INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE ENTRY of plasma-borne macromolecules such as low-density lipoprotein into the arterial wall has been implicated as a causative process in the long-term development of atherosclerosis (9, 41), and the mechanisms of molecular transport have been the subject of intensive experimental and theoretical study over the last 50 years. Much attention has focused on quantifying endothelial permeability (1, 6, 20, 30, 33, 39) and subsequent arterial distribution by potential diffusive and convective mechanisms (3, 8, 15, 24, 32, 37) for low-density lipoprotein and for more convenient tracer macromolecules such as serum albumin (2, 4, 5, 7, 10, 11, 13, 29, 32, 35) and horseradish peroxidase (16, 18, 25-28).

The arterial distribution of solutes depends on mechanisms of transport such as interstitial diffusion and convection, the latter arising from the transmural hydrostatic pressure gradient and the hydraulic conductivity of the arterial wall. Potential anatomic barriers such as the endothelium or adventitia impact the transport of solutes between the arterial wall and the lumen flow or extravascular capillaries in surrounding tissues. Despite decades of research, the importance of various mechanisms and structures that control transport remains elusive. For example, some have reported or assumed that 1) diffusion exclusively controls transmural transport (2, 5, 12, 14, 33), 2) diffusive mechanisms dominate only in healthy arteries but that convective forces become significant after endothelial injury and denudation (7, 23, 25, 29, 38), or 3) convection is always important (10, 19, 24, 35, 36). We previously examined the vascular transport of heparin, a potent inhibitor of vascular smooth muscle cell proliferation, and showed that convection was insignificant in mediating transport and distribution in the thin rat abdominal aorta (23). However, we also predicted that the significance of convective relative to diffusive forces should increase with the thickness of the artery. The introduction of vascular dimension as a governing parameter might help resolve the role of convective arterial transport.

The regulation of molecular transport is important to local vascular drug-delivery schemes, which administer drug directly to either the endovascular or perivascular aspect of the artery. The success of local release systems in transferring drug to the arterial wall depends on tailoring their designs to the mechanisms of drug distribution. For example, if convection is as or more important than diffusion, then the deposition following endovascular delivery will exceed that from perivascular delivery, because convection and diffusion are aligned in the former and opposed in the latter (Fig. 1) (23). On the other hand, if convective forces are insignificant, drug deposition would not be influenced by the aspect of delivery. Only recently have the quantitative concepts in the atherogenesis literature been applied to the local delivery of vasotherapeutic compounds (22, 23).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Schematic showing directions in which diffusion and convection act in transmural transport following endovascular and perivascular drug delivery. Diffusion always moves drug away from the point of release (shaded region), and convection, which arises from transmural hydrostatic pressure gradient, is always directed across arterial wall from inside to out. Shaded region represents heparin in either endovascular or perivascular compartment of in vitro perfusion apparatus or alternatively the hydrogel drug-release systems deployed on endovascular or perivascular surfaces of rabbit iliac arteries in vivo.

In this study, we examined heparin deposition under conditions that ranged successively from the highly idealized calf carotid artery in vitro to the intact rabbit iliac artery in vivo. We elucidated several mechanisms that impact arterial drug transport after local administration: 1) the competing effects of convection and diffusion, 2) the impact of the endothelium as a barrier to solutes and a modulator of convective flows, and 3) the loss of drug from the immediate arterial environment to the lumen flow or extra-arterial microvessels. Whereas the first mechanism is best studied in an in vitro preparation, the in vivo experiments allow us to explore the barrier function of the endothelium without uncertainty of their state. In addition, these in vivo studies enable us to compare endovascular and perivascular delivery in a setting where drug can be absorbed by nonarterial structures not present in our idealized in vitro perfusions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

