Portland Intensive Insulin Therapy During Living Donor Liver Transplantation: Association with Postreperfusion Hyperglycemia and Clinical Outcomes

Many liver transplant recipients experience intraoperative hyperglycemia after graft reperfusion. Accordingly, we introduced the Portland intensive insulin therapy (PoIIT) in our practice to better control blood glucose concentration (BGC). We evaluated the effects of PoIIT by comparing with our conventional insulin therapy (CoIT). Of 128 patients who underwent living donor liver transplantation (LDLT) during the phaseout period of CoIT, 89 were treated with the PoIIT and 39 were treated with CoIT. The primary outcome was hyperglycemia (BGC > 180 mg/dL) during the intraoperative postreperfusion phase. The secondary outcomes were postoperative complications such as infection. The incidence of hyperglycemia (22.5% vs. 53.8%, p = 0.001) and prolonged hyperglycemia for >2 hours (7.9% vs. 30.8%, p = 0.002) was significantly lower in PoIIT group than in CoIT group. A mixed linear model further demonstrated that repeatedly measured BGCs were lower in PoIIT group (p < 0.001). The use of PoIIT was significantly associated with decreases in major infections (OR = 0.23 [0.06–0.85], p = 0.028), prolonged mechanical ventilation (OR = 0.29 [0.09–0.89], p = 0.031), and biliary stricture (OR = 0.23 [0.07–0.78], p = 0.018) after adjustments for age, sex, and diabetes mellitus. In conclusion, the PoIIT is effective for maintaining BGC and preventing hyperglycemia during the intraoperative postreperfusion phase of living donor liver transplantation with potential clinical benefits.

doses by accounting for patient insulin sensitivity that is evaluated according to the rate of BGC change at a set insulin dose. That is, calculated insulin dose differs according to the rate of BGC change even if the measured BGCs are the same 15 . Thus, we had deduced that the PoIIT effectively controls rapid intraoperative hyperglycemic changes in liver transplant recipients and introduced the PoIIT to our practice in January 2015. Among various algorithms with different target BGC ranges, we chose the PoIIT with the lowest target range (80-120 mg/dL) to rapidly control hyperglycemia and minimize the duration of intraoperative hyperglycemia 16 . We also considered that the risk of hypoglycemia is low due to frequent intraoperative BGC checkups 17 . Herein, we evaluated the effects of the PoIIT by comparing results with those of our conventional insulin therapy (CoIT) in recipients who underwent liver transplantation during the phaseout period of CoIT.

Materials and Methods
Subjects and data collection. We reviewed the medical records of 128 patients who underwent their first adult-to-adult living donor liver transplantations at our institution during the phaseout period of CoIT (implantation period of the PoIIT) between January 2015 and July 2017. Of these, 39 recipients were treated with CoIT and 89 recipients were treated with the PoIIT. During the phaseout period, the PoIIT was gradually established as a standard protocol by ensuring patient safety and compliance with attending anesthetists (Supplementary Table S1). All data were collected from computerized medical records or a liver transplant database (prospectively collected) and were anonymized and de-identified prior to analysis. The Institutional Review Board of Samsung Medical Center approved this retrospective study (SMC 2016-11-016) and waived the requirement for written informed consent. All procedures in this study were performed in accordance with the relevant guidelines and regulations. No grafts were procured from prisoners. All surgeries were performed at our hospital by surgeons in the Department of surgery, Samsung Medical Center.
Perioperative glycemic management. Recipients fasted starting the evening before surgery and 5% dextrose solution was infused at a rate of 80 mL/h during the fasting period. Preoperative oral carbohydrate supplements were not provided. During surgery, arterial BGCs were measured in combination with other arterial blood-derived parameters using a blood gas/chemistry analysis device (RAPIDLAB1265, Siemens Healthcare Diagnostics Inc., Berlin, Germany) on an hourly basis, as well as at the following additional time points: start of the anhepatic phase, immediately before graft reperfusion, and 5 and 30 minutes after graft reperfusion, as described previously 11 . The blood gas/chemistry analysis device was near each patient and clinicians were able to reach the device within 1 minute after arterial blood sampling. Insulin therapy was not indicated for controlling BGC before graft reperfusion because BGC generally decreases during the anhepatic phase due to the lack of hepatic glucose production 18 Fig. S1) 16 . After ICU arrival, glycemic management was performed by liver transplant surgeons based on a standardized protocol that is independent from intraoperative glycemic protocols. BGCs were measured at ICU arrival and every 8 hours thereafter using a blood gas/chemistry analysis device (RAPIDLAB1265 Anesthetic management. Anesthesia was performed based on a standardized institutional protocol as described previously 11 . In short, mechanical ventilation was delivered at a tidal volume of 8 mL per ideal body weight (kg) using a mixture of medical air and oxygen with positive end-expiratory pressure set at 6 mmHg. The respiratory rate was adjusted as needed to maintain normocapnea. Vasoactive drugs were used to maintain mean arterial pressure >70 mmHg. Metabolic acidosis was corrected with sodium bicarbonate when the base deficit was >10 mEq/L. Body core temperature was maintained using a whole body-sized warm blanket, airway humidifiers, and fluid warming devices. Transfusion of allogeneic blood was strictly controlled based on a restrictive and prophylactic policy, with each blood component transfused separately according to its respective indication. Blood salvage was routinely used for intraoperative autotransfusion irrespective of the presence of a hepatic tumor 20 .
Perioperative surgical procedures. Acceptance criteria for liver donation were age ≤65 years, body mass index <35 kg/m 2 , macrosteatosis ≤30%, and residual liver volume ≥30%. Individuals with any type of hepatitis or fibrosis were excluded from donation. All grafts consisted of segments 5-8 excluding the middle hepatic vein trunk. Graft implantation was performed using the piggyback technique. After the portal vein anastomosis was completed, the graft was reperfused by consecutively unclamping the hepatic vein and portal vein. The hepatic artery was subsequently anastomosed, followed by biliary anastomosis. Immunosuppression was performed based on a quadruple regimen consisting of methylprednisolone, basiliximab, mycophenolate mofetil, and tacrolimus as described previously 21 . In particularly, recipients received methylprednisolone 500 mg intravenously before graft reperfusion for the induction of immunosuppression and 500 mg daily until postoperative day 2, followed by a tapered dose of 60 mg per day for 5 days, and then 8 mg twice per day for 1 month.
Variables and statistical analysis. The primary outcome was hyperglycemia during the intraoperative postreperfusion phase. Hyperglycemia was defined when a BGC was >180 mg/dL based on previous research 22

