Evaluation of the antiadhesion potential of UV cross-linked gelatin
lms in a rat abdominal model Shojiro Matsuda a,b , Naomi Se a,c , Hiroo Iwata a, *, Yoshito Ikada d a Institute for Frontier Medical Sciences, Kyoto University, Kawahara-cho 53, Shogoin, Sakyo-ku, Kyoto 606-8507, Japan b Research and Development Department, GUNZE Ltd., Natsumegaichi 46, Aono-cho, Ayabe, Kyoto 623-8513, Japan c JMS Company, Limited, Kako-machi 12-17, Naka-ku, Hiroshima 730-8652, Japan d Faculty of Medical Engineering, Suzuka University of Medical Science, Kataoka-cho 1001, Suzuka, Mie 510-0293, Japan Received 27 November 2000; accepted 20 November 2001 Abstract Among ve kinds of rat adhesion models tested, the following model was selected. The epigastric vein 2.5 cm from the midline of the abdomen was cut by sharp scissors, and the lateral side of the cut epigastric vein was ligated using a 30 silk suture. This model could be easily prepared and gave a rate of adhesion formation of 90%, which was useful for screening antiadhesive materials. For the kinetic study of tissue adhesion in this model, an injured site was covered with a non-degradable poly(vinyl alcohol) (PVA) lm. The incidence rate of adhesion was 18%, when the PVA lm covered the injured site for 2 days. This suggests that an antiadhesive barrier should cover the injured site for at least 2 days. The antiadhesion efcacy of cross-linked gelatin lms were evaluated using this adhesion model. The UV cross-linked gelatin lm which was designed to exist for 2 days but to disappear at day 3 in the rat abdominal cavity showed the highest antiadhesion efcacy. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Peritoneal adhesion; Rat; Antiadhesion; Gelatin lm 1. Introduction Enormous efforts have been devoted to prevention of tissue adhesion by improving surgical procedures and using antiadhesion materials. However, tissue adhesion continues to be a cause of post-surgical complications. Placing a physical barrier between an injured site and the adjacent tissues is a direct approach to prevent adhesion. Various kinds of materials made of animal tissues, biological materials and synthetic polymers have been reported to be effective in reducing adhesion both in animal models and in clinical practices [1]. Some have been commercially and clinically used [25]. Non- absorbable synthetic materials such as silicone and polytetrauoroethylene sheets have been employed as antiadhesive materials [6,7]. Although they can effec- tively isolate an injured site from adjacent tissues and prevent adhesion, because of their non-degradability they may cause foreign body reactions, inducing strong connective tissue formation which again evokes tissue adhesion. Coating the injured surface with viscous polymer solutions such as polyvinylpyrolidone, carboxy- methylcellulose [8], dextran [9] and hyaluronic acid [10] also have been reported to be effective. However, the stay at the applied site depends on circumstances, sometimes the polymer solutions may be readily washed out before the injured surface becomes no longer susceptible to adhesion formation. Sepralm s is a hydrophilic lm composed of sodium hyaluronate and carboxymethylcellulose. It is claimed that this lm adheres to a wound surface and mechanically separates the tissue during the post-operative healing phase. But, some surgeons claimed difculty in handling because of its brittleness and low mechanical properties. Interceed s is a knitted fabric composed of oxidized regenerated cellulose and is reported to signicantly reduce adhesion when blood contamination is avoided during applica- tion [3,11]. As blood evacuation cannot always be done effectively in a clinical situation, surgeons are not always satised with Interceed s in the prevention of adhesion. An antiadhesive material should be designed to stay at the injured site during the post-operative wound *Corresponding author. Tel.: +81-75-751-4119; fax: +81-75-751- 4144. E-mail address: iwata@frontier.kyoto-u.ac.jp (H. Iwata). 0142-9612/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 1 ) 0 0 4 1 8 - 5 healing phase. The aims of the present study were (1) to establish an animal model that can form adhesion with high reproducibility, (2) to determine the critical time when the injured surface no longer becomes susceptible to adhesion formation in the model, (3) to design an antiadhesive lm from gelatin on the basis of these ndings, and (4) to examine its efcacy using the model. 