In vitro perfusion. Calf carotid arteries were excised at a slaughterhouse, immediately placed in phosphate-buffered saline with 0.01 mM calcium and 0.1 mM magnesium (Sigma) at 4°C, and stored for no more than 3 h. The arteries were cleaned of excess fat and fascia, and ~1.5-cm long segments were cannulated at each end with polyethylene tubing (1.57 mm ID, 2.08 mm OD, Clay Adams). Just before cannulation, some arteries were denuded with three passes of an inflated 3-Fr embolectomy catheter (Baxter). After cannulation, the integrity of the artery was assessed by connecting one cannula to an elevated bag of Ringer solution, sealing the other cannula, and inspecting the artery for leaks under a dissecting microscope (23). Both cannulas were clamped to a rigid frame while the vessel was expanded under physiological pressure so that the inflated length was maintained throughout the perfusion. The artery was then placed in a perfusion apparatus as previously described (Fig. 2) (23). The arterial wall was used to separate two fluid compartments: endovascular and perivascular. Krebs-Henseleit buffer (Sigma) flowed from an elevated reservoir through the artery and a throttle valve and into a lower reservoir, where it was pumped back to the elevated reservoir thus forming the endovascular compartment. The artery was immersed in a perivascular bath, which was filled with the same buffer. The entire perfusion system was placed within a closed cabinet, which was maintained at 37°C and 100% relative humidity. The perfusion apparatus simulated plasma flow through the artery and allowed the application of any desired concentration of the drug to the endovascular or perivascular aspect and established a one-dimensional transmural concentration gradient (Fig. 1). The transmural hydrostatic pressure gradient imposed on the artery was adjustable between 0 and 100 cmH₂O. The endovascular volume was 100 ml for all perfusions, whereas the perivascular volume was 12 ml for perivascular and 100 ml for endovascular administration.

View larger version (28K): [in this window] [in a new window]   Fig. 2.   Schematic diagram of in vitro perfusion apparatus. Perfusate circulates between upper and lower reservoirs through a calf carotid artery immersed in a perivascular bath. An overflow line keeps hydrostatic head constant at the desired level regardless of pump speed. An outflow needle is used to visualize and estimate volume flow rate (23). Schematic diagram does not show a stir bar in the perivascular bath, a thermometer, a temperature-regulating water jacket, and a humidifier for the surrounding air.

[³H]heparin (0.12 µCi/ml, 0.17 ng/ml, NEN-DuPont, 6-20 kDa, mol mass 13.5 kDa) was applied to either the perivascular or endovascular compartments, and the artery was perfused for 1 h. This time was long enough for a substantial amount of drug to be deposited and yet was insufficient for heparin to fully penetrate the arterial wall, thus minimizing the confounding effects of clearance from the opposite surface of the blood vessel from the site of application (2, 3, 31). The volumetric flow rate of perfusate was maintained between 3 and 4 ml/min, which resulted in negligible heparin-transfer boundary layer resistances between the lumen flow and the arterial wall (23). The pressure loss along the length of the artery was undetectably small. Three 50-µl samples were taken from both compartments at the start and end of each perfusion experiment. After the perfusion, adsorbed drug was removed by either flushing 3 ml of fresh buffer through the lumen following endovascular delivery or dipping the artery once in clean buffer following perivascular delivery. The artery was cut into five segments, and the middle three segments were freeze-dried, weighed, solubilized with Soluene 350 (Packard), and prepared for measurement of deposited [³H]heparin through liquid scintillation spectrometry with Hionic Fluor (Packard). Histologic frozen sections were cut from the two end arterial segments and stained with Verhoeff's elastin stain. The average medial thickness was determined through computer-assisted morphometric analysis (23).

Perfusion experiments were performed with eight combinations of the following conditions: perivascular or endovascular administration of heparin, native or denuded arteries, and a transmural hydrostatic pressure gradient of 0 or 100 cmH₂O. Ten arteries were perfused under each of these eight conditions. Deposition in each vessel was measured three times in longitudinal segments. The deposition in each artery was taken to be the average of the three segments and is reported as the mass of drug in the artery normalized by both the dry mass of tissue and the applied concentration in either the perivascular or endovascular compartment. The latter was steady and the same for all experiments.

