Portal Pressure <15 mm Hg Is a Key for Successful Adult Living Donor Liver Transplantation Utilizing Smaller Grafts than Before

Transcription

1 LIVER TRANSPLANTATION 16: , 2010 ORIGINAL ARTICLE Portal Pressure <15 mm Hg Is a Key for Successful Adult Living Donor Liver Transplantation Utilizing Smaller Grafts than Before Yasuhiro Ogura, Tomohide Hori, Walid M. El Moghazy, Atsushi Yoshizawa, Fumitaka Oike, Akira Mori, Toshimi Kaido, Yasutsugu Takada, and Shinji Uemoto Division of Hepatobiliary-Pancreatic Surgery and Transplantation, Department of Surgery, Graduate School of Medicine, Kyoto University, Kyoto, Japan To prevent small-for-size syndrome in adult-to-adult living donor liver transplantation (A-LDLT), larger grafts (ie, right lobe grafts) have been selected in many transplant centers. However, some centers are investigating the benefits of portal pressure modulation. Five hundred sixty-six A-LDLT procedures using right or left lobe grafts were performed between 1998 and In 2006, we introduced intentional portal pressure control, and we changed the graft selection criteria to include a graft/recipient weight ratio >0.7% instead of the original value of >0.8%. All recipients were divided into period I ( , the era of unintentional portal pressure control; n ¼ 432) and period II ( , the era of intentional portal pressure control; n ¼ 134). The selection of small-for-size grafts increased from 7.8% to 23.9%, and the selection of left lobe grafts increased from 4.9% to 32.1%. Despite the increase in the number of smaller grafts in period II, 1-year patient survival was significantly improved (87.9% versus 76.2%). In 129 recipients in period II, portal pressure was monitored. Patients with a portal pressure <15 mm Hg demonstrated better 2-year survival (n ¼ 86, 93.0%) than patients with a portal pressure 15 mm Hg (n ¼ 43, 66.3%). The recovery from hyperbilirubinemia and coagulopathy after transplantation was significantly better in patients with a portal pressure <15 mm Hg. In conclusion, our strategy for A-LDLT has changed from larger graft based A-LDLT to controlled portal pressure based A-LDLT with smaller grafts. A portal pressure <15 mm Hg seems to be a key for successful A-LDLT. Liver Transpl 16: , VC 2010 AASLD. Received September 29, 2009; accepted February 14, See Editorial on Page 695 Adult-to-adult living donor liver transplantation (A- LDLT) has spread all over the world in the last 10 years. Our institute started pediatric living donor liver transplantation (LDLT) in 1990, and the recipient population gradually shifted from pediatric patients to adult patients. The concept of small-for-size (SFS) partial liver grafts used for LDLT was reported in Liver grafts with a graft weight/standard liver volume ratio less than 40% or a graft/recipient weight ratio (GRWR) less than 0.8% have been considered SFS grafts, and functional impairments such as prolonged cholestasis, ascites, coagulopathy, and encephalopathy secondary to SFS grafts have been described as small-for-size syndrome (SFSS). 2 Because SFS grafts and SFSS have been shown to lead to poorer outcomes and higher morbidity, larger grafts (ie, right lobe grafts) have become the standard graft type in the A-LDLT setting at many transplant Abbreviations: A-LDLT, adult-to-adult living donor liver transplantation; ALT, alanine aminotransferase; BA, biliary atresia; BUN, blood urea nitrogen; CI, confidence interval; Cre, creatinine; CVP, central venous pressure; GRWR, graft/recipient weight ratio; HR, hazard ratio; LDLT, living donor liver transplantation; MELD, Model for End-Stage Liver Disease; MHV, middle hepatic vein; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; PT-INR, prothrombin time international normalized ratio; PV, portal vein; SAL, splenic arterial ligation; SD, standard deviation; SFS, small for size; SFSS, small-for-size syndrome. Address reprint requests to Yasuhiro Ogura, MD, Division of Hepatobiliary-Pancreatic Surgery and Transplantation, Department of Surgery, Graduate School of Medicine, Kyoto University, 54 Kawara-cho, Shogoin, Sakyo-ku, Kyoto City, Kyoto, Japan Telephone: þ ; FAX: þ ; oguchan@kuhp.kyoto-u.ac.jp DOI /lt Published online in Wiley InterScience ( VC 2010 American Association for the Study of Liver Diseases.

