cholangiocarcinoma: are north american surgical...

Cholangiocarcinoma: Are North American SurgicalOutcomes Optimal?

Andrew P Loehrer, MD, Michael G House, MD, FACS, Attila Nakeeb, MD, FACS, E Molly Kilbane, RN,Henry A Pitt, MD, FACS

BACKGROUND: Cholangiocarcinomas are deadly and require complex decisions as well as major surgery. Afew referral centers have reported good results, but no robust, risk-adjusted outcomes dataare available. The aims of this study were to analyze the surgical outcomes of a very largecohort of patients undergoing operations for cholangiocarcinoma in North America.

STUDY DESIGN: The American College of Surgeons National Surgical Quality Improvement Program Partic-ipant Use File was queried for patients with bile duct cancers. Patients (n¼ 839) were classifiedas having intrahepatic (36.5%), perihilar (34.4%), or distal (29.1%) cholangiocarcinomas bythe type of procedure performed. Observed and expected (O/E) morbidity and mortality rates,O/E indices, and regression-adjusted risk factors were determined.

RESULTS: Mortality was highest for perihilar tumors that were managed with hepatectomy and biliary-enteric anastomosis (11.9%) and lowest for distal cholangiocarcinomas (1.2%). After riskadjustment, mortality was considerable greater than expected for patients undergoing hepa-tectomy with biliary-enteric anastomosis (O/E ¼ 3.0) or hepatectomy alone (O/E ¼ 2.4).

CONCLUSIONS: This analysis suggests that postoperative outcomes are best for distal and worst for perihilar cholan-giocarcinomas, and hepatectomy for bile duct cancers is associated with a 2- to 3-foldmortality risk.We conclude that North American surgical outcomes can be improved for patients with proximalcholangiocarcinomas. (J Am Coll Surg 2013;216:192e200.� 2013 by the American College ofSurgeons)

Cholangiocarcinoma represents a rare but increasinglycommon malignancy, accounting for nearly 3% ofall gastrointestinal tumors and 10% to 15% of all primaryhepatobiliary cancer.1,2 Approximately 5,000 new patientsare diagnosed with cholangiocarcinoma in the UnitedStates each year.3 Survival rates for both intrahepatic andextrahepatic cholangiocarcinoma have improved, butoptimal surgical outcomes remain problematic, in largepart because of the challenging location of the primarytumors.4 Overall 5-year survival rates for patients under-going surgical resection range from 10% to 50%, depend-ing on stage, location of tumor, and adequacy of surgical

resection.5 Survival has been shown to improve with R0surgical resection, yet these procedures remain highlymorbid.6,7 In addition, postoperative complications fromresection of hilar cholangiocarcinoma are associated withdecreased long-term survival.8 Most studies to date arisefrom small single or multi-institutional series. Althoughprevious studies have identified specific factors associatedwith morbidity from pancreatectomy, data are limitedon current outcomes of surgery for cholangiocarcinoma.9

The primary aim of this study is to analyze the surgicaloutcomes of patients undergoing operations for cholan-giocarcinoma in a North American cohort.

METHODSThe American College of Surgeons (ACS) NationalSurgical Quality Improvement Program (NSQIP) isa prospective, multicenter clinical registry that wascreated to provide feedback on risk-adjusted outcomesto hospitals for quality-improvement purposes. Thesampling strategy, data abstraction procedures, variablescollected, and structure have already been published.10-13

Through nationally trained surgical clinical reviewers,the program collects detailed information on patientdemographics, preoperative risk factors and laboratory

CME questions for this article available athttp://jacscme.facs.org

Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein,Editor-in-Chief, has nothing to disclose.

Presented at the International Hepato-Pancreato-Biliary AssociationMeeting, Paris, France, July 2012.

Received September 14, 2012; Revised October 29, 2012; AcceptedNovember 5, 2012.From the Department of Surgery, Indiana University, Indianapolis, IN.Correspondence address: Henry A Pitt, MD, FACS, Department ofSurgery, Indiana University, 535 Barnhill Dr, RT 130D, Indianapolis,IN 46202. email: [email protected]

192ª 2013 by the American College of Surgeons ISSN 1072-7515/12/$36.00

Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jamcollsurg.2012.11.002

values, operative variables and postoperative events usingstandardized definitions.14

Data were gathered from the ACS NSQIP ParticipantUse Files (PUF) from January 1, 2005 to December 31,2009. This time period represents the first 5 years of datafrom ACS NSQIP. During this time, 243 centers partic-ipated, 54% were academic. To assure the high quality ofthe data, ACS NSQIP conducts inter-rater reliabilityaudits of the surgical clinical reviewers, requires 6 web-based training modules, and provides an online decisionsupport system. The overall disagreement rate for all136 variables is only 2.5%. In addition to the case inclu-sion and exclusion criteria described here, hospital inclu-sion and exclusion criteria are imposed for the PUF. Onlycases included in the morbidity and mortality observed/expected (O/E) index analyses are included in the PUF.A site can be excluded from the O/E calculation andthe PUF if the 30-day follow-up rate is <80%, theinter-rater reliability audit score is >5%, or <200 caseshave been submitted in the calendar year.Patients older than 16 years of age who underwent

