A surgical model of composite cystoplasty with cultured urothelial cells: a controlled study of gross outcome and urothelial phenotype

Authors


Prof. J. Southgate, Jack Birch Unit for Molecular Carcinogenesis, Department of Biology, University of York, York YO10 5YW, UK.
e-mail: js35@york.ac.uk

Abstract

OBJECTIVES

To study the outcome of composite cystoplasty using cultured urothelial cells combined with de-epithelialized colon or uterus in a porcine surgical model, using appropriate controls, and to characterize the neo-epithelium created by composite cystoplasty.

MATERIALS AND METHODS

Urothelial cells were isolated and propagated in vitro from open bladder biopsies taken from nine female minipigs. Cohesive sheets of confluent urothelial cells were transferred to polyglactin carrier meshes and sutured to de-epithelialized autologous colon in four animals and de-epithelialized autologous uterus in five. These composite segments were then used for augmentation cystoplasty. Conventional colocystoplasty, de-epithelialized colocystoplasty and sham operations were carried out in six control animals. After killing the animals at ≈ 90 days the bladders were removed for examination and immunohistochemical analysis, using a panel of antibodies against cytokeratins and urothelial differentiation-associated antigens.

RESULTS

Macroscopically, the bladders augmented with composite segments derived from uterine muscle had no evidence of shrinkage or contracture. Histological analysis showed that in four of five composite uterocystoplasties, the neo-urothelium was stratified and had a transitional morphology, although in some areas coverage was incomplete. Immunohistochemical analysis showed evidence of squamous differentiation in both native and augmented segments. All composite and de-epithelialized colonic segments showed significant contraction with poor urothelial coverage, reflecting the unsuitability of the thin-walled porcine colon for de-epithelialization.

CONCLUSIONS

The functional and macroscopic outcome of bladder augmentation with a composite derived from cultured urothelium and de-epithelialized smooth muscle of uterine origin endorses the feasibility of composite cystoplasty.

Abbreviations
NHU

normal human urothelial (cells)

NPU

normal porcine urothelial (cells).

INTRODUCTION

The use of intestinal segments for bladder augmentation or substitution (enterocystoplasty) has been a major advance in the surgical management of congenital and acquired abnormalities of the bladder. However, the benefits of enterocystoplasty are offset by well-documented, relatively common and potentially serious complications, including mucus production, stone formation, chronic low-grade infection and metabolic disturbance (reviewed in [1]). These problems are attributable to the lining of the intestine, an absorptive, mucus-secreting epithelium that is not adapted to prolonged contact with urine. The ideal material for bladder reconstruction would combine the compliance afforded by smooth muscle with the non-absorptive barrier lining of normal urothelium.

Despite extensive research into alternatives to conventional enterocystoplasty very few techniques have been translated into clinical practice. Ureterocystoplasty achieves the goal of a urothelium-lined augmentation but is effectively confined to a few patients with a combination of gross ureteric dilatation and an ipsilateral nonfunctioning kidney [2]. Seromuscular enterocystoplasty coupled with autoaugmentation also creates a urothelium-lined augmentation [3,4]. However, in clinical practice the role of this technique may be limited by the difficulty of detrusor myectomy in a contracted or heavily trabeculated neuropathic bladder, and by the limited potential to increase the capacity of small bladders by autoaugmentation.

We previously reported the development of reliable cell-culture systems capable of generating large areas of normal human urothelium [5–7], with the aim of developing a tissue-engineering approach to bladder reconstruction. Other groups have described tissue-engineering approaches for the development of whole-bladder replacements [8] or biomaterials for bladder wall substitution [9,10]. However, as the complications of enterocystoplasty are largely attributable to the epithelium rather than the intestinal smooth muscle, we favoured the concept of ‘composite’ cystoplasty, in which autologous urothelium is cultured in vitro and combined with de-epithelialized bowel at the time of reconstruction [11,12]. This concept has been endorsed by studies in which we showed that in vitro-propagated normal human urothelial (NHU) cells can be induced to stratify and differentiate when recombined with a de-epithelialized stroma in organ culture [13].

