Journal of Surgical Oncology 2014;110:585–591
Outcomes in Biliary Malignancy BAS GROOT KOERKAMP, MD PhD1 AND YUMAN FONG, MD2* 1
Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands 2 City of Hope Medical Center, Duarte, California
The biliary malignancies that are reviewed here are gallbladder cancer (GBC), intrahepatic cholangiocarcinoma (IHC), and perihilar cholangiocarcinoma (PHC). The focus is on outcomes after potentially curative resection of biliary malignancies. Key outcomes are postoperative mortality, median and 5‐year overall survival (OS), recurrence‐free survival, and recurrence patterns. Poor prognostic factors for recurrence and survival as well as prognostic models are also discussed. The incidence of biliary malignancies in the United States is about 5 in 100,000. Postoperative mortality for resection of GBC and IHC is similar to that of liver resections for other indications. However, 90 day postoperative mortality after liver resection for PHC is about 10%. For GBC, median OS depends strongly on the T‐stage and ranges from 8 months (pT3) to 79 months (pT1b). Median OS after resection for IHC is about 30 months, and for PHC about 38 months. The majority of patients with biliary malignancies develop a recurrence after resection. Patients with GBC recur early with a median time to recurrence of 12 months, versus about 20 months for IHC and PHC. In patients with resected IHC or PHC locoregional recurrence was the only site of recurrence in about 60% of patients, versus 15% in patients with GBC. Poor prognostic factors after resection of all biliary malignancies include the presence of lymph node metastasis, a positive surgical resection margin, and moderate or poor tumor differentiation. Several prognostic nomograms have been developed to predict long‐ term outcomes of biliary cancer resection.
J. Surg. Oncol. 2014;110:585–591. ß 2014 Wiley Periodicals, Inc.
KEY WORDS: klatskin; cholangiocarcinoma; bile duct cancer; gallbladder cancer
INTRODUCTION Biliary malignancy includes tumors of the bile ducts (cholangiocarcinoma) and tumors of the gallbladder. Three subtypes of cholangiocarcinoma are distinguished, based on where in the bile duct the cancer arises. Distal cholangiocarcinoma (DC) is arising from the common bile duct, between the confluence of the cystic duct with the common hepatic duct, and the ampulla of Vater. Perihilar cholangiocarcinoma (PHC) is arising from the common hepatic duct distal to the cystic duct confluence, the right or left hepatic duct, or the second‐order bile ducts. The second‐order bile ducts are the bile ducts that come together to form the right or left hepatic duct. Intrahepatic cholangiocarcinoma (IHC) is a tumor within the liver, arising from the bile ducts proximal to the second‐order bile ducts. Cancers arising from the cystic duct are not considered cholangiocarcinoma but rather are classified as gallbladder cancer (GBC). These four biliary malignancies each have a separate American Joint Committee on Cancer (AJCC) staging system since the most recent 7th edition [1]. In previous editions, PHC and DC were staged together as extrahepatic cholangiocarcinoma, and IHC was staged together with hepatocellular carcinoma. The annual incidence of biliary malignancies in the Western world is about 5 per 100,000 with about 12,000 new patients each year in the United States [2]. GBC is the most common biliary cancer. For patients with cholangiocarcinoma who undergo surgical exploration, PHC is most common (50%), followed by DC (40%), and IHC (10%) [3]. However, the incidence of IHC is increasing [4]. Unfortunately, most patients with biliary malignancies have locally advanced or distant metastatic disease at the time of presentation [5]. These patients are unlikely to benefit from a surgical resection. The focus of this review is on outcomes of patients who underwent a potentially curative resection. DC is not covered in this review, because it’s diagnostic work‐up, treatment, and outcomes are more similar to other periampullary cancers such as pancreatic cancer. Outcomes of interest include postoperative mortality, overall survival (OS), recurrence‐free survival (RFS), and sites of recurrence. Outcomes may differ depending
ß 2014 Wiley Periodicals, Inc.
on prognostic factors such as demographics, tumor markers, imaging characteristics, or pathology findings. A prognostic model or risk score combines important prognostic factors to predict outcomes such as 5‐year OS. Accurate ascertainment of outcomes in biliary malignancy is a prerequisite for clinical auditing and continued quality improvement. Prediction of outcomes using prognostic factors and models can guide and improve patient management. For example, a patient with an excellent predicted RFS after potentially curative resection of a biliary malignancy is unlikely to benefit from adjuvant treatment. Outcomes for patients with biliary malignancies are available from several sources: registries (e.g., SEER), patient series, or randomized controlled trials. Outcomes from registries and patient series constitute the main source of this review, because few randomized controlled trial for patients with biliary malignancies have been published.
