World J Surg DOI 10.1007/s00268-014-2891-6
ORIGINAL SCIENTIFIC REPORT
Blood Transfusion is Associated with Recurrence of Hepatocellular Carcinoma After Hepatectomy in Child–Pugh Class A Patients Noboru Harada • Ken Shirabe • Takashi Maeda Hiroto Kayashima • Teruyoshi Ishida • Yoshihiko Maehara
•
Ó Socie´te´ Internationale de Chirurgie 2014
Abstract Introduction Previous reports have indicated an association between blood transfusion and prognosis of hepatocellular carcinoma (HCC) after hepatectomy. However, clinicopathological biases were not adjusted in these studies. We aimed to clarify the effect of blood transfusions in patients with HCC and Child–Pugh class A after hepatectomy by using inverse probability of treatment weighting (IPTW) analysis for selection bias control. Materials and methods We enrolled 479 patients with primary HCC and Child–Pugh class A retrospectively (91 transfused and 388 nontransfused patients) who underwent curative hepatectomy. After adjusting for different covariate distributions for both groups by IPTW, we analyzed the prognostic outcomes. Results In the unweighted analyses, overall survival (OS) rate of transfused patients was significantly lower than in nontransfused patients (P \ 0.0001). Recurrence-free survival (RFS) rate of transfused patients was significantly lower than that of nontransfused patients (P = 0.0024). Multivariate analysis showed that blood transfusion was an independent prognostic factor of OS and RFS. The different distributive covariates between the two groups were age, presence of liver cirrhosis, serum level of alpha-fetoprotein, maximum tumor diameter, and amount of intraoperative blood loss. After IPTW by these covariates, OS rate of transfused patients was not significantly lower than those of nontransfused patients, whereas RFS rate of transfused patients remained significantly lower than those of nontransfused patients (P = 0.038, adjusted HR 1.45; 95 % CI 1.0–2.1). Conclusions These results suggest that blood transfusion was associated with recurrence of HCC after hepatectomy in patients with HCC and Child–Pugh class A.
Abbreviations HCC Hepatocellular carcinoma IPTW Inverse probability of treatment weighting OS Overall survival RFS Recurrence-free survival
AFP ICG-R15 HCV TRIM HR CI
alpha-Fetoprotein Indocyanine green retention rate at 15 min Hepatitis C virus Transfusion-related immunomodulation Hazard ratio Confidence interval
N. Harada (&) T. Maeda H. Kayashima T. Ishida The Department of Surgery, Hiroshima Red Cross Hospital and Atomic Bomb Survivors Hospital, Hiroshima, Japan e-mail: [email protected]
Introduction
K. Shirabe Y. Maehara The Department of Surgery and Medical Science, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan
Hepatocellular carcinoma (HCC) is the sixth most prevalent cancer and the third most frequent cause of cancer-related death in the world [1]. In Japan, population-based survival
123
World J Surg
rates of HCC, accounting for 90 % of liver cancer cases, increased statistically during 1993–1996 and 2000–2002. These outcomes were considered a result of improved diagnostic and therapeutic strategies for HCC [2]. The recent decrease in liver cancer incidence and mortality was interpreted as a result of the decreasing trend in hepatitis C infection. However, the long-term outcomes of HCC after hepatectomy are still not satisfactory although the surgical management has made progress [3]. In hepatic resection, we have still possibility of blood transfusion when we experienced much more intraoperative bleeding than we intend. Regarding the surgical management of HCC, there has been considerable interest in studying the effect of blood transfusions on the recurrence for HCC after hepatectomy to improve the postoperative prognosis. These observations have led to the speculation that perioperative blood transfusions, by immunomodulation reactions, may have a deleterious effect on the recurrence and survival of patients undergoing hepatectomy for HCC. Although several studies have indicated that perioperative blood transfusion after hepatectomy is a prognostic factor of HCC outcomes [4, 5], these study may be limited by selection biases, such as liver function, presence of liver cirrhosis, preoperative level of alpha-fetoprotein (AFP), tumor size, tumor number, vascular invasion, tumor differentiation, and intraoperative blood loss. Especially, intraoperative blood loss was reported to be an independent prognostic factor of tumor recurrence and death [6]. To obtain confirmative evidence, a well-designed randomized controlled trial is necessary; however, such a trial cannot be performed owing to current practice guidelines and ethical concerns. We hypothesized that a cohort study with inverse probability of treatment weighting (IPTW) and a regression method could balance the underlying different distributive covariates between transfused and nontransfused patients. IPTW weights the samples using the propensity score to reduce the confounding that frequently occurs in cohort studies of the effects of treatment on outcome, and enables estimation of marginal or population-average treatment effects [7]. In this study, we investigated the association between perioperative blood transfusion and the recurrence-free survival (RFS) or overall survival (OS) in patients with HCC and Child–Pugh class A after hepatectomy by using IPTW to control selection bias.
Patients and methods Patients and surgical procedures We enrolled 479 consecutive patients with primary HCC that underwent curative hepatectomy at the Department of
123
Surgery, Hiroshima Red Cross Hospital and Atomic-bomb Survivors Hospital between March 1989 and September 2013. All patients had a confirmed pathological diagnosis of HCC. The type of hepatectomy was selected based on liver function and tumor extension [8, 9]. Liver function was assessed by Child–Pugh classification [10] and indocyanine green retention rate at 15 min (ICG-R15). Hepatectomy procedures were described elsewhere [9, 11]. Written informed consent was obtained from all patients for surgical treatment according to institutional guidelines. The study protocol conformed to the updated ethical guidelines of the 2013 Declaration of Helsinki and was approved by our Institutional Review Board. Definitions Curative resection was defined as complete macroscopic and microscopic tumor removal. Resected liver specimens were stained with Masson’s trichrome and examined by two pathologists at our hospital. The METAVIR scoring system [12] was used to diagnose liver cirrhosis (F4). Perioperative blood transfusion was defined as a transfusion of red cell concentrate. Transfusions of other blood products, including fresh-frozen plasma, platelets, or albumin, were not considered blood transfusions. Major hepatectomy was defined as resection of 2 or more subsegments, and minor resection, including partial resection, involved less than 2 subsegments. Serious postoperative complications were defined as Clavien–Dindo grade III or higher [13]. Postoperative liver failure was defined according to the guidelines of the International Study Group of Liver Surgery [14]. Survival and recurrence Patients underwent blood tests and computed tomography every 3 months. Recurrence was diagnosed based on imaging findings. Patients with intrahepatic recurrence were managed with ablative therapies, such as radiofrequency ablation and percutaneous ethanol injection, transcatheter arterial chemoembolization, or surgery. In case of death, survival time after surgery and cause of death were recorded. Postoperative survival time and recurrence status were recorded. Statistical analysis For continuous variables, parametric analyses were performed using Student’s t test. Wilcoxon rank-sum test was used for nonparametric analyses. Categorical variables were compared using v2 or Fisher’s exact tests. The Kaplan–Meier method was used to construct OS and RFS curves and survival curves were compared by log-rank tests.
World J Surg
To identify independent factors of OS and RFS rate, the factors found to be significant on univariate analysis were subjected to multivariate analysis using a Cox proportional hazards model. We used IPTW analysis to overcome possible bias of different covariates distributions among the nontransfused and transfused groups. Propensity scores by IPTW were calculated using a logistic regression model to predict the probability of each patient receiving blood transfusion on the basis of the 12 clinicopathological variables. Variables entered in the IPTW model were age, gender, anti-hepatitis C virus (anti-HCV) antibody, frequency of liver cirrhosis, ICG-R15, AFP levels, maximum tumor diameter, number of tumors, histological grading (well vs. moderately or poorly differentiated HCC), microscopic vascular invasion, extent of hepatic resection (major vs. minor hepatectomy), and intraoperative blood loss. All enrolled patients qualified as Child–Pugh class A. After IPTW were created, the two groups were then balanced by means of IPTW and the differences in OS and RFS rates among each group were tested by Cox regression analyses. Statistical significance was set at P value less than 0.05. Statistical analyses were performed using the R statistical programming environment (the Comprehensive R Archive Network, http://cran.md.tsukuba.ac.jp) and JMP 9.0 software (SAS Institute, Cary, NC). We used four basic R programming packages (Rcmdr, survival, RcmdrPlugin.survival, and Epi) to perform the IPTW analyses in this study.