2 PORTAL PRESSURE <15 MM HG FOR SUCCESSFUL A-LDLT 719 TABLE 1. Clinical Features of Adult-to-Adult Living Donor Liver Transplantation Variables Period I (n ¼ 432) Period II (n ¼ 134) P Value Recipient factors Gender (male:female) 226:206 70: Age, years (mean 6 SD) < Body weight, kg (mean 6 SD) MELD score (mean 6 SD) Etiology Viral hepatitis 84 (19.4%) 28 (20.9%) Other liver cirrhosis 22 (5.1%) 12 (9.0%) Hepatocellular carcinoma 144 (33.3%) 48 (35.8%) Cholestatic (PBC/PSC/BA) 91 (21.1%) 21 (15.7%) Autoimmune hepatitis 5 (1.2%) 5 (3.7%) Metabolic 10 (2.3%) 3 (2.2%) >0.999 Fulminant hepatic failure 46 (10.6%) 8 (6.0%) Other 30 (6.9%) 9 (6.7%) Donor factors Donor gender (male:female) 237:195 75: Donor age, years (mean 6 SD) Graft factors Graft type (right:left) 411:21 91:43 < Graft weight, g (mean 6 SD) < GRWR, % (mean 6 SD) < ABO blood type compatibility (identical:compatible:incompatible) 296:74:62 79:25: centers. 3 There are 2 options for selecting right lobe grafts: right lobe grafts without the middle hepatic vein (MHV) and right lobe grafts with the MHV. However, harvesting a larger graft from a living donor is sometimes a high-risk operation, and there have been some reports of living donor deaths. 4 Recently, we have thought that it is more desirable to save A-LDLT recipients with smaller liver grafts to reduce the risk to living donors. The concept of graft selection in A-LDLT has changed in our institute; in other words, smaller grafts can be selected to save recipients under certain conditions. Several animal models have demonstrated that better liver regeneration can be observed with portal vein pressure control in large hepatectomy models, and this leads to better outcomes than those of non portal pressure control groups. 5,6 Although the importance of portal pressure during A-LDLT has been recognized in the past several years, 7-13 this is the first large series study of the impact of the graft size and survival after clinical A- LDLT with intentional portal pressure control. PATIENTS AND METHODS Patients and Graft Type Kyoto University Hospital started pediatric LDLT in 1990, and the number of adult cases (age > 18 years) has increased since The cumulative number of LDLT procedures performed at our institute has exceeded 1350 so far. Five hundred sixty-six standard A-LDLT procedures using right or left lobe grafts were performed between 1998 and 2008 (Table 1). A-LDLT using posterior segment grafts (n ¼ 11), auxiliary partial orthotopic liver transplantation (n ¼ 8), dual liver transplantation (n ¼ 1), and domino liver transplantation (n ¼ 10) were excluded from this study. Donor and recipient surgical procedures have been described in detail elsewhere Briefly, according to computed tomography scan volumetric analysis, a right lobe or left lobe graft was selected. In the case of a right lobe graft, drainage veins from anterior segments, such as V5, V8, or the MHV, were preserved in the graft for reconstruction. In the case of a left lobe graft, the MHV was always included in the graft. The caudate lobe was generally included in the left lobe graft, unless the graft size was sufficient without the caudate lobe. The graft was perfused with cold histidine tryptophan ketoglutarate solution (Custodiol, Essential Pharmaceuticals, LLC, Newtown, PA). Immediately after perfusion of the preservation solution, all resected liver grafts were measured. This actual graft weight was used for the calculation of GRWR. Recipient total hepatectomy was performed without venovenous bypass with preservation of the inferior vena cava. To maximize the hepatic venous orifice, venoplasty was generally added at the back table. After venoplasty, end-to-end anastomosis of the hepatic veins and reconstruction of the portal vein between the graft portal vein and the recipient portal vein trunk were performed. The graft hepatic artery was anastomosed to the recipient artery with surgical loupes. Immediately after all vessel reconstructions, intraoperative Doppler ultrasound was performed to confirm good triphasic hepatic venous outflow, sufficient portal flow, and good intrahepatic arterial flow. Any vascular complications leading to high portal pressure (eg, hepatic venous outflow block and portal

3 720 OGURA ET AL. Figure 1. Annual number of standard A-LDLT procedures using right or left lobe grafts. After 1998 (the year in which right lobe grafts were introduced), the main graft type in A- LDLT was the right lobe graft. Since 2006 (the year in which intentional portal pressure was introduced), left lobe graft selection for A-LDLT has increased. This change in the graft selection concept has developed into less dependence on right lobe grafts. vein anastomotic strictures) were excluded by this Doppler ultrasound examination. As right lobe grafts used to be our standard graft selection for A-LDLT between 1998 and 2006 (period I), the majority of the grafts selected for A-LDLT were right lobe grafts (n ¼ 411/432, 95.1%), and the number of left lobe grafts in our institute was small (n ¼ 21/432, 4.9%). However, we changed the graft selection concept in 2006, so left lobe graft selection increased in the last 3 years (period II); in other words, only 4.9% of the grafts were left lobe grafts before the conceptual change, but this increased to 32.1% (43 of 134) after the conceptual change. Consequently, the dependence on right lobe grafts dramatically decreased after this change in the institutional graft selection criteria (Fig. 1). GRWR varied from 0.49% to 2.14% (1.11% %) in this series, and the Model for End-Stage Liver Disease (MELD) score ranged from (period I) to (period II; P ¼ 0.012). Portal Pressure Control A large number of patients have been treated with intentional portal pressure control since 2006, and 134 cases were analyzed. The portal pressure was monitored via the small jejunal vein branch with an 18-G central venous catheter during the