The era of experimental small bowel transplantation was heralded by the seminal paper of Dr Alexis Carrell, who in 1902 successfully transplanted segments of canine small bowel with vascular anastomoses to the great vessels of the neck. In 1959 Lillehei et al. successfully autografted autograft canine small bowel, using the mesenteric vessels to revascularize the bowel, whilst the first rodent model was described by Monchik and Russell in 1971.

The increasing technical success of renal transplantation in the late 1960s led to the establishment of several experimental small bowel transplantation programmes, but enthusiasm waned following the failure of all clinical attempts in humans. Interest was only reawakened by the improved immunosuppression offered by cyclosporin A in preventing rejection of renal allografts in the early 1980s, and there is now considerable activity in this field, with more clinical trials of small bowel transplantation being reported. At present the over-riding clinical problem is acute rejection of the graft, but many other potential problems are being addressed in the laboratory. These include:

1.Methods of reducing the incidence of rejection;

2.Early diagnosis and effective treatment of rejection episodes;

3.Graft-versus-host disease. Intestinal transplants contain more immunocompetent cells than any other solid organ graft, and if rejection can be prevented, graft-versus-host disease may become a problem;

4.Absorptive function of the graft. This depends on functional integrity of the gut wall, adequate venous and lymphatic drainage, and the establishment of effective peristalsis;

5.Techniques for optimal preservation of the graft;

6.The significance of the portosystemic shunt resulting from venous drainage directly into the systemic circulation;

7.The maintenance of the ‘barrier’ function of the gut mucosa against external pathogens following transplantation.

Monchik and Russell used inbred rat strains and their F1 hybrid offspring (semiallogeneic combinations) to produce unopposed rejection or graft-versus-host disease (Fig.1) 719. Recently there has been a tendency to study fully allogeneic models (in which both graft-versus-host disease and rejection may occur) as being more representative of the clinical situation. The use of mice has been limited to non-vascularized transplants of fetal tissue; pigs and dogs are used for studies in large animals. Fetal tissue can be transplanted as a non-vascularized graft, and is usually placed in the subcutaneous tissue of the back, from where it may subsequently be removed and anastomosed to the recipient's own gut. Although this allows the study of the immunological events following such a procedure, the relevance to clinical practice is limited.

Two basic experimental models of vascularized grafts are currently used: heterotopic, in which the ends of the graft are brought out to the skin of the anterior abdominal wall as a Thiry-Vella fistula, and orthotopic in which, after resection of most of the recipient's own gut, the proximal end of the graft is anastomosed to the recipient's duodenum and the distal end of the graft is anastomosed to the recipient's terminal ileum.

Part or all of the small intestine is removed, along with its vascular supply (portal vein and superior mesenteric artery with an aortic patch). The lumen is flushed to remove intestinal contents and the graft vasculature may be flushed with cooled saline. Using microvascular techniques, the donor aortic cuff is anastomosed to the recipient aorta. In the case of an orthotopic graft, venous drainage is established by anastomosis of the portal vein of the graft to that of the recipient. When the graft is placed heterotopically the venous drainage may either be into the recipient inferior vena cava or into the portal vein.

Portal venous drainage is favoured by some investigators: rat cardiac and renal allografts may show enhanced survival when venous drainage is directed into the portal circulation, and there is concern about the desirability of creating a portosystemic shunt in metabolically compromised patients. Whilst prolonged graft survival is documented in semiallogeneic rat models with portal venous drainage, this has not been observed in fully allogeneic combinations or in studies using larger animals. Present experimental evidence suggests that although metabolic disturbances can be detected if the venous drainage is into the systemic circulation, these may not be significant in clinical practice.

The recipient of an orthotopic graft depends on the transplanted gut for nutrition, making this model especially useful for nutritional and metabolic studies. The heterotopic model has the advantages that direct access to the graft lumen is possible via the stomata and that the nutritional state of the animal is independent of graft function. In the long term, however, the intestinal wall atrophies and the ‘barrier’ function of the mucosa is lost.

The technical failure rate of 10 to 30 per cent for small animal models is mostly due to vascular complications and intestinal anastomotic leaks. Despite the relative ease of creating vascular anastomoses in large animals, technical failure rates of 15 to 50 per cent are reported, to which intussusception and volvulus contribute significantly.

Histological examination of a rejecting graft at 6 to 7 days shows a diminution in the height of the villi and loss of villous epithelium. The brush border is lost and goblet cells are depleted. These changes are followed by widespread epithelial loss and infiltration of monocytes and macrophages into the lamina propria. Complete infiltration of the graft by inflammatory cells, with widespread destruction of the graft architecture, occurs by the ninth day. The nature of the infiltrating cells has been extensively studied using immunohistological techniques but the reported results are not particularly consistent.

There are several technical problems in the use of histology in both clinical and experimental practice. Firstly, a full thickness biopsy is essential if accurate assessment is to be made. Secondly, because of changes resulting from local trauma, biopsies from the stoma may not be representative of changes elsewhere. Thirdly, procurement of an adequate biopsy sample from the depths of the lumen of the graft may result in perforation and peritonitis. These problems have prompted the search for a simple, reliable, and rapid test to assess rejection. Although glucose, maltose, and xylose absorption are decreased during rejection, doubts have been expressed about the specificity of these observations and these changes are not seen until after the onset of histological evidence of rejection. The serum level of the mucosal enzyme N-acetyl hexosaminidase is increased during rejection, at the same time as or before the earliest detectable histological changes. This is a rapid biochemical assay which may prove to be useful in clinical practice. Intestinal absorption of low molecular weight polyethylene glycol (usually almost non-absorbable) can be assessed by measuring urinary levels. Increased levels are detected at the same time as or immediately before histological evidence of rejection becomes evident. The transmural potential difference across the graft wall is reduced during rejection. This decrease is apparent at the same time as the earliest histological changes. Monocyte procoagulant activity (a measure of the ability of peripheral blood monocytes to shorten the clotting time of human plasma) is enhanced during rejection and graft-versus-host disease, the effect just preceding histological changes. In rodents a simple and specific indicator of rejection which reliably precedes histological changes is the presence of an abdominal mass, caused by swelling of the graft and mesenteric lymph nodes.

