| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Departments of Radiology (S.O.T., T.J.C., R.G.D.), Radiation Oncology (R.D.T., S.V.F.), Medicine (K.A.B.), and Surgery (M.F.), Indiana University Medical Center, UH0279, 550 N University Blvd, Indianapolis, IN 46202-5253. From the 1998 RSNA scientific assembly. Received July 8, 1998; revision requested August 6; revision received November 23; accepted February 16, 1999. Address reprint requests to S.O.T. (e-mail: streroto@iupui.edu).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Six-millimeter polytetrafluoroethylene arteriovenous grafts were created bilaterally in six dogs. After 1 month, Wallstents spanning the venous anastomosis were placed to accelerate restenosis. Gamma irradiation (12 Gy) was delivered endoluminally to one of the two grafts by using an iridium 192 source; thus, each animal served as its own control. Fistulography was performed monthly for 10 months or until graft thrombosis, with measurement of stenosis at each time point. At the conclusion of the study period, the treated area was examined histologically, and a computer model was used to calculate the volume of intimal hyperplasia.
RESULTS: Delayed stent migration resulted in exclusion of one dog. In the remaining five dogs, maximum stenosis across all time intervals was less for the treated side (P < .04), and the volume of intimal hyperplasia was less for the treated side (P < .045). In one animal studied at 1 year, this trend reversed in terms of percentage stenosis but not total neointimal volume.
CONCLUSION: Brachytherapy with 192Ir (gamma) delivered at the time of stent placement reduces restenosis in this hemodialysis graft model, but, depending on the parameter evaluated (stenosis vs total volume of neointima), the benefit may wane or even reverse with time.
Index terms: Animals • Dialysis, shunts, 80.42, 92.457, 93.457 • Grafts, interventional procedures, 92.1268, 93.1268 • Grafts, stenosis or thrombosis, 92.1268, 92.458, 93.1268, 93.458 • Iridium, radioactive
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Over the past 15 years, balloon angioplasty has become an accepted, approved treatment alternative to traditional surgical revision of hemodialysis-related venous stenosis (3–6). Balloon angioplasty is thought by many to be preferable to surgical revision because of its lower cost, because it is an outpatient procedure, because it can be easily repeated when restenosis occurs, and, most important, because it results in preservation of other veins for future access. Using percutaneous transluminal angioplasty, investigators have documented decreased graft thrombosis rates and prolonged access life (6–8). Despite its numerous advantages, venous angioplasty does have a number of drawbacks—most important, restenosis. Recurrence rates following balloon angioplasty are high, reportedly 37%–62% at 6 months (3–6,9). To our knowledge, efforts at controlling restenosis with stents (10–12) have been unsuccessful to date.
Brachytherapy, the endovascular delivery of radiation therapy, has been introduced as a possible means of controlling restenosis in coronary arteries (13–19) and peripheral arteries (20,21) following percutaneous treatments for stenosis, including percutaneous transluminal angioplasty and stent placement. To our knowledge, few data are available concerning the effects of brachytherapy on venous restenosis (22,23); however, the basic restenotic process of smooth muscle cell hyperplasia and proliferation is similar in arterial and venous lesions (1). The purpose of this study was to determine whether iridium 192 brachytherapy could be used to prevent in-stent stenosis in a canine hemodialysis graft model.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Twelve femoral artery–to–femoral vein arteriovenous grafts were created (M.F.) according to the method of Fillinger et al (24–27) in six adult mongrel dogs weighing 20–30 kg. The animals fasted for 12 hours prior to surgery or any percutaneous procedure, as required for anesthesia. Prophylactic cefazolin (1 g; Kefzol; Marsam Pharmaceuticals, Cherry Hill, NJ) was administered intramuscularly. Anesthesia was induced with thiopental sodium (Pentothal; Abbott Laboratories, North Chicago, Ill; 20–30 mg per kilogram of body weight). An endotracheal tube was inserted, and inhalation anesthesia (isoflurane [Isoflo]; Rhone-Poulenc Chemicals, Avonmouth, Bristol, UK) was administered to effect.
