Experimental investigation of two techniquesof arterial microanastomosis used to managea small-to-large diameter discrepancy*

R.F. Rickard a,c,*, G.H.C. Engelbrecht b, D.A. Hudson c

aDepartment of Plastic Surgery, Derriford Hospital, Plymouth, PL6 8DH, UKbDepartment of Surgery, Division of Surgical Research, University of Cape Town, South AfricacDepartment of Surgery, Division of Plastic, Reconstructive and Maxillofacial Surgery, University of Cape Town,South Africa

Received 30 December 2010; accepted 2 March 2011

KEYWORDSMicrosurgery;Anastomosis;Microanastomosis;Size mismatch;Anastomotic technique

Summary Introduction: A limiting factor in the use of perforators as recipient vessels is thesmall-to-large diameter mismatch often encountered. Mismatches less than 1:1.5 may bemanaged by dilatation of the smaller vessel and by differentially-spaced suture bites. Beyondthis ratio, little evidence exists to direct the choice of end-to-end anastomotic technique.Following in silico work and the characterisation of a rodent superficial caudal epigastric/femoral artery model, we conducted an experimental series examining two techniques - anoblique section of the smaller vessel and invaginating the smaller vessel inside the larger.Materials and methods: A paired design was used. To test for a difference in patency of >5%required a total of 156 animals (312 anastomoses). Side and technique were randomised. Twoinvestigators performed the anastomoses. A single revision was permitted. Anastomoses weretimed and patency was tested at one hour, one week and at six weeks.Results: There was no significant difference in patency at each of the three time points(p Z 0.8026, 0.2963 and 0.8137). The invagination technique was significantly faster toperform (p < 0.0001). There was a significant association between the investigator and bothpatency and the time taken to complete an anastomosis. Independent of the investigator,a revision was more likely to be necessary with the oblique end-to-end technique, and a revi-sion having been performed showed a highly significant association with an anastomosis havingfailed at 1 h (p < 0.0001, OR 33.333).

* This study was presented at the Fifth Congress of the World Society for Reconstructive Microsurgery, Okinawa, Japan, June 2009.* Corresponding author. Department of Plastic Surgery, Derriford Hospital, Plymouth, PL6 8DH, UK. Tel.: þ44 (0) 1752 431519; fax: þ44 (0)

1752 763042.E-mail address: Rory.Rickard@phnt.swest.nhs.uk (R.F. Rickard).

Journal of Plastic, Reconstructive & Aesthetic Surgery (2011) 64, 1088e1095

1748-6815/$-seefrontmatterª2011BritishAssociationofPlastic,ReconstructiveandAestheticSurgeons.PublishedbyElsevierLtd.All rightsreserved.doi:10.1016/j.bjps.2011.03.011

Conclusions: In the management of microarterial size discrepancy between 1:1.5 and 1:2.5, aninvaginating anastomosis is faster to perform and produces comparable patency in a rat model.ª 2011 British Association of Plastic, Reconstructive and Aesthetic Surgeons. Published byElsevier Ltd. All rights reserved.

Introduction

The use of perforators as recipient vessels in microsurgicalbreast reconstruction,1e4 and more latterly in lower limbreconstruction,5 has led to a small-to-large diameterdiscrepancy being found increasingly often, in particularwhen performing an anastomosis between a perforatingand an axial vessel. Some experimental evidence exists tosuggest that arterial anastomotic patency decreases withincreasing diameter mismatch,6e8 and this is corroboratedby some clinical reports.9,10 No good experimental or clin-ical evidence, however, is available to direct the choice ofend-to-end anastomotic technique in this situation.

We have previously attempted to answer the question ofbest method of arterial anastomosis by in silico modellingof flow lines and wall shear stresses in computer-generatedanastomotic models.11 Conclusions from these studies,however, are limited by their inorganic geometries and non-compliant walls, and by the use of Newtonian fluidcharacteristics.

We have subsequently characterized an experimentalanimal model that produces a paired, small-to-large arte-rial diameter discrepancy in the region of 1:1.5 to 1:2.5,without the use of interposition vein grafts.12,13 Insummary, the model uses the femoral axis of outbred Wis-tar (HsdOla:WI or HsdHan!:WIST) strain rats. All axial bloodflow is directed into the Superficial Caudal Epigastric Artery(SCEA), which can then be anastomosed, end-to-end, to thedistal Femoral Artery (FA) to produce a suitable sizemismatch (Figure 1). The model also approximates arterialperforator flow characteristics in man.