Results Clinicopathological characteristics of patients with primary HCC and Child–Pugh class A who underwent hepatectomy Differences between characteristics of transfused and nontransfused patients are summarized in Table 1. Transfused patients had younger age (mean 64.4 vs. 68.0, P \ 0.001), higher frequency of liver cirrhosis (79.1 vs. 55.7 %, P \ 0.001), higher serum AFP levels (median 34.7 vs. 14.2 ng/mL, P = 0.0014), a greater maximum tumor diameter (mean 41.3 vs. 29.7 mm, P \ 0.001), and greater intraoperative blood loss (mean 1,324 vs. 360 mL, P \ 0.001) than nontransfused patients. OS and RFS The median duration of follow-up after surgery for the 479 patients was 50.3 months (range 0.5–230.3 months). According to the unweighted (crude) analyses, OS rates of transfused patients were significantly lower than that of nontransfused patients (P \ 0.0001). The 5- and 10-year
Table 1 Clinicopathological characteristics of primary HCC patients with Child–Pugh class A who underwent hepatectomy Variables
Nontransfused (n = 388)
Transfused (n = 91)
P
Age (years), mean (range)
68.0 (34–87)
64.4 (42–82)
\0.001
Gender 261/127
58/33
Anti-HCVAb positive
Male/female
256 (66.0 %)
62 (68.1 %)
0.70
Liver cirrhosis
216 (55.7 %)
71(79.1 %)
\0.001
ICG-R15 (%), mean (range)
17.2 (0.9–73.7)
18.6 (1.9–52.2)
0.17
AFP (ng/mL), median (range)
14.2 (1.6–450,000)
34.7 (0–30,890)
0.0014
Maximum tumor diameter (mm), mean (range)
29.7 (0.6–12.5)
41.3 (0.8–16.0)
Number of tumors, mean (range) Histological grading
1.3 (1–7)
1.2 (1–4)
0.27
51/337
16/75
0.27
227 (58.5 %)
50 (54.9 %)
0.54
103/285
32/59
0.1
Intraoperative blood loss (mL), mean (range)
360 (15–2,250)
1,324 (826,086)
Blood transfusion (mL), mean (range)
0
935 (200–6,400)
–
Postoperative serious complications (Clavien–Dindo grade III or higher)
17 (4.4 %)
7 (7.7 %)
0.19
Well/moderately or poorly Microscopic vascular invasion
0.52
\0.001
Extent of hepatic resection Major hepatectomy/ minor hepatectomy
\0.001
HCC hepatocellular carcinoma, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alphafetoprotein
OS rates in the transfused group were 55.4 and 20.1 %, respectively; whereas, the OS rates were 73.9 and 44.6 %, respectively, in the nontransfused group (Fig. 1a). The RFS rate of transfused patients was significantly lower than that of nontransfused patients (P = 0.0024). The 5- and 10-year RFS rates in the transfused group were 24.3 and 10.9 %, respectively; whereas the RFS rates in the nontransfused group were 37.7 and 20.2 %, respectively (Fig. 1b). Clinicopathological factors were evaluated to determine which were associated with worse OS in this population (Table 2). Univariate analyses showed that liver cirrhosis (P \ 0.0001), ICG-R15 [ 20 % (P = 0.0002), maximum tumor diameter [50 mm (P = 0.032), multiple tumors (P = 0.002), microscopic vascular invasion (P = 0.0041),
123
World J Surg
Percent overall survival (%)
(a)
(Crude OS)
100 Non-transfused patients (n = 388) 80 P < 0.0001 60 40 Transfused patients (n = 91) 20 0 0
1
2
3
4
5
6
7
8
9
10
Percent recurrence free survival (%)
Years after surgery (year)
(b)
(Crude RFS)
100 P = 0.0024 80 Non-transfused patients (n = 388)
60 40 20
IPTW analysis
Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Fig. 1 Prognosis of 479 patients who underwent liver resection for primary hepatocellular carcinoma with and without perioperative blood transfusion (n = 91 vs. n = 388). a Unweighted (crude) Kaplan–Meier curve for percent overall survival rate after hepatectomy. Overall survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P \ 0.0001). b Unweighted (crude) Kaplan–Meier curve for percent recurrent-free survival rate after hepatectomy. Recurrentfree survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P \ 0.0024)
intraoperative blood loss [1,000 mL (P = 0.0068), and blood transfusion (P \ 0.0001) were significantly associated with worse OS, whereas age, gender, anti-HCVAb positive, serum AFP levels, histological grading, and extent of hepatic resection were not. Multivariate Cox proportional hazards analysis showed that liver cirrhosis (P = 0.011; hazard ratio [HR] 1.50; 95 % confidence interval [CI] 1.10–2.06), ICG-R15 [ 20 % (P = 0.006; HR 1.58; 95 % CI 1.14–2.18), maximum tumor diameter [50 mm (P = 0.0015; HR 1.81; 95 % CI 1.13–2.78), multiple tumors (P = 0.004; HR 1.71; 95 % CI 1.19–2.41), microvascular invasion (P = 0.006; HR 1.53; 95 % CI 1.13–2.08), and blood transfusion (P = 0.008; HR 1.76; 95 % CI 1.16–2.62) were independent predictors of worse OS.
123
Clinicopathological factors were also evaluated to determine which were associated with worse recurrence in this population (Table 3). Univariate analyses showed that liver cirrhosis (P \ 0.0001), ICG-R15 [ 20 % (P = 0.027), serum AFP levels [400 ng/mL (P = 0.0014), maximum tumor diameter [50 mm (P = 0.038), multiple tumors (P = 0.0003), histological grading (moderately or poorly differentiated HCC; P = 0.029), microscopic vascular invasion (P = 0.0034), and blood transfusion (P = 0.0024) were significantly associated with worse RFS, whereas age, gender, anti-HCVAb positive, extent of hepatic resection, and intraoperative blood loss were not. Multivariate Cox proportional hazards analysis showed that liver cirrhosis (P = 0.005; HR 1.41; 95 % CI 1.11–1.82), serum AFP levels [400 ng/mL (P = 0.03; HR 1.43; 95 % CI 1.04–1.94), multiple tumors (P = 0.001; HR 1.63; 95 % CI 1.23–2.13), histological grading (moderately or poorly differentiated HCC, P = 0.082; HR 1.39; 95 % CI 0.96–2.05), microvascular invasion (P = 0.04; HR 1.30; 95 % CI 1.01–1.69), and blood transfusion (P = 0.009; HR 1.46; 95 % CI 1.10–1.91) were independent predictors of worse RFS.