recipient operation. The optimal portal pressure was set below 20 mm Hg after all vessel anastomoses of the liver graft were completed. If it was more than 20 mm Hg after the reflow of the liver graft, intentional portal pressure control was performed. Intentional portal pressure control was done mainly by concurrent splenectomy and additionally by the creation of a portosystemic shunt if it was indicated. LigaSure Atlas (Valleylab, Boulder, CO), which is more frequently used during laparoscopic splenectomy, was employed for dissection around the spleen to avoid excess bleeding. We used to perform splenic arterial ligation (SAL), 8 but we did not adopt SAL recently. Large spontaneous portosystemic shunts, such as splenorenal shunts and gastric/esophageal varices (collateral vessels), should be ligated, although this ligation may increase portal venous pressure. As long as the portal pressure was kept less than the target portal pressure (20 mm Hg), those collaterals were ligated for the improvement of portal flow and the prevention of the portal venous steal phenomenon when the graft resistance was increased postoperatively. However, those collateral vessels could not be ligated or could be partially ligated when the portal pressure was shown to be more than 20 mm Hg by collateral test clamping. In a group of patients with hepatitis C, our previous data and the data of another group 17 demonstrated that the platelet count recovery was limited when severe thrombocytopenia existed. Because these patients would require ribavirin/interferon combination therapy for the posttransplant recurrence of hepatitis C in the future, concurrent splenectomy was carried out to treat thrombocytopenia despite any portal pressure level. Clinical Laboratory Data The serum total bilirubin level and prothrombin time international normalized ratio (PT-INR), which are the parameters used for the determination of SFSS, were collected from patients medical charts during period II preoperatively, on postoperative days 1, 2, 3, 4, 5, 6, 7, at 1 and 3 months, and at 1 year. For the risk factor analysis, 22 items, including recipient factors, graft factors, ABO blood type compatibility, preoperative laboratory data, and operative factors, were also collected and analyzed (Table 2). Ascites Data Daily ascites amounts were collected for all recipients until 2 weeks after A-LDLT during period II. The total amounts of the 2 weeks were divided by 14 days to obtain the average daily ascites output. SFSS The Clavien definition of SFSS was used for the analysis; that is, dysfunction [the presence of 2 of the following on 3 consecutive days: bilirubin >100 lmol/l (5.84 mg/dl), international normalized ratio >2, and encephalopathy grade 3 or 4] of a small partial liver graft (GRWR < 0.8%) during the first postoperative week after the exclusion of other causes was considered SFSS. 18 Statistical Analysis All values are expressed as means and standard deviations. Categorical variables were compared with the

4 PORTAL PRESSURE <15 MM HG FOR SUCCESSFUL A-LDLT 721 TABLE 2. Univariate and Multivariate Proportional Hazards Model for Patient Survival After Adult-to-Adult Living Donor Liver Transplantation Hazard Univariate Analysis 95% Confidence Hazard Multivariate Analysis 95% Confidence Variables Ratio Interval P Value Ratio Interval P Value Recipient factors Age Gender Female Male Child-Pugh classification score MELD score Graft factors Graft type Right lobe Left lobe * GRWR < 0.8% % GRWR < 1% GRWR 1% ABO blood type compatibility ABO-identical ABO-compatible * ABO-incompatible Preoperative laboratory data Total bilirubin ALT PT-INR Cre BUN Operative factors Cold ischemic time * Warm ischemic time Initial portal pressure Final portal pressure <15 mm Hg mm Hg * * Splenectomy No Yes Portal pressure modulation No Yes *P < chi-square test or Fisher s exact test, whereas continuous variables were compared with the unpaired t test. For the patient survival analysis, the Kaplan- Meier method with the log-rank test was used. A P value <0.05 was considered statistically significant. The contribution of risk factors to patient survival was analyzed with Cox proportional hazards regression. Data were analyzed with Stat View for Windows version 5.0 and SPSS for Windows version 16 (SAS Institute, Inc.). RESULTS Clinical Features of A-LDLT Characteristics of the recipients and donors are listed in Table 1. There were no significant differences between period I and period II with respect to the recipient gender, recipient body weight, etiology, donor gender, donor age, and ABO blood type incompatibility. The average age was significantly older (P < ), and the mean MELD score was significantly higher (P ¼ 0.012) in period II. Left lobe grafts were selected more (P < ) in period II. The mean GRWR of period I was 1.15%, whereas that of period II was 0.96% (P < ). Graft Size and Type During period I (prior to intentional portal pressure control), only 7.8% of the A-LDLT procedures (34 of 432) used liver grafts with a GRWR < 0.8 (ie, SFS grafts). In contrast, during period II (the era of intentional portal pressure control), the selection of SFS