Using this model, a study was designed to examine twotechniques of anastomosing arteries with a small-to-largediameter mismatch. The techniques examined were:obliquely sectioning the smaller vessel to produce a largercircumference,14e16 and invaginating the smaller vesselinside the larger.12,17,18

The following null hypothesis was formed: "In anasto-mosis of arteries of unequal diameter, where a small-to-large diameter mismatch exists of between 1:1.5 and 1:2.5,there is no difference in: (a) patency rate, or; (b) speed ofcompletion, between an invaginating anastomosis and anoblique end-to-end anastomosis in a rodent model."

Methods

Statistical design

Because of the paired nature of the study, the McNemartest19 for analysing matched-pair data was used to calcu-late the number of animals needed to test the primaryoutcome measure, i.e. patency. A patency rate differenceof 5% was deemed clinically relevant and pragmatically

possible to test. Power was set at 80% and a significancelevel of 5% was used. Using these parameters, a sample sizeof 156 animals was calculated. Allowing for animal lossesthroughout the study prompted an increase in thesenumbers by 10% to 172 animals.

Animals

Ethical approval for the study was gained from the AnimalEthics Committee of the University of Cape Town. MaleWistar strain rats were obtained from the breeding colonyof the Central Research Facility, University of Stellenbosch.This colony was established in 2001 from HsdOla:WI rats,and maintained as an outbred colony.

Anaesthesia

Animals were weighed and underwent gaseous induction ofanaesthesia. Anaesthesia was maintained with parenteralketamine HCl (Anaket-V", Centaur Laboratories, Bryan-ston, SA) and xylazine HCl (Rompun", Bayer (Pty) Ltd.,Isando, SA). Bupivacaine HCl (Marcaine", Adcock Ingram,Bryanston, SA) was infiltrated locally following woundclosure, and buprenorphine HCl (Temgesic", Schering-Plough, Johannesburg, SA) was administered prior torecovery from anaesthesia. Animals were rehydrated priorto recovery by the administration of 2 mL of 0.9% w/v NaCl(BjBraun Medical (Pty) Ltd., Randburg, Gauteng 2125, SA.http://www.bbraun.co.za) via the dorsal penile vein.

Figure 1 An oblique end-to-end anastomosis between thesuperficial caudal epigastric artery (SCEA) and the distalfemoral artery (FA) in the Wistar rat model used. Flow is fromleft to right. Scale bar Z 1 mm. Narrow arrowhead Z tiearound the FA immediately distal to the origin of the SCEA.Wide arrowhead Z anastomosis.

Experimental investigation of two techniques of arterial microanastomosis 1089

Study design

The desire to facilitate extrapolation of results beyond thoseof a single investigator meant that more than one investi-gator was necessary. The study was therefore divided intotwo, with two investigators each putting 86 animals throughthe study. Order of execution in terms of which side (left orright) was operated on first and which technique (invagina-tion or oblique end-to-end) was used on the first side wasdetermined by computer-generated nested randomisation.Randomisation was concealed until vessels were dissectedfree. The investigator was not randomised by serial number,but carried out the procedures simultaneously, allowing forother research commitments.

If flow through the anastomosis was not observedimmediately following clamp release, a single revision wasperformed. This was to more closely replicate normalclinical practice. At one hour following completion of theanastomosis a single ‘empty-and-refill’ test20e22 was usedto determine patency. At one week and at six weeks,animals were re-anaesthetised and the anastomosis wasdissected free from surrounding tissues. The absence ofthrombus at the anastomosis, judged by microscopic visualinspection, and a successful ‘empty-and-refill’ test wereused to determine patency at these time points. Animalswere killed at this point by an intravenous injection ofsodium pentobarbitone (Euthanaze", Bayer (Pty) Ltd.,Isando, South Africa. http://www.bayer.co.za).

The time taken to complete an anastomosis wasmeasured from division of the vessels to release of theclamps (see Technical Details of Anastomoses, below).

