INTRODUCTION

Despite all the advances in anesthesia and surgery, operative blood loss during liver transplantation (LT) remains a major concern, associated with increased morbidity and reduced patient and graft survival⁹,¹⁹,²¹,²³,²⁶. Given the complexity of the procedure, the multifaceted origin, and the volume of bleeding, it remains urgent to identify high-risk patients and properly monitor hemostasis disorders in order to improve outcomes⁴,⁵,²².

Regardless of their limitations in cirrhotic patients and the growing use of cell-based viscoelastic tests, plasma-based coagulation assays (PBCAs) remain the most used tests to monitor hemostatic competence during LT³,²³,²⁴,²⁹. However, some studies have shown no correlation between PBCAs and bleeding parameters³,⁷,¹⁷. Although the international normalized ratio (INR) is an important marker of severity in acute and chronic liver diseases (CLDs), it does not seem to accurately reflect the complexity of hemostatic interactions in various clinical situations, such as CLD, trauma, hepatic surgery, and bleeding patients. Its ability to properly monitor perioperative hemostasis disorders has thus been questioned¹⁴,²⁷,²⁹.

In fact, the amount of bleeding may be largely influenced by other factors unrelated to the coagulation status — such as adhesions, portal hypertension, and surgical expertise — over which PBCAs have no predictive capacity¹⁴,²⁵. In this study, we aimed to evaluate whether INR and activated partial thromboplastin time (aPTT) can predict fresh frozen plasma (FFP) transfusion in patients undergoing LT.

METHODS

After approval by the ethics committee of the institution (number 083.08.10), the requirement for written informed consent was waived.

This is a retrospective cohort study of adult (>18 years old) cirrhotic patients who underwent deceased donor LT surgery over a 9-year period at Walter Cantídio Hospital, Universidade Federal do Ceará, Fortaleza (CE), Brazil.

Exclusion criteria

Exclusion criteria included cases of combined transplants, repeated LT procedures, non-cirrhotic recipients, transplants for fulminant liver failure, patients on anticoagulants or platelet inhibitors, and patients with incomplete medical records.

Surgical technique, anesthetic management, and intraoperative monitoring

Whole liver grafts were used from deceased brain-dead donors. Under general anesthesia, LT was performed using the piggyback technique. An arterial line was placed in the left radial artery, and a central line was inserted into the right internal jugular vein. The fluid regimen consisted of lactated Ringer’s solution (475 mL) mixed with 25 mL of human albumin 20%, infused at the anesthetist’s discretion based on hemodynamic and tissue perfusion variables. Blood loss was aspirated, anticoagulated immediately (30,000 IU of heparin in 1,000 mL saline), processed in a cell saver (Compact-Advanced^®, Dideco, Mirandola, Italy), and reinfused into the patient when recovered.

Laboratory and coagulation assays

Arterial samples were collected at four fixed time points (preoperatively within 6 h of surgery and at the end of each phase of the surgery: preanhepatic, anhepatic, and neohepatic) and stored in citrated tubes. The acid–base status and plasma levels of electrolytes, glucose, hemoglobin (Hb), and hematocrit (Ht), and standard coagulation tests (platelet count, fibrinogen by the Clauss method, aPTT, and INR) were monitored.

Platelet-poor plasma was obtained by centrifugation for 15 min at 3500 rpm and 25°C. Tests were performed on the Sysmex CA1500 (Sysmex Corporation^®, Kobe, Japan). Thromborel^® S and Actin^® FS (Siemens Healthcare Diagnostics^®, Marburg, Germany) were used to determine INR and aPTT, respectively. The International Sensitivity Index (ISI) values from Thromborel^® S were lot- and analyzer-specific, ranging from 0.93 to 1.12 across the study period. Coagulometer and reagents were used as per the manufacturer’s instructions. When blood samples were rendered incoagulable, the following values were assigned: INR=9.99 and aPTT=120 s. The turnaround time for the routinely measured tests was approximately 30 min.

Coagulation monitoring

In addition to PBCAs, periodic visual assessments of the surgical field and communication with the surgeon were performed by the attending anesthesiologist to assess the presence of microvascular bleeding¹².

