Int. J. Oral Maxillofac. Surg. 2003; 32: 188–197doi:10.1054/ijom.2002.0365, available online at http://www.sciencedirect.com

Research paperTissue Engineering

Intraoral grafting of an ex vivoproduced oral mucosaequivalent: a preliminary reportK. Izumi, S. E. Feinberg, A. Iida, M. Yoshizawa: Intraoral grafting of an ex vivoproduced oral mucosa equivalent: a preliminary report. Int. J. Oral Maxillofac. Surg.2003; 32: 188–197. � 2003 International Association of Oral and MaxillofacialSurgeons. Published by Elsevier Science Ltd. All rights reserved.

Abstract. The objective of this study was to assess the efficacy of the use of anex vivo produced oral mucosa equivalent (EVPOME) for intraoral graftingprocedures. Autogenous keratinocytes were harvested from a punch biopsy 4 weeksprior to surgery, placed in a serum-free culture system and seeded onto a humancadaveric dermal equivalent, AlloDerm�. Thirty patients with either apremalignant or cancerous lesion were triaged into two groups, depending on thestage of disease: Group 1: EVPOME or Group 2: AlloDerm�, control without anepithelial layer. Clinically, EVPOME grafts were easy to handle and showedexcellent compliance on grafting. Both, EVPOME and AlloDerm� grafts, showeda 100% take rate. At 6 days post-grafting, the EVPOME clinically showed changesindicating vascular ingrowth and had cytologic evidence of the persistence ofgrafted cultured keratinocytes on the surface. The EVPOME grafts had enhancedmaturation of the underlying submucosal layer associated with rapid epithelialcoverage when compared to the AlloDerm� grafts at biopsies taken at 28 dayspost-grafting. In summary, EVPOME appears to be an acceptable oral mucosalsubstitute for human intraoral grafting procedures and results in a more favorablewound healing response than AlloDerm� alone.

K. Izumi1,2, S. E. Feinberg2, A. Iida1,M. Yoshizawa1

1Department of Oral and Maxillofacial Surgery,Course for Oral Life Science, NiigataUniversity Postgraduate School of Medicaland Dental Sciences, Japan; 2Department ofOral and Maxillofacial Surgery, University ofMichigan Medical Center, Ann Arbor, MI, USA

Key words: tissue-engineering; oral mucosaequivalent; AlloDerm®; oral keratinocytes;human clinical trial.

Accepted for publication 17 December 2002

Introduction

Preprosthetic and reconstructive pro-cedures often result in open wounds inthe mouth during intraoral surgery. Afree mucosal graft used for reconstruc-tion or closure of the open wounds canprevent microbial infection, excessivefluid loss, and foreign material contami-nation or relapse secondary to woundcontracture. Oral mucosa or skin graftsboth require harvesting of tissue from asecond surgical site resulting in increasepatient morbidity. In addition, oralmucosa is limited in supply while split-thickness skin grafts, though more easily

0901-5027/03/020188+10 $30.00/0 � 2003 Internatio

available, contain adnexal structures,and express a different pattern of kerati-nization9. Technological advancementsin the field of tissue-engineering haveallowed the development of innovativeapproaches to the fabrication of skinand mucosal substitutes17. Over the lastdecade, two prevalent skin substitutemethods have been frequently used inoral and maxillofacial surgery. One is amucosal epithelial sheet fabricated bythe technique developed by G andR4. However, several dis-advantages of the use of sheets of epi-thelial cells exist. First is the difficulty inhandling of the friable epithelial sheets.

nal Association of Oral and Maxillofacial Surgeons. Pu

Second is the use of an irradiated xeno-geneic 3T3 mouse fibroblast feeder layerthat is necessary to enhance growth ofcultured keratinocytes. Unfortunately,the feeder layer has the potential totransfer xenogenous components to theco-cultured autologous cells of thegrafted patients4,18,19. A second skinsubstitute technique uses a bi-layeredartificial dermis that consists of collagen-glycosaminoglycan/silastic sheet1,14 thatis an ‘off-the-shelf’ material that isavailable at the time of surgery. Itsmajor shortcomings are its difficultywith suturing and its susceptibility toinfection17.

blished by Elsevier Science Ltd. All rights reserved.

Intraoral grafting of an ex vivo produced oral mucosa equivalent 189

Our research team has been able toovercome these disadvantages by thedevelopment of a human ex vivoproduced oral mucosa equivalent(EVPOME) in a serum-free culture sys-tem without a feeder layer that has excel-lent handling characteristis9. OurEVPOME is composed of a stratifiedlayer of human autogeneous oral kerati-nocytes seeded on top of an acellular,human cadaver dermis, AlloDerm�. It isthe use of AlloDerm� as the dermalbase that confers enhanced handling

characteristics and assists in complianceof the graft.

The elective nature of maxillofacialsurgical procedures allows the flexibilityand timing to coordinate the manufac-ture of a human ex vivo producedengineered-tissue of adequate dimen-sions within sufficient time to allow itsuse in the reconstruction of mucosaltissues within the oral cavity.

