Association Between Red Blood Cell Distribution Width and Intraoperative Transfusion in Patients Undergoing Living Donor Liver Transplantation: a Retrospective Single-center Cohort Study


 Background: Living donor liver transplantation (LDLT) has been associated with massive bleeding and increased blood transfusion requirements. However, information on indicators predicting bleeding and transfusion in LDLT is limited. Recent studies have reported a correlation between red cell distribution width (RDW) and bleeding risk. This study investigates the association between RDW and intraoperative blood transfusion in patients receiving LDLT.Methods: This retrospective study analyzed 2548 patients who underwent LDLT between January 2010 and October 2019. The patients were divided into four groups based on preoperative RDW quartiles: Q1 (<13.9), Q2 (13.9–15.2), Q3 (15.2–17.4), and Q4 (≥17.4), and transfusion requirement and surgical outcomes were assessed. The risk factors for intraoperative transfusion were evaluated by multivariate regression analysis. The predictive power of RDW was assessed through receiver operating characteristic (ROC) and integrated discrimination improvement (IDI) analysis.Results: There were significant differences in incidence of intraoperative transfusion according to preoperative RDW quartile (Q1 vs Q2 vs Q3 vs Q4: 47.3% vs. 78.1% vs. 91.8% vs. 96.2%, P<0.001). Q4 had poor surgical outcomes, such as acute kidney injury (adjusted odds ratio [OR]: 1.91, 95% CI: 1.44–2.54, P<0.001). In the multivariate logistic analysis, RDW, age, sex, diabetes mellitus, coronary artery disease, model for end-stage liver disease scores, and total ischemic time were risk factors for transfusion. In ROC and IDI analysis, RDW had predictive power for intraoperative transfusion (P=0.023 in IDI).Conclusions: Preoperative RDW is a potential predictor of intraoperative transfusion and postoperative acute kidney injury in patients receiving LDLT.


Introduction
Liver transplantation (LT) is one of the most complex surgeries for patients with end-stage liver disease and represents a great challenge for surgeons and anesthesiologists. [1] Historically, LT has been associated with massive intraoperative bleeding and large transfusions. [2] Although recent advances in surgical techniques, anesthetic skills, and perioperative management have reduced intraoperative bleeding, [3,4] LT is still associated with high bleeding risk and increased blood transfusion requirements. [5] Despite numerous efforts to nd predictors of intraoperative bleeding and transfusion in LT, only limited information is available to date. [6,7] The red blood cell distribution width (RDW) is a measure of the variation in the size of red blood cells in a blood sample and part of a standard complete blood count.
[8] RDW is a simple and objective indicator of patient survival and complications even in acute and chronic diseases such as heart failure, stroke, and various cancers. [9][10][11][12][13] Higher RDW indicates more than expected variations in red blood cell size. Several recent studies have reported that RDW is correlated with bleeding risk and transfusion for some diseases and procedures. [14][15][16] However, there are few studies on the association between preoperative RDW and intraoperative transfusion in LT. Therefore, in this study, we evaluated the association between preoperative RDW and intraoperative blood transfusion in patients who underwent living donor liver transplantation (LDLT).

Study design and patient population
The institutional review board of the Asan Medical Center (Protocol No. 2021 − 0243) approved this retrospective study, and because of the retrospective nature of our study, the requirement for written informed consent was waived. We analyzed all the patients who underwent LDLT for end-stage liver disease between January 2010 and October 2019. The following patients were excluded: patients aged < 18 or ≥ 80 years; patients with hematologic disorders; patients taking anticoagulants, such as aspirin, warfarin, or antiplatelet agents; patients undergoing emergency surgery; and patients with incomplete or missing data.

