See Editorial on Page 241
Risk of waitlist mortality in patients with primary sclerosing cholangitis and bacterial cholangitis
Article first published online: 11 MAR 2013
Copyright © 2013 American Association for the Study of Liver Diseases
Volume 19, Issue 3, pages 250–258, March 2013
How to Cite
Goldberg, D. S., Camp, A., Martinez-Camacho, A., Forman, L., Fortune, B. and Reddy, K. R. (2013), Risk of waitlist mortality in patients with primary sclerosing cholangitis and bacterial cholangitis. Liver Transpl, 19: 250–258. doi: 10.1002/lt.23587
This work was supported by the National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases (grant F32 1-F32-DK-089694-01 to David S. Goldberg).
- Issue published online: 11 MAR 2013
- Article first published online: 11 MAR 2013
- Accepted manuscript online: 5 DEC 2012 02:43AM EST
- Manuscript Accepted: 18 NOV 2012
- Manuscript Received: 13 SEP 2012
Patients with primary sclerosing cholangitis (PSC) are at increased risk for bacterial cholangitis because of biliary strictures and bile stasis. A subset of PSC patients suffer from repeated episodes of bacterial cholangitis, which can lead to frequent hospitalizations and impaired quality of life. Although waitlist candidates with PSC and bacterial cholangitis frequently receive exception points and/or are referred for living donor transplantation, the impact of bacterial cholangitis on waitlist mortality is unknown. We performed a retrospective cohort study of all adult waitlist candidates with PSC who were listed for initial transplantation between February 27, 2002 and June 1, 2012 at the University of Pennsylvania and the University of Colorado–Denver. During this period, 171 PSC patients were waitlisted for initial transplantation. Before waitlisting, 38.6% (66/171) of the patients had a history of bacterial cholangitis, whereas 28.0% (44/157) of the patients with at least 1 Model for End-Stage Liver Disease update experienced cholangitis on the waitlist. During follow-up, 30 patients (17.5%) were removed from the waitlist for death or clinical deterioration, with 46.7% (14/30) developing cholangiocarcinoma. Overall, 12 of the 82 waitlist candidates (14.6%) who ever had an episode of cholangitis were removed for death or clinical deterioration, whereas 18 of the 89 candidates (20.2%) without cholangitis were removed (P = 0.34 for a comparison of the 2 groups). No patients were removed because of bacterial cholangitis. In multivariate competing-risk models, a history of bacterial cholangitis was not associated with an increased risk of waitlist removal for death or clinical deterioration (subhazard ratio = 0.67, 95% confidence interval = 0.65–0.70, P < 0.001). In summary, waitlist transplant candidates with PSC and bacterial cholangitis do not have an increased risk of waitlist mortality. The data call into question the systematic granting of exception points or referral for living donor transplantation due to a perceived risk of increased waitlist mortality. Liver Transpl 19:250–258, 2013. © 2013 AASLD.
endoscopic retrograde cholangiopancreatography
inflammatory bowel disease
Model for End-Stage Liver Disease
magnetic resonance imaging
primary sclerosing cholangitis
United Network for Organ Sharing.
Primary sclerosing cholangitis (PSC) is a chronic cholestatic disorder characterized by progressive biliary strictures and ultimately leading to the development of liver failure or cholangiocarcinoma. As a result of biliary strictures and bile stasis, patients with PSC are at increased risk for bacterial cholangitis. Among PSC patients, bacterial cholangitis is more commonly seen in those with a history of biliary tract instrumentation [ie, endoscopic retrograde cholangiopancreatography (ERCP)] and/or dominant bile duct strictures. Furthermore, bacterial cholangitis may be associated with secondary complications, including endocarditis and hepatic abscess formation. A subset of patients with PSC (<10%) will suffer from repeated bouts of bacterial cholangitis, which may result in frequent hospitalizations and thus impaired quality of life. PSC is clearly associated with a considerable amount of morbidity; however, the impact of bacterial cholangitis on the mortality of patients with PSC is unknown.
