R

Id

DY

a

VV

lbacm9m

2

0d

The American Journal of Surgery (2010) 200, 270–275

esearch

ntra-abdominal administration of bevacizumabiminishes intra-peritoneal adhesions

ejan Ignjatovica,*, Kristine Aaslandb, Marianne Pettersena, Stale Sundc,in Chend, Milan Spasojevice, Jens Marius Nesgaarda

Department of Gastrointestinal Surgery, Vestfold Hospital, Tonsberg, Norway; bLaboratory Animal Unit, Norwegian School ofeterinary Science, Oslo, Norway; cDepartment of Pathology, Forde Central Hospital, Forde, Norway; dDepartment of Pathology,

estfold Hospital, Tonsberg, Norway; eNarvik Hospital, University North Norway, Narvik, Norway

AbstractAIM: To determine the effect of a single dose of bevacizumab on adhesion formation in the rat

cecum abrasion model.METHODS: The cecum and parietal peritoneum of 38 male Wistar rats were abraded to promote

adhesion formation. The rats were randomized into 2 groups: group 1 received bevacizumab (2.5mg/kg) intraperitoneally, and group 2 received saline. On day 30 animals were killed, adhesions scored,and histopathological samples taken.

RESULTS: There was no wound dehiscence; there were 2 incision hernias (5.3%), 1 per group.Thirty-seven animals developed adhesions (97.4%). Adhesion grade and severity scores were signifi-cantly different between groups 1 and 2 at 2.7:1.6 (P � .018) and 3.8:2.7 (P � .007), respectively.There was no difference in adhesion square area (27.7:25.0%; P � .16), location (P � 1.00), or number(2.1:1.3; P � .06). Histopathology confirmed the statistical difference between groups (P � .049), anda highly significant correlation between results was shown (r � .758; P � .0001).

CONCLUSION: A single dose of intraperitoneal bevacizumab significantly reduces grade and se-verity of abdominal adhesions in the cecum abrasion rat model.© 2010 Elsevier Inc. All rights reserved.

KEYWORDS:Vascular endothelialgrowth factor;Adhesion;Postoperative;Peritoneum;Wistar rats;Comparative animalstudy

aa

irmbeimiag

Abdominal adhesions remain a common clinical prob-em. They are often the reason for lengthy hospital staysoth in cases of surgery for acute bowel obstruction, as wells for costly, time-consuming diagnostics and treatment ofhronic abdominal pain. In acute small bowel obstructionortality rates can be as high as 30%.1 Data for 1988 reveal

48,000 hospital days and 1.18 billion dollars for this treat-ent.2 This makes the effective prevention and treatment of

* Corresponding author. Tel.: �47 33 34 20 00; fax: �47 33 34 39 45.E-mail address: dejign@siv.noManuscript received February 25, 2009; revised manuscript August 7,

a009

002-9610/$ - see front matter © 2010 Elsevier Inc. All rights reserved.oi:10.1016/j.amjsurg.2009.08.038

bdominal adhesions not only a professional issue but alsofinancial burden for the health system.2

The effect of vascular endothelial growth factor (VEGF)s not confined only to a potent angiogenic cytokine, butather this substance has multiple effects on several crucialechanisms in adhesion formation. The role of VEGF has

een demonstrated in restorative tissue processes such asarly inflammatory responses, wound repair, and remodel-ng through its effect on fibroblast function.3 Leukocytes,acrophages, and T lymphocytes have also been implicated

n this process and data demonstrate that CD4� T cells playcentral role in adhesion formation.4,5 Apart from angio-

enesis, VEGF facilitates increased vascular permeability

nd the deposition of fibrinogen and subsequent cellular

mmb

2f

M

L

aNAfCuUU

A

R3fgSwBdto�tcomir

A

taatSo

S

c

twmigucpitsidNgBAajtna

P

sggiwdc

A

adoiebaa

A

w(Tgnd

271D. Ignjatovic et al. Bevacizumab reduces peritoneal adhesion formation

igration. Bevacizumab is a recombinant humanizedonoclonal IgG1 antibody that binds to and inhibits the

iological activity of human VEGF.6

The aim of this study is to evaluate the effect of a single.5-mg/kg intraperitoneal dose of bevacizumab on adhesionormation in the cecum abrasion rat model.

