Healing of polytetrafluoroethylene arterial grafts is influenced by graft porosity Michael A. Golden, M D , a Stephen IC Hanson, P h D , b Thomas R. Kirkman, B A , a Peter A. Schneider, M D , b and Alexander W. Clowes, M D , a Seattle, Wash. and
La Jolla, Calif. The importance of porosity in synthetic arterial graft healing has not been adequately defined. To determine the effect of porosity on graft healing, we measured the extent of cellular response at late times in 4 mm internal diameter polytetrafluoroethylene grafts of varying porosity (between 10 and 90 ~m internodal distance) inserted into the arterial system of baboons. After 1 and 3 months the grafts were retrieved and examined for endothelial coverage, intimal thickening, and endothelial cell and smooth muscle cell proliferation. The pattern of intimal healing with endothelial cells and smooth muscle cells was only related to porosity in the sense that there was an abrupt switch in the pattern of healing as porosity was increased from 30 to 60 ~m. In low porosity grafts (10 and 30 ~tm internodal distances) endothelial coverage of the luminal surface was incomplete and, along with intimal thickening, was limited to graft near the anastomosis. In high porosity grafts (60 and 90 ~m internodal distances) luminal endothelial coverage was complete, and intimal thickening was uniformly distributed throughout the graft. The highest porosity graft studied (90 ~m) developed areas of focal loss of endothelial cells at late time periods. In this limited series there does appear to be an optimal porosity for polytetrafluoroethylene grafts near 60 ~m, since 10 and 30 ~m grafts fail to achieve luminal endothelial cell coverage, and 90 ~m grafts exhibit instability of the intima with focal endothelial cell loss. (J VAs¢ SuRG 1990;11:838-45.) The quest for a satisfactory synthetic substitute for small arteries continues since available grafts have a relative high incidence of thrombosis. An ideal synthetic vascular prosthesis should have a nonthrombogenic flow surface. It also should form only a small amount o f intimal thickening, and the intima that forms should be stable, with no tendency to late degeneration, endothelial loss, or progressive stenosis resulting in late thrombosis. The early studies ofWesolowski et al.1 supported the conclusion that porous synthetic grafts are incorporated better into the surrounding tissue than
impervious grafts. This fibrous ingrowth produces an intima of fibroblasts, fibrin, and blood-borne cells and, therefore, seems to improve the patency rates. However, Berger et al? noted that in humans porous Dacron grafts are largely covered by fibrin, and develop fibrous tissue coverage (presumably endothelial) only at the anastomoses. These observations were confirmed by Goldman et al. 3 and Stratton et al. 4 They demonstrated that Dacron grafts col); tinue to take up indium 1 l l - l a b e l e d platelets at late times after graft implantation. These results suggest that porous Dacron grafts continue to be thrombogenic and do not develop an endothelial luminal surface.
From the Department of Surgery' and the Regional Primate Research Center University of Washington School of Medicine, and the division of Cardiovascular Disease and Thrombosis,b Department of Basic and Clinical Research, Scripps Clinic and Research Foundation. Supported by grants HL 30946, HL 31950, HL 31469, and RR 00166 from the National Institutes of Health, United States Public Health Service, and a grant from The American Heart Association-Washington affdiate. Presented at the Fourth AnnualMeeting of The Western Vascular Society, Kauai Hawaii, Jan. 18-22, 1989. Reprint requests: AlexanderW. Clowes, MD, Department of Surgery, University of Washington, School of Medicine, RF-25, Seattle, WA 98195. 24/6/18047 838
In contrast, animal models o f graft healing provide support for the concept that spontaneous endothelial coverage can occur even though it has not been observed in man. In studies of Dacron grafts in a nonhuman primate model, Florey et al. s observed patches of endothelium along the grafts and suggested that the endothelium in these patches might be derived from capillaries invading the grafts. The phenomenon of transmural healing was also observed in dogs by Florian et al.6 and Kusaba et al. 7 Our laboratory confirmed this phenomenon in a baboon model with porous Dacron and polytetrafluoroethy-
Volume 11 Number 6 June 1990
lene (PTFE; 60 p~m internodal distance) grafts, and demonstrated that the source of the endothelial cells (ECs) is indeed transmural capillaries derived from the surrounding granulation tissue. 8'9 Initially it was not clear why porous Dacron grafts could develop an endothelial lining in baboons but not in man. A partial explanation was provided by the observation that endothelialization is usually patchy and often incomplete even at late times in the Dacron grafts. However, it should be noted that the 60 Ixm PTFE grafts complete the process of endothelialization at 2 weeks. A number of possibilities might account for the differences in healing patterns. The incompleteness of Dacron graft healing may be due to its increased thrombogenicity, the greater instability of its early luminal surface, the increased porosity of the Dacron, or to release of toxic product from the Dacron. The influence of porosity on graft healing, and i-"~~articular on PTFE graft healing, has been unclear. On the one hand, Campbell et al. 1° reported that porous PTFE grafts develop thrombosis more readily than less porous PTFE grafts. On the other hand, we have demonstrated a rather dramatic increase in the rate and extent of endothelial coverage in 60 txm PTFE grafts compared to 30 ~m grafts.9'11 Taken together, these observations support the hypothesis that there is an optimal porosity for PTFE vascular grafts. The purpose of this study was to determine if there is an optimal PTFE porosity in terms of late healing, with specific reference to endothelialization and intimal thickening. Our results suggest that 60 ~m internodal distance may be the optimal porosity for PTFE grafts. MATERIAL A N D M E T H O D S