The current study was designed to establish an in vivo surgical model of composite cystoplasty using autologous urothelium propagated in vitro and combined surgically with either de-tubularized colon or uterus. A porcine model was chosen as being relevant to clinical practice in terms of body size and weight, and because preliminary studies showed that, unlike rodent urothelial cells, normal porcine urothelial (NPU) cells closely resembled NHU cells in terms of in vitro characteristics. In addition, laboratory rodents and rabbits have a marked propensity for stone formation after bladder augmentation.

The porcine colon is thin-walled and the muscularis mucosae poorly defined, whereas the bicornuate porcine uterus is a tubular structure, with a muscle wall comparable in thickness to human sigmoid and descending colon. For this reason, in addition to composite and conventional colocystoplasty, the bladder was augmented using urothelium combined with de-epithelialized segments of de-tubularized porcine uterus as a surrogate for colonic smooth muscle. The use of uterine segments had the added advantages of a lower risk of bacterial infection than with bowel and avoidance of the potential risks associated with intestinal anastomosis.

MATERIALS AND METHODS

The study comprised 15 female Yucatan or Gottingen minipigs weighing 15–30 kg. All animals were maintained on a standard porcine diet. Procedures were carried out in full compliance with statutory regulations. The 15 animals were divided into five groups as follows. (A) Composite colocystoplasty (four); in these procedures the bladder was incised and augmented with a de-epithelialized colonic segment combined with autologous tissue engineered urothelium. (B) Composite uterocystoplasty (five); the bladder was incised and augmented with a de-epithelialized uterine segment combined with autologous tissue engineered urothelium. (C) Conventional colocystoplasty (‘control’ group, two); conventional augmentation cystoplasty using a non de-epithelialized colonic segment. (D) De-epithelialized colocystoplasty (‘control’ group, two); in these procedures the augmenting segment comprised de-epithelialized colon with no tissue-engineered urothelium. (E) Sham operation (‘control’ group, two); the bladder was exposed, opened and then re-closed, with no form of augmentation.

ANAESTHESIA

Food, but not water, was withheld for 12 h before surgery; general anaesthesia was induced by inhaled 5% halothane (AstraZeneca, Cheshire, UK) and by intravenous propofol (5–10 mg/kg; AstraZeneca). After tracheal intubation all animals breathed spontaneously. Anaesthesia was maintained with 2–5% halothane and oxygen via a modified Bain circuit. Buprenorphine (300 µg; Schering-Plough, Hertfordshire, UK) or diclofenac sodium (100 mg; Novartis Pharmaceuticals, Surrey, UK) was administered for postoperative analgesia.

UROTHELIAL TISSUES AND CELL CULTURE

The urinary bladder was exposed through a small lower midline abdominal incision and a full-thickness section of ≈ 2 cm2 removed. The tissue was collected into sterile transport medium (Hank's balanced salt solution containing 10 mm HEPES pH 7.6 and 20 KIU aprotonin) and used for harvest of the urothelium (Fig. 1a,b).

Figure 1.

A diagrammatic representation of sequential steps involved in composite uterocystoplasty; see text for details.

Cell culture techniques for NPU culture were based on our previously described procedures for NHUs [6,7] except that dispase was used to isolate urothelial cell sheets. In brief, bladder tissue specimens were washed twice in fresh transport medium. Excess stroma was removed by sharp dissection and the urothelium separated by 16 h incubation at 4 °C in 2% (w/v) dispase II (Boehringer Mannheim, Roche Diagnostics, Lewes, UK) in Hank's balanced salt solution. Urothelial sheets were incubated in collagenase type IV (100 U/mL; Sigma, Poole, UK) for 20 min at 37 °C and disaggregated by pipetting. Washed cells were seeded into Primaria® tissue culture flasks (Becton Dickinson, Cowley, UK) at 2 × 105 cells/cm2 in keratinocyte serum-free medium containing bovine pituitary extract and epidermal growth factor (Gibco BRL, Paisley, UK). Cholera toxin (30 ng/mL, Sigma) was used to enhance the plating efficiency [6]. Growth medium was changed at 24 h and thereafter three times weekly. Cells were passaged at near-confluency with trypsin-versene, as described [6,7]. For bladder reconstructions, urothelial cell cultures were maintained in culture for 2–4 weeks and used at passages 3 or 4. Stocks of cells were cryopreserved at passage 1 to provide a backup in case of loss through contamination.