aterials and Methods

icense

The study was performed at the Laboratory Animal Unitt the Norwegian School of Veterinary Science in Oslo,orway. The animal unit is licensed by the Norwegiannimal Research Authority (NARA) and the Association

or Assessment and Accreditation of Laboratory Animalare (AAALAC). The study was approved by the animalnit local competent person, by NARA, and by the Animalnit Ethics Committee of the Institutional Animal Care andse Committee (IACUC).

nimals, housing, and husbandry

The animals were 42 male Wistar rats from Charlesiver Laboratories, 12 to 13 weeks of age, weighing 250 to00 g. The rats were housed in type IV Macrolon rat cages,rom North Kent Plastic Cages (Rochester, Kent, UK), inroups of 4. They lived on standard aspen bedding fromcanbur BK (Nittedal, Norway) and they were given Celle-att nesting material, a red Teckniplast tunnel from ScanburK, and a gnawing-stick in wood. They were given stan-ard SDS feed from Scanbur BK and tap water (which isested for microbes once a month) ad libitum. The light wasn from 8:00 AM to 8:00 PM and the temperature was 21°C

2°C; the room had 20 air changes within 24 hours andhe humidity was 45 � 5%. The cages and bedding werehanged 2 times per week and the water was changednce per day by a skilled technician. The rats werearked on the tail with green permanent marker to identify

ndividual animals. The rats were acclimatized to the envi-onment at the unit for 2 weeks before surgery.

nesthesia

The rats were fasted from the night before surgery. Anes-hesia was obtained by isoflurane gas, given in a chamber usingirflow 200 mL and 2.5% isoflurane. When anesthesia waschieved the rats were moved to a mask giving isoflurane andhen they were given Temgesic (Buprenorfin 0.3 mg/mL,chering-Plough, Brussels, Belgium) .05 mg/kg subcutane-usly (SC), carprofen 5 mg/kg SC, and NaCl 30 mL/kg SC.

urgery

All surgical procedures were performed under aseptic

onditions by a single surgeon (D.I.). The adhesion forma- s

ion was induced between the cecum and anterior abdominalall through a mini-laparotomy. A standardized lesion wasade both on the wall of the cecum and on the correspond-

ng parietal peritoneum through stroking and scraping withauze and pinching of the serosal surface with tweezersntil local hemorrhage was induced to the site of injury. Theecum was placed back into the abdomen. A catheter waslaced through the abdominal wall, outside the laparotomyn the superior right abdominal quadrant, and then the peri-oneum, fascia, and skin were closed with 2/0 absorbableutures. The skin was sutured intracutaneously. After clos-ng the abdomen, the test material was injected intra-ab-ominally through the catheter. Group A rats were givenaCl and represent the control group. Group B rats wereiven 2.5 mg/kg bevacizumab (Avastin, Roche 25 mg/mL,asel, Switzerland) diluted to 5 mg/mL with .9% NaCl.fter the injection, the catheters were removed and the

bdomen massaged to achieve equal spreading of the in-ected material. The duration of the total procedure, andherefore the duration of the anesthesia, was standardized too more than 30 minutes. Animals that died were immedi-tely replaced and operated on the same day.

ostoperative care

The rats were allowed to awake from the anesthesia in aingle cage with a towel and a warming bottle. All rats wereiven a second injection of Temgesic 6–8 hours after sur-ery, on the evening of the day of surgery, and a thirdnjection on the morning of the day after surgery. The ratsere observed closely by a skilled technician during theays after surgery to look for symptoms of pain and toheck the wound for swelling and bleeding.

fter surgery

Four weeks after surgery the rats were given isofluranenesthesia and then humanely killed by cervical dislocationuring anesthesia. They were opened through the excisionf the anterior abdominal wall, and the adhesions weredentified and graded. The abdominal organs were thenxcised from the posterior abdominal wall and placed in 4%uffered formaldehyde solution together with the anteriorbdominal wall with undisturbed adhesions. Photographs ofdhesions were taken during the procedure.

