Of 150 patients who underwent damage control (DC), 58 had a single, planned additional operation for delayed biliary reconstruction (DBR), 26 had a single, unplanned additional operation during which the bile duct was reconstructed, and 57 had multiple additional operations for various complications. Five patients died before any additional operation, and 4 patients underwent additional transplant at the first additional operation. OLT indicates orthotopic liver transplant.
eTable 1. Comparison of Recipient Characteristics
eTable 2. Comparison of Recipient Pretransplant Acuity
eTable 3. Comparison of Donor and Operative Characteristics
eTable 4. Postoperative Outcomes
eFigure. Receiver Operating Characteristic (ROC) Curve Demonstrates a C-Statistic of 0.75 for the Multivariate Model of Predictors of Damage Control
DiNorcia J, Lee MK, Harlander-Locke MP, Xia V, Kaldas FM, Zarrinpar A, Farmer DG, Yersiz H, Hiatt JR, Busuttil RW, Agopian VG. Damage Control as a Strategy to Manage Postreperfusion Hemodynamic Instability and Coagulopathy in Liver Transplant. JAMA Surg. 2015;150(11):1066-1072. doi:10.1001/jamasurg.2015.1853
Damage control (DC) with intra-abdominal packing and delayed reconstruction is an accepted strategy in trauma and acute care surgery but has not been evaluated in liver transplant.
To evaluate the incidence, effect on survival, and predictors of the need for DC using intra-abdominal packing and delayed biliary reconstruction in patients with coagulopathy or hemodynamic instability after liver allograft reperfusion.
Design, Setting, and Participants
We performed a retrospective analysis of adults undergoing liver transplant at a large transplant center from February 1, 2002, through July 31, 2012.
Main Outcomes and Measures
Predictors of DC, effects on graft, and patient survival.
Of 1813 patients, 150 (8.3%) underwent DC during liver transplant, with 84 (56.0%) requiring a single additional operation for biliary reconstruction and abdominal closure and 57 (38.0%) requiring multiple additional operations. Compared with recipients without DC, patients requiring DC had greater Model for End-stage Liver Disease scores (33 vs 27; P < .001); more frequent pretransplant hospitalization (72.0% vs 47.9%; P < .001), intubation (33.3% vs 19.9%; P < .001), vasopressors (23.2% vs 10.9%; P < .001), renal replacement therapy (49.6% vs 30.3%; P < .001), and prior major abdominal operations (48.3% vs 21.9%; P < .001), including prior liver transplant (29.3% vs 8.9%; P < .001); greater operative transfusion requirements (37 vs 13 units of packed red blood cells; P < .001); worse intraoperative base deficit (10.3 vs 8.4; P = .03); more frequent postreperfusion syndrome (56.2% vs 27.3%; P < .001); and longer cold (430 vs 404 minutes; P = .04) and warm (46 vs 41 minutes; P < .001) ischemia times. Patients who underwent DC followed by a single additional operation for biliary reconstruction and abdominal closure had similar 1-, 3-, and 5-year graft survival (71%, 62%, and 62% vs 81%, 71%, and 67%; P = .26) and patient survival (72%, 64%, and 64% vs 84%, 75%, and 70%; P = .15) compared with recipients not requiring DC. Multivariate predictors of DC included prior liver transplant or major abdominal operation, longer pretransplant recipient and donor length of stay, greater Model for End-stage Liver Disease score, and longer warm and cold ischemia times (C statistic, 0.75).
Conclusions and Relevance
To our knowledge, this study represents the first large report of DC as a viable strategy for liver transplant recipients with coagulopathy or hemodynamic instability after allograft reperfusion. In DC recipients not requiring additional operations, outcomes are excellent and comparable to 1-stage liver transplant.
