- Email: [email protected]
Patency in canine inferior vena cava grafting: Effects of graft material, size, and endothelial seeding
Patency in canine inferior vena cava grafting: Effects of graft material, size, and endothelial seeding M a l c o l m H e r r i n g , M.D., A u s t i n G a r d n e r , M.D., P a m e l a Peigh, M . D . , D a v i d M a d i s o n , M . D . , Sally B a u g h m a n , B.S., J o h n B r o w n , M.D., and J o h n Glover, M.D., Indianapolis, Ind. We studied 117 inferior vena cava (IVC) replacements in dogs to determine the effects of graft material, graft size, endothelial seeding, and cultured endothelial linings on graft patency. As a control, the IVC was removed and reimplanted in 11 dogs. Dacron (n = 7) and expanded polytetrafluoroethylene (e-PTFE) grafts (n = 12) were seeded immediately with the use of enzymatically derived autogenous jugular vein endothefium. Cultured linings were prepared for e-PTFE grafts (n = 9) by inoculating the graft with jugular endothelium and nurturing the lining in tissue culture for I 4 to 30 days before implantation. Unseeded grafts (n = 27) were prepared according to the manufacturer's recommendations. These six methods of preparation were tested in grafts measuring 6 m m I.D. and 60 m m in length. Other sizes were tested with a Latin square study design. After 30 to 60 days the grafts were perfusion fixed and studied with light and transmission electron microscopy. Patency was determined by contrast cavography after 7 and 30 days. Patency in the IVC reimplantation was 100% compared with 28.0% of the e-PTFE (p = 0.00i) and none of the Dacron grafts that measured 6 m m I.D. and 60 m m long. e-PTFE and Dacron graft patency also differed significantly (p = 0.035). Seeded and culture-lined e-PTFE grafts in that same size were patent in 31.6% compared with 16.7% of unseeded e-PTFE. With grafts measuring 80 m m long, three of the five e-PTFE grafts were patent between 3 and 7 days. All progressed to occlusion by 30 days and compared poorly with all other graft sizes tested (2.6% progression to occlusion [p = 3 × 10-s]). Recanalization was not seen in 10 occluded grafts that were followed for 60 days. The histologie features of seeded grafts differed remarkably from grafts previously studied in the arterial circulation and from culture-lined and unseeded venous prostheses in that 60% had prominent large, random, endothelium-lined channels within the inner capsule. Larger graft diameters (p = 0.009) and the omission of an endothelial surface treatment (p = 0.004) were associated with anastomotic subendothelial fibrous hyperplasia. We conclude that graft material is the major determinant of patency in IVC replacements, that an extensive endothelial surface promotes patency, but that simply seeding e-PTFE or Dacron grafts with l 0 s endothelial cells does not provide sufficient endothelium to alter early patency. Graft dimensions are also critical to long-term patency and the development of subendothelial fibrous hyperplasia. Graft recanalization is not a major feature in the healing of synthetic venous replacements. (J VAse SURG 1984; 1:877-887.)
Synthetic grafts are not used frequently in the venous system because o f high rates o f occlusion. 1 H o w e v e r , Richardson et al.2 were able to prevent occlusion o f Dacron venous grafts in dogs. The "grafts were spontaneously lined with endothelium From the Indiana University School of Medicine. Presented at the Thirty-eighth Annual Meeting of the Society for Vascular Surgery, Atlanta, Ga., June 7-8, 1984. Supported by the Indiana Affiliate of the American Heart Asso~- clarion, grant No. 5788146, and the American Heart Association, grant No. 83-755. Reprint requests: Malcolm Herring, M.D., 8402 Harcourt Rd., Suite 613, Indianapolis, IN 46260.
by implantation in the arterial system for 12 weeks prior to their use as venous grafts. The purpose o f this study was to determine the extent to which pa. tency o f venous grafts was determined by (1) type o f material, (2) length and diameter o f grafts, (3) immediate seeding o f endothelium, and (4) culture lining o f grafts with endothelium prior to implantation. MATERIAL AND METHODS One hundred seventeen mongrel dogs, weighing 15 to 25 kg, were studied. Each graft treatment was tested in grafts measuring 6 men I.D. and 60 m m in 877
878
Journal of VASCULAR SURGERY
Herring et al.
Table I. Patency of Dacron and e-PTFE grafts* Graft IVC e-PTFE e-FFFE e-PTFE Dacron Dacron
Treatment Reimplant Cultured Seeded Unseeded Seeded Unseeded
Patent
Table III. e-PTFE patency (30 days)
Occluded
6 3 3 1 0 0
0 5 7 5 3 10
Inner diameter of graft Graft length (ram) 80 60 40
6 mm 0/5 unseeded 3/10 seeded 5/7 cultured
10 mm
14 mm
6/6 seeded 2/6 cultured 2/3 unseeded
3/6 cultured 6/6 unseeded 3/6 seeded
* Six-millimeter I.D. and 60 mm long at 30 days.
Table II. e-PTFE patency (3-7 days) Inner diameter ofgraft Graft length (ram) 80 60 40
6 mm 3/5 unseeded 3/10 seeded 6/8 cultured
10 mm 6/6 seeded 3/6 cultured 2/3 unseeded
14 mm 3/6 cultured 6/6 unseeded 4/7 seeded
length (Table I). An extension of the size and graft preparation testing was made using a Latin square study design, summarized in Tables II and III. The graft diameters were compared with the host inferior vena cava (IVC) diameters by operative caliper measurement, and the residual graft diameters were determined by measuring the dye column on venacavograms performed 30 days postoperatively. Each dog was anesthetized with secobarbital. The infrarenal IVC was replaced aseptically with a 6 cm length of autogenous IVC (n = 11), seeded Dacron (Sauvage-knitted, Bard Implants Division, USCI, Billerica, Mass.) (n = 7), unseeded Dacron (n = 15), seeded expanded polytetrafluoroethylene (e-PTFE) (W. L. Gore & Associates, Flagstaff, Ariz.) (n = 12), culture-lined e-PTFE (n = 9), and unseeded e-PTFE (n = 12). Heparin (100 U / k g ) was given prior to implantation (Table I). Following implantation neither protamine nor further anticoagulant drugs were given. Knitted Dacron grafts were seeded by a technique described previously? Briefly, the external jugular veins were gently and aseptically removed. They were inverted on steel rods and placed in a 0.2% solution of crude collagenase (Worthington type II; Worthington Diagnostics, Freehold, N.J.) in phosphate-buffered saline (PBS) solution for 20 minutes at room temperature. The endothelium was irrigated from the vein wall with PBS, separated by centrifugation, rinsed, and resuspended in PBS. One third of this suspension was added to each of three aliquots of blood used to preclot the Dacron graft.
Finally the graft was rinsed with a heparin-blood mixture just prior to its implantation. The unseeded Dacron grafts were prepared in the same way except that the cell suspension was omitted. One half of the e-PTFE grafts were seeded. The seeding process was modified for e-PTFE because the material could not be seeded successfully by the method that we used for Dacron? Whole blood was introduced into the e-PTFE tube, which was the~ occluded at each end. The graft was squeezed between the fingertips until blood appeared on the outer surface. The blood was permitted to clot, and the excess was stripped from the tube. The cell suspension was introduced into the e-PTFE graft, which was again occluded at both ends and placed on its side. After 5 minutes, it was turned over for another 5 minutes. After inoculation, the excess solution was allowed to run out of the graft, and the graft was installed. The unseeded e-PTFE grafts were autoclaved and air dried just prior to implantation, according to the manufacturer's recommendations. Cultured linings were prepared on e-PTFE grafts. Three different methods were used to prepare the graft surface for cell culture. In each method the ~ material was autoclaved (10 minutes, 15 psi, 220 ° C, in &ionized water) prior to treatment. In one method the inner surface was coated with a c o i l s gen-rich gel (Difco Laboratories, Inc., Detroit, Mich.) (n = 2). In the second method the surface was soaked in plasma; the excess plasma was permit-' ted to drip out, and the residual plasma was converted to fibrin by using 5 units of thrombin (Parke-Davis, Detroit, Mich.) in the graft lumen (n = 5). In the third method the material was exposed to radio frequency plasma discharge in an argon atmosphere for 10 minutes, ~ and then treated" with thrombin-coagulated plasma and 10/xg/cm ~of plasma-derived fibronectin (Collaborative Research, Waltham, Mass.) (n = 1). Once the graft surface was prepared for culture, the lumen was inoculated with 106 autogenous endothelial cells that had been maintained in culture from 14 to 30 days. Tissu~ culture medium was prepared for canine cultures
880
Journal of VASCULAR SURGERY
Herring et al.
