Multivisceral transplantation (MVTx) is defined as the exenteration of the native viscera followed by an en-bloc transplantation of stomach, liver, pancreas and small bowel. It is proposed as a radical therapeutic option for extensive abdominal pathology that is otherwise untreatable (1). In this procedure, a single cluster of organs is implanted on a combined arterial patch including celiac trunk (CT) and superior mesenteric artery (SMA) with venous outflow being provided through the inferior vena cava (IVC). Currently, the most frequent indication for MVTx is diffuse portomesenteric thrombosis (DPMT) (2). DPMT is defined as complete thrombosis of the portomesenteric vessels, resulting in severe portal hypertension and aberrant collateral circulation, with a major risk of gastro-intestinal bleeding (3). DPMT can present in two clinical forms which we propose to classify in Type I [associated with end-stage liver disease, seen in some liver transplant (LTx) candidates] and type II (no advanced liver disease justifying LTx, but life-threatening complications caused by portal hypertension) (Figure 1). The proportion of MVTx among all forms of intestinal transplantations (ITx) has increased up to 21% (4) but the survival remains inferior (5). The reasons for this are both medical and surgical. Medically, MVTx patients are often severely weakened in the pre-operative phase. After transplantation, they suffer from higher incidences of infection and graft-versus-host disease (GVHD) compared to other types of ITx (6). Surgically, MVTx represents the most invasive abdominal procedure that can be performed. In case of DPMT, MVTx is even more challenging due to the severe bleeding from the engorged collateral circulation that inevitably occurs during the resection phase (7). In type I DPMT, bleeding can be further exacerbated by liver disease-induced coagulopathy and severe adhesions. Over the last 10 years, survival rates have improved, mainly due to better patient selection, refinement in immunosuppressive protocols, use of extensive infectious prophylaxis (8), and accumulated surgical experience. However, MVTx is only performed in a few ITx centers worldwide and the technical aspects of the procedure have not been described in detail.

We aim to review our experience in MVTx, report technical challenges in detail, and propose surgical strategies to overcome them in order to improve the outcome.

The study was a retrospective cohort analysis of a prospectively maintained database of ITx performed at the University Hospitals Leuven, Belgium between 01/01/2000 and 31/12/2020. Any patient receiving a MVTx transplantation was included. All patients remained in follow-up at our institution per protocol described elsewhere (9). Donor data included: age, gender, weight, cause of death, BMI, ABO, Cytomegalovirus status and days admitted. Recipient data included: age, gender, admission status, cause of DPMT, ABO compatibility, DPMT type, cold ischemia time (CIT), warm ischemia time, survival and outcomes including postoperative complications. Outcome data not already present in our prospective database, was extracted from patients' files maintained at our institution. These included medical and surgical complications. Complications were defined according to the Clavien-Dindo classification (10). All data is presented as median (range) unless stated otherwise.

Out of 22 ITx patients performed at our center between 2000 and 2020, 11 (50%) were isolated ITx, 6 (27%) were combined liver bowel transplants, and 5 patients underwent a MVTx (23%). Age at time of MVTx was 47 years (23–62) with a BMI of 24 kg/m² (15–33). They were all male. All had symptomatic DPMT -Yerdel grade 4 (3)- with recurrent life-threatening gastrointestinal bleeding and portal congestion. Underlying causes are listed in Table 1. Two patients were hospitalized at the time of transplantation (Patient 1: upper gastrointestinal bleeding and Patient 3 for liver decompensation). Patient 2 additionally had intestinal failure and recurrent ascites. The lab-MELD score was 13 (10–32). Patient 3 had severe liver disease (Type I DPMT) while the other patients did not (Type II DPMT). Patient 3 also had severe renal failure. All patients received a graft from an ABO identical or compatible brain-dead donor. Additional donor data is listed in Table 2. All donors underwent a standardized pre-treatment protocol described elsewhere (9). University of Wisconsin preservation solution was used in the first 3 donors (6–7 liters) while institute George Lopez-1 preservation solution was used in the last two donors (6 liters) (13). The CIT was 331 min (316–416).

All embolization procedures were uneventful. Embolization in the first 3 patients (using embolic agents) took 80 min (70–90). Plug embolization of the CT and SMA in the last two patients was performed in 35 and 50 min, respectively.