The objective of this study wasto evaluate the efficacy and clinicalhandling of EVPOME for intraoral

reconstruction. To our knowledge, thisis the first reported clinical study usinga tissue-engineered ex vivo producedhuman oral mucosa composed of bothan epithelial and dermal component forintraoral grafting procedures with anexperimental protocol that includes aparallel control group, AlloDerm�,without epithelium.

Experimental design andmethodology

Fabrication of Ex vivo Produced OralMucosa Equivalent (EVPOME)

The development and characteristicsof EVPOME have been previouslydescribed9. EVPOME is a compositehuman oral mucosa equivalent consist-ing of a dermal component composed ofa human cadaver dermis, AlloDerm�

(LifeCell, Branchburg, NJ, USA) that isseeded with autogenous human oralkeratinocytes to form an overlyingstratified parakeratinized epitheliallayer. Briefly, our protocol for the manu-facture of the EVPOME is the following.First, in the out patient clinic, underlocal anaesthesia, a 5�5 mm punchbiopsy of keratinized oral mucosa istaken from an uninvolved site in the oralcavity. The biopsy is taken at a timepoint prior to surgery to allow fabrica-tion of an EVPOME that has appropri-ate dimensions for the planned surgicalreconstruction. Oral keratinocytes aredissociated from the biopsy specimenand expanded in a serum-free definedculture medium, MCDB153 (Sigma, StLouis, MO, USA) containing a calciumconcentration of 0.06 mM and porcinepituitary extract (COSMO BIO, Tokyo,Japan). The AlloDerm� is pre-soakedin 5 �g/cm2 human type IV collagen(Becton Dickinson Labware, Bedford,MA, USA) for 3 h prior to seeding toenhance adherence of the seeded kerati-nocytes (1.25�105 cells/cm2). Once asufficient number of oral keratinocyteshas been harvested, they are seededonto the collagen coated AlloDerm�

(thickness of 0.007 to 0.020 in). Thecomposites of oral keratinocytes andAlloDerm� are then cultured, sub-merged, for 4 days to form a continuousepithelial monolayer. At this time, theconcentration of calcium in the culturemedium is raised to 1.2 mM to enhancekeratinocyte differentiation. When theequivalents are raised to an air-liquidinterface to encourage stratification ofthe epithelial monolayer, they are cul-tured for an additional 7 days resulting

Fig. 1. A. Photomicrograph of day 11 EVPOME illustrating epithelial cell stratification withevidence of parakeratinization (haematoxylin and eosin staining, original magnification(�250). B. EVPOME illustrating its tensile strength.

190 Izumi et al.

in the formation of a parakeratinizedlayer in preparation for intraoral graft-ing of the EVPOME (Fig. 1A).

This optimal fabrication protocol forthe EVPOME, i.e., the number of daysin culture, was determined by experi-mental studies that involved the graftingof different stages of epithelial develop-ment of the EVPOME into SCID mice.Grafted EVPOMEs were then evaluatedfor microvessel ingrowth within theunderlying AlloDerm� layer, sincere-vascularization is an important factorin free graft survival8. At this stage ofdevelopment the EVPOME possessedexcellent handling characteristics thatallowed it to be easily transferred fromthe culture flask to the oral cavity (Fig.1B). These EVPOME grafts have beenreferred to as day 11 grafts, and day 10to 14 grafts are the ones that were usedin this reported study.

Experimental design

All patients enrolled in the study, atNiigata University, signed a consentform in which they agreed to be part ofthe investigation to evaluate EVPOMEsfor intraoral grafting procedures. Studysubjects were selected from patientsdiagnosed with premalignant lesions ororal cancer who were to undergo surgi-cal excision of the lesion followed byreconstruction. Patients were triagedto Group 1, EVPOMEs, or Group 2,AlloDerm�, control group without anepithelial layer, depending on theurgency of surgery, since the EVPOME

required additional time to fabricate.The use of a control group, AlloDerm�

without an epithelial layer, allowed us toassess the efficacy of the use of a strati-fied layer of oral epithelial keratinocyteson the dermal equivalent.

Inclusion and exclusion criteria

Inclusion criteria were: (1) over 18 yearsof age and (2) diagnosed as oral squa-mous cell carcinoma or oral premalig-nant lesion such as dysplasia orcarcinoma in situ. Exclusion criteriawere: (1) serologically positive for syphi-lis, HIV or Hepatitis B or C virus, (2)uncontrolled diseases such as diabetes,pregnancy, and not medically able towithstand a general anaesthetic (ASAclass III or higher), (3) immuno-deficiency, (4) patients receiving radia-tion therapy, and (5) cases wherethe surgical defect could be repaired,primarily, without tension.

Study population

Fifteen patients were recruited for eachgroup, for a total of 30 patients. Patientswere triaged to each group depending onthe urgency for need of the surgicalexcision and physical condition of sub-ject so as not to compromise patientcare.