Fig. 1. Toluidine blue- stained section of unseeded graft (G) measuring I4 mm I.D. and 60 mm long and its anastomosis with IVC (V). Thickness (T) of subendothelial fibrous hyperplasia was measured at its thickest point along arrow. (Original magnification × 30.)
grafts and 12.27 + 2.19 mm in recipients with grafts that subsequently occluded. Surface preparation. All Dacron grafts failed regardless of surface preparation. Considering only grafts 6 mm I.D. and 60 mm long, cultured and seeded e-PTFE performed insignificantly better (31.6% patent) than unseeded e-PTFE (16.7% patent) (Table I). To determine whether any graft pretreatment method was superior for culture-lined grafts, all culture-lined grafts were considered, regardless of size. In the grafts of sizes other than 6 rnm I.D. and 60 mm long, 13 of 27 (48.2%) were patent, and no advantage was detectable with gel, plasma-thrombin, or plasma-thrombin-tibronectin pretreatments (p = 0.2). Clot-free surface area. The grossly clot-free surface area did not differ between the e-PTFE treatment groups. Seeded, culture-lined, and unseeded grafts demonstrated 45.45 + 33.26%, 46.15 + 25.55%, and 48.83 + 38.48% clot-free surface, respectively. Pannus ingrowth. The ingrowth of cellular pannus from each anastomosis of e-PTFE grafts
varied widely, ranging as high as 0.31 mm/day. It averaged 0.13 _+ 0.08 mm/day for culture-lined, 0.12 _ 0.06 ram/day for seeded, and 0.07 _+ 0.03 mm/day for unseeded grafts, but these differences were insignificant (p = 0.2). Inferior venacavography. Early venacavograms disclosed 38 patent grafts, nine stenotic grafts (of which six were IVC replants), and 43 occluded grafts. Four progressed from patent to occluded, four from patent to stenosed, and one from stenosed to occluded by 30 days. O f the 13 venacavograms performed at 60 days, 10 grafts had been previously occluded. All 10 remained occluded without any cavographic evidence of recanalization. Histology. We studied microscopic sections of the grafts that were patent when they were removed. Scattered midgraft endothelium was seen in 4 of 7 (57%) of unseeded e-PTFE grafts. Endothelium predominated (>50% of the surface) on the midgraft in 10 of 12 (83%) seeded e-PTFE grafts. Prelined grafts had predominant endothelium in 9 of 14 (64%). Endothelial inoculation by seeding or culture lining effected an endothelial lining about as
Volume 1 Number 6 November 1984
with medium 199 and Earles' balanced salt solution (M. A. Bioproducts, Walkersville, Md.), 25 mM HEPES buffer, 1% L-glutamine, 1% vitamin supplement (GIBCO, Madison, Wise.), penicillin and streptomycin, 100 U / m l (GIBCO), and 20% fetal calf serum (FCS). Cultures were maintained in 35 mm tissue culture plates (Costar, Cambridge, Mass.) in 5% CO2 and 95% O2 at 37 ° C. The medium was changed on the first day after inoculation and three times weekly thereafter. The grafts were placed in the tissue culture environment for 5 to 7 days prior to implantation. Just before implantation, an end segment o f each culture-lined graft was removed for histologic evaluation. An estimate of the fraction of the surface that was covered by cells was made. 6 Inferior venacavography was performed between 3 and 7 days postoperatively. It was repeated 30 days postoperatively whether or not the previous cavogram demonstrated patency. Thirteen dogs underwent an additional cavogram 60 days postoperatively in an attempt to identify recanalization, v The patent grafts were removed for study after 30 days. The animals were anesthetized, and the grafts were perfusion fixed with 2.5% glutaraldehyde. Graft occlusions were identified if the perfusate would not pass through the lumen under physiologic pressures or if there was no luminal opacification during venacavography. A photograph of each graft flow surface was made. A planimetric analysis of each photograph was used to determine the fraction of the surface area that was clot-free.6 Pannus ingrowth was measured from each anastomosis along the axis of the graft. Three to four measurements, separated by 2 mm each, were made at each anastomosis. Each patent graft underwent a histologic evaluation of the midgraft segment. Light microscopy allowed the analysis of the general patterns of healing. Transmission electron microscopy (TEM) was used in selected instances in which cell identification was in doubt. Scanning electron microscopy (SEM) was used to determine the nature of the flow surface itself. The grafts and adjacent tissues were categorized into four histologic levels for purposes of analysis: (1) The surface was defined as the material or cell layer in immediate contact with the flowing blood. (2) The inner capsule was defined as the tissue or material forming between the surface and the graft. (3) The interstitium was defined as the tissue or material that formed within the pores o f the graft material. (4) The outer capsule was defined as the tissue or material that formed between the outer margin of
Patency in canine inferior vena cava grafeing
879
the graft and the perigraft fat. Each level was graded histologically for the presence of endothelium, fibroblasts, smooth muscle cells, leukocytes, erythrocytes, fibrin-platelet clot, and other pericellular material. Cells were identified as fibroblasts if their shape ranged from generally triangular to spindle shaped and if they had rough endoplasmic reticulum (RER), dense nuclear material, and large amounts of extracellular matrix. Anastomotic subendothelial fibrous hyperplasia was analyzed by examining longitudinal histologic sections including the anastomoses. The distance between the graft and the flow surface was measured with an eyepiece micrometer at the point of maximum thickness (Fig, 1). The patency and surface treatment data were organized into contingency tables and subjected to chi-square analysis. The length, diameter, and surface treatments o f e-PTFE grafts were determined by Latin square analysis of variance,s The extent of pannus ingrowth, the clotfree surface areas, the preimplant estimates of cell coverage,9 the thickness of the subendothelial fibrous hyperplasia, and the inner capsule thicknesses were compared by one-way analysis of variance. The subendothelial fibrous hyperplasia was compared at the inflow and outflow anastomoses with the use of the paired t test. RESULTS Graft material. Six IVC reimplantations (Table I) remained patent. Five additional dogs died during the study period after IVC reimplantation. In each successful case the graft remained narrow (3 to 4 ram) throughout the study period. When compared with prostheses measuring 6 mm I.D. and 60 mm in length, the patency ofIVC reimplantations was better than the 28% patency experienced with e-PTFE (p = 0.001) and the universal failure of Dacron grafts (p = 1.3 x 10-5). The differences in patency between the e-PTFE and the Dacron grafts was also significant (p = 0.035). Graft size. The adverse effects of graft length were detected only in the narrow-diameter grafts. Of the five e-PTFE grafts measuring 6 mm I.D. and 80 mm in length, three remained patent at 3 to 7 days (Table II) only to occlude by 30 days (Table III). Only one of the 38 other grafts (2.6%) progressed to occlusion (p = 3 x 10-8). The diameter of the graft did not correlate significantly with patency. Nor was there a correlation o f the host vessel diameter with patency, since the caliper measurement of the host IVC diameter averaged 13.10---1.49 mm in recipients with patent
Volume 1 Number 6 November 1984
Patency in canine inferior vena cava grafting
881
Fig. 2. Toluidine blue- stained section of seeded e-PTFE graft (6 mm I.D. and 60 mm long). Note surface endothelium (arrow) and large endotheliurn-lined vascular channels (V). The surrounding cells (C) proved to be fibroblasts on transmission electron microscopy (Fig. 4). Note thickness of lining that forms on PTFE graft (G). (Original magnification x220.)
frequently as did the unseeded preparations of e-PTFE. The extent o f endothelial surfacing of culture-lined e-PTFE grafts at the time of implantaion was 18.5% of the surface area for patent and 13.3% for occluded grafts (p = 0.4). O f the nine implants that had an analysis o f surface coverage before implantation and that remained patent, five had histologically detectable endothelium preoperatively, and only one failed to demonstrate surface endothelium when it was removed. Four had no demonstrable endothelium preoperatively, but two demonstrated surface endothelium after 1 month in the IVC. Endothelium also lined vascular channels that formed in the inner linings of seeded grafts. The channels were present in 6 of i0 grafts (60%) that were patent when they were perfusion fixed. The channels did not communicate with other vascular channels in the outer linings of the grafts; they were often quite large (Fig. 2), as compared with the vasa
vasorum that we described in arterial grafts~°; they measured up to 100 ~m in diameter. These vascular channels were oriented in a seemingly random pattern. Fibroblasts, identified by electron microscopy, surrounded these channels and underpinned the surface endothelium (Fig. 3). The culture-lined grafts most often had only a monolayer o f endothelium directly adjacent to the e-PTFE grafts (Fig. 4). Anastomotic subendothelial fibrous hyperplasia was noted in all o f the e-PTFE grafts (Fig. 1). The lengths of file grafts did not correlate with the thicknesses of hyperplasia, but the hyperplasia was thicker at the outflow than the inflow anastomoses. The hyperplasia averaged 0.75 mm in thickness at the inflow and 0.99 mm at the outflow anastomoses (p = 0.045). Two other variables were associated with anastomotic hyperplasia, specifically the application o f endothelium to the surface (by seeding or culture lining) prior to implantation and the diameter of the graft.
Journal of VASCULAR SURGERY
882 Herring et al.
Fig. 3. Transmission electron micrograph of subendothelial inner lining of seeded e-PTFE graft (6 mm I.D. and 60 mm long). Connective tissue cells are characteristic of fibroblasts. Note relationship of IVC lumen (L) to developing vascular channel (V). A mast cell (M) is seen just beneath luminal endothelium (E), and endothelium (e) is seen lining vascular channel. (Original magnification × 6160.)