Damage control (DC) is an accepted strategy to treat seriously injured patients1- 4 and is increasingly used to manage critically ill patients with nontraumatic abdominal emergencies.5- 7 The principles of DC include abbreviated laparotomy to control blood loss and contamination, resuscitation in the intensive care unit, and planned additional operation.8- 10 Damage control avoids the vicious cycle perpetuated by hypothermia, acidosis, and coagulopathy and allows for the restoration of normal physiologic mechanisms before definitive surgical management.11,12 Although associated with significant morbidity, DC provides indisputable benefit to properly selected patients, with reported survival rates after severe trauma ranging from 58% to 90%.13
Patients undergoing orthotopic liver transplant (OLT) experience varying degrees of hypothermia, acidosis, and coagulopathy after reperfusion depending on the difficulty of the hepatectomy, volume of blood transfusions, presence of postreperfusion syndrome (PRS), and function of the allograft.14- 17 In routine cases, the physiologic derangements are corrected intraoperatively, and the transplant is completed in a single operation; in challenging cases, the physiologic derangements may persist, leading to worsening hemodynamic instability, bleeding, and coagulopathy.
Damage control is a valuable option for the challenging liver transplant. The strategy includes completion of portal venous and hepatic arterial anastomoses for allograft reperfusion, deferral of biliary reconstruction, intra-abdominal packing, and resuscitation in the intensive care unit before packing removal, biliary reconstruction, and closure of the abdomen within 48 hours. Although DC is widely used in trauma and acute care surgery, it has not been reported in liver transplantation. The specific aims of this study were to evaluate the use of DC, effects on survival, and predictors of need for DC in a large series of patients undergoing OLT in the post–Model for End-stage Liver Disease (MELD) era.
We retrospectively reviewed a prospectively maintained transplant database and identified all adult patients (aged ≥18 years) who underwent OLT (primary and additional transplants) from February 1, 2002, through July 31, 2012, at the University of California, Los Angeles (UCLA). The UCLA Institutional Review Board approved the study. Informed consent was not required.
Recipient variables analyzed included age, sex, medical and surgical comorbidities, MELD score, pretransplant hospitalization status and length of hospital stay (LOS), pretransplant need for mechanical ventilation, vasopressor support, and renal replacement therapy (RRT). Medical comorbidities included hypertension, diabetes mellitus, coronary artery disease, and presence of ascites. Coronary artery disease was defined as a history of myocardial infarction, prior percutaneous or open revascularization, or nonocclusive coronary atherosclerosis detected on angiography. Surgical comorbidities included prior abdominal operations, categorized as minor (pelvic or gynecologic surgery, laparoscopic cholecystectomy, appendectomy, and inguinal or ventral hernia repairs) and major (any exploratory laparotomy, gastric, small or large intestinal operation, open cholecystectomy, and liver surgery, including prior OLT).
Donor variables included age, sex, serum sodium level, number of vasopressors, LOS, and graft type (such as whole or split cadaveric, donation after cardiac death, living donor grafts, and expanded criteria donor [ECD] grafts based on criteria published previously).18 Donor LOS was calculated from the day of hospital admission to the day of organ procurement. Operative variables included cold ischemia time (CIT), warm ischemia time (WIT), transfusion of packed red blood cells (PRBCs), worst base deficit, PRS, use of venovenous bypass, and need for bolus or infusion of vasopressors. Postreperfusion syndrome was defined as a decrease in the mean arterial pressure greater than 30% below baseline, lasting at least 1 minute and occurring within the first 5 minutes of reperfusion.19
Variables collected as measures of postoperative morbidity included the development of graft nonfunction, hepatic artery thrombosis, portal vein thrombosis, and biliary and infectious complications. Graft nonfunction was defined as the need for additional transplant during the index admission or recipient death due to all-cause graft failure and included true primary nonfunction and graft loss due to hepatic artery thrombosis, portal vein thrombosis, and other perioperative factors. Outcome measures included posttransplant LOS, 30-day mortality, and overall graft and patient survival. Pretransplant LOS was calculated from the day of admission to the day of transplant; posttransplant LOS was calculated from transplant to the day of discharge from the hospital.
A DC strategy with deferral of biliary reconstruction and intra-abdominal packing was based on a joint decision between the surgeon and anesthesiologist in critically ill patients with ongoing hypothermia, acidosis, coagulopathy, transfusion requirements, and/or hemodynamic instability that required vasopressor support after arterial reperfusion. After meticulous hemostasis, ongoing coagulopathy was treated with antifibrinolytic agents, including tranexamic acid and aminocaproic acid. The abdomen was temporarily packed with laparotomy pads, and the patient was warmed to a goal temperature greater than 35°C. If the physiologic derangements persisted beyond 60 minutes despite hemostasis and resuscitation, then biliary reconstruction was deferred, the abdomen was packed with laparotomy pads, the fascia was left open, and the skin was closed. The patient was brought to the intensive care unit for resuscitation and stabilization. A blood transfusion requirement greater than 6 to 8 units of PRBCs in 24 hours prompted consideration for urgent additional laparotomy.20 Otherwise, additional operation for packing removal, biliary reconstruction, and closure was planned within 48 hours when normal physiologic mechanism was restored.