In 12 clinicopathological variables, including age, gender, anti-HCV antibody, frequency of liver cirrhosis, ICG-R15, serum AFP levels, maximum tumor diameter, number of tumors, histological grading, microscopic vascular invasion, extent of hepatic resection, and operative blood loss, the different distributive covariates between the two groups were age (P \ 0.001), liver cirrhosis (P \ 0.001), serum AFP levels (P = 0.0014), maximum tumor diameter (P \ 0.001), and intraoperative blood loss (P \ 0.001) (Table 1). After IPTW by these covariates, weighted OS rates of transfused patients were not significantly lower than those of nontransfused patient (P = 0.26, adjusted HR 1.33; 95 % CI 0.80–2.24, Fig. 2a); whereas weighted RFS rate of transfused patients remained significantly lower than those of nontransfused patients (P = 0.038, adjusted HR 1.45; 95 % CI 1.0–2.1, Fig. 2b). Postoperative complications There was one in-hospital death in the transfused group secondary to liver failure, and one in-hospital death in the nontransfused group secondary to sepsis. There were no significantly different rates of serious postoperative complications (Clavien–Dindo grade III or higher) between the transfused group and the nontransfused group (7.7 vs. 4.4 %; P = 0.19). Seven serious postoperative complications were recorded in the transfused group, including refractory ascites (n = 2), grade C liver failure (n = 1),
World J Surg Table 2 Univariate and multivariate analysis of overall survival Variables
Univariate analysis 5-year survival
Multivariate analysis P value
HR (95 % CI)
P value
Age (years) ^60/[60
73.0/69.1
0.14
71.6/66.9
0.77
70.2/69.7
0.68
59.6/79.2
\0.0001
1.50 (1.10–2.06)
0.011
74.9/58.2
0.0002
1.58 (1.14–2.18)
0.006
72.9/52.6
0.16
71.9/54.0
0.032
1.81 (1.13–2.78)
0.015
57.2/73.6
0.002
1.71 (1.19–2.41)
0.004
83.3/67.8
0.29
64.0/77.8
0.041
1.53 (1.13–2.08)
0.006
70.7/69.7
0.22
73.0/54.9
0.0068
1.05 (0.67–1.63)
0.83
55.4/73.9
\0.0001
1.76 (1.16–2.62)
0.008
Gender Male/female Anti-HCVAb positive Positive/negative Liver cirrhosis Yes/no ICG-R15 (%) ^20/[20 AFP (ng/mL) ^400/[400 Maximum tumor diameter (mm) ^50/[50 Multiple tumors Yes/no Histological grading Well/moderately or poorly Microscopic vascular invasion Yes/no Extent of hepatic resection Major/minor hepatectomy Intraoperative blood loss (mL) ^1000/[1000 Blood transfusion Transfused/nontransfused
CI confidence interval, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alpha-fetoprotein
ileus (n = 1), bile leakage (n = 1), postoperative bleeding (n = 1), and refractory pleural effusion (n = 1). Seventeen serious postoperative complications were recorded in the nontransfused group, including surgical site infection (n = 4), bile leakage (n = 3), refractory ascites (n = 2), grade B liver failure (n = 2), cholangitis (n = 1), refractory pleural effusion (n = 1), pneumonia (n = 1), pneumothorax (n = 1), perforation of the jejunum (n = 1), and sepsis (n = 1).’’
Discussion In this study, we analyzed the effect of blood transfusion on prognostic factors of HCC after hepatectomy in patients with Child–Pugh class A by using IPTW. Many studies have reported on the association between the blood transfusions and prognosis of HCC [4, 5]. Several investigators have reported that perioperative blood transfusions
promote the recurrence of HCC and decrease disease-free and OS rates after hepatectomy [15, 16]. However, their studies were affected by selection biases, including liver function, presence of liver cirrhosis, characteristics, and operative parameters. The effect of blood transfusions could be evidence by a well-designed randomized controlled trial. However, it is not easy to conduct such a trial because of current practice guidelines and ethical concerns. Therefore, we opted for a cohort study with IPTW and a regression method to balance underlying different distributive covariates between transfused and nontransfused patients. As for liver function, we selected the patients with primary HCC and Child–Pugh class A in this study because of the accepted indications of hepatectomy for HCC, including Child–Pugh class A, performance status 0, and early stage (e.g., single nodule and tumor diameter B2 cm) based on the guideline by Barcelona Clinic Liver Cancer Group [17] and American Association for the Study of Liver Disease [18]. Furthermore, we adjusted the
123
World J Surg
(a)
(Weighted OS)
Percent overall survival (%)
100
Non-transfused patients (n = 388)
80 P= 0.26 60 40 20 Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Percent recurrence free survival (%)
(b)
(Weighted RFS)
100 P= 0.038 80 Non-transfused patients (n = 388) 60 40 20 Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Fig. 2 a Weighted (IPTW) Kaplan–Meier curve for percent overall survival rate after hepatectomy after IPTW analyses. Overall survival rate of transfused patients (bolded black line) was not significantly lower than that of nontransfused patients (dotted line P = 0.26). b Weighted (IPTW) Kaplan–Meier curve for percent recurrent-free survival rate after hepatectomy after IPTW analyses. Recurrent-free survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P = 0.038)
background characteristics of patients, tumors, and operating factors such as age, presence of liver cirrhosis, AFP, maximum tumor diameter, and intraoperative blood loss. Especially, we focused on intraoperative blood loss because Katz et al. reported that blood loss was an independent prognostic factor [6]. Therefore, the amount of intraoperative bleeding can be one of the greatest confounding factors. To the best of our knowledge, no previous reports have adjusted for intraoperative bleeding. It was reported that perioperative blood transfusions did not influence the overall and disease-free survival rate in patients with HCC by using propensity score analysis [19]. However, this variable was not adjusted between transfusion and nontransfusion groups in their study and matching
123
on the propensity scores, as a statistical method, was limited for sample selection. In nonexperimental studies of medical interventions, IPTW estimation is now commonly used to control for confounding [20, 21]. Conversely, propensity scores can be used to reduce confounding, that is, matching on the propensity score. This involves forming matched sets of treated and untreated subjects who have similar values of propensity scores [22]. Once these propensity score-matched groups have been formed, the sample size becomes small and a bias that the matched group has been formed based on the smaller group comes into play. Therefore, we adopted the IPTW adjusted analysis to control selection bias in our study. In the weighted variables such as age, presence of liver cirrhosis, AFP, maximum tumor diameter, and intraoperative blood loss, we used a Cox regression model to regress survival between transfused and nontransfused patients, and used a robust variance estimator [7, 23, 24]. Although this model adjusted the variables, including intraoperative blood loss that varied widely in each operation, the recurrence rate of transfused patients significantly remained to be worse than that of nontransfused patients. However, the negative effect of blood transfusion on OS vanished when the background confounding factors were adjusted. Other independent factors, such as ICG15-R, serum AFP levels, and maximum tumor diameter may have strong effects on OS because these factors were the different independent factors of OS from those of RFS. These results suggested that blood transfusions could have a negative effect on the recurrence of HCC after hepatectomy. Therefore, we recommend avoiding blood transfusions during hepatectomy if the patient vital signs are stable and tolerable for hepatectomy when in the presence of considerable blood loss during surgery. Our results show a dramatic decrease in the RFS rate at about 5 years after hepatectomy, as a result of highweighted patients with low propensity scores developing recurrent HCC. Although IPTW analysis minimizes selection bias, it may result in an artificial imbalance of weighted patients [25]. This reflects a limitation of using IPTW analysis to estimate outcomes. In transfused patients, the most reasonable cause of the worsened recurrence rate was the immunosuppressive effect to the host that occurs after a transfusion (transfusion-related immunomodulation: TRIM). Results of experimental animal models suggest that allogeneic blood transfusion-associated immunomodulation is an immunologically mediated biological effect associated primarily with the infusion of allogeneic leukocytes [26]. Thus, TRIM may have negative effects, such as tumor recurrence, owing to allogeneic blood transfusion by immunomodulatory mechanisms. Moreover, TRIM has also been
World J Surg Table 3 Univariate and multivariate analysis of recurrence-free survival Variables
Univariate analysis 5-year survival
P value
Multivariate analysis HR (95 % CI)
P value
Age (years) ^60/[ 60
34.4/34.9
0.93
34.1/36.8
0.44
34.5/36.3
0.67
23.8/44.2
\0.0001
1.41 (1.11–1.82)
0.005
38.3/26.1
0.027
1.23 (0.94–1.59)
0.13
36.7/23.9
0.0014
1.43 (1.04–1.94)
0.03
36.1/25.6
0.038
1.41 (0.96–2.02)
0.078
19.3/39.1
0.0003
1.63 (1.23–2.13)
0.001
43.5/33.4
0.029
1.39 (0.96–2.05)
0.082
31.7/39.6
0.0034
1.30 (1.01–1.69)
0.04
40.3/32.9
0.34
36.0/29.3
0.17
24.3/37.7
0.0024
1.46 (1.10–1.91)
0.009
Gender Male/female Anti-HCVAb positive Positive/negative Liver cirrhosis Yes/no ICG-R15 (%) ^20/[20 AFP (ng/mL) ^400/[400 Maximum tumor diameter (mm) ^50/[50 Multiple tumors Yes/no Histological grading Well/moderately or poorly Microscopic vascular invasion Yes/no Extent of hepatic resection Major/minor hepatectomy Intraoperative blood loss (mL) ^1000/[1000 Blood transfusion Transfused/nontransfused
CI confidence interval, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alpha-fetoprotein
associated with allogeneic blood transfusion by proinflammatory mechanisms. TRIM effects may be mediated by allogeneic mononuclear cells, white blood cell-derived soluble mediators, and soluble HLA peptides circulating in allogeneic plasma [27]. Additionally, iron overload secondary to blood transfusion may have deleterious effects on postoperative patients with HCC. It has become increasingly evident that excess body iron may accelerate the progress of liver fibrosis and recurrence of HCC [28]. According to multiple regression analysis, blood transfusions have a negative effect on postoperative liver function in patients with HCC within 2 years [29]. In conclusion, we found that the RFS rate of transfused patients remained significantly lower than those of nontransfused patients after adjusting the control bias by using IPTW. Our results suggest that tumor recurrence is associated with blood transfusions post-hepatectomy for patients with primary HCC and Child–Pugh class A.
