Results After Propensity Score Matching
Characteristics after propensity score matching analysis are shown in Table 1. Twenty-one of the 24 patients who received adjuvant immunotherapy were matched with 21 patients who did not receive therapy after covariate adjustment. Therefore, three patients who received immunotherapy and 69 patients who did not receive the therapy were excluded because their propensity scores could not be matched. The study group of 42 patients was well matched; in particular, all covariates that significantly affected overall survival in the entire study group were equally distributed over the two matched groups. Matched patients who received adjuvant immunotherapy had a similar age (56 vs. 57; P=0.62), donor age (36 vs. 36; P=0.51), graft-to-recipient weight ratio (0.89 vs. 0.85; P=0.72), score of MELD (13 vs. 11; P=0.73), blood loss (3270 vs. 3420 mL; P=0.53), and rate of postoperative bile leakage (14.3% vs. 19.0%; P=0.67) to those of matched patients who did not receive adjuvant immunotherapy. Similarly, other clinical variables were similar for patients who did and did not receive the therapy. Of the 21 patients who did not receive immunotherapy, six (28.6%) developed bacteremia. Of the 21 patients who received immunotherapy, one (4.8%) developed bacteremia. The incidence of bacteremia was significantly lower in the adjuvant immunotherapy group than in the nonadjuvant immunotherapy group, respectively (P=0.038) (Table 2). The 1-year survival rate tended to be higher in the patients who had received adjuvant immunotherapy (100%) than in those who had not received the therapy (80%) (P=0.10; Table 2). The sources and causative pathogens of bacteremia are shown in Tables 3 and 4, respectively, and were as follows: one patient with primary bacteremia (MRSA), two patients with catheter-related infections (Escherichia coli, and Serratia sp), two patients with peritonitis (MRSA and Escherichia coli), and two patients with cholangitis (coagulase-negative Staphylococcus and Escherichia coli). There was no difference in postoperative clinical data, including duration of central venous catheterization, hospital stay, and intensive care unit stay between the two groups (Table 5).
Bacteremia is one of the main complications of liver transplantation, and it is also an important factor influencing the mortality associated with liver transplantation (6, 7). Many studies have revealed that the period of hospitalization and stay in the intensive care unit is longer for patients with postoperative bacteremia (4, 11, 12).
Adjuvant immunotherapy using liver allograft-derived lymphocytes was principally performed to prevent HCC recurrence. As these lymphocytes may have antibacterial properties, this study was performed as a secondary aim to investigate whether this adjuvant immunotherapy reduced the incidence of posttransplant bacteremia in LDLT recipients. In this one-to-one matching study using propensity score, we have shown that adjuvant immunotherapy using liver allograft-derived lymphocytes significantly reduced the incidence of postoperative bacteremia after LDLT. We have previously shown that stimulation with IL-2 significantly increased the expression of tumor necrosis factor-related apoptosis-inducing ligand on liver NK cells exerting anti-HCV and tumoricidal effects (9, 10). We have also confirmed that the number of interferon (IFN)-γ-secreting cells, including NK and NKT cells, in the peripheral blood of the liver transplant patients who received adjuvant immunotherapy was significantly higher than that in the peripheral blood of patients who did not receive the therapy at 14 days after LDLT (10). It is expected that the circulating activated NK and NKT cells in the peripheral blood may prevent postoperative bacteremia by exerting an antibacterial and antifungal effect, and the effect for preventing the postoperative infections may last during at least 2 weeks after LDLT. The adjuvant immunotherapy might reduce the number of postoperative infections as it involved the use of IL-2 stimulated NK and NKT cells. NK cell has protective functions for bacterial infection such as Staphylococcus aureus and Pseudomonas aeroginosa by NK cytotoxicity and augmentation of phagocytosis by macrophages by IFN-γ and tumor necrosis factor production secreted by NK cells (13, 14). NKT cells also have protective effects for bacterial infection by IFN-γ production (15). In this study, no adverse effects, such as graft-versus-host disease, were observed. To prevent graft-versus-host disease, we added the anti-CD3 monoclonal antibody to the culture medium a day before the inoculation. As the current immunosuppressive regimen following after LDLT reduces the adaptive immune components, adjuvant immunotherapy may be a promising approach for reducing the posttransplant bacteremia.