2. Materials and methods 2.1. Materials Six-week-old Wistar rats were purchased from Shi- mizu Laboratory Animal Supply Co., Ltd. (Kyoto, Japan). All animal procedures were in accordance with institutional guidelines of Kyoto University on animal experimentation. Gelatin, extracted using the alkaline method from bovine bone (type I collagen), was donated by Nitta Gelatin Co. Ltd. (Osaka, Japan). Antiadhesive efcacy was determined for lms prepared from this. A poly(vinyl alcohol) (PVA) hydrogel lm of 70 wt% water content was prepared by low temperature crystal- lization in a water/dimethyl sulfoxide mixed solvent [12] and used for the kinetic study of tissue adhesion formation. To prepare gelatin lms, gelatin powder was dissolved in distilled water to a concentration of 10 wt%. The solution was cast on glass plates and allowed to dry in air to obtain lms of 0.1 mm thickness. Cross-linking was introduced by exposing both the sides of lms to ultraviolet (UV) light in air 60 cm from an UV lamp (Toshiba GL-15 (15 W)) for different time periods. The lms were sterilized by ethyleneoxide gas for the animal experiment. 2.2. Rat adhesion model To set up an animal model to evaluate the efcacy of antiadhesion materials, the adhesion incidence of ve adhesion models were evaluated. In total, 52 rats were used in this study. Rats were anesthetized by inhalation of diethyl ether and their ventral hair was scraped using a razor. Using the aseptic technique, a 3 cm incision was made on the midline of the abdominal wall, and both abdominal walls were reected and similar adhesion models were made on each of the abdominal walls. One site was excluded because of technical failure in preparation of the adhesion model. A total of 103 sites were used in this study. Five kinds of adhesion models were prepared by varying the following factors, the presence or absence of a 30 silk suture knot, location of the injury (2.5 or 1.5 cm from the midline incision), injury spot (on the epigastric vein or abdominal wall) and cutting tissue by sharp scissors. The combinations of these factors in the models are given in Table 1. After preparing the models, the middle line incision was closed using 30 silk suture. The animals were maintained under free access to food and water. Seven days after surgery, the incision was reopen to examine the injured site for the incidence of adhesion by the naked eye. Some of the model animals (e) were sacriced at 12 h, 1, 3 and 7 days after surgery to observe the adhesion formation process at the injured site. The abdominal wall of the injured site was removed and xed in 10 wt% formalin solution. The tissues were processed by the standard procedure for histological examinations and their thin sections were examined after staining with hematoxylineosin (HE). 2.3. Kinetic study on adhesion prevention PVA lms with 2 3 cm 2 and 2 mm thickness were sterilized with 70 vol% ethanol for 2 h and then swollen in sterilized phosphate buffered saline at 251C for 1 h. The adhesion models (e) were made on both sides of the abdominal walls. A PVA hydrogel lm was applied and xed with 50 nylon suture at four corners to cover the injured site. In the preliminary experiments, the existence of the nylon suture induced tissue reactions Table 1 Incidence rates of adhesion formation 7 days after surgery in various rat abdominal adhesion models Adhesion model Incidence of adhesion 7 days after surgery Suture knot Injury location from the midline (cm) Injury spot Cutting No. of model sites Incidence rate (%) Adhesion (+) Adhesion () (a) Removed 2.5 Vein Yes 4 16 20 b (b) Present 1.5 Vein Yes 8 12 40 b (c) Present 2.5 Wall Yes 13 7 65 NS (d) a Present 2.5 Vein No 19 4 83 NS (e) Present 2.5 Vein Yes 18 2 90 a One site was excluded because of technical failure during preparation of the adhesion model. b Signicantly different from model (e) at the 0.1% level. NS; not signicantly different from model (e) at the 5% level. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2902 which spread over the injured model site. Therefore, to reduce the effect of the nylon suture at the injured site, relatively larger lms were applied in this kinetic study than the lms used in the antiadhesion efcacy study. At 1, 2 and 3 days after closure, the abdomen was reopened, and both the PVA lm and nylon sutures were carefully removed not to disturb the injured model surface, and the abdomen was closed again. One site at 2 days and all sites at 3 days were excluded, because the adhesion formation had already occurred to the nylon suture and PVA lm at the reopening. In total, 21 model sites were evaluated. Seven days after the rst surgery, all rats were sacriced for the assessment of adhesion by the naked eye. Some animals were sacriced at 12, 24, 36 and 48 h after the lm application to observe the wound healing process at the injured site histologically. The tissues were processed by the standard procedure for histological examinations and their thin sections were examined after staining with hematoxylineosin (HE). 2.4. In vivo degradation of UV cross-linked gelatin lms Cross-linked gelatin lms by UV irradiation for 5, 10, 20 and 40 h were used in the biodegradation study. In total, 11 rats were used in this study. Each rat received implantation of three pieces of lms with 1 1.5 cm 2 of known dry weight intraperitoneally. At predetermined times after implantation, the lms were explanted with the surrounding tissue. This tissue was carefully removed not to damage the lm. The lms were completely dried under vacuum conditions and the dry weights of the lms were measured. The extent of in vivo degradation was expressed as the mean7SD of the percentage of the remaining dry weight. 2.5. Antiadhesive efcacy evaluation Forty-one rats were used in this study. Adhesion models (e) were made on both sides of the abdominal walls. Three sites were not used because of technical failure in preparation of the adhesion model. In total, 79 model sites were used to evaluate the antiadhesive efcacy of cross-linked gelatin. The animals were divided into ve experimental groups. Group-1 (G-1) consisted of animals that underwent the experimental adhesion protocol of the model (e) without application of an experimental antiadhesive sample. These animals constituted the untreated control group. Group-2 (G-2) was treated with the protocol with application of a non- cross-linked gelatin lm as described later. Group-3 (G- 3), Group-4 (G-4) and Group-5 (G-5) were treated with the same protocol, with application of UV 10, 20 and 40 h irradiated gelatin lm, respectively. In G-2 to G-5, an antiadhesive sample of 1 1.5 cm 2 was placed on the injured site of the model (e) without xation by suturing. The animals were maintained under free access to food and water. Animals from the ve groups were randomly operated by one surgeon. Two or three animals from each group were operated per day. The assessment of adhesion was examined 7 days after the operation by the naked eye. When tissue adhesion was observed at the model site, it was judged as the occurrence of tissue adhesion regardless of adhesion severity. For the histological study, some rats were sacriced 1 day and 7 days after antiadhesive barrier application and the model sites were removed and xed in 10 wt% formalin solution. The tissue were processed by the standard procedure for histological examinations and their thin sections were examined after staining with HE. 2.6. Statistical analysis The chi-squared test, Yates correction and Fishers exact test were employed to determine the signicant differences of frequency between two groups. The signicance which existed at Po0:05; Po0:01 and Po0:001 was expressed by *, ** and ***, respectively. 3. Results 3.1. Adhesion model Five kinds of adhesion models were prepared as given in Table 1. The macroscopic appearances at the injured site are shown in Fig. 1 for model (e) just after preparation and 7 days after surgery. The greater omentum adhered to the suture knot left on the abdominal wall and spread over the surrounding tissue. In most adhesions in all models, the greater omentum was the principal tissue responsible for adhesion, although the mesentery was also engaged in some adhesions. The incidence rates of adhesion are listed in Table 1 for each model. Adhesion depended on the location of Fig. 1. Abdominal wall adhesion model (e) (see also Table 1). (A) Immediately after surgery, arrow: silk suture knot, (B) 7 days after surgery. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2903 the surgical trauma and the existence of a suture knot. A suture knot left on the site at 2.5 cm from the midline of the abdomen caused adhesion at a rate >80% in models (d) and (e). However, the incidence rate in model (b), in which a suture knot was left on the site at 1.5 cm from the midline, was signicantly lower than that of model (e). This may be due to the movement of the bowel facing the suture knot and the surgical trauma. Although the location of the model site was an important factor, the existence of a suture knot more markedly affected the incidence of adhesion. Model (a), in which the surgical trauma was made 2.5 cm from the midline, but the suture knot was removed, demonstrated the lowest incidence rate. The incidence rate of model (a) was also signicantly lower than that of model (e). This suggests that the existence of a suture knot was the main factor in the incidence of adhesion. In model (e), the epigastric vein was cut by sharp scissors and its distal end was ligated using a 30 silk suture. A small amount of blood exuded from the proximal end of the vein. When the injury point was not on the epigastric point (model (c)), or cutting by scissors was not done (model (d)), the incidence rate was lower than that of model (e). Since model (e) presented the highest adhesion inci- dence, and might represent more realistic surgical events than model (d), model (e) was employed in the following experiments. However, no signicant difference was found in the adhesion incidence rate between models (e) vs. (c) and (e) vs. (d). The tissues of model (e) were explanted at predeter- mined time intervals. The photomicrographs of their thin sections stained by HE are shown in Fig. 2. At 12 h after surgery, the suture knot was partially covered with an amorphous material. At day 1, the greater omentum had already adhered to the suture knot, but the adhesion was not strong, so that it could be released by weak mechanical disturbance. At day 3, the population of neutrophils decreased, but more monocytes and bro- blasts inltrated into the area surrounding the suture, and rm granulation tissue with neovascularization was formed between the greater omentum and the abdom- inal wall. At day 7, the granulation tissue became larger and rmly covered the abdominal wall surrounding the suture knot. 3.2. Kinetic study on adhesion prevention The injured site of model (e) was covered with a PVA lm xed by 50 nylon monolament suturing and then the lm and the suture were removed at 1, 2 and 3 days after the operation. The incident rates of adhesion formation at day 7 after rst surgery are shown in Fig. 3. The incidence rate of adhesion of the 1 day and 2 days PVA covered model was signicantly lower than that of the non-covered model. The incidence rate of the 2 days PVA covered model was lower than that of 1 day, but there was no signicant difference. Coverage of the injured site with a hydrogel sheet for 3 days was expected to drastically decrease the incidence of adhe- sion, but this was not the case. At 3 days after coverage by PVA lm, it was found that the greater omentum was markedly adhered to the nylon suture, the surrounding PVA lm and abdominal wall tissue. Removal of these materials from the model site might be traumatic. A kinetic study was not carried out beyond 3 days. This suggests that the foreign materials might become nuclei for new adhesion. Fig. 2. Histological observation of the wound site of adhesion model (e). Thin sections were stained with HE. Fig. 3. Kinetic study of adhesion formation. A PVA lm was placed on the wound part for various times after injury. After those periods, the lm was removed. Seven days after the rst surgery, all rats were sacriced for the assessment of adhesion by the naked eye. The incidence rate of adhesion formation in the group without PVA lm application was cited from that of model (e) in Table 1. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2904 Microscopic examinations of the tissue healing process of the model (e) are shown in Fig. 4. At 12 h after PVA lm application, the suture knot was partially covered with an amorphous substance and was almost free from cell inltration. At 24 h after surgery, neutrophils inltrated into the substance covering the suture and the surrounding tissue, and a brin-like exudate fully covered the suture. The tissue healing continued under the PVA lm and granulation tissue completely covered the suture knot 48 h after lm application. These ndings indicate that antiadhesive materials should cover the injured site for at least 48 h as far as in this animal model. 