The median graft implantation time (warm ischemia) was 38 min (24–44). In patient 1, a cuff of native stomach was left in place and a proximal gastro-gastric anastomosis was performed. A segment of ascending colon was transplanted and a distal colo-colic anastomosis was performed. In the other patients, an esophago-gastric anastomosis was performed proximally and an ileocolonic anastomosis distally without transplanting colon. There were no intraoperative deaths. Intra-operative transfusion requirement was 3 units of blood (0–5), despite low pretransplant hemoglobin levels of 7.4 g/dL (5.6–10.7). The first 3 patients received a gastrostomy and jejunostomy while the last two patients only received a nasogastric tube. All patients received a double loop ileostomy in the right lower quadrant. In all recipients, primary closure was possible and thus no rectus fascia was used. At the end of the MVTx, patient 3 received an additional kidney transplantation placed retroperitoneally in the left lower quadrant through a separate hockey stick incision using standard technique. Total recipient procedure time was 588 min (510–780).

MVTx involves a resection of all splanchnic organs followed by en-bloc transplantation. The technique was first developed in animal models by Starzl et al. in the 1960s (15, 16) and was then applied in man for the first time in Pittsburgh by Starzl et al. (1) and in Chicago by Williams et al. (17) in 1989.

Since then, untreatable DPMT has been the leading indication for MVTx (2). Although experience has been expanding in the last 2 decades (5, 6), the first large series for MVTx for DPMT was only recently described in 25 patients (7). The reported 72% 5-year survival was encouraging despite a high complication rate. In the present case series, we describe in detail the techniques used for MVTx for DPMT, and propose potential solutions for the encountered surgical problems.

The most challenging part of MVTx is the removal of the native organs, particularly in DPMT due to the massively congested collateral circulation (8). In the largest series described so far, Vianna et al. reported a median peri-operative blood transfusion of 29 units (5–146) (7). To limit bleeding, the arterial inflow through the CT and the SMA should be suppressed as soon as possible. The difficulty is that this exposure requires extensive mobilization of the viscera. This can in itself cause severe blood loss and hemodynamic instability. To accelerate access to the CT and SMA, some authors recommend transecting either the transverse colon and pancreas (2) or the esophagus (18) followed by mass clamping of vascular inflow, but this too can cause major bleeding.

The major advantage of our pre-operative embolization technique is that complete suppression of arterial flow is obtained before surgery starts. Because the portal circulation is interrupted, the abdomen becomes a virtually bloodless field (12). This was demonstrated previously in cirrhotic patients, where temporary occlusion of splanchnic arterial inflow greatly reduced portal pressure gradients (19).

One limitation of the embolization technique in type 1 DPMT (Figure 1) is that it does not allow an attempt at an isolated liver transplantation (LTx). With the back-up of a MVTx graft Vianna et al. were able to perform an isolated LTx in 6 out of 30 cases (19%) using either thrombo-endovenectomy or venous jump graft (7). The remaining intestinal graft was then transplanted into another recipient. In our patients however, pretransplant imaging showed chronic, complete portal vein thrombosis (i.e., Yerdel grade 4, including the splenic vein) which would have made a standard LTx impossible (3, 20). The only remaining option would have been a cavo-portal transposition, a procedure that does not resolve portal hypertension and leads to persisting gastrointestinal bleeding and ascites (20, 21).

In case of type II DPMT (Figure 1), an isolated LTx would not be appropriate since liver function is usually next to normal. In these cases, MVTx is the only therapy to replace the “failing portal system” if it cannot be decompressed by interventional or surgical shunts (22, 23).

In our first 3 patients, we used a mixture of embolic agents to embolize the CT and SMA. It took up to 90 min to completely occlude all vascular side branches. In order to shorten the embolization time, the Cambridge team modified our technique and used vascular plugs instead of embolic agents (24). A vascular plug may also allow a more selective embolization, for example sparing the left gastric artery (or even the hepatic artery to preserve the stomach and the liver in modified MVTx). One risk, however, is the migration of a plug into the aorta as we faced in patient 4 (Figure 11). Plug migration has previously been described after other embolization procedures (25, 26) and this risk is particularly high when combined with surgical manipulation (27). In our patient we opted for a conservative therapy because it was a wide meshed plug (first generation Amplatzer^©) in a high flow location without hemodynamic or embolic complications. In case endovascular plug extraction is considered, it should be done early after implantation (within 3 days). At a later stage, surgical extraction would be the only option (25).