Requirements for functional hemostasis

A prophylactic infusion of epsilon-aminocaproic acid (EACA), 20 mg/kg/h, was maintained intraoperatively. The esophageal temperature was kept above 36°C, and ionic calcium levels were maintained above 1.1 mEq/L. Sodium bicarbonate was administered to keep bicarbonate levels around 20 mEq/L. Transfusions of salvaged blood units and packed red blood cells (PRBCs) were guided by laboratory results to maintain Ht above 25%. Intraoperative blood salvage was utilized in all cases, except for those involving neoplasms or contamination of the abdominal cavity.

Diagnosis and management of hemostatic defects

Transfusion trigger: Regardless of the PBCA results, hemostatic blood components were not transfused unless microvascular bleeding was detected.

Diagnosis and management: Once microvascular bleeding was observed, the PBCA results were rank-ordered according to the deviation from values commonly considered adequate (INR <1.5, platelet count >50,000/mm³, and fibrinogen >100 mg/dL)¹. Coagulopathy associated with an isolated elevation of INR was managed with coagulation factor replacement using FFP. When the coagulopathy was associated with an isolated low platelet count or fibrinogen levels, platelet concentrates or cryoprecipitate, respectively, was administered. In cases of multiple coagulation abnormalities, the most severe one was treated first. If the coagulopathy persisted after the first transfusion, other components were sequentially administered according to the severity of laboratory abnormalities. If coagulation tests indicated simultaneous deficiencies in coagulation factors and fibrinogen, FFP was initially transfused alone, as it contains both deficient components. If the FFP transfusion did not resolve the coagulopathy, a supplemental transfusion of cryoprecipitate was administered.

Analysis of predictors of intraoperative fresh frozen plasma use during liver transplantation

Patients who received intraoperative FFP transfusions were categorized as the “FFP group” and those who did not as the “non-FFP group.” Demographic, surgical, laboratory, transfusion, and donor data were compared. Patients on the waiting list with hepatic tumors or other special conditions received exceptional points, leading to higher (corrected) Model for End-stage Liver Disease (MELD) scores⁴.

Postoperative analysis

Survival at 1 week, 1 month, and 1 year, intensive care unit (ICU) stay duration until death or discharge to the wards, and early (<6 h) reoperations for microvascular bleeding were compared between the groups.

Statistical data analysis

Descriptive statistical parameters were calculated for quantitative variables (mean and standard deviation [SD]) and qualitative variables (absolute and relative frequencies) for both groups. Baseline characteristics were compared using the χ² test for categorical variables and Student’s t-test for normally distributed variables. Multiple logistic regression analysis was conducted to identify independent preoperative risk factors for FFP transfusions during LT. The Mann-Whitney test was used to compare ICU stay durations between the groups. Survival rates for both groups were calculated using Kaplan-Meier analysis, and a log-rank test was applied to assess survival differences. Receiver operating characteristic (ROC) curves were generated to evaluate the performance of PBCAs in predicting FFP transfusions during the entire LT and at each phase of surgery. For ROC curves specific to each phase of surgery, only patients who had not yet received an FFP transfusion at the time of blood sampling were included. The area under the ROC curve (AUC) was calculated. Cutoff values associated with the best sensitivity and specificity were identified. Diagnostic accuracy was categorized as fail (AUC≤0.6), poor (0.6<AUC≤0.7), fair (0.7<AUC≤0.8), good (0.8<AUC≤0.9), or excellent (0.9<AUC≤1.0). The significance level was set at p<0.05. Analyses were performed using Microsoft Excel 2003 (Microsoft Inc., Seattle, WA, USA), SPSS for Windows 15.0 (SPSS Inc., Chicago, IL, USA), and SAS 8.0 statistical software (SAS Institute Inc., Cary, NC, USA).

RESULTS

Patients’ characteristics

Of the 507 patients initially included in the study, 176 met the exclusion criteria, and 34 were excluded from the analysis due to incomplete data. Consequently, data from 297 patients were analyzed and they were divided into FFP group (n=88) and non-FFP group (n=209) (Figure 1).

Demographic, surgical, laboratory, transfusion, and donor data are shown in Table 1. Patients in the FFP group were younger and had significantly higher uncorrected MELD scores, lower aspartate aminotransferase values, a lower incidence of hepatocellular carcinoma, and worse preoperative values of Hb, Ht, and fibrinogen. Additionally, the FFP group had lower Hb levels by the end of surgery. This group also received more units of platelets and PRBCs per patient and showed statistically significantly higher transfusion rates for all blood products compared to the non-FFP group.