Secondary outcome measures

Secondary outcome measures were:

a. The incidence of a revision being required,b. The influence on patency of a revision having been

performed,c. The influence on patency of the investigator,d. The difference in speed of completion between inves-

tigators, and;e. The influence on patency of the side (left or right) on

which the anastomoses were performed.

Technical Details of Anastomoses

Vessel preparationVessel dissection and preparation was carried out bilater-ally. A binocular operating microscope (SMZ-10, NikonCorporation, Tokyo, Japan. http://www.nikon.com) wasused throughout. The side order and technique were thenassigned from the computer-generated randomisation list.The FA was tied off immediately distal to the origin of theSCEA using 10-O nylon suture material (Dafilon" Black,G1118781, BjBraun Medical (Pty) Ltd.). Care was taken inthe placement of this ligature, the aim being to avoida blind-ended pouch that might cause recirculating flow atthe ostium of the SCEA. The SCEA was then tied off 7 mmdistal to its origin using 10-O nylon suture material. A

double approximating clamp (Acland Model ABB-2V, S&TAG, Neuhausen, Switzerland. http://www.microsurgery.ch)was placed on the FA proximal to the origin of the SCEA,and on the distal FA at its bifurcation into popliteal andsaphenous arteries. For the invagination technique, the FAwas divided immediately distal to the ligature. For theoblique end-to-end technique the FA was divided a further1 mm distal to the ligature. This difference was to allow forthe extra 1 mm of distal FA required to invaginate the SCEAin the invagination technique, in an attempt to remove thepotential confounding factors of differences in anastomotictension and length of distal FA before its bifurcation.

The SCEA was then divided 6 mm distal to its origin. Forthe invaginating anastomosis this was performed as a 90"

transverse division. For the oblique end-to-end anasto-moses, the vessel was transected at a 45" angle.

The approximating clamp was then separated and thelumens of both vessels were flushed with heparinised saline(1000 U heparin in 100ml 0.9%w/v NaCl). In accordance withconventional microsurgical technique, the adventitia wastrimmed back far enough from the vessel ends to prevent anyfronds from entering the vessel lumen. The vessels weredilated with a 0.3 mm diameter vessel dilator (Model D5aZs,S&T AG). In the side undergoing invagination, a cuff of theadventitia of the SCEA was slipped back for a distance of1 mm, in themanner of pulling back a sweater sleeve. A 10-Onylon suture swaged onto a 50 mm, 3 mm 3/8 circle taperpoint round-bodied needle (Dafilon" Black, G1118781,BjBraun Medical (Pty) Ltd.) was used for both techniques.

Invaginating anastomosisThe invaginating anastomosis technique was based on thatdescribed by Duminy.23 It is performed by first passing theneedle through the full thickness of the wall of the FA onelarger vessel diameter from its end (approximately 1 mm inthis model). A small, transverse, tangential bite is thentaken of the end of the SCEA, without breaching the intima(Figure 2(i)). The needle is then passed back through thelumen of the FA and the suture is brought out immediatelybeside itself (Figure 2(ii)). The SCEA is then invaginatedinside the FA by drawing on this suture, and this is veryloosely tied by a single reef knot. It is important to keepthis suture loose in order to prevent crimping of the end ofthe SCEA inside the FA. The approximating clamp is thenturned over and the adventitia of the SCEA is pulledforward to meet the end of the FA. An interrupted suture isthen placed through the end of the FA and through theadventitia and media of the SCEA, at a point 180" from theinvaginating suture, in a manner similar to the ‘sleeve’anastomosis of Duminy (Figure 2(iii)). The clamp is thenreturned to its original position and two further interruptedsutures placed in a similar manner, at 120" from this secondsuture (Figure 2(iv)). These two sutures are additional tothe method of Duminy and are necessary because of thesize discrepancy. Thus a total of four sutures are requiredto complete the invaginating anastomosis.

Oblique end-to-end anastomosisThe oblique end-to-end anastomosis is performed by theuse of interrupted, full-thickness suture bites in theconventional manner, placing the first suture throughthe FA and the heel of the obliquely cut SCEA (Figure 3(i)).