Animals. Male baboons (Papio cyncephalus/anubis) at approximately 2 years of age were used in the study. Animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 80-23, revised 1985). Grafts. Expanded PTFE grafts with 4 mm (ID), and 10, 30, 60, and 90 Ixm internodal distances (unreinforced and not available for clinical use; W. L. Gore & Assoc., Inc., Elkton, Md.) were used in the graft healing studies. Graft healing studies Graft placement. Under general endotracheal anesthesia with halothane, 52 PTFE graft segments were placed in the aortoiliac circulation. Each animal
Porosity and polytetrafluoroethylene arterial graft healing 839 received two grafts. The grafts had porosities of 10, 30, 60, or 90 Ixm internodal distances. The graft segments measured 6 to 8 cm in length. Before systemic anticoagulation with 150 units/kg of heparin, the 60 and 90 Ixm porosity grafts were preclotted by means of standard techniques. Anastomoses were constructed in an end-to-side fashion with continuous 6-0 polypropylene suture (Davis & Geck, Wayne, N.J.). The proximal anastomosis was made to the distal aorta and the distal anastomosis to the proximal external iliac artery. The aorta distal to the proximal anastomosis was ligated, as was the distal common iliac artery, without disturbing the internal iliac artery. Morphology. Animals were killed at 1 and 3 months after graft placement. Tritiated thymidine (New England Nuclear, Inc., Billerica, Mass.) (0.5 mCi/kg body weight) was injected intramuscularly at 17, 9, and 1 hour before death. One half hour before fixation, heparin (3000 units) and Evans blue (50 mg/kg) were injected intravenously. The animals were killed with an anesthetic overdose, and their arteries were flushed with lactated Ringer's solution and then fixed by perfusion at 100 mm Hg, via an axillary artery cannula with 2.5% glutaraldehyde in 0.1 mol/L phosphate buffer, pH 7.4. The aortoiliac grafts and adjacent native arteries were excised, and further immersion fixed with 2.5% glutaraldehyde for at least 24 hours. None of the grafts were dilated. Both grafts were then extensively washed in 0.1 mol/L phosphate buffer, pH 7.4, and then 0.1 mol/L phosphate buffer with 0.1 mol/L glycine. One of each pair of iliac grafts were prepared for en face light microscopic autoradiography and scanning electron microscopy by bisecting the graft longitudinally and pinning it out on a Teflon sheet. It was dehydrated with graded alcohol solutions and dried by the critical-point method. Half of each specimen was mounted on studs and sputter coated with gold-paladium for examination of the luminal surface by scanning electron microscopy. The remaining half of the dried specimen was processed for autoradiography as described below. From these specimens it was possible to obtain the luminal surface morphologic characteristics of the grafts, the extent of intimal coverage of the luminal surface, and the endothelial thymidine labeling indexes. The other aortoiliac graft was sectioned along its entire length at 0.5 cm intervals. These samples were embedded in paraffin and sectioned perpendicularly to the long axis of the graft. Histologic sections were used for morphometric measurements of intimal cross-sectional area and smooth muscle cell (SMC)
Journal of VASCULAR SURGERY
Golden et al.
~///I////////////////////////////////////////////////~, I~ 1 month
o .o_ E
0 1.0 2.0 Proximal anastomosis Centimeters
3.8 Distal anastomosis
Fig. 1. Graph displays the extent of luminal endothelial coverage of the PTFE grafts at 1 and 3 months after implantation. The coverage is measured in centimeters +_ SEM from the heel of the anastomosis to the growing edge of endothelium for the 10 and 30 ~m porosity grafts, which derive their endothelium from the anastomoses. (n = 3 for all groups except n = 5 for 90 wm grafts at 3 months). Small focal defects in the endothelium were observed in the 90 p~m group at 3 months. autoradiography. Intimal areas from the cross sections along the length of each graft were obtained by means of a camera lucida and planimetric analysis. Autoradiography. Unstained, deparaffinized, histologic cross section slides were dipped in Kodak (Eastman Kodak Co., Rochester, N.Y.) NTB-2 emulsion, allowed to dry, and were kept in lighttight boxes for 14 days at 4 ° C. The slides were developed with Kodak D-19 developer and fixed. The sections were then stained with hematoxylin and evaluated at ( × 100) power under oil. Every nucleus with five or more overlying silver grains was considered labeled. The SMC thymidine labeling index was obtained by dividing the number of labeled nuclei by the total number of nuclei and multiplying by 100. One cross section was evaluated from each block. To assess the proliferation of the luminal surface cells, longitudinally bisected specimens were pinned flat onto Teflon sheets, dehydrated and critical point dried as described above. Dilute photographic emulsion (three parts Kodak NTB-2 to even parts water) was dripped onto the luminal surface and allowed to dry. The specimens were then processed as described above. These specimens were viewed under a light microscope at × 25 and the number of labeled cells per field recorded for the perianastomotic area on the graft side of the anastomosis. The total cell density was obtained from scanning electron micrographs of corresponding specimens not coated with emulsion.