dhesion assessment

Adhesions were assessed between organs and abdominalall and between the organs themselves by 2 surgeons

M.P. and D.I.) independently for location, extent, and type.he procedure of adhesion assessments included photo-raphy, discussion, and staging through consensus. Theumber of adhesions between organs/abdominal wall wasetermined by counting individual adhesion bands. For as-

essment of the extent of the adhesions between organs and

at(saiwatcs

H

mf

ssclaTsiAbdpgg(ts

Flofi(

272 The American Journal of Surgery, Vol 200, No 2, August 2010

bdominal wall, the latter was divided into 4 quarters, andhe presence of adhesions in these quarters was noted�25%, 25%–50%, 50%–75%, �75%). The type of adhe-ions was determined according to the method of Zühlke etl,7 whereby grade 0 indicates no adhesions and grade IVndicates firm extensive adhesions that are only dissectibleith sharp instruments, with organ damage almost unavoid-

ble. If different adhesion types were present, the highestype was scored. The severity of adhesion formation wasalculated by multiplying the extent and type of adhe-ions.8,9

istology

The en block removed and formaldehyde fixed speci-ens were then re-examined by 1 of the pathologists (Y.C.)

or adhesion formation. Representative samples from adhe-

→

→*

*c

►

►

*

A

→

→

→

C

igure 1 (A) Fibrous band of adhesion. Granulation tissue wymphocytes (arrowheads) and mast cells (asterices). Histologicalriginal magnification � 200. (B) Fibrous adhesion between outbrosis). HE, original magnification � 40. (C) Scarce fibrous ad

arrows), without significant granulation tissue. Histological score 1 (sca

ions were taken out after staining with dye and the tissueamples embedded in paraffin blocks. Sections for micros-opy were cut at 4 �m and stained according to standardaboratory procedure with hematoxylin–eosin (HE) in anutomated stainer (Leica, ST 5020, Nussloch, Germany).wo HE slides were prepared from each block, and deeperections were taken in selected cases. All cases were exam-ned by light microscopy by 2 of the authors (S.S. and Y.C.).dherences were categorized as histological types I–IVased on the presence and extent of fibrosis. Grade I wasefined as scarce fibrosis of localized extent, grade II im-lied scarce but extensive fibrosis (at least 2 fibrous bands),rade III indicated pronounced but localized fibrosis, andrade IV was pronounced and extensive fibrosis. Significantextensive) fibrosis was defined as the presence of granula-ion tissue with distinct capillaries within the area of adhe-ion (Fig. 1A and B). The fibrotic areas often contained

→←

B

apillaries (arrows) and bundles of collagen (C). Infiltration of4 (severe fibrosis, more than 1 focus). Hematoxylin-eosin (HE),rs of gut wall (arrows). Histological score 3 (1 focus of severebetween gut wall and serosal-layered abdominal adipose tissue

ith cscore

er layehesions

rce fibrosis in 1 focus). HE, original magnification � 200.

dligmfifiaspwm

sbwsws

S

iotcepw

S

Sw

wnTb

R

dgswmfsdphidstsiagnsnah(g

m

Fwf

Fw

273D. Ignjatovic et al. Bevacizumab reduces peritoneal adhesion formation

istinct collagen fibers and mild–moderate infiltration ofymphocytes; mast cells within the infiltrates were strikingn some cases. Another common finding was foreign bodyranulomas with giant cells containing highly birefringentaterial in polarized light, probably derived from gauzebers; these granulomas occurred sometimes within thebrous adhesions and sometimes along the serosal surface,nd were not used for lesion scoring. Cases of scarce fibro-is were characterized by subtle changes only, with inter-osition of abdominal fat tissue between serosal surfaces,ith only small and discrete areas of fibrosis with no orinimal granulation tissue (Fig. 1C).Histological scoring was performed independently of

urgical grading. All cases were examined microscopicallyy each of the pathologists; thereafter, the series of slidesere reviewed by both together. Consensus on the final

core was achieved in all cases. If several types of adhesionsere present in 1 animal, all were noted; however, the most

evere was used for the statistical analysis.