In vivo deposition measurements. Heparin deposition was compared 90 min after endovascular or perivascular administration to the rabbit iliac artery in vivo through hydrogel drug-delivery devices of similar composition and geometry (Fig. 1). This time duration was sufficient for quasi-steady-state transvascular heparin transport to be established in the 80-µm-thick rabbit iliac artery (23). Perivascular hydrogel-release devices were formed in molds and wrapped around isolated arteries, and endovascular devices were formed in situ. Hydrogels were formed by cross-linking a prepolymer solution using a photoreactive technique (17). The prepolymer consisted of a backbone of polyethylene glycol (3.3 kDa) with lactates on both ends (an average of 5 lactates per molecule) and capped with acrylate (Focal, Lexington, MA). This prepolymer was dissolved in 90 mM triethanolamine (30% wt/wt, Aldrich) to which N-vinylpyrrolidone (2 µl/ml, Aldrich) and [³H]heparin (20 µCi/ml, NEN-DuPont) were added.

For perivascular delivery, male New Zealand White rabbits (2.75-3.25 kg) were anesthetized with an intramuscular injection of ketamine (50 mg/kg) and xylazine (15 mg/kg). They were maintained with intravenous and intramuscular boluses of anesthetic as needed as well as ~10 ml · kg¹ · h¹ drip of lactated Ringer solution. The iliac arteries were exposed through a midline abdominal incision and displacement of the intestinal viscera, and arterial segments were isolated from the iliac bifurcation to the inguinal ligament. Estimated blood loss during arterial exposure and throughout the subsequent experiment was always <5 ml.

Perivascular hydrogel-release devices were made fresh before implantation. Eosin Y (20 µg/ml, Sigma) was added to the heparin-containing polymer solutions, and this mixture was injected into 70-µm thick planar glass molds where it was photopolymerized with an argon laser (488-514 nm, 70 mW/cm², American Laser). The resulting films were cut into ~7-mm wide strips, and two strips were folded over proximal and distal segments of each iliac artery. The abdomen was sutured closed to provide complete contact between the release device and the surrounding tissues and to prevent dehydration. The abdomen was reopened 90 min later, and just before each device was removed, the corresponding iliac artery was clamped with a hemostat next to the iliac bifurcation. This allowed for vessels to be removed without disrupting the flow to the contralateral artery, enabling drug delivery to persist for the duration of the experiment, and ensuring that the adventitial surfaces would not become contaminated with blood. In a separate experimental group, both left and right iliac arteries were balloon denuded with three passes of an inflated 3-Fr embolectomy catheter passed retrogradely from a femoral arteriotomy.

For endovascular heparin delivery, rabbits were anesthetized as described above and maintained on inhaled halothane (1-3% in O₂) anesthesia throughout the procedure. A specialized double-balloon hydrogel-delivery catheter (Focal) was inserted through a carotid arteriotomy and advanced to the iliac arteries under fluoroscopic guidance (34). Once in the iliac artery, the catheter was advanced so that both the proximal and distal balloons were beyond the iliac bifurcation. The endothelium was removed by inflation of the distal balloon and withdrawal of the catheter 35-mm toward the aortic bifurcation a total of three times. After denudation was completed, the catheter was repositioned, and the balloons were inflated, isolating a 25-mm segment of the artery within the deendothelialized zone. This vascular segment was subject to sequential 3-ml flushes of saline, initiator solution (eosin Y, 20 µg/ml), saline, and the liquid prepolymer solution. At the conclusion of the injection sequence, a fiber-optic element within the catheter delivered laser light (514 nm, American Laser) to the endoluminal surface, forming a 70-µm thick layer of heparin-containing hydrogel on the vessel wall. After the deposition procedure, the balloons were deflated, the catheter was withdrawn, and the procedure was repeated in the contralateral iliac artery. Endovascular hydrogels were formed on both iliac arteries within 5 min of each other. The abdomen was opened shortly before excision, and the iliac arteries were isolated and removed as described above.

Immediately after the tissue was excised in all drug-delivery experiments in vivo, plasma samples were collected to determine heparin concentration. The iliac arteries were cut into proximal and distal segments and stored at 80°C. At the time of processing, they were freeze-dried, weighed, solubilized with Soluene 350, and prepared for measurement of deposited [³H]heparin through liquid scintillation spectrometry with Hionic Fluor (Packard). The deposition is reported as the amount of drug normalized by both the dry mass of tissue and the initial concentration of the drug in the hydrogel-release device. For endovascular delivery, the heparin deposition in each artery was taken to be the average of the proximal and distal segments. Three rabbits were studied, each with an endovascular hydrogel device on each iliac artery, resulting in six independent data points. For perivascular delivery to either native or denuded blood vessels, four segments of iliac artery per rabbit were wrapped by their own drug-releasing hydrogel sheet. Three rabbits were studied in each group, and therefore there were 12 independent deposition measurements. In separate experiments, the average thickness of the endovascular hydrogel was determined repeatedly to be 70 µm, both initially and at 24 h, through microscopic analysis and computer-assisted morphometry of 100-µm thick frozen sections.