Allogeneic transfusion
Red blood cells (units)  26 . Major infections included septicemia, peritonitis, pneumonia, and tissue-invasive cytomegalovirus disease. During the postoperative period, the first BGC or potassium concentration measured each day was used for analysis. Continuous variables are expressed as mean ± standard deviation or median (25th percentile-75th percentile) and were analyzed by t-test or the Mann-Whitney test. Repeatedly measured continuous variables were analyzed by a mixed linear model with the Bonferroni correction. Categorical variables are expressed as frequency (%) and were analyzed by chi-square test, Fisher's exact test, or binary logistic regression, as appropriate. In terms of postoperative complications, potential compounding effect from age, sex, and diabetes mellitus was adjusted during multivariable analysis and the risk of type I error or the false discovery rate was controlled by using the Benjamini-Hochberg procedure 27 . All reported p values were 2-sided, and p < 0.05 was considered statistically significant. Analyses were performed using SPSS version 23.0 (IBM, Armonk, NY, USA) or SAS version 9.2 (SAS institute, Cary, NC, USA).

Postoperative blood glucose and potassium concentrations.
There were no significant differences between the two groups regarding blood glucose and potassium concentrations after the cessation of intraoperative glycemic protocols and the start of the ICU glycemic protocol (Supplementary Fig. S2), indicating that a single ICU glycemic protocol was performed similarly for the two groups. Postoperative complications. As shown in Fig. 4a, repeatedly measured postoperative ASTs were significantly lower in PoIIT group than in CoIT group p = 0.015, indicating less hepatocyte injury in relation to the use of the PoIIT. In particular, AST level immediately after ICU arrival was 257 ± 159 IU/L in PoIIT group and 390 ± 333 IU/L in CoIT group. Repeatedly measured ALTs were insignificantly lower in PoIIT group (p = 0.412) (Fig. 4b). In particular, ALT level immediately after ICU arrival was 213 ± 167 IU/L in PoIIT group and 340 ± 347 IU/L in CoIT group. As shown in Table 3