3.3. Dependence of the biodegradation rate of gelatin lms on UV irradiation time Gelatin was employed as a material to prepare antiadhesive lms, because it had been clinically used as a biomaterial and is known to biodegrade after implantation. Non-cross-linked gelatin lms were ra- pidly swollen with water and dissolved into ascites within several hours in the abdominal cavity. Therefore, gelatin lms were cross-linked by UV irradiation to retard their biodegradation. Fig. 5 shows that the biodegradation of the gelatin lms treated with UV light for different periods in the rat abdominal cavity. The biodegradation rate decreased the longer the UV irradiation time. Cross-linked gelatin lms adsorbed ascites and became hydrogel after the implantation. Films deformed but existed as a single piece, and its volume decreased with increasing implantation periods. Five hours UV irradiated gelatin lm was not found after 1 day implantation. Similarly, 10 h UV irradiated gelatin lm was not found after 3 days implantation. In conjunction with the ndings of the kinetic studies with PVA, 10 h irradiation appeared to be promising for preparing antiadhesive lms. 3.4. Evaluation of the antiadhesion efcacy The antiadhesion efcacy of gelatin was evaluated using model (e). The materials were applied onto the injured site of model (e) and left there without xation by suturing. The incidence rates of adhesion are summarized in Table 2 for each material. Among the four antiadhesive application groups, only G-3 showed a signicantly lower incidence rate of adhesion than the control group (G-1). The incidence rates become higher in the following order G-3, G-2, G-4 and G-5. This suggests that the optimal UV irradiation time for the lowest incidence rate of adhesion may exist between 0 and 20 h. G-5 showed a signicantly higher incidence rate than G-3. Macroscopic views of the model site with the 10 h UV irradiated gelatin lm just after application and at day 7 are shown in Fig. 6(A) and (B), respectively. The model site looked clear and smooth, suggesting that it had been fully covered with regener- ated tissue. Photomicrographs of the tissue sections are shown in Fig. 6(C) and (D) for the gelatin lm irradiated for 10 h at days 1 and 7 after application, respectively. The gelatin lm could be clearly identied and the amorphous tissue layer that was stained deeper than the gelatin lm could be seen around the suture knot at day 1. Few neutrophils inltrated between the lm and the abdominal wall. At day 7, the gelatin lm had already been fully degraded and sparse granulation tissue including monocytes and broblasts covered the suture. The macroscopic and microscopic views of the model site with the 40 h UV irradiated gelatin lm application at day 7 are shown in Fig. 6(E) and (F), respectively. Fig. 4. Histological observation of the wound site covered by the PVA hydrogel lm. Thin sections were stained with HE. Fig. 5. In vivo degradation of the cross-linked gelatin lms with different UV irradiation times. Films irradiated for 5 and 10 h could not be found by naked eye observation after 1 day and 3 days, respectively. Irradiation time; J: 5 h, n: 10 h, &: 20 h and : 40 h. Error bar indicates the standard deviation. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2905 The lms were completely encapsulated by the granula- tion tissue and the capsule tissue was covered by the surrounding greater omentum. 4. Discussion Various kinds of adhesion models have been devel- oped using animals, but most have specic problems. As discussed by Bakkum et al. [13], a reliable animal model that can be reproduced and scored objectively, and that possess clinical relevancy concerning the tissue involved and pathogenesis of the adhesions, is sought. In the present study, ve kinds of adhesion models were made on the rat abdominal wall and compared to nd the most suitable model that can be easily prepared and give a high incidence rate of adhesion. Model (e) in which the epigastric vein was cut 2.5 cm from the midline using sharp scissors and the distal end of the epigastric vein was ligated with a 30 silk suture gave a 90% incidence rate of adhesion. This high incidence rate appears to be because the model site involved a surgical trauma and a foreign material (a suture knot). The ischemic area and exuded blood have also been considered to induce adhesion, but these factors did not exert signicant effects on the adhesion incidence rates in the present study. The present model could be easily prepared and gave a high incidence rate, suggesting that this model is useful for screening antiadhesive materials. Fig. 7 shows the tissue reactions against