The hyperplasia at the outflow anastomoses averaged 0.921 + 0.920 mm for lined, 0.751 + 0.581 m m for seeded, and 2.684 + 1.279 mm for unseeded e-PTFE (p = 0.004); however, at the inflow anastomoses the differences were insignificant. The diameter o f the graft was an even greater determinant of the thickness of hyperplasia (Table IV). The thickness of inflow or outflow anastomotic hyperplasia may be interpolated for other graft diameters using Fig. 5. The hyperplasia usually bridged the anastomoses, but at least 75% of the hyperplasia was within the graft itself and not within the host vein (Fig. 1). Exclusions f r o m the study. Twenty-six dogs were excluded from the study because they died before the 30th postoperative day. One dog died o f an anesthetic problem at 30 days, before inferior venacavography could be completed. That dog's graft
was considered an early patency but was excluded from the late patency and histologic considerationg" The majority o f the deaths occurred in the first 72 hours after their implant from anesthetic or bleeding complications. Five (19.2%) related directly to technical difficulties with the IVC replantation. Considerable constriction and shortening o f the replanted IVC was observcd in every case (Fig. 6). ~ There was no discernible association of mortality rates with the type or preparation of the synthetic grafts. DISCUSSION Graft material was the most significant deter -; minant of patency in canine IVC replacement surgery. The IVC-reimplanted segments developed substantial and persistent narrowing that began with vasospasm when the scgment was resected. Despite
Volume 1 Number 6 November 1984
883
Patency in canine inferior vena eava grafting
Fig. 4. Toluidine blue-stained section of culture-lined e-PTFE graft (6 mm I.D. and 60 mm long). Note monolayer of endothelium (E) directly adherent to the e-PTFE (G). (Original magnification x 300.) the resultant flow restriction, the IVC grafts remained patent. Clearly, the autogenous endothclialized IVC is superior to either o f the synthetics that we tested. Dacron-knitted grafts failed uniformly, but the e-PTFE grafts represented a substantial improvement over Dacron. Dacron and e-PTFE grafts differ in their chemistry, the smoothness o f their flow surface, their porosity, and their method o f preparation. One can only speculate as to which of these features differentiates the per%rmance o f the two materials in the IVC. The size o f the graft is also important. Previous investigators have recommended that venous grafts be isodiametric, n but somewhat oversized grafts performed as well in this study. The best patency was accomplished with seeded, isodiametric grafts, all o f which remained patent. Although isodiametric and oversized grafts were patent more often than undersized e-PTFE grafts (64.7% vs. 36.4%), these differences were insignificant. Nevertheless, late occlusions were significantly more common in long, narrow-diameter grafts. This finding could not be adequately explained by the data from this study, but it suggests that certain combinations o f graft length and diameter are adverse to IVC patency. The diameter o f the graft had an even greater effect on anastomotic healing. The greater the di-
Table IV. Subendothelial anastomotic fibrous hyperplasia Diameter
Thickness of A S F H
Standard
(ram)
(ram)
deviation
No.
0.357 0.911 2.365*
0.267 0.454 1.175
10 6 9
0.355 0.896 1.347"
0.182 0.463 0.911
10 8 8
Outflow 6 10 14 Inflow 6 10 14
*Fourteen-millimeter diameter differs significantlyfrom 6 and 10 mm grafts (p < 0.01).
ameter, the greater was the development o f luminal obstruction by anastomotic subendothelial fibroplasia. Whether the graft diameter be large or small, the first step in this hyperplastic process involved the migration o f fibroblasts from the adjacent host IVC onto the para-anastomotic graft lumen. We postulate that the turbulence at each anastomosis was increased as the graft diameter increasingly differed from the host vessel. In this scenario, thrombin and platelets would have been recruited by anastomotic turbulence 12 to sensitize and to stimulate cell divi-
lournal of VASCULAR SURGERY
884 Herring et al.
.
OUTFLOW INFLOW
~ I
_.. o /
2
ul z
1
s
j
x
.
O
.
.
.
.
6 8 10 12 14 GRAFT DIAMETER (MM)
®
10
" ~i~>~ @OUTFLOW
<"
MM
"~
LM
~--'~
<"
<-
INFLOW
Fig. 5, A. Thickness of anastomotic subendothelial fibroplasia in e-PTFE grafts is plotted on the ordinate, and diameter (ram) is plotted on the abscissa. B, Location and average thickness of anastomotic subendothelial fibrous hyperplasia depicted schematically for e-PTFE grafts of different sizes. Note relative constriction of IVC to accommodate 6 mm I.D. grafts and, by contrast, constriction of 14 mm grafts to accommodate smaller IVC. This difference may displace anastomotic turbulence into host vessel in instance of 6 mm grafts, at which site endothelial inhibition of platelet release can occur.
sion in the migrating fibroblasts. Activation of the platelets by contact with the e-PTFE surface may partly explain the prominence o f hyperplasia at the outflow compared with the inflow anastomoses. Endothelial seeding failed to confer patency on Dacron and e-PTFE grafts. Like our experience with small artery replacements, 1~ an insufficient proportion of the total grafted flow surface is covered with endothelium after seeding to effect protection against early occlusion. Moreover, studies with rain platelets suggest that a newly seeded graft is more thrombogenic than an unseeded one.14 Since seeded arterial grafts demonstrated markedly improved
Fig. 6. Inferior venacavogram oflVC reimplantation after 30 days. Note narrowing of reimplanted segment, designated by arrows at anastomoses. Narrowing persisted from moment segment of IVC was removed prior to reimplantation.
thromboresistance within 24 hours, early patency might be achieved in the IVC grafts by the addition o f a suitable anticoagulant. The cutture-lined graft~" failed to achieve more than a nominal endothelial lining at the time of implantation. This problem was encountered with cultured e-PTFE arterial grafts as well. 13 Further technical development will be needed to achieve a satisfactory endothelial lining in tissue culture. " Even though endothelial seeding or culture had little effect on graft patency, it decreased the anastomotic subendothelial fibrous hyperplasia substantially. Seeded or cultured endothelial islands may not have inhibited thrombin and platelet effects on the e-PTFE surface during the first few hours of ~' flow, but they would be expected to hasten such inhibition within a few hours o f implantation. This inhibition by endothelium would be expected to shorten the exposure o f the migrating fibroblasts to circulating mitogens, thereby limiting hypertrophy.
Volume 1 Number 6 November1984
We conclude that graft material is the major determinant o f patency in I V C replacements, that an extensive endothelial surface p r o m o t e s patency, b u t that simply seeding e - P T F E or D a c r o n grafts with 10 s endothelial cells does n o t provide sufficient end o t h e l i u m to alter early patency. Graft dimensions are also critical to l o n g - t e r m p a t e n c y and the develo p m e n t o f subendothelial fibrous hyperplasia. Graft recanalization is n o t a m a j o r feature in the healing o f synthetic venous replacements. We gratefully acknowledge the support provided by the St. Vincent Hospital Foundation, St. Vincent Hospital and Health Care Center, Indianapolis, Ind.; W. L. Gore and Associates, Flagstaff, Ariz.; and Bard Implants Division, USCI, Billerica, Mass., and the technical assistance provided by Mr. Anthony Hubbard, Mr. Michael Arnold, and Mrs. Rebecca Compton.
REFERENCES
1. Hiratzka L, Wright C. Graft materials in the venous system: Recent advances. In: Wright C, Hiratzka L, et al, eds. Vascular grafting. Boston: John Wright, PSG, CO, 1983. 2. Richardson JV, Wright CB, Hiratzka LF. The role of endothelium in the patency of small venous substitutes. J Surg Res 1980; 28:556-62. 3. Burkel W, Ford l, Vinter D, Kahn R, Graham L, Stanley J. Endothelial seeding of enzymatically derived and cultured cells on prosthetic grafts. In: Stanley J, Burkel W, et al, eds. Biologic and synthetic vascular prostheses. New York: Grune & Stratton, Inc, 1982:631-51.