There was no statistically significant difference between the groups regarding ICU stay or the incidence of early reoperations for microvascular bleeding. Kaplan-Meier survival estimates coupled with a log-rank test showed a non-significant trend toward better long-term survival among patients who did not receive FFP transfusions (Figure 2 and Table 2).

Multivariate analysis to predict the use of fresh frozen plasma during liver transplantation

Although many variables exhibited some influence on the use of FFP during LT in the univariate analysis (Table 1), multivariate logistic regression analysis (Table 3) showed that preoperative Ht (odds ratio [OR]=0.90, p<0.001), preoperative fibrinogen (OR=0.99, p<0.001), and absence of hepatocellular carcinoma (OR=3.57, p=0.004) were the only statistically independent risk factors for FFP transfusions. Based on these results, it can be concluded that a 1% increase in preoperative Ht results in a 10% reduction in the chance of using FFP transfusions. Additionally, each 1 mg/dL increase in preoperative fibrinogen decreases the chance of receiving FFP transfusions by 1%. Finally, cirrhotic patients without hepatocellular carcinoma have a 3.57 times higher chance of requiring FFP transfusions compared to those prioritized for transplantation due to this neoplasm.

Plasma-based coagulation assays to predict the need for fresh frozen plasma transfusions during liver transplantation

Preoperative PBCAs, even at their best cutoffs, had poor accuracy in predicting the use of FFP during LT (AUC, sensitivity, and specificity values barely exceeded 60%). When evaluated phase-by-phase perioperatively, PBCAs also demonstrated poor accuracy (AUC, sensitivity, and specificity values did not exceed 70%) (Figure 3 and Table 4).

Additionally, Table 5 summarizes the cutoff values associated with specificity as close as possible to 90%, identified from the ROC curves in Figure 3. The 90%-specificity cutoff values identified for samples collected in the preanhepatic, anhepatic, and neohepatic phases of LT were 2.14, 2.62, and 3.52 for INR and 50.5, 80.2, and 119.5 for aPTT, respectively.

DISCUSSION

Transfusion practices in LT are characterized by wide variability, and differences in how coagulation abnormalities are managed with FFP and platelets may be even more significant than with the use of PRBCs²⁰. Similar to other groups, our team employs a “wait-and-see” approach of rescue transfusion therapy, instead of prophylactic or preventive therapy, combined with maintaining hemostatic prerequisites, utilizing antifibrinolytics, using a cell saver, and relying on a skilled surgical team. Adopting such restrictive practices may help reduce FFP transfusion rates, as these may worsen LT perioperative results¹¹,¹⁸,¹⁹.

Our overall reoperation rate for bleeding (9.1%) was in accordance with previous descriptions, although lower rates have been reported. These lower rates may reflect more effective transfusion protocols¹¹,¹⁸. Long-term survival was longer in patients who did not receive FFP transfusions, although there was no significant difference between the groups in terms of survival and other postoperative outcomes. The adverse effects of FFP transfusions on perioperative results are not universal and seem to be dose-related. Therefore, our sample size may have been too small to detect such adverse effects, given the low transfusion rates presented. Additionally, potential differences in patient populations and criteria for group allocation may have contributed to these results.

In the FFP group, the transfusion of all blood products was more prevalent, and more units were transfused per patient, except for cryoprecipitate. These findings may reflect a more fragile hemostatic status or greater exposure to intraoperative challenges to hemostasis. Patients in this group had a lower incidence of hepatocellular carcinoma and higher uncorrected MELD scores⁴. The severity of liver disease is a well-known factor associated with transfusions during LT. Additionally, patients in this group had lower preoperative Hb and fibrinogen levels, both of which play critical roles in functional hemostasis¹⁵,³⁰. The FFP group may have had a more fragile baseline hemostatic status, coupled with greater exposure to negative influences on hemostasis during LT, such as larger blood loss and dilution, which were not evaluated in this study²⁰.