The Miami group recently performed preoperative embolization in 3 MVTx in patients with a hostile abdomen (28). The first patient died intraoperatively after a difficult and lengthy exenteration and mass transfusion related coagulopathy (97 units transfused). Distal migration of the plug into the gastroduodenal artery resulting in re-arterialization of the liver may also have played a role. To avoid plug migration for their next two cases, they switched to distal embolization using embolic agents. Furthermore, they also spare the hepatic artery to shorten the anhepatic phase, and the left gastric artery to prevent gastro-gastric anastomotic leaks. However, sparing tributaries of the CT (or the SMA) maintains a persistent arterial and portal flow and may reduce the hemostatic effect of the embolization (29).

We feel that the risk of migration does not outweigh the benefits of a shorter embolization time. In our last case, the risk of intra-aortic migration was minimized by placing the plug slightly more distally and by clamping the CT and SMA proximal to the plugs. In addition, we now verify the position of the plugs fluoroscopically at the end of the transplantation. In case of plug migration, it can then be removed endovascularly (30) or through an arteriotomy (25).

Most centers remove the native viscera including the liver, in one bloc. In contrast, our exenteration method proceeds in two-steps. We first transect the liver hilum en-bloc (a maneuver without risk of portal congestion due to the preoperative embolization) and then remove the gastrointestinal tract (stomach, duodenum, pancreas, spleen, and intestines) while leaving the liver in situ. We then implant the donor aortic tube onto the infra-renal aorta, while the devascularized liver is still in place. When the graft is ready for implantation, we start veno-venous bypass and the liver is removed. The MVTx graft can then be implanted. This two-step exenteration technique and veno-venous bypass preserves normal systemic venous flow and hemodynamic stability.

For implantation, we use the caval replacement technique instead of the more commonly used “piggy back” technique. Both techniques have been used in isolated LTx and there is no evidence for the superiority of either (31). However, in the context of MVTx we believe that the classical caval replacement technique has the potential advantage to “anchor” the graft and prevent retrohepatic IVC rotation which is known to occur after MVTx (32).

For arterial inflow, we anastomose the donor patch end-to-end to the already placed infra-renal aortic conduit (Figures 8A,B). Arterial implantation onto the supra-celiac aorta either directly or through a conduit has also been described (18). The advantage of the infra-renal implantation is that this area is easier to reach and renal ischemia is avoided. In LTx, these conduits have a high thrombosis risk because blood must flow against gravity (33). However, in MVTx the outflow resistance is much lower, making these conduits less prone to thrombosis. Indeed, arterial thrombosis occurs in only 2.8% of all ITx compared to 9% in LTx alone (34, 35).

The arterial conduit is prone to infections, given that some intestinal contamination is inevitable. In our series, patient 3 developed a mycotic aneurysm after 5 months. This was corrected before any rupture could occur using aortic and iliac arterial homografts (14) (see Figure 10). The reported incidence of mycotic aneurysms was 2.5% in a large series of composite visceral allografts (n = 7/285) (34). Unlike in our case, most patients presented with an acute rupture and hemodynamic instability. In this setting, temporary endovascular occlusion can stop the bleeding and allow sufficient time for surgical repair (34). However, despite prompt treatment, this complication is often fatal. In an attempt to reduce perioperative contamination, we now perform extensive abdominal lavage (10 liters of warm physiological solution) with antibiotics (Meropenem) and antifungal medication (Amphotericin B) at the end of the procedure. Also, all patients receive 1 week of antibiotic prophylaxis.

Needless to say, construction of well vascularized intestinal anastomoses is crucial. The first patient developed a leak at the gastro-gastric junction, possibly secondary to the ischemia of the remaining cuff of native stomach, inherent to the complete embolization of the CT. Since then, we remove the native stomach completely and we perform an esophago-gastrostomy. An alternative to keep the upper part of the native stomach well vascularized would be to selectively embolize the CT and spare the left gastric artery (12, 28).