Acknowledgments The authors thank the co-medical staff (O.T., S.Y., and M.T.) in Hiroshima Red Cross Hospital and Atomic Bomb Survivor Hospital for helping with the data collection. Conflict of interest of interest.
The authors declare that they have no conflict
References 1. Forner A, Llovet JM, Bruix J (2012) Hepatocellular carcinoma. Lancet 379(9822):1245–1255 2. Katanoda K, Matsuda T, Nishimoto H et al (2013) An updated report of the trends in cancer incidence and mortality in Japan. Jpn J Clin Oncol 43(5):492–507 3. Shirabe K, Takeishi K, Maehara Y et al (2011) Improvement of longterm outcomes in hepatitis C virus antibody-positive patients with hepatocellular carcinoma after hepatectomy in the modern era. World J Surg 35(5):1072–1084. doi:10.1007/s00268-011-1013-y 4. Asahara T, katayama K, Itamoto T et al (1999) Perioperative blood transfusion as a prognostic indicator in patients with
123
World J Surg
5.
6.
7.
8. 9.
10.
11.
12.
13.
14.
15.
16.
hepatocellular carcinoma. World J Surg 23(7):676–680. doi:10. 1007/PL00012367 Makino Y, Yamanoi A, Nagasue N et al (2000) The influence of perioperative blood transfusion on intrahepatic recurrence after curative resection of hepatocellular carcinoma. Am J Gastroenterol 95(5):1294–1300 Katz SC, Shia J, Dematteo RP et al (2009) Operative blood loss independently predicts recurrence and survival after resection of hepatocellular carcinoma. Ann Surg 249(4):617–623 Austin PC (2012) The performance of different propensity score methods for estimating marginal hazard ratios. Stat Med 32:2837–2849 Makuuchi M, Kosuge T, Kawasaki S et al (1993) Surgery for small liver cancers. Semin Surg Oncol 9(4):298–304 Takeishi K, Shirabe K, Maehara Y et al (2011) Clinicopathological features and outcomes of young patients with hepatocellular carcinoma after hepatectomy. World J Surg 35(5):1063– 1071. doi:10.1007/s00268-011-1017-7 Pugh RN, Murray-Lyon IM, Williams R et al (1973) Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 60(8):646–649 Yamashita YI, Tsuijita E, Maehara Y et al (2013) Trends in surgical results of hepatic resection for hepatocellular carcinoma: 1,000 consecutive cases over 20 years in a single institution. Am J Surg 13:522–529 Poynard T, Bedossa P, Opolon P (1997) Natural history of liver fibrosis progression in patients with chronic hepatitis C. The OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups. Lancet 349(9055):825–832 Dindo D, Demartines N, Clavien PA (2004) Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 240(2):205–213 Rahbari NN, Garden OJ, Padbury R et al (2011) Posthepatectomy liver failure: a definition and grading by the International Study Group of Liver Surgery (ISGLS). Surgery 149(5):713–724 Yamamoto J, Kosuge T, Makuuchi M et al (1994) Perioperative blood transfusion promotes recurrence of hepatocellular carcinoma after hepatectomy. Surgery 115(3):303–309 Hanazaki K, Kajikawa S, Amano J et al (2005) Perioperative blood transfusion and survival following curative hepatic
123
17.
18.
19.
20.
21. 22.
23. 24.
25. 26.
27.
28. 29.
resection for hepatocellular carcinoma. Hepatogastroenterology 52(62):524–529 Forner Al, Reig ME, Bruix J et al (2010) Current strategy for staging and treatment: the BCLC update and future prospects. Semin Liver Dis 30(1):61–74 Bruix J, Sherman M, American Association for the Study of Liver Diseases (2011) Management of hepatocellular carcinoma: an update. Hepatology 53(3):1020–1022 Kuroda S, Tashiro H, Ohdan H et al (2012) No impact of perioperative blood transfusion on recurrence of hepatocellular carcinoma after hepatectomy. World J Surg 36(3):651–658. doi:10. 1007/s00268-012-1425-3 Robins JM, Herna´n MA, Brumback B (2000) Marginal structural models and causal inference in epidemiology. Epidemiology 11(5):550–560 Sato T, Matsuyama Y (2003) Marginal structural models as a tool for standardization. Epidemiology 14(6):680–686 Rosenbaum PR, Rubin DB (1985) Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 39:33–38 Lin DY, Wei LJ (1989) The robust inference for the proportional hazards model. J Am Stat Assoc 84:1074–1078 Joffe MM, Ten Have TR, Kimmel SE et al (2004) Model selection, confounder control, and marginal structural models: review and new applications. Am Stat 58:272–279 Halpern EF (2014) Behind the numbers: inverse probability weighting. Radiology 271(3):625–628 Blajchman MA (1998) Immunomodulatory effects of allogeneic blood transfusions: clinical manifestations and mechanisms. Vox Sang 74(Suppl 2):315–319 Vamvakas EC, Blajchman MA (2007) Transfusion-related immunomodulation (TRIM): an update. Blood Rev 21(6): 327–348 Kew MC (2014) Hepatic iron overload and hepatocellular carcinoma. Liver Cancer 3(1):31–40 Itasaka H, Yamamoto K, Matsumata T et al (1995) Influence of blood transfusion on postoperative long-term liver function in patients with hepatocellular carcinoma. Hepatogastroenterology 42(5):465–468
ORIGINAL SCIENTIFIC REPORT
Blood Transfusion is Associated with Recurrence of Hepatocellular Carcinoma After Hepatectomy in Child–Pugh Class A Patients Noboru Harada • Ken Shirabe • Takashi Maeda Hiroto Kayashima • Teruyoshi Ishida • Yoshihiko Maehara
•
Ó Socie´te´ Internationale de Chirurgie 2014
Abstract Introduction Previous reports have indicated an association between blood transfusion and prognosis of hepatocellular carcinoma (HCC) after hepatectomy. However, clinicopathological biases were not adjusted in these studies. We aimed to clarify the effect of blood transfusions in patients with HCC and Child–Pugh class A after hepatectomy by using inverse probability of treatment weighting (IPTW) analysis for selection bias control. Materials and methods We enrolled 479 patients with primary HCC and Child–Pugh class A retrospectively (91 transfused and 388 nontransfused patients) who underwent curative hepatectomy. After adjusting for different covariate distributions for both groups by IPTW, we analyzed the prognostic outcomes. Results In the unweighted analyses, overall survival (OS) rate of transfused patients was significantly lower than in nontransfused patients (P \ 0.0001). Recurrence-free survival (RFS) rate of transfused patients was significantly lower than that of nontransfused patients (P = 0.0024). Multivariate analysis showed that blood transfusion was an independent prognostic factor of OS and RFS. The different distributive covariates between the two groups were age, presence of liver cirrhosis, serum level of alpha-fetoprotein, maximum tumor diameter, and amount of intraoperative blood loss. After IPTW by these covariates, OS rate of transfused patients was not significantly lower than those of nontransfused patients, whereas RFS rate of transfused patients remained significantly lower than those of nontransfused patients (P = 0.038, adjusted HR 1.45; 95 % CI 1.0–2.1). Conclusions These results suggest that blood transfusion was associated with recurrence of HCC after hepatectomy in patients with HCC and Child–Pugh class A.