Anesthetic technique
Hemodynamic monitoring and general anesthesia were performed according to our institutional standards. [17] Brie y, anesthesia was maintained using des urane or sevo urane, a mixture of 50% O 2 and 50% air, and continuous intravenous infusion of fentanyl. Invasive arterial-venous pressure monitoring was performed by radial and femoral arterial catheterization. Central venous pressure monitoring was routinely performed, and a pulmonary arterial catheter was inserted to monitor hemodynamic variables via a Vigilance monitor (Vigilance II, Edwards Lifesciences LLC). During anesthesia, uids or vasopressors were administered by an anesthesiologist based on the patient's mean blood pressure (MBP) and hemodynamics. In cases of low systemic vascular resistance, MBP was maintained by the continuous infusion of inotropic agents, such as norepinephrine, vasopressin, or terlipressin. Plasma and 5% or 20% albumin were administered during anesthesia. If the plasma hemoglobin level was < 8 g/dL or massive bleeding was expected due to intraoperative bleeding, packed red blood cells (PRBCs) were transfused, and hemoglobin level was maintained at > 10 g/dL in patients with ischemic heart disease.

Data collection and outcome assessment
Patient characteristics and perioperative variables were collected using the medical record system of our institution. Patient characteristics included age, sex, body mass index (BMI), diabetes mellitus (DM), hypertension (HTN), coronary artery disease (CAD), and model for end-stage liver disease (MELD) scores.
Variables related to patients' tumor etiology included hepatitis B virus (HBV), hepatitis C virus (HCV), alcoholic liver cirrhosis, and combined hepatocellular carcinoma (HCC). Donor-related variables included age, sex, and BMI.
Preoperative laboratory values included levels of white blood cells (WBCs), hemoglobin, platelet, brinogen, albumin, aspartate transaminase (AST), alanine transaminase (ALT), total bilirubin, sodium, serum creatinine (sCr), international normalized ratio (INR), and estimated glomerular ltration rate (eGFR). sCr levels were checked daily from postoperative day 1 to day 7 to con rm acute kidney injury (AKI). Data on preoperative neutrophil-lymphocyte ratio (NLR), platelet-lymphocyte ratio (PLR), prognostic nutritional index (PNI), C-reactive protein/albumin ratio, and RDW were also collected. NLR was de ned as the ratio of absolute neutrophil count to absolute lymphocyte count, and PLR was de ned as the ratio of absolute platelet count to absolute lymphocyte count, respectively. PNI was calculated using serum albumin level and total lymphocyte count. [18] Intraoperative variables included operation time, total ischemic time, post-reperfusion syndrome, total volume of uid administered, and urine output. Data on intraoperative and postoperative transfusion, massive transfusion, hospital stay, postoperative AKI, intensive care unit (ICU) stay, 1-year graft failure, 1year survival, and overall survival were collected. Massive transfusion was de ned as ≥ 10 units of PRBCs within 24 hours, ≥ 4 units of PRBCs within 1 hour, or replacement of > 50% of the total blood volume by blood products within 3 hours. [19] Total blood volume for adults was based on Gilcher's rule of ve for blood volume. [19] Primary and secondary endpoints The primary endpoints were to compare intraoperative transfusion according to quartiles of preoperative RDW. The secondary endpoints were to analyzed risk factors associated with intraoperative transfusion and evaluate surgical outcomes, such as postoperative AKI. Postoperative AKI was de ned by the Kidney Disease Improving Global Outcomes classi cation: sCr level increased by at least 1.5 times of baseline value within 7 days after surgery or an sCr level increase of 0.3 mg/dL within 48 hours after surgery. [20] In addition, we evaluated the predictive power of RDW for intraoperative blood transfusion through receiver operating characteristic (ROC) and integrated discrimination improvement (IDI) analyses.

Statistical analysis
Data are described as means ± standard deviations, medians (interquartile ranges), or numbers

Results
Of 2658 patients who underwent LDLT, 110 patients were excluded according to the exclusion criteria. In total, 2,548 patients were included and divided into four groups by RDW quartile: Q1 (< 13.9, n = 609), Q2  Table 1. The groups in higher quartiles were younger (P = 0.001), more likely to be female (P < 0.001), more likely to have a history of DM (P = 0.007), and less likely to have a history of HTN (P < 0.001). Moreover, these groups had higher MELD scores (P < 0.001), a higher incidence of alcoholic liver cirrhosis, lower BMI (P = 0.003), and a lower incidence of HBV, HCV, and HCC. There was no signi cant association between preoperative RDW and donor-related variables, such as age (P = 0.621), sex (P = 0.755), and BMI (P = 0.773) ( Table 1). Values are expressed as means ± standard deviations, medians (interquartile ranges), or absolute numbers (percentages).