DISCUSSION Dr. H e n r y Haimovici (New York, N.Y.). I enjoyed Dr. Herring's experimental models for inferior vena cava (IVC) replacements. His study demonstrates again the difficulties o f achieving consistent graft patency in such replacements. Seeding of these synthetic grafts altered very little the rates o f thrombosis, thus indicating that a neointimal lining as an initial feature is not sufficient to "overcome the thrombogenic factors. Some years ago my associates, Drs. Steinman and A1pert, and I carried out an experimental study on the use of a temporary arteriovenous (AV) fistula on long-term patency o f Dacron, Teflon, and autogenous vein grafts. These were all used as bypass implants around the inferior v e n a cava (Arch Surg i966; 93:747). The results were assessed periodically by serial venograms, done immediately, 3 to 8 days, and 4 to 6 months postoperatively, as well as by the anatomic appearance of the grafts. The venograms o f the synthetic grafts taken 3 to 8 days after implantation disclosed in about half o f them some degree of partial thrombosis together with collateral
latency in canine inferior vena cava grafCing
885
4. Herring MB, Gardner A, Glover J. Seeding endothefium onto canine arterial prostheses: The effects of graft design. Arch Surg i979; 14:679-82. 5. Hahn AW, Yasuda HK, James WJ, Nichols MF, Sadhir RK, Sharma AK, Pringle OA, York DH, Chaxlson EJ. Glow discharge polymers as coatings for implanted devices. Biomed Sci Instmm 1981; 17:109-13. 6. Herring MB, Evans D, Baughman S, Glover J. The quantiration of cultured cellular surface coverage: Applications for transparent and opaque surfaces. J Biomed Mater Res 1984; 18:567-76, 7. Hiratzka L, Wright C. Experimental and clinical results of grafts in the venous system; a current review. J Surg Res 1978; 25:542-61. 8. Cochran WG, Cox GM. Experimental designs, ed. 2. New York: John Wiley & Sons, Inc, 1957: 95-114. 9. Colton T. Statistics in medicine. Boston: Little, Brown & CO, i974. 10. Herring MB, Gardner AL, Clover J. A single-staged technique for seeding vascular grafts with autogenous endothelium. Surgery 1978; 84:498-504. 11. ScherckIP, Kerstein MD, Stansel Jr HC. The current status of vena caval replacement. Surgery 1974; 76:209-33. 12. Turitto VT, Baumgarmer HR. Inhibited platelet adhesions and irreversible thrombus formation under high shear conditions. Trans Am Soc Artif Intern Organs 1978; 24:71926. 13. Herring MB, Hubbard A, Baughman S, Smith D, Miller B, Dilley R, Gardner A, Glover J. Endothelium-lined small artery prostheses: A preliminary report. J Am Soc Artif Intern Organs 1983; 6:93-i02. 14. Allen BT, Long JA, Welch MI, Hopkins KT, Sicard CA, Clark RE. Effect of aspirin therapy and its withdrawal on control and endothelial cell seeded grafts. Surg Forum 1983; 34:470-2.
venous channels. Subsequently, the lumen regained a normal diameter concomitant with the disappearance of the collateral circulation. After discontinuation of the AV fistula (22 to 40 days postoperatively), an 83% patency rate of the synthetic grafts was achieved. O f the autogenous veins, one thrombosed at i3 days but completely recanalized later. All vein grafts remained patent. The temporary AV fistula, through increasing blood pressure and rapid blood flow, provided the hemodynamic factors necessary for the initial critical phase for the healing of the grafts and their resulting patency. In addition we found in most instances a perigraft fibrous reaction with resulting rigidity of the graft that may also have played a role in enhancing patency. Thus I would like to ask the authors whether they had noted a perigraft fibrosis and whether there was any correlation with patency o f their grafts. Dr. James C. Stanley (Ann Arbor, Mich.). This is a very complex study, and I found it dif~cult to interpret some of the data. The difficulty arose not so much with the
886
Herring et aL
authors' presentation but with the number o f variables in some of the study's subgroups. It may be important to note that 26 of the 117 experimental subjects were excluded because they died within the first 30 days, the majority occurring within the first 72 hours. If the attrition o f any of these animals could be related to the thrombogenicity of their caval grafts, then they might better have been included in the survival data when analyzed. The importance of endothelium in maintaining an inferior vena caval prosthesis has been clearly established by the earlier work of Wright and his colleagues, and it should be noted that the best results in the current report were found among seeded isodiametric grafts. It is also important to realize that incomplete endothelial coverage provides little reliable benefit regarding patency, a fact not too surprising given the low-flow situation o f the venous circulation. The Michigan group has had considerable experience in the past with cultivating endothelial cells in grafts prior to their implantation, and it may be worthwhile to note that our patency rates in the arterial circulation o f grafts prepared in this manner were considerably less than when grafts were seeded with the use o f a preclot method. Certainly, endothelial cells adherent to grafts in the culture environment do not appear to have the same tenacity or viability as those trapped in preclot blood. It is not surprising that caval grafts with cultivated endothelial cell linings frequently failed. Finally, I am not sure I would agree with the concept espoused by a number o f other investigators that endothelial cell seeding actually increases early graft thrombogenicity as documented by increased platelet accumulation following the graft implantation. Nevertheless, it is obvious that endothelial cell seeding affords no immediate benefit as to antithrombogenicity in either the venous or arterial circulations. This leads to the obvious question whether it would not be appropriate to intervene pharmacologically with either anticoagulants or antiplatelet agents, or perhaps even with the creation of an AV fistula in the case o f small venous conduits, in an effort to mainrain early patency of venous conduits. We may be somewhat naive to believe that endothelial cells are going to ensure graft patency before they have a chance to proliferate and evolve as a confluent endothelial surface. Dr. Eugene F. Bernstein (La Jolla, Calif.). I would like to congratulate Dr. Herring; he has tackled a very difficult job in attempting to replace the IVC, probably the most difficult large vessel in the body to replace. I think it is also clear that when one looks at the data he presented that endothelial cell seeding does not appear to be the best current answer to this problem. Finally, Drs. Plate and Eldof pointed out earlier how important this problem is, since the American approach to the treatment of iliofemoral venous thrombosis is anticoagulation and some o f the patients will end up with clinical problems related to IVC obstruction. Our approach to this challenge was a little different. With my associates; Drs. Chan and Barden, a combination
Journal of VASCULAR SURGERY
o f PTFE with external stents and AV fistulas was used to bypass the occluded IVC. We used a long graft because we believed that the problem frequently involved both the lilac vein and the vena cava, and both should be bypassed. The experimental model included a ligature o f the cava and either an endto-end or end-to-side anastomosis of a stented PTFE graft with an AV fistula o f the femoral vessels in the ipsilateral groin. In a control group of experiments in which bypasses were done without an AV fistula, the grafts all thrombosed. But in the presence o f AV fistula, the patency rate by venography at 1 and 8 weeks was 83%, and then following takedown of the fistula at 25 weeks, it was 75%. What seems clearly demonstrated is the virtue o f stenting in the venous system, as well as the benefit of an adjunctive AV fistula. I would like to ask Dr. Herring whether he has considered these adjuncts in addition to an endothelial cell lining? Perhaps the combination of all these technique~ will help solve this difficult problem. Dr. Creighton B. Wright (Cincinnati, Ohio). Dr. Robert Hobson and I started some 10 years ago with venous grafting in the femoral system of dogs and performed a total of 172 o f these types of grafts. We were able to achieve substantial success with autogenous vein grafts in the venous system. A totally healed prosthesis from a dog's descending thoracic aorta could be satisfactorily replanted into the venous circulation with a high degree ofpatency, and thus I thought that Dr. Herring's series might be even better with endothelial seeding. Since he did not achieve total endothelialization of his grafts, the lack o f seeding might account for some of the diminution in patency that he reported. I would like to comment on the spiral vein technique that Ray Chin and our group studied experimentally, and Drs. Doty, Baker, and others of us have used clinically for superior vena cava bypass and replacement of segments of large peripheral veins. This technique is well described and useful clinically. (Wright CB, Doty DB. Spiral veto grafting: The technique. In: Bergan JJ, Yao JST, eds. Operative techniques in vascular surgery. New York: Grune & Stratton, Inc, 1980:307-10.) Dr. Hobson and I were able to improve patency in segmental venous replacement with a number of unfavorable conduits by the inclusion o f either small or large AV fistulas. At the present time there is not an individual ideal venous replacement graft other than autogenous, and I think Dr. Herring is on the right track with a fully endothelialized graft. I hope he will share with us his additional thoughts about adjuvants such as ibuprofen (Motrin), aspirin, and dipyridamole (Persantine) or other antiplatelet drugs and AV fistulas as a means of getting these less than ideal grafts to endothelialize fully in situ in the canine inferior vena cava or other veins. Dr. Peter Gloviczki (Rochester, Minn.). Similar ex-
Volume 1 Number 6 November1984
periments have been carried out in our laboratory at the Mayo Clinic. We implanted externally supported expanded polytetrafluoroethylene grafts into the inferior vena cava o f dogs and obtained patency rates similar to those presented by Dr. Herring. We had 25% patency at 1 week, i month, and 3 months, as confirmed by venography. In contrast to their study, however, all of our grafts were isodiametric, with an internal diameter of 10 mm. If we used a femoral AV fistula, as Dr. Haimovici also suggested, or an AV fistula plus antiplatelet treatment, our patency rate significantly improved to 75% at 3 months. We also analyzed the cross-sectional area of the harvested specimens at 3 months and found a decrease of luminal diameter to 59%, which suggests that thrombus formation in the grafts is significant; therefore a largediameter graft may be beneficial. Also, the addition of a temporary AV fistula appears to improve patency. When we implanted spiral vein grafts, our patency rate was further improved. In fact, 6-month patency of spiral "Tein grafts was 100% when antiplatelet agents and a temporary AV fistula were used. Since autogenous vein o f adequate size is rarely available, we think that spiral vein grafts should be considered for replacement o f large veins before using prosthetic material. Dr. Herring (closing). I would like to thank all the discussants for their provocative questions. Dr. Haimovici has asked about AV fistulas and perigraft fibrosis. The improved patency that Dr. Haimiovici noted with perigraft fibrosis may relate to the apparent benefit of graft rigidity in the venous system. We have seen improved patency in the inferior vena cava with externally supported PTFE grafts in other experiments, but we have not quantitated perigraft fibrosis sufficiently to determine whether it has an effect. Dr. Stanley asked about the number of variables in the experiments that we reported, and I think the discussants
Patency in canine i@rior vena cava grafting
887
have pointed out for me that there are many variables influencing the success of venous replacement grafting. We began naively thinking that the technique of endothelial seeding might prove a panacea for synthetic venous replacement. When our early seeding experiments failed to confer patency, we planned additional experiments. Because of the number of variables involved and the cost of studying each variable individually, we decided to use a Latin square study design. We did not feel that the animal deaths related to the graft surface nor the patency of graft, rather to anesthetic and bleeding problems. Most of the deaths related to problems with IVC reimplantation. Because of the spasm that developed in the reimplanted segment, there was a great discrepancy in size between the implant and undisturbed segments. We are uncertain as to why the cultured cells failed to maintain graft patency. The cells may have been lost from the surface of the graft because of mechanical effects or perhaps because o f the complement-mediated effects. Apparently, the granulocytic response in the venous system is substantially greater than that in the arterial system. Therefore complement-mediated effects may be extremely important in venous grafting. Other pharmacologic strategies might be directed toward complement functions and used in addition to the platelet inhibitory drugs that Dr. Wright mentioned. The question of an AV fistula is an excellent one. We elected not to introduce yet another variable into our study design, particularly since fistula flow would be difficult to control. The consensus that I hear from Drs. Stanley, Haimovici, Bernstein, Wright, and Gloviczki is that perhaps by adding all of these techniques together, supplemental drugs, external wrapping o f the graft, endothelial seeding, and AV fistula, we would be able to achieve a much higher rate of prosthetic patency in the venous system.