Abbreviations HCC Hepatocellular carcinoma IPTW Inverse probability of treatment weighting OS Overall survival RFS Recurrence-free survival
AFP ICG-R15 HCV TRIM HR CI
alpha-Fetoprotein Indocyanine green retention rate at 15 min Hepatitis C virus Transfusion-related immunomodulation Hazard ratio Confidence interval
N. Harada (&) T. Maeda H. Kayashima T. Ishida The Department of Surgery, Hiroshima Red Cross Hospital and Atomic Bomb Survivors Hospital, Hiroshima, Japan e-mail: [email protected]
Introduction
K. Shirabe Y. Maehara The Department of Surgery and Medical Science, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan
Hepatocellular carcinoma (HCC) is the sixth most prevalent cancer and the third most frequent cause of cancer-related death in the world [1]. In Japan, population-based survival
123
World J Surg
rates of HCC, accounting for 90 % of liver cancer cases, increased statistically during 1993–1996 and 2000–2002. These outcomes were considered a result of improved diagnostic and therapeutic strategies for HCC [2]. The recent decrease in liver cancer incidence and mortality was interpreted as a result of the decreasing trend in hepatitis C infection. However, the long-term outcomes of HCC after hepatectomy are still not satisfactory although the surgical management has made progress [3]. In hepatic resection, we have still possibility of blood transfusion when we experienced much more intraoperative bleeding than we intend. Regarding the surgical management of HCC, there has been considerable interest in studying the effect of blood transfusions on the recurrence for HCC after hepatectomy to improve the postoperative prognosis. These observations have led to the speculation that perioperative blood transfusions, by immunomodulation reactions, may have a deleterious effect on the recurrence and survival of patients undergoing hepatectomy for HCC. Although several studies have indicated that perioperative blood transfusion after hepatectomy is a prognostic factor of HCC outcomes [4, 5], these study may be limited by selection biases, such as liver function, presence of liver cirrhosis, preoperative level of alpha-fetoprotein (AFP), tumor size, tumor number, vascular invasion, tumor differentiation, and intraoperative blood loss. Especially, intraoperative blood loss was reported to be an independent prognostic factor of tumor recurrence and death [6]. To obtain confirmative evidence, a well-designed randomized controlled trial is necessary; however, such a trial cannot be performed owing to current practice guidelines and ethical concerns. We hypothesized that a cohort study with inverse probability of treatment weighting (IPTW) and a regression method could balance the underlying different distributive covariates between transfused and nontransfused patients. IPTW weights the samples using the propensity score to reduce the confounding that frequently occurs in cohort studies of the effects of treatment on outcome, and enables estimation of marginal or population-average treatment effects [7]. In this study, we investigated the association between perioperative blood transfusion and the recurrence-free survival (RFS) or overall survival (OS) in patients with HCC and Child–Pugh class A after hepatectomy by using IPTW to control selection bias.
Patients and methods Patients and surgical procedures We enrolled 479 consecutive patients with primary HCC that underwent curative hepatectomy at the Department of
123
Surgery, Hiroshima Red Cross Hospital and Atomic-bomb Survivors Hospital between March 1989 and September 2013. All patients had a confirmed pathological diagnosis of HCC. The type of hepatectomy was selected based on liver function and tumor extension [8, 9]. Liver function was assessed by Child–Pugh classification [10] and indocyanine green retention rate at 15 min (ICG-R15). Hepatectomy procedures were described elsewhere [9, 11]. Written informed consent was obtained from all patients for surgical treatment according to institutional guidelines. The study protocol conformed to the updated ethical guidelines of the 2013 Declaration of Helsinki and was approved by our Institutional Review Board. Definitions Curative resection was defined as complete macroscopic and microscopic tumor removal. Resected liver specimens were stained with Masson’s trichrome and examined by two pathologists at our hospital. The METAVIR scoring system [12] was used to diagnose liver cirrhosis (F4). Perioperative blood transfusion was defined as a transfusion of red cell concentrate. Transfusions of other blood products, including fresh-frozen plasma, platelets, or albumin, were not considered blood transfusions. Major hepatectomy was defined as resection of 2 or more subsegments, and minor resection, including partial resection, involved less than 2 subsegments. Serious postoperative complications were defined as Clavien–Dindo grade III or higher [13]. Postoperative liver failure was defined according to the guidelines of the International Study Group of Liver Surgery [14]. Survival and recurrence Patients underwent blood tests and computed tomography every 3 months. Recurrence was diagnosed based on imaging findings. Patients with intrahepatic recurrence were managed with ablative therapies, such as radiofrequency ablation and percutaneous ethanol injection, transcatheter arterial chemoembolization, or surgery. In case of death, survival time after surgery and cause of death were recorded. Postoperative survival time and recurrence status were recorded. Statistical analysis For continuous variables, parametric analyses were performed using Student’s t test. Wilcoxon rank-sum test was used for nonparametric analyses. Categorical variables were compared using v2 or Fisher’s exact tests. The Kaplan–Meier method was used to construct OS and RFS curves and survival curves were compared by log-rank tests.
World J Surg
To identify independent factors of OS and RFS rate, the factors found to be significant on univariate analysis were subjected to multivariate analysis using a Cox proportional hazards model. We used IPTW analysis to overcome possible bias of different covariates distributions among the nontransfused and transfused groups. Propensity scores by IPTW were calculated using a logistic regression model to predict the probability of each patient receiving blood transfusion on the basis of the 12 clinicopathological variables. Variables entered in the IPTW model were age, gender, anti-hepatitis C virus (anti-HCV) antibody, frequency of liver cirrhosis, ICG-R15, AFP levels, maximum tumor diameter, number of tumors, histological grading (well vs. moderately or poorly differentiated HCC), microscopic vascular invasion, extent of hepatic resection (major vs. minor hepatectomy), and intraoperative blood loss. All enrolled patients qualified as Child–Pugh class A. After IPTW were created, the two groups were then balanced by means of IPTW and the differences in OS and RFS rates among each group were tested by Cox regression analyses. Statistical significance was set at P value less than 0.05. Statistical analyses were performed using the R statistical programming environment (the Comprehensive R Archive Network, http://cran.md.tsukuba.ac.jp) and JMP 9.0 software (SAS Institute, Cary, NC). We used four basic R programming packages (Rcmdr, survival, RcmdrPlugin.survival, and Epi) to perform the IPTW analyses in this study.