INTRAHEPATIC CHOLANGIOCARCINOMA Diagnosis and Treatment The annual incidence of IHC in the United States is about 1 per 100,000. At an early stage patients are typically asymptomatic. At later stages, patients present with non‐specific symptoms such as weight loss, malaise, and abdominal discomfort [6]. Jaundice is less common in patients with IHC. Tumor markers have suboptimal sensitivity and specificity for the diagnosis of IHC: a CA 19‐9 of more than 100 U/ml has a sensitivity of only 53% [7]. A diagnosis of IHC can typically be
*Correspondence to: Yuman Fong, MD, Department of Surgery, City of Hope Medical Center, 1500 East Duarte Road, Duarte, CA 91010, USA. Fax: þ1 626 471 9212. E‐mail: [email protected] Received 12 June 2014; Accepted 29 July 2014 DOI 10.1002/jso.23762 Published online in Wiley Online Library (wileyonlinelibrary.com).
586
Groot Koerkamp and Fong
made on imaging after ruling out extrahepatic primary malignancies, in particular colon cancer. On CT, IHC is characterized by a hypodense mass in the non‐contrast phase, and peripheral enhancement in the arterial phase with progressive enhancement on later phases. On MRI, IHC is hypointense on T1‐ and hyperintense on T2‐weighted images. Most tumors (85%) are mass‐forming on imaging; the other morphological subtypes are periductal which grows along the bile ducts without forming a clear mass, and intraductal which grows within the bile duct. A biopsy is not required prior to surgical resection, but is recommended prior to chemotherapy. FDG‐PET was found to change surgical management in about 25% of patients [8]. Resection is recommended in fit patients without extrahepatic disease on imaging, in whom all disease can be removed with negative margins and an adequate liver remnant volume [6]. Patients with locally advanced or distant metastases should be considered for gemcitabine and cisplatin [9]. Patients with unresectable disease confined to the liver can also be considered for locoregional therapy such as transarterial chemo‐embolization (TACE) or hepatic artery infusion pump chemotherapy (HAIP) [10].
Postoperative Mortality Resection for IHC requires at least a hemihepatectomy in about 75% of patients, and a bile duct resection and reconstruction in about 25% of patients [11]. Nevertheless, in specialized high volume centers, the mortality rates can be as low as 1.2% [12].
Overall Survival The prognosis of patients with unresectable IHC is poor with about 5 to 10% of patients alive at 5 years after diagnosis [13]. Between 1983 and 2010, only about 15% of patients with IHC in the SEER database underwent a liver resection [14]. After potentially curative resection the 5‐year OS ranged from 21% to 35% in the five largest series (both Western and Asian) with each more than 200 patients; the median OS ranged from 18 to 39 months [11,15–18]. When one outlier of these five studies was excluded [16], the 5‐year overall OS was on average about 32%, and the median OS about 30 months. Outcomes of liver transplantation in patients with IHC are worse than for PHC. Even in highly selected patients the 5‐year OS was only about 30%, mainly due to early recurrence [19,20]. However, a more recent matched cohort study showed that cirrhotic patients with uninodular IHC smaller than 2 cm and patients with hepatocellular cholangiocarcinoma have similar long‐term survival as patients transplanted for hepatocellular carcinoma [21].
Recurrence Free Survival and Recurrence Patterns Recurrence free survival (RFS) varied widely in a recent systematic review: 5‐year RFS varied from 2% to 39%, and the median RFS from 7 to 34 months [22]. Studies that found a low incidence of recurrence had a shorter follow‐up or performed less frequent postoperative imaging to detect recurrence. The liver was the most common site of initial recurrence. In a multi‐institutional series of 301 patients, 54% had developed a recurrence with a median RFS of 20 months [23]. The initial recurrence was intrahepatic only in 61%, extrahepatic only in 21%, and both in 19%. Extrahepatic recurrence most commonly involved lung, peritoneum, or lymph nodes.