Synthetic grafts are not used frequently in the venous system because o f high rates o f occlusion. 1 H o w e v e r , Richardson et al.2 were able to prevent occlusion o f Dacron venous grafts in dogs. The "grafts were spontaneously lined with endothelium From the Indiana University School of Medicine. Presented at the Thirty-eighth Annual Meeting of the Society for Vascular Surgery, Atlanta, Ga., June 7-8, 1984. Supported by the Indiana Affiliate of the American Heart Asso~- clarion, grant No. 5788146, and the American Heart Association, grant No. 83-755. Reprint requests: Malcolm Herring, M.D., 8402 Harcourt Rd., Suite 613, Indianapolis, IN 46260.
by implantation in the arterial system for 12 weeks prior to their use as venous grafts. The purpose o f this study was to determine the extent to which pa. tency o f venous grafts was determined by (1) type o f material, (2) length and diameter o f grafts, (3) immediate seeding o f endothelium, and (4) culture lining o f grafts with endothelium prior to implantation. MATERIAL AND METHODS One hundred seventeen mongrel dogs, weighing 15 to 25 kg, were studied. Each graft treatment was tested in grafts measuring 6 men I.D. and 60 m m in 877
878
Journal of VASCULAR SURGERY
Herring et al.
Table I. Patency of Dacron and e-PTFE grafts* Graft IVC e-PTFE e-FFFE e-PTFE Dacron Dacron
Treatment Reimplant Cultured Seeded Unseeded Seeded Unseeded
Patent
Table III. e-PTFE patency (30 days)
Occluded
6 3 3 1 0 0
0 5 7 5 3 10
Inner diameter of graft Graft length (ram) 80 60 40
6 mm 0/5 unseeded 3/10 seeded 5/7 cultured
10 mm
14 mm
6/6 seeded 2/6 cultured 2/3 unseeded
3/6 cultured 6/6 unseeded 3/6 seeded
* Six-millimeter I.D. and 60 mm long at 30 days.
Table II. e-PTFE patency (3-7 days) Inner diameter ofgraft Graft length (ram) 80 60 40
6 mm 3/5 unseeded 3/10 seeded 6/8 cultured
10 mm 6/6 seeded 3/6 cultured 2/3 unseeded
14 mm 3/6 cultured 6/6 unseeded 4/7 seeded
length (Table I). An extension of the size and graft preparation testing was made using a Latin square study design, summarized in Tables II and III. The graft diameters were compared with the host inferior vena cava (IVC) diameters by operative caliper measurement, and the residual graft diameters were determined by measuring the dye column on venacavograms performed 30 days postoperatively. Each dog was anesthetized with secobarbital. The infrarenal IVC was replaced aseptically with a 6 cm length of autogenous IVC (n = 11), seeded Dacron (Sauvage-knitted, Bard Implants Division, USCI, Billerica, Mass.) (n = 7), unseeded Dacron (n = 15), seeded expanded polytetrafluoroethylene (e-PTFE) (W. L. Gore & Associates, Flagstaff, Ariz.) (n = 12), culture-lined e-PTFE (n = 9), and unseeded e-PTFE (n = 12). Heparin (100 U / k g ) was given prior to implantation (Table I). Following implantation neither protamine nor further anticoagulant drugs were given. Knitted Dacron grafts were seeded by a technique described previously? Briefly, the external jugular veins were gently and aseptically removed. They were inverted on steel rods and placed in a 0.2% solution of crude collagenase (Worthington type II; Worthington Diagnostics, Freehold, N.J.) in phosphate-buffered saline (PBS) solution for 20 minutes at room temperature. The endothelium was irrigated from the vein wall with PBS, separated by centrifugation, rinsed, and resuspended in PBS. One third of this suspension was added to each of three aliquots of blood used to preclot the Dacron graft.
Finally the graft was rinsed with a heparin-blood mixture just prior to its implantation. The unseeded Dacron grafts were prepared in the same way except that the cell suspension was omitted. One half of the e-PTFE grafts were seeded. The seeding process was modified for e-PTFE because the material could not be seeded successfully by the method that we used for Dacron? Whole blood was introduced into the e-PTFE tube, which was the~ occluded at each end. The graft was squeezed between the fingertips until blood appeared on the outer surface. The blood was permitted to clot, and the excess was stripped from the tube. The cell suspension was introduced into the e-PTFE graft, which was again occluded at both ends and placed on its side. After 5 minutes, it was turned over for another 5 minutes. After inoculation, the excess solution was allowed to run out of the graft, and the graft was installed. The unseeded e-PTFE grafts were autoclaved and air dried just prior to implantation, according to the manufacturer's recommendations. Cultured linings were prepared on e-PTFE grafts. Three different methods were used to prepare the graft surface for cell culture. In each method the ~ material was autoclaved (10 minutes, 15 psi, 220 ° C, in &ionized water) prior to treatment. In one method the inner surface was coated with a c o i l s gen-rich gel (Difco Laboratories, Inc., Detroit, Mich.) (n = 2). In the second method the surface was soaked in plasma; the excess plasma was permit-' ted to drip out, and the residual plasma was converted to fibrin by using 5 units of thrombin (Parke-Davis, Detroit, Mich.) in the graft lumen (n = 5). In the third method the material was exposed to radio frequency plasma discharge in an argon atmosphere for 10 minutes, ~ and then treated" with thrombin-coagulated plasma and 10/xg/cm ~of plasma-derived fibronectin (Collaborative Research, Waltham, Mass.) (n = 1). Once the graft surface was prepared for culture, the lumen was inoculated with 106 autogenous endothelial cells that had been maintained in culture from 14 to 30 days. Tissu~ culture medium was prepared for canine cultures
880
Journal of VASCULAR SURGERY
Herring et al.
Fig. 1. Toluidine blue- stained section of unseeded graft (G) measuring I4 mm I.D. and 60 mm long and its anastomosis with IVC (V). Thickness (T) of subendothelial fibrous hyperplasia was measured at its thickest point along arrow. (Original magnification × 30.)