5 722 OGURA ET AL. Figure 2. Plot of each GRWR value used for A-LDLT from 1998 to During period I, only 7.8% of the grafts were SFS grafts (GRWR < 0.8%). However, 23.9% of A-LDLT grafts had a GRWR < 0.8% during period II (the era of intentional portal pressure control). Open squares indicate left lobe grafts, and open circles indicate right lobe grafts. The left lobe graft population had a tendency to be smaller than the right lobe graft population. GRWR was smaller in period II (0.92% %, n 5 134) than period I (1.15% %, n 5 432; P < ). grafts increased up to 23.9% (32 of 134 A-LDLT procedures; Fig. 2). This shift to smaller grafts was due to more selection of left lobe grafts (Fig. 1), which demonstrated smaller GRWRs (open squares, Fig. 2) than right lobe graft (open circles, Fig. 2). The GRWRs during periods I and II were calculated to be 1.15% % and 0.92% %, respectively, and this was statistically significant (P < ). Overall Survival Even though we selected smaller grafts in period II, the LDLT outcomes were not jeopardized. The overall 1-, 3-, and 5-year survival rates after A-LDLT in period I were 76.2%, 71.1%, and 68.8%, respectively, and the 1- and 2-year survival rates in period II were 87.9% and 81.6%, respectively; this was a statistically improved outcome (P ¼ 0.01; Fig. 3A). However, when survival in each period was analyzed by graft size (GRWR < 0.8 or GRWR 0.8), the survival curve did not demonstrate a separate line clearly, and this indicates that GRWR may not play a crucial role in recent A-LDLT outcomes (P ¼ 0.22 in period I and P ¼ 0.97 in period II; Fig. 3B). Portal Pressure and Outcomes The portal pressure was monitored during the recipient operation. Between 2006 and 2008, the portal pressure was monitored in a total of 129 cases and retrospectively analyzed. After the reflow of liver grafts, intentional portal pressure control (n ¼ 85/129, 65.9%) was done mainly by concurrent splenectomy to lower the portal pressure to <20 mm Hg. Splenectomy was performed 84 times, and SAL was performed once. Because 8 cases underwent splenectomy prior to graft reperfusion on account of severe thrombocytopenia of hepatitis C, the portal pressure both before and after intentional portal pressure control was recorded for 77 of 85 portal pressure modulations. The portal pressure prior to intentional portal pressure control was mm Hg, and the portal pressure after intentional control was mm Hg; this indicated that intentional portal pressure control could effectively reduce the portal pressure by 4.4 mm Hg. Four cases required a portosystemic shunt (inferior mesenteric vein to left renal vein) in addition to splenectomy. The requirement rate for splenectomy was 67.7% (21 of 31) for a GRWR < 0.8% and 64.3% (63 of 98) for a GRWR 0.8, and this was not significantly different (P ¼ 0.72). Furthermore, the requirement for splenectomy was not statistically different according to the graft type: 67.7% (29 of 43) in left lobe grafts and 64.0% (55 of 86) in right lobe grafts (P ¼ 0.70). These results suggest that SFS grafts as well as larger grafts require intentional portal pressure control to achieve a final portal pressure <20 mm Hg despite the graft type. Three cases could not achieve a portal pressure <20 mm Hg, and the remaining 126 cases did achieve a portal pressure <20 mm Hg (Fig. 4). In general, larger grafts are considered to have better compliance, which helps to decrease portal pressure after reflow. This is one of the reasons that larger liver grafts (right lobe grafts) have been selected for A-LDLT for many years. However, when we analyzed the correlation of the GRWR and final portal pressure, we found that they were independent of each other (r ¼ 0.007; Fig. 4). In Fig. 4, open circles indicate living recipients, and crosses indicate dead recipients. When these outcomes were analyzed with the GRWR (<0.8% or 0.8%) and final portal pressure (<15 mm Hg or 15 mm Hg), the GRWR did not show a statistical difference with respect to the proportion of dead patients (10.0% versus 13.2%; P ¼ 0.636). In contrast, the final portal pressure showed a significant difference with respect to the recipient outcome (P ¼ 0.002). There was a higher proportion of dead patients with a final portal pressure 15 mm Hg (12 deaths, 28.0%), and the proportion of dead patients with a final portal pressure <15 mm Hg (4 deaths, 4.7%) was small. These results indicate that the final portal pressure could be controlled despite any GRWR by splenectomy, and patient outcome was associated with the final portal pressure, not with the GRWR. As our initial portal pressure control strategy was set at 20 mm Hg, we first analyzed patient outcomes at 20 mm Hg. The patient survival rate of the wellcontrolled portal pressure group (<20 mm Hg, n ¼ 126) was 88.4% and 84.8% at 1 and 2 years, respectively, and that of the poorly controlled portal pressure group (20 mm Hg, n ¼ 3) was 66.7% at 1 and 2 years. Among 3 recipients with poorly

6 PORTAL PRESSURE <15 MM HG FOR SUCCESSFUL A-LDLT 723 Figure 3. (A) Comparison of patient survival after A-LDLT in periods I and II. The outcomes of period II were statistically improved versus those of period I (P ). (B) Comparison of patient survival after its division by a GRWR of 0.8% in periods I and II. Patient survival after A-LDLT with a GRWR < 0.8% (dashed line) and patient survival after A-LDLT with a GRWR 0.8% (solid line) in each period were not statistically different (P in period I and P in period II); this suggests that GRWR may not have played a crucial role in recent A-LDLT outcomes. controlled portal pressure (20 mm Hg), 2 survivors had to overcome SFSS and infection, and 1 expired because of a severe infection. Because of the sample size mismatching, this analysis was statistically inappropriate. For that reason, although we set <20 mm Hg as the target portal pressure during period II, we analyzed those 129 cases retrospectively with a final portal pressure of 15 mm Hg. There were 86 cases with a portal pressure < 15 mm Hg and 43 cases with a portal pressure 15 mm Hg. The MELD score for patients with a portal pressure <15 mm Hg was , and the MELD score for patients with a portal pressure 15 mm Hg was ; these scores were not statistically different (P ¼ 0.894). The 1- and 3- year survival rates were 95.2% and 93.0% when the portal pressure was well controlled (<15 mm Hg); in contrast, the rates were 73.0% and 66.3% when the portal pressure was 15 mm Hg (P ¼ ; Fig. 5). Portocaval Pressure Gradient and Outcomes The central venous pressure (CVP), measured at the same time as the final portal venous pressure, was mm Hg (for the group with a portal pressure <15 mm Hg) and mm Hg (for the group with a portal pressure 15 mm Hg; P ¼ 0.04), and the portocaval pressure gradient (final portal pressure CVP) was mm Hg (for the group with a portal pressure <15 mm Hg) and mm Hg (for the group with a portal pressure 15 mm Hg; P < ). This suggested that both the CVP and portocaval pressure gradient were higher in the group with the higher final portal pressure. However, when we divided and analyzed our study group by a low portocaval gradient (<9 mm Hg, n ¼ 86) and a high portocaval gradient (9 mm Hg, n ¼ 43), the 1- and 3-year survival rates were 90.4% and 88.4% with the low portocaval gradient and 82.9% and 74.6% with the high portocaval gradient; this was not statistically significant (P ¼ 0.163; Fig. 6). Bilirubin and Prothrombin Time Graft functional impairments, such as prolonged cholestasis and coagulopathy, were compared between patients with final portal pressures <15 mm Hg and 15 mm Hg in period II. There was no significant difference in the preoperative total bilirubin level between the 2 groups (9.8 6

7 724 OGURA ET AL. Figure 4. Between 2006 and 2008, the portal pressure was monitored in 129 A-LDLT patients. The final portal pressure and GRWR were plotted. A total of 126 cases (97.7%) achieved a portal pressure <20 mm Hg. The final portal pressure was well controlled despite the various GRWR ranges, and this indicates that the portal pressure could be controlled by not only the GRWR but also other modalities. Open circles and crosses indicate living and dead patients, respectively. There was a significantly higher proportion of deaths observed for patients with a final portal pressure 15 mm Hg versus the low portal pressure population (P ). Figure 5. Patient survival curve in period II based on a final portal pressure of 15 mm Hg. Patients with a portal pressure <15 mm Hg had statistically better outcomes than those with a portal pressure 15 mm Hg (1 year, 95.2% versus 73.0%; 2 years, 93.0% versus 66.3%; P ). 1.2 versus mg/dl; P ¼ 0.852). However, from 2 days through 3 months after transplantation, the total bilirubin levels of patients with a portal pressure 15 mm Hg were significantly higher than those of patients with a portal pressure <15 mm Hg (P < 0.05; Fig. 7A). PT-INR was also compared, and the preoperative data showed no significant difference ( versus ; P ¼ 0.131). Similarly to the bilirubin data, coagulopathy was confirmed from day 1 through 1 month after transplantation in patients with a portal pressure 15 mm Hg (P < 0.05; Fig. 7B). Ascites Reduction by Portal Pressure Control Another symptom of SFSS is prolonged and large-volume ascites after liver transplantation. The average daily amount of ascites in the first 2 weeks after A- Figure 6. Patient survival curve in period II based on a portocaval gradient of 9 mm Hg. High and low portocaval gradients did not show a significant difference for patient survival (1 year, 90.4% versus 82.9%; 2 years, 88.4% versus 74.6%; P ). LDLT in period II was calculated to be ml (portal pressure <15 mm Hg) and ml (portal pressure 15 mm Hg); this demonstrated statistically significant ascites reduction by portal pressure control (P ¼ 0.01; Fig. 8). Morbidity and Mortality The mortality rate in period II was 12.4% (n ¼ 16/ 129; Table 3). Four expired recipients showed a final portal pressure <15 mm Hg, and another 12 expired recipients showed a portal pressure 15 mm Hg. Splenectomy was performed in 9 of these 16 expired patients. The causes of death were infection/sepsis (n ¼ 6), graft failure due to chronic or humoral rejection (n ¼ 6), cerebral bleeding (n ¼ 1), pulmonary hemorrhage (n ¼ 1), thrombotic microangiopathy (n ¼ 1), and primary lung cancer (n ¼ 1). The correlation of infection/sepsis and splenectomy was not clear. SFSS is known to lead to infection/sepsis, but the 6 deaths due to infection/sepsis were related to surgical complications (n ¼ 2, days 53 and 62), sepsis after cerebral bleeding (n ¼ 1, day 40), Aspergillus infection (n ¼ 2, days 48 and 328), and Pneumocystis carinii (n ¼ 1, day 242). It was difficult to prove if these infectious complications were associated with SFSS or general infection in immunocompromised patients. SFSS was diagnosed in 12.9% (4 of 31 SFS grafts with a GRWR <0.8%) by the Clavien definition. Two of the 4 patients expired after SFSS complications (postoperative days 48 and 76). The complication of early portal vein thrombus was experienced in 2 cases (1.7%, postoperative days 2 and 5), and both had undergone splenectomy during A-LDLT. The onset of portal vein thrombus was not identified, but both cases were successfully treated by surgery without any liver damage. Risk Factors for Recipient Death After A-LDLT Twenty-two items, including recipient factors, graft factors, ABO blood type compatibility, preoperative laboratory data, and operative factors were analyzed with univariate analysis and then multivariate analysis (Table 2). In univariate analysis, the graft type [left

8 PORTAL PRESSURE <15 MM HG FOR SUCCESSFUL A-LDLT 725 Figure 8. Less ascites production was achieved by intentional portal pressure control. The average daily amount of ascites in the first 2 weeks after A-LDLT in period II was calculated, and it is represented graphically with a box plot. A portal pressure <15 mm Hg yielded an average daily ascites volume of ml, and a portal pressure 15 mm Hg yielded a daily ascites volume of ml; this demonstrated statistically significant ascites reduction by intentional portal pressure control (P ). Figure 7. Serial changes in (A) the total bilirubin levels and (B) the PT-INR levels in A-LDLT recipients. Hyperbilirubinemia and coagulopathy were observed postoperatively in the group with a portal pressure 15 mm Hg. The group with a portal pressure <15 mm Hg showed significantly better improvement of the total bilirubin and PT-INR levels (*P < 0.05). lobe; hazard ratio (HR) ¼ 2.662; 95% confidence interval (CI) ¼ ; P ¼ 0.033], cold ischemic time (HR ¼ 1.003; 95% CI ¼ ; P ¼ 0.037), and final portal pressure 15 mm Hg (HR ¼ 5.931; 95% CI ¼ ; P ¼ 0.001) were significantly associated with worse patient survival, and an ABO-compatible combination demonstrated a decreased risk in comparison with other blood type combinations (HR ¼ 0.324; 95% CI ¼ ; P ¼ 0.025). However, multivariate analysis revealed that only a final portal pressure 15 mm Hg (HR ¼ 4.25; 95% CI ¼ ; P ¼ 0.008) significantly affected patient survival after A-LDLT. Moreover, univariate analysis clearly showed that graft size and splenectomy were not risk factors for patient survival after A-LDLT. DISCUSSION As SFS grafts and SFSS are widely accepted, many transplant centers have selected larger grafts, such as right lobe grafts with or without the MHV, instead of smaller grafts, such as left lobe grafts. 