Demographics of Group 1 (graftedwith EVPOME) were five males, and tenfemales. The ages of patients rangedfrom 49 to 84 years with a mean age of64.2�2.7 years. Lesions were located in

the tongue in 13 cases and alveolar gin-giva in five cases (locations in three casesof oral leukoplakia were multiple). Thelesions were histologically diagnosed assquamous cell carcinoma in six cases (Tstage: T3=1, T2=1, T1=4), epithelialdysplasia in eight and oral lichen planusin one. Eight patients were non-smokers,three had a positive smoking history butdid not smoke at the time of surgery andfour were current smokers in which onepatient was classified as a heavy smoker(An index calculated by ‘number of cur-rently smoked cigarettes per day’�‘years of smoking at the time of excision’is 1000 or higher). In contrast, Group 2(AlloDerm� without seeded oral kerati-nocytes) contained ten males, and fivefemales. The ages of patients rangedfrom 41 to 84 years with a mean age of68.0�3.0 years. Lesions were located inthe tongue in 11 cases (T stage: T2=4,T1=7) as well as the buccal mucosa,floor of mouth, oropharyngeal mucosaand alveolar gingiva. Histological diag-noses of the lesions were; 11 cases ofsquamous cell carcinomas, four casesof epithelial dysplasias with two of thesecases presenting with melanotic pig-mentation. Eleven patients were non-smokers and four were current smokerswith one patient classified as a heavysmoker (Table 1).

Treatment protocol and follow-up

All oral lesions were resected with theuse of electrocautery. In all cases surgi-cal margins of 10 mm around cancer or5 mm around premalignant lesion weredone resulting in exposure of either theunderlying muscle or periosteum. Bothgroups, EVPOME or AlloDerm�, with-out epithelium, grafts were secured inplace with 4-0 black-silk sutures andthen stabilized by a gauze bolster press-ure dressing (Fig. 2) or with a surgicalstent.

Patients were transnasally fed via anasogastric tube until postoperative day7 and then placed on a soft diet. At day6 postoperatively, the bolster or stentwas removed and at postoperative day 7the sutures were removed. On average,patients in both groups were discharged2 weeks after surgery.

Outcome measures

At day 6, postoperatively, after removalof the gauze bolster or surgical stent, themost central portion of the surface of thegraft was scraped for cytological exami-nation to assess for the presence of

Table 1. Research subjects of the EVPOME (Group 1) and AlloDerm� alone (Group 2).

Group 1EVPOME (15 pts)

Group 2AlloDerm�alone (15 pts)

Mean age (range) 64.2�2.7 (49 to 84) 68.0�3.0 (41 to 84)

Male/female ratio 5/10 10/5

Site Tongue (13)* Tongue (11)Alveolar gingiva (5)* Buccal mucosa (1)

Floor of mouth (1)Alveolar gingiva (1)

Oropharyngeal mucosa(1)

Diagnoses Sq. cell carcinoma (6) Sq. cell carcinoma (11)Epithelial dysplasia (8) Epithelial dysplasia (4)Oral lichen planus (1)

Smoking history Non-smoker (8) Non-smoker (11)Previous smoker (3)† Current smoker (4)‡Current smoker (4)‡

Surface area (cm2) to be covered 12.41�1.70 11.28�1.83

Follow-up (range) 5–17 months 4–18 months

*Several patients had multiple sites thus number of sites is greater than 15.†Previous history of smoking; presently a non-smoker.‡One smoker is this group of four was classified as a ‘heavy smoker’ (an index showed 1000 orhigher).

Intraoral grafting of an ex vivo produced oral mucosa equivalent 191

grafted keratinocytes. Grafted sites of allpatients were evaluated by clinical obser-vation and photographs every day untildischarge. Three variables were evalu-ated to assess patient and graft accept-ance. First, postoperative pain wasassessed by a visual analog scale (VAS)(� =0, + =1–24, + + =25–49, + + +=50–100). Secondly, ‘pull off strength’was used to evaluate graft adhesivenessto the underlying layer at the peripheryof the wound margin at day 7 postopera-tively (� =non-adherent, + =adherentbut easily removed, + + =adherent diffi-cult to remove, + + + =adherent unable

to remove). Third, the percent of epi-thelial coverage of the graft, clinicallyevident at 4 weeks after transplantation,was calculated by the formula of A-B/A(%). ‘A’ was the wound area trans-planted with EVPOME or AlloDerm�

and ‘B’ was the area with exposedAlloDerm� (lack of epithelial coverage).Areas of the grafts that had epithelialcoverage appeared red, while white oropaque regions at the grafted site wereseen with AlloDerm� which lacked anepithelial layer (� =0–50%, + =51–75%, + + =76–99%, + + + =100%). Thisclinical assessment was subsequently

confirmed, by a postoperative histologi-cal biopsy. The ethical committee inNiigata University Dental Hospital per-mitted only a single postoperative biopsyto evaluate the grafts. Therefore, in lieuof multiple biopsies from the same sub-ject, we altered the timing of the onebiopsy in every ten cases. In fifteen cases,a biopsy was performed at 4 weeks post-operatively, while the last five were doneat 2 weeks. The postoperative biopsywas taken from the periphery of thegrafts in which both the EVPOMEand AlloDerm� clinically appeared, topossess an epithelial layer.

