Results
Descriptive Analysis
The complexity of the study cohort is demonstrated in Table 1. With ages ranging from 19 to 61 years at time of primary BS, 77% were Caucasian and 85% were female. With a diversified origin of referral, 42% were national and international patients and most of the regional cases had their BS done outside both UPMC and CCF. Most subjects had moderate to severe obesity with BMI up to 89 kg/m. RYGB was the main surgical procedure and laparoscopy was the most commonly utilized technical approach. Multiple bariatric and subsequent surgical interventions were observed with a maximum of 44 abdominal operations. Short gut syndrome developed in 54% of the patients. The severity of disease gravity is further compounded by the coexistence of liver failure (4%), renal failure (5%), portomesenteric venous thrombosis (6%), major psychiatric disorders (20%), prothrombotic state (69%), and significant hepatic pathology (73%). In addition, impaired skeletal health was documented in 19 (61%) of 31 studied patients. The mean duration of TPN was 21 ± 33 months with a range of 2 to 252.
Causes of GF are categorized in Figure 2. Visceral vascular occlusion, including strangulated internal hernias and primary vascular thrombosis was the precipitating cause of gut loss in 52 (94%) of the patients with Type-I GF (Fig. 2A). The remaining 3 patients suffered iatrogenic vascular injury at time of BS (n = 2) or at a later date during resection of mesenteric desmoids (n = 1). Intermittent abdominal pain with frequent visits to the emergency room was documented in most patients with internal hernia. A hypercoagulable state with hereditary disorders, including deficiencies of protein C, S, and antithrombin-III and mutations of factors V and II were detected in most of the patients with primary vascular occlusion who underwent hematologic studies. Nonetheless, thrombophilia was positive in 69% of the 87 screened patients (Table 1). Pregnancy preceded the catastrophic event in 5 females.
In Type-II GF patients, gastrointestinal fistulae and loss of gut continuity due to stapling or external diversion were the main causes of GF with 21 (49%) patients experiencing more than one pathology (Fig. 2B). The causes of the negative energy balance with Type-III were motility disorders, restrictive intolerance, and malabsorption (Fig. 2C). The dysmotility was de novo or globally developed in patients who underwent BS for gastroparesis or gastroesophageal reflux disease. Interestingly, two of the de novo dysmotility patients experienced symptoms after the development of neurological deficits due to severe malnutrition. The clinical syndrome of restrictive intake includes inability to eat, postprandial pain, anorexia, and other post-BS eating disorders. Refractory malabsorption was observed in patients with jejunoileal bypass and those with long RYGB alimentary limb.
The results of the univariate statistical analysis comparing the three types of GF among the study population and surgical cohort are shown in Table 1 and Table 2, respectively. Despite being a component of each of the 3 types of GF, short gut syndrome was the most significant describing feature of Type-I. Its variable association with Type-II and Type-III GF was due to prior surgical interventions in patients with multiple abdominal pathology with a single example of ultrashort gut in the Type-II category. With multivariate analysis, short gut syndrome (P = 0.000), time of referral from onset of GF (P = 0.003), surgical interventions (0.008), and multiple prior bariatric procedures (P = 0.047) continued to be the most significant defining variables reflecting the underlying etiology and pathophysiology of gastrointestinal failure in each type.
Restorative surgery with autologous reconstruction and visceral transplantation was performed in 116 (89%) of the 131 study patients with clinical features summarized in Table 2. Of the 15 nonsurgical group, 6 were Type-I/II; 2 were prohibitive operative risk, 2 were awaiting surgery, 1 was successfully reconstructed elsewhere, and 1 was lost to FU. The 9 remaining Type-III patients received successful medical management (n = 8) or continued to require nutritional support because of insurance denial for surgery (n = 1).
Visceral transplantation was required in 23 (20%) of the 116 surgical patients with a total of 25 allografts, 21 (84%) liver-free, and 4 (16%) liver contained (Table 3). Of the liver-free allografts; 3 were modified multivisceral that were given to patients with prior total gastrectomy (n = 2) and severe gastroparesis as part of global dysmotility (n = 1). The liver-contained allografts were transplanted in 3 patients with TPN-associated liver failure and 1 recipient after failure of primary isolated intestinal graft.