surgical resection for cholangiocarcinoma were identifiedusing CPT codes and postoperative ICD-9 diagnosiscodes. Patients were considered eligible if carrying a post-operative ICD-9 diagnosis for malignancy of the liverand intrahepatic bile duct (ICD-9 code 155), malignancyof the intrahepatic bile duct (ICD-9 code 155.1), malig-nancy of the extrahepatic bile duct (ICD-9 code 156.1),malignancy of other specified sites of gallbladder andextrahepatic bile ducts (ICD-9 code 156.8), or malig-nancy of biliary tract part unspecified site (ICD-9 code156.9). Patients with malignancy of the gallbladder(ICD-9 code 156.0) were excluded from analyses.Subjects were then grouped based on type of procedureand location of primary tumor, consistent with publishedmethodologies previously.15 Intrahepatic lesions wereclassified as those with a postoperative diagnosis of malig-nancy of the intrahepatic bile duct and/or liver and whounderwent hepatectomy. Perihilar lesions were defined asthose with a postoperative diagnosis of malignancy of theextrahepatic bile ducts, biliary tract part unspecified, orother sites of gallbladder and extrahepatic bile ducts

who underwent a biliary-enteric anastomosis (BEA)with or without hepatectomy. Distal lesions were classi-fied as those with postoperative diagnosis of malignancyof extrahepatic bile ducts who underwent a pancreatec-tomy or pancreaticoduodenectomy. During the studyperiod, no specific pathologic, cholangiographic, orimaging data were available in the ACS NSQIP databaseto characterize the patients. Subjects were excluded if theydid not possess both an inclusion ICD-9 diagnosis codeand a CPT code for one of these procedures.Primary outcomes of interest were mortality rates,

overall morbidity rates, and specific complication rates.Observed to expected mortality indices were determinedfor each grouping of procedures using the previously vali-dated NSQIP risk-adjustment models. The probabilitiesof morbidity and mortality are based on a logistic regres-sion analysis that uses the patients’ preoperative character-istics as the independent or predictive variables. Theseprobability analyses are performed independently ondata from each calendar year. The observed morbidityor mortality was divided by the expected (probability)morbidity or mortality to obtain the morbidity andmortality O/E indices. Although expected rates are estab-lished via risk-adjustment models for all general surgerycases, O/E indices have been shown to be valid for hep-atopancreatobiliary procedures.16

Multivariate logistic regression models were then usedto determine risk factors for overall and procedure-specific morbidity, mortality, and prolonged hospitaliza-tion. Model goodness-of-fit was assessed using theHosmer-Lemeshow test, analysis of R2 and residuals, inaddition to F-tests and t-tests. Where appropriate, 95%confidence intervals were reported and a p value of�0.05 was considered statistically significant. All analyseswere performed using Stata 12.1 (Stata Corp).

RESULTSOur query of the NSQIP PUF for 2005 through 2009identified 843 patients who underwent surgery for intra-hepatic cholangiocarcinoma (n ¼ 306 [36.3%]), perihilarcholangiocarcinoma (n ¼ 289 [34.3%]), or distal cholan-giocarcinoma (n ¼ 242 [28.7%]) based on the type ofprocedure performed (Fig. 1A). Use of hepatectomywith biliary enteric anastomosis for perihilar diseaseranged from 46.3% to 60.0% per year with no changein rates during the 5 years studied (Fig. 1B). Additionalbaseline characteristics are shown in Table 1.

Morbidity

Overall and procedure-specific morbidity rates and O/Eindices are shown in Figure 2A and B, respectively. Rates

Abbreviations and Acronyms

ACS ¼ American College of SurgeonsBEA ¼ biliary-enteric anastomosisIQR ¼ interquartile rangeNSQIP ¼ National Surgical Quality Improvement ProgramO/E ¼ observed/expectedPUF ¼ Participant Use File

Vol. 216, No. 2, February 2013 Loehrer et al Cholangiocarcinoma in North America 193

of specific complications and overall risk factors for overallmorbidity are shown in Table 2 and Table 3, respectively.Morbidity for all patients was 41.8% with a risk-

adjusted O/E morbidity index of 1.1 (95% CI,1.0e1.1). The most frequent complications includedsepsis or septic shock (n ¼ 125 [14.9%]), return to oper-ating room (n ¼ 54 [6.4%]), and superficial or deepsurgical site infection (n ¼ 53 [6.3%]). Selected riskfactors for morbidity (shown in Table 3) include >7 Upacked RBCs transfused intraoperatively, body massindex (calculated as kg/m2) <18.5 (underweight), oper-ating time >5 hours, and history of COPD.Patients who underwent a hepatectomy (n¼ 305) expe-

rienced amorbidity rate of 29.4%with a risk-adjustedO/Emorbidity index of 0.8 (95% CI, 0.8e0.9). The mostcommon complications after hepatectomy included sepsisor septic shock (n ¼ 25 [8.2%]), return to operatingroom (n ¼ 15 [4.9%]), and organ space infection (n ¼13 [4.3%]).Patients undergoing hepatectomy and biliary-enteric