1090 R.F. Rickard et al.

Figure 2 Construction of the invaginating anastomosis. The upstream vessel is on the left in all diagrams. (i) After passingthrough the full thickness of the larger, downstream vessel wall one larger vessel diameter from its end, a small tangential bite istaken of the smaller, upstream vessel. (ii) The suture is passed back out of the larger vessel immediately beside itself and thesmaller vessel is invaginated by drawing on this suture. (iii) A suture taking the full thickness of the larger vessel wall, and onlytunica adventitia and tunica media in the smaller vessel, is placed 180" from the invaginating suture. (iv) Two further sutures areplaced in a similar manner to this second suture, at 120" from it.

Figure 3 Construction of the oblique end-to-end anastomosis. The upstream vessel is on the left in all diagrams. (i) The firstsuture is passed through the ‘heel’ of the obliquely-sectioned smaller, upstream vessel. (ii) The second suture is placed in the ‘toe’.(iii) Back wall sutures placed by Harashina’s method to allow visualisation of the lumen at all times.28 (iv) The front wall iscompleted in a similar manner.

Experimental investigation of two techniques of arterial microanastomosis 1091

The second suture is then placed in the toe of the SCEA(Figure 3(ii)). The approximating clamp is then turned overand two interrupted full-thickness sutures are placed in theback wall, using Harashina’s method to allow visualisationof the lumen at all times (Figure 3(iii)).28 The clamp is thenreturned to its original position and two further sutures areplaced in the front wall in a similar manner. Thus a total ofsix sutures are used to accomplish the oblique end-to-endanastomosis (Figure 3(iv)).

Timing of procedure

Anastomoses were timed from division of the SCEA untilrelease of the clamps by the use of a digital stopwatch.

Statistical analysis

Primary statistical analysis of patency results was per-formed by the McNemar test online at http://www.graphpad.com/quickcalcs. Minitab" 15 for Windows"(Minitab Inc., State College, PA, USA. http://www.minitab.com) was used for all other analyses. Results are reportedas means # SD.

Results

Mean animal body weight was 387.4 # 53.4 g. All 172animals were available for study at one hour. At one week,16 animals had been lost, leaving 156 available for study atthis point. At six weeks, 127 animals were available forstudy.

Patency by technique - primary analysis

Of 344 anastomoses, a total of 18 failed to run at one hour,nine of each technique. At one week, 20 oblique end-to-end and 27 invagination anastomoses had failed. At sixweeks, 14 oblique end-to-end anastomoses had failed, and12 invagination anastomoses. Analysis by McNemar test ofpatency at one hour, one week and at six weeks did notshow a significant difference between oblique end-to-endand invagination techniques (p valuesZ 0.8026, 0.2963 and

0.8137 respectively; Odds Ratios Z 1.000, 0.650 and 1.250respectively) (Table 1).

Patency by technique - secondary analysis

Patency at one hour was further analysed by Binary LogisticRegression, looking at the influence of side, technique,investigator, when in the series the animal was operatedon, and whether a revision of the anastomosis had beenperformed.

Side did not show a significant association with one hourpatency (p Z 0.135). Similarly, technique did not showa significant association with patency (p Z 0.303).

The position in series was analysed by dividing thecomplete series approximately into thirds of 58, 57 and 57animals. The position in series showed a significant associ-ation with patency, when the last 57 animals werecompared to the first 58 (p Z 0.050, OR 6.3). In otherwords, an anastomosis was more likely to fail in the firstthird than in the last, independent of all other factors.There was no significant association between first andmiddle thirds (p Z 0.228).

Whether or not a revision had been performed showeda highly significant association with an anastomosis havingfailed at one hour (p < 0.0001, OR 33.333).

Timing

Timings are plotted at Figure 4. The time taken to completean oblique end-to-end anastomosis varied from 11 min24 sec to 42 min 5 sec. The mean value was 19 min 37 sec(#4 min 26 sec). Time taken to complete the invaginationtechnique varied from 5 min 9 sec to 25 min 52 sec. Themean value was 12 min 44 sec (#3 min 26 sec).

The length of time that it took to complete the anas-tomoses was analysed by a four-factor ANOVA, looking atthe influence of side, technique, investigator and when inthe experimental series the anastomoses were done (theseries again being divided approximately into thirds of 58,57 and 57 animals each).

Side and position in series showed no influence on howlong it took to do an anastomosis (p Z 0.402 and 0.103respectively).