UROTHELIAL CELL TRANSFER

For use in composite cystoplasty urothelial cells were seeded and grown to confluence in 11 6-cm Petri dishes (total area 28.3 cm2); 24 h before harvest 0.5 mol/L CaCl2 was added to the growth medium to a final concentration of 2.3 mmol/L to increase urothelial cell:cell cohesion, as exogenous Ca++ induces urothelial cell stratification and desmosome formation, and increases the expression of E-cadherin [6]. Cells were detached as intact sheets by incubation in 2% (w/v) dispase II and transferred to 30 × 35 mm2 pieces of polyglactin 910 woven mesh (Ethicon Ltd, Edinburgh, UK) with the apical surface of the urothelium in contact with the mesh, as described previously [14]. Because of contraction of the urothelial sheets after detachment from the substrate, on average, two or three urothelial sheets were needed to cover each piece of polyglactin mesh. The urothelium-polyglactin complexes were transferred aseptically to the operating theatre and kept moist in medium until use.

COMPOSITE COLO- AND UTEROCYSTOPLASTY

Intravenous antibiotic (cefuroxime 750 mg, Glaxo Wellcome, UK) was administered after inducing anaesthesia. A 10–15 cm segment of colon (four animals; group A) or uterus (five animals; group B) was isolated on a vascular pedicle and the remaining bowel or tubular uterus re-anastomosed in a single layer with 3/0 or 4/0 polydioxanone. The isolated segment was de-tubularized by opening along its antemesenteric border to produce a flat ‘plate’ of ≈ 80 cm2 and de-epithelialized using a combination of sharp and blunt dissection. In the uterine segments de-epithelialization was facilitated by a clearly visible line that demarcated the mucosa from deeper layers (Fig. 1c). By contrast, de-epithelialization of the colon was technically difficult and resulted in a flimsy seromuscular plate of 1–2 mm thick.

The complexes of urothelium and polyglactin were sutured to the de-epithelialized surface (Fig. 1d) with 6/0 polydioxanone with the cellular face of the mesh-cell support in contact with the stroma, thus ensuring that the urothelial cell sheets retained their normal polarity. The dome and lateral walls of the bladder were incised to accommodate the 10–15 cm augmenting segment. A continuous suture of 2/0 polydioxanone was used to suture the composite segment to the bladder, as in a clam cystoplasty. Before completing the augmentation a 250-mL low-pressure silicone balloon (Medasil, Leeds, UK) was positioned in the bladder, with the filling port exiting anteriorly through the native bladder wall and inflated with 100 mL sterile saline (Fig. 1e). A suprapubic catheter was used for bladder drainage in all animals and infant-feeding tubes were also used as ureteric catheters in some. Catheters were tunnelled subcutaneously to exit on the lateral abdominal wall and were allowed to drain freely without collection. From induction of anaesthesia to recovery the mean (range) duration of the procedure was 4.5 (4–6) h.

CONTROL ANIMALS

Two control animals (group C) had a conventional colocystoplasty using the technique described above, except that the augmenting segment consisted of intact, epithelialized colon. An intravesical balloon was used after surgery in one animal whilst the augmentation had no indwelling intravesical balloon in the second.

In a further two control animals (group D) the augmentation was by a segment of de-epithelialized colon but with no attachment of a urothelium/polyglactin complex. As in group C an intravesical balloon was used in one animal only.

Finally, two control animals (group E) had the bladder incised as above but closed with a continuous suture, with no augmentation; a balloon was left in the bladder in one.

In all animals, antibiotics (cephradine, 50 mg/kg, Bristol-Myers Squibb Pharmaceuticals, UK) were continued for 14 days, when the intravesical balloons were removed under a further short general anaesthetic. Animals were kept for 90 days after surgery. Voiding patterns were observed and serial venous blood samples taken to assess any changes in variables of renal function and acid-base status. After death the bladders were removed, examined macroscopically for evidence of contracture of the augmentation and fixed in 10% formalin.

HISTOLOGY AND IMMUNOHISTOLOGY

Paraffin wax-embedded tissues were prepared from normal (control) porcine bladder and uterus, and from specimens taken from several areas of the composite cystoplasty, corresponding to the native ureter, native bladder, uterine/bladder junction and the augmented segment. Sections were stained for haematoxylin and eosin, or Alcian Blue, and counterstained with Nuclear Fast Red for acid mucins.