After discharge, all patients wereevaluated weekly for the first month,bi-weekly for the next 2 months and thenmonthly up to 12 months. At each post-operative visit, direct clinical obser-vation and photographs were used toevaluate and document wound healing.The time period required to completeepithelial coverage was determined bythe number of days between the dayof excision and the day full epithelialcoverage was achieved. Adverse eventswere evaluated by daily reports duringhospitalization and at each visit afterdischarge, as well as by a serum chemis-try profile and a complete blood cellcount at day 7, 14, 28 days, 3, 6 months,postoperatively.

Statistical analysis

Mann–Whitney U test was used todetermine the statistical significancebetween the two groups: EVPOME andAlloDerm�. Values of P<0.05 were con-sidered to be statistically significant.Data was expressed as the mean�SEM.

Results

There were a few demographic differ-ences between, Group 1, EVPOME andGroup 2, AlloDerm�. In Group 1, themean interval between punch biopsy andexcision, equal to days it took to fabri-cate the EVPOME prior to surgery, was27.7�1.4 days whereas the mean inter-val between the first visit and excisionfor Group 2, AlloDerm�, was 13.7�1.5days. This difference was statistically sig-nificant (P<0.01). This occurred becauseour treatment protocol triaged patientswho had a clinical diagnosis of carci-noma, T1 or T2 into a group that hadthe lesion excised as soon as possiblewhile the EVPOME cases were sched-uled electively. Our triage treatmentpolicy resulted in a larger sample ofmalignant lesions and males in Group 2,

Fig. 2. Clinical photographs showing a representative surgical procedure. (Top left) Outer solidline is the incision line used to excise the lesion with a 5 mm surgical margin from the lesion thatis delineated by an inner solid line. (Top right) The lesion is excised and an open wound iscreated. (Bottom left) Two circular pieces of EVPOME are applied onto the open wound andsutured into place. (Bottom right) Gauze bolster is placed on the EVPOME.

Table 2. Postoperative outcomes of the EVPOME (Group 1) and AlloDerm� alone (Group 2).

Group 1EVPOME (15 pts)

Group 2AlloDerm�alone (15 pts)

Postoperative pain � (15) � (15)

Adhesiveness + + + (15) + + + (13)+ + (0) + + (2)

Epithelial coverage + + + (13) + + + (0)+ + (2) + + (12)+ (0) + (3)

Required days for epithelialization 27.4�1.2 46.0�2.8

Postoperative pain: + and � units used for Visual Analog Scale (VAS) values ranging asfollows; � = 0, + =1–24, + + =25–49, + + + =50–100.Adhesive: + and � units represent ‘Pull off strength’; � =non-adherent, + =adherent buteasily removed, + + =adherent difficult to remove, + + + =adherent unable to remove.Epithelial coverage: + and � unit for % of epithelial coverage of grafted area at four weekspost-grafting; � =0–50%, + =51–75%, + + =76–99%, + + + =100%.Required days for epithelialization: Values were expressed as the mean�SEM. P value of<0.05 indicates statistical significance.

192 Izumi et al.

AlloDerm�. In contrast, patient’s with apositive smoking history, which plays arole in etiology of malignant lesionsand can affect graft success rate, wasequal in both groups. The average sur-face area surgically covered by thegrafted EVPOME was 12.41�1.70 cm2,while the mean area of coverage ofAlloDerm� was 11.28�1.83 cm2. Therewas no significant difference between thetwo groups.

Postoperative course for both groupsof patients was uneventful. There wereno significant differences between bothgroups in regard to postoperative painor adhesion of the graft to the under-lying tissue while epithelial coverageoccurred more rapidly in Group 1 (Table2). Patients who had lesions excised fromthe tongue, in either group, experiencedoccasional paraesthesia (dull sensationand/or numbness). AlloDerm� control

grafts experienced more tethering andindurations of the tongue, secondary towound contraction, than EVPOMEgrafts. Complete epithelial coverage wasachieved in the EVPOME grafts by27.4�1.2 days while the AlloDerm�

grafts required 46.0�2.8 days. The dif-ference was statistically significant(P<0.01).

At 6 days post-surgery, after removalof pressure dressings, a cytologicalsample was taken at the centre of thegraft, which revealed healthy single andaggregates of epithelial cells on the sur-face of EVPOME, indicating the persist-ence of cultured oral keratinocytes onthe grafts (Fig. 3A). In contrast, thecytological smear taken at the centre ofthe AlloDerm�, showed only numerousneutrophils within a mucinous exudatebut no evidence of viable epithelial cells(Fig. 3B). The surface of the trans-planted EVPOME exhibited a dark redhue that was found to correlate withvascular ingrowth into the dermalequivalent from the underlying host tis-sue (Fig. 4A). The AlloDerm� grafts, incontrast, were faintly red, indicating alesser degree of vascular infiltration (Fig.4B). At the second week post-grafting,after removal of the bolus or stent, hostgranulation tissue was seen to extendover the wound periphery in all patients(Fig. 4A,B). In several patients a portionof the surface of the AlloDerm�, wasseen to have a ‘feather’ appearance dur-ing the second week after surgery (post-operative days 8 to 14). Histologicalexamination showed this to be necrotictissue infiltrated with bacteria (notshown).