tudy design

The randomization method was block randomization us-ng Random Allocation software v 1.00 (Isfahan Universityf Medical Sciences, Iran). The researchers were blinded tohe treatment and control groups. The sample size wasalculated with SSD v 7.3 software (Henrik Lehman, Haug-sund, Norway). The sample size for a 2-sided test, studyower 80%, to detect a 20% difference between the groupsas 38 (19 animals in each arm).

tatistics

Analysis was done using the Statistical Package for theocial Sciences (SPSS, Chicago, IL) software. Percentages

igure 2 Adhesions between the cecum and anterior abdominalall, macroscopic grade 2 (arrow). The laparotomy can be seen

ree of adhesions.

ere compared using Student t test and continuous variables o

ith the Mann–Whitney test. Chi-square test was used foronparametric values and Spearman test for correlation.he P values given are 2-sided; P � .05 was considered toe the limit of significance.

esults

A total of 42 rats were operated. Four animals died, 2uring anesthesia and 2 in the following 2 days after sur-ery. All were immediately replaced but not included in thetudy. There were no symptoms before death; all 4 under-ent necropsy without significant findings. Serology forycoplasma, carbacillus, Sendai, and pneumonia virus

rom mice (PVM) were negative. Deaths were not related tourgery. The remaining 38 animals recovered without inci-ent and resumed preoperative physical activity and feedingatterns by postoperative day 2. There was no wound de-iscence; 2 animals developed an incision hernia (5.3%), 1n each group. All animals, except 1, developed intra-ab-ominal adhesions (97.4%). Most animals developed adhe-ions between the cecum and small bowel (73.7%), omen-um (47.3%), and abdominal wall (23.6%) and there was notatistical significant difference between the groups accord-ng to location of the adhesions (Figs. 2 and 3). However,dhesion grades were significantly different between theroups (Fig. 4) (mean grade 2.7:1.6; P � .018). There waso statistical difference in the area occupied by the adhe-ions (mean surface 27.7:25.0%; P � .1628), nor in theumber of adhesions (mean number 2.1:1.3; P � .064),lthough this difference was nearly significant. The ad-esion severity score gave a high significant differencemean score � area 3.8:2.7; P � .0086) between theroups (Fig. 5).

All specimens were eligible for histopathology. Theean adhesion grade at histology was 3.55 (SD .69) and

igure 3 Adhesions of the omentum to the anterior abdominalall, macroscopic grade 3 (arrow). The adhesion is to the laparot-

my site.

2ddp.sn1

1tw(w

C

btda

btilfsso

wptrwlrarciada

ftb

FTg

Fga

Ftl

274 The American Journal of Surgery, Vol 200, No 2, August 2010

.63 (SD 1.46) for control and test groups, respectively. Theifference between the groups was statistically significantlyifferent (P � .049). Moreover, a highly significant andositive correlation has been demonstrated (r � .758; P �

0001) between the macroscopic grading and the micro-copic grading (Fig. 6). Mean histological grade was sig-ificantly higher (P � .0001) than macroscopic grading for.022 grades.

Granuloma formation was noted in 23 (60.5%) animals,2 (31.5%) in the control group and 11 (28.9%) in thereatment group. Granulation tissue within the adhesionsas noted in 31 (81.6%) animals, 15 (39.5%) and 16

42.1%) in the control and test groups, respectively. Thereas no statistical difference between groups.

igure 4 Y axis: adhesion grade. X axis: C � control group;� test group. The test group had a significantly lower adhesion

rades, P � .018.

igure 5 Y axis: adhesion severity score. X axis: C � controlroup; T � test group. The test group had a significantly lower

vdhesion severity score, P � .0086.

omments

The main finding of this study is that a single dose ofevacizumab, consisting of a half therapeutically adminis-ered dose for the treatment of patients with metastaticisease, seems to hinder the development of abdominaldhesions.