A transverse incision was made in each groin 3 cm below the inguinal ligament. The femoral artery and vein were mobilized. By using nontapered 6-mm reinforced expanded polytetrafluoroethylene (Gore-Tex; W.L. Gore, Flagstaff, Ariz) graft material, a 25-cm loop-type arteriovenous shunt was created by means of end-to-side anastomosis with a running 7-0 expanded polytetrafluoroethylene suture. The grafts were tunneled subcutaneously. Intraoperative medication consisted of only a flush solution containing heparin. The wounds were closed in layers by using polyglycolic acid suture (Dexon; Davis-Feck, Manati, Puerto Rico). After surgery, 500 mg of cefadroxil (Cefa-Tabs; Fort Dodge Laboratories, Fort Dodge, Iowa) was given orally daily for 5 days. The shunts were palpated daily to assess patency.
After 4 weeks of graft maturation, the animals were returned to the radiology research laboratory for stent placement (T.J.C., S.O.T.) and brachytherapy (R.D.T.). Anesthesia and preprocedural medication were identical to those described earlier. By using a sterile technique, access to the grafts was obtained by direct percutaneous puncture, and a 7-F sheath (Pinnacle; manufactured by Terumo and distributed by Boston Scientific, Natick, Mass) was placed in each graft directed toward the venous anastomosis.
After initial fistulography to determine the exact location of the venous anastomosis, 10-mL-diameter, 42-mm-long Wallstents (Schneider, Minneapolis, Minn) were placed percutaneously to span the venous anastomosis and were centered on the anastomosis. Stents were used as a part of the experimental model; previous work using this model had indicated that the most reliable way of producing stenosis in this model was to place Wallstents in this position, and the intimal hyperplasia produced was indistinguishable from that seen in hemodialysis access-related stenoses with or without stents (28). Angioplasty was not performed before or after stent placement. During stent placement, iodinated contrast material (iohexol [Omnipaque 350]; Nycomed, Princeton, NJ) and a flush solution containing heparin were used as needed. Post–stent placement baseline fistulography was performed by using spot radiography.
After stent placement, one graft in each animal was randomly assigned to receive brachytherapy; in this way, each animal served as its own control. Identical 6-F carrier catheters (Nucleotron, Columbia, Md) were placed through each stent via the 7-F sheath and were positioned by using a dummy seed cable (Nucleotron) under fluoroscopic guidance so that the radiation dose would be delivered to the area with the stent and to the vein 2.5 mm central to the stent, with a total length of radiation delivery of 7.5 cm (Fig 1). The carrier catheters on the untreated side served as a control for the manipulation involved in the delivery of brachytherapy.
|
|
The grafts were examined by using contrast material fistulography (T.J.C.) at 1-month intervals following brachytherapy to assess development of neointimal hyperplasia and/or stenosis. Note that scheduling conflicts precluded performance of several of the interval fistulographic examinations (Table).
|
Measurements
Measurement of intimal hyperplasia and stenoses was performed by direct measurement on the spot radiographs by using a micrometer (S.O.T.). Because relative measurements were used, correction for magnification was unnecessary. Measurements were obtained at the site of maximum stenosis within the treatment area by using the stent diameter as a reference. Percentage stenosis was calculated as follows: percentage = (1 - MLD/S) x 100, where MLD is the measured minimum lumen diameter and S is the stent diameter measurement. Measurements were made at a time remote from the performance of fistulography by an investigator who did not participate in the brachytherapy delivery and who was blinded to the side of irradiation.
Histologic Examination
Excised grafts, stents, and veins were fixed in formalin. Gross photographs of each sample were taken. Prior to sectioning, a silk suture was threaded within the outer (adventitial) fibrous layer through the entire length of each explant. This allowed proper orientation of each section with respect to adjacent sections for computer modeling (see earlier). Transverse sections were made in 5-mm increments extending 2 cm proximal to and 2 cm distal to the venous anastomosis or to the end of the stent. When cutting through the portions of the vein or graft with the stent, a razor blade was used to first cut through the soft tissue and then a wire cutter was used to cut each of the stent wires. A gross photograph was taken of each section. Magnified gross photographs were taken by using a dissecting microscope with a camera attachment.
Stent wires were individually removed by using fine forceps under the dissecting microscope. All sections were processed through paraffin, and three 5-µm sections were cut from each block. The sections were stained with hematoxylin-eosin. Neointimal thickness was approximated by using an ocular micrometer on the microscope.
Blinded analysis of the histologic sections was performed by a pathologist with extensive experience in stent histologic analysis.
Computer Modeling
Computer modeling based on the histologic specimens was also used to determine an estimated volume of intimal hyperplasia. This was used to try to overcome variations in the degree of neointima from point to point in the graft that have been seen in previous experiments (28). Since marked variations in intimal thickness may be encountered with this model, sampling error might be high if histologic sections are used for comparison. These variations occur in cross-section as well as longitudinally.
