Perioperative Blood Transfusion as a Poor Prognostic Factor After Aggressive Surgical Resection for Hilar Cholangiocarcinoma

Background Blood transfusion is linked to a negative outcome for malignant tumors. The aim of this study was to evaluate aggressive surgical resection for hilar cholangiocarcinoma (HCCA) and assess the impact of perioperative blood transfusion on long-term survival. Methods Sixty-six consecutive major hepatectomies with en bloc resection of the caudate lobe and extrahepatic bile duct for HCCA were performed using macroscopically curative resection at our institute from 2002 to 2012. Clinicopathologic factors for recurrence and survival were retrospectively assessed. Results Overall survival rates at 1, 3, and 5 years were 86.7, 47.3, and 35.7 %, respectively. In univariate analysis, perioperative blood transfusion and a histological positive margin were two of several variables found to be significant prognostic factors for recurrence or survival (P < 0.05). In multivariate analysis, only perioperative blood transfusion was independently associated with recurrence (hazard ratio (HR) = 2.839 (95 % confidence interval (CI), 1.370–5.884), P = 0.005), while perioperative blood transfusion (HR = 3.383 (95 % CI, 1.499–7.637), P = 0.003) and R1 resection (HR = 3.125 (95 % CI, 1.025–9.530), P = 0.045) were independent risk factors for poor survival. Conclusions Perioperative blood transfusion is a strong predictor of poor survival after radical hepatectomy for HCCA. We suggest that circumvention of perioperative blood transfusion can play an important role in long-term survival for patients with HCCA.


Introduction
Surgical resection 1,2 or liver transplantation 3 remains the only curative treatment that can offer long-term survival for hilar cholangiocarcinoma (HCCA). HCCA involves the confluence of the bilateral hepatic ducts where the main portal and hepatic arterial branches are close, and thus major hepatectomy is often required to obtain a disease-free margin status. Simultaneous caudate lobectomy with extended resection of the hilar bile ducts is now the procedure of choice because the bile ducts of the caudate lobe are direct tributaries from the bilateral hepatic ducts. 4,5 In fact, several reports have shown that hepatectomy with en bloc resection of the caudate lobe and extrahepatic bile duct, 6,7 lymph node dissection, 8 or resection of portal vein 9,10 or hepatic artery 11,12 for HCCA is more likely to result in disease-free resection. Preoperative percutaneous transhepatic portal vein embolization (PTPVE) has also helped to make major hepatectomy safer and applicable to more patients. 13 -18 Despite these preoperative and surgical advances for achieving disease-free resection, many patients still have microscopic positive resection margins. This is of concern because margin status is a critical prognostic factor and cases with microscopic invasive carcinoma at the ductal margin have poor survival. 6,7,16,17,19 However, aggressive and complex procedures such as hepatectomy with en bloc resection of the caudate lobe and extrahepatic bile duct for HCCA cause excessive intraoperative loss of blood that requires perioperative blood transfusion, which may also result in negative outcomes for malignant tumors of the liver. 20 However, the effects of blood transfusion on survival after aggressive surgery for HCCA have not been widely studied.
In this study, we surveyed retrospective data in 66 consecutive hepatectomies with en bloc resection of the caudate lobe and extrahepatic bile duct for HCCA performed at our institute. The main aim was to identify prognostic factors for recurrence and long-term survival after aggressive surgery for HCCA, with a particular focus on perioperative blood transfusion.