grafts and 12.27 + 2.19 mm in recipients with grafts that subsequently occluded. Surface preparation. All Dacron grafts failed regardless of surface preparation. Considering only grafts 6 mm I.D. and 60 mm long, cultured and seeded e-PTFE performed insignificantly better (31.6% patent) than unseeded e-PTFE (16.7% patent) (Table I). To determine whether any graft pretreatment method was superior for culture-lined grafts, all culture-lined grafts were considered, regardless of size. In the grafts of sizes other than 6 rnm I.D. and 60 mm long, 13 of 27 (48.2%) were patent, and no advantage was detectable with gel, plasma-thrombin, or plasma-thrombin-tibronectin pretreatments (p = 0.2). Clot-free surface area. The grossly clot-free surface area did not differ between the e-PTFE treatment groups. Seeded, culture-lined, and unseeded grafts demonstrated 45.45 + 33.26%, 46.15 + 25.55%, and 48.83 + 38.48% clot-free surface, respectively. Pannus ingrowth. The ingrowth of cellular pannus from each anastomosis of e-PTFE grafts
varied widely, ranging as high as 0.31 mm/day. It averaged 0.13 _+ 0.08 mm/day for culture-lined, 0.12 _ 0.06 ram/day for seeded, and 0.07 _+ 0.03 mm/day for unseeded grafts, but these differences were insignificant (p = 0.2). Inferior venacavography. Early venacavograms disclosed 38 patent grafts, nine stenotic grafts (of which six were IVC replants), and 43 occluded grafts. Four progressed from patent to occluded, four from patent to stenosed, and one from stenosed to occluded by 30 days. O f the 13 venacavograms performed at 60 days, 10 grafts had been previously occluded. All 10 remained occluded without any cavographic evidence of recanalization. Histology. We studied microscopic sections of the grafts that were patent when they were removed. Scattered midgraft endothelium was seen in 4 of 7 (57%) of unseeded e-PTFE grafts. Endothelium predominated (>50% of the surface) on the midgraft in 10 of 12 (83%) seeded e-PTFE grafts. Prelined grafts had predominant endothelium in 9 of 14 (64%). Endothelial inoculation by seeding or culture lining effected an endothelial lining about as
Volume 1 Number 6 November 1984
with medium 199 and Earles' balanced salt solution (M. A. Bioproducts, Walkersville, Md.), 25 mM HEPES buffer, 1% L-glutamine, 1% vitamin supplement (GIBCO, Madison, Wise.), penicillin and streptomycin, 100 U / m l (GIBCO), and 20% fetal calf serum (FCS). Cultures were maintained in 35 mm tissue culture plates (Costar, Cambridge, Mass.) in 5% CO2 and 95% O2 at 37 ° C. The medium was changed on the first day after inoculation and three times weekly thereafter. The grafts were placed in the tissue culture environment for 5 to 7 days prior to implantation. Just before implantation, an end segment o f each culture-lined graft was removed for histologic evaluation. An estimate of the fraction of the surface that was covered by cells was made. 6 Inferior venacavography was performed between 3 and 7 days postoperatively. It was repeated 30 days postoperatively whether or not the previous cavogram demonstrated patency. Thirteen dogs underwent an additional cavogram 60 days postoperatively in an attempt to identify recanalization, v The patent grafts were removed for study after 30 days. The animals were anesthetized, and the grafts were perfusion fixed with 2.5% glutaraldehyde. Graft occlusions were identified if the perfusate would not pass through the lumen under physiologic pressures or if there was no luminal opacification during venacavography. A photograph of each graft flow surface was made. A planimetric analysis of each photograph was used to determine the fraction of the surface area that was clot-free.6 Pannus ingrowth was measured from each anastomosis along the axis of the graft. Three to four measurements, separated by 2 mm each, were made at each anastomosis. Each patent graft underwent a histologic evaluation of the midgraft segment. Light microscopy allowed the analysis of the general patterns of healing. Transmission electron microscopy (TEM) was used in selected instances in which cell identification was in doubt. Scanning electron microscopy (SEM) was used to determine the nature of the flow surface itself. The grafts and adjacent tissues were categorized into four histologic levels for purposes of analysis: (1) The surface was defined as the material or cell layer in immediate contact with the flowing blood. (2) The inner capsule was defined as the tissue or material forming between the surface and the graft. (3) The interstitium was defined as the tissue or material that formed within the pores o f the graft material. (4) The outer capsule was defined as the tissue or material that formed between the outer margin of
Patency in canine inferior vena cava grafeing
879
the graft and the perigraft fat. Each level was graded histologically for the presence of endothelium, fibroblasts, smooth muscle cells, leukocytes, erythrocytes, fibrin-platelet clot, and other pericellular material. Cells were identified as fibroblasts if their shape ranged from generally triangular to spindle shaped and if they had rough endoplasmic reticulum (RER), dense nuclear material, and large amounts of extracellular matrix. Anastomotic subendothelial fibrous hyperplasia was analyzed by examining longitudinal histologic sections including the anastomoses. The distance between the graft and the flow surface was measured with an eyepiece micrometer at the point of maximum thickness (Fig, 1). The patency and surface treatment data were organized into contingency tables and subjected to chi-square analysis. The length, diameter, and surface treatments o f e-PTFE grafts were determined by Latin square analysis of variance,s The extent of pannus ingrowth, the clotfree surface areas, the preimplant estimates of cell coverage,9 the thickness of the subendothelial fibrous hyperplasia, and the inner capsule thicknesses were compared by one-way analysis of variance. The subendothelial fibrous hyperplasia was compared at the inflow and outflow anastomoses with the use of the paired t test. RESULTS Graft material. Six IVC reimplantations (Table I) remained patent. Five additional dogs died during the study period after IVC reimplantation. In each successful case the graft remained narrow (3 to 4 ram) throughout the study period. When compared with prostheses measuring 6 mm I.D. and 60 mm in length, the patency ofIVC reimplantations was better than the 28% patency experienced with e-PTFE (p = 0.001) and the universal failure of Dacron grafts (p = 1.3 x 10-5). The differences in patency between the e-PTFE and the Dacron grafts was also significant (p = 0.035). Graft size. The adverse effects of graft length were detected only in the narrow-diameter grafts. Of the five e-PTFE grafts measuring 6 mm I.D. and 80 mm in length, three remained patent at 3 to 7 days (Table II) only to occlude by 30 days (Table III). Only one of the 38 other grafts (2.6%) progressed to occlusion (p = 3 x 10-8). The diameter of the graft did not correlate significantly with patency. Nor was there a correlation o f the host vessel diameter with patency, since the caliper measurement of the host IVC diameter averaged 13.10---1.49 mm in recipients with patent
Volume 1 Number 6 November 1984
Patency in canine inferior vena cava grafting
881
Fig. 2. Toluidine blue- stained section of seeded e-PTFE graft (6 mm I.D. and 60 mm long). Note surface endothelium (arrow) and large endotheliurn-lined vascular channels (V). The surrounding cells (C) proved to be fibroblasts on transmission electron microscopy (Fig. 4). Note thickness of lining that forms on PTFE graft (G). (Original magnification x220.)
frequently as did the unseeded preparations of e-PTFE. The extent o f endothelial surfacing of culture-lined e-PTFE grafts at the time of implantaion was 18.5% of the surface area for patent and 13.3% for occluded grafts (p = 0.4). O f the nine implants that had an analysis o f surface coverage before implantation and that remained patent, five had histologically detectable endothelium preoperatively, and only one failed to demonstrate surface endothelium when it was removed. Four had no demonstrable endothelium preoperatively, but two demonstrated surface endothelium after 1 month in the IVC. Endothelium also lined vascular channels that formed in the inner linings of seeded grafts. The channels were present in 6 of i0 grafts (60%) that were patent when they were perfusion fixed. The channels did not communicate with other vascular channels in the outer linings of the grafts; they were often quite large (Fig. 2), as compared with the vasa
vasorum that we described in arterial grafts~°; they measured up to 100 ~m in diameter. These vascular channels were oriented in a seemingly random pattern. Fibroblasts, identified by electron microscopy, surrounded these channels and underpinned the surface endothelium (Fig. 3). The culture-lined grafts most often had only a monolayer o f endothelium directly adjacent to the e-PTFE grafts (Fig. 4). Anastomotic subendothelial fibrous hyperplasia was noted in all o f the e-PTFE grafts (Fig. 1). The lengths of file grafts did not correlate with the thicknesses of hyperplasia, but the hyperplasia was thicker at the outflow than the inflow anastomoses. The hyperplasia averaged 0.75 mm in thickness at the inflow and 0.99 mm at the outflow anastomoses (p = 0.045). Two other variables were associated with anastomotic hyperplasia, specifically the application o f endothelium to the surface (by seeding or culture lining) prior to implantation and the diameter of the graft.
Journal of VASCULAR SURGERY
882 Herring et al.
Fig. 3. Transmission electron micrograph of subendothelial inner lining of seeded e-PTFE graft (6 mm I.D. and 60 mm long). Connective tissue cells are characteristic of fibroblasts. Note relationship of IVC lumen (L) to developing vascular channel (V). A mast cell (M) is seen just beneath luminal endothelium (E), and endothelium (e) is seen lining vascular channel. (Original magnification × 6160.)
The hyperplasia at the outflow anastomoses averaged 0.921 + 0.920 mm for lined, 0.751 + 0.581 m m for seeded, and 2.684 + 1.279 mm for unseeded e-PTFE (p = 0.004); however, at the inflow anastomoses the differences were insignificant. The diameter o f the graft was an even greater determinant of the thickness of hyperplasia (Table IV). The thickness of inflow or outflow anastomotic hyperplasia may be interpolated for other graft diameters using Fig. 5. The hyperplasia usually bridged the anastomoses, but at least 75% of the hyperplasia was within the graft itself and not within the host vein (Fig. 1). Exclusions f r o m the study. Twenty-six dogs were excluded from the study because they died before the 30th postoperative day. One dog died o f an anesthetic problem at 30 days, before inferior venacavography could be completed. That dog's graft
was considered an early patency but was excluded from the late patency and histologic considerationg" The majority o f the deaths occurred in the first 72 hours after their implant from anesthetic or bleeding complications. Five (19.2%) related directly to technical difficulties with the IVC replantation. Considerable constriction and shortening o f the replanted IVC was observcd in every case (Fig. 6). ~ There was no discernible association of mortality rates with the type or preparation of the synthetic grafts. DISCUSSION Graft material was the most significant deter -; minant of patency in canine IVC replacement surgery. The IVC-reimplanted segments developed substantial and persistent narrowing that began with vasospasm when the scgment was resected. Despite
Volume 1 Number 6 November 1984
883
Patency in canine inferior vena eava grafting
Fig. 4. Toluidine blue-stained section of culture-lined e-PTFE graft (6 mm I.D. and 60 mm long). Note monolayer of endothelium (E) directly adherent to the e-PTFE (G). (Original magnification x 300.) the resultant flow restriction, the IVC grafts remained patent. Clearly, the autogenous endothclialized IVC is superior to either o f the synthetics that we tested. Dacron-knitted grafts failed uniformly, but the e-PTFE grafts represented a substantial improvement over Dacron. Dacron and e-PTFE grafts differ in their chemistry, the smoothness o f their flow surface, their porosity, and their method o f preparation. One can only speculate as to which of these features differentiates the per%rmance o f the two materials in the IVC. The size o f the graft is also important. Previous investigators have recommended that venous grafts be isodiametric, n but somewhat oversized grafts performed as well in this study. The best patency was accomplished with seeded, isodiametric grafts, all o f which remained patent. Although isodiametric and oversized grafts were patent more often than undersized e-PTFE grafts (64.7% vs. 36.4%), these differences were insignificant. Nevertheless, late occlusions were significantly more common in long, narrow-diameter grafts. This finding could not be adequately explained by the data from this study, but it suggests that certain combinations o f graft length and diameter are adverse to IVC patency. The diameter o f the graft had an even greater effect on anastomotic healing. The greater the di-
Table IV. Subendothelial anastomotic fibrous hyperplasia Diameter
Thickness of A S F H
Standard
(ram)
(ram)
deviation
No.