anastomosis (n ¼ 144) for perihilar tumors had higheroverall morbidity (47.8%) and a risk-adjusted O/Emorbidity index of 1.3 (95% CI, 1.2e1.3). Biliary-enteric anastomosis alone (n ¼ 149) had a morbidityrate of 43.5% and a risk-adjusted O/E morbidity indexof 1.1 (95% CI, 1.1e1.2). The primary complicationsfor both groups of procedures included sepsis or septicshock (n ¼ 61 [41%]), return to operating room(n ¼ 27 [18.2%]), and deep or superficial surgical siteinfection (n ¼ 22 [14.8%]). Controlling for confound-ing factors, patients undergoing partial hepatectomy

and BEA had significantly greater odds of morbiditycompared with BEA alone (odds ratio ¼ 1.8; 95%CI, 1.0e3.0; p ¼ 0.049).Pancreatectomy for distal bile duct tumors (n ¼ 243)

was associated with a morbidity rate of 47.1% witha risk-adjusted O/E morbidity index of 1.1 (95% CI,1.1e1.1). The most frequent complications includedsepsis or septic shock (n ¼ 39 [16.0%]), deep or superfi-cial surgical site infection (n ¼ 22 [9.1%]), and pneu-monia (n ¼ 8 [3.3%]).

Table 1. Patient Population

Patient characteristics

Age, y, median (IQR) 65 (57e73)

n %

Sex

Male 494 58.9

Female 345 41.1

Race, n (%)

White 261 80.8

Black, non-Hispanic 15 4.6

Hispanic 16 5.0

Asian/Pacific Islander 13 4.0

Other 18 5.6

Body mass index*

Underweight (<18.5) 19 2.3

Normal weight (18.5e25) 301 35.9

Overweight (25e30) 283 33.7

Obese (30e35) 142 16.9

Morbidly obese (35) 94 11.2

ASA class

1 6 0.7

2 231 27.5

3 549 65.4

4 53 6.3

Cardiovascular comorbiditiesy 489 58.3

Antihypertensive medication 441 52.6

Diabetes 158 18.9

Previous percutaneous cardiac intervention 54 6.4

Previous cardiac surgery 49 5.8

Respiratory comorbiditiesz 178 21.2

Smoker 154 18.4

COPD 38 4.5

Preoperative body mass loss of >10% 113 13.5

Steroid use within 30 d of operation 19 2.3

*Calculated as kg/m2.yDefined as history of congestive heart failure, angina, myocardial infarc-tion, diabetes, hypertension, previous cardiac catheterization or coronaryartery bypass.zDefined as history of COPD or smoking in addition to presence ofpreoperative pneumonia or dependency on mechanical ventilation.ASA, American Society of Anesthesiologists; IQR, interquartile range.

16.9 17.5

29.136.5

0

10

20

30

40

50

Hep Hep + BEA BEA Panc

% P

atie

nts

47.4 46.3 49.453.660

010203040506070

2005 2006 2007 2008 2009

% P

atie

nts

B

A

Figure 1. (A) Distribution of procedure types for all patients and (B)rates of hepatectomy for perihilar cholangiocarcinoma during thestudy period.

194 Loehrer et al Cholangiocarcinoma in North America J Am Coll Surg

Mortality

Mortality by procedure and complications are shown inTable 4, and risk factors for mortality are shown inTable 5. Overall and procedure-specific mortality ratesandO/E indices are shown in Figure 3A andB, respectively.Overall mortality rate for all procedures was 5.4% with

an O/E mortality index of 1.6 (95% CI, 1.5e1.8).Mortality for patients with no postoperative complicationswas <2% for the entire cohort and each individual proce-dure. However, mortality was higher if patients had a post-operative return to the operating room, if sepsis or septicshock developed, or if patients experienced other postop-erative complications (Table 4). Controlling for con-founding factors, reoperation was associated withsignificantly greater odds of mortality (odds ratio ¼ 3.7;95% CI, 1.4e9.3; p ¼ 0.006) (Table 5). Additional riskfactors for 30-day mortality included intraoperative trans-fusion of >7 U packed RBCs, previous cardiac surgery,history of chronic obstructive pulmonary disease, plateletsof <150,000 mL3, and preoperative bilirubin >3 mg/dL.Return to the operating room, sepsis or septic shock, and

other complications were each independently associatedwith increased odds of death (p < 0.05).Patients undergoing hepatectomy had a mortality rate

of 5.2% with a risk-adjusted O/E mortality index of2.4 (95% CI, 1.9e3.0). Of those who died, renal failuredeveloped in 2, and the remaining deaths were associatedwith failure to wean from mechanical ventilation, postop-erative coma, bleeding, sepsis, or reasons not specified.Hepatectomy and BEA (n ¼ 146) were associated with

12.0% mortality and a risk-adjusted O/E mortality indexof 3.0 (95% CI, 2.6e3.5). Sepsis or septic shock devel-oped in 8 patients who died within 30 days and theremaining deaths were associated with postoperativecardiac arrest (n ¼ 2 [1.4%]), postoperative bleeding(n ¼ 2 [1.4%]), or causes not specified. Biliary-entericanastomosis alone (n ¼ 150) was associated with 6.1%mortality and an O/E mortality index of 1.5 (95% CI,1.2e1.7). The most common complications associatedwith death included sepsis in 2 patients and cardiac arrestin 1 patient. Controlling for confounding factors, patientsundergoing partial hepatectomy and BEA had significantlygreater odds of 30-day mortality as compared with BEAalone (odds ratio ¼ 4.0; 95% CI, 1.4e11.3; p ¼ 0.010).Management of distal cholangiocarcinoma via pancre-

atectomy was associated with 1.2% mortality and risk-adjusted O/E mortality ratio of 0.4 (95% CI, 0.3e0.4).Only 3 deaths occurred in this group and they were asso-ciated with sepsis and cardiac arrest.