In this study, we adopted a one-to-one matching study using propensity score analysis to overcome bias due to the different distribution of covariates among patients from the two groups (patients who received adjuvant immunotherapy and those who did not). Bert et al. (7) have shown that gender, kidney transplant, intraoperative transfusions, MELD score, return to surgery, retransplantation, and biliary complications are associated with bacteremia. Kim et al. (8) have revealed an association between bacteremia and age, intravenous catheterization, United Network for Organ Sharing status IIA, and posttransplant hemodialysis. The clinical characteristics that may influence outcomes tended to differ between the two groups in the whole study (Table 1). The proportion of the older patients with HCC was significantly higher in the adjuvant immunotherapy group than those in no adjuvant immunotherapy group because the liver transplant patients who received adjuvant immunotherapy with IL-2-stimulated lymphocytes were limited to patients with HCC or HCV. The scores of MELD were significantly lower in the adjuvant immunotherapy group than in the nonadjuvant immunotherapy group. A high MELD score is known to indicate a major risk of postoperative infections after liver transplantation (16, 17). However, after matching using propensity scores analysis, there were no differences in clinical variables influencing outcomes such as age, MELD score, underlying liver disease, operative factors, and postoperative operative factors between the two groups. Iida et al. (6) have shown that the presence of preoperative massive pleural effusion or ascites requiring drainage were independent risk factors for postoperative bacteremia. In this study, there was no significant difference in the preoperative presence of ascites between the two groups. Although there was a significant difference in the incidence of bacteremia between the two groups, this study was unable to reveal a significant difference in outcomes, likely due to small study size. It is possible that the learning curve of individual surgeons or of the entire institution may be associated with an incidence of septic complications. Second, there was a selection bias confounding our results between the two groups, because adjuvant immunotherapy was selected into only patients with HCC or HCV, thus larger study is required to confirm the positive effect of adjuvant immunotherapy on outcomes of liver transplant recipients.
Bacteremia episodes were predominantly caused by gram-positive bacilli, which accounted for 55.6% of all isolated infections. Methicillin-resistant staphylococci were commonly observed among the gram-positive organism, whereas Escherichia coli strains were found to be the most prevalent species among the isolated gram-negative rods. During the 1990s, MRSA had become endemic and emerged as a major pathogen in many transplant centers. However, in the late 1990s, the incidence of gram-negative bacterial infection increased, probably because of the use of prophylactic antibiotics (18). In contrast, our study has shown that gram-positive bacteria have remained as the major pathogens (8, 19). Our results support recent studies that report coagulase-negative staphylococcus as the major cause of bacteremia (6). Catheter-related infections were found to occur frequently after surgery. Appropriate management of the catheter, which disrupts the organization of the skin epithelium and the mucosa, and early catheter removal are important for lowering the incidence of catheter-related infections.
In conclusion, we found that adjuvant immunotherapy using liver allograft-derived lymphocytes significantly reduced the incidence of postoperative bloodstream infections after LDLT in the setting of one-to-one matching study using propensity score. As the immunosuppressive regimen currently used after LDLT reduces the adaptive immune components, adjuvant immunotherapy using liver allograft-derived lymphocytes may be a promising approach for reducing posttransplant bacteremia.
MATERIALS AND METHODS
We retrospectively studied 114 consecutive patients who had undergone LDLT at Hiroshima University Hospital, from April 2004 to December 2009. The LDLTs had been performed after the approval of the Liver Transplantation Ethics Committee of Hiroshima University, and informed consents were obtained from all the patients.
Perioperative Management Strategy
The procedures for donor evaluation, donor surgery, recipient surgery, and perioperative management followed in our hospital have been described in previously published studies (20–22). A central venous catheter (triple lumen catheter, SA series, Arrow Corporation, Japan) was place in the internal jugular vein at induction of general anesthesia.
Antimicrobial prophylaxis consisted of intravenous cefmetazole (1.0 g) administration immediately before surgery and every 6 hr during surgery; thereafter, a dosage of 2 g/day was maintained for 5 days. Vancomycin (0.5 g/day, orally) was administered for 3 days before surgery. Itraconazole (200 mg/day) was orally administered for 7 days before surgery as prophylaxis against fungal infections. Trimethoprim/sulfamethoxazole (80/400 mg/day) was orally administered after surgery as prophylaxis against Pneumocystis.
The basic immunosuppression regimen comprised tacrolimus and methylprednisolone. If liver function stabilized, then the patients were weaned off the steroids 2 to 3 months after the operation. Rejection episodes were mainly treated with methylpredonisolone.
The use of adjuvant immunotherapy involving activated liver allograft-derived lymphocytes was approved by the Clinical Institutional Ethical Review Board of Hiroshima University, and the immunotherapy was started from January 2006. Gradient centrifugation with Ficoll-Paque was performed to isolate liver mononuclear cells from the perfusate effluents of liver grafts obtained from healthy donors. Liver mononuclear cells were cultured with human recombinant IL-2 for 3 days. A day before the infusion, the CD3-positive cell fraction was incubated with an anti-CD3 monoclonal antibody for opsonization. The purity of the isolated fractions was assessed by flow cytometric analysis, and the viability of the cells was assessed by dye- exclusion test before injection. The cells were suspended along with 5% human serum albumin in 0.9% sodium chloride; 3 days after liver transplantation, the cell suspension was then injected into patients with HCV or HCC (10).
Definitions of Bacteremia
Infections were defined according to the criteria proposed by the Centers for Disease Control (23). For this study, we included cases in which the infections had developed within 3 months after surgery and documented the first episode of bacteremia. Isolation of bacteria (other than common skin contaminants) from a single blood culture in the presence of clinical symptoms or signs of infection was considered proof of bacteremia. Bacteremia caused by common skin contaminants was considered significant only if the organism was also isolated from two blood cultures and was associated with clinical signs of infection. The bacteremia source was determined on the basis of clinical criteria and isolation from a clinically significant site of infection of the same organism found in the blood isolate on the basis of species identification and antibiotics susceptibility results. Bacteremia was classified as primary if it was of unknown origin (no physical, radiological, or pathological evidence of a definite infection source). Catheter-related infections were documented when the blood isolate was cultured from the catheter tip and blood cultures obtained by the catheter were found to be the same as the peripheral blood culture. Briefly, blood cultures were taken out by puncture. The blood culture was as follows: a 10-mL blood sample was aseptically inoculated at bedside into each bottle of a set of aerobic and anaerobic blood culture bottles. Bottles were incubated at 37°C for a period of 5 days or until a positive reading was detected by the instrument. Bacteremia was classified as primary if it was of unknown origin.
Continuous variables were compared using the Mann-Whitney test. Categorical data were compared using the χ2 test or Fisher's exact test, as appropriate. Overall survival analyses were carried out using the Kaplan-Meier method; comparisons between different groups were carried out using the log-rank test. To overcome bias due to the different distribution of covariates among patients from the two groups (patients who received adjuvant immunotherapy and those who did not), a one-to-one match was created using propensity score analysis (24, 25). The propensity score represents the probability of each individual patient being assigned to a particular condition in a study given a set of known covariates. Propensity scores are used to reduce selection bias by equating groups on the basis of these covariates and are used to adjust for selection bias in observational studies through matching. Variables entered in the propensity model were age, sex, donor age, the presence of HCC, previous surgery, graft-to-recipient weight ratio, MELD score, operative factors (blood loss during operation and operation time), and postoperative complications (biliary leakage and reoperation). The model was then used to obtain a one-to-one match by using the nearest-neighbor matching method. We used a matching algorithm based on linear predictive values without replacement, until all possible matches had been formed. Initially, matching was performed to five decimal points, followed by 4, 3, 2, and 1 decimal point matching; cases whose propensity score deviated more than 0.10 were considered unmatched (26, 27) Patients with unmatchable propensity scores were excluded from further analysis. Once the matched groups were obtained, differences in postoperative infection and prognosis were further analyzed to assess the unbiased influence of adjuvant immunotherapy on postoperative infection. Overall, survival analysis was performed within each matched subgroup to assess the influence of adjuvant immunotherapy on postoperative infection amended from the confounding factors. All analyses were performed using the SPSS 16.0, and P values less than 0.05 were considered statistically significant.
1. Dummer JS, Hardy A, Poorsattar A, et al. Early infections in kidney, heart, and liver transplant recipients on cyclosporine. Transplantation
1983; 36: 259.
2. Wade JJ, Rolando N, Hayllar K, et al. Bacterial and fungal infections after liver transplantation: An analysis of 284 patients. Hepatology
1995; 21: 1328.
3. Kim YJ, Kim SI, Wie SH, et al. Infectious complications in living-donor liver transplant recipients: A 9-year single-center experience. Transpl Infect Dis
2008; 10: 316.
4. Asensio A, Ramos A, Cuervas-Mons V, et al. Effect of antibiotic prophylaxis on the risk of surgical site infection in orthotopic liver transplant. Liver Transpl
2008; 14: 799.
5. Iinuma Y, Senda K, Fujihara N, et al. Surgical site infection in living-donor liver transplant recipients: A prospective study. Transplantation
2004; 78: 704.
6. Iida T, Kaido T, Yagi S, et al. Posttransplant bacteremia
in adult living donor liver transplant recepients. Liver Transpl
2010; 16: 1379.
7. Bert F, Larroque B, Paugam-Burtz C, et al. Microbial epidemiology and outcome of bloodstream infections in liver transplant recipients: An analysis of 259 episodes. Liver Transpl
2010; 16: 393.
8. Kim SI, Kim YJ, Jun YH, et al. Epidemiology and risk factors for bacteremia
in 144 consecutive living-donor liver transplant recipients. Yonsei Med J
2009; 50: 112.
9. Ishiyama K, Ohdan H, Ohira M, et al. Difference in cytotoxicity against hepatocellular carcinoma between liver and peripheral natural killer cells
in humans. Hepatology
2006; 43: 362.
10. Ohira M, Ishiyama K, Tanaka Y, et al. Adoptive immunotherapy with liver allograft-derived lymphocytes induces anti-HCV activity after liver transplantation in humans and humanized mice. J Clin Invest
2009; 119: 3226.
11. Reed A, Herndon JB, Ersoz N, et al. Effect of prophylaxis on fungal infection and costs for high-risk liver transplant recipients. Liver Transpl
2007; 13: 1743.
12. Garcia Prado ME, Matia EC, et al. Surgical site infection in liver transplant recipients: Impact of the type of perioperative prophylaxis. Transplantation
2008; 85: 1849.
13. Emoto M, Yoshizawa I, Emoto Y, et al. Rapid development of a gamma interferon-secreting glycolipid/CD1d-specific Vα14+
T-cell subset after bacterial infection. Infect Immun
2006; 74: 5903.
14. Borchers MT, Harris NL, Wesselkamper SC, et al. The NKG2D-activating receptor mediates pulmonary clearance of pseudomonas aeroginosa. Infect Immun
2006; 74: 2578.
15. Small CL, McCormick S, Gill N, et al. NK cells play a critical protective role in host defense against extracellular staphylococcus aureus bacterial infection in the lung. J Immunol
2008; 180: 5558.
16. Merion RM, Schaubel DE, Dykstra DM, et al. The survival benefit of liver transplantation. Am J Transplant
2005; 5: 307.
17. Ishigami M, Honda T, Okumura A, et al. Use of the model for end-stage liver disease (MELD) score to predict 1-year survival of Japanese patients with cirrhosis and to determine who will benefit from living donor liver transplantation. J Gastroenterol
18. Singh N. Impact of current transplantation practices on the changing epidemiology of infections in transplant recipients. Lancet Infect Dis
2003; 3: 156.
19. Saner FH, Olde Damink SW, Pavlakovic G, et al. Pulmonary and blood stream infections in adult living donor and cadaveric liver transplant patients. Transplantation
2008; 85: 1564.
20. Itamoto T, Emoto K, Mitsuda H, et al. Safety of donor right hepatectomy for adult-to-adult living donor liver transplantation. Transpl Int
2006; 19: 177.
21. Tashiro H, Ohdan H, Itamoto T, et al. Using recipient's middle hepatic vein for drainage of the right paramedian sector in right liver graft. Transplantation
2008; 86: 1565.
22. Tashiro H, Itamoto T, Sasaki T, et al. Biliary complications after duct-to-duct biliary reconstruction in living-donor liver transplantation
: Causes and treatment. World J Surg
2007; 31: 2222.
23. Garner JS, Jarvis WR, Emori TG, et al. CDC definitions for nosocomial infections. Am J Infect Control
1988; 16: 128.
24. Zinsmeister AR, Connor JT. Ten common statistical errors and how to avoid them. Am J Gastroenterol
2008; 103: 262.
25. Layer P, Zinsmeister AR, DiMagno EP. Effects of decreasing intraluminal amylase activity on starch digestion and postprandial gastrointestinal function in humans. Gastroenterology
1986; 91: 41.
26. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med
1997; 127: 757.
27. D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med
1998; 17: 2265.
Keywords:© 2011 Lippincott Williams & Wilkins, Inc.
Living-donor liver transplantation; Natural killer cells; Bacteremia; Adjuvant immunotherapy