Poor Prognostic Factors Several poor prognostic factors after potentially curative resection have been identified and confirmed in multiple studies. A pooled analysis of these prognostic factors was recently performed using seven studies including more than 2,000 patients [22]. The most important Journal of Surgical Oncology
prognostic factors for OS in univariate analyses are presented in Table I. The prognostic value of a positive surgical margin was not statistically significant in this pooled analysis. However, another study showed that a positive surgical margin is a strong poor prognostic factor with a hazard ratio (HR) of 5.55 (95% CI: 3.57–8.33) [15]. Independent poor prognostic factors in more than one study were: multiple tumors, vascular invasion, lymph node metastasis, large tumor size, and high serum CA19‐9. The presence of poor prognostic factors does not preclude the possibility of long‐term survival. Patients with lymph node metastasis, multiple tumors, or vascular invasion had a median OS of about 20 months; however, 5‐year OS was 27% in the presence of one of these factors and 18% in the presence of two or three risk factors [11]. Independent risk factors for recurrence in a large multi‐institutional study were macrovascular invasion, lymph node metastasis, and tumor size more than 5 cm [23].
Prognostic Models The AJCC staging system for IHC is based on the following poor prognostic factors: multiple tumors, vascular invasion, direct invasion of extrahepatic structures, periductal invasion (versus mass‐forming lesions), lymph node metastasis, and distant metastasis. Stage I tumors are solitary without vascular invasion; stage II involves multiple tumors or a tumor with vascular invasion; stage III tumors demonstrate direct invasion of extrahepatic structures; and stage IV tumors have periductal invasion, or nodal or distant metastatic disease. Figure 1 presents stage‐specific survival for patients with IHC. Several prognostic models for patients with IHC have recently been published. Wang et al. from Hong Kong created a nomogram to predict 3‐year and 5‐year OS based on 367 patients with resected IHC [18]. The nomogram consisted of seven prognostic factors: multiple tumors, vascular invasion, lymph node metastasis, tumor diameter, direct invasion of adjacent structures or local metastasis, and tumor markers (CEA and CA19‐9). The c‐statistic of this model was 0.74, which was superior to other models including the AJCC staging system. Hyder et al. used a multinational database of 514 patients with resected IHC to predict 3‐year and 5‐year OS [24]. The prognostic factors in this nomogram were: multiple tumors, vascular invasion, lymph node metastasis, tumor diameter, age, and cirrhosis. These factors were also part of the nomogram of Wang et al., except for age and cirrhosis. The c‐statistic of this model was 0.69, again superior to the AJCC staging system. Jiang et al. from Shanghai developed a prognostic score to predict 5‐ year OS based on 344 patients with resected IHC [16]. The score only required prognostic factors that are available prior to resection: multiple tumors, tumor diameter, tumor boundary on imaging (distinct versus obscure), alkaline phosphatase, and CA 19‐9. Based on this score, patients were categorized in groups that differed in 5‐year OS: low risk (49%), intermediate risk (26%), high risk (10%), and extremely high risk (0%). All three prognostic models require external validation. Only the model of Jiang et al. could inform the decision whether or not to proceed TABLE I. Pooled Univariate Analysis of Poor Prognostic Factors for OS After Resection of IHC (n ¼ 2132) [22] Prognostic factor
HR
HR 95% confidence interval
Multiple tumors Vascular invasion Lymph node metastasis Large tumor size Poor tumor differentiation Older age Positive surgical margin
1.70 1.87 2.09 1.09 1.41 1.10 1.06
1.43–2.02 1.44–2.42 1.80–2.43 1.02–1.16 1.17–1.71 1.03–1.17 0.49–2.32
Surgery for Biliary Cancer
587
Several controversies remain regarding the optimal management of patients with GBC. A preoperative PET scan may avoid futile surgery in some patients [30]. Distant metastasis was found in 3 out of 23 patients with incidental GBC [8]. Excision of port‐sites at the time of definitive resection for GBC has been recommended because of wound recurrences. However, resection of port‐sites is a disfiguring procedure, and no survival benefit was found in a large study [31]. Anatomical segment 4B and 5 resection (instead of a 2–3 cm wedge resection) has been recommended, but two studies found no survival difference. [32,33] Open resection of GBC is the current standard of care, although some have recommended laparoscopic or robotic resection for GBC [34,35].