The computer modeling and interpretation were done in a blinded fashion. Hematoxylin-eosin–stained slides of the axial sections were input via a flatbed scanner (ScanJet 3c; Hewlett Packard, Palo Alto, Calif) and were enhanced by using a commercial graphics software package (Microsoft Imager 1.0; Microsoft, Seattle, Wash). The images were then digitized into a commercial three-dimensional radiation therapy treatment planning system (Render Plan 3-D; Precision Therapy International, North Miami Beach, Fla), where the vessels were virtually reconstructed in three dimensions. By using the computer model, the volume of remaining lumen could be calculated and subtracted from that of the bare stent lumen, which allowed calculation of the volume of neointimal hyperplasia present. This technique is an extension of planimetry used widely in studies of neointimal hyperplasia, in which three-dimensional rendering techniques similar to those used for computed tomography are applied.
Statistical Evaluation
The two-tailed paired Student t test was used to compare the volume of neointima between the treated and untreated legs. The difference in percentage stenosis between the two legs was plotted against the number of days since surgery for each dog to identify changes over time. In looking at the difference in percentage stenosis between the two legs, one dog had both grafts thrombose within 1 day of each other at 118 days after brachytherapy. To not possibly bias the results toward the alternative hypothesis, four new data points at months 6, 7, 8, and 9 were added, with a difference in percentage stenosis of 0 between the two legs. Four new points were added to make this dog have the same number of measurements as the mean number of 7 for the other four dogs. A weighted paired t test was used to compare the mean difference in percentage stenosis between the two legs across all times for each dog. The test was weighted by the number of times the measurements were made. The number of days since surgery was grouped into monthly categories, which were based on 30 days in a month plus or minus 15 days. A weighted paired t test was also used to compare the mean difference in percentage stenosis between the two legs of each dog during the time intervals. The test was weighted by the number of measurements in each dog during the time interval. In only one case did a dog have two measurements made within the same month. Differences were considered statistically significant if P was less than .05.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The mean volume of neointima determined with computer modeling was higher in the untreated legs (0.32 cm3 ± 0.14 vs 0.46 cm3 ± 0.19, P < .045). In all five dogs, the neointimal volume was greater in the untreated leg than in the treated leg. Across all time intervals, the percentage stenosis was higher for untreated legs (P < .04). When looking at each of the time intervals, the untreated leg showed a significantly greater percentage stenosis during the third (P < .02), fourth (P < .015), and seventh (P < .05) months. There was no significant difference between the treated and untreated legs during the first (P = .09), second (P = .12), fifth (P = .28), sixth (P = .13), eighth (P = .38), or ninth (P = .27) months. There was only one observation in the 10th and 12th months. Note that in all but one animal, the percentage stenosis was always higher in the control group. In this animal, this effect was reversed beginning at the 9-month point and persisting at 1 year.
All sections showed varying amounts of neointima consisting of mature fibrous connective tissue. Figure 2 shows representative sections from each group. Even within individual sections, there was marked variation in neointimal thickness, ranging from 0 to 1,950 µm, as has been demonstrated in prior studies by using this model (28). Focal chronic inflammation was seen in two specimens, both in the treatment arm, but was immediately adjacent to microperforations due to stent wires and was not thought to be due to the brachytherapy. Similar inflammation has been seen previously in nonirradiated specimens by using this model (28).
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Relatively little is known about the effects of brachytherapy in the venous system and hemodialysis circuits. A clinical trial of gamma brachytherapy in patients with failing hemodialysis grafts is reportedly under way (32), but results of this trial are not yet available. Smith et al (22) reported a trend toward reduction of intimal hyperplasia at 1 month in a sheep arteriovenous graft model with a rhenium 186–filled balloon (a beta emitter). Waksman et al (23) reported results of a pilot study in failing human hemodialysis grafts. They used 192Ir following balloon angioplasty at a dose of 14 Gy in 18 grafts in 11 patients. At a mean follow-up of 44 weeks, 11 of 18 sites remained patent. These preliminary human data are not particularly promising, although the authors indicated they believed suboptimal centering accounted for the poor results.
Nowhere is a feasible approach to the reduction of restenosis needed more than in hemodialysis access. Even the best results of balloon angioplasty in hemodialysis-related venous stenosis are measured in months, with a 38%–63% 6-month primary patency rate reported in several large series (3–6). Currently available vascular stents appear to be no better than balloon angioplasty, as determined in three prospective randomized trials (10–12). Medical treatments designed to reduce restenosis in hemodialysis access have been equally disappointing (2).