0.357 0.911 2.365*
0.267 0.454 1.175
10 6 9
0.355 0.896 1.347"
0.182 0.463 0.911
10 8 8
Outflow 6 10 14 Inflow 6 10 14
*Fourteen-millimeter diameter differs significantlyfrom 6 and 10 mm grafts (p < 0.01).
ameter, the greater was the development o f luminal obstruction by anastomotic subendothelial fibroplasia. Whether the graft diameter be large or small, the first step in this hyperplastic process involved the migration o f fibroblasts from the adjacent host IVC onto the para-anastomotic graft lumen. We postulate that the turbulence at each anastomosis was increased as the graft diameter increasingly differed from the host vessel. In this scenario, thrombin and platelets would have been recruited by anastomotic turbulence 12 to sensitize and to stimulate cell divi-
lournal of VASCULAR SURGERY
884 Herring et al.
.
OUTFLOW INFLOW
~ I
_.. o /
2
ul z
1
s
j
x
.
O
.
.
.
.
6 8 10 12 14 GRAFT DIAMETER (MM)
®
10
" ~i~>~ @OUTFLOW
<"
MM
"~
LM
~--'~
<"
<-
INFLOW
Fig. 5, A. Thickness of anastomotic subendothelial fibroplasia in e-PTFE grafts is plotted on the ordinate, and diameter (ram) is plotted on the abscissa. B, Location and average thickness of anastomotic subendothelial fibrous hyperplasia depicted schematically for e-PTFE grafts of different sizes. Note relative constriction of IVC to accommodate 6 mm I.D. grafts and, by contrast, constriction of 14 mm grafts to accommodate smaller IVC. This difference may displace anastomotic turbulence into host vessel in instance of 6 mm grafts, at which site endothelial inhibition of platelet release can occur.
sion in the migrating fibroblasts. Activation of the platelets by contact with the e-PTFE surface may partly explain the prominence o f hyperplasia at the outflow compared with the inflow anastomoses. Endothelial seeding failed to confer patency on Dacron and e-PTFE grafts. Like our experience with small artery replacements, 1~ an insufficient proportion of the total grafted flow surface is covered with endothelium after seeding to effect protection against early occlusion. Moreover, studies with rain platelets suggest that a newly seeded graft is more thrombogenic than an unseeded one.14 Since seeded arterial grafts demonstrated markedly improved
Fig. 6. Inferior venacavogram oflVC reimplantation after 30 days. Note narrowing of reimplanted segment, designated by arrows at anastomoses. Narrowing persisted from moment segment of IVC was removed prior to reimplantation.
thromboresistance within 24 hours, early patency might be achieved in the IVC grafts by the addition o f a suitable anticoagulant. The cutture-lined graft~" failed to achieve more than a nominal endothelial lining at the time of implantation. This problem was encountered with cultured e-PTFE arterial grafts as well. 13 Further technical development will be needed to achieve a satisfactory endothelial lining in tissue culture. " Even though endothelial seeding or culture had little effect on graft patency, it decreased the anastomotic subendothelial fibrous hyperplasia substantially. Seeded or cultured endothelial islands may not have inhibited thrombin and platelet effects on the e-PTFE surface during the first few hours of ~' flow, but they would be expected to hasten such inhibition within a few hours o f implantation. This inhibition by endothelium would be expected to shorten the exposure o f the migrating fibroblasts to circulating mitogens, thereby limiting hypertrophy.
Volume 1 Number 6 November1984
We conclude that graft material is the major determinant o f patency in I V C replacements, that an extensive endothelial surface p r o m o t e s patency, b u t that simply seeding e - P T F E or D a c r o n grafts with 10 s endothelial cells does n o t provide sufficient end o t h e l i u m to alter early patency. Graft dimensions are also critical to l o n g - t e r m p a t e n c y and the develo p m e n t o f subendothelial fibrous hyperplasia. Graft recanalization is n o t a m a j o r feature in the healing o f synthetic venous replacements. We gratefully acknowledge the support provided by the St. Vincent Hospital Foundation, St. Vincent Hospital and Health Care Center, Indianapolis, Ind.; W. L. Gore and Associates, Flagstaff, Ariz.; and Bard Implants Division, USCI, Billerica, Mass., and the technical assistance provided by Mr. Anthony Hubbard, Mr. Michael Arnold, and Mrs. Rebecca Compton.
REFERENCES
1. Hiratzka L, Wright C. Graft materials in the venous system: Recent advances. In: Wright C, Hiratzka L, et al, eds. Vascular grafting. Boston: John Wright, PSG, CO, 1983. 2. Richardson JV, Wright CB, Hiratzka LF. The role of endothelium in the patency of small venous substitutes. J Surg Res 1980; 28:556-62. 3. Burkel W, Ford l, Vinter D, Kahn R, Graham L, Stanley J. Endothelial seeding of enzymatically derived and cultured cells on prosthetic grafts. In: Stanley J, Burkel W, et al, eds. Biologic and synthetic vascular prostheses. New York: Grune & Stratton, Inc, 1982:631-51.
DISCUSSION Dr. H e n r y Haimovici (New York, N.Y.). I enjoyed Dr. Herring's experimental models for inferior vena cava (IVC) replacements. His study demonstrates again the difficulties o f achieving consistent graft patency in such replacements. Seeding of these synthetic grafts altered very little the rates o f thrombosis, thus indicating that a neointimal lining as an initial feature is not sufficient to "overcome the thrombogenic factors. Some years ago my associates, Drs. Steinman and A1pert, and I carried out an experimental study on the use of a temporary arteriovenous (AV) fistula on long-term patency o f Dacron, Teflon, and autogenous vein grafts. These were all used as bypass implants around the inferior v e n a cava (Arch Surg i966; 93:747). The results were assessed periodically by serial venograms, done immediately, 3 to 8 days, and 4 to 6 months postoperatively, as well as by the anatomic appearance of the grafts. The venograms o f the synthetic grafts taken 3 to 8 days after implantation disclosed in about half o f them some degree of partial thrombosis together with collateral
latency in canine inferior vena cava grafCing
885
4. Herring MB, Gardner A, Glover J. Seeding endothefium onto canine arterial prostheses: The effects of graft design. Arch Surg i979; 14:679-82. 5. Hahn AW, Yasuda HK, James WJ, Nichols MF, Sadhir RK, Sharma AK, Pringle OA, York DH, Chaxlson EJ. Glow discharge polymers as coatings for implanted devices. Biomed Sci Instmm 1981; 17:109-13. 6. Herring MB, Evans D, Baughman S, Glover J. The quantiration of cultured cellular surface coverage: Applications for transparent and opaque surfaces. J Biomed Mater Res 1984; 18:567-76, 7. Hiratzka L, Wright C. Experimental and clinical results of grafts in the venous system; a current review. J Surg Res 1978; 25:542-61. 8. Cochran WG, Cox GM. Experimental designs, ed. 2. New York: John Wiley & Sons, Inc, 1957: 95-114. 9. Colton T. Statistics in medicine. Boston: Little, Brown & CO, i974. 10. Herring MB, Gardner AL, Clover J. A single-staged technique for seeding vascular grafts with autogenous endothelium. Surgery 1978; 84:498-504. 11. ScherckIP, Kerstein MD, Stansel Jr HC. The current status of vena caval replacement. Surgery 1974; 76:209-33. 12. Turitto VT, Baumgarmer HR. Inhibited platelet adhesions and irreversible thrombus formation under high shear conditions. Trans Am Soc Artif Intern Organs 1978; 24:71926. 13. Herring MB, Hubbard A, Baughman S, Smith D, Miller B, Dilley R, Gardner A, Glover J. Endothelium-lined small artery prostheses: A preliminary report. J Am Soc Artif Intern Organs 1983; 6:93-i02. 14. Allen BT, Long JA, Welch MI, Hopkins KT, Sicard CA, Clark RE. Effect of aspirin therapy and its withdrawal on control and endothelial cell seeded grafts. Surg Forum 1983; 34:470-2.