Postoperative Mortality In the 80s and 90s major liver resections, combined with extrahepatic bile duct resections in jaundiced patients were commonly performed for GBC patients. This contributed to a postoperative mortality of about 4% in high volume centers [36]. More recently, a 2 cm wedge resection of segment 4B and 5 is typically performed in non‐jaundiced patients without a bile duct resection. The mortality for such minor liver resections is about 1% in high‐volume centers, similar to liver resections for other indications. In a SEER study, however, the postoperative mortality was 4% [37]. Fig. 1. Stage‐specific (AJCC, 7th edition) overall survival after resection of IHC (n ¼ 212) [15]. The control group includes patients without resection or with an R2 resection. Reproduced with permission from Wolters Kluwer Health.
with a potentially curative resection. For example, patients with an extremely high risk score may be unlikely to benefit from a resection. The two other models require prognostic factors that are only available after resection.
GALLBLADDER CANCER Diagnosis and Treatment The annual incidence of GBC in the United States is about 3 per 100,000. GBC is found at pathologic review in about 1% of patients undergoing laparoscopic cholecystectomy for cholelithiasis [25,26]. These incidental GBCs represent about two‐thirds of patients with potentially curable GBC. At a more advanced stage, patients with GBC typically present with right upper quadrant pain, weight loss, and nausea. About 40% of these patients are jaundiced at the time of presentation due to involvement of the common bile duct. On cross‐sectional imaging a gallbladder and/or liver mass is typically seen in symptomatic patients. GBC arising from a gallbladder polyp represents only a small proportion of GBC patients. Staging for all patients with GBC involves a CT or MRI of the chest, abdomen, and pelvis. Patients with distant metastases, including nodal metastases beyond the hepatoduodenal ligament, are unlikely to benefit from a surgical resection. The same is true for patients with a T4 tumor, involving the main portal vein, common hepatic artery, or at least two extrahepatic organs. Patients presenting with jaundice are also unlikely to benefit from a resection [27]. These patients with locally advanced or distant metastases should be considered for gemcitabine and cisplatin [9]. All other patients, who are fit for surgery, are expected to benefit from a potentially curative (non‐anatomical) 2–3 cm wedge resection of segments 4B and 5 of the liver with a lymphadenectomy of the hepatoduodenal ligament [28]. Re‐resection is recommended after diagnosis of an incidental GBC, except for patients with T1a disease who rarely develop a recurrence [29]. Journal of Surgical Oncology
Overall Survival In a natural history study of patients with GBC, the median OS was 5 months [38]. A study using SEER data evaluated treatments and outcomes for patients with resectable GBC (after excluding patients with M1 or T4 disease). Only 16% of GBC patients underwent a radical resection including a liver resection (as recommended by the NCCN guidelines) [28], instead of a simple cholecystectomy alone [37]. They concluded that compliance with NCCN guidelines was poor. Patients had a superior median OS after a radical resection versus a simple cholecystectomy for both T2 tumors (53 versus 16 months) and T3 tumors (11 versus 8 months) [37]. In a German registry study, the superior median OS after radical resection for T1b tumors was 79 versus 42 months [39]. Due to the retrospective nature of these studies, however, the differences in OS could be partly attributable to differences between patients at baseline (i.e., selection bias).
Recurrence Free Survival and Recurrence Patterns The median time to recurrence after a potentially curative resection for GBC was only 12 months (as compared to 20 months for PHC) [40]. Of those patients that recurred, 85% had a distant recurrence with or without a concomitant locoregional recurrence; only 15% had a locoregional recurrence without distant recurrence. The peritoneum was among the sites of recurrence in about 50% of patients that recurred. Other common sites of recurrence included the liver, lung, and abdominal wall. Extrahepatic bile duct resection in non‐jaundiced GBC patients has been recommended to avoid bile duct recurrences. However, an isolated recurrence at the extrahepatic bile duct was never found in 26 non‐jaundiced patients who underwent a radical resection without extrahepatic bile duct resection [41]. Risk factors for recurrence are positive lymph node status, positive resection margin, and moderate or poor tumor differentiation. Patients with T2 tumors at the hepatic (versus the peritoneal) side of the gallbladder are more likely to recur in the liver, even after radical resection [42].