A major concern (especially with regard to absorption of fat-soluble compounds such as cyclosporin A) is whether lymphatic drainage of intestinal allografts will be adequate. Studies in canine models have shown that lymphatic drainage commences at 2 weeks and is fully established by 4 weeks. In the rat drainage can be observed as early as 3 days after syngeneic transplantation, but full flow is not seen until 14 days. Drainage may be either by spontaneous lympholymphatic communication or development of lymphovenous anastomoses at the porta hepatis (as has been demonstrated in pigs).

It was shown in the 1960s that denervation of the canine intestine resulted in temporarily abnormal motility and gut function, whether the denervation resulted from autotransplantation or merely division of the splanchnic nerves. Myoelectrical activity returns on day 2 in unimmunosuppressed canine allografts and only ceases when the graft is destroyed by rejection. In immunosuppressed animals, however, electrical activity is maintained even in chronically rejected grafts. In canine allografts intraluminal pressure increases until day 6 or 7 and undergoes a marked reduction on day 8 post-transplantation, coinciding with the onset of overt rejection. In the rat, smooth muscle contractility and response to electrical stimulation and autonomic agonists is not impaired in syngeneic small bowel transplants. Since initiation and co-ordination of peristaltic activity takes place principally within the gut wall itself, it is likely that contractility and motility of small bowel transplants will be adequate (but not necessarily normal).

The effect of immunosuppressive regimens in the rat is very dependent on whether semi- or fully allogeneic strain combinations are used: it is much more difficult to obtain long-term survival in the latter. Prior to the introduction of cyclosporin A, conventional immunosuppression failed to prolong graft survival significantly in any model. Cyclosporin A monotherapy produces long-term survival, or at least marked prolongation of survival, in various semiallogeneic and fully allogeneic rat models. The experience in large animal models has been less encouraging, with cyclosporin A significantly prolonging survival time, but producing few long-term dog or pig survivors. Whilst acute rejection is seen less often, chronic rejection is responsible for a considerable number of failures. The addition of other immunosuppressive agents such as azathioprine, antilymphocyte globulin, graft irradiation, portal drainage, and recipient splenectomy has not improved results significantly. The new macrolide immunosuppressive agent FK506 appears to be much more effective than cyclosporin A in preventing rejection of rat intestinal allografts, and its use in clinical small bowel transplantation in Pittsburgh appears encouraging.

Preoperative donor-specific transfusion in combination with cyclosporin treatment improves survival in some fully allogeneic rat combinations, but not in others. Treatment with monoclonal antibody directed against the interleukin–2 receptor prolonged survival in one semiallogeneic rat combination, an effect which was enhanced by concurrent administration of cyclosporin A.

Graft-versus-host disease must be considered a potential problem in small bowel transplantation because of the vast numbers of immunocompetent cells transferred within the Peyer's patches and mesenteric lymph nodes of the graft. The disease may readily be seen in appropriate semiallogeneic rat models and also in certain fully allogeneic strain combinations. Graft-versus-host disease has been observed infrequently in dogs, and has never been reported in pigs. Graft-versus-host disease may be prevented either by reducing the lymphoid component of the graft or by the use of immunosuppressive agents, such as cyclosporin A or FK506. Three methods are available for reducing the amount of lymphoid tissue transplanted: surgical excision of the mesenteric lymph nodes (a relatively simple procedure in the rat); irradiation of the graft before implantation; or transplantation of only half of the donor small intestine.

Although rat allografts may be stored safely for up to 24 h by a combination of intravascular and luminal flushing, canine experiments have suggested a maximum storage time of up to 12 h. Copious flushing of the graft lumen appears to be an essential part of any preservation protocol.

Present evidence suggests that, in the absence of rejection, absorption by the graft of all nutrients is adequate. Although subtle differences can be detected in water and electrolyte absorption by rat isografts and allografts, the animals remain healthy and gain weight normally. Similarly, in the absence of rejection the mucosal barrier is maintained.

Rejection (in which there is dramatic loss of the brush border and epithelial lining of the graft) results in marked reduction in absorptive function. The barrier function of the graft is severely compromised, allowing bacterial translocation into the host.

Carrel A. La technique operatoire des anastomoses vasculaire et la transplantation des visceres. Lyon Med, 1902; 98: 859.

Lillehei RC, Goott B, Miller FA. The physiological response of the small bowel of the dog to ischemia including prolonged in vitro preservation of the bowel with successful replacement and survival. Ann Surg, 1959; 150: 543–60.

Milland PR, Dennison A, Hughes DA, Collin J, Morris PJ. Morphology of intestinal allograft rejection and the inadequacy of mucosal biopsy in its recognition. Br J Exp Pathol, 1986; 67: 687–98.

Monchik GJ, Russell PS. Transplantation of small bowel in the rat: technical and immunological considerations. Surgery, 1971; 70: 693–702.

Schraut WH. Current status of small-bowel transplantation. Gastroenterology, 1988; 94: 525–38.