Table 1 Primary analysis of patency, at (a) one hour, (b) one week, and; (c) six weeks (McNemar test with continuitycorrection). Odds ratios given with 95% confidence intervals. A positive result represents a patent anastomosis.

Oblique End-to-End Oblique End-to-End Oblique End-to-End

+ - + - + -

+ 155 8 163 + 116 13 129 + 105 10 115

- 8 1 9 - 20 7 27 - 8 4 12

Inv

ag

ina

tio

n

163 9 172

Inv

ag

ina

tio

n

136 20 156

Inv

ag

ina

tio

n

113 14 127

a. Patency at 1 hour.

p value = 0.8026,

X2 = 0.063, DF = 1.

OR = 1.000 (0.327–3.057)

b. Patency at 1 week.

p value = 0.2963,

X2 = 1.019, DF = 1.

OR = 0.650 (0.297-1.373)

c. Patency at 6 weeks.

p value = 0.8137,

X2 = 0.056, DF = 1.

OR = 1.250 (0.444-3.645)

1092 R.F. Rickard et al.

Technique showed a highly significant influence on thetime taken to perform an anastomosis, independent ofother factors (p < 0.0001, means; Oblique end-to-end Z 19.62 min, Invagination Z 12.73 min, difference inmeans Z 6.89 min).

Revision rate

A total of 37 anastomoses failed to run immediately andwererevised. Of these, 11 occurred after invagination and 26 afteran oblique end-to-end anastomosis. The number of revisionsperformed was analysed by Binary Logistic Regression,looking at the influence of side, technique, and investigator.

Side and investigator did not show any association with arevision having being done (pZ 0.645 and 0.902 respectively).

Technique showed a significant influence on a revisionhaving been performed (p Z 0.010, OR Z 2.66, a revisionwas more likely to have been done with an oblique end-to-end technique).

The likelihood of a revision failing at one hour was ana-lysed by technique. Of the 26 oblique end-to-end anasto-moses that underwent revision, 19 were patent at one hour.Of the 11 invagination anastomoses requiring revision, sixwere patent at one hour. Analysis of this sample of 37 anas-tomoses showed no significant difference in revisionoutcome by technique (p Z 0.4737, Chi-square) (Table 2).

If a revision had not been attempted, and if it is assumedthat anastomotic failure immediately following clamprelease would have remained a failure at one hour, thena further analysis is possible. An additional 19 failures can

Figure 4 Scatterplot of time taken to complete an anastomosis by position in series. Open circles Z Oblique end-to-endtechnique. Filled circles Z Invagination technique. Dotted line Z mean of all oblique end-to-end technique times. Solidline Z mean of all invagination technique times.

Experimental investigation of two techniques of arterial microanastomosis 1093

be included with the eight oblique end-to-end recorded andan additional six with the eight invagination failures. Thisanalysis indicates that the difference in one hour patencybetween the oblique ETE and invagination is not quitestatistically significant (p-value Z 0.061, OR Z 1.929,McNemar) (Table 3).

Influence of the investigator

Who the investigator was showed a significant associationwith one hour patency overall (pZ 0.022, OR 4.545, A moresuccessful than B, Binary Logistic Regression).

Who the investigator was also showed a significantassociation with speed of completion of the anastomoses,independent of all other factors (p Z 0.0037, difference inmeans Z 1.27 min, B faster than A, ANOVA).

Discussion

Patency

The study presented here was designed to test the nullhypothesis that there is less than 5% patency difference

between an invagination technique and an oblique end-to-end technique in rodent arteries with a small-to-largediameter mismatch in the region of 1:1.5e1:2.5. Primaryanalysis has not shown significant patency differences atone hour, one week or at six weeks in this experiment.Secondary analysis of one hour patency has agreed with thisfinding. Null hypothesis (a), therefore, is not rejected.

Secondary analysis of possible confounding factors orinfluences on patency in this study has shown that:

(1) A learning curve existed for both investigators, and;(2) The investigator who took longer to complete an

anastomosis achieved a higher success rate with bothtechniques.

Although success improved overall, between the first 58animals with the last 57, investigators neither slowed down,nor did they become faster at performing an anastomosis.Therefore they must simply have become more dextrous. Itmay be concluded from finding (2) that investigator A tookgreater care in performing both techniques.