Poor Prognostic Factors Several poor prognostic factors have been identified and confirmed in multiple studies. Table II presents independent poor prognostic factors
588
Groot Koerkamp and Fong
TABLE II. Independent Poor Prognostic Factors for Disease‐Specific Survival After Resection of GBC (n ¼ 122) [43] Prognostic factor
HR
HR 95% confidence interval
Lymph node metastasis
Outcomes in Biliary Malignancy BAS GROOT KOERKAMP, MD PhD1 AND YUMAN FONG, MD2* 1
Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands 2 City of Hope Medical Center, Duarte, California
The biliary malignancies that are reviewed here are gallbladder cancer (GBC), intrahepatic cholangiocarcinoma (IHC), and perihilar cholangiocarcinoma (PHC). The focus is on outcomes after potentially curative resection of biliary malignancies. Key outcomes are postoperative mortality, median and 5‐year overall survival (OS), recurrence‐free survival, and recurrence patterns. Poor prognostic factors for recurrence and survival as well as prognostic models are also discussed. The incidence of biliary malignancies in the United States is about 5 in 100,000. Postoperative mortality for resection of GBC and IHC is similar to that of liver resections for other indications. However, 90 day postoperative mortality after liver resection for PHC is about 10%. For GBC, median OS depends strongly on the T‐stage and ranges from 8 months (pT3) to 79 months (pT1b). Median OS after resection for IHC is about 30 months, and for PHC about 38 months. The majority of patients with biliary malignancies develop a recurrence after resection. Patients with GBC recur early with a median time to recurrence of 12 months, versus about 20 months for IHC and PHC. In patients with resected IHC or PHC locoregional recurrence was the only site of recurrence in about 60% of patients, versus 15% in patients with GBC. Poor prognostic factors after resection of all biliary malignancies include the presence of lymph node metastasis, a positive surgical resection margin, and moderate or poor tumor differentiation. Several prognostic nomograms have been developed to predict long‐ term outcomes of biliary cancer resection.
J. Surg. Oncol. 2014;110:585–591. ß 2014 Wiley Periodicals, Inc.
KEY WORDS: klatskin; cholangiocarcinoma; bile duct cancer; gallbladder cancer
INTRODUCTION Biliary malignancy includes tumors of the bile ducts (cholangiocarcinoma) and tumors of the gallbladder. Three subtypes of cholangiocarcinoma are distinguished, based on where in the bile duct the cancer arises. Distal cholangiocarcinoma (DC) is arising from the common bile duct, between the confluence of the cystic duct with the common hepatic duct, and the ampulla of Vater. Perihilar cholangiocarcinoma (PHC) is arising from the common hepatic duct distal to the cystic duct confluence, the right or left hepatic duct, or the second‐order bile ducts. The second‐order bile ducts are the bile ducts that come together to form the right or left hepatic duct. Intrahepatic cholangiocarcinoma (IHC) is a tumor within the liver, arising from the bile ducts proximal to the second‐order bile ducts. Cancers arising from the cystic duct are not considered cholangiocarcinoma but rather are classified as gallbladder cancer (GBC). These four biliary malignancies each have a separate American Joint Committee on Cancer (AJCC) staging system since the most recent 7th edition [1]. In previous editions, PHC and DC were staged together as extrahepatic cholangiocarcinoma, and IHC was staged together with hepatocellular carcinoma. The annual incidence of biliary malignancies in the Western world is about 5 per 100,000 with about 12,000 new patients each year in the United States [2]. GBC is the most common biliary cancer. For patients with cholangiocarcinoma who undergo surgical exploration, PHC is most common (50%), followed by DC (40%), and IHC (10%) [3]. However, the incidence of IHC is increasing [4]. Unfortunately, most patients with biliary malignancies have locally advanced or distant metastatic disease at the time of presentation [5]. These patients are unlikely to benefit from a surgical resection. The focus of this review is on outcomes of patients who underwent a potentially curative resection. DC is not covered in this review, because it’s diagnostic work‐up, treatment, and outcomes are more similar to other periampullary cancers such as pancreatic cancer. Outcomes of interest include postoperative mortality, overall survival (OS), recurrence‐free survival (RFS), and sites of recurrence. Outcomes may differ depending
ß 2014 Wiley Periodicals, Inc.
on prognostic factors such as demographics, tumor markers, imaging characteristics, or pathology findings. A prognostic model or risk score combines important prognostic factors to predict outcomes such as 5‐year OS. Accurate ascertainment of outcomes in biliary malignancy is a prerequisite for clinical auditing and continued quality improvement. Prediction of outcomes using prognostic factors and models can guide and improve patient management. For example, a patient with an excellent predicted RFS after potentially curative resection of a biliary malignancy is unlikely to benefit from adjuvant treatment. Outcomes for patients with biliary malignancies are available from several sources: registries (e.g., SEER), patient series, or randomized controlled trials. Outcomes from registries and patient series constitute the main source of this review, because few randomized controlled trial for patients with biliary malignancies have been published.