For patients waitlisted for liver transplantation, prioritization on the waitlist is determined by the Model for End-Stage Liver Disease (MELD) score. The MELD score, whose calculation is based on a patient's international normalized ratio, bilirubin level, and creatinine level, does not account for complications of liver disease. As a result, waitlist candidates with PSC and specifically those suffering from bacterial cholangitis are frequently referred for living donor liver transplantation and/or receive MELD exception points after a petition to a regional review board.[5, 6] This is despite evidence showing that even after we account for exception points, MELD scores, and differences in living donor liver transplantation, waitlist candidates with PSC have a lower risk of waitlist removal for death or clinical deterioration than candidates with other forms of end-stage liver disease. Additionally, there are no data for evaluating whether waitlist candidates with PSC who suffer from episodes of bacterial cholangitis have an increased risk of mortality.[4, 5] Although there are published recommendations for guiding regional review boards on which patients with PSC and recurrent bacterial cholangitis should receive exception points, these are not evidence-based.
National transplant registry data available from the United Network for Organ Sharing (UNOS) do not capture data on episodes of bacterial cholangitis, and this limits our ability to evaluate its association with waitlist mortality. Accordingly, using patient-level data from 2 large liver transplant centers, we sought to determine whether (1) waitlist transplant candidates with PSC and bacterial cholangitis have an increased risk of waitlist mortality and (2) whether or not current practices of exception point allocation for patients with PSC and bacterial cholangitis are justified.
PATIENTS AND METHODS
We performed a retrospective cohort study of all adult (≥18 years old) waitlist transplant candidates diagnosed with PSC at the University of Pennsylvania and the University of Colorado–Denver between February 27, 2002 and June 1, 2012. At each center, the respective transplant databases were queried to identify waitlist candidates with a diagnosis of PSC. The diagnosis of PSC was confirmed on the basis of a constellation of clinical, biochemical, radiological, and histological features consistent with published guidelines.
All PSC patients waitlisted for initial transplantation on or after February 27, 2002 were included. The start date was chosen to limit the analysis to the MELD era. We excluded patients listed for retransplantation.
For each study subject, a detailed medical record review was conducted at each center (D.S.G., A.C., and A.M.-C.). Relevant demographic and clinical data were recorded from the chart: age, self-reported race, sex, date of the diagnosis of PSC, history of inflammatory bowel disease (IBD; defined by a review of medical endoscopy or pathology reports and/or documentation of a medical record of confirmatory endoscopy and pathology findings consistent with ulcerative colitis or Crohn's disease), laboratory data, and cirrhosis on radiographic imaging (presence or absence).
Subjects were defined as having cirrhosis if liver biopsy and/or a radiographic study [ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI)] produced findings consistent with cirrhosis (including a nodular contour of the liver surface and/or caudate lobe hypertrophy). Detailed data on waitlist removal, including the reasons for removal (based on UNOS coding criteria) and the specific causes of death and/or clinical deterioration for patients removed from the waitlist because they died or were deemed too sick or medically unsuitable for transplantation, were obtained.
Because of the complexity of the diagnosis of bacterial cholangitis in the setting of PSC, we chose to define an episode of cholangitis as any documentation in a physician's medical record stating that a patient had an episode of bacterial cholangitis. We did not use elevated alkaline phosphatase levels and/or jaundice or magnetic resonance cholangiopancreatography/ERCP abnormalities to further define cholangitis episodes because patients with PSC commonly have baseline abnormalities in one of these parameters. As a result, we defined bacterial cholangitis as any documentation in a physician's medical record stating that a patient had an episode of bacterial cholangitis. Although these episodes were commonly associated with fever, right upper quadrant pain, worsened jaundice, and/or bacteremia in the absence of an alternative source of infection, no specific criteria were used because of the retrospective nature of the study.
The primary outcome was waitlist removal for death or clinical deterioration, which was defined as the patient being deemed too sick or medically unsuitable for transplantation. Outcomes were initially determined through a review of the transplant databases at the respective transplant centers, but all waitlist removals were subsequently confirmed with a detailed medical record review to verify the accuracy of the transplant databases.