Statistics. All data are means ± SE. Heparin deposition for the various manipulations to the calf carotid artery and rabbit iliac artery was determined through an ANOVA using the Bonferroni-Dunn test, and data were deemed statistically significant when P < 0.05.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Figure 3 shows [³H]heparin deposition in calf carotid arteries following a 1-h perfusion in vitro for both native and denuded arteries, with transmural hydrostatic pressure gradients of 0 or 100 cmH₂O, and for endovascular (Fig. 3A) or perivascular (Fig. 3B) heparin administration. In the absence of a transmural hydrostatic pressure gradient, the deposition following endovascular delivery to denuded arteries was indistinguishable from native arteries. In the absence of a pressure gradient, heparin deposition within the arterial wall was indistinguishable following perivascular and endovascular delivery, whether examined in native or denuded arteries. The deposition with endovascular delivery was significantly increased by 85% from the addition of 100 cmH₂O pressure gradient, and after denudation this increase was 180%. In contrast, although statistically insignificant, the addition of this pressure gradient decreased the deposition following perivascular delivery by 17 and 13% in both native and denuded arteries, respectively. Perfusion experiments with a physiological pressure gradient showed that the deposition with endovascular application was 69 and 98% higher than with perivascular administration for native and denuded arteries, respectively. The average medial thickness of these calf carotid arteries was determined from histological sections from each end of each artery to be 415 ± 86 µm (average ± SD, n = 160 sections). There was no statistical difference in thickness between native and balloon-deendothelialized vessels.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   In vitro heparin deposition in calf carotid arteries 1 h after endovascular (A) and perivascular application (B). Arteries were either left intact (native) or balloon deendothelialized (denuded) and either subjected to a physiological or no transmural pressure gradient (P). Ten arteries were perfused under each condition. Each artery was divided into 3 axial segments, and the deposition in each segment was averaged to form 1 data point. Deposition is drug mass in artery normalized by both dry tissue mass of arterial segment and applied heparin concentration (n = 10 arterial segments, means ± SE).

Deposition of [³H]heparin in rabbit iliac arteries from an interfacially formed endovascular hydrogel was 7.7 times higher than that observed for the same release device placed as a perivascular wrap (Fig. 4). Deposition was statistically indistinguishable for perivascular delivery to both native and balloon-denuded arteries. This comparison could not be made with endovascular delivery, because removal of the endothelium was necessary for efficient formation of the interfacial hydrogel.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   In vivo heparin deposition in rabbit iliac arteries 90 min after either perivascular (PV) or endovascular (EV) release-device deployment. Both of these devices were 70-µm thick sheets made from same polymer matrix (see text). For perivascular delivery, both iliac arteries of 3 rabbits were balloon denuded, and both arteries of 3 rabbits were native. Four independent PV heparin-delivery devices were applied to segments of iliac arteries in each rabbit (n = 12 arterial segments, means ± SE). One independent EV heparin-delivery device was applied to each iliac artery in 3 rabbits, and proximal and distal samples from each artery were averaged to form 1 data point (n = 6 arterial segments, means ± SE). Deposition is normalized by both dry mass of artery and initial heparin concentration in hydrogel.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Evaluation of the extent and processes that govern the transvascular transport of macromolecules is central to our understanding of both the accumulation of atherogenic lipids and proteins in the vessel wall and the potential treatment of proliferative vascular diseases with exogenous vasotherapeutic compounds. These phenomena have been studied extensively in the context of the former but only recently have been applied to the latter (22, 23). In a previous study, we demonstrated that convective forces were insignificant at controlling heparin deposition in thin vessels with an intact endothelium (23). We did, however, also predict that the impact of convective forces on heparin deposition should increase with vessel thickness, and therefore the relative benefit of endovascular delivery of this drug should likewise increase. This paper sought to validate these predictions and directly compare heparin deposition following perivascular and endovascular delivery under circumstances that ranged from highly idealized in vitro perfusions to realistic clinically relevant in vivo conditions. This process allowed us to decipher the anatomic and physiological mechanisms that govern heparin deposition and distribution. With these insights, drug-delivery systems can now be rationally designed to work synergistically with the mechanisms of transport to more effectively treat vascular diseases.