INTRAHEPATIC CHOLANGIOCARCINOMA Diagnosis and Treatment The annual incidence of IHC in the United States is about 1 per 100,000. At an early stage patients are typically asymptomatic. At later stages, patients present with non‐specific symptoms such as weight loss, malaise, and abdominal discomfort [6]. Jaundice is less common in patients with IHC. Tumor markers have suboptimal sensitivity and specificity for the diagnosis of IHC: a CA 19‐9 of more than 100 U/ml has a sensitivity of only 53% [7]. A diagnosis of IHC can typically be
*Correspondence to: Yuman Fong, MD, Department of Surgery, City of Hope Medical Center, 1500 East Duarte Road, Duarte, CA 91010, USA. Fax: þ1 626 471 9212. E‐mail: [email protected] Received 12 June 2014; Accepted 29 July 2014 DOI 10.1002/jso.23762 Published online in Wiley Online Library (wileyonlinelibrary.com).
586
Groot Koerkamp and Fong
made on imaging after ruling out extrahepatic primary malignancies, in particular colon cancer. On CT, IHC is characterized by a hypodense mass in the non‐contrast phase, and peripheral enhancement in the arterial phase with progressive enhancement on later phases. On MRI, IHC is hypointense on T1‐ and hyperintense on T2‐weighted images. Most tumors (85%) are mass‐forming on imaging; the other morphological subtypes are periductal which grows along the bile ducts without forming a clear mass, and intraductal which grows within the bile duct. A biopsy is not required prior to surgical resection, but is recommended prior to chemotherapy. FDG‐PET was found to change surgical management in about 25% of patients [8]. Resection is recommended in fit patients without extrahepatic disease on imaging, in whom all disease can be removed with negative margins and an adequate liver remnant volume [6]. Patients with locally advanced or distant metastases should be considered for gemcitabine and cisplatin [9]. Patients with unresectable disease confined to the liver can also be considered for locoregional therapy such as transarterial chemo‐embolization (TACE) or hepatic artery infusion pump chemotherapy (HAIP) [10].
Postoperative Mortality Resection for IHC requires at least a hemihepatectomy in about 75% of patients, and a bile duct resection and reconstruction in about 25% of patients [11]. Nevertheless, in specialized high volume centers, the mortality rates can be as low as 1.2% [12].
Overall Survival The prognosis of patients with unresectable IHC is poor with about 5 to 10% of patients alive at 5 years after diagnosis [13]. Between 1983 and 2010, only about 15% of patients with IHC in the SEER database underwent a liver resection [14]. After potentially curative resection the 5‐year OS ranged from 21% to 35% in the five largest series (both Western and Asian) with each more than 200 patients; the median OS ranged from 18 to 39 months [11,15–18]. When one outlier of these five studies was excluded [16], the 5‐year overall OS was on average about 32%, and the median OS about 30 months. Outcomes of liver transplantation in patients with IHC are worse than for PHC. Even in highly selected patients the 5‐year OS was only about 30%, mainly due to early recurrence [19,20]. However, a more recent matched cohort study showed that cirrhotic patients with uninodular IHC smaller than 2 cm and patients with hepatocellular cholangiocarcinoma have similar long‐term survival as patients transplanted for hepatocellular carcinoma [21].
Recurrence Free Survival and Recurrence Patterns Recurrence free survival (RFS) varied widely in a recent systematic review: 5‐year RFS varied from 2% to 39%, and the median RFS from 7 to 34 months [22]. Studies that found a low incidence of recurrence had a shorter follow‐up or performed less frequent postoperative imaging to detect recurrence. The liver was the most common site of initial recurrence. In a multi‐institutional series of 301 patients, 54% had developed a recurrence with a median RFS of 20 months [23]. The initial recurrence was intrahepatic only in 61%, extrahepatic only in 21%, and both in 19%. Extrahepatic recurrence most commonly involved lung, peritoneum, or lymph nodes.
