Abstract
Background: It is of great clinical concern to preoperatively predict the need for blood transfusions during hepatectomy for hepatocellular carcinoma (HCC). Patients and Methods: A total of 168 consecutive patients undergoing elective hepatectomy for HCC were retrospectively reviewed. We investigated preoperative factors potentially influencing intraoperative blood transfusion and established a predictive scoring system for intraoperative blood transfusion. Results: Thirty-eight patients (22.6%) received red cell blood transfusion during surgery. A preoperative predicting scoring system for blood transfusion was constructed using the following four factors: platelet count <10×104/mm3 (2 points), α-fetoprotein ≥80 ng/ml (1 point), tumor size ≥4.0 cm (1 point), and major hepatectomy (1 point). The nomogram showed an area under the curve (AUC) of 0.760. This scoring system was highly predictive for blood transfusion (AUC=0.758). When the score was 0 points, the incidence of intraoperative blood transfusion was 3%. The rate increased to 10% and 38% when the score was 1 and 2 points, respectively, and reached 45% when the score was 3 points or more. Conclusion: This predictive scoring system would be useful for preoperatively assessing the need for intraoperative blood transfusions during hepatectomy for HCC.
Despite recent advances in surgical techniques and newly-developed hemostatic devices, hepatectomy remains one of the most hemorrhagic procedures in abdominal surgery (1, 2); as patients sometimes develop large amounts of blood loss during surgery and consequently require blood transfusions. Most patients with hepatocellular carcinoma (HCC) undergoing hepatectomy suffer from hepatitis with cirrhosis. Therefore they sometimes experienced unexpectedly large amounts of blood loss requiring for blood transfusions, even when they undergo standard hepatectomy.
Recently, autologous blood donation has been recommended for patients requiring intraoperative blood transfusions because homologous transfusions carry risks of infectious or immunological complications (3, 4), in addition to conferring a poor prognosis (5). On the other hand, autologous blood transfusions are associated with problems such as bacterial contamination (6), heart strain (7), or non-cost-effectiveness (8). From the viewpoint of the efficient and effective usage of autologous transfusions, autologous blood donation should be considered in patients at a high risk of requiring intraoperative blood transfusions.
Although some of the risk factors indicating the need for intraoperative blood transfusion during hepatectomy for HCC have been clarified (9-12), it remains difficult to preoperatively predict such a need during hepatectomy for HCC. It is important to clarify the preoperative predictors influencing this need and to preoperatively discriminate patients requiring intraoperative blood transfusions for the efficient use of autologous blood donation.
This retrospective study was designed to determine the preoperative predictive factors of intraoperative blood transfusions during hepatectomy for HCC. Moreover, this study was undertaken to develop a predictive scoring system for blood transfusions during hepatectomy using these factors.
Patients and Methods
One hundred and sixty-eight consecutive patients underwent hepatectomy for HCC between January 2001 and December 2010. All patients were retrospectively reviewed. There were 128 males and 40 females. The mean (±SD) age was 64.7 (±10.3) years (range 20 to 85 years). The underlying liver diseases included cirrhosis in 70 patients (41.6%) and non-cirrhosis in 98 patients (58.4%). According to Child's classification modified by Pugh et al. (13), 163 patients (97.0%) were grouped in class A, and five patients (3.0%) were in class B. Twenty-seven patients were seropositive for hepatitis B surface antigen and 96 patients were seropositive for antibody to hepatitis C. The mean (+SD) tumor diameter was 4.1 (+3.1) cm (range 0.4 to 24.0 cm). 130 patients (77.3%) had a solitary tumor, and 38 patients (22.7%) had multiple tumors. In patients bearing more than one tumor, the largest tumor was chosen as the representative tumor.
The type of hepatectomy and the tumor location were defined according to the classification of Couinaud's definition (14) of liver segmentation. Major hepatectomy included right trisectionectomy (resection of segments I, IV, V, VI, VII, and VIII), left trisectionectomy (resection of segments I, II, III, IV, V, and VIII), right hepatectomy (resection of segments V, VI, VII, and VIII), left hepatectomy (resection of segments II, III, and IV), central bisectionectomy (resection of segments IV, V, and VIII), right anterior sectionectomy (resection of segments V and VIII), right posterior sectionectomy (resection of segments VI, and VII), and left medial sectionectomy (resection of segment IV). Minor hepatectomy included left lateral sectionectomy (resection of segments II and III), segmentectomy, and limited resection. In patients undergoing multiple resections, the most extensive procedure was considered to be the main type of hepatectomy. The areas of resected liver were as follows: less than one section in 77 patients, one or more sections in 50 patients, and two or more sections in 41 patients. Preoperative transarterial chemoembolization (TACE) was performed in 19 patients.