Surgical Techniques
All surgical procedures were performed with an open approach because of previous multiple abdominal operations and existing complex abdominal pathology. External ureteric stents were utilized for patients with hostile lower abdomen. Tissue dissection was mostly sharp with generous use of thermal hemostatic tools. Excision of all scar and granulation tissues was essential with careful preservation of the segmental visceral blood supply and any aberrant vasculature particularly of the liver. All anastomoses were tension free and hand sewn in two anatomic layers using standard techniques. Pyloroplasty was performed for patients with foregut reconstruction. Surgical mesh was removed with abdominal wall debridement. Closure was done in a single layer using nonabsorbable material particularly in patients with infected abdomen.
Most of the surgical patients required more than one anatomical reconstruction with a total of 317 different procedures (Table 4). Combined foregut and midgut reconstructions were often required in Type-I and Type-III GF patients with RYGB. The operative time ranged from 4 to 14 hours with a median length of hospital stay of 17 days (range: 7–115).
Autologous Reconstruction
Of the 317 total surgical procedures, 198 (62%) were autologous reconstruction; 191 (96%) primary; and 7 (4%) with an interposition alimentary conduit. Foregut reconstruction was performed in 88 (76%) patients (Table 4). Techniques were primary in 81 (92%) patients (Fig. 3A) and with an interposition alimentary conduit in 7 (8%) (Fig. 3B). Primary reconstruction was gastrogastric (n = 67), standard Roux-en-Y gastrojejunostoy (n = 6), esophagogastric (n = 5), and gastroplasty (n = 3). The interposition conduits were jejunal (n = 2) and colonic (n = 5). It was paramount to preserve the main left gastric artery and its branches particularly in the presence of interrupted gastroepiploic and short gastric vessels. Great attention was directed towards any vascular anomalies including an aberrant left hepatic artery. All efforts were made to avoid blunting of the angle of His.
(Enlarge Image)
Figure 3.
A, Surgical techniques utilized for foregut reconstruction; hand-sewn gastrogastric and esophagogastric anastomoses (a-c) with pyloroplasty. Note interruption of the short gastric vessels with preservation of the main left gastric artery along the lesser curvature of the retained stomach. Gastroplasty (d-e) was required in patients with nondivided partitioned stomach (d) and those with strictured incisura angularis after sleeve gastrectomy (e). B, Interposition alimentary conduits utilizing jejunum (a) and colon (b-d) for final foregut reconstruction (a-c) or as a first-stage operation in preparation for a modified multivisceral transplant that included the stomach, duodenum, pancreas, and intestine (d). C, Midgut reconstruction and bowel lengthening determined by the length of remaining intestine. Duodenocolonic anastomosis in patients with ultrashort gut syndrome (a). Multiple enteroenteric anastomoses with different alignments guided by discrepancy in the intestinal diameter (b). STEP in patients with less than 100cm residual small bowel (c). STEP, serial transverse enteroplasty.
Midgut and hindgut reconstruction was the most common operation with a total of 110 procedures (Table 4). Re-establishment of mid gut continuity was required in 100 patients and the anastomoses were performed between duodenum or upper jejunum and remnant ileum or colon utilizing different alignment techniques (Fig. 3C). Hindgut reconstruction was performed in 10 (9%) patients with colocolic (n = 9) or colorectal (n = 1) anastomoses.
Reversal of Bariatric Surgery
BS was reversed in 84 (72%) and retained in 32 (28%) patients with pertinent data summarized in Table 2. Of the 32 retained bariatric procedures, 23 (72%) were RYGB, 7 (22%) were SG, 1 was laparoscopic adjustable gastric banding, and 1 was biliopancreatic diversion. The decision to preserve BS was guided by the type of GF, presence of short gut syndrome, reversibility of bariatric anatomy, indications for BS, and patient's wish to maintain the weight reduction procedure.