Patients requiring DC were further categorized as needing one planned additional operation for delayed biliary reconstruction or having had multiple complications, including a single unplanned additional operation for ongoing bleeding during which biliary reconstruction was performed, multiple subsequent operations, or graft failure with additional transplant or death.
Comparisons were made between patients with and without DC at the time of OLT. Continuous variables were compared using the Mann-Whitney test and summarized as medians and interquartile ranges. Categorical variables were compared using χ2 or Fisher exact tests and summarized as percentages. Graft and patient survival curves were computed using Kaplan-Meier methods and compared using log-rank tests. To allow comparison of similarly challenging cohorts, propensity analyses using logistic regression were performed by matching patients requiring DC with controls not requiring DC on both preoperative factors (1:2 on recipient MELD score, LOS, and prior abdominal surgery) and intraoperative factors (1:1 on units of PRBCs, worst base deficit, and PRS). A multivariable logistic regression model was used to evaluate the effect of 14 recipient (age, sex, hypertension, diabetes, coronary artery disease, ascites, prior major abdominal operation, prior OLT, MELD score, pretransplant hospitalization and LOS, intubation, vasopressors, and RRT), 6 donor (age, sex, terminal serum sodium, number of vasopressors, LOS, and graft type), and 2 operative variables (CIT and WIT) on the need for DC. Missing values were singly imputed using the Markov Chain Monte Carlo regression method. The final model was selected using the backward procedure for variable selection with a liberal P < .15 as the retention criterion. Linearity was evaluated by fitting splines. Variables that did not conform to the linearity assumption were modeled using best thresholds. The best threshold for a given continuous predictor was defined as the value of the predictor that maximized the χ2 statistic for the odds ratio (OR) under the final logistic model. All effects were assumed to be additive on the log odds scale. Results are summarized as ORs and 95% CIs. P ≤ .05 was considered statistically significant. Model accuracy was summarized using the concordance C statistic.
Of 1813 adult patients who underwent OLT during the study period, 150 (8.3%) required DC during OLT (Figure 1). Of these 150, 84 (56.0%) required a single additional operation for removal of packing and completion of the biliary anastomosis, with 58 (38.7%) as a planned procedure when the patient was stabilized and 26 (17.3%) as an unplanned additional operation due to ongoing bleeding and hemodynamic instability. Multiple subsequent additional operations were required in 57 patients (38.0%), 5 patients (3.3%) died before any additional operation, and 4 patients (2.7%) underwent additional transplant at the second laparotomy. The median time to the first additional operation was 2 days.
Baseline recipient characteristics and pretransplant acuity are listed in eTable 1 and eTable 2 in the Supplement. Compared with recipients not requiring DC, patients undergoing DC were significantly more likely to have prior major abdominal operations (48.3% vs 21.9%; P < .001), including prior OLT (29.3% vs 8.9%; P < .001); greater MELD scores (33 vs 27; P < .001); more frequent pretransplant hospitalization (72.0% vs 47.9%; P < .001); longer pretransplant LOS (19 vs 8 days; P < .001); and greater need for mechanical ventilation (33.3% vs 19.9%; P < .001), vasopressors (23.2% vs 10.9%; P < .001), and RRT (49.6% vs 30.3%; P < .001) before OLT. No significant differences were found in age, sex, medical comorbidities, or presence of ascites among the groups.