Primary endpoints
The incidence of intraoperative transfusion according to RDW quartile is shown in Table 1   Values are expressed as means ± standard deviations, medians (interquartile ranges), or absolute numbers (percentages).
According to the ROC curve analysis for intraoperative transfusion, the addition of hemoglobin levels to model 1 (MELD scores) improved the area under the curve (AUC: 0.901, P < 0.001) ( Table 4 and Fig. 1). The addition of RDW to model 2 (MELD scores + hemoglobin) showed no signi cant improvement in AUC (P = 0.394) ( Table 4 and Fig. 1). However, an IDI of 0.003 (P = 0.023) indicated that adding RDW to model 2 signi cantly improved the model's predictive power in terms of a positive difference between increased average sensitivity and any potential increase in average 1-speci city (Table 4). Figure 2 shows the Kaplan-Meier curve according to preoperative RDW quartiles (log-rank test; P = 0.021). One-year survival was signi cantly different between RDW Q4 and the other quartiles. Values are expressed as means ± standard deviations, medians (interquartile range), or absolute numbers (percentages).

Discussion
In the present study, we found a signi cant difference in the incidence of intraoperative transfusion between RDW quartiles (47.3%, 78.1%, 91.8%, and 96.2% in Q1, Q2, Q3, and Q4, respectively) in patients who underwent LDLT. There was a positive correlation between RDW quartile, massive transfusions, postoperative transfusions, and postoperative AKI. The incidence of intraoperative transfusions and postoperative AKI increased signi cantly, even after adjusting for potential confounding factors. In addition, IDI analysis showed that RDW, MELD scores, and hemoglobin levels had better predictive power for intraoperative transfusion.
Despite the recent advancements in surgical and anesthetic techniques, intraoperative transfusion and massive bleeding remain major problems in LDLT. Massive bleeding is a predictor of poor surgical outcomes [21] and is linked to increased morbidity and mortality. [22] Blood transfusion is correlated with poor graft survival, infection, and increased complications such as renal failure, bacterial sepsis, and allergic reactions. [23,24] The preoperative identi cation of factors that can predict the need for blood transfusion and the risk of massive bleeding during LDLT can improve surgical outcomes.
Previous studies have analyzed predictors of bleeding and transfusion during LT. [6,7,21,[25][26][27][28][29] Araújo and colleagues reported that preoperative INR, hemoglobin levels, age, and liver pathology were signi cant predictors of intraoperative blood transfusion in orthotopic liver transplantation (OLT); however, the predictive power of these factors was low. [7] Yuasa and colleagues reported that age, weight, CRP, hematocrit, and total bilirubin were preoperative risk factors for massive blood loss in LDLT. [21] Other studies found a low correlation between bleeding and INR in patients with chronic liver disease. [29,30] The association between the severity of liver disease (Child-Turcotte-Pugh and MELD scores) and perioperative bleeding is controversial. [4,31] Moreover, there is limited information on the risk factors for transfusion and bleeding in LDLT. Our study is clinically signi cant as the rst to investigate the link between preoperative RDW and intraoperative transfusion in LDLT patients.
In our study, the incidence of perioperative blood transfusion was signi cantly associated with preoperative RDW. Additionally, the inclusion of RDW in the model consisting of MELD scores and hemoglobin levels, which are known transfusion risk factors, improved the predictive power for intraoperative blood transfusion, suggesting a strong association between preoperative RDW and perioperative bleeding in LDLT.