Intestinal Lengthening
Serial transverse enteroplasty (STEP) was utilized in 10 patients. All but one had Type-I GF. The operation was performed twice in one patient. An illustration is shown in Fig. 3C (c) with the technique described elsewhere. All patients were off TPN at a mean FU of 32 ± 15 months.
Visceral Transplantation
The standard recipient operation (Figure 3, Supplemental Digital Content, http://links.lww.com/SLA/A846) was performed in all but two recipients. The previously utilized native colon for foregut reconstruction was partially retained at the time of transplant in two recipients. In one patient, the colonic conduit connected the second part of native duodenum to the allograft jejunum (Figure 4, Supplemental Digital Content, http://links.lww.com/SLA/A846). The technical complexity of the second patient is illustrated in Figure 4. Upon referral, the patient had complete foregut disruption (Fig. 4A) with loss of native stomach that resulted in cervical esophagostomy with external drainage of the distal esophagus and duodenum. In preparation for transplantation, the native colon was used for foregut reconstruction to temporary restore continuity and control the intra-abdominal leak (Fig. 4B). At transplantation, the esophagocolonic anastomosis was preserved retaining a well-vascularized colonic segment via the marginal artery. The interposition conduit was utilized to lengthen the shortened abdominal esophagus (Fig. 4C).
(Enlarge Image)
Figure 4.
A case of extensive foregut loss 5 years after gastric bypass with 3-stage restorative surgery. The previously performed cervical fistula was repaired with primary esophageal reconstruction (A). The survived native colon was utilized to reconstruct the abdominal esophagus and remaining duodenum to control intraabdominal leaks in preparation for visceral transplantation (B). Modified multivisceral transplantation including the stomach, duodenum, pancreas, and intestine with preservation of the previously established esophagocolonic anastomosis as an alimentary conduit between the short abdominal esophagus and the transplanted stomach (C). The blood supply of the colonic conduit was maintained via a well-preserved marginal artery.
Complications
A total of 19 (16%) patients developed postoperative surgical complications (Table 4). Of these, 10 were gastrointestinal leaks (4 anastomotic and 6 thermal injuries) that required abdominal re-exploration with successful repair in 9. The remaining patient developed gastric venous infarction due to concomitant interruption of gastroesophageal varices and died of sepsis-induced liver failure (Table 5). The other 9 surgical morbidities were intra-abdominal and wound infections (n = 5), recurrent bowel obstruction (n = 2), shunt thrombosis in a combined liver-intestinal recipient (n = 1), and fatal pulmonary embolism (n = 1).
Of the transplant recipients, 1 developed PTLD 4 months after transplant and was successfully treated with anti-CD20 antibody. Another patient developed de novo adenocarcinoma of the native stomach 4 years after retransplantation. The patient was treated with partial gastrectomy and radical lymph node dissection with no evidence of recurrence 4 years after surgery.
Survival
With a mean FU of 137 ± 91 months (range: 2–440) from date of BS, 18 (14%) of the 131 study patients died; 2 before and 16 after surgical intervention. From date of BS, all-cause cumulative risk of mortality was 3% at 5 years, 10% at 10 years, 21% at 20 years, and 48% at 30 years (Fig. 5A). With surgical treatment, including transplantation, the overall patient survival was 96% at 1 year, 84% at 5 years, and 72% at 10 and 15 years (Fig. 5B). Patients with Type-III GF achieved better survival than those with Type-I/II (Fig. 5C). After visceral transplantation, patient survival was 91% at 1 year and 69% at 5 years with respective overall graft survival rates of 88% and 64% (Fig. 5D).
(Enlarge Image)
Figure 5.
Kaplan-Meier cumulative mortality and survival rates for the total study population and surgical cohorts, respectively. A, Cumulative mortality of the 131 study patients from the date of bariatric surgery. Estimated all-cause mortality rates were 3% at 5 years, 10% at 10 years, 21% at 20 years, and 48% at 30 years. B, Overall cumulative patient survival after restorative surgery including transplantation from date of surgery. C, Patient survival after restorative surgery according to type of GF. D, Patient and graft survival after visceral transplantation. GF, gut failure.