A biopsy taken at 2 weeks post-grafting, from an area that clinicallyappeared epithelialized in both groups,showed histologically the submucosallayer to have an intense inflammatoryinfiltrate among a bed of immaturegranulation tissue with diluted, leakyvenules (Fig. 5A,B). The overlying epi-thelial layer in Group 1, EVPOME,showed a well-ordered, thick spinouslayer, indicative of hyperplasia (Fig. 5C),whereas that in Group 2, AlloDerm�

alone, showed a more random irregularregenerative pattern (Fig. 5D). At the 4week biopsy, taken from an epithelial-ized area in both groups, Group 1,EVPOME, showed a more mature, dif-ferentiated, overlying epithelium in con-trast to, Group 2, AlloDerm�, whichlacked the uniformity and epithelia regu-larity (Fig. 6A,B). An inflammatoryresponse was noted within the dermallayer of both groups as documented

Fig. 3. Cytological smear of grafts. A: Squamous cells seen on the smear, suggestive of thepresence of cultured keratinocytes (Papanicolaou staining). B: Numerous neutrophils withinmucinous exudates with no evidence of keratinocytes being present (Papanicolaou staining).

Intraoral grafting of an ex vivo produced oral mucosa equivalent 193

by the plethora of blood vessels perpen-dicular to the overlying epithelium(Fig. 6A,B). In Group 1, the dermalcomponent showed fewer numbers ofinflammatory cells and large venulesladen with ‘tall’ endothelial cells (Fig.6C), and the presence of elongatedspindle-shaped fibroblasts among colla-gen bundles (Fig. 6E). Group 2, in con-trast, showed evidence of a persistentintense inflammatory response with

numerous giant cells and eosinophilspresent within the dermis (Fig. 6D,F,compared with Fig. 6C,E).

There were no clinical signs of aforeign body response to either theEVPOME or AlloDerm� grafts withhealing eventually occurring unevent-fully. Total serum chemistry profile aftergrafting showed no systemic evidence ofan immunologic or systemic response tothe transplants.

Discussion

Recently, tissue-engineered oral mucosasubstitutes have been applied onto openwounds in the oral cavity1,4,11,14,18–20.These oral mucosa substitutes can beclassified into three types17: (1) sheetsof cultured autologous keratino-cytes4,11,18,19, (2) an artificial dermiscomposed of collagen and/or gly-cosaminoglycans1,14, or (3) a compositeoral mucosa, containing both an epi-thelial and dermal layer20. These studieshave shown favorable outcomes usingthe different types of grafts in recon-structive oral and maxillofacial surgerybut have several major deficiencies intheir experimental design and protocol;the absence of a control group and asmall sample size, respectively. To ourknowledge, our investigation is the firststudy evaluating the clinical applicationof a composite human ex vivo producedoral mucosa equivalent (EVPOME) witha parallel control group, AlloDerm�,without epithelium with a sufficientsample size to perform statisticalanalysis.

The EVPOME utilized in our clinicalprotocol is a composite or full-thicknessoral mucosa equivalent fabricated withan epithelial and dermal layer. To date,there has only been one other report onthe intraoral use of a bilayered culturedoral mucosa. In that study the dermalequivalent utilized was composed of aType I collagen gel repopulated withfibroblasts20. A major disadvantage ofthe use of collagen, intraorally, is itsfragility and early breakdown secondaryto the plethora of lytic enzymes presentwithin the oral cavity3. In addition, theresurfacing of the oral cavity is moredifficult to accomplish than skin becauseof a moist environment laden withnumerous microbial organisms, andan undulating surface which can com-promise graft stability. The use ofAlloDerm�, as a dermal equivalent,enhances the durability and complianceof the EVPOME as well as improvingthe handling characteristics assisting inmanufacture, intraoral placement andstabilization of the graft at the surgicalsite.

The use of an irradiated xenogeneic3T3 mouse feeder layer, utilized in otherstudies, has been shown to optimize thein vitro culturing of keratinocytes, whileeffecting stem cell maintenance resultingin a longer life-span, and a more prolif-erative cell population16. The ability toincrease and maintain the number ofstem cells to sustain proliferation and

Fig. 4. Typical appearances of EVPOME and AlloDerm� grafts after transplantation. A:EVPOME, Group 1 (the same case as shown in Fig. 2); Epithelialization starts from woundedge. Note the full epithelial coverage obtained by 4 weeks post-grafting. Postoperative biopsyis taken from the site shown by a white arrowhead. B: AlloDerm�, Group 2; Note epithelialcoverage not completed at 4 weeks post-grafting. Postoperative biopsy is taken from the siteshown by an arrow.

194 Izumi et al.

renewal of the keratinocyte populationwould be beneficial in enhancing themanufacture of the EVPOME. Unfortu-nately the optimization of keratinocytegrowth should not be at the cost of anincrease risk of cross contamination ofthe cultured human keratinocytes withxenogeneic material from the irradiated3T3 feeder layer, especially in light thatthe majority of oral and maxillofacialsurgical procedures are elective in natureand the surgery can be timed to assure

the availability of an adequate popu-lation of autologous keratinocytes.