the model site with or without an antiadhesive material which were elucidated from the ndings mentioned above. An antiadhesive barrier should cover the injured site until the site is no longer susceptible to adhesion and then the barrier should quickly disappear from the site. Other- wise, tissue adhesion would take place at the wound site, accompanying foreign body reactions to the barrier materials, which can cause another focus of adhesion. For development of antiadhesive barriers, it is very important to know the period of time that is required for the injured tissue to change into a non-adherent surface during the wound healing process. By covering an injured site with a silicone lm for predetermined periods, Rodeheaver et al. [14] deter- mined the time period necessary for healing the injured abdominal surface of rats, so that the tissue was no longer susceptible to adhesion. They showed that a material that remained on the injured surface for at least 36 h after injury was required for an effective reduction in adhesion incidence. In the present study, a PVA lm was employed to cover the model surface instead of the silicone lm, because it was reported to induce less granulation than silicone [15]. The PVA hydrogel was Fig. 6. The macroscopic and microscopic view of the gelatin lm applied abdominal wall adhesion model. (A) Immediately after application of 10 h UV irradiated gelatin lm. (B) 1 week after application of 10 h UV irradiated gelatin lm. (C) 1 day after application of 10 h UV irradiated gelatin lm. (D) 1 week after application of 10 h UV irradiated gelatin lm. (E) and (F) 1 week after application of 40 h UV irradiated gelatin lm. Table 2 Antiadhesive efcacy of gelatin lms Group code Materials tested Incidence of adhesion 7 days after surgery No. of model sites Incidence rate (%) Adhesion (+) Adhesion () G-1 a Without any agents 15 2 G-2 Gelatin lm without UV irradiation 10 6 G-3 a Gelatin lm irradiated with UV for 10 h 8 11 G-4 a Gelatin lm irradiated with UV for 20 h 10 3 G-5 Gelatin lm irradiated with UV for 40 h 13 1 a One site of each group was excluded because of technical failure during preparation of the adhesion model. **Signicantly different at the 1% level. No mark; not signicantly different at the 5% level. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2906 also reported to allow low molecular weight substances, such as oxygen and nutrients, to freely pass through it [16] and, thus, it is expected not to hinder the wound healing process. In the present model, the incidence rate of adhesion was 18%, when the PVA lm remained in place for 2 days. When the PVA barrier remained in place for a longer time, the nylon suture used to x the PVA lm induced foreign body reactions and thus allowed adhesion to take place at the periphery of the lm. Both the ndings of Rodeheaver et al. [14] and the present ones suggest that an antiadhesive barrier should cover the injured site for at least 2 days to prevent the adhesion of injured tissues. In contrast to cross-linked gelatin lms, non-cross- linked gelatin lm readily absorbs ascites, rapidly dissolves at body temperature, and hence cannot stay in place as a lm for more than several hours. As a result, the antiadhesive effect of the non-cross-linked gelatin lm was limited, although its incidence rate of adhesion was lower than that of the controls. It has been reported that the incidence of post-operative adhesion can be effectively decreased by simple application of viscous solutions of polysaccharides, such as carbox- ymethylcellulose [8], dextran [9] or hyaluronic acid. These viscous solutions will cover the injured surface for a certain period of time. A similar benet may be obtained when a non-cross-linked gelatin lm is applied. A kinetic study using a PVA lm suggested that a barrier should stay in place for at least 2 days. Gelatin lms with various bioabsorption rates were prepared by controlling the UV cross-link. It was expected at the beginning of this study, that a lm that remained intact for a longer time could more effectively prevent adhesion. However, the gelatin lm prepared by 10 h UV irradiation and absorbed within 3 days, could most effectively prevent adhesion among the antiadhesive barriers examined in the present study. The antiadhesive efcacy of gelatin lms became less efcient as the UV irradiation time became longer than 10 h. This is probably because the prolonged existence of gelatin lms at the injury