Length of hospitalization

Overall median length of inpatient hospital stay was of8 days (interquartile range [IQR] 6 to 15 days). Medianlength of hospitalization was 7 days (IQR 5 to 9 days) forpatients undergoing hepatectomy, 11 days (IQR 7 to 18days) for those receiving hepatectomy and BEA, 8 days(IQR 6 to 16 days) for BEA alone, and 11 days (IQR8 to 17.5 days) for patients undergoing pancreatectomy.Patients with postoperative complications had markedly

higher length of hospitalization as compared with thosewithout complications (Table 6). Postoperative return tothe operating room was associated with 2- to 3-fold longerhospitalization and patients with superficial or deep surgical

Table 2. Postoperative Morbidity Rates for Cholangiocarcinoma

Complications

All Hepatectomy Hepatectomy + BEA BEA Pancreatectomy

n % n % n % n % n %

Sepsis/septic shock 125 14.9 25 8.2 34 23.9 27 18.4 39 16.0

Return to OR 54 6.4 15 4.9 17 12.0 10 6.8 12 4.9

Superficial/deep SSI 53 6.3 9 2.9 9 6.3 13 8.8 22 9.0

Organ space infection 35 4.2 13 4.3 11 7.8 4 2.7 7 2.9

Other 84 10.0 28 9.2 11 7.8 10 6.8 35 14.3

BEA, biliary-enteric anastomosis; OR, operating room; SSI, surgical site infection.

47.8 47.1

29.4

43.541.8

0102030405060

All Hep Hep + BEA BEA Panc

% P

atie

nts

1.1

0.8

1.3

1.1 1.1

0.60.70.80.9

11.11.21.31.4


O/E

M

orb

id

ity

In

de

x

A

B

Figure 2. (A) Morbidity rates and (B) morbidity observed-to-expectedindices after surgery for cholangiocarcinoma.


site infections had median lengths of stay that were2 days longer than those patients with no complications.Other independent risk factors contributing to prolongedlength of stay (increasing hospitalization by �0.5days) included preoperative insulin-controlled diabetes(p ¼ 0.007), preoperative dyspnea (p ¼ 0.023), history ofmyocardial infarction (p ¼ 0.038), partially or totallydependent functional status (p ¼ 0.001), or preoperativebilirubin >1.5 mg/dL (p¼ 0.011).

DISCUSSIONCholangiocarcinoma is increasingly common acrossNorth America and continues to be associated with signif-icant morbidity and mortality. To our knowledge, thisarticle represents the largest series to date of surgicaloutcomes for patients with cholangiocarcinoma. We esti-mate that only 10% of the operations performed duringthe 5 years were included in this analysis. Morbiditywas seen in 41.8% of all patients and mortality occurredin 5.4% of all patients. The most frequent complicationsincluded sepsis or septic shock, wound infections, andreturn to the operating room. As illustrated with previousstudies, increased intraoperative blood transfusion was themost consistent risk factor for morbidity across all proce-dures.17 Mortality was most frequently associated withsepsis or septic shock, postoperative cardiac arrest, renalfailure, or postoperative bleeding.With respect to specific procedures, hepatectomy was

associated with a 29.4% morbidity rate and 5.2%

mortality. Hepatectomy and BEA were associated with47.8% morbidity and 12.0% mortality. Postoperativemorbidity was seen after 43.5% of patients undergoingBEA and the accompanying mortality rate was 6.1%.Patients receiving pancreatectomy for distal lesions hada considerable morbidity rate of 47.1%, but relativelymodest mortality of 1.2%. The cause for excess mortalityfor the patients with perihilar tumors undergoing hepatec-tomy likely is multifactorial. Possible explanations includepreoperative jaundice and cholangitis, the requirement foran extended hepatectomy in a significant proportion ofpatients; postoperative bile leaks; and postoperative liverfailure from inadequate functional liver reserve.Consistent with previous studies from around the world,

we found considerable morbidity and mortality acrossprocedures (Table 7).18-38 Most reports to date havestemmed from single or multi-institutional studies fromspecialty centers. Lang and colleagues from Germany re-ported similar postoperative morbidity and mortality inpatients with intrahepatic cholangiocarcinoma who under-went more extensive surgery or required reoperation.6 Theprevious largest single-center report of cholangiocarcinomafrom Johns Hopkins also demonstrated 2- to 3-fold greaterrates of sepsis for patients with perihilar lesions ascompared with intrahepatic or distal tumors.5 We reporta higher morbidity rate for distal lesions, but favorablemortality compared with previous studies.Only 49% of operations for perihilar lesions included

a partial hepatectomy in this series. Other recent singleor multi-institutional series from Europe, Asia, and the