In our investigation, several obser-vations indicated that the graftedEVPOME had advantages over the useof AlloDerm� without epithelium. First,was the early revascularization noted bythe increase in ‘redness’ of the EVPOMEgraft; second, cytological evidence of thepresence of intact squamous cells withinthe centre of the EVPOME graft atpostoperative day 6; third, a greater

reduction in dermal inflammatoryresponse, histologically, over time whencompared to the AlloDerm� alone, andfourth, the more rapid and mature epi-thelial coverage, observed clinically andnoted histologically, of the EVPOMEgraft.

On postoperative day 6 after removalof the pressure dressing, it was observedthat the grafted AlloDerm�, withoutepithelium, appeared to have an ‘ulcera-tive’ appearance secondary to the lack of

Fig. 5. Representative histologic findings of the postoperative biopsy at 2 weeks in EVPOME (A,C) and AlloDerm� (B,D) grafts (haematoxylinand eosin staining). A, B: Numerous eosinophils, lymphocytes, plasma cells, multinucleated giant cells and plumped fibroblasts are seen in thedeeper portion of the submucosal layer (original magnification �400). C: Epithelial layer demonstrates a thick spinous layer and elongated reteridges. Immature, highly vascular granulation tissue is present within the subjacent epithelial layer (original magnification �50). D: The basalcell layer shows an increase of nucleus/cytoplasm ratio and disordered cellular arrangement. Note prominent inflammatory cell infiltration isevident beneath epithelium (original magnification �75).

Intraoral grafting of an ex vivo produced oral mucosa equivalent 195

Fig. 6. Photomicrograph of histologic find-ings of the postoperative biopsy at 4 weeks inEVPOME (A,C,E) and AlloDerm� (B,D,F)grafts (haematoxylin and eosin staining). A:In the EVPOME a thick differentiated epi-thelial layer is present. A decreased inflamma-tory cell infiltration is present within thedermal layer in which immature is beingreplaced by fibrous connective tissue (originalmagnification �50). B: There is disarray ofthe epithelial layer of the AlloDerm�. As wellan intense inflammatory response is still per-sistent. Numerous diluted blood vessels arenoted by their arrangement perpendicular tothe overlying epithelium (original magnifi-cation �60). C: A decrease in number ofinflammatory cells is seen within theEVPOME. The thickness and diameter of theblood vessels beneath the epithelium arereduced (arrows) (original magnification�300). D: The AlloDerm� alone shows thecontinued presence of neutrophils, lym-phocytes, macrophages and plumped fibro-blasts beneath epithelium (originalmagnification �300). E: The EVPOMEshows few plumped fibroblasts and eosi-nophils within in the deeper portion of thedermis (original magnification �300). F: Inthe deeper portion of the AlloDerm�, anintense cellular infiltration containing numer-ous eosinophils (arrows) is observed (originalmagnification �300).

196 Izumi et al.

an epithelial layer. In contrast, thegrafted EVPOME had a ‘slough off’appearance of its superficial epitheliallayer as is usually seen in free gingivalgrafts. It is known that this ‘sloughedoff’ tissue usually represents loss of themore differentiated superficial stratifiedcells of the transplant. Unfortunately, abiopsy was not taken of the centre of theEVPOME graft to confirm the presenceof an intact basal layer. It is, therefore,assumed that basal cells of the epitheliallayer continued to remain attached tothe underlying dermal substrate. It ismost likely the presence of the persistentepithelial basal cells of the EVPOMEgrafts that were influential in allowing amore rapid healing to occur. This, inturn, resulted in a better colour renditionof the EVPOME graft with the adjacentmucosa as well as a marked reduction inthe underlying inflammatory responsewithin the dermal component.

The advantages noted with the use ofthe EVPOME grafts may be the resultof the presence of an intact and matureepithelial layer secreting keratinocyte-derived cytokines, such as vascularendothelial growth factor (VEGF) thateffect dermal vascularity. We haveobserved that human oral keratinoctyescan naturally secrete high endogenouslevels of VEGF in culture (data notshown). The release of VEGF andassociated cytokines may have moder-ated the inflammatory response,enhanced cell growth, neoangiogenesis,and wound healing while minimizinggraft contracture2,5,13. According toY et al.21, the initial 2–3 days aftertransplantation are critical to graft sur-vival and is predicated on vascularingrowth into the transplants via theprocess of inosculation in associationwith neoangiogenesis. The presence ofnumerous intact acellular blood vesselchannels, within the dermal layer, thatare maintained within the architecture ofthe AlloDerm� after preparation mayalso have assisted in the early revascu-larization of the dermal equivalent inboth groups.