Poor Prognostic Factors Several poor prognostic factors after potentially curative resection have been identified and confirmed in multiple studies. A pooled analysis of these prognostic factors was recently performed using seven studies including more than 2,000 patients [22]. The most important Journal of Surgical Oncology
prognostic factors for OS in univariate analyses are presented in Table I. The prognostic value of a positive surgical margin was not statistically significant in this pooled analysis. However, another study showed that a positive surgical margin is a strong poor prognostic factor with a hazard ratio (HR) of 5.55 (95% CI: 3.57–8.33) [15]. Independent poor prognostic factors in more than one study were: multiple tumors, vascular invasion, lymph node metastasis, large tumor size, and high serum CA19‐9. The presence of poor prognostic factors does not preclude the possibility of long‐term survival. Patients with lymph node metastasis, multiple tumors, or vascular invasion had a median OS of about 20 months; however, 5‐year OS was 27% in the presence of one of these factors and 18% in the presence of two or three risk factors [11]. Independent risk factors for recurrence in a large multi‐institutional study were macrovascular invasion, lymph node metastasis, and tumor size more than 5 cm [23].
Prognostic Models The AJCC staging system for IHC is based on the following poor prognostic factors: multiple tumors, vascular invasion, direct invasion of extrahepatic structures, periductal invasion (versus mass‐forming lesions), lymph node metastasis, and distant metastasis. Stage I tumors are solitary without vascular invasion; stage II involves multiple tumors or a tumor with vascular invasion; stage III tumors demonstrate direct invasion of extrahepatic structures; and stage IV tumors have periductal invasion, or nodal or distant metastatic disease. Figure 1 presents stage‐specific survival for patients with IHC. Several prognostic models for patients with IHC have recently been published. Wang et al. from Hong Kong created a nomogram to predict 3‐year and 5‐year OS based on 367 patients with resected IHC [18]. The nomogram consisted of seven prognostic factors: multiple tumors, vascular invasion, lymph node metastasis, tumor diameter, direct invasion of adjacent structures or local metastasis, and tumor markers (CEA and CA19‐9). The c‐statistic of this model was 0.74, which was superior to other models including the AJCC staging system. Hyder et al. used a multinational database of 514 patients with resected IHC to predict 3‐year and 5‐year OS [24]. The prognostic factors in this nomogram were: multiple tumors, vascular invasion, lymph node metastasis, tumor diameter, age, and cirrhosis. These factors were also part of the nomogram of Wang et al., except for age and cirrhosis. The c‐statistic of this model was 0.69, again superior to the AJCC staging system. Jiang et al. from Shanghai developed a prognostic score to predict 5‐ year OS based on 344 patients with resected IHC [16]. The score only required prognostic factors that are available prior to resection: multiple tumors, tumor diameter, tumor boundary on imaging (distinct versus obscure), alkaline phosphatase, and CA 19‐9. Based on this score, patients were categorized in groups that differed in 5‐year OS: low risk (49%), intermediate risk (26%), high risk (10%), and extremely high risk (0%). All three prognostic models require external validation. Only the model of Jiang et al. could inform the decision whether or not to proceed TABLE I. Pooled Univariate Analysis of Poor Prognostic Factors for OS After Resection of IHC (n ¼ 2132) [22] Prognostic factor
HR
HR 95% confidence interval
Multiple tumors Vascular invasion Lymph node metastasis Large tumor size Poor tumor differentiation Older age Positive surgical margin
1.70 1.87 2.09 1.09 1.41 1.10 1.06
1.43–2.02 1.44–2.42 1.80–2.43 1.02–1.16 1.17–1.71 1.03–1.17 0.49–2.32
Surgery for Biliary Cancer
587
Several controversies remain regarding the optimal management of patients with GBC. A preoperative PET scan may avoid futile surgery in some patients [30]. Distant metastasis was found in 3 out of 23 patients with incidental GBC [8]. Excision of port‐sites at the time of definitive resection for GBC has been recommended because of wound recurrences. However, resection of port‐sites is a disfiguring procedure, and no survival benefit was found in a large study [31]. Anatomical segment 4B and 5 resection (instead of a 2–3 cm wedge resection) has been recommended, but two studies found no survival difference. [32,33] Open resection of GBC is the current standard of care, although some have recommended laparoscopic or robotic resection for GBC [34,35].