Impact of convection. The in vitro perfusion experiments demonstrated that hydraulic convective mechanisms are significant determinants of transvascular heparin transport, because deposition increased by 85% following endovascular administration to native arteries when subjected to a hydrostatic pressure gradient of 100 cmH₂O (Fig. 3A). In this scenario, hydraulic convective flows are aligned with the concentration gradient and assist drug distribution (Fig. 1). These data support our earlier predictions that the ratio of convective forces to diffusive forces increases with medial thickness and with arterial injury (23). In that study we demonstrated that convection is an insignificant force at determining heparin distribution in the thin rat abdominal aorta. We expected that convection will be far more significant with soluble compounds like heparin in the 450-µm thick calf carotid and indeed in the 300-µm human coronary artery than in the 40-µm rat abdominal aorta. Experimental evidence that the balance between diffusive and convective forces may depend on arterial thickness dates back to 1962, when Duncan et al. (10) reported that increased blood pressure increased arterial uptake of albumin in the thick canine ascending aorta, but that this enhancement progressively diminished as one examined thinner arteries more distally.

Although with perivascular administration convective forces oppose diffusive forces and when present should lower deposition (Fig. 1), the measured 13 and 17% decreases were statistically insignificant (Fig. 3B). The short duration of these experiments allows examination of the impact of adding transmural pressure on deposition near the aspect of drug application. It should be noted that the calf carotid artery is thick relative to its diameter, and that although the transmural hydraulic volume flux is constant, the area normal to this flow increases from the intima to the adventitia. The convective hydraulic velocity is inversely proportional to the area normal to the flow, and therefore it is lower near the perivascular surface than near the intima by the ratio of the inner to outer arterial radii. In the calf carotid artery, the hydraulic velocity in the media near the adventitia might be up to 25% lower than near the intima, and thus convective forces are more likely to be observed with endovascular rather than perivascular application. In the in vivo experiments, the hydrogel-release devices encircled the artery and had an unknown but finite hydraulic conductivity, making assessment of convective forces impossible.

Endothelium modulates transport. The endothelial monolayer can potentially impact the distribution of any applied compound in two independent ways. First, the endothelium is a significant barrier to transmural hydraulic flux, and its disruption will increase convective forces. Convective currents within the artery are determined by the hydraulic conductivity of the arterial media and endothelial monolayer, and in a vessel the size of the calf carotid artery removal of the endothelium might let transmural hydraulic velocities increase by 50% (40). Indeed, in vitro deposition of endovascularly applied heparin in this artery with a transmural pressure gradient of 100 cmH₂O was enhanced following endothelial removal (Fig. 3A) precisely through this mechanism. Second, the endothelial monolayer may be a significant barrier that can directly limit the entry of solute into the blood vessel wall and/or slow the loss of applied drug out of the artery. In the absence of a pressure gradient, the deposition of endovascularly applied heparin after endothelial denudation did not increase (Fig. 3A). Therefore, the endothelium may not be a significant barrier to the transport of heparin in vitro. The possibility exists that the endothelium is not completely intact in vitro and that intercellular gaps may allow solutes to pass that might be restricted in vivo. However, as demonstrated above, the monolayer does modulate pressure-driven hydraulic currents, and therefore should be largely intact, exerting a large fraction of its normal resistance to solute transport.

In vivo experiments support the limited role of the endothelium as a barrier to heparin transport. With perivascular delivery, any potential endothelial resistance to transport would prevent the loss of drug to the lumen flow and would result in elevated deposition in native over-denuded arteries. Yet our data show that the deposition was not significantly different with and without the endothelium (Fig. 4), suggesting that the endothelial resistance to heparin transport is immeasurably low in vivo. Therefore, the in vitro and in vivo experiments together suggest that the endothelium modulates deposition by controlling hydraulic flows rather than as a direct anatomic barrier to heparin diffusion. One potential reconciliation for this seemingly incongruous statement may be that active transcytotic transport may negate the tight endothelial barrier to solutes but not to solvent.