In analysis of the revision of unsuccessful anastomoses,the following may be concluded:

(3) A significantly higher revision rate was observed in theoblique end-to-end technique.

(4) If an anastomosis was unsuccessful at the first attempt,it was significantly less likely to be successful ata further attempt, but;

(5) No significant difference in revision success wasobserved between techniques, and;

(6) If in the study a revision had not been permitted,a greater number of invagination anastomoses mayhave been patent at one hour, although this differencewas not quite statistically significant.

Findings (3), (4) and (5), taken together, suggest that theoblique end-to-end technique was technically more diffi-cult to perform in vessels of this calibre, in this model. Thisconclusion agrees with reported opinions of other authorswho have examined invagination techniques in comparisonto the conventional interrupted end-to-end micro-anastomosis of equal-sized vessels.23, 24 However, technicaldifficulties found with the oblique end-to-end were notmore difficult to remedy when compared to the invagina-tion technique, and the overall success rate at one hourfollowing revision did not show a significant differencebetween techniques.

Including successful revisions in the one hour failurenumbers indicate that any difference in one hour patencyshould be considered as not quite statistically significant.This analysis suffers the disadvantage of being based on anassumption that immediate failure would have translatedinto failure at one hour. Furthermore, the experiment wasnot designed with this analysis in mind.

Timing

Independent of all other factors, the invagination tech-nique was faster to perform than the oblique end-to-end. Inthis study, the oblique end-to-end took approximately

Table 3 Analysis of patency at one hour where successfulrevisions have been included in the failure category.A positive result represents a patent anastomosis. P val-ue Z 0.0609, X2 Z 3.512, df Z 1, OR Z 1.929 (95%CI Z 0.976e3.979) (McNemar test).

Oblique End-to-End

+ -

+ 130 27 157

- 14 1 15

Inv

ag

ina

tio

n

144 28 172

Table 2 Two-by-two contingency table for analysis ofrevision successes. X2 Z 0.513; df Z 1. P value Z 0.4737.(Chi-square).

Ob

liq

ue

ET

E

Inv

ag

ina

tio

n

Successful 19 6 25

Unsuccessful 7 5 12

26 11 37

1094 R.F. Rickard et al.

one-and-a-half times longer to complete. Null hypothesis(b) is therefore rejected.

There was a significant difference in cadence betweeninvestigators, which may have influenced patency rates.There was no difference in time taken by position in series.

The finding of less time taken to complete the invagi-nation technique is logical in that four, as opposed to six,sutures were required. Whilst a greater number of suturesare required to complete an invagination technique in thissituation than in equal-sized vessels, this finding is consis-tent with previous reports of invagination techniques inequal-sized vessels.24e27

Conclusions

Within the limits of the study, it can be concluded that theinvagination technique provides at least similar patency tothe oblique end-to-end technique, whilst being easier andsignificantly faster to perform in this model.

Acknowledgements

The authors are grateful to Mr Nolan Hendricks, Mr WillemRyneveldt and Mr Trevor Finch of the Department ofSurgical Research, University of Cape Town for their assis-tance in the experimental series and in animal care andhusbandry. Professor George Gettinby, Professor of Statis-tics and Modelling Science at the University of Strathclydeperformed the statistical analysis.

Conflicts of interest

None.

Funding

Funding was provided in part by the Royal Navy of theUnited Kingdom of Great Britain and Northern Ireland. Theopinions expressed in this article are those of the authorsand not of the UK Ministry of Defence.

References

1. Park MC, Lee JH, Chung J, Lee SH. Use of internal mammaryvessel perforator as a recipient vessel for free TRAM breastreconstruction. Ann Plast Surg 2003;50:132e7.

2. Haywood RM, Raurell A, Perks AG, et al. Autologous free tissuebreast reconstruction using the internal mammary perforatorsas recipient vessels. Br J Plast Surg 2003;56:689e91.

3. Hamdi M, Blondeel P, Van Landuyt K, Monstrey S. Algorithm inchoosing recipient vessels for perforator free flap in breastreconstruction: the role of the internal mammary perforators.Br J Plast Surg 2004;57:258e65.