The indications for hepatectomy and the types of operative procedures were usually determined based on the patient's liver function, primarily assessed using the Makuuchi Criteria, which comprise preoperative measurements of ascites, the serum bilirubin level, and the indocyanine green retention rate at 15 min (ICGR15) (15). Anatomical resection required access to the major hepatic veins. Blood transfusion refers to transfusions of packed red blood cells in the present study. The indications for blood transfusion were assessed by the anesthesiologists and surgeons. Unstable vital signs or a hemoglobin value of less than 8.0 mg/dl were used as indicators for transfusion.
Univariate analyses of risk factors for intraoperative blood transfusion were performed using the categorized variables. The cut-off values of continuous variables for differentiation between the group with blood transfusion and the group without blood transfusion were determined by receiver operating characteristics (ROC) analysis. Fisher's exact test or the Chi-squared test was used to compare categorical data between groups.
A multivariate analysis was performed using a logistic regression analysis with a forward stepwise selection model. All factors whose p-value equaled 0.05 or less in the univariate analysis were included in the multivariate model. A p-value of 0.05 or less was considered to be statistically significant. All patients were then assigned an estimated probability calculated based on the logistic regression model. ROC curves and the corresponding area under the curve (AUC) were used to evaluate how the predictive model performed on the test data (16). The AUC was substituted with Harrell's concordance index, which was used in this analysis (17). A value of 1.0 indicates a perfect prediction, whereas a value of 0.5 is equivalent to prediction by the toss of a coin. The AUC is a measure of the diagnostic accuracy such that values between 0.5 and 0.7 indicate low accuracy, values between 0.7 and 0.9 indicate moderate accuracy, and values >0.9 indicate high accuracy (18).
The results of the univariate analysis of factors affecting intraoperative blood transfusion.
Results of logistic regression analysis for intraoperative blood transfusion.
To make the predictive scoring system easily and readily available in the clinical setting, we further developed a score for intraoperative blood transfusion using standardized variables based on the regression coefficient of the logistic regression model. The parameter estimates derived from the regression coefficient were then standardized and rounded to the nearest integer to produce the coefficients ultimately used in the score. Stratification of the patients was performed based on this score. Considering the score to be a continuous variable, its predictive accuracy was quantified with the AUC, and the AUC of the predictive scoring system was compared to that of the estimated probability calculated based on the logistic regression model.
The statistical analyses were performed using a statistical analysis software package (SPSSII 11.0, SPSS Inc., Chicago, IL, USA).
Results
Thirty-eight patients (22.6%) received red cell blood transfusions during surgery. The univariate analysis identified tumor size ≥4.0 cm (p=0.002), platelet count <10.0×104/mm3 (p=0.028), major hepatectomy (p=0.004) and an α-fetoprotein level ≥80 ng/ml (p=0.007) as risk factors for intraoperative blood transfusion during hepatectomy, as shown in Table I.
A logistic regression analysis of these four variables identified preoperative tumor size ≥4.0 cm (p=0.045), platelet count <10.0×104/mm3 (p=0.002), major hepatectomy (p=0.041) and an α-fetoprotein level ≥80 ng/ml (p=0.032) to be independently-correlated risk factors of intraoperative blood transfusion, as shown in Table II.
The logistic regression model provided the estimated probability of intraoperative blood transfusion during hepatectomy. This probability was equal to y=1/(1+e−Z), where z=−2.833+1.732 × (0, platelet count ≥10×104/mm3; 1, platelet count <10×104/mm3) + 1.074 × (0, minor hepatectomy; 1, major hepatectomy) + 0.943 × (0, tumor size <4.0 cm; 1, tumor size ≥4.0 cm) + 0.914 × (0, an α-fetoprotein level <80 ng/ml; 1, an α-fetoprotein level ≥80 ng/ml), and e is the mathematical constant and base value of the natural logarithm. The associated ROC curve (Figure 1a) showed that the AUC indicated a predictive accuracy of 0.760.
We further developed a predictive score for intraoperative blood transfusion using each standardized variable based on the regression coefficient of the logistic regression model. A patient receives 2 points for platelet count <10×104/mm3, and 1 point for α-fetoprotein ≥80 ng/ml, tumor size ≥4.0 cm, and major hepatectomy. The scores for each patient ranged from 0 up to 5 points. Considering this score to be a continuous variable, the ROC curve (Figure 1b) demonstrated that the AUC of this score was 0.758, which was slightly lower than that of the estimated probability calculated based on the logistic regression model. In addition, this predictive score exhibited moderate accuracy as a discriminating test. The percentage of patients who actually received intraoperative blood transfusions during surgery for each score (Figure 2) showed that the incidence of intraoperative blood transfusion gradually increased as the score increased. When the score was 0 points, the incidence of intraoperative blood transfusion was 3%. The rate increased to 10% and 38% when the score was 1 and 2 points, respectively, and reached 45% when the score was more than or equal to 3 points or more.