Donor and operative characteristics are listed in eTable 3 in the Supplement. Compared with recipients without DC, recipients requiring DC had significantly longer CIT (430 vs 404 minutes; P = .04) and WIT (46 vs 41 minutes; P < .001), greater blood transfusion requirements (37 vs 13 units of PRBCs; P < .001), greater base deficit (10.3 vs 8.4; P = .03), more frequent PRS (56.2% vs 27.3%; P < .001), and more frequent use of venovenous bypass (64.7% vs 36.3%; P < .001), vasopressor boluses (72.4% vs 29.8%; P < .001), and infusions (85.9% vs 60.2%; P < .001). No significant differences were found in donor age, sex, terminal serum sodium level, number of vasopressors, LOS, or graft type, with a trend toward increased need for DC for grafts with greater ECD scores (ECD score 0, 5.9% DC; ECD score 1: 8.3% DC; ECD score 2: 10.5% DC; and ECD score 3 or 4; 10.3%; overall P = .053).
Postoperative outcomes are given in eTable 4 in the Supplement. Compared with patients not requiring DC, patients requiring DC had longer posttransplant LOS (53 vs 31 days; P < .001) and greater incidence of infection (49.3% vs 21.4%; P < .001), graft nonfunction (18.7% vs 3.3%; P < .001), and mortality within 30 days (15.3% vs 4.0%; P < .001). No significant differences were found in the incidence of hepatic artery thrombosis, portal vein thrombosis, or biliary complications between the groups.
Kaplan-Meier graft and patient survival estimates are given in Figures 2, 3, and 4. Compared with recipients undergoing standard 1-stage OLT, patients requiring DC (Figure 2) had inferior 1-, 3-, and 5-year graft survival (51%, 44%, and 43% vs 81%, 71%, and 67%; P < .001) and patient survival (62%, 55%, and 54% vs 84%, 75%, and 70%; P < .001). However, DC patients who underwent a planned, single subsequent operation for delayed biliary reconstruction had similar 1-, 3-, and 5-year graft survival (71%, 62%, and 62% vs 81%, 71%, and 67%; P = .26) and patient survival (72%, 64%, and 64% vs 84%, 75%, and 70%; P = .15) compared with 1-stage OLT recipients and significantly superior 1-, 3-, and 5-year graft survival (71%, 62%, and 62% vs 38%, 33%, and 30%; P < .001) and patient survival (72%, 64%, and 64% vs 55%, 50%, and 47%; P < .001) compared with DC recipients with multiple complications (Figure 3).
Figure 4 compares 1-, 3-, and 5-year graft survival estimates among 1-stage OLT and DC recipients based on number of additional operations. Patients undergoing DC requiring a single additional operation had equivalent graft survival compared with patients undergoing 1-stage OLT who subsequently underwent additional operations for other indications (66%, 59%, and 56% vs 73%, 64%, and 60%; P = .52); similarly, DC recipients requiring multiple additional operations had poor but equivalent graft survival compared with patients undergoing 1-stage OLT requiring multiple additional operations for other indications (35%, 29%, and 29% vs 48%, 43%, and 40%; P = .06).
Despite propensity matching to controls not requiring DC based on preoperative MELD score, LOS, and prior abdominal surgery, patients undergoing DC were significantly more likely to require venovenous bypass (65.3% vs 45.3%; P < .001), require intraoperative blood transfusions (35 vs 13 units of PRBCs; P < .001), and had greater PRS (57.3% vs 26.3%; P < .001), worse intraoperative base deficits (10.2 vs 7.8; P < .001), and significantly worse postoperative outcomes, including LOS (31.0 vs 26.0 days; P = .04), infection (49.3% vs 29.9%; P < .001), and 1-, 3-, and 5-year graft survival (52.0%, 46.0%, and 44.0% vs 73.0%, 61.0%, and 56.0%; P < .001) and patient survival (62.0%, 56.0%, and 54.0% vs 75.0%, 63.0%, and 58.0%; P = .08) compared with their matched cohort. When matched to controls not requiring DC on intraoperative factors, patients undergoing DC had worse postoperative outcomes, including graft nonfunction, death within 7 days, need for additional unplanned additional operations, and inferior graft and patient survival.
Multivariable logistic regression analysis identified 7 factors significantly associated with DC during OLT (Table). These factors included prior OLT (OR, 4.74; 95% CI, 2.90-7.75; P < .001), prior major abdominal operation (OR, 2.53; 95% CI, 1.51-4.24; P < .001), pretransplant LOS of 30 days or longer (OR, 2.48; 95% CI, 1.53-4.02; P < .001), WIT of 45 minutes or longer (OR, 2.19; 95% CI, 1.53-3.15; P < .001), donor LOS of 3 days or longer (OR, 1.76; 95% CI, 1.13-2.74; P = .01), MELD (OR, 1.04 per point increase; 95% CI, 1.02-1.05; P < .001), and CIT (OR, 1.002 per minute increase; 95% CI, 1.001-1.003; P = .006). The C statistic for the model was 0.75 (eFigure in the Supplement).