Immunohistochemistry included appropriate controls, as described previously [13], using a panel of antibodies against markers of urothelial differentiation and proliferation (Table 1). Antigens were retrieved as required (Table 1). Antibodies were selected on the basis of reactivity with paraffin wax-processed tissues and showing equivalent immunoreactivity patterns on porcine and human tissues. Antibody binding was visualized using an indirect streptavidin avidin-biotin immunoperoxidase method (Dako, High Wycombe, UK). Endogenous avidin-binding sites were blocked with avidin/biotin blocking reagents (Vector Laboratories, Peterborough, UK).

Table 1.  Immunohistology of normal porcine urothelium and uterocystoplasty reconstructions
Reagent*SpecificityNormal tissueNative part of bladderSurgical augmentation
BasalInterSuperBasalInterSuperBasalInterSuper
  1. The intensity of reaction was graded subjectively from – (negative), +/– (equivocal), + (weak), + + (moderate) to strong (+ + +). *Where indicated, antibody labelling of paraffin wax-embedded tissue sections was after antigen retrieval protocols as follows: T, 10 min digestion in 0.1% (w/v) trypsin in 0.1% (w/v) CaCl2 pH 7.6; M, microwave heating in 10 mmol/L citrate buffer pH 6.0 for 13 min; T+M, 1 min trypsinization followed by microwaving; N, no antigen retrieval required.

LP1K (M)CK7+ + ++ + ++ + ++ + ++ + ++ + ++++
LE41 (M)CK8++ + +++ + +++
KS13.1 (M)CK13+ + ++ + ++ + ++ + ++ + ++/–+ + ++ + +
LL002 (T+M)CK14+ + ++ + ++ + ++ + ++ + ++ + +
LL025 (T+M)CK16
CK20.8 (T)CK20+ + ++
Anti-AUM (N)AUM+ + ++ ++
MIB-1 (M)Proliferating cell nuclei<1%1–10%1–10%
Alcian blueAcid mucins+ ++ ++

RESULTS

The composite cystoplasty proceeded as planned in eight pigs but in one of the five uterocystoplasty procedures technical difficulties led to insufficient urothelium being available at the time of surgery. This resulted in incomplete coverage of the intestinal segment used for augmentation.

All 15 animals survived the surgery and the period immediately afterward with no significant complications, and voided with a normal frequency and pattern. Two animals subsequently died at 6 and 8 weeks; one in group A developed adhesive intestinal obstruction, whilst a control animal in group D died from presumed urinary infection.

MACROSCOPIC APPEARANCES

In group A, there was fibrosis and contraction of the augmenting segment to less than half of the original area in all four of the surviving animals at ≈ 90 days. There was no evidence of intravesical mucus and no calculi in the bladders. In group B at 90 days the external examination of the bladder in all five animals showed a clear demarcation of the serosal surfaces of bladder and uterine components. The augmenting segment was supple, retained its original surface area and showed no evidence of either fibrotic contracture or calculus formation (Fig. 2). In group C there was no contraction of the augmentation in either animal, but in both the bladder contained copious mucus and bulky matrix calculi. In group D, in the remaining animal, contraction and macroscopic appearances suggestive of fibrosis were evident at the time of death. In group E, in both animals the bladder was macroscopically normal at death except for a well-healed bladder incision.

Figure 2.

The macroscopic appearance of the augmented bladder showing the external aspect before (a) and the internal aspect of the inflated bladder after fixing in 10% formalin (b).

PHENOTYPIC CHARACTERISTICS OF NATIVE PORCINE TISSUES

Urothelium from ureter and bladder had similar morphology; the basal layer comprised closely packed, polarized cells with dark-staining nuclei, surmounted by an intermediate layer of three to five larger cells. The superficial layer consisted of flattened cells, each overlying several intermediate cells (Fig. 3a).

Figure 3.

Haematoxylin and eosin-stained histological sections illustrating normal porcine urothelium (a), uterine epithelium (b) and the epithelium lining both the augmented (c) and native (d) segments of reconstructed bladders. Final magnification × 100.