Discussion
In the present study, we developed a predictive scoring system to assess the need for intraoperative blood transfusion during hepatectomy using four independent preoperative factors. When the score is 0 points, the rate of intraoperative blood transfusion is 3%, however, the rate increases to 45% when the score is ≥3 points or more. This score is easily and quickly available in the clinical setting, and surgeons can determined the predictive risk for intraoperative blood transfusion preoperatively for individual patients. This predictive score exhibits good discriminator's quality (AUC=0.758), and it is classified as possessing moderate accuracy according to Swets' classification (18).
Receiver operating characteristic curve for the estimated probability of intraoperative blood transfusion calculated based on the logistic regression model (a) and on the preoperative score (b). AUC: Area under the curve.
In the present study, the preoperative predicting scoring system for intraoperative blood transfusion was constructed using platelet count <10×104/mm3, major hepatectomy, α-fetoprotein ≥80 ng/ml and tumor size ≥4.0 cm. A low preoperative platelet count is well-known to be related to portal hypertension and results in hyper-splenism and hepatic fibrosis (19, 20). Platelet count is associated with a cirrhotic status in relation to hepatic disease. Several reports have mentioned that thrombocytopenia has a high prevalence among patients with advanced liver disease, and bleeding related to invasive procedures occurs most frequently in patients with severe thrombocytopenia (21). Splenectomy and transcatheter partial splenic artery embolization (PSE) has been reported to be effective in treating portal hypertension and thrombocytopenia in recent years. In particular, recent advances in laparoscopic approach minimize the surgical stress of the splenectomy. To prevent the perioperative blood transfusion, splenectomy or PSE prior to hepatic resection may be a safe procedure for the patient with HCC who has liver cirrhosis and secondary hypersplenism (22-24).
Major hepatectomy has already been reported in other studies as a risk factor for a large amount of blood loss or the need for intraoperative transfusion during hepatectomy (2, 9, 10, 12, 25). Major hepatectomy requires access to and encirclement of the hepatic veins, which results in a large amount of blood loss. Moreover, major hepatectomy usually involves larger transection surface than minor hepatectomy. In recent years, inferior vena cava clamping (26) or intraoperative blood salvage (27) were introduced to reduce the central venous pressure during hepatecotmy. In patients with these risk factors, achieving outflow control before performing liver transection is desirable.
Percentage of patients requiring intraoperative blood transfusion at each score.
Similar to the results of some previous reports, the tumor size independently affected the need for blood transfusion under multivariate analysis (2, 10, 28). On the other hand, a high α-fetoprotein level is also an independent risk factor of intraoperative blood transfusion. A high α-fetoprotein level has been reported to be related to the presence of multiple tumors, and portal venous and hepatic venous invasion (29, 30), and may indicate a more advanced tumor status, which requires more aggressive hepatic resection, leading to the development of large amounts of blood loss requiring blood transfusion. Our findings suggest that the necessity for intraoperative blood transfusion will be affected not only by the status of liver cirrhosis or type of hepatectomy but also by the status of the primary hepatic tumor.
A British consensus conference on autologous transfusion stated that autologous blood donation should be considered only if the likelihood of transfusion exceeds 50% (31). From the viewpoint of the efficient and effective usage of autologous transfusions, it is not necessary to consider autologous blood donation when our score is less than 2 points.
The experience of the surgeon can influence the rates of success and complications (32). Hepatectomy requires a sophisticated technique that is different from that used in digestive surgery (32). Trainee status of the surgeon has been identified to be a significantly favorable factor for decreased liver transection times (33). Junior trainees should be given more practice on low-scoring patients with appropriate supervision.
It is true that preoperative anemia may be a risk factor for blood transfusion during hepatectomy, whereas the status of anemia does not relate to the difficulty of hepatectomy. Moreover, having preoperative anemia prevents patients from donating blood before surgery. Thus, in this study, we excluded the status of anemia from our analysis. Retrospective analysis in a single Institution is a limitation of the present study. Additional external validation is needed to confirm that this score is applicable to other surgical teams.
In conclusion, this study has established a predictive scoring system to assess the need for intraoperative blood transfusion during hepatectomy, with good discriminating ability using only preoperative factors. The present predictive scoring system will be useful for the perioperative management of patients undergoing hepatectomy.
- Received November 1, 2013.
- Revision received November 14, 2013.
- Accepted November 15, 2013.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