Although DC with intra-abdominal packing and temporary closure has gained popularity in the treatment of patients with multisystem trauma,4,13,21,22 hepatic trauma,23- 31 and nontrauma,5- 7,32,33 few data exist regarding its use in liver transplant, with the literature limited to case reports and small case series.34- 36 This study is the largest contemporary single-center experience with DC in liver transplant. We report graft and patient outcomes and define predictors of the need for DC.
Recipients requiring DC had significantly greater acuity of illness compared with patients not requiring DC, with greater MELD scores and need for hospitalization, mechanical ventilation, vasopressors, and RRT before transplant.37 Compounding the critical illness, nearly 50% of recipients requiring DC had a history of prior major abdominal surgery, including 30% with prior OLT. Adhesions from prior surgery, particularly prior OLT, increase the difficulty of dissection and bleeding during the transplant, which in part explain the more frequent use of venovenous bypass, longer CIT and WIT, and larger transfusion requirements in the recipients who required DC. In addition, patients who required DC had more profound physiologic derangement and hemodynamic instability, including greater base deficit, more frequent PRS, and greater use of vasopressors intraoperatively.
A major objective of our study was to evaluate the association of DC with posttransplant outcomes. Recipients who underwent DC had greater incidence of infection, graft nonfunction, and death within 30 days. In the literature, the DC strategy improves outcomes over prolonged definitive operations in critically ill patients, but the morbidity and mortality remain high.7 To further evaluate the inferior outcomes in the DC cohort, we performed 2 propensity-matched analyses. To control for preoperative acuity, patients requiring DC were matched to patients not requiring DC on laboratory MELD score, LOS, and prior abdominal surgery. Although there was excellent matching on these preoperative acuity measures, patients not requiring DC had significantly less need for intraoperative venovenous bypass and transfusions and had less PRS and acidosis, indicating that the intraoperative environment that dictated the need for DC was not recapitulated in those not requiring DC. When matching on intraoperative factors (PRBCs, PRS, and worst base deficit), DC recipients continued to have significantly greater base deficit at the end of the operation, signifying that additional factors unaccounted for in the analysis led to the need for DC in this subset of patients. On the basis of these analyses, the increased morbidity and mortality in our series are likely less attributable to the DC strategy itself and more a reflection of the high acuity of our transplant recipients.38
Overall, patients who underwent DC had inferior 1-, 3-, and 5-year graft and patient survival compared with recipients who underwent 1-stage OLT. However, to address whether the need for this additional planned operation was detrimental in itself, DC recipients undergoing a planned single additional operation for biliary reconstruction and abdominal closure were compared with patients who underwent 1-stage OLT and revealed equivalent survival. Furthermore, DC recipients with a single additional operation (planned or unplanned) had equivalent survival to 1-stage OLT recipients who subsequently required additional operation, and survival was uniformly poor among DC and 1-stage OLT recipients who required multiple additional operations. Collectively, these data suggest that the DC strategy itself is not deleterious to outcomes and in properly selected patients is an effective strategy when critical illness precludes 1-stage transplant.
Finally, we analyzed perioperative factors that might predict the use of DC in liver transplant. Prior OLT and prior major abdominal operation were the most significant predictors. Prior transplant and other major abdominal surgery portend a difficult hepatectomy and potential for raw surface bleeding. Moreover, increased donor and recipient pretransplant LOS and greater recipient MELD scores also predicted DC, reflecting how prolonged donor hospitalization affects liver allograft function and recipient illness and debility contribute to a challenging operation. The WIT and CIT also were important. Increased time to liver allograft reperfusion (WIT) increased the odds of requiring DC more than 2-fold, and CIT increased the odds of requiring DC by 0.2% per each additional minute. Prolonged WIT is a surrogate for a difficult operation; prolonged CIT may contribute to increased graft ischemia and reperfusion injury causing postreperfusion graft dysfunction and coagulopathy that may lead to the need for DC.