Patients Between January 2002 and December 2012, 108 patients with H C C A w e r e a d m i t t e d t o t h e D e p a r t m e n t o f
Gastroenterological Surgery at Hirosaki University Hospital. Nineteen patients did not undergo any surgery due to highly advanced unresectable disease or poor liver function during preoperative workup. Laparotomy was conducted in the remaining 89 patients. Twenty-three of the patients were excluded because three patients underwent minor hepatic resection (only segment 4) with caudate lobe and extrahepatic bile duct; one patient had R2 (macroscopic residual tumor) despite left hepatectomy with caudate lobe and extrahepatic bile duct; nine patients underwent palliative bile duct resection; two patients underwent palliative pancreatoduodenectomy; and eight patients were not able to undergo resection due to liver metastases or intraoperative detection of highly local advanced disease. The remaining 66 patents underwent major hepatectomies (hemihepatectomy, central bisectionectomy, or more extensive resection) with en bloc resection of the caudate lobe and extrahepatic bile duct in addition to macroscopically curative resection (R0 or R1 resection) were enrolled in this study. HCCA was defined as a tumor involving the primary ductal confluence. Tumors invading the hepatic hilar region but located predominantly in the liver parenchyma and gall bladder were not included in the definition of HCCA. All clinical information was collected based on prospectively maintained data and analyzed retrospectively. Demographic data included information on gender, age, mode of presentation, pre-resection interventions, and investigations. Preoperative laboratory values included serum carbohydrate antigen 19-9 (CA19-9) as a tumor marker and indocyanine green retention value at 15 min (indocyanine green (ICG)-R15) as a marker of hepatic function reserve. Operative records were reviewed for operative details, operative time, intraoperative estimated blood loss, and blood transfusion requirements.

Preoperative Evaluation and Workup
The location and extent of the disease were evaluated by ultrasonography, enhanced computed tomography (CT), magnetic resonance imaging (MRI) with magnetic resonance cholangiopancreatography (MRCP), endoscopic retrograde cholangiography (ERC), and percutaneous transhepatic cholangiography (PTC). Enhanced CT was performed before biliary drainage, which in turn was performed if obstructive jaundice (total bilirubin (T-bil) >3 mg/dL) was present. Endoscopic biliary drainage (EBD) was preferred, but percutaneous transhepatic biliary drainage (PTBD) was performed when EBD was not successful. Biliary drainage was performed primarily for the future liver remnant (FLR) lobe; however, in cases of bilateral biliary sepsis, both lobes were drained. Biliary drainage was undertaken until T-bil decreased to <2 mg/dL. In patients with cholangitis, surgery was postponed until alleviation of inflammation. Enhanced CT of the chest, abdomen, and pelvis was used to identify metastatic lesions. More recently, positron emission tomography (PET) has also been used for this purpose. Unresectable disease was defined as the presence of extensive bilobular metastases or extrahepatic metastases other than regional lymph node involvement or peritoneal dissemination.
The total liver volume (TLV) and the part of the hepatic segment to be resected were calculated using CT volumetry. 21 PTPVE was indicated when the FLR was estimated to be <40 % and was performed 3 to 4 weeks before scheduled liver resection, since PTPVE has been suggested to be useful for inducing compensatory hypertrophy of the FLR. 13 All patients underwent CT within 4 weeks after PTPVE. ICG tests were performed in all cases before surgery. For patients who underwent PTPVE, ICG tests were performed just before and 3 to 4 weeks after PTPVE. ICG (0.5 mg/kg body weight) was administered via a peripheral vein and venous blood was sampled before and 5, 10, and 15 min after injection. The plasma clearance rate of ICG (KICG) was calculated by linear regression analysis of plasma ICG concentrations. 22 Total liver function was evaluated based on KICG, and the KICG of the FLR was calculated using the following formula: KICG×FLR volume/TLV. 18

