].

This study has several limitations. Even if the VHRU accuracy is good for wall layer thickness overall, the arterial adventitial and venous IMTs of small peripheral vessels are, nevertheless, close to the limit of ultrasound resolution (0.03–0.05 mm) and challenged by some technical variability. Furthermore, the study included relatively young Fontans in good clinical condition, without significant heart failure or signs of chronic venous insufficiency or other significant hemodynamic morbidities. Only retrospective, although recent, exercise test data were included. The low number of Fontans with gastrointestinal morbidities precludes any conclusions relating to the vascular findings. On the other hand, we included a sufficient amount of age-matched healthy controls to optimize the statistical power. Our study included relatively young Fontans operated with the fenestrated extracardiac tunnel only, and may, therefore, not apply to older Fontans completed with other techniques. In the majority of the Fontans, the most recent invasive catheterization was performed more than a decade ago, precluding analyses in relation to invasive pressure data. Therefore, these results may only be applied to contemporary hemodynamically stable young patients that have adapted to the Fontan circulation.

In conclusion, our results show significant changes in the morphology and size of the peripheral arteries and veins in adolescents that have adapted to the Fontan circulation. Further prospective studies are needed to clarify the role of these changes in the pathogenesis and the prediction of the development of significant hemodynamic morbidities in Fontans long term.