PBCA results, regardless of the sampling time, showed poor performance in predicting FFP transfusions during LT (Table 5). This inaccuracy has been previously described, although some of these studies used coagulation scores rather than PBCA results as predictors, and most evaluated blood loss or survival as the predicted outcomes¹⁰,¹¹. Additionally, most previous studies restricted their analysis to preoperative coagulation tests and did not evaluate intraoperative measures. Furthermore, some of them employed conventional PBCA cutoffs, which were originally defined for non-cirrhotic patients.

Cirrhosis is commonly associated with clinical bleeding and abnormalities in PBCAs, making it tempting to infer a cause-and-effect relationship between these observations in both clinical and surgical scenarios²⁸,³¹. However, prothrombin time (PT)/aPTT has several limitations in correlating with microvascular bleeding²⁹. Despite these limitations, PBCAs are still the most commonly used tools for implementing goal-directed transfusion therapy. By identifying which part of coagulation is altered, selecting the appropriate blood component, and confirming therapeutic effectiveness, PBCAs may help reduce the use of blood products compared to relying solely on the visual assessment of the surgical field⁸,³².

The multivariate analysis supported the poor predictive performance of PBCAs, demonstrating that the only independent predictors of FFP transfusions during LT were preoperative Ht, preoperative fibrinogen levels, and the absence of hepatocellular carcinoma. According to this and other studies, these predictors can identify a group of LT patients who are more severely ill and more likely to receive transfusions⁸,¹⁶,³².

Global coagulation tests have shown good negative predictive values for bleeding in the early postoperative period. However, their isolated use to predict bleeding has consistently shown poor positive predictive values. To overcome this limitation, the authors have recommended correlating pathological results with clinically significant bleeding. This highlights the importance of clinical evaluation of microvascular bleeding over isolated coagulation test results, especially when triggering transfusions. As a result, microvascular bleeding has become increasingly accepted as a formal indication for FFP transfusions, to the detriment of relying solely on PBCA results, and has been included as a key step in hemostasis protocols²,⁶. This approach adds a layer of safety against laboratory errors and avoids potentially unnecessary transfusions, although a certain degree of subjectivity is inherent in most clinical transfusion studies⁶,¹⁰,¹³.

Additionally, coagulation assays have become important in guiding when not to transfuse FFP. Guidelines from the American Society of Anesthesiology regarding perioperative blood transfusions state that the transfusion of FFP is not indicated if PT, INR, and aPTT are normal². In line with this recommendation, we evaluated critical PBCA values in patients who did not receive FFP transfusions. For this purpose, we arbitrarily chose cutoff levels that maximized specificity (90%), with the known trade-off of reduced sensitivity and overall prediction accuracy. In our scenario, for each phase of surgery, we found that 90% of patients who did not receive FFP transfusions had results below the calculated cutoffs. Clinically, this suggests that only 10% of patients who did not receive FFP transfusions reached the cutoffs shown in Table 5. Thus, patients with a dry surgical field who exceed these 90%-specificity cutoffs deserve frequent and meticulous monitoring to identify coagulopathy as it develops.

Limitations of the study

Our study has some limitations. First, it is a single-center, non-randomized retrospective cohort study. However, controlled studies in patients with significant blood loss are difficult to conduct, as it would be unethical to include a control group that receives no hemostatic therapy. Second, the multifactorial etiology of coagulopathy during LT complicates the task of isolating all factors influencing PBCA abnormalities. Third, similar to other transfusion decision-making studies, the diagnosis of microvascular bleeding may involve some degree of subjectivity, and time lags may have hindered the real-time assessment of coagulation status.

Despite these limitations, we emphasize that this is a large investigation to assess the correlation between PBCA and FFP transfusions during LT, potential cutoffs, and their accuracy in such prediction. The proposal of high-specificity cutoff values for PBCAs in cirrhotic patients undergoing LT is also intriguing.

CONCLUSIONS

Our results showed that the significant predictors for FFP transfusions were preoperative Ht, preoperative fibrinogen, and the absence of hepatocellular carcinoma. FFP transfusions did not influence survival, ICU stay, or the incidence of reoperations. PBCAs, regardless of the cutoffs adopted or the time of sampling during LT, had a poor correlation with intraoperative FFP transfusions. Therefore, in LT centers where global hemostasis tests are unavailable, PBCA values should not be used to trigger FFP transfusions but may be used as a reference for when not to transfuse FFP.

REFERENCES

Edited by

Publication Dates

History