Endothelial cell thymidine labeling indexes were determined for readings of at least 50 fields (five along the graft and 10 across). Intima from selected perfusion-fixed grafts was embedded in Epon, and thin sections were cut perpendicular to the luminal surface. Transmission electron microscopy was performed on these sections. Statistical analysis Nonpararnetric methods were used for all statistical testing. 12 Comparisons of labeling indexes and intimal area at 1 and 3 months was by Kruskal-Wall]s test and analysis of variance (ANOVA). The sign.~icance level for pairwise testing between porosity categories was controlled by Dunn's procedure. A Jonckhere-Terpstra test for ordered alternatives was applied to detect trends across levels o f graft porosity within each time period. Wilcoxon matched pairs analysis was used to compare graft endothelial or SMC proliferation to native artery levels. RESULTS Healing in aortoiliac grafts At 1 month the endothelial coverage as determined by the ability of the graft to exclude Evans blue was complete in the 60 and 90 p~m aortoiliac PTFE grafts; in the 10 and 30 wm grafts onty the ends were covered and the central part stained blue (Fig. 1). At 3 months one pair o f three pair of 30 ~m grafts was healed, and none o f the 10 ~m grafts
Volume 11 Number 6 June 1990
Porosity and polytetrafluoroethylene graB healing 841
Fig. 2. Scanning electron micrographs of a 30 Ixm porosity PTFE graft at 1 month shows (A) the endothelial lining of the healed area that excluded the Evan's blue dye, and (B) the endothelial growing edge (arrows). (Original magnification × 800, bar = 50 I~m.)
was healed. The pair of 30 ~m grafts that healed were particularly short, measuring 3.0 cm long from heel to heel. The 60 Ixm grafts remained fully covered whereas three of five 90 txm grafts contained multiple focal blue spots. Morphology. The presence of endothelium in regions not stained with Evans blue was confirmed by scanning electron microscopy. In the blue regions of the less porous grafts (10 and 30 txm) a carpet of microthrombi containing platelets, fibrin, and occasional blood-borne cells (possibly macrophages) was observed (Fig. 2). In the focal blue stained spots in the 3-month 90 ~m grafts, we found that the endothelium was missing and had been replaced by
thrombus. The absence ofendothetium and the presence of platelets were confirmed when these blue regions were studied in cross section by transmission electron microscopy (Fig. 3). Histologic cross sections displayed progressive cellular intimal thickening in all grafts in regions covered by endothelium. We have previously shown that most of these cells stain positive for alpha-smooth muscle action and therefore are of smooth muscle origin. 9,1a The 10 and 30 ~m PTFE grafts appeared to heal by ingrowth of endothelium and SMCs from the ends of the graft, and the 60 and 90 ~m grafts appeared to heal by transmural ingrowth of capillary cells. These different patterns of healing produced
Journal of VASCULAR SURGERY
Golden et al.
Fig. 3. Transmission electron micrographs of a 90 ~m porosity PTFE graft at 3 months shows (A) an endothelial cell E at the luminal surface of an intact area of intima, and (B) platelets (P) attached to the luminal surface of the intima in an area of endothelial cell loss. Beneath the platelets can be seen smooth muscle cells (S). (Original magnification (A) × 23,000, (B) ×44,000.) humps o f intima at the ends in the 10 and 30 p~m grafts, and uniform intimal thickening along the entire lengths o f the 60 and 90 p~m grafts. T o permit comparisons between the four porosity groups o f grafts, we measured the amount o f intimal area at 2 m m from the heel o f the anastomosis (Table I). As noted before, intimal thickening was present at 1 month and increased by three months. 9'iX Even though there were early but not significant differences, the area at 3 months was the same in all four types o f graft. There was no difference in the amount o f intimal area in the proximal and distal ends o f the grafts (data not shown). T o determine if there was
a correlation between intimal area and graft porosity, a trend test was applied. There appeared to be a weak correlation between increasing porosity and intimal area (1 month: 0.09 > p > 0.05; 3 month: p = 0.12) at both time points. A u t o r a d i o g r a p h y . Endothelial cell proliferation was greater for all graft porosities at 1 month than at 3 months, but even at 3 months did not decrease to the background levels o f native aorta (aortic endothelial thymidine index