We used Fisher's exact test and the chi-square test for dichotomous variables and the Student t test or the Wilcoxon rank-sum test for continuous variables (according to their distributions) to compare waitlist candidates with bacterial cholangitis to candidates without bacterial cholangitis and to compare candidates who were removed for death or clinical deterioration to candidates who were not.
In evaluating whether bacterial cholangitis affects the risk of waitlist mortality, one must consider the competing risk of transplantation because it influences the probability that a waitlist candidate will be removed from the waitlist for death or clinical deterioration. Thus, we fit competing-risk Cox regression models with waitlist removal as the outcome and transplantation as the competing risk.[9-11] All other outcomes were treated as censors. Waitlist candidates still on the waitlist at the end of follow-up were censored on that date.
The primary covariate of interest was a history of bacterial cholangitis before waitlisting or on the waitlist. We grouped together all episodes of bacterial cholangitis because the objective of this study was to determine whether cholangitis, regardless of its timing, was associated with worse outcomes. In addition, a history of cholangitis, regardless of its timing, is commonly used as a criterion for the granting of exception points. Secondary models were used to evaluate the association between waitlist mortality and a history of bacterial cholangitis separately before listing and on the waitlist.
Potential covariates included sex, race (white versus nonwhite because of the small sample sizes for nonwhite patients), age at listing, MELD score at listing, and a history of portal hypertensive complications (ascites, variceal bleeding, hepatic encephalopathy, or spontaneous bacterial peritonitis) before listing. Covariates were first tested in a single-item fashion; we included any variable with a P value ≤ 0.2 in the final model. The final models used a stepwise variable-selection process to retain variables with P values ≤ 0.1. Potential confounders were included if the variable changed the hazard ratio of bacterial cholangitis by 10%. We used robust standard errors to account for correlations due to patient clustering by transplant center. Covariates were reported as subhazard ratios because of the use of competing-risk models.[9, 11]
Competing-risk survival curves comparing the risk of waitlist removal for death or clinical deterioration and accounting for the competing risk of transplantation were constructed with the stcurve function in Stata.
We chose to exclude the small number of patients with missing data from the final multivariate models (n = 10) because of the limited variables that could be used for multiple imputation.
Institutional review board approval was obtained from both the University of Pennsylvania and the University of Colorado–Denver. All statistical analyses were performed with Stata 12.0 (College Station, TX).
From February 27, 2002 to June 1, 2012, 171 patients with PSC were listed for initial transplantation at the University of Pennsylvania (n = 83) and the University of Colorado–Denver (n = 88). The median age was 48 years [interquartile range (IQR) = 37–58 years], and waitlist candidates with PSC were predominantly white [83.6% (143/171)] and male [73.1% (125/171)].
At the time of listing, 76.6% (131/171) had a documented history of IBD, with a significantly greater proportion of waitlist candidates at the University of Pennsylvania with IBD (85.5% versus 68.2%, P = 0.007; Table 1). Nearly 80% of those with IBD [79.4% (104/131)] had ulcerative colitis.