Discussion
In this study, we demonstrated that the PoIIT was superior to our CoIT for controlling acute hyperglycemic disturbance after graft reperfusion and for preventing intraoperative hyperglycemia during the postreperfusion phase in living donor liver transplantations. Hyperglycemia risk was reduced by 60% in relation to the use of the PoIIT (22.5% vs. 53.8%). Although greater amounts of insulin were used with the PoIIT, no recipients developed clinically relevant hypoglycemia. PoIIT showed an additional advantage by preventing increases in potassium concentration after graft reperfusion. Moreover, the use of the PoIIT was associated with improved post-transplant clinical courses in terms of hepatocyte injury, major infection, prolonged mechanical ventilation, and biliary stricture. Among the various IITs 16 , our team selected the PoIIT because it was designed for surgical patients, while many IITs are for medical ICU patients, and because the PoIIT has been well-validated as an effective intraoperative glycemic protocol in cardiac surgeries during which glycemic environment and patient insulin sensitivity change considerably during surgery 13,14 . Moreover, a previous study has suggested that hypoglycemia risk is lower in the PoIIT than in other IITs, particularly when insulin sensitivity is impaired 16 . Nonetheless, it was unclear whether the PoIIT would be effective for use in liver transplantation. Some studies of patients undergoing liver surgeries have highlighted the necessity of developing better intraoperative glycemic strategies by reporting the difficulties in controlling BGC and negative clinical impacts of intraoperative hyperglycemia 4,5,30 . In a previous study comparing 60 liver transplant recipients with mean intraoperative BGC of <150 mg/dL and 124 recipients with mean intraoperative BGC of ≥150 mg/dL, higher intraoperative BGC was associated with post-transplant infections and mortality 5 . Another study of 680 recipients demonstrated that intraoperative hyperglycemia (BGC > 200 mg/dL) was associated with surgical site infection 4 . Despite the need for better intraoperative glycemic protocols for liver transplantation, there have been no studies evaluating the efficacy and/or safety of particular insulin infusion protocols. The mechanisms underlying the association of the use of PoIIT with improved post-transplant clinical outcomes might be attributable to decreased intraoperative BGCs and hyperglycemia. First, acute transient hyperglycemia or BGC fluctuation even during short periods disturbs the innate immune system by inhibiting neutrophil migration, phagocytosis, and complement function, stimulating inflammatory cytokines, and decreasing microvascular reactivity 7,8 . The deterioration of innate immune response may promote infection progress. Second, previous research demonstrated that only a transient period of acute hyperglycemia is necessary to aggravate ischemia reperfusion injury via oxidative stress and impaired tissue microcirculation 3,9 . A previous study of critically ill patients demonstrated that hepatocyte mitochondrial ultrastructure and function can be protected by preventing acute hyperglycemia with tight glycemic control 31 . Moreover, insulin may directly mitigate hepatocyte injury by serving as a scavenger of free radicals generated during the ischemia reperfusion process 32 . Third, decreased hepatic ischemia reperfusion injury could also benefit the bile duct 23 . Ischemia reperfusion injury to the microvasculature of the bile duct arteriolar plexus causes biliary epithelial cell damage; consequently, inflammatory cells penetrate between epithelial cells and basement membranes, and result in biliary stricture 33 , Moreover, insulin is an essential molecule for the regeneration of hepatic tissues including the bile duct 34,35 . A previous study in rats demonstrated that the degree of liver regeneration after hepatectomy was lower in subjects with decreased insulin sensitivity and suggested the benefit of insulin administration 36 . Another study demonstrated that early graft regeneration after LDLT was improved in relation to postoperative insulin administration 37 . Fourth, acute hyperglycemia-induced oxidative stress is systemic and can affect various tissues and organs in remote areas including the diaphragm 38 , which supports the association between the use of the PoIIT and decreased risk of prolonged mechanical ventilation. In addition, a direct anabolic effect of insulin on respiratory muscles may play a role: previous research demonstrated that insulin administration promotes skeletal muscle protein uptake and synthesis, improving skeletal muscle function 39 .
This study had several limitations. First, due to its retrospective nature, we are unable to establish a direct cause-and-effect relationship between the use of the PoIIT and improved post-transplant outcomes, although some of the underlying mechanisms can be hypothesized based on previous studies. In addition, we were unable to include more variables in the multivariable model due to the limited event number 40 . Thus, the association between strict glycemic management and post-transplant outcomes warrants further research with sufficient sample size. Nonetheless, we performed the Bonferroni correction and the Benjamini-Hochberg procedure to decrease the risk of type I error. Moreover, the sufficient sample size to assess the primary outcome was an advantage of the current study: the power of the 50% reduction of an assumed intraoperative hyperglycemia rate with the PoIIT was 83%. Second, there was a potential risk of selection bias because the use of the PoIIT or CoIT was determined at the discretion of attending anesthetists. In this regard, we tested and confirmed the absence of significant differences among 4 attending anesthetists who covered all recipients evaluated in the current   study regarding graft quantity and quality, donor factors, recipient factors, and preoperative laboratory findings (Supplementary Table S3). Third, only 25.8% of recipients with the PoIIT reached the target range (80-120 mg/ dL) before ICU arrival, during about 4 hours after the initiation of the PoIIT, most likely due to profound hepatic injury and severe insulin insensitivity. Thus, it remains unclear that how fast the target BGC range can be achieved and how safely the target range can be maintained once it is achieved. It should be noted that IIT-induced severe hypoglycemia and hypokalemia can cause poor outcomes 17,41 . In addition, identifying the optimal time window for the use of the PoIIT warrants further research. Nonetheless, the results of the current study suggest that transient use of PoIIT during the intraoperative postreperfusion phase and prevention of acute hyperglycemia for the critical time window may provide clinical benefits. The use of the PoIIT during the intraoperative postreperfusion phase was associated with decreases in intraoperative hyperglycemia and postoperative complications such as infections, prolonged mechanical ventilation, and biliary stricture. We observed no increases in the risks of hypoglycemia or hypokalemia in relation to the use of the PoIIT. Our findings suggest that the PoIIT effectively and safely controls acute hyperglycemic change after graft reperfusion and prevents hyperglycemia during LDLT, resulting in potential clinical benefits.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.