Postoperative Mortality In the 80s and 90s major liver resections, combined with extrahepatic bile duct resections in jaundiced patients were commonly performed for GBC patients. This contributed to a postoperative mortality of about 4% in high volume centers [36]. More recently, a 2 cm wedge resection of segment 4B and 5 is typically performed in non‐jaundiced patients without a bile duct resection. The mortality for such minor liver resections is about 1% in high‐volume centers, similar to liver resections for other indications. In a SEER study, however, the postoperative mortality was 4% [37]. Fig. 1. Stage‐specific (AJCC, 7th edition) overall survival after resection of IHC (n ¼ 212) [15]. The control group includes patients without resection or with an R2 resection. Reproduced with permission from Wolters Kluwer Health.
with a potentially curative resection. For example, patients with an extremely high risk score may be unlikely to benefit from a resection. The two other models require prognostic factors that are only available after resection.
GALLBLADDER CANCER Diagnosis and Treatment The annual incidence of GBC in the United States is about 3 per 100,000. GBC is found at pathologic review in about 1% of patients undergoing laparoscopic cholecystectomy for cholelithiasis [25,26]. These incidental GBCs represent about two‐thirds of patients with potentially curable GBC. At a more advanced stage, patients with GBC typically present with right upper quadrant pain, weight loss, and nausea. About 40% of these patients are jaundiced at the time of presentation due to involvement of the common bile duct. On cross‐sectional imaging a gallbladder and/or liver mass is typically seen in symptomatic patients. GBC arising from a gallbladder polyp represents only a small proportion of GBC patients. Staging for all patients with GBC involves a CT or MRI of the chest, abdomen, and pelvis. Patients with distant metastases, including nodal metastases beyond the hepatoduodenal ligament, are unlikely to benefit from a surgical resection. The same is true for patients with a T4 tumor, involving the main portal vein, common hepatic artery, or at least two extrahepatic organs. Patients presenting with jaundice are also unlikely to benefit from a resection [27]. These patients with locally advanced or distant metastases should be considered for gemcitabine and cisplatin [9]. All other patients, who are fit for surgery, are expected to benefit from a potentially curative (non‐anatomical) 2–3 cm wedge resection of segments 4B and 5 of the liver with a lymphadenectomy of the hepatoduodenal ligament [28]. Re‐resection is recommended after diagnosis of an incidental GBC, except for patients with T1a disease who rarely develop a recurrence [29]. Journal of Surgical Oncology
Overall Survival In a natural history study of patients with GBC, the median OS was 5 months [38]. A study using SEER data evaluated treatments and outcomes for patients with resectable GBC (after excluding patients with M1 or T4 disease). Only 16% of GBC patients underwent a radical resection including a liver resection (as recommended by the NCCN guidelines) [28], instead of a simple cholecystectomy alone [37]. They concluded that compliance with NCCN guidelines was poor. Patients had a superior median OS after a radical resection versus a simple cholecystectomy for both T2 tumors (53 versus 16 months) and T3 tumors (11 versus 8 months) [37]. In a German registry study, the superior median OS after radical resection for T1b tumors was 79 versus 42 months [39]. Due to the retrospective nature of these studies, however, the differences in OS could be partly attributable to differences between patients at baseline (i.e., selection bias).
Recurrence Free Survival and Recurrence Patterns The median time to recurrence after a potentially curative resection for GBC was only 12 months (as compared to 20 months for PHC) [40]. Of those patients that recurred, 85% had a distant recurrence with or without a concomitant locoregional recurrence; only 15% had a locoregional recurrence without distant recurrence. The peritoneum was among the sites of recurrence in about 50% of patients that recurred. Other common sites of recurrence included the liver, lung, and abdominal wall. Extrahepatic bile duct resection in non‐jaundiced GBC patients has been recommended to avoid bile duct recurrences. However, an isolated recurrence at the extrahepatic bile duct was never found in 26 non‐jaundiced patients who underwent a radical resection without extrahepatic bile duct resection [41]. Risk factors for recurrence are positive lymph node status, positive resection margin, and moderate or poor tumor differentiation. Patients with T2 tumors at the hepatic (versus the peritoneal) side of the gallbladder are more likely to recur in the liver, even after radical resection [42].
Poor Prognostic Factors Several poor prognostic factors have been identified and confirmed in multiple studies. Table II presents independent poor prognostic factors
588
Groot Koerkamp and Fong
TABLE II. Independent Poor Prognostic Factors for Disease‐Specific Survival After Resection of GBC (n ¼ 122) [43] Prognostic factor
HR
HR 95% confidence interval
Lymph node metastasis