4. Saint-Cyr M, Chang DW, Robb GL, Chevray PM. Internalmammary perforator recipient vessels for breast reconstruc-tion using free TRAM, DIEP, and SIEA flaps. Plast Reconstr Surg2007;120:1769e73.

5. Hong JP, Koshima I. Using perforators as recipient vessels(supermicrosurgery) for free flap reconstruction of the kneeregion. Ann Plast Surg 2010;64:291e3.

6. Buchler U, Buncke HJ. Experimental microvascular autografts.In: Serafin D, Buncke HJ, editors. Microsurgical compositetissue transplantation. St. Louis: CV Mosby; 1979. p. 83e9.

7. Monsivais JJ. Microvascular grafts: effect of diameterdiscrepancy on patency rates. Microsurgery 1990;11:285e7.

8. Harris JR, Seikaly H, Calhoun K, Daugherty E. Effect of diam-eter of microvascular interposition vein grafts on vesselpatency and free flap survival in the rat model. J Otolaryngol1999;28:152e7.

9. Godina M. Preferential use of end-to-side arterial anastomosesin free flap transfers. Plast Reconstr Surg 1979;64:673e82.

10. Chuang DC, Jeng SF, Chen HT, Chen HC, Wei FC. Experience of73 free groin flaps. Br J Plast Surg 1992;45:81e5.

11. Rickard RF, Meyer C, Hudson DA. Computational modeling ofmicroarterial anastomoses with size discrepancy (small-to-large). J Surg Res 2009;153:1e11.

12. de la Pena-Salcedo JA, Cuesy C, Lopez-Monjardin H. Experi-mental microvascular sleeve anastomosis in size discrepancyvessels. Microsurgery 2000;20:173e5.

13. Rickard RF, Wilson J, Hudson DA. Characterization of a rodentmodel for the study of arterial microanastomoses with sizediscrepancy (small-to-large). Lab Anim 2009;43:350e6.

14. Brener BJ, Raines JK, Darling RC. The end-to-end anastomosisof blood vessels of different diameters. Surg Gynecol Obstet1974;138:249e50.

15. Sabapathy SR. Microsurgery suture techniques. Vessels. In:Wei FC, Mardini S, editors. Flaps and reconstructive surgery.Philadephia: Saunders; 2009. p. 81e92.

16. Daniel RK, Terzis JK. Reconstructive microsurgery. Boston:Little, Brown & Co; 1977.

17. de la Pena-Salcedo JA, Lopez-Monjardin H. Sleeve anastomosisin head and neck reconstruction. Microsurgery 2000;20:193e4.

18. Nakayama Y, Soeda S, Iino T, Uchida A. Is the sleeve anasto-mosis a risky technique? Br J Plast Surg 1987;40:288e94.

19. McNemar Q. Note on the sampling error of the differencebetween correlated proportions or percentages. Psychome-trika 1947;12:153e7.

20. Hayhurst JW, O’Brien BM. An experimental study of microvas-cular technique, patency rates and related factors. Br J PlastSurg 1975;28:128e32.

21. Acland RD. Microsurgery practice manual. St Louis: CV Mosby;1980.

22. Petry JJ, French TS, Wortham KA. The effect of the "patencytest" on arterial endothelial surface. Plast Reconstr Surg 1986;77:960e4.

23. Duminy FJ. A new microvascular "sleeve" anastomosis. J SurgRes 1989;46:189e94.

24. Lauritzen C. A new and easier way to anastomose microvessels.An experimental study in rats. Scand J Plast Reconstr Surg1978;12:291e4.

25. Lauritzen C, Bagge U. A technical and biomechanical comparisonbetween two types of microvascular anastomoses. An experi-mental study in rats. Scand JPlast Reconstr Surg1979;13:417e21.

26. Hyland WT, Botens SR, Minasi JS. A re-appraisal and modifi-cation of the Lauritzen technique of microvascular anasto-moses. Br J Plast Surg 1981;34:451e3.

27. Stamatopoulos C, Biemer E, Stock W, Zechner W. Microvascularanastomosis by invagination: an experimental study. J Car-diovasc Surg (Torino) 1982;23:130e4.

28. Harashina T. Use of the untied suture in microvascular anas-tomoses. Plast Reconstr Surg 1977;59:134e5.

Experimental investigation of two techniques of arterial microanastomosis 1095