Although there was a statistically significant correlationetween the surgeons’ adhesion scorings and the his-opathological grading of fibrosis, there were cases of def-nite discrepancy between the 2. In general, sampling prob-ems for histology can explain such cases. For example, aew very focal adhesions detected on the fixed organ blockeem to have been overlooked by the surgeons on the freshpecimen, while some adhesions can be too thin to sampleptimally.

The overall data collected in this study correlate wellith data previously published. Two similar, previouslyublished studies were performed with species specific an-ibodies to VEGF; in these studies, the test period wasestricted to 7 and 14 days.10,11 Our study was performedith a humanized antibody, in a species where abundant

iterature suggests a similarity in effect of bevacizumab inats and humans.12,13 Our follow-up period was 4 weeks,nd the adhesion maturation process can be affected by theeabsorbed circulating bevacizumab since it remains in cir-ulation up to 6 weeks. Compared with the previous stud-es,10,11 our findings may imply a lesser area of adhesionsnd a lower adhesion maturation level. It is difficult toetermine whether this is the effect of the reabsorbed bev-cizumab, which remains to be tested in further studies.

The change in the perception of abdominal adhesionsrom inert to dynamic structures seems to have occurred athe shift of the century. Human peritoneal adhesions haveeen proved to be highly cellular, vascularized, and inner-

igure 6 Y axis: macroscopic adhesion score. X axis: his-opathological adhesion score. A highly significant positive corre-ation between the groups was found (r � .758; P � .0001).

ated.14-17 The presence of blood vessels and nerves (both

msfpwecctri

astwobaaafibufmsVa

doWmp

A

dAftF

R

1

1

1

1

1

1

1

1

1

1

2

275D. Ignjatovic et al. Bevacizumab reduces peritoneal adhesion formation

yelinated and nonmyelinated) implies that mature adhe-ions are more permanent structures, easily differentiatedrom early adhesions. Furthermore, animal studies haverovided a timetable for vessel and nerve ingrowth to 2 to 4eeks.15 The results of our study also show that no differ-

nce was found between the groups as far as wound dehis-ence and abdominal wall hernia are concerned. These dataorrelate well with conclusions of clinical studies, showinghat selective therapeutic targeting of VEGF does not dis-upt operative wound healing in a clinically important fash-on.18

It has been demonstrated that fewer adhesions developfter laparoscopic bowel resection19 and laparoscopic adhe-iolysis20 than after conventional surgery. The type ofrauma inflicted to the cecum in our study correlates wellith the type of trauma inflicted while performing adhesi-lysis, where the surgeon usually does not use diathermyut rather blunt dissection and scissors. In this context, theddition of a single intraperitoneal dose of bevacizumabfter performing a safe laparoscopic adhesiolysis could bettempted for the prevention of adhesions in patients suf-ering from chronic abdominal pain and/or in those suffer-ng from recurrent small bowel obstructions. The fact thatevacizumab has unwanted side effects does not exclude itsse but rather narrows the patient group that could benefitrom a single intra-abdominal application, and can causeodifications in the operative technique (eg, use of muscle

plitting trocar insertion). The clinical success and safety ofEGF neutralization in the treatment of malignant diseases

dds further momentum to such a statement.From our study, we conclude that a single intraperitoneal

ose of bevacizumab diminishes both the grade and severityf abdominal adhesions in the rat cecum abrasion model.e consider this model of value for further studies on theechanisms involved in adhesion prevention within the

eritoneum.

cknowledgment

Special thanks to Dr Erik Skaaheim Haug and Dr Half-an Aass (Vestfold Hospital) for their support and to Prof.. Smith from the Norwegian School of Veterinary Science

or his unselfish advice and help. For technical assistance,he authors thank Aslaug Vagstad and Sverre Nordgård,orde Central Hospital.

eferences

1. Agresta F, Piazza A, Michelet I, et al. Small bowel obstruction.

Laparoscopic approach. Surg Endosc 2000;14:154–6.