Table 4. Complications and Postoperative Mortality for Cholangiocarcinoma

Complications

Mortality, % (with complications, n)

All Hepatectomy Hepatectomy þ BEA BEA Pancreatectomy

None 1.4 (488) 1.9 (216) 1.7 (60) 1.2 (83) 0.8 (129)

Return to OR 25.9 (54) 33.3 (15) 23.5 (17) 50 (10) 0.0 (12)

Sepsis/septic shock 9.6 (125) 1.4 (25) 23.5 (34) 7.4 (27) 2.6 (39)

Other (excluding SSI) 14.3 (84) 21.4 (28) 36.4 (11) 10.0 (10) 2.9 (35)

BEA, biliary-enteric anastomosis; OR, operating room; SSI, surgical site infection.

Table 3. Pre- and Perioperative Risk Factors for Postoperative Morbidity for Cholangiocarcinoma

Risk factor

Odds ratio (with risk factor, n)


>7 U packed RBC 10.8* (27) 11.5* (15) 2.0 (7) 1.0 (1) 1.9* (4)

Underweight 3.4* (19) 14.3* (4) d (4)y d (3)y 3.6* (8)

Operative time >5 h 3.3* (482) 4.5* (112) 1.0 (118) 1.0 (76) 3.3* (176)

History of COPD 2.8* (38) 2.9* (16) 2.2 (6) d (6)y 2.9* (10)

Dyspnea 1.8* (71) 1.8 (28) 0.5 (8) 2.0 (16) 1.8* (19)

Bilirubin >3 mg/dL 1.5* (226) 1.1 (40) 4.8* (46) 2.9* (65) 1.8* (75)

*p � 0.05.yAll cases with risk factor associated with morbidity.BEA, biliary-enteric anastomosis.


United States report much higher rates of hepatectomy,ranging from 82% to 100% (Table 7).30,31,33 However,each of these reports are from highly specialized centersor extended periods of time. Although our data had nosignificant variation in hepatectomy rates during the5-year study period, the lower rate of hepatectomy pre-sented here could stem from considerable variation inpractice patterns across the >200 NSQIP hospitalsfrom which data were collected.Despite having mortality rates comparable with

previous studies, the NSQIP also affords the uniqueopportunity to compare O/E morbidity and mortalityratios using validated risk-adjustment models. Particularlystriking in our analysis were the O/E mortality indices forpatients undergoing hepatectomy (O/E index ¼ 2.4) orhepatectomy and BEA (O/E index ¼ 2.9). Controllingfor confounding factors, patients undergoing hepatec-tomy and BEA had 3-fold greater odds of mortality rela-tive to hepatectomy alone (odds ratio ¼ 3.3; 95% CI,1.3e8.7; p ¼ 0.015). In these patients, sepsis or septicshock occurred in 27.3% of patients who subsequentlydied and was independently associated with 57.6% highermortality (p < 0.001). Similarly, the mortality rate andO/E mortality index for patients with perihilar cholangio-carcinoma undergoing partial hepatectomy were alsosignificantly higher than those receiving BEA alone.The addition of a partial hepatectomy for perihilar lesionsincreased the mortality nearly 4-fold. These findingssuggest that patients undergoing hepatectomy and BEAare at particularly high risk of 30-day mortality relativeto either hepatectomy or BEA alone.As concerning as the elevated mortality associated with

intrahepatic andperihilar cholangiocarcinoma is the discor-dance with standard practice for perihilar lesions. Multiplestudies to date have shown increased survival for perihilarcholangiocarcinoma when patients undergo R0 resectionthat usually involves at least a partial hepatectomy,

including resection of the caudate lobe.18-20 However, ourfindings also suggest that nearly 51% of patients with peri-hilar tumors fail to undergo hepatectomy at initial opera-tion, with no significant changes in rate of hepatectomyduring the study period. Although present NSQIP datalack oncology-specific variables, such as tumor size andextent, lymph node status, and surgical margins, the highpercentage of patients treated with BEA alone is concerningfor potentially inadequate resection for these tumors. Giventhe survival advantage of R0 resections, this variation fromstandard practice presents a unique opportunity to betteroptimize surgical management of perihilar lesions.Although our findings suggest suboptimal surgical

outcomes for patients with cholangiocarcinoma, the studydoes have a number of limitations. The NSQIP PUF has

Table 5. Risk Factors for 30-Day Postoperative Mortality for Cholangiocarcinoma

Risk factors

Odds ratio (with risk factor, n)


Complications

Return to OR 10.8* (54) 27.1* (15) 14.2* (17) 64.2* (10) 1.0 (12)

Sepsis/septic shock 5.5* (125) 1.0 (25) 17.2* (34) 55.7 (28) 1.0 (39)

Other (excluding SSI) 12.2* (84) 48.6* (28) 37.7* (11) 184.1* (10) 15.2 (35)

>7 U packed RBC 7.4* (27) 131.3* (15) 1.6 (7) 5.5* (1) 1.0 (4)

Previous cardiac surgery 3.5* (49) 4.7 (20) 7.1* (6) 1.0 (9) 1.0 (14)

COPD 3.4* (38) 1.6 (16) 2.2 (6) 25.7* (6) 1.0 (10)

Platelets <150 mL3 3.3* (61) 2.3 (32) 3.7 (7) 1.0 (12) 1.0 (10)

Bilirubin >3 mg/dL 2.8* (226) 1.7 (40) 0.8 (46) 1.0 (65) 0.5 (75)

*p � 0.05.BEA, biliary-enteric anastomosis; OR, operating room; SSI, surgical site infection.