An intact epithelium has been notedto create a barrier effect of the overly-ing dermis protecting it from bacterialinvasion6. In addition, the intact kerati-nocyte layer on the EVPOME couldhave been a source of other releasedgrowth factors, such as transforminggrowth factor alpha (TGF�) and epider-mal growth factor (EGF) that couldinfluence host epithelial cells prolifer-ation and migration from the margin ofthe wounds thus enhancing epithelial

coverage resulting in less of an inflam-matory response within the dermal com-ponent12. The presence of a persistentepithelium on the EVPOME trans-plants would also potentially make themuseful as a device for grafting of geneti-cally modified keratinocytes that maybe beneficial for various therapeuticapplications15.

There were no systemic complicationsseen in either Group 1 or 2 after graft-ing. Postoperative pain was minimal inboth groups of patients with evidence ofslight neurosensory deficits observed insome of the cases involving reconstruc-tion of the tongue. Kangesu10 noted thatnerve regeneration, into a grafted skinequivalent, composed of de-epidermizeddermis and cultured keratinocytes, fol-lowed re-vascularization. Thus, since thetongue contains a rich nerve plexuswithin its musculature longer follow upmay see a resolution to this adverseoutcome.

In our study, it was noted that theAlloDerm� grafts, without epithelium,resulted in a more indurated and con-tracted wound than seen with theEVPOME grafts, post transplantation.Severe wound contraction, which is seenin healing by second intention, can resultin post-grafting morbidity, leading tofunctional disturbances in speech byrestriction of the mobility of the tongue.Garner demonstrated that early epider-mal coverage prevented excessive colla-gen synthesis within the underlyingdermal fibroblasts and can minimizewound contracture7. The smaller degreeof wound contracture seen with theEVPOME grafts may be the result of theestablishment of an intact and matureepithelial layer influencing underlyingcollagen synthesis within the dermalcomponent, AlloDerm�. It also appearsthat the persistence of the culturedautologous keratinocytes on theEVPOME played a key role in assistingin its integration into the underlying hosttissue. In addition the inflammatorycellular infiltrate within the dermisdecreased more dramatically from week2 through 4, postoperatively, in theEVPOME, in contrast to the persistentinflammatory infiltrate seen in Group 2,AlloDerm� without an epithelial layer.The evidence of less induration at thewound site in the EVPOME graft maybe related to a more physiologic repairof the underlying dermal layer secondaryto the presence of an intact epitheliallayer that has the capability to secretepro-inflammatory cytokines and growthfactors resulting in a more moderate and

shortened inflammatory phase of woundrepair.

Retrospective evaluation of ourclinical study with grafted EVPOMEs,revealed several design and logisticaldeficiencies:

1. Only one biopsy was allowed and itwas not taken at the same timeperiod.

2. Biopsies were taken from the periph-ery of the graft (from areas that‘appeared’ to have an epithelial layer)and not from the graft centre. Thisdid not allow us to assess the absenceor presence of a basal epitheliallayer on either type of graft at a sitethat would be less influenced, thecentre, by migration of cells fromthe periphery of the reconstructedwound.

3. The use of specific stains may haveadded additional information.

4. A blinded evaluator did not clinicallyassess the grafts and would haveadded more objectivity to the clinicalevaluation.

5. A thinner dermal equivalent mayhave resulted in a faster vascularingrowth but may also have impairedthe handling characteristics (curlingup of the margins during manu-facture). This issue will have to beevaluated in future studies.

6. The majority of the surgical sites werethe tongue, a highly mobile organ. Anon-mobile bony recipient site, suchas a maxillary or mandibular ridge,may have been more predictablein stabilization of the grafts thusenhancing re-vascularization andmaintaining the epithelial layer.

7. Elimination of the use of pituitaryextract in the manufacturing processof the EVPOME to create a moredefined and ‘clean’ system. This pointhas, in fact, been accomplished in ourlaboratory and will be incorporatedinto future investigations.

8. We are planning, in the future, arandomized, controlled multi-centrestudy using patients with non-malignant lesions that would allowmore similarity between the groupsbeing evaluated.

In summary, it appears that theEVPOME graft may be a more accept-able oral mucosa substitute for humanintraoral grafting procedures resultingin a more favorable wound healingresponse than the AlloDerm� graftwithout an epithelial layer. The advan-tage of the EVPOME over the

Intraoral grafting of an ex vivo produced oral mucosa equivalent 197

AlloDerm� may have been due to thepersistence of a transplanted epithelialbasal cell layer resulting in the develop-ment of a more rapid and mature epi-thelium. The persisted basal layer ofgrafted keratinocytes and/or the initialpresence of an intact epithelial layer,developed in vitro, at the time ofgrafting of the EVPOME, may havegiven it a decided advantage in eventualrepair and healing of the surgicalwound by the release of cytokines andgrowth factors that are known to besecreted by keratinoctyes. Additionalstudies will give us more insight intothe potential benefits of the use of atissue engineered human oral mucosalequivalent as an aid in surgical recon-struction and as a vehicle for use ingene therapy.

Acknowledgments. We would like tothank Dr Satoshi Maruyama andProfessor Takashi Saku for their help inevaluating the histology of the biopsiesand Masaaki Hoshino for excellent tech-nical assistance. The authors thankProfessors Ritsuo Takagi and ChikaraSaito for their helpful suggestions. Wealso thank the Oral & MaxillofacialSurgeons in Niigata University for theircordial patient care and cooperationwith this study.