The individual causes of the 16 surgical mortalities are summarized in Table 5 with reference to date of bariatric and restorative surgery. The 2 nonsurgical patients died of drug overdose and liver failure at 6 and 3 years from date of BS, respectively. Accordingly, drug overdose was the precipitating cause of death in 4 (22%) of the total mortalities.
The causes of the 11 lost visceral allografts were patient death (n = 5) and allograft enterectomy (n = 6) (Table 3). Chronic and acute rejection was the cause of 5 allograft enterectomies 5 to 97 months after transplant. The remaining graft was lost due to intra-abdominal infection with ruptured pseudoaneurysm of the arterial graft 8 days after transplant.
Nutritional and Metabolic Outcome
Nutritional autonomy was achievable in 92 (81%) of the 113 total survivors; 9 with medical management and 83 after surgical treatment (Fig. 6A). In addition, 8 of the 18 total mortalities achieved nutritional autonomy before death. Reversal or amelioration of hepatic steatosis was observed after discontinuation of TPN as demonstrated in Figure 5, Supplemental Digital Content, http://links.lww.com/SLA/A846. In 90 patients with body weight measurements at last FU, 3 (3%) were underweight, 36 (40%) had normal weight, 29 (32%) were overweight, and 22 (24%) were obese. Of the obese patients, 62% were class-I, 19% were class-II and 19% were class-III. Of the 9 medically managed patients, 1 was underweight, 1 had normal weight, 4 were overweight, and 3 were obese with class-I.
(Enlarge Image)
Figure 6.
Nutritional autonomy among current survivors after restorative surgery including transplantation. A, Achievement of gastrointestinal nutritional autonomy defined by freedom from total parental nutrition (TPN). B, BMI of 60 patients with restoration of nutritional autonomy at time of BS, referral, and last FU with a mean (SD) BMI of 49±12, 26±7, and 27±6, respectively. The remaining 23 patients with BMI values that were not obtainable at one of the 3 study points had a current mean BMI of 25±5. The dotted horizontal lines mark BMI of 18.5 kg/m (lower), 24.9 kg/m (middle), and 29.9 kg/m (upper). BS, bariatric surgery; BMI, body mass index.
With surgical treatment, nutritional autonomy was restored in 87% of the transplant recipients and 82% of the autologous reconstruction plus bowel lengthening-alone patients patients with an overall rate of 83% (Fig. 6A). Grouping of BMI after restorative surgery with and without transplantation is shown in Table 4. Interestingly, 31% of the patients in each subcohort were overweight and 23% were morbidly obese. All of the 3 obese transplant recipients were class-I whereas 4 (25%) of the 16 reconstructed-alone patients were class-II (n = 2) and III (n = 2). A paired scatter plot of BMI at time of BS, referral, and last FU is shown in Figure 6B.
Concerning the 21 patients with compromised nutritional autonomy, 4 have yet to undergo surgery, and 17 underwent surgical intervention; 15 underwent autologous reconstruction and 2 underwent transplant. Of the 15 reconstructed patients, 3 had recent surgery, 5 were awaiting transplant, and 7 were not candidates for transplant because of poor behavioral health and psychosocial support. The two transplant recipients had intermittent TPN therapy for graft dysfunction or underwent allograft enterectomy awaiting retransplantation.
Most current survivors had serum albumin, vitamins, iron, and zinc levels within normal range with partial supplementation in nearly 50%, including the transplant recipients (Table 4). Patients with osteoporosis were under treatment at last FU.
Quality of Life
No health-related quality of life survey was conducted in this study. It is imperative to emphasize that 20% of the study patients experienced major psychiatric disorders at referral with 10% of the patients receiving continuous intravenous narcotics via an implantable pump. Discontinuation of narcotics was achievable in most patients at last FU. It is reasonable to expect that restoration of gut autonomy with discontinuation of TPN and repair of complex gastrointestinal fistulae will improve the different quality of life physical and emotional domains. Full satisfaction was expressed by most patients. Interestingly, one of the reconstructed patients with reversal of BS and one of the transplant recipients requested consideration for a second bariatric operation particularly with the growing favorable publicity of sleeve gastrectomy.