Results Clinicopathological characteristics of patients with primary HCC and Child–Pugh class A who underwent hepatectomy Differences between characteristics of transfused and nontransfused patients are summarized in Table 1. Transfused patients had younger age (mean 64.4 vs. 68.0, P \ 0.001), higher frequency of liver cirrhosis (79.1 vs. 55.7 %, P \ 0.001), higher serum AFP levels (median 34.7 vs. 14.2 ng/mL, P = 0.0014), a greater maximum tumor diameter (mean 41.3 vs. 29.7 mm, P \ 0.001), and greater intraoperative blood loss (mean 1,324 vs. 360 mL, P \ 0.001) than nontransfused patients. OS and RFS The median duration of follow-up after surgery for the 479 patients was 50.3 months (range 0.5–230.3 months). According to the unweighted (crude) analyses, OS rates of transfused patients were significantly lower than that of nontransfused patients (P \ 0.0001). The 5- and 10-year
Table 1 Clinicopathological characteristics of primary HCC patients with Child–Pugh class A who underwent hepatectomy Variables
Nontransfused (n = 388)
Transfused (n = 91)
P
Age (years), mean (range)
68.0 (34–87)
64.4 (42–82)
\0.001
Gender 261/127
58/33
Anti-HCVAb positive
Male/female
256 (66.0 %)
62 (68.1 %)
0.70
Liver cirrhosis
216 (55.7 %)
71(79.1 %)
\0.001
ICG-R15 (%), mean (range)
17.2 (0.9–73.7)
18.6 (1.9–52.2)
0.17
AFP (ng/mL), median (range)
14.2 (1.6–450,000)
34.7 (0–30,890)
0.0014
Maximum tumor diameter (mm), mean (range)
29.7 (0.6–12.5)
41.3 (0.8–16.0)
Number of tumors, mean (range) Histological grading
1.3 (1–7)
1.2 (1–4)
0.27
51/337
16/75
0.27
227 (58.5 %)
50 (54.9 %)
0.54
103/285
32/59
0.1
Intraoperative blood loss (mL), mean (range)
360 (15–2,250)
1,324 (826,086)
Blood transfusion (mL), mean (range)
0
935 (200–6,400)
–
Postoperative serious complications (Clavien–Dindo grade III or higher)
17 (4.4 %)
7 (7.7 %)
0.19
Well/moderately or poorly Microscopic vascular invasion
0.52
\0.001
Extent of hepatic resection Major hepatectomy/ minor hepatectomy
\0.001
HCC hepatocellular carcinoma, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alphafetoprotein
OS rates in the transfused group were 55.4 and 20.1 %, respectively; whereas, the OS rates were 73.9 and 44.6 %, respectively, in the nontransfused group (Fig. 1a). The RFS rate of transfused patients was significantly lower than that of nontransfused patients (P = 0.0024). The 5- and 10-year RFS rates in the transfused group were 24.3 and 10.9 %, respectively; whereas the RFS rates in the nontransfused group were 37.7 and 20.2 %, respectively (Fig. 1b). Clinicopathological factors were evaluated to determine which were associated with worse OS in this population (Table 2). Univariate analyses showed that liver cirrhosis (P \ 0.0001), ICG-R15 [ 20 % (P = 0.0002), maximum tumor diameter [50 mm (P = 0.032), multiple tumors (P = 0.002), microscopic vascular invasion (P = 0.0041),
123
World J Surg
Percent overall survival (%)
(a)
(Crude OS)
100 Non-transfused patients (n = 388) 80 P < 0.0001 60 40 Transfused patients (n = 91) 20 0 0
1
2
3
4
5
6
7
8
9
10
Percent recurrence free survival (%)
Years after surgery (year)
(b)
(Crude RFS)
100 P = 0.0024 80 Non-transfused patients (n = 388)
60 40 20
IPTW analysis
Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Fig. 1 Prognosis of 479 patients who underwent liver resection for primary hepatocellular carcinoma with and without perioperative blood transfusion (n = 91 vs. n = 388). a Unweighted (crude) Kaplan–Meier curve for percent overall survival rate after hepatectomy. Overall survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P \ 0.0001). b Unweighted (crude) Kaplan–Meier curve for percent recurrent-free survival rate after hepatectomy. Recurrentfree survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P \ 0.0024)
intraoperative blood loss [1,000 mL (P = 0.0068), and blood transfusion (P \ 0.0001) were significantly associated with worse OS, whereas age, gender, anti-HCVAb positive, serum AFP levels, histological grading, and extent of hepatic resection were not. Multivariate Cox proportional hazards analysis showed that liver cirrhosis (P = 0.011; hazard ratio [HR] 1.50; 95 % confidence interval [CI] 1.10–2.06), ICG-R15 [ 20 % (P = 0.006; HR 1.58; 95 % CI 1.14–2.18), maximum tumor diameter [50 mm (P = 0.0015; HR 1.81; 95 % CI 1.13–2.78), multiple tumors (P = 0.004; HR 1.71; 95 % CI 1.19–2.41), microvascular invasion (P = 0.006; HR 1.53; 95 % CI 1.13–2.08), and blood transfusion (P = 0.008; HR 1.76; 95 % CI 1.16–2.62) were independent predictors of worse OS.
123
Clinicopathological factors were also evaluated to determine which were associated with worse recurrence in this population (Table 3). Univariate analyses showed that liver cirrhosis (P \ 0.0001), ICG-R15 [ 20 % (P = 0.027), serum AFP levels [400 ng/mL (P = 0.0014), maximum tumor diameter [50 mm (P = 0.038), multiple tumors (P = 0.0003), histological grading (moderately or poorly differentiated HCC; P = 0.029), microscopic vascular invasion (P = 0.0034), and blood transfusion (P = 0.0024) were significantly associated with worse RFS, whereas age, gender, anti-HCVAb positive, extent of hepatic resection, and intraoperative blood loss were not. Multivariate Cox proportional hazards analysis showed that liver cirrhosis (P = 0.005; HR 1.41; 95 % CI 1.11–1.82), serum AFP levels [400 ng/mL (P = 0.03; HR 1.43; 95 % CI 1.04–1.94), multiple tumors (P = 0.001; HR 1.63; 95 % CI 1.23–2.13), histological grading (moderately or poorly differentiated HCC, P = 0.082; HR 1.39; 95 % CI 0.96–2.05), microvascular invasion (P = 0.04; HR 1.30; 95 % CI 1.01–1.69), and blood transfusion (P = 0.009; HR 1.46; 95 % CI 1.10–1.91) were independent predictors of worse RFS.