3,19 Even when a larger right lobe is selected, the GRWR can sometimes be smaller than the ideal value if the recipient body weight is too high. To overcome SFS issues, Boillot et al. 10 first reported portosystemic shunts to prevent graft congestion and failure of liver grafts by overperfusion. Our group and another group also have tried permanent portosystemic shunts in several cases with favorable outcomes However, we do not use this procedure now because of (1) the technical difficulty, (2) the uncontrollable flow volume between the liver direction and inferior vena cava direction, and (3) the potential risk of the portal flow steal phenomenon (away from the liver grafts). Ito et al. 8 analyzed the relationship of the portal venous pressure and the outcome after LDLT. In their study, an elevated portal pressure in the early phase was strongly associated with poor patient survival that was attributable, at least in part, to SFS grafts. After this observation, they tried to modify the portal pressure and determine the effect on recipient survival. Splenic artery ligation (SAL) was adopted to control the portal pressure, and better outcomes were observed with this portal pressure modulation. Those data showed that the portal pressure modification was key for improving LDLT results. However, as the effect of SAL on portal pressure was not as strong as the effect of splenectomy, 9 we have not recently used SAL at our institute. Other reports have mentioned better regeneration and less graft damage with a lower portal pressure after A-LDLT. 12 An early postoperative portal pressure elevation greater than 20 mm Hg has been associated with rapid graft hypertrophy, higher peripheral blood hepatocyte growth factor levels, lower portal vascular endothelial growth factor levels, and SFSS (poor outcomes, graft dysfunction with hyperbilirubinemia, coagulopathy, and severe ascites). They concluded

9 726 OGURA ET AL. TABLE 3. Summary of Sixteen Recipient Deaths in Period II Age Graft GRWR Final Portal Vein Pressure Patient Loss Case (years) Gender Type (%) Splenectomy (mm Hg) SFSS (days) Cause of Death 1 57 Male Left lobe 0.75 No 8 No 195 Primary lung cancer 2 42 Male Right lobe 0.85 No 12 No 62 Infection/sepsis (surgical complications) 3 59 Male Right lobe 0.90 Yes 13 No 599 Graft failure due to chronic rejection 4 25 Female Right lobe 1.26 No 14 No 53 Graft failure due to severe rejection 5 28 Female Left lobe 0.80 No 15 No 195 Graft failure due to chronic rejection 6 53 Male Right lobe 1.04 Yes 15 No 100 Infection/sepsis 7 47 Female Left lobe 1.11 Yes 16 No 40 Infection/sepsis (after cerebral bleeding) 8 44 Female Left lobe 0.77 No 17 Yes 76 Thrombotic microangiopathy 9 56 Male Right lobe 0.88 No 17 No 328 Infection/sepsis (Aspergillus infection) Female Left lobe 0.70 Yes 17 No 242 Infection/sepsis (Pneumocystis carinii) Male Right lobe 1.41 No 17 No 11 Cerebral bleeding Male Right lobe 1.19 Yes 17 No 53 Infection/sepsis (surgical complications) Female Left lobe 1.13 Yes 18 No 703 Graft failure due to chronic rejection Female Left lobe 0.96 Yes 18 No 9 Graft failure due to humoral rejection Female Left lobe 1.06 Yes 18 No 5 Pulmonary hemorrhage Female Left lobe 0.74 Yes 26 Yes 48 Infection/sepsis (Aspergillus infection) that adequate liver regeneration requires well-controlled portal venous pressure and flow, which are reflected by the clearance of hepatocyte growth factor and elevated vascular endothelial growth factor levels. This is the first report of a large series of SFS grafts and intentional portal vein pressure control by splenectomy. SFS grafts, such as partial liver grafts in A- LDLT, have been considered to induce SFSS with various degrees of severity. Four cases in period II were categorized as SFSS by the Clavien SFSS definition. 18 However, we have shown that SFS grafts can be used with a favorable result if the portal pressure is controlled adequately (<15 mm Hg); this prevents individual symptoms of SFSS such as hyperbilirubinemia, coagulopathy, and posttransplant ascites output (Figs. 7 and 8). The current study implies that adequate portal pressure after reflow of liver grafts is <15 mm Hg. It is true that we cannot manipulate the portal pressure up and down. There are only 2 options: no modulation or modulation by splenectomy. However, we have shown that a portal pressure <15 mm Hg leads to a better outcome. Moreover, so far we have never experienced hypoperfusion of portal flow due to portal pressure that is too low. Therefore, we believe that this 2-step portal pressure control method is sufficient to improve A-LDLT outcomes. In this study, we also tried to analyze the impact of patient survival after A-LDLT. In univariate analysis, recipient factors (age, gender, Child-Pugh classification score, and MELD score) and preoperative laboratory data (total bilirubin, alanine aminotransferase, PT-INR, creatinine, and blood urea nitrogen) did not affect the A-LDLT outcomes. The graft type (right versus left lobe), ABO blood type compatibility, and cold ischemic time were significantly