To prevent gastric outlet obstruction and gastroesophageal reflux in a denervated stomach, a pyloromyotomy and a Nissen fundoplication are performed. Additional advantage of the later is that it also protects the proximal esophago-gastric anastomosis.

Obliteration of the CT and SMA (by tying or embolization) leaves the inferior mesenteric artery open and the sigmoid and rectum remain vascularized. It remains important to transect the sigmoid well distal from the demarcation line and allow a safe margin to avoid anastomotic ischemia and leakage as seen in patient 1.

A few centers advise the transplantation of the colon, either partially or completely, to augment the gastrointestinal resorptive function (36). This trend is seen globally in the ITx registry and seems to result in less episodes of dehydration and faster weaning from parenteral nutrition (4). However, our patients had a sufficient length of native sigmoid and rectum, obviating the need for additional colon transplantation and its associated morbidity (37, 38).

At the end of the procedure, a distal ileostomy is routinely constructed in order to provide easy access for endoscopic biopsies. We prefer a loop ileostomy that can be closed easily after a few months. In an effort to reduce ostomy-related morbidity, the Miami group either creates an hybrid ostomy using an excluded ileal segment or avoided an ostomy altogether (36). However, this limits access to the graft for surveillance and may put the distal ileocolonic anastomosis at risk for leakage.

Essential in the success of MVTx is the maintenance of hemodynamic stability during the entire procedure. Firstly, limiting the exenteration/ischemic phase is crucial to reduce the risk of acidosis and metabolic instability. The use of intra-operative renal replacement therapy has been advocated by Cambridge and Miami to eliminate circulating waste products from ischemic organs during exenteration, and better maintain metabolic homeostasis until the new graft is functioning properly (24, 28). Equally important to the success of the procedure, is a short CIT (ideally <5 h). The recipient should be anesthetized immediately after approval of the donor organs. The patient should not remain anhepatic for too long (ideally <2 h). For this, a perfect coordination between donor and recipient team is essential.

The inclusion of the spleen in the MVTx graft and the fate of the native spleen is another contentious issue. Proponents of transplanting the spleen point to its potential to reduce rejection and infections, especially in children (39, 40). However, transplanting the spleen increases the risk of GVHD (41). The Madrid team preserves the native spleen without transplanting the donor spleen in children (42). This has the theoretical advantage of preventing infections while reducing the risk of GVHD. However, not all patients have a functional spleen and its preservation can substantially increase exenteration time. In the setting of DPMT, spleen preservation is virtually impossible. At our center, both native and donor spleen are removed. To prevent infections, we vaccinate our patients for S. pneumoniae, N. meningitidis, H. influenzae type b and influenza virus prior to MVTx (43). No infections or thrombosis associated with asplenia were seen in our patients. We also did not see GVHD or lymphoproliferative disorders.

The intestine is extremely immunogenic and prone to rejection (44). However, the risk of severe rejection after MVTx is lower compared to isolated ITx due to the immunoprotective effect of the liver (5, 8). In our series, only one patient had a severe grade III rejection. Early asymptomatic rejection was detected on protocol biopsy in 3 patients but could easily be controlled with corticosteroids.

Several limitations with this study have to be acknowledged. As this is a rare entity, the numbers are low which makes generalization difficult. Also, this study limits itself to one indication of MVTx, namely DPMT, which is also the most challenging one. Certain surgical aspects, such as the embolization, are specifically useful in this category and may not be extrapolated to all MVTx procedures. Only patients that actually underwent MVTx (after multidisciplinary approval) were included in the study which leads to an inevitable selection bias.

In conclusion, MVTx is the only definitive treatment in selected patients with refractory DPMT. We present a series of strategies to decrease the risk associated to this extremely invasive procedure. Preoperative embolization, sequential native organ extraction, standard use of venous-venous bypass and synchronization between donor and recipient teams all contribute to reduce perioperative mortality. MVTx can also be applied in otherwise untreatable extensive intra-abdominal diseases.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Ethics Board of the University Hospitals Leuven (Registration Number: S61313). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

EC, LC, NG, and JP: conception and design. EC and LC: administrative support. IJ, TV, DM, and JP: provision of study materials or patients. EC, LC, NG, GD, and JP: collection and assembly of data. EC, LC, NG, and JP: data analysis and interpretation. All authors: manuscript writing and final approval of manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.