In 12 clinicopathological variables, including age, gender, anti-HCV antibody, frequency of liver cirrhosis, ICG-R15, serum AFP levels, maximum tumor diameter, number of tumors, histological grading, microscopic vascular invasion, extent of hepatic resection, and operative blood loss, the different distributive covariates between the two groups were age (P \ 0.001), liver cirrhosis (P \ 0.001), serum AFP levels (P = 0.0014), maximum tumor diameter (P \ 0.001), and intraoperative blood loss (P \ 0.001) (Table 1). After IPTW by these covariates, weighted OS rates of transfused patients were not significantly lower than those of nontransfused patient (P = 0.26, adjusted HR 1.33; 95 % CI 0.80–2.24, Fig. 2a); whereas weighted RFS rate of transfused patients remained significantly lower than those of nontransfused patients (P = 0.038, adjusted HR 1.45; 95 % CI 1.0–2.1, Fig. 2b). Postoperative complications There was one in-hospital death in the transfused group secondary to liver failure, and one in-hospital death in the nontransfused group secondary to sepsis. There were no significantly different rates of serious postoperative complications (Clavien–Dindo grade III or higher) between the transfused group and the nontransfused group (7.7 vs. 4.4 %; P = 0.19). Seven serious postoperative complications were recorded in the transfused group, including refractory ascites (n = 2), grade C liver failure (n = 1),
World J Surg Table 2 Univariate and multivariate analysis of overall survival Variables
Univariate analysis 5-year survival
Multivariate analysis P value
HR (95 % CI)
P value
Age (years) ^60/[60
73.0/69.1
0.14
71.6/66.9
0.77
70.2/69.7
0.68
59.6/79.2
\0.0001
1.50 (1.10–2.06)
0.011
74.9/58.2
0.0002
1.58 (1.14–2.18)
0.006
72.9/52.6
0.16
71.9/54.0
0.032
1.81 (1.13–2.78)
0.015
57.2/73.6
0.002
1.71 (1.19–2.41)
0.004
83.3/67.8
0.29
64.0/77.8
0.041
1.53 (1.13–2.08)
0.006
70.7/69.7
0.22
73.0/54.9
0.0068
1.05 (0.67–1.63)
0.83
55.4/73.9
\0.0001
1.76 (1.16–2.62)
0.008
Gender Male/female Anti-HCVAb positive Positive/negative Liver cirrhosis Yes/no ICG-R15 (%) ^20/[20 AFP (ng/mL) ^400/[400 Maximum tumor diameter (mm) ^50/[50 Multiple tumors Yes/no Histological grading Well/moderately or poorly Microscopic vascular invasion Yes/no Extent of hepatic resection Major/minor hepatectomy Intraoperative blood loss (mL) ^1000/[1000 Blood transfusion Transfused/nontransfused
CI confidence interval, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alpha-fetoprotein
ileus (n = 1), bile leakage (n = 1), postoperative bleeding (n = 1), and refractory pleural effusion (n = 1). Seventeen serious postoperative complications were recorded in the nontransfused group, including surgical site infection (n = 4), bile leakage (n = 3), refractory ascites (n = 2), grade B liver failure (n = 2), cholangitis (n = 1), refractory pleural effusion (n = 1), pneumonia (n = 1), pneumothorax (n = 1), perforation of the jejunum (n = 1), and sepsis (n = 1).’’
Discussion In this study, we analyzed the effect of blood transfusion on prognostic factors of HCC after hepatectomy in patients with Child–Pugh class A by using IPTW. Many studies have reported on the association between the blood transfusions and prognosis of HCC [4, 5]. Several investigators have reported that perioperative blood transfusions
promote the recurrence of HCC and decrease disease-free and OS rates after hepatectomy [15, 16]. However, their studies were affected by selection biases, including liver function, presence of liver cirrhosis, characteristics, and operative parameters. The effect of blood transfusions could be evidence by a well-designed randomized controlled trial. However, it is not easy to conduct such a trial because of current practice guidelines and ethical concerns. Therefore, we opted for a cohort study with IPTW and a regression method to balance underlying different distributive covariates between transfused and nontransfused patients. As for liver function, we selected the patients with primary HCC and Child–Pugh class A in this study because of the accepted indications of hepatectomy for HCC, including Child–Pugh class A, performance status 0, and early stage (e.g., single nodule and tumor diameter B2 cm) based on the guideline by Barcelona Clinic Liver Cancer Group [17] and American Association for the Study of Liver Disease [18]. Furthermore, we adjusted the
123
World J Surg
(a)
(Weighted OS)
Percent overall survival (%)
100
Non-transfused patients (n = 388)
80 P= 0.26 60 40 20 Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Percent recurrence free survival (%)
(b)
(Weighted RFS)
100 P= 0.038 80 Non-transfused patients (n = 388) 60 40 20 Transfused patients (n = 91) 0 0
1
2
3
4
5
6
7
8
9
10
Years after surgery (year)
Fig. 2 a Weighted (IPTW) Kaplan–Meier curve for percent overall survival rate after hepatectomy after IPTW analyses. Overall survival rate of transfused patients (bolded black line) was not significantly lower than that of nontransfused patients (dotted line P = 0.26). b Weighted (IPTW) Kaplan–Meier curve for percent recurrent-free survival rate after hepatectomy after IPTW analyses. Recurrent-free survival rate of transfused patients (bolded black line) was significantly lower than that of nontransfused patients (dotted line P = 0.038)
background characteristics of patients, tumors, and operating factors such as age, presence of liver cirrhosis, AFP, maximum tumor diameter, and intraoperative blood loss. Especially, we focused on intraoperative blood loss because Katz et al. reported that blood loss was an independent prognostic factor [6]. Therefore, the amount of intraoperative bleeding can be one of the greatest confounding factors. To the best of our knowledge, no previous reports have adjusted for intraoperative bleeding. It was reported that perioperative blood transfusions did not influence the overall and disease-free survival rate in patients with HCC by using propensity score analysis [19]. However, this variable was not adjusted between transfusion and nontransfusion groups in their study and matching
123
on the propensity scores, as a statistical method, was limited for sample selection. In nonexperimental studies of medical interventions, IPTW estimation is now commonly used to control for confounding [20, 21]. Conversely, propensity scores can be used to reduce confounding, that is, matching on the propensity score. This involves forming matched sets of treated and untreated subjects who have similar values of propensity scores [22]. Once these propensity score-matched groups have been formed, the sample size becomes small and a bias that the matched group has been formed based on the smaller group comes into play. Therefore, we adopted the IPTW adjusted analysis to control selection bias in our study. In the weighted variables such as age, presence of liver cirrhosis, AFP, maximum tumor diameter, and intraoperative blood loss, we used a Cox regression model to regress survival between transfused and nontransfused patients, and used a robust variance estimator [7, 23, 24]. Although this model adjusted the variables, including intraoperative blood loss that varied widely in each operation, the recurrence rate of transfused patients significantly remained to be worse than that of nontransfused patients. However, the negative effect of blood transfusion on OS vanished when the background confounding factors were adjusted. Other independent factors, such as ICG15-R, serum AFP levels, and maximum tumor diameter may have strong effects on OS because these factors were the different independent factors of OS from those of RFS. These results suggested that blood transfusions could have a negative effect on the recurrence of HCC after hepatectomy. Therefore, we recommend avoiding blood transfusions during hepatectomy if the patient vital signs are stable and tolerable for hepatectomy when in the presence of considerable blood loss during surgery. Our results show a dramatic decrease in the RFS rate at about 5 years after hepatectomy, as a result of highweighted patients with low propensity scores developing recurrent HCC. Although IPTW analysis minimizes selection bias, it may result in an artificial imbalance of weighted patients [25]. This reflects a limitation of using IPTW analysis to estimate outcomes. In transfused patients, the most reasonable cause of the worsened recurrence rate was the immunosuppressive effect to the host that occurs after a transfusion (transfusion-related immunomodulation: TRIM). Results of experimental animal models suggest that allogeneic blood transfusion-associated immunomodulation is an immunologically mediated biological effect associated primarily with the infusion of allogeneic leukocytes [26]. Thus, TRIM may have negative effects, such as tumor recurrence, owing to allogeneic blood transfusion by immunomodulatory mechanisms. Moreover, TRIM has also been