1.6

2.43

1.5

0.40

0.51

1.52

2.53

3.54


O/E

M

orta

lity

In

de

x

5.4 5.2

1.2

6.1

12.0

02468

101214


% P

atie

nts

A

B

Figure 3. (A) Mortality rates and (B) mortality observed-to-expectedindices after surgery for cholangiocarcinoma.


a number of deliberate limitations that aim to ensurepatient privacy, as well as limitations imposed by resourceconstraints. These limitations pertain to the genericnature of most outcomes measures; the exclusion ofpatients under 17 years of age and the lumping togetherof those aged older than 90 years; the fact that follow-upis limited to 30 days; the removal of absolute dates; thedeidentification of participating sites; the absence ofrecords of many preventative measures; the relativelysmall number of participating hospitals; incomplete setsof some preoperative laboratory data; and limited infor-mation on pathological diagnosis and cancer therapies.Although long-term overall survival is influenced by oper-ative mortality and postoperative complications,8 tumorfactors such as stage and adequacy of surgical resectionalso clearly influence cancer-specific survival.Although data on preoperative chemoradiation within

NSQIP were variable during this study period, limiteduse of neoadjuvant therapy for cholangiocarcinomagreatly reduces the likelihood of such data omission intro-ducing uncontrolled confounding into our analysis. TheACS is currently implementing hepatectomy-specific vari-ables that will improve the clinical granularity of data forpatients undergoing surgery for cholangiocarcinoma.Procedure-specific variables, including use of preoperativestents, portal vein embolization, data to calculate Modelfor End-Stage Liver Disease score, hepatitis history, livertexture, operative technique, inflow occlusion, pathologicdata, drain management, postoperative bile leaks39 andpostoperative liver failure40 were not included in thecurrent analysis. Although these variables will be collectedbeginning in 2013, they were not reflected in this analysis,and their absence might have contributed to the higherthan expected mortality. The high O/E mortality indicesin this report might be secondary to inadequate, althoughextensive, risk adjustment, especially for extended hepa-tectomy.9,41 Future procedure-specific variable collectionwill likely provide greater insight into which preoperative,operative, postoperative, and oncology-specific factorscontributing to adverse postoperative outcomes.

To improve surgical outcomes for patients with peri-hilar cholangiocarcinomas, a number of strategies needto be implemented and/or studied. Regionalization ofhepatectomy has occurred to some degree in North Amer-ica. However, additional regionalization of these high-riskoperations needs to occur. In addition, improved patient

Table 6. Impact of Complications on Length of Hospitalization after Surgery for Cholangiocarcinoma

Complications

Length of stay, d, median (IQR)


None 7 (5e10) 6 (5e7) 8 (6e12) 7 (5e12) 8 (7e11)

Return to OR 21 (13e35) 11 (7e23) 22 (13e36) 24 (15e35) 27 (17e44)

Sepsis/septic shock 14 (8e21) 8 (6e15) 16.5 (10e27) 10 (8e22) 14 (12e21)

Superficial/deep SSI 9 (7e13) 8 (7e15) 7 (7e15) 8 (6e9) 10 (7e18)

Organ space infection 10 (7e14) 8 (7e10) 7 (7e11) 16 (11.5e20) 16 (11e23)

Other 15.5 (9e21) 10 (6e18) 18 (12e32) 13.5 (8e21) 16 (12e21)

BEA, biliary-enteric anastomosis; IQR, interquartile range; OR, operating room; SSI, surgical site infection.

Table 7. Previously Published Mortality Rates after Surgeryfor Cholangiocarcinoma