In the multivariate logistic regression analysis, preoperative RDW, preoperative hemoglobin level > 8 g/dL, age, female sex, DM, CAD, MELD scores, and total ischemic time were signi cantly associated with intraoperative transfusion. Preoperative hemoglobin level has been reported as a risk factor for blood transfusion in many previous studies. [7,26,28] Transfusions occur when hemoglobin levels are low, and in our study, blood transfusions were also based on hemoglobin less than 8; therefore, hemoglobin is one of the strongest triggers of the need for transfusions. Age and MELD scores are risk factors for blood transfusion in LT, which may be due to the patient's systemic condition, comorbidities, and disease severity. [7,21,25,32] A recent study reported that female patients had a higher risk of intraoperative blood transfusion than male patients because of sex differences in preoperative hematocrit and blood volume. [33] DM is related to coagulopathy, microvasculopathy, and a hypercoagulable state, [34,35] which may increase the risk of intraoperative blood transfusions. CAD is associated with anticoagulants and, therefore, increases the risk of surgical bleeding.[36, 37] However, patients taking anticoagulants were excluded from our analysis. The relationship between CAD and transfusion is thought to be the result of a high standard of transfusion in CAD patients (hemoglobin level > 10 g/dL). Total or warm ischemic time is a risk factor for blood loss in LT patients, which might be associated with postreperfusion syndrome, a possible cause of increased transfusion requirements. [25,38] In ammatory reactions caused by various cytokines can trigger an abnormal clotting system, resulting in hypercoagulation, increasing intraoperative bleeding risk. [39,40] Decreased liver function associated with nutritional de ciencies and in ammatory status may exacerbate disease severity [41] and increase the risk of bleeding. [42] Recent studies have shown that elevated RDW may be associated with bleeding risk.
[ [14][15][16] The mechanism by which RDW increases is not yet clear, but it has been reported to be associated with anemia, in ammatory responses and oxidative stress. [43] Increased RDW is a sign of a nutritional de cit, such as a de ciency of iron, folic acid, or vitamins B-12, which may indicate macrocytic anemia and may increase blood transfusion requirements during surgery. [44] The inhibition of erythrocyte maturation by in ammatory cytokines can increase the risk of blood transfusions by causing abnormal erythropoietin function and anisocytosis, which may be associated with anemia and thrombotic conditions. [45] In our study, patients with higher RDW had poor surgical outcomes, such as prolonged hospital and ICU stay, AKI, increased risk of 1-year graft failure, and lower 1-year and overall survival. Postoperative AKI was signi cantly associated with RDW, even after adjusting for potentially confounding variables.
Consistent with previous studies, our results suggest that high RDW is signi cantly associated with poor surgical prognosis. [10,12] This study has some limitations. First, the major limitations of this study are those inherent to its retrospective nature. Thus, the possibility of reporting undocumented factors and a potential bias associated with patient selection and recall existed. However, we tried to reduce the impact of confounding factors by adjusting for variables that could affect the outcome. Second, our patients were admitted to one of the largest LT center in the world, and the LT team had extensive experience, performing more than 300 LDLTs per year since 2010.
[46] Therefore, our results may differ from those of studies conducted in other institutions or from multicenter studies and may be di cult to apply to patients undergoing OLT. Third, there are no studies on the accurate validation of preoperative RDW cutoff values that predict surgical outcomes. Therefore, further well-designed research on this topic is needed.

Conclusion
We found that high preoperative RDW was strongly associated with intraoperative blood transfusion and postoperative AKI in patients undergoing LDLT. These results indicate that preoperative RDW can be a useful predictor of intraoperative blood transfusion and postoperative AKI in LDLT recipients.

Declarations
Ethics approval and consent to participate The need for informed consent from individual patients was waived owing to the retrospective nature of the study.

Consent for publication
Not applicable  Kaplan-Meier survival curve for 1-year survival according to quartiles of preoperative red cell distribution width. The survival rate was signi cantly poor in patients with preoperative RDW level of quartile 4 than in those with preoperative RDW level of quartile 1, 2, and 3 (log-rank test; p=0.021).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.