World J Surg Table 3 Univariate and multivariate analysis of recurrence-free survival Variables
Univariate analysis 5-year survival
P value
Multivariate analysis HR (95 % CI)
P value
Age (years) ^60/[ 60
34.4/34.9
0.93
34.1/36.8
0.44
34.5/36.3
0.67
23.8/44.2
\0.0001
1.41 (1.11–1.82)
0.005
38.3/26.1
0.027
1.23 (0.94–1.59)
0.13
36.7/23.9
0.0014
1.43 (1.04–1.94)
0.03
36.1/25.6
0.038
1.41 (0.96–2.02)
0.078
19.3/39.1
0.0003
1.63 (1.23–2.13)
0.001
43.5/33.4
0.029
1.39 (0.96–2.05)
0.082
31.7/39.6
0.0034
1.30 (1.01–1.69)
0.04
40.3/32.9
0.34
36.0/29.3
0.17
24.3/37.7
0.0024
1.46 (1.10–1.91)
0.009
Gender Male/female Anti-HCVAb positive Positive/negative Liver cirrhosis Yes/no ICG-R15 (%) ^20/[20 AFP (ng/mL) ^400/[400 Maximum tumor diameter (mm) ^50/[50 Multiple tumors Yes/no Histological grading Well/moderately or poorly Microscopic vascular invasion Yes/no Extent of hepatic resection Major/minor hepatectomy Intraoperative blood loss (mL) ^1000/[1000 Blood transfusion Transfused/nontransfused
CI confidence interval, HCVAb hepatitis C virus antibody, ICG-R15 indocyanine green retention rate at 15 min, AFP alpha-fetoprotein
associated with allogeneic blood transfusion by proinflammatory mechanisms. TRIM effects may be mediated by allogeneic mononuclear cells, white blood cell-derived soluble mediators, and soluble HLA peptides circulating in allogeneic plasma [27]. Additionally, iron overload secondary to blood transfusion may have deleterious effects on postoperative patients with HCC. It has become increasingly evident that excess body iron may accelerate the progress of liver fibrosis and recurrence of HCC [28]. According to multiple regression analysis, blood transfusions have a negative effect on postoperative liver function in patients with HCC within 2 years [29]. In conclusion, we found that the RFS rate of transfused patients remained significantly lower than those of nontransfused patients after adjusting the control bias by using IPTW. Our results suggest that tumor recurrence is associated with blood transfusions post-hepatectomy for patients with primary HCC and Child–Pugh class A.
Acknowledgments The authors thank the co-medical staff (O.T., S.Y., and M.T.) in Hiroshima Red Cross Hospital and Atomic Bomb Survivor Hospital for helping with the data collection. Conflict of interest of interest.
The authors declare that they have no conflict
References 1. Forner A, Llovet JM, Bruix J (2012) Hepatocellular carcinoma. Lancet 379(9822):1245–1255 2. Katanoda K, Matsuda T, Nishimoto H et al (2013) An updated report of the trends in cancer incidence and mortality in Japan. Jpn J Clin Oncol 43(5):492–507 3. Shirabe K, Takeishi K, Maehara Y et al (2011) Improvement of longterm outcomes in hepatitis C virus antibody-positive patients with hepatocellular carcinoma after hepatectomy in the modern era. World J Surg 35(5):1072–1084. doi:10.1007/s00268-011-1013-y 4. Asahara T, katayama K, Itamoto T et al (1999) Perioperative blood transfusion as a prognostic indicator in patients with
123
World J Surg
5.
6.
7.
8. 9.
10.
11.
12.
13.
14.
15.
16.
hepatocellular carcinoma. World J Surg 23(7):676–680. doi:10. 1007/PL00012367 Makino Y, Yamanoi A, Nagasue N et al (2000) The influence of perioperative blood transfusion on intrahepatic recurrence after curative resection of hepatocellular carcinoma. Am J Gastroenterol 95(5):1294–1300 Katz SC, Shia J, Dematteo RP et al (2009) Operative blood loss independently predicts recurrence and survival after resection of hepatocellular carcinoma. Ann Surg 249(4):617–623 Austin PC (2012) The performance of different propensity score methods for estimating marginal hazard ratios. Stat Med 32:2837–2849 Makuuchi M, Kosuge T, Kawasaki S et al (1993) Surgery for small liver cancers. Semin Surg Oncol 9(4):298–304 Takeishi K, Shirabe K, Maehara Y et al (2011) Clinicopathological features and outcomes of young patients with hepatocellular carcinoma after hepatectomy. World J Surg 35(5):1063– 1071. doi:10.1007/s00268-011-1017-7 Pugh RN, Murray-Lyon IM, Williams R et al (1973) Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 60(8):646–649 Yamashita YI, Tsuijita E, Maehara Y et al (2013) Trends in surgical results of hepatic resection for hepatocellular carcinoma: 1,000 consecutive cases over 20 years in a single institution. Am J Surg 13:522–529 Poynard T, Bedossa P, Opolon P (1997) Natural history of liver fibrosis progression in patients with chronic hepatitis C. The OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups. Lancet 349(9055):825–832 Dindo D, Demartines N, Clavien PA (2004) Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 240(2):205–213 Rahbari NN, Garden OJ, Padbury R et al (2011) Posthepatectomy liver failure: a definition and grading by the International Study Group of Liver Surgery (ISGLS). Surgery 149(5):713–724 Yamamoto J, Kosuge T, Makuuchi M et al (1994) Perioperative blood transfusion promotes recurrence of hepatocellular carcinoma after hepatectomy. Surgery 115(3):303–309 Hanazaki K, Kajikawa S, Amano J et al (2005) Perioperative blood transfusion and survival following curative hepatic
123
17.
18.
19.
20.
21. 22.
23. 24.
25. 26.
27.
28. 29.
resection for hepatocellular carcinoma. Hepatogastroenterology 52(62):524–529 Forner Al, Reig ME, Bruix J et al (2010) Current strategy for staging and treatment: the BCLC update and future prospects. Semin Liver Dis 30(1):61–74 Bruix J, Sherman M, American Association for the Study of Liver Diseases (2011) Management of hepatocellular carcinoma: an update. Hepatology 53(3):1020–1022 Kuroda S, Tashiro H, Ohdan H et al (2012) No impact of perioperative blood transfusion on recurrence of hepatocellular carcinoma after hepatectomy. World J Surg 36(3):651–658. doi:10. 1007/s00268-012-1425-3 Robins JM, Herna´n MA, Brumback B (2000) Marginal structural models and causal inference in epidemiology. Epidemiology 11(5):550–560 Sato T, Matsuyama Y (2003) Marginal structural models as a tool for standardization. Epidemiology 14(6):680–686 Rosenbaum PR, Rubin DB (1985) Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 39:33–38 Lin DY, Wei LJ (1989) The robust inference for the proportional hazards model. J Am Stat Assoc 84:1074–1078 Joffe MM, Ten Have TR, Kimmel SE et al (2004) Model selection, confounder control, and marginal structural models: review and new applications. Am Stat 58:272–279 Halpern EF (2014) Behind the numbers: inverse probability weighting. Radiology 271(3):625–628 Blajchman MA (1998) Immunomodulatory effects of allogeneic blood transfusions: clinical manifestations and mechanisms. Vox Sang 74(Suppl 2):315–319 Vamvakas EC, Blajchman MA (2007) Transfusion-related immunomodulation (TRIM): an update. Blood Rev 21(6): 327–348 Kew MC (2014) Hepatic iron overload and hepatocellular carcinoma. Liver Cancer 3(1):31–40 Itasaka H, Yamamoto K, Matsumata T et al (1995) Influence of blood transfusion on postoperative long-term liver function in patients with hepatocellular carcinoma. Hepatogastroenterology 42(5):465–468