2. Ray NF, Larsen JW Jr, Stillman RJ, et al. Economic impact of hos-pitalizations for lower abdominal adhesiolysis in the United States in1988. Surg Gynecol Obstet 1993;176:271–6.

3. Diamond MP, El-Hammady E, Munkarah A, et al. Modulation of theexpression of vascular endothelial growth factor in human fibroblasts.Fertil Steril 2005;83:405–9.

4. Ar’Rajab A, Mileski W, Sentementes JT, et al. The role of neutrophilsin peritoneal adhesion formation. J Surg Res 1996;61:143–6.

5. Chung DR, Kasper DL, Panzo RJ, et al. CD4� T cells mediate abscessformation in intra-abdominal sepsis by an IL-17-dependent mecha-nism. J Immunol 2003;170:4411.

6. Cahill RA, Redmond HP. Cytokine orchestration in post-operativeperitoneal adhesion formation. World J Gastroenterol 2008;14:4861– 6.

7. Zühlke HV, Lorenz EM, Straub EM, et al. [Pathophysiology andclassification of adhesions]. Langenbecks Arch Chir Suppl II VerhDtsch Ges Chir 1990:1009–16.

8. Swank DJ, Hop WC, Jeekel J. Reduction, regrowth, and de novoformation of abdominal adhesions after laparoscopic adhesiolysis: aprospective analysis. Dig Surg 2004;21:458.

9. Swank DJ, Swank-Bordewijk SC, Hop WC, et al. Laparoscopic ad-hesiolysis in patients with chronic abdominal pain: a blinded random-ised controlled multi-centre trial. Lancet 2003;361:2250.

0. Saltzman AK, Olson TA, Mohanraj D, et al. Prevention of postoper-ative adhesions by an antibody to vascular permeability factor/vascularendothelial growth factor in a murine model. Am J Obstet Gynecol1996;174:1502–6.

1. Molinas CR, Binda MM, Carmeliet P, et al. Role of vascular endo-thelial growth factor receptor 1 in basal adhesion formation and incarbon dioxide pneumoperitoneum-enhanced adhesion formationafter laparoscopic surgery in mice. Fertil Steril 2004;82 (Suppl3):1149 –53.

2. Bäuerle T, Hilbig H, Bartling S, et al. Bevacizumab inhibits breastcancer-induced osteolysis, surrounding soft tissue metastasis, and an-giogenesis in rats as visualized by VCT and MRI. Neoplasia 2008;10:511–20.

3. Raatschen HJ, Simon GH, Fu Y, et al. Vascular permeability duringantiangiogenesis treatment: MR imaging assay results as biomarker forsubsequent tumor growth in rats. Radiology 2008;247:391–9.

4. Herrick SE, Mutsaers SE, Ozua P, et al. Human peritoneal adhesionsare highly cellular, innervated, and vascularized. J Pathol 2000;192:67–72.

5. Sulaiman H, Gabella G, Davis C, et al. Growth of nerve fibres intomurine peritoneal adhesions. J Pathol 2000;192:396–403.

6. Jin K, Zhu Y, Sun Y, et al. Vascular endothelial growth factor (VEGF)stimulates neurogenesis in vitro and in vivo. Proc Natl Acad SciU S A 2002;99:11946 –50.

7. Sulaiman H, Gabella G, Davis C, et al. Presence and distribution ofsensory nerve fibers in human peritoneal adhesions. Ann Surg 2001;234:256–61.

8. Scappaticci FA, Fehrenbacher L, Cartwright T, et al. Surgical woundhealing complications in metastatic colorectal cancer patients treatedwith bevacizumab. J Surg Oncol 2005;91:173–80.

9. Schippers E, Tittel A, Ottinger A, et al. Laparoscopy versus laparot-omy: comparison of adhesion-formation after bowel resection in acanine model. Dig Surg 1998;15:145–7.

0. Tittel A, Treutner KH, Titkova S, et al. Comparison of adhesionreformation after laparoscopic and conventional adhesiolysis in an

animal model. Langenbecks Arch Surg 2001;386:141–5.