associated with worse patient survival, but multivariate analysis did not reveal that those factors significantly affected patient survival. According to multivariate analysis, only a final portal pressure 15 mm Hg was highlighted as an independent predictor for A-LDLT outcomes. As our new portal pressure control was started in 2006, it is true that we cannot confirm the safety of splenectomy in the long term even though its shortterm effects for adult LDLT outcomes are excellent. Two major concerns after splenectomy are severe infections and vascular complications. Sulfamethoxazole/trimethoprim was given as infection prophylaxis, but the pneumococcal polysaccharide vaccine has not been given to splenectomized A-LDLT recipients so far. Even though we have not yet encountered severe infections related to splenectomy, we have to consider a vaccination plan for those recipients. Warfarin is

10 PORTAL PRESSURE <15 MM HG FOR SUCCESSFUL A-LDLT 727 Figure 9. Current strategy for intentional portal pressure control at Kyoto University Hospital. The target portal pressure is set below 15 mm Hg after reflow. Splenectomy is the first procedure to reduce portal pressure. If the portal pressure is >15 mm Hg after splenectomy, additional shunt creation (eg, inferior mesenteric vein to left renal vein shunting) should be considered. Large collaterals should be ligated to prevent portal flow steal in A-LDLT, although it may increase the portal venous pressure. generally administered to prevent portal vein thrombosis to A-LDLT recipients who have undergone splenectomy. Lusebrink et al. 23 reported that there was a marked difference in the frequency of infectious episodes (61.7% versus 25.3%) that resulted in a decreased survival rate (77.5% versus 95.4%) for splenectomized deceased donor liver transplant recipients. Although we need close long-term follow-up for A-LDLT recipients who have undergone splenectomy, our univariate analysis for the short term has demonstrated that splenectomy does not have a negative impact on patient survival. Our current portal pressure control flowchart for A- LDLT is shown in Fig. 9. Splenectomy is the first procedure for reducing portal pressure. If it is not sufficient, a portosystemic shunt can be created. 24 Large spontaneous portosystemic shunts, such as splenorenal shunts and gastric/esophageal varices (collateral vessels), should be ligated, as long as the portal pressure is kept less than 15 mm Hg, for the improvement of portal flow and for the prevention of the portal venous steal phenomenon at the time of postoperative severe rejection However, those collateral vessels cannot be ligated or can be partially ligated when the portal pressure is shown to be more than 15 mm Hg by collateral test clamping. Of course, the portal venous pressure during A- LDLT is affected by many factors, such as the CVP, hepatic venous outflow and hepatic vein reconstruction procedures, graft type, quality, and compliance, congestion volume, portal inflow volume, and portosystemic shunts. These factors determine both the portal venous pressure and flow simultaneously. Yagi et al. 28 reported that optimal portal venous circulation (portal venous pressure, portal venous flow, and graft compliance) is required for better outcomes. In this study, we focused on portal pressure because an appropriate portal venous pressure is known to trigger liver regeneration and excessive shear stress induces liver injury. 12,29 Moreover, this simple monitoring of the portal venous pressure is more practical than complicated multiple measurements in clinical liver transplantation settings, and our data clearly demonstrate that a final portal pressure <15 mm Hg correlates with better outcomes in A-LDLT. The portocaval pressure gradient is influenced by many operative factors similarly to the final portal venous pressure, but it cannot predict the LDLT outcome (Fig. 6). By intentional portal pressure control, the final portal pressure can be controlled well for any graft size (Fig. 4). Our minimally acceptable GRWR at the time of living donor evaluation was 0.8% before However, we currently require a GRWR greater than 0.7% because SFSS caused by small grafts can be avoided or controlled by portal pressure control. Therefore, our selection of left lobes as grafts has been increasing for the past 3 years. The advantage of left lobe selection is donor safety, which is one of the most important issues in A-LDLT. We are still investigating how far we can go. The smallest graft used for successful A-LDLT had a GRWR of 0.49%. To extend the small limits of GRWR, not only technical refinements and portal pressure modulation but also other issues, such as preservation and ischemic/reperfusion injury, should be clarified. We think that this concept of intentional portal pressure control can be applied not only in the A- LDLT setting but also in the deceased donor liver transplant setting and split liver transplant setting. 30 If a liver graft from a deceased donor is or is estimated to be too small for the patient, a liver graft of a marginal size may be used for the recipient with a favorable outcome by the use of additional portal pressure control procedures. We believe that this procedure can be a strong tool for size mismatching in all types of liver transplantation. In conclusion, our strategy for A-LDLT has changed from larger graft based A-LDLT to controlled portal pressure based A-LDLT using smaller grafts. A portal pressure < 15 mm Hg seems to be a key for successful A-LDLT. REFERENCES 1. Kiuchi T, Kasahara M, Uryuhara K, Inomata Y, Uemoto S, Asonuma K, et al. Impact of graft size mismatching on graft prognosis in liver transplantation from living donors. Transplantation 1999;67: Kiuchi T, Tanaka K, Ito T, Oike F, Ogura Y, Fujimoto Y, et al. Small-for-size graft in living donor liver transplantation: how far should we go? Liver Transpl 2003;9:S29-S35.