First author, yearResected,

nLiver

resection, %Mortality,

%

Intrahepatic

Casavilla, 199721 34 100 7

Madariaga, 199822 34 100 6

Valverde, 199923 30 100 3

Inoue, 200024 52 100 2

Weber, 200125 33 100 3

DeOliveira, 20075 34 100 2

Lang, 20096 83 100 7

Lanthaler, 201026 25 100 4

Present series 305 100 5

Perihilar

Jarnagin, 200127 80 62 10

Nishio, 200528 301 90 8

Dinant, 200629 99 33 15

DeOliveira, 20075 173 30 5

Endo, 200830 101 82 5

Regimbeau, 201131 56 100 8

Boland, 201232 27 30 0

Nuzzo, 201233 440 85 9


Distal

Bortolasi, 200034 15 0 0

Cheng, 200635 112 0 0

Murakami, 200736 36 0 0

DeOliveira, 20075 229 0 3

Choi, 200937 94 0 4

Qiao, 201138 122 0 7



selection will result in lower morbidity and mortality.Increased use of measurements to assess functional liverreserve will also result in less postoperative hepatic failure.Increased understanding of the impact of preoperativecholangitis on postoperative outcomes needs to be gainedby all specialists involved with preoperative biliarydrainage. The exact role of preoperative portal vein embo-lization also needs to be determined for patients whorequire an extended hepatectomy. Optimal methods toimprove nutrition preoperatively also need to be devel-oped. Careful operative technique to avoid blood lossand perioperative transfusions is also key to improvedoutcomes. Advanced hepato-pancreato-biliary trainingprograms to provide superspecialists and enhance pre-and intraoperative decision making also should improveoutcomes. In addition, drain management can be impor-tant for controlling bile leaks and preventing organ spaceinfections and sepsis.

CONCLUSIONSOur data reinforce the considerable morbidity andmortality associated with operativemanagement of cholan-giocarcinoma in a large North American cohort. However,the observed morbidity and mortality particularly forresections involving hepatectomy remain considerablyhigher than would be expected after risk adjustment.Patients receiving a BEA in addition to a hepatectomyare at particularly high risk of 30-day mortality. In addi-tion, more than half of all patients with perihilar lesionsare likely failing to receive optimal oncologic resectionsthat should include at least a partial hepatectomy. On theother hand, the decision by some surgeons to avoid a hepa-tectomy for perihilar tumors might have been due toconcerns about increased risk of this procedure. Additionalefforts as outlined here are needed to improve the postop-erative outcomes for patients undergoing hepatectomy forproximal and perihilar lesion in addition to increasing theuse of hepatectomy for perihilar cholangiocarcinoma.

Author Contributions

Study conception and design: Loehrer, House, Nakeeb, PittAcquisition of data: Loehrer, KilbaneAnalysis and interpretation of data: Loehrer, KilbaneDrafting of manuscript: LoehrerCritical revision: House, Nakeeb, Kilbane, Pitt

REFERENCES

1. Vauthey JN, Blumgart LH. Recent advances in the managementof cholangiocarcinomas. Semin Liver Dis 1994;14:109e114.

2. Shaib YH, El-Serag HB, Davila JA, et al. Risk factors of intra-hepatic cholangiocarcinoma in the United States: a case-control study. Gastroenterology 2004;128:620e626.

3. Lazaridis KN, Gores GJ. Cholangiocarcioma. Gastroenter-ology 2005;128:1655e1667.

4. Nathan H, Pawlik TM, Wolfgang CL, et al. Trends in survivalafter surgery for cholangiocarcinoma: a 30-year population-based SEER database analysis. J Gastrointest Surg 2007;11:1488e1497.

5. DeOliveira ML, Cunningham SC, Cameron JL, et al. Cholan-giocarcinoma: thirty-one-year experience with 564 patients ata single institution. Ann Surg 2007;245:755e762.

6. Lang H, Sotiropoulos GC, Sgourakis G, et al. Operations forintrahepatic cholangiocarcinoma: single-institution experienceof 158 patients. J Am Coll Surg 2009;208:218e228.

7. Endo I, House MG, Klimstra DS, et al. Clinical significance ofintraoperative bile duct margin assessment for hilar cholangio-carcinoma. Ann Surg Onc 2008;15:2104e2112.

8. Chauhan A, House MG, Pitt HA, et al. Post-operativemorbidity results in decreased long-term survival after resec-tion for hilar cholangiocarcinoma. HPB 2011;13:139e147.

9. Parikh P, Shiloach M, Cohen ME, et al. Pancreatectomy riskcalculator: an ACS-NSQIP resource. HPB 2010;12:488e497.

10. ACS-NSQIP Participant Use File user’s guide. https://acsnsqip.org/puf/docs/ACS_NSQIP_Participant_User_Data.File_User_Guide.pdf. Accessed on May 25, 2012.

11. ACS-NSQIPprogram specifics: surgical case inclusion/exclusionoverview. http://acsnsqip.org/main/program_case_inclusion_exclusion.asp. Accessed on May 25, 2012.

12. ACS-NSQIP program specifics: surgical clinical nurse reviewertraining. http://acsnsqip.org/main/program_nurse_training.asp.Accessed on May 25, 2012.

13. ACS-NSQIPprogram specfics: ACSNSQIP data; participant usedata file. http://acsnsqip.org/puf/PufRequestHomepage.aspx.Accessed on May 25, 2012.

14. Khuri SF, Henderson WG, Daley J, et al. Successful implemen-tation of the Department of Veteran Affairs National SurgicalQuality Improvement Program in the private sector: the PatientSafety in Surgery Study. Ann Surg 2008;248:329e336.

15. Nakeeb A, Pitt HA, Sohn TA, et al. Cholangiocarcinoma:a spectrum of intrahepatic, perihilar, and distal tumors. AnnSurg 1996;224:463e475.

16. Pitt HA, Kilbane M, Strasberg SM, et al. ACS-NSQIP has thepotential to create an HPB-NSQIP option. HPB 2009;11:405e413.