In paraffin wax-embedded sections porcine urothelium was reactive with antibodies against cytokeratins CK7, CK8, CK13 and CK20, but not CK14 and CK16 (Fig. 4). Both CK7 and CK13 were expressed throughout all layers, although the latter was most intense in the late intermediate and superficial cell layers. CK8 was not expressed in the basal layer, weakly expressed in the intermediate and moderately expressed in the superficial layer. CK20 expression was restricted to superficial cells. The anti-AUM antibody reacted with superficial cells, where it localized to the apical membrane. The few MIB-1-positive cells were confined to the basal layer. Alcian Blue staining was weak and confined to the superficial urothelium.

Figure 4.

Immunohistochemical labelling of CK13 (a,c,e) and CK14 (b,d,f) on normal porcine urothelium (a–b) and sections from native (c–d) and augmented (e–f) portions of the uterocystoplasty. In all cases CK13 was expressed throughout the urothelium and most intensely in the late intermediate and superficial cell layers. By contrast, CK14 was negative in normal urothelium (b), but was expressed intensely throughout both the native (d) and augmented (f) portions of the uterocystoplasty. Final magnification × 70.

The uterus consisted of a cellular stroma containing many tubular glands, lined by a simple, tall columnar epithelium (Fig. 3b). Uterine surface epithelium expressed CK7 and CK8 only; there was no reactivity with antibodies against CK13, CK14, CK16 or CK20. CK8 was also strongly expressed in the epithelium of the endometrial glands. Very occasional cells were positive with MIB-1 antibody.

PHENOTYPIC CHARACTERISTICS OF PORCINE TISSUES FROM AUGMENTED BLADDERS

In group A, in the three surviving animals the augmented segment was extensively covered by a tall columnar epithelium of gastrointestinal appearance, suggesting re-growth of colonic epithelium. In one animal 10–20% of the augmentation was covered by an epithelium of multilayered transitional appearance. There was evidence of an inflammatory cell infiltrate in the submucosal layers but no epithelial denudation. The urothelium in the native bladder segment appeared normal.

The expression of CK8 and CK18 only by columnar epithelium confirmed the intestinal-type epithelium; there was no expression of urothelial differentiation antigens. The small areas of epithelium with transitional appearance in one animal showed weak expression of CK7, and expression of CK8 and CK13 that resembled the pattern seen in native urothelium. There was no reactivity against anti-CK20 and anti-AUM antibodies.

In group B the histological examination of the augmentation revealed consistent but incomplete epithelial coverage; 20–50% of the surface of the augmentation was denuded. In four animals the epithelium was stratified and had the morphological features of a multilayered, transitional-type epithelium (Fig. 3c). In the one animal, where insufficient urothelium had been applied, the epithelium lining the augment was exclusively a columnar uterine epithelium. Residual endometrial glands were seen in the underlying stroma in all animals. Infiltration by mononuclear and polymorphonuclear inflammatory cells was evident throughout the bladders, including in the stroma of the native portion of the bladder distant from the augmentation (Fig. 3d). Inflammatory infiltrates were most prominent in areas of epithelial denudation in the augment.

By immunohistology (Table 1), CK7 expression was retained in the native portion of the reconstruction, but was only expressed weakly in the epithelium lining the augment. CK8 and CK13 expression remained largely unaltered in both native and augmented segments, with most intense expression in the superficial and late intermediate cell layers (Fig. 4).

There was a striking difference in CK14 expression, which was not expressed in any part of the normal porcine urothelium but strongly expressed throughout the epithelium lining the augmented segment, and in the native portion of the bladder remote from the augmentation (Fig. 4). CK16 reactivity remained absent. There were more basal cells in the augment with MIB-1 positivity, indicating a higher level of proliferation than in normal urothelium.

CK20 expression was absent from the superficial cells of the augmentation, and was less consistently positive in the superficial cell layer of the native bladder. There was weak apical membrane reactivity with anti-AUM antibody in superficial cells of the augment. Alcian Blue staining was unaltered.

In group C both augmented bladders had a clear histological demarcation on their internal aspect between the urothelium-lined native bladder and intestinal epithelium-lined augment. There were inflammatory cells in the subepithelial layers throughout most of the bladder. Each epithelium type retained its normal pattern of immunoreactivity with the panel of antibodies, apart from absence of CK20 expression in the urothelium and weaker reactivity with the anti-AUM antibody.