Surgical Technique
All operations were performed after T-bil had decreased to <2 mg/dL. Preoperative autologous blood donation was performed in patients with preoperative hemoglobin level of >11 g/dL and no hypotension or severe cardiac disease, at the discretion of the anesthesiologist. Firstly, en bloc dissection of the regional lymph nodes surrounding the hepatoduodenal ligament, behind the pancreatic head, and around the common hepatic artery was uniformly performed, followed by skeletonization of the hepatic hilus. After complete mobilization of the hepatic lobe to be resected, the caudate lobe was completely separated from the inferior vena cava. Liver transection was carried out using CUSA (Aloka, Tokyo, Japan) or the forceps clamp-crushing method based on the surgeon's preference during hepatic artery and portal vein clamping for 15 min with 5-min intervals, which is known as Pringle's maneuver. All patients underwent hemihepatectomy, central bisectionectomy, or more extensive resection with en bloc resection of the caudate lobe and extrahepatic bile duct. Biliary continuity was reconstructed with hepaticojejunostomy using a Roux-en-Y jejunal limb brought up in a transmesenteric fashion. Combined pancreatoduodenectomy was performed if the tumor extended below the intrapancreatic portion of the distal bile duct. 23 Reconstruction for hepatopancreatoduodenectomy was conducted by a modified Child's method with an end-to-side anastomosis. Concomitant portal vein resection and reconstruction have been applied aggressively in cases with suspected macroscopic invasion during surgery. 9,10 Concomitant hepatic artery resection and reconstruction have been similarly applied, although this is still controversial. 11,12,24 -26 All autologous blood collected preoperatively was transfused intraoperatively. Allogeneic red blood cells and fresh frozen plasma (FFP) were transfused intraoperatively at the discretion of the anesthesiologist.

Histopathological Evaluation
Pathology reports were reviewed to determine tumor histological grade, margin status, and the presence of microvascular, lymphovascular and perineural invasion. Tumors were staged using the tumor-node-metastasis (TNM) Classification of Malignant Tumors of the International Union Against Cancer (7th edition, 2009) for proximal extrahepatic bile duct cancer. 27 Therefore, intrahepatic metastasis, peritoneal deposits and positive para-aortic nodes were classified as distant metastasis (M1). R0 resection was defined as negative ductal and radial margins on histology; R1 resection as the histological presence of a tumor at any margin; and R2 resection as the macroscopic presence of a tumor at any margin or M1. Histopathological evidence of invasive carcinoma, dysplastic lesions including carcinoma in situ (CIS), or normal epithelium was recorded for proximal and distal ductal margins. Radial margins were assessed in the resected specimen only. Dysplastic lesions and CIS were defined as R0 resection because it is difficult to distinguish CIS from high-grade dysplastic lesions pathologically, 28 and the presence of CIS at the ductal margin has no impact on survival compared with a negative ductal margin in extrahepatic cholangiocarcinoma. 29 -34 Postoperative Course Patients were not routinely admitted to the intensive care unit postoperatively but were monitored overnight in the general ward. In principle, allogeneic red blood cell transfusion was performed for asymptomatic patients if the hemoglobin level fell to <6.5 g/dL. However, patients with additional specific risk factors such as severe cardiac disease or hypotension often received blood transfusion when the hemoglobin level fell to <7 g/dL. FFP was used at the discretion of the attending surgeon and was most commonly administered for prothrombin time (PT) <60 %. Liver function tests were obtained routinely on postoperative days 1, 3, and 6 and as clinically indicated thereafter. Hepatic failure was defined as T-bil of >7.0 mg/dL postoperatively, and hyperbilirubinemia was defined as T-bil of >5.0 mg/dL. 35 Hence, the postoperative maximum T-bil and minimum PT were assessed. Postoperative complications were defined and graded according to the validated Clavien classification system 36 as grades 1 and 2 and grades 3-5 for minor and major complications, respectively. Postoperative mortality was defined as death as an inpatient or within 30 days of surgery.

Adjuvant Treatment
Since 2008, all patients except UICC stage I cases were scheduled to receive S-1, an oral fluoropyrimidine derivative, as adjuvant chemotherapy for about 1 year following surgery, whereas those prior to 2007 did not receive chemotherapy.

Follow-up After Surgical Resection
Patients were followed regularly at the outpatient clinic every 1-3 months. Clinical examinations were performed, tumor markers (carcinoembryonic antigen and CA19-9) and liver function were checked monthly, and CT scans of the chest, abdomen, and pelvis were performed every 3 months until year 2. Thereafter, clinical examinations, tumor marker tests, liver function tests, and CT scans were carried out every 6 months until year 5. Subsequently, patients attended a clinic for an annual examination. MRI and PET-CT were carried out if recurrence was suspected in routine follow-up.