17. Ball CG, Pitt HA, Kilbane ME, et al. Peri-operative bloodtransfusion and operative time are quality indicators for pan-creatoduodenectomy. HPB 2010;12:465e471.

18. Aljiffry M, Abdulelah A, Walsh M, et al. Evidence-basedapproach to cholangiocarcinoma: a systematic review of thecurrent literature. J Am Coll Surg 2008;208:134e147.

19. Hasegawa S, Ikai I, Fujii H, et al. Surgical resection of hilarcholangiocarcinoma: analysis of survival and postoperativecomplications. World J Surg 2007;31:1256e1263.

20. Nakeeb A, Tran KQ, Black MJ, et al. Improved survival inresected biliary malignancies. Surgery 2002;132:555e564.

21. Casavilla FA, Marsh JW, Iwatsuki S, et al. Hepatic resectionand transplantation for peripheral cholangiocarcinoma. J AmColl Surg 1997;185:429e436.

22. Madariaga JR, Iwatsuki S, Todo S, et al. Liver resection forhilar and peripheral cholangiocarcinomas: a study of 62 cases.Ann Surg 1998;227:70e79.

23. Valverde A, Bonhomme N, Farges O, et al. Resection of intra-hepatic cholangiocarcinoma: a Western experience. J Hepato-biliary Pancreat Surg 1999;6:122e127.


24. Inoue K, Makuuchi M, Takayama T, et al. Long-term survivaland prognostic factors in the surgical treatment of mass-forming type cholangiocarcinoma. Surgery 2000;127:498e505.

25. Weber SM, Jarnagin WR, Klimstra D, et al. Intrahepatic chol-angiocarcinoma: resectability, recurrence pattern, andoutcomes. J Am Coll Surg 2001;193:384e391.

26. Lanthaler M, Biebl M, Strasser S, et al. Surgical treatment ofintrahepatic cholangiocarcinoma: a single center experience.Am Surg 2010;76:411e417.

27. Jarnagin WR, Fong Y, DeMatteo RP, et al. Staging, respect-ability, and outcomes in 225 patients with hilar cholangiocar-cinoma. Ann Surg 2001;234:507e519.

28. Nishio H, Nagino M, Nimura Y. Surgical management ofhilar cholangiocarcinoma: the Nagoya experience. HPB2005;7:259e262.

29. Dinant S, Gerhards MF, Rauwas EAJ, et al. Improvedoutcome of resection of hilar cholangiocarcinoma (Klatskintumor). Ann Surg Onc 2006;13:872e880.

30. Endo I, Gonen M, Yopp AC, et al. Intrahepatic cholangiocar-cinoma: rising frequency, improved survival, and determinantsof outcome after resection. Ann Surg 2008;248:84e96.

31. Regimbeau JM, Fuks D, Le Treut YP, et al. Surgery for hilarcholangiocarcinoma: a multi-instutional update on practiceand outcome by the AFC-HC study group. J GastrointestSurg 2001;15:480e488.

32. Boland B, Kim A, Nissen N, Colquhoun S. Cholangiocarci-noma: aggressive surgical intervention remains justified. AmSurg 2012;78:157e160.

33. Nuzzo G, Giuliante F, Ardito F, et al. Improvement in perio-perative and long-term outcome after surgical treatment forhilar cholangiocarcinoma: results of an Italian multicenteranalysis of 440 patients. Arch Surg 2012;147:26e34.

34. Bortolasi L, Burgart LJ, Tsiotos GG, et al. Adenocarcinoma ofthe distal bile duct: a clinicopathologic outcome analysis aftercurative resection. Dig Surg 2000;17:36e41.

35. Cheng Q, Luo X, Zhang B, et al. Distal bile duct carcinoma:prognostic factors after curative surgery: a series of 112 cases.Ann Surg Onc 2006;14:1212e1219.

36. Murakami Y, Uemura K, Hayashidani Y, et al. Pancreatoduo-denectomy for distal cholangiocarcinoma: prognostic impactof lymph node metastasis. World J Surg 2007;31:337e342.

37. Choi SB, Park SW, Kim KS, et al. The survival and prognosticfactors for middle and distal bile duct cancer following surgicalresection. J Surg Onc 2009;99:335e342.

38. Qiao QL, Zhang TP, Guo JC, et al. Prognostic factors afterpancreatoduodenectomy for distal bile duct cancer. Am Surg2011;77:1445e1448.

39. Koch M, Garden OJ, Padbury R, et al. Bile leakage after hep-atobiliary and pancreatic surgery: a definition and grading ofseverity by the International Study Group of Liver Surgery.Surgery 2011;149:680e688.

40. Rahbari NN, Garden OJ, Padbury R, et al. Posthepatectomyliver failure: a definition and grading by the International StudyGroup of Liver Surgery (ISGLS). Surgery 2011;148:713e724.

41. Kneuertz P, Pitt HA, Bilimoria KY, et al. Risk of morbidityand mortality following hepato-pancreato-biliary surgery.J Gastrointest Surg 2012;16:1680e1685.


cholangiocarcinoma: are north american surgical...

Documents