In group D the epithelium of the augmentation in the surviving animal was solely intestinal in appearance. There were areas of epithelial loss associated with florid polymorphonuclear cell infiltration, but no suggestion of ingrowth of urothelium from the native bladder.

In group E the two bladders were histologically indistinct from normal urothelium. Immunohistological markers were expressed in a pattern identical to that in normal porcine urothelium.

DISCUSSION

We describe the first surgical model of composite cystoplasty; by developing an autologous urothelium-lined smooth muscle surrogate, this approach has the potential to overcome the complications of conventional enterocystoplasty, which primarily reflects the inappropriate exposure of intestinal epithelium to urine. The complications of conventional colocystoplasty were clearly shown, with the bladders of both control animals containing mucus and extensive matrix calculi at the time of death.

The mini-pig is well accepted as a surgical model for man, as it is of equivalent size and anatomy. NPU cells also behaved very similarly to NHU cells in vitro and required only minor modifications of cell culture technique. The main inconsistency with the pig as a model was that the porcine colon has a very thin muscular wall, with a poorly defined plane of dissection between the mucosa and submucosa, making de-epithelialization of the porcine colon difficult. This meant that the colon was unsuited to composite enterocystoplasty in the porcine experimental model, and there was marked contracture and fibrosis of the composite colonic segments in all five animals, probably reflecting ischaemia and tissue damage during de-epithelialization, and possibly compounded by inflammatory changes on exposure to urine. By substituting the porcine uterus, which has very similar characteristics to human colon in terms of its muscularity and plane of dissection, we developed a more appropriate porcine surgical model that can be used to refine and study the outcome of composite cystoplasty. Nevertheless, that residual endometrial glands were present in the four augmentations in which there was otherwise good urothelial coverage almost certainly reflects difficulties in achieving complete de-epithelialization. In the porcine uterus endometrial glands may extend into superficial muscle, thus increasing the difficulty of excising the epithelium completely.

De-epithelialization has been highlighted as an area of concern in the development of several animal and human models of bladder augmentation [3,4,15,16]. Although the potential for re-growth of native epithelium is inherent in any technique involving de-epithelialization, experience in clinical colorectal surgical practice suggests that more effective de-epithelialization can be achieved in the human sigmoid colon, the proposed material for composite enterocystoplasty, than in porcine uterine segments. In clinical practice the feasibility of surgically removing the mucosa of the human colon is well-established, as in the widely used Soave endorectal pull-through procedure for treating Hirschprung's disease, and in various procedures in adult colorectal surgery. Indeed, seromuscular enterocystoplasty, where a de-epithelialized colonic segment is combined with native urothelium after detrusorectomy and autoaugmentation, has been used in clinical practice by Gonzales et al.[4] after experimental studies in a canine model. Unfortunately, seromuscular colocystoplasty is not well suited to small or trabeculated bladders of the sort commonly encountered in neuropathic dysfunction.

Although urothelium has a very slow turnover time in situ, it has a high regenerative capacity. We and others have reported that human urothelium can be rapidly generated in vitro because of its high proliferative rate that is thought to reflect the normal wound-healing response [17]. The present study shows that it is feasible to generate adequate amounts of normal urothelium by cell culture, to make composite cystoplasty a viable technique. However, the extensive uterine epithelial re-growth in one animal, where technical difficulties had resulted in insufficient urothelial cell sheets to achieve adequate coverage at the time of surgery, emphasizes the importance of timing the composite cystoplasty operation to coincide with the availability of adequate urothelial sheets. It also highlights the need for strict quality-control procedures to regulate the use of cultured cells in tissue-engineering techniques.