Survival and Prognostic Factors
The following data were reviewed to identify prognostic factors: (1) preoperative clinical factors: gender, age, jaundice, preoperative CA19-9 value, preoperative ICG-R15, FLR Disease-free survival (DFS) was defined as the time from the operation to the date of disease recurrence. Overall survival (OS) was defined as the time from the operation to the time of death due to recurrence or the last follow-up time.

Statistical Analysis
For variables associated with perioperative blood transfusion, continuous variables are expressed as median and interquartile range and were compared by Mann-Whitney U test, and categorical variables were compared by Pearson Chi-square test with a Fisher exact test or Yates continuity correction where appropriate in univariate analysis. Variables with a significant relationship with perioperative blood transfusion in univariate analysis were used in a multivariate logistic regression model. DFS and OS were calculated by the Kaplan-Meier method, and differences were evaluated by log-rank test. Only variables that were significant in univariate analysis were included in multivariate analysis to identify independent predictors of DFS and OS. This analysis was performed using a Cox proportional hazards regression model. P < 0.05 was considered significant in all analyses. All analyses were performed using PASW Statistics ver.18.0 for Windows (SPSS, Inc., Chicago, IL, USA).

Discussion
HCCA is a rare cancer with an estimated incidence of 0.8 and 1.2 per 100,000 in females and males, respectively. Recent developments in surgical techniques and perioperative management have led to a significant improvement in the resectability rate and survival for HCCA, but morbidity and a poor prognosis are still of concern. However, complete surgical removal of cancer tissues by surgical resection or liver transplantation is the only curative treatment, since there is no evidence of efficacy of chemotherapy or radiotherapy. In a recent review of reports of surgical resection for HCCA, Ito et al. 37   to 95 % and 11 to 45 %, respectively, with morbidity of 14 to 76 % and mortality of 0 to 19 %. In our study, the curative resection rate using major hepatectomy with en bloc resection of the caudate lobe and extrahepatic bile duct was 81.8 %, which is relatively favorable compared with previous studies. Similar, we found a somewhat favorable 5-year survival rate of 35.7 %, morbidity of 47 %, and mortality of 1.5 % in the current series. Many clinicopathological factors, including positive margin status, lymph node involvement, higher T-stage, moderate or poorly differentiated tumor grade, perineural invasion, microvascular invasion and lymphovascular invasion, can have a negative impact on long-term outcome after resection for HCCA. 6,7,16,17,19,38 -46 Of these factors, a positive margin status has frequently been identified as a risk factor associated with reduced survival in patients with HCCA. Complete resection with negative histologic margins is the only modifiable factor among potential risk factors, and thus is the primary goal of surgical treatment. 37 Aggressive surgery with concomitant major hepatic resection is thought to be more effective for acquisition of a negative margin, compared with local resection. 6,7,42,46,47 Consistent with this view, aggressive surgery in this study resulted in a relatively high R0 resection rate of 81.8 %, and cases with R0 resection had a significantly better prognosis compared with those with R1 resection.
The current study also indicated that a requirement for perioperative blood transfusion was a strong independent risk factor for both recurrence and poor survival after surgical resection for HCCA. In addition, intraoperative blood loss was significantly associated with blood transfusion requirements. Allogeneic blood transfusions are known to induce host immunosuppression. Some older reports have suggested some positive effects of blood transfusion, including a beneficial effect on graft prolongation after renal transplantation 48 and reduction of recurrence of Crohn's disease. 49 In contrast, since Burrows et al. first reported that blood transfusion was associated with adverse oncologic outcomes after resection of colorectal cancer, 50 many negative effects of blood transfusion In liver surgery, there is a high incidence of blood transfusion because intraoperative blood loss is often large. Immunomodulation due to perioperative blood transfusion has been linked to cancer recurrence, particularly in surgery for hepatocellular carcinoma 66 -71 and colorectal liver metastases. 72,73 However, developments in surgical techniques, perioperative management, and anesthetic protocols have markedly improved outcomes after liver surgery. 74,75 Nevertheless, the procedure is often accompanied by substantial intraoperative blood loss and a requirement for blood transfusion perioperatively or in the postoperative course, especially in extended hepatectomy or surgery for liver cirrhosis.