The results of our study suggest a possible role for brachytherapy in hemodialysis access–related stenosis. Gamma brachytherapy in this established animal model yielded significant reductions in in-stent restenosis, as measured as maximum percentage stenosis measurements at multiple time points during the study, with a consistent trend toward reduction even when statistical significance was not achieved. This effect was confirmed by volumetric computer modeling, which we believe more accurately conveys the difference in volume of intimal hyperplasia as opposed to stenosis at a single site. However, in the single animal followed up for 1 year, the treated side became the more stenotic side at 9 months. It should be noted that this late reversal could conceivably indicate a delayed detrimental effect of brachytherapy, which will need even longer-term studies to evaluate. Since this represents only a single observation, one must be vary cautious when trying to interpret it. The possible waning or even late detrimental benefit of brachytherapy in this dialysis graft model suggests that to obtain a durable result, treatment might need to be repeated at intervals after the initial intervention. This will need to be studied with further animal experiments and/or human clinical trials.
The animal model used here is an established model of hemodialysis access. This model has been shown to result in intimal hyperplasia indistinguishable from that seen in human dialysis grafts (24–28). Two downsides of this model are that it is inconsistent in its production of intimal hyperplasia and, in the natural state (ie, without stents), tends to develop stenosis at a relatively slow pace. Further, there tends to be greater accumulation of intimal hyperplasia within the graft as opposed to within the vein, which makes it slightly different from human grafts. Since we were studying in-stent restenosis as opposed to graft-vein anastomotic stenosis, this consideration may be less important. In a study in which the same model was used, Trerotola et al (28) showed that when Wallstents were placed across the venous anastomosis in this model, restenosis occurred at an accelerated rate: By 5 months, five of six grafts treated with Wallstents had thrombosis as compared with four of six untreated grafts and three of six grafts treated with Gianturco stents. Further, the mean percent stenosis in the Wallstent group was consistently higher at all time intervals. Thus, to create a more consistent and accelerated model, grafts with Wallstents in place served as the experimental model. Hence, the results obtained here can be directly applied only to the reduction of in-stent restenosis, although we believe they would apply to lesions treated with angioplasty alone as well, since such lesions are histologically indistinguishable from in-stent restenosis. Similar studies in coronary arteries have shown reductions in in-stent restenosis following brachytherapy (19–21) or when radioactive stents were used (29–31).
It is interesting that, despite the use of Wallstents as part of the model, only one of five dogs developed intimal hyperplasia at the rate seen by Trerotola et al, and the remaining four dogs' grafts remained patent for up to 1 year. The explanation for this disparity is unclear. Unlike in prior trials (24–28), aspirin was not administered to the animals on a daily basis. We chose not to use aspirin in this study to eliminate a potentially confounding variable. Conflicting data exist with regard to the role of aspirin in hemodialysis access failure; however, findings of at least one trial have suggested that aspirin worsens graft survival (33). If this is true, the discrepancy in neointimal hyperplasia between this and prior studies by using this model might be explained on this basis.
Due to the small number of animals in the study, we were able to perform histologic analysis at only a single data point. Many studies of brachytherapy in coronary arteries have used shorter end points, usually in the 1-month range. By sampling at the 9-month time point, it is possible that we missed greater differences in neointimal volume occurring at an earlier time point. We chose to follow the grafts for a longer time to determine whether any late detrimental effects of brachytherapy might be evident, as well as to determine whether the effect of a single treatment with brachytherapy would be durable. While no detrimental effects were observed histologically, the late reversal of results seen in a single animal suggests that further observations at this time point should be made before universally accepting the concept of brachytherapy on the basis of findings of shorter-term studies.
The mechanism of action of brachytherapy in the arterial system is thought to be inhibition of smooth muscle cell proliferation (14,16,17,29,30). Studies have shown that despite inhibition of smooth muscle cells, endothelial regrowth is not inhibited (15,17,30). Since restenosis in hemodialysis access appears to be due to mechanisms similar to those in arteries (1,27), the mechanism of action of brachytherapy in veins is likely similar. The exact dosing regimen and timing of brachytherapy have yet to be determined, as has the ideal source. Theoretically, applying the dose several days after the intervention would be ideal, as this is when smooth muscle cells are most actively proliferating: Waksman et al (17) found greater inhibition or intimal hyperplasia when brachytherapy was given 48 hours after instead of immediately after intervention. For practical reasons, timing of brachytherapy must by necessity coincide with the intervention; to our knowledge, this has been the approach taken by most investigators. In dialysis access grafts, however, due to the superficial nature of the lesion, it would be possibly to apply external beam therapy at a time remote from the intervention; more work is needed in this area. External beam therapy has been shown to reduce restenosis in animal arterial models (34,35).