venous channels. Subsequently, the lumen regained a normal diameter concomitant with the disappearance of the collateral circulation. After discontinuation of the AV fistula (22 to 40 days postoperatively), an 83% patency rate of the synthetic grafts was achieved. O f the autogenous veins, one thrombosed at i3 days but completely recanalized later. All vein grafts remained patent. The temporary AV fistula, through increasing blood pressure and rapid blood flow, provided the hemodynamic factors necessary for the initial critical phase for the healing of the grafts and their resulting patency. In addition we found in most instances a perigraft fibrous reaction with resulting rigidity of the graft that may also have played a role in enhancing patency. Thus I would like to ask the authors whether they had noted a perigraft fibrosis and whether there was any correlation with patency o f their grafts. Dr. James C. Stanley (Ann Arbor, Mich.). This is a very complex study, and I found it dif~cult to interpret some of the data. The difficulty arose not so much with the
886
Herring et aL
authors' presentation but with the number o f variables in some of the study's subgroups. It may be important to note that 26 of the 117 experimental subjects were excluded because they died within the first 30 days, the majority occurring within the first 72 hours. If the attrition o f any of these animals could be related to the thrombogenicity of their caval grafts, then they might better have been included in the survival data when analyzed. The importance of endothelium in maintaining an inferior vena caval prosthesis has been clearly established by the earlier work of Wright and his colleagues, and it should be noted that the best results in the current report were found among seeded isodiametric grafts. It is also important to realize that incomplete endothelial coverage provides little reliable benefit regarding patency, a fact not too surprising given the low-flow situation o f the venous circulation. The Michigan group has had considerable experience in the past with cultivating endothelial cells in grafts prior to their implantation, and it may be worthwhile to note that our patency rates in the arterial circulation o f grafts prepared in this manner were considerably less than when grafts were seeded with the use o f a preclot method. Certainly, endothelial cells adherent to grafts in the culture environment do not appear to have the same tenacity or viability as those trapped in preclot blood. It is not surprising that caval grafts with cultivated endothelial cell linings frequently failed. Finally, I am not sure I would agree with the concept espoused by a number o f other investigators that endothelial cell seeding actually increases early graft thrombogenicity as documented by increased platelet accumulation following the graft implantation. Nevertheless, it is obvious that endothelial cell seeding affords no immediate benefit as to antithrombogenicity in either the venous or arterial circulations. This leads to the obvious question whether it would not be appropriate to intervene pharmacologically with either anticoagulants or antiplatelet agents, or perhaps even with the creation of an AV fistula in the case o f small venous conduits, in an effort to mainrain early patency of venous conduits. We may be somewhat naive to believe that endothelial cells are going to ensure graft patency before they have a chance to proliferate and evolve as a confluent endothelial surface. Dr. Eugene F. Bernstein (La Jolla, Calif.). I would like to congratulate Dr. Herring; he has tackled a very difficult job in attempting to replace the IVC, probably the most difficult large vessel in the body to replace. I think it is also clear that when one looks at the data he presented that endothelial cell seeding does not appear to be the best current answer to this problem. Finally, Drs. Plate and Eldof pointed out earlier how important this problem is, since the American approach to the treatment of iliofemoral venous thrombosis is anticoagulation and some o f the patients will end up with clinical problems related to IVC obstruction. Our approach to this challenge was a little different. With my associates; Drs. Chan and Barden, a combination
Journal of VASCULAR SURGERY
o f PTFE with external stents and AV fistulas was used to bypass the occluded IVC. We used a long graft because we believed that the problem frequently involved both the lilac vein and the vena cava, and both should be bypassed. The experimental model included a ligature o f the cava and either an endto-end or end-to-side anastomosis of a stented PTFE graft with an AV fistula o f the femoral vessels in the ipsilateral groin. In a control group of experiments in which bypasses were done without an AV fistula, the grafts all thrombosed. But in the presence o f AV fistula, the patency rate by venography at 1 and 8 weeks was 83%, and then following takedown of the fistula at 25 weeks, it was 75%. What seems clearly demonstrated is the virtue o f stenting in the venous system, as well as the benefit of an adjunctive AV fistula. I would like to ask Dr. Herring whether he has considered these adjuncts in addition to an endothelial cell lining? Perhaps the combination of all these technique~ will help solve this difficult problem. Dr. Creighton B. Wright (Cincinnati, Ohio). Dr. Robert Hobson and I started some 10 years ago with venous grafting in the femoral system of dogs and performed a total of 172 o f these types of grafts. We were able to achieve substantial success with autogenous vein grafts in the venous system. A totally healed prosthesis from a dog's descending thoracic aorta could be satisfactorily replanted into the venous circulation with a high degree ofpatency, and thus I thought that Dr. Herring's series might be even better with endothelial seeding. Since he did not achieve total endothelialization of his grafts, the lack o f seeding might account for some of the diminution in patency that he reported. I would like to comment on the spiral vein technique that Ray Chin and our group studied experimentally, and Drs. Doty, Baker, and others of us have used clinically for superior vena cava bypass and replacement of segments of large peripheral veins. This technique is well described and useful clinically. (Wright CB, Doty DB. Spiral veto grafting: The technique. In: Bergan JJ, Yao JST, eds. Operative techniques in vascular surgery. New York: Grune & Stratton, Inc, 1980:307-10.) Dr. Hobson and I were able to improve patency in segmental venous replacement with a number of unfavorable conduits by the inclusion o f either small or large AV fistulas. At the present time there is not an individual ideal venous replacement graft other than autogenous, and I think Dr. Herring is on the right track with a fully endothelialized graft. I hope he will share with us his additional thoughts about adjuvants such as ibuprofen (Motrin), aspirin, and dipyridamole (Persantine) or other antiplatelet drugs and AV fistulas as a means of getting these less than ideal grafts to endothelialize fully in situ in the canine inferior vena cava or other veins. Dr. Peter Gloviczki (Rochester, Minn.). Similar ex-
Volume 1 Number 6 November1984
periments have been carried out in our laboratory at the Mayo Clinic. We implanted externally supported expanded polytetrafluoroethylene grafts into the inferior vena cava o f dogs and obtained patency rates similar to those presented by Dr. Herring. We had 25% patency at 1 week, i month, and 3 months, as confirmed by venography. In contrast to their study, however, all of our grafts were isodiametric, with an internal diameter of 10 mm. If we used a femoral AV fistula, as Dr. Haimovici also suggested, or an AV fistula plus antiplatelet treatment, our patency rate significantly improved to 75% at 3 months. We also analyzed the cross-sectional area of the harvested specimens at 3 months and found a decrease of luminal diameter to 59%, which suggests that thrombus formation in the grafts is significant; therefore a largediameter graft may be beneficial. Also, the addition of a temporary AV fistula appears to improve patency. When we implanted spiral vein grafts, our patency rate was further improved. In fact, 6-month patency of spiral "Tein grafts was 100% when antiplatelet agents and a temporary AV fistula were used. Since autogenous vein o f adequate size is rarely available, we think that spiral vein grafts should be considered for replacement o f large veins before using prosthetic material. Dr. Herring (closing). I would like to thank all the discussants for their provocative questions. Dr. Haimovici has asked about AV fistulas and perigraft fibrosis. The improved patency that Dr. Haimiovici noted with perigraft fibrosis may relate to the apparent benefit of graft rigidity in the venous system. We have seen improved patency in the inferior vena cava with externally supported PTFE grafts in other experiments, but we have not quantitated perigraft fibrosis sufficiently to determine whether it has an effect. Dr. Stanley asked about the number of variables in the experiments that we reported, and I think the discussants
Patency in canine i@rior vena cava grafting
887
have pointed out for me that there are many variables influencing the success of venous replacement grafting. We began naively thinking that the technique of endothelial seeding might prove a panacea for synthetic venous replacement. When our early seeding experiments failed to confer patency, we planned additional experiments. Because of the number of variables involved and the cost of studying each variable individually, we decided to use a Latin square study design. We did not feel that the animal deaths related to the graft surface nor the patency of graft, rather to anesthetic and bleeding problems. Most of the deaths related to problems with IVC reimplantation. Because of the spasm that developed in the reimplanted segment, there was a great discrepancy in size between the implant and undisturbed segments. We are uncertain as to why the cultured cells failed to maintain graft patency. The cells may have been lost from the surface of the graft because of mechanical effects or perhaps because o f the complement-mediated effects. Apparently, the granulocytic response in the venous system is substantially greater than that in the arterial system. Therefore complement-mediated effects may be extremely important in venous grafting. Other pharmacologic strategies might be directed toward complement functions and used in addition to the platelet inhibitory drugs that Dr. Wright mentioned. The question of an AV fistula is an excellent one. We elected not to introduce yet another variable into our study design, particularly since fistula flow would be difficult to control. The consensus that I hear from Drs. Stanley, Haimovici, Bernstein, Wright, and Gloviczki is that perhaps by adding all of these techniques together, supplemental drugs, external wrapping o f the graft, endothelial seeding, and AV fistula, we would be able to achieve a much higher rate of prosthetic patency in the venous system.