11 728 OGURA ET AL. 3. Tanaka K, Ogura Y. Small-for-size graft and small-forsize syndrome in living donor liver transplantation. Yonsei Med J 2004;45: Trotter JF, Adam R, Lo CM, Kenison J. Documented deaths of hepatic lobe donors for living donor liver transplantation. Liver Transpl 2006;12: Francavilla A, Zeng Q, Polimeno L, Carr BI, Sun D, Porter KA, et al. Small-for-size liver transplanted into larger recipient: a model of hepatic regeneration. Hepatology 1994;19: Iida T, Yagi S, Taniguchi K, Hori T, Uemoto S. Improvement of morphological changes after 70% hepatectomy with portocaval shunt: preclinical study in porcine model. J Surg Res 2007;143: Lo CM, Liu CL, Fan ST. Portal hyperperfusion injury as the cause of primary nonfunction in a small-for-size liver graft-successful treatment with splenic artery ligation. Liver Transpl 2003;9: Ito T, Kiuchi T, Yamamoto H, Oike F, Ogura Y, Fujimoto Y, et al. Changes in portal venous pressure in the early phase after living donor liver transplantation: pathogenesis and clinical implications. Transplantation 2003;75: Sato Y, Yamamoto S, Oya H, Nakatsuka H, Tsukahara A, Kobayashi T, et al. Splenectomy for reduction of excessive portal hypertension after adult living-related donor liver transplantation. Hepatogastroenterology 2002;49: Boillot O, Delafosse B, Mechet I, Boucaud C, Pouyet M. Small-for-size partial liver graft in an adult recipient; a new transplant technique. Lancet 2002;359: Shimada M, Ijichi H, Yonemura Y, Harada N, Shiotani S, Ninomiya M, et al. The impact of splenectomy or splenic artery ligation on the outcome of a living donor adult liver transplantation using a left lobe graft. Hepatogastroenterology 2004;51: Yagi S, Iida T, Taniguchi K, Hori T, Hamada T, Fujii K, et al. Impact of portal venous pressure on regeneration and graft damage after living-donor liver transplantation. Liver Transpl 2005;11: Ikegami T, Shimada M, Imura S, Arakawa Y, Nii A, Morine Y, et al. Current concept of small-for-size grafts in living donor liver transplantation. Surg Today 2008;38: Tanaka K, Inomata, Y, Kaihara, S. Living-Donor Liver Transplantation. Barcelona, Spain: Prous Science; 2003: Tanaka K, Inomata, Y, Kaihara, S. Living-Donor Liver Transplantation. Barcelona, Spain: Prous Science; 2003: Yamaoka Y, Washida M, Honda K, Tanaka K, Mori K, Shimahara Y, et al. Liver transplantation using a right lobe graft from a living related donor. Transplantation 1994;57: Kishi Y, Sugawara Y, Akamatsu N, Kaneko J, Tamura S, Kokudo N, et al. Splenectomy and preemptive interferon therapy for hepatitis C patients after livingdonor liver transplantation. Clin Transplant 2005;19: Dahm F, Georgiev P, Clavien PA. Small-for-size syndrome after partial liver transplantation: definition, mechanisms of disease and clinical implications. Am J Transplant 2005;5: Liu CL, Fan ST. Adult-to-adult live-donor liver transplantation: the current status. J Hepatobiliary Pancreat Surg 2006;13: Masetti M, Siniscalchi A, De Pietri L, Braglia V, Benedetto F, Di Cautero N, et al. Living donor liver transplantation with left liver graft. Am J Transplant 2004;4: Takada Y, Ueda M, Ishikawa Y, Fujimoto Y, Miyauchi H, Ogura Y, et al. End-to-side portocaval shunting for a small-for-size graft in living donor liver transplantation. Liver Transpl 2004;10: Yamada T, Tanaka K, Uryuhara K, Ito K, Takada Y, Uemoto S. Selective hemi-portocaval shunt based on portal vein pressure for small-for-size graft in adult living donor liver transplantation. Am J Transplant 2008;8: Lusebrink R, Blumhardt G, Lohmann R, Bachmann S, Knoop M, Lemmens HP, et al. Does concomitant splenectomy raise the mortality of liver transplant recipients? Transpl Int 1994;7(suppl 1):S634-S Sato Y, Yamamoto S, Takeishi T, Kato T, Nakatsuka H, Kobayashi T, et al. Inferior mesenteric venous left renal vein shunting for decompression of excessive portal hypertension in adult living related liver transplantation. Transplant Proc 2004;36: Shirouzu Y, Ohya Y, Tsukamoto Y, Yamamoto H, Lee KJ, Okajima H, et al. How to handle a huge portosystemic shunt in adult living donor liver transplantation with a small-for-size graft: report of a case. Surg Today 2009; 39: Sadamori H, Yagi T, Matsukawa H, Matsuda H, Shinoura S, Umeda Y, et al. The outcome of living donor liver transplantation with prior spontaneous large portasystemic shunts. Transpl Int 2008;21: Lee SG, Moon DB, Ahn CS, Kim KH, Hwang S, Park KM, et al. Ligation of left renal vein for large spontaneous splenorenal shunt to prevent portal flow steal in adult living donor liver transplantation. Transpl Int 2007;20: Yagi S, Iida T, Hori T, Taniguchi K, Yamamoto C, Yamagiwa K, et al. Optimal portal venous circulation for liver graft function after living-donor liver transplantation. Transplantation 2006;81: Sato Y, Ichida T, Yamamoto S, Hirano K, Kobayashi T, Oya H, et al. Shear stress theory and small-for-size graft in adult living related liver transplantation. Transplant Proc 2003;35: Heaton N. Small-for-size liver syndrome after auxiliary and split liver transplantation: donor selection. Liver Transpl 2003;9:S26-S28.