This study provides proof of principle for composite enterocystoplasty, as combining a urothelium/polyglactin complex with a de-epithelialized smooth muscle stroma resulted in a functional, compliant augmentation. In the four animals for which adequate urothelial cells had been available at the time of surgery, the neo-epithelium lining the composite uterine segment was stratified and had the morphological appearance of urothelium. The transplanted urothelial cells were integrally attached to the stroma, which is likely to have been facilitated by transferring the cultured urothelial cells onto the de-epithelialized uterine muscle in the form of polarized urothelial cell sheets. Other workers have attempted to establish coverage of a de-epithelialized stroma with cultured urothelial cells. In one study, limited epithelial coverage of de-epithelialized colonic segments was found in only two of 12 rabbits and the derivation of the neo-epithelium was not characterized as being urothelial [16]. In an ovine model, the seeding of cultured urothelial cells onto de-epithelialized stomach or colon maintained in a subcutaneous pouch resulted in development of a uroplakin III-positive neo-epithelium. Nevertheless, fibrosis and inflammation were apparent after 2 weeks, despite lack of contact with urine [18]. Urothelial transdifferentiation to a glandular phenotype has been reported after combining urothelium with rectal mesenchyme [15], providing a cautionary note for heterotypic epithelial-stromal recombinations in reconstructive procedures. In the present model of composite uterocystoplasty there was no evidence of transdifferentiation towards a uterine phenotype. Our data suggest that the presence of columnar epithelium corresponds to areas of incomplete de-epithelialization, whereas squamous differentiation in the neo-epithelium is most likely a result of reactive inflammatory processes, as there were identical changes in the native portion of the bladder.

The results of this first surgical study of composite cystoplasty indicate some important considerations for future refinement of the technique. At a macroscopic level the composite segments showed no evidence of inflammation, graft contraction or stone formation. However, histological examination revealed the presence of an underlying active inflammatory infiltrate, both in the native portion of the bladder and in the composite segment. This may reflect a low-level infection secondary to urinary stasis that was not evident by conventional clinical criteria. A degree of stasis (predisposing to bacteriuria and infection) is an unavoidable consequence of bladder augmentation or substitution, and whilst this can be largely overcome in the clinical setting by regular intermittent catheterization, this is not feasible in an animal model. Alternatively, the findings might reflect a low-grade inflammatory response in the stroma caused by exposure to urine in the early phases of regeneration, when the barrier function of the urothelial complex had not fully developed or the urothelial coverage was incomplete. The urothelium of both native and augmented segments showed evidence of squamous differentiation in association with reduced or incomplete expression of markers of terminal urothelial differentiation, including the AUM, which functionally is thought to contribute to urinary barrier function [19]. In this study, urothelial cells were transferred from in vitro to in vivo environments as cell sheets, after exposure to physiological calcium to induce the formation of intercellular bonds. We showed previously that such treatment does not induce terminal urothelial differentiation [6] and suggest that transplanting a more differentiated urothelium would contribute to the success of composite cystoplasty by rapid instatement of a functional urinary barrier.

Lima et al.[20] reported that the contraction which invariably occurs when a de-epithelialized colonic segment is used for augmentation can be prevented by postoperative distension of the bladder with an intravesical balloon. However, the nature of the neo-epithelium in the model reported by Lima et al. has not been adequately characterized. The present data allow no meaningful conclusions on the effect of using an intravesical balloon, as this was used routinely in the nine animals undergoing composite cystoplasty, with the aim of gently distending the augmenting segment and maintaining apposition between the urothelium/polyglactin complex and de-epithelialized surface. Although one control animal was augmented with de-epithelialized colon and an intravesical balloon, and died from presumed infection at 4 weeks, it is unclear whether this was a direct consequence, as there were no adverse effects in the other controls in which balloons had been used. A balloon was used in only one of the two sham-operated animals, with no difference in recovery between them, and the macroscopic and histological findings were normal in both at death. A partially filled spherical tissue expander was used for bladder distension, whereas the intravesical balloon designed by Lima et al. is likely to be better suited for further experimental studies and eventual clinical use.

In conclusion, we have established a model for composite cystoplasty suggesting that the approach of using autologous cultured urothelial cells to reline a de-epithelialized smooth muscle tissue is feasible, although it requires further refinement to improve the outcome. Future work will address limiting the exposure of non-barrier tissues to urine by (a) transplanting in vitro-generated urothelium with superior differentiated and functional properties, and (b) improving contact between urothelium and stroma through the use of a more compliant biomaterial carrier and better designed intravesical balloons.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the immunohistology work of Mrs Christine Gascoigne and the theatre and husbandry skills of Mrs Carol Knott and Mr Andrew Horner. This work was supported Action Research and Ethicon Ltd. J.S. is supported by York Against Cancer. Antibodies LP1K, LE41, LL002 and LL025 were supplied by Cancer Research UK, Lincoln's Inn Fields, London, and the anti-AUM antiserum was the generous gift of Dr T-T Sun, New York University Medical Center, New York.

Ancillary