site induced foreign body reactions and resulted in adhesion. As demonstrated above, the gelatin lm that was designed to exist for 2 days but to disappear at day 3 in the rat abdominal cavity showed a signicantly lower incidence rate of adhesion than the controls. Although longer periods of covering were expected to reduce the incidence of adhesion, the lm that was not absorbed within 3 days did not show better antiadhesion efcacy. This may be explained in terms of foreign body reactions induced by the antiadhesive material itself as schematically shown in Fig. 7. In this antiadhesive efcacy study, G-3 showed the highest antiadhesive efcacy. However, the incidence rate of adhesion was still 42%. For the development of an ideal antiadhesive that can completely prevent adhesion, a bioabsorbable material that has an optimal absorbability and that induces little tissue reactions so as not to induce adhesion formation by itself is required. References [1] Wiseman D. Polymers for the prevention of surgical adhesions. In: Domb AJ, editor. Polymeric site-specic pharmacotherapy. Arlington, TX: Wiley, 1994. p. 370421. [2] Diamond MP, Linsky CB, Cunningham T, Constantine B, diZerega GS, DeCherney AH. A model for sidewall adhesions in the rabbit: reduction by an absorbable barrier. Microsurg 1987;8:197200. [3] Interceed (TC7) Adhesion Barrier Study Group. Prevention of postsurgical adhesions by Interceed (TC7), an absorbable adhe- sion barrier: a prospective, randomized multicenter clinical study. Fertil Steril 1989;51:9338. [4] Diamond MP. The sepralm adhesion study group. Reduction of adhesions after uterine myomectomy by Sepralm membrane (HAL-F): a blinded, prospective, randomized, multicenter clinical study. Fertil Steril 1996;66:90410. [5] Becker JM, Dayton MT, Fazio VW, Beck DE, Stryker SJ, Wexner SD, Wolff BG, Roberts PL, Smith LE, Sweeney SA, Moore M. Prevention of postoperative abdominal adhesions by a sodium hyaluronate-based bioresorbable membrane: a Fig. 7. Schematic drawing of the tissue reactions against the model site with or without an antiadhesive material. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2907 prospective, randomized, double-blind multicenter study. J Am Coll Surg 1996;183:297306. [6] Stark HH, Boyes JH, Johnson L, Ashworth CR. The use of paratenon, polyethylene lm, or silastic sheeting to prevent restricting adhesions to tendons in the hand. J Bone Jt Surg 1977;59:90813. [7] Haney AF, Hesla J, Hurst BS, Kettel LM, Murphy AA, Rock JA, Rowe G, Schlaff WD. Expanded polytetrauoroethylene (Gore- Tex Surgical Membrane) is superior to oxidized regenerated cellulose (Interceed TC7+) in preventing adhesions. Fertil Steril 1995;63:10216. [8] Jaacobi Y, Israel AA, Goldberg EP. Prevention of postoperative abdominal adhesions by tissue precoating with polymer solutions. J Surg Res 1993;55:4226. [9] Adhesion study group. Reduction of postoperative pelvic adhe- sions with intraperitoneal 32% dextran 70: a prospective, randomized clinical trial. Fertil Steril 1983;40:6129. [10] Goldberg EP, Burns JW, Jaacobi Y. Prevention of postoperative adhesions by precoating tissues with dilute sodium hyaluronate solutions. In: Diamond MP, Reid RL, Di Zerega GS, Linsky CD, editors. Gynecologic surgery and adhesion prevention. New York: Wiley, 1993. p. 191204. [11] Linsky CB, Diamond MP, DeCherney AH, diZerega GS, Cunningham T. Effect of blood on the efcacy of barrier adhesion reduction in the rabbit uterine horn model. Infertility 1988;1:27380. [12] Cha WI, Hyon SH, Ikada Y. Transparent poly(vinyl alcohol) hydrogel with high water content and high strength. Makromo- lekulare Chemie 1992;193:191325. [13] Bakkum EA, van Blitterswijk CA, Dalmeijer RAJ, Trimbos JB. A semiquantitative rat model for intraperitoneal postoperative adhesion formation. Gynecol Obstet Invest 1994;37:99105. [14] Harris ES, Morgan RF, Rodeheaver GT. Analysis of the kinetics of peritoneal adhesion formation in the rat and evaluation of potential antiadhesive agents. Surgery 1995;117:6639. [15] Inoue K, Fujisato T, Gu YJ, Burczak K, Sumi S, Kogire M, Tobe T, Uchida K, Nakai I, Maetani S, Ikada Y. Experimental hybrid islet transplantation: application of polyvinyl alcohol membrane for entrapment of islets. Pancreas 1992;7:5628. [16] Burczak K, Fujisato T, Hatada M, Ikada Y. Protein permeation through poly(vinyl alcohol) hydrogel membranes. Biomaterials 1994;15:2318. S. Matsuda et al. / Biomaterials 23 (2002) 29012908 2908