This work was supported by Grant-in-Aid for Scientific Research (No.12771216) from the Ministry of Edu-cation, Science and Culture, Japan,Grant for the Promotion of NiigataUniversity Research Projects, and Grantfrom Kato Memorial BioscienceFoundation (KI) and US PHS GrantNIDCR/NIH R01 DE 13417 (SEF).

References

1. B K, M K, I T. Theuse of a new bilayer artificial dermis forvestibular extension. Br J Oral MaxillofacSurg 1998: 36: 457–459.

2. B LF, Y K-T, B B, Y TK,S DR, D HF, V W L. Expression of vascular per-meability factor (vascular endothelialgrowth factor) by epidermal keratino-cytes during wound healing. J Exp Med1992: 176: 1375–1379.

3. C KH, A HT, P KC, CJH, K SW, S MW, K KH,C PH, E HC, Y JI.Reconstruction of human hard-palatemucosal epithelium on de-epidermizeddermis. J Dermatol Sci 2000: 22: 117–124.

4. D L M, A E, M M,C R, M PE, CA, F AT. Evidence that human oralepithelium reconstituted in vitro andtransplanted onto patients with defects inthe oral mucosa retains properties of theoriginal donor site. Transplantation 1990:50: 454–459.

5. D M, Y K-T, N JA,V W L, B LF, B B,E BM, L S, DHF. Keratinocyte-derived vascular per-meability factor (vascular endothelialgrowth factor) is a potent mitogen fordermal microvascular endothelial cells. JInvest Dermatol 1995: 105: 44–50.

6. F V, M D, A O,A M, M F, AM, J J, S M, H-Y J. Rapid healing of venous ulcersand lack of clinical rejection with anallogeneic cultured human skin equiva-lent. Arch Dermatol 1998: 134: 293–300.

7. G WL. Epidermal regulation ofdermal fibroblast activity. Plast ReconstrSurg 1998: 102: 135–139.

8. I K, F SE, T H,M CL. Evaluation of transplantedtissue-engineered oral mucosa equivalentto SCID mice. Tissue Engineering Feb2003: in press.

9. I K, T H, M CL,F SE. Development and charac-terization of a tissue-engineered humanoral mucosa equivalent produced in aserum-free culture system. J Dent Res2000: 79: 798–805.

10. K T, M S, T G, GX-H, N HA, P JM, GCJ, L IM. Nerve and blood vesselgrowth in response to grafted dermis andcultured keratinocytes. Plast ReconstrSurg 1998: 101: 1029–1038.

11. L G, S R. Tissue-engineered mucosa graft for reconstruc-tion of the intraoral lining after freeing ofthe tongue: a clinical and immunohisto-logic study. J Oral Maxillofac Surg 2001:59: 169–177.

12. M P. Wound healing-aiming forperfect skin regeneration. Science 1997:276: 75–81.

13. M S, N H, S R,G C, L I. Transplantation ofkeratinocytes in the treatment of wounds.Am J Surg 1995: 170: 75–83.

14. O S, M N, H S,K R, F K. A newly developedcollagen/silicone bilayer membrane as amucosal substitute: a preliminary report.Br J Oral Maxillofac Surg 1997: 35:85–91.

15. O’T G, M D, BMF, L R, P M. A review oftherapeutic angiogenesis and considera-tion of its potential applications to plasticand reconstructive surgery. Br J PlastSurg 2001: 54: 1–7.

16. P G, R R, S G,B S, G L, Z G,M G, D L M. The control ofepidermal stem cells (holoclones) in thetreatment of massive full-thickness burnswith autologous keratinocytes culturedon fibrin. Transplantation 1999: 68: 868–879.

17. P TJ. New skin for old. ArchDermatol 1998: 134: 344–349.

18. R GM, T AM,S J, B EH, W MJH,V A. Use of cultured mucosalgrafts to cover defects caused by vestibu-loplasty: An in vitro study. J OralMaxillofac Surg 1995: 53: 872–878.

19. T C-Y, U M, H K, H K,H Y, S Y, T K,T S. Clinical results of cultured epi-thelial cell grafting in the oral and maxil-lofacial region. J Cranio-Maxillofac Surg1997: 25: 4–8.

20. U M, E K, K T. In vitrofabrication of bioartificial mucosa forreconstruction of oral mucosa: basicresearch and clinical application. AnnPlast Surg 1991: 27: 540–549.

21. Y DM, G KM, W HG.Species-specific in situ hybridization withfluorochrome-labeled DNA probes tostudy vascularization of human skingrafts on athymic mice. J Burn CareRehabil 1996: 17: 305–310.

Address:Kenji Izumi, DDS, PhDDepartment of Oral and Maxillofacial

SurgeryUniversity of Michigan Medical Center,

UH-B1-2041500 E. Medical Center Dr.Ann ArborMI 48109-0018USATel: +1-734-615-8708Fax: +1-734-936-5941E-mail: kennie@umich.edu