|Variable||University of Pennsylvania (n = 83)||University of Colorado–Denver (n = 88)||Combined Data for 2 Centers||P Valuea|
|Age at listing (years)b||48 (32–58)||38 (37–58)||48 (37–58)||0.98|
|Race [n (%)]||0.44a|
|White||67 (80.7)||76 (86.4)||143 (83.6)|
|Black||14 (16.9)||8 (9.1)||22 (12.9)|
|Asian||1 (1.2)||2 (2.3)||3 (1.7)|
|Other||1 (1.2)||2 (2.3)||3 (1.7)|
|Male sex [n (%)]||57 (69.5)||68 (77.3)||125 (73.1)||0.25|
|History of IBD [n (%)]||71 (85.5)||60 (68.2)||131 (76.6)||0.007|
|IBD subtype [n/N (%)]||0.49|
|Ulcerative colitis||57/71 (80.3)||47/60 (78.3)||104/131 (79.4)|
|Crohn's disease||13/71 (18.3)||10/60 (16.7)||23/131 (17.6)|
|Indeterminate||1/71 (1.4)||3/60 (5.0)||4/131 (3.1)|
|International normalized ratiob||1.2 (1.1–1.4)||1.14 (1.0–1.3)||1.2 (1.1–1.4)||0.008|
|Total bilirubin (mg/dL)b||3.6 (1.9–10.6)||2.3 (1.1–8.2)||3.3 (1.4–10.1)||0.06|
|Creatinine (mg/dL)b||0.8 (0.7–1.0)||0.9 (0.7–1.0)||0.8 (0.7–1.0)||0.14|
|Laboratory MELD scoreb||15 (10–20)||12 (7–17)||14 (9–19)||0.01|
|Aspartate aminotransferase (U/L)b||110 (73–161)||85.5 (53–126)||98 (63–156)||0.01|
|Platelet count (1000/μL)b||173 (101–238)||180 (121–287)||178 (110–281)||0.13|
|Variceal bleeding [n (%)]||18 (21.7)||10 (11.4)||28 (16.4)||0.07|
|Spontaneous bacterial peritonitis [n (%)]||1 (1.2)||1 (1.1)||2 (1.1)||0.97|
|Ascites [n (%)]||31 (37.4)||23 (26.1)||54 (31.6)||0.12|
|Encephalopathy [n (%)]||14 (16.9)||12 (13.6)||26 (15.2)||0.56|
|Radiographic evidence of cirrhosis [n (%)]||51 (61.5)||41 (46.6)||92 (53.8)||0.08|
Listing Laboratory and Clinical Data
The median laboratory MELD score at the time of listing was 14 (IQR = 9–19), with waitlist candidates at the University of Pennsylvania having significantly higher laboratory MELD scores at listing [15 (IQR = 10–20) versus 12 (IQR = 7–17), P = 0.01]. The median serum creatinine level at listing was 0.8 mg/dL (IQR = 0.7–1.0), the total bilirubin level was 3.3 mg/dL (IQR = 1.4–10.1), and the international normalized ratio was 1.2 (IQR = 1.1–1.4; Table 1).
Before waitlisting, 39.2% (67/171) of the waitlist candidates had experience a hepatic decompensation, with 31.6% (54/171) having ascites, 16.4% (28/171) having variceal bleeding, 15.2% (26/171) having hepatic encephalopathy, and 1.1% (2/171) having spontaneous bacterial peritonitis (Table 1). According to imaging before waitlisting (few patients underwent liver biopsy because it was not needed for the diagnosis of PSC, so radiographic diagnosis data are presented), 53.8% (92/171) of the waitlist candidates with PSC had documented cirrhosis. None of these variables differed by center.
Episodes of Bacterial Cholangitis
Before waitlisting, 38.6% (66/171) of the waitlist candidates had a reported history of bacterial cholangitis. Similar proportions of waitlist candidates experienced at least 1 episode of bacterial cholangitis at the 2 centers (Table 2). At each center, among those with a history or bacterial cholangitis prior to waitlisting, greater than 50% experienced more than 1 episode (the median number of episodes could not be calculated because the discrete number of episodes was not available for every patient).
|Bacterial Cholangitis||n/N (%)|
|Before listinga||66/171 (38.6)|
|Number of episodes before listing|
|Multiple or severalb||16/66 (24.2)|
|On waitlistc||44/157 (28.0)|
|Number of episodes on waitlist|
|Multiple or severalb||4/44 (9.1)|
Among the waitlist candidates who received at least 1 UNOS MELD update while they were waitlisted, 28.0% (44/157) had an episode of bacterial cholangitis after listing, with a significantly greater proportion of waitlist candidates at the University of Colorado–Denver suffering from bacterial cholangitis (Table 2).
Overall, 48.0% (82/171) of