The doses used in most studies to date that we know of have ranged from 7 to 56 Gy, with doses of 12–14 Gy being the most commonly used (13–21). Findings of one study (18) showed increasing benefit in the coronary arteries up to but not beyond 28 Gy; however, for safety reasons, lower doses have been advocated. Bottcher et al (20), in human superficial femoral arteries, used gamma brachytherapy with an 192Ir source to prevent restenosis in stents used to treat atherosclerotic lesions, with excellent results at follow-up from 3 to 27 months; the dose used was 12 Gy. In the clinical study by Teirstein et al (13) of coronary brachytherapy, the dose was calculated at between 8 and 30 Gy, depending on the distance from the source.
We chose to use 12 Gy, as it was the dose most commonly used in preclinical studies and clinical studies. It should be noted, however, that no centering device was used; thus, the delivered dose varied depending on the course of the catheter in the treated area (Fig 1). The maximum dose delivered could not be reliably determined, while the minimum dose was 6 Gy. Whether centering the source within the vessel lumen would have improved results will require further study.
Finally, with respect to the choice of radiation source, both gamma and beta emitters have been used. Beta emitters have the advantage of less exposure to the operator and do not require the use of an afterloader (18,36). However, longer treatment times are required if beta emitters are used, although this can be reduced by using liquid beta emitting sources in balloons (36). Use of beta-emitting stents obviates this problem (29–31). Gamma emitters have the advantage of being readily available, as they are used clinically for such indications as treatment of keloids and prevention of heterotopic bone and pterygia. Further, the delivery of the dose takes only a matter of seconds. We chose to use a 192Ir source in this study because it was readily available. Because of dosing and radiation protection issues, it is likely that beta emitters will be the preferred sources for brachytherapy in clinical applications (36).
Many investigators have used maximum intimal thickness as the primary determinant of efficacy (14–19,29–31). While this may be applicable to the coronary acute balloon injury model, we did not think it applied to the present model because of the wide variations in intimal thickness observed. We believe the computer model allows a more meaningful description of the extent of the restenotic process, as it allows determination of the total volume of neointimal hyperplasia and is less affected by variations in neointimal thickness. The importance of the model can be seen in Figure 3: Clearly, there is a much greater volume of intimal hyperplasia in the untreated graft, yet this is poorly conveyed by measurement of percent stenosis or in a single histologic section. The model allows us to quantify this visual impression. There is an important limitation to the model, however: It makes assumptions about data points between actual sections; the more sections obtained, the less this limitation. We chose to section at 5-mm intervals; below this would be prohibitively expensive. Nonetheless, we believe the data from the model are representative of the actual neointimal volumes. A potential improvement for further study might be the application of this model to intravascular ultrasonographic (US) data, thus allowing a greater number of data points. However, findings of one recent study (37) showed that under experimental conditions, angiography and intravascular US were equivalent in determining in-stent stenosis; thus, we believe our use of angiographic measurements in this experiment is valid.
|
|
Practical applications: Our preliminary data with this animal model suggest that gamma brachytherapy can reduce in-stent restenosis in hemodialysis access. We believe the results are sufficiently compelling to warrant further animal study; however, the long-term (>1-year) effects and optimal dosing of brachytherapy remain to be determined.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  Y. Wang, M. Krishnamoorthy, R. Banerjee, J. Zhang, S. Rudich, C. Holland, L. Arend, and P. Roy-Chaudhury Venous stenosis in a pig arteriovenous fistula model--anatomy, mechanisms and cellular phenotypes Nephrol. Dial. Transplant., February 1, 2008; 23(2): 525 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. Terry, D. K. Blumenthal, S. Sikharam, L. Li, T. Kuji, S. E. Kern, and A. K. Cheung Evaluation of histological techniques for quantifying haemodialysis arteriovenous (AV) graft hyperplasia Nephrol. Dial. Transplant., November 1, 2006; 21(11): 3172 - 3179. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. O. Trerotola and L. A. Turmel-Rodrigues Off the Beaten Path: Transbrachial Approach for Native Fistula Interventions Radiology, March 1, 2001; 218(3): 617 - 619. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