Few studies have focused on the negative impact of blood transfusion on outcome after liver resection for HCCA. This type of hepatectomy is more complex and difficult to perform and more risky than that for hepatocellular carcinoma or liver metastasis because concomitant extrahepatic bile duct resection with lymph node dissection and biliary reconstruction are required for HCCA. This can lead to a large amount of blood loss and a need for blood transfusion in comparison with simple hepatectomy. Thus, the median blood loss in the current study was 1778.5 ml, which is a relatively large amount. A few studies have found that perioperative blood transfusion is a negative predictor for morbidity or mortality after resection for HCCA. 76 -78 The association between blood transfusion and long-term survival after resection for HCCA has only been examined in two previous studies, 79,80 as far as we are aware. In 40 patients undergoing surgical resection for HCCA, Liu et al. 79 found a significant association between blood transfusion and poor survival in univariate analysis, but blood transfusion could not be confirmed as an independent predictor in multivariate analysis. In a recent study of 83 patients with HCCA, Young et al. 80 found that perioperative blood transfusion was a significant independent determinant of reduced survival after surgery by multivariate analysis. Similarly, in the present study, perioperative blood transfusion was a strong independent predictor for recurrence and poor survival after liver resection for HCCA. Thus, this study might be thought of as the second to show a significant independent correlation between blood transfusion and prognosis in patients with HCCA. One difference between this study and that of Young et al. 80 is that all of our patients underwent major hepatectomy with caudate lobe and extrahepatic bile duct resection, whereas Young et al. included patients who underwent local resection alone, in addition to cases treated with aggressive surgery. 80 The mechanism underlying the adverse effect of blood transfusion is unclear, but experimental and clinical studies have demonstrated that blood transfusion suppresses host immunity via reduction of natural killer cell activity and cytotoxic T cell function, increases suppressor T cell activity, and decrease helper/suppressor (T4/T8) lymphocyte ratios. 81,82 Also, leaching of biologically active substance from cells into stored blood products occurs due to normal physiological aging and metabolic processes, and these leached bioactive substances have immunomodulatory effects that promote cell growth and angiogenesis, and may therefore have a direct effect on tumor growth. 83 For this reason, the immunosuppressive effect of blood transfusion may play a major role in recurrence of a malignant tumor. Goeppert et al. recently showed that the presence of both intratumoral T and B lymphocytes is correlated with longer survival in cholangiocarcinoma, and that the prognosis was linked to inflammation. These data provide a solid basis for the understanding of the biological role of inflammatory infiltrates in cholangiocarcinoma and for functional and clinical studies exploring modulation of the inflammatory response in cholangiocarcinoma patients. 84 The results indicate that an immunosuppressive status may affect recurrence and survival in patients with cholangiocarcinoma, even though the adverse effects of blood transfusion on host immunity are still unclear.
The current study has several limitations. First, it was performed retrospectively and not as a randomized controlled study, and included a relatively small number of patients who received blood transfusion. However, it is very difficult to design a prospective randomized trial to investigate the relationship between blood transfusion and HCCA outcome. Second, we did not analyze the association between transfusion quantity and survival because the small sample size of patients receiving blood transfusion may have resulted in a loss of statistical power. Third, we also did not analyze the influence of perioperative FFP transfusion on HCCA prognosis because almost all patients received administration of FFP during surgery or within at least 7 days postoperatively. In summary, we identified perioperative blood transfusion and a histologic positive margin as strong independent predictors for a poor prognosis following aggressive surgery with concomitant major hepatectomy for HCCA. Thus, circumvention of perioperative blood transfusion is likely to play an important role in long-term survival for patients with HCCA. It is widely believed that an aggressive surgical approach to gain a negative margin is associated with a favorable prognosis. However, better long-term survival may be achieved through vigilant surgery to reduce blood loss and avoid blood transfusion when possible. More effective preoperative or postoperative adjuvant therapy and surgery for oncological control of advanced HCCA is also required to improve future outcomes.