Endovascular vs. perivascular delivery. The in vitro perfusion apparatus allowed us to apply drug equally to both aspects of the artery and to examine deposition without the complications of losses at the boundaries encountered in vivo. Without a transmural pressure gradient, the heparin deposition after perivascular application was not significantly different from endovascular delivery for both native and denuded arteries (Fig. 3). Thus the resistances to heparin flux at both boundaries are roughly equivalent. Because the endothelial resistance to heparin transport has been shown to be negligible in vitro and immeasurably small in vivo, the resistance of the adventitia must also be small. With the addition of the transmural hydrostatic pressure gradient to the calf carotid arteries, however, convective forces augmented the deposition with endovascular administration and tended to inhibit that with perivascular administration.

Heparin deposition following endovascular administration to the rabbit iliac artery from a 70-µm thick sheet of hydrogel was 7.7-fold higher than following perivascular delivery from an equivalent device (Fig. 4). The in vitro perfusion experiments with a physiological pressure gradient showed that the deposition with endovascular application was 69 and 98% higher than with perivascular administration for native and denuded arteries, respectively (Fig. 3). Thus the difference between endovascular and perivascular heparin deposition appears to be greater in vivo than in vitro. Because the in vitro perfusion experiments demonstrated that there are few anatomic barriers to heparin distribution, and because the hydrogels potentially dampen transmural convective flows, another mechanism is required to explain the larger difference in deposition in vivo. One possibility is that whereas the concentration of drug at the endovascular and perivascular surface of the artery was precisely controlled in vitro, heparin deployed from the hydrogel-release devices in vivo was lost to surrounding tissues. These losses to other structures explain the greater deposition in vitro than in vivo. Furthermore, these losses may differ between the endovascular and perivascular surfaces of the artery. For example, the endovascular hydrogel was subject to losses to rapid flow in the lumen, and alternatively interstitial fluids provided a low-resistance pathway away from the perivascular hydrogel. Indeed, the clearance from extra-arterial capillaries and lymphatics has been shown to be a substantial sink for perivascularly applied drugs (21). The in vivo data suggest that the losses from perivascular dilution and clearance outweighed the losses from the lumen flow, because heparin deposition with endovascular delivery exceeded that of perivascular administration (Fig. 4). The endovascular devices were formed directly adherent to the intima in vivo, and tightly adhering perivascular release devices might have negated some of this difference.

The in vivo studies also illustrate some potential distinctions in current endovascular and perivascular controlled-release technologies. Perivascular release devices that adhere to the adventitia are not currently available, and therefore, our in vivo data compare the state-of-the-art means of delivering drugs to both aspects of the artery. The increased loss of drug to the perivascular space reflects a low-resistance pathway that includes both the contributions of extra-arterial capillaries and loosely adherent hydrogels. Despite these seeming disadvantages, perivascular drug reservoirs have a theoretically limitless volume, whereas endovascular reservoirs are limited to a small fraction of the volume of the lumen. These are important considerations, because the size of the drug-delivery depot ultimately dictates the dose and duration of drug delivery. Although in the rabbit iliac experiments the deposition with endovascular delivery exceeded that with perivascular delivery, the plasma drug concentrations at the end of the experiment were over 20-fold lower in the latter than the former, and thus the effects of heparin are expected to be more confined to the local arterial environment in the latter.

In summary, heparin has been administered to the calf carotid artery in vitro, both in the absence and presence of physiological transmural hydrostatic pressure gradients and to the rabbit iliac artery in vivo. These experiments have demonstrated that in thick blood vessels, convection is potentially as important a mechanism of heparin distribution as diffusion. The endothelium and adventitia are not direct barriers to heparin diffusion, but the former does influence the magnitude of convective forces within the media. This has implications for local vascular drug delivery, because convective forces augment deposition from the endovascular aspect and inhibit deposition from the perivascular aspect. In addition, the deposition following heparin delivery from the endovascular aspect of the artery exceeded the deposition from the perivascular aspect, because the loss of drug to interstitial fluids outside the artery exceeded that to the lumen flow.