The use of ECD or marginal livers also correlated with an increased need for DC. Although categorization by the ECD score did not reach statistical significance, our multivariate analysis revealed that 3 of the 4 factors contributing to the ECD score (donor LOS, CIT, and WIT) were independently significant in predicting DC, supporting the notion that the use of ECD livers, particularly in high-acuity recipients, may influence intraoperative outcomes and the need for DC. Our data predict the need for DC in a recipient with prior major abdominal surgery, protracted hospitalization, and high MELD score who is receiving a liver with prolonged ischemia times from a donor with a protracted hospitalization. Our transplant program has previously reported that MELD score, pretransplant septic shock, cardiac risk, and comorbidities were independent predictors of futile outcome after liver transplant.39 Coupled with that study, the data from the current study may inform donor-recipient matching to optimize outcomes in high-acuity patients and perhaps avoid a futile transplant.
In the trauma literature, common indications for DC include hypothermia (temperature <35°C), severe metabolic acidosis (pH <7.30), significant bleeding requiring massive transfusion (>10 units of PRBCs), and coagulopathy on laboratory results or presenting as nonsurgical bleeding.40 Although we did not use specific trigger points to initiate DC, DC was strongly considered in recipients who remained refractorily hypothermic (temperature <35°C) and acidemic (pH <7.25) and who required more than 20 to 25 units of PRBCs with ongoing raw surface bleeding. Ultimately, consideration for DC for an individual patient required the collective decision between the surgeon and anesthesiologist, assessing the patient’s temperature, acid-base and volume status, transfusion requirement, coagulopathy, and hemodynamic and cardiac stability. Defining and validating more specific perioperative parameters that might indicate when to perform DC in liver transplant are areas for future study.
The main limitation of this study is its retrospective, single-center design. Although variation in patient management over time is a potential bias in any retrospective study, we minimized this effect by limiting the analysis to post-MELD era recipients. Our high-acuity recipients, including patients undergoing additional transplants, represent a study group that may be different from other transplant centers, and DC in liver transplant may not be universally needed.
We report, to our knowledge, the largest single-center series of DC in liver transplant. Damage control is a viable option for the management of challenging recipients with intraoperative hypothermia, acidosis, coagulopathy, and hemodynamic instability after allograft reperfusion. It may be required in up to 8% of high-acuity recipients, with favorable outcomes equivalent to 1-stage OLT in patients requiring a single additional operation. We identify important recipient and perioperative factors that accurately predict the need for DC and may help in the early identification of patients most likely to benefit from a 2-stage OLT procedure.
Accepted for Publication: April 25, 2015.
Corresponding Author: Vatche G. Agopian, MD, Dumont-UCLA Liver Cancer and Transplant Centers, Pfleger Liver Institute, Department of Surgery, David Geffen School of Medicine at University of California, Los Angeles, 757 Westwood Plaza, Ste 8501-B, Los Angeles, CA 90095 (firstname.lastname@example.org).
Published Online: August 26, 2015. doi:10.1001/jamasurg.2015.1853.
Author Contributions: Dr Agopian had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: DiNorcia, Xia, Zarrinpar, Farmer, Busuttil, Agopian.
Acquisition, analysis, or interpretation of data: DiNorcia, Lee, Harlander-Locke, Xia, Kaldas, Zarrinpar, Farmer, Yersiz, Hiatt, Agopian.
Drafting of the manuscript: DiNorcia, Harlander-Locke, Hiatt, Agopian.
Critical revision of the manuscript for important intellectual content: DiNorcia, Lee, Xia, Kaldas, Zarrinpar, Farmer, Yersiz, Hiatt, Busuttil, Agopian.
Statistical analysis: DiNorcia, Lee, Agopian.
Administrative, technical, or material support: Harlander-Locke, Kaldas, Zarrinpar, Farmer, Yersiz, Agopian.
Study supervision: Farmer, Yersiz, Busuttil, Agopian.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the 86th Annual Meeting of the Pacific Coast Surgical Association; February 21, 2015; Monterey, California.
Additional Contributions: Daniela Markovic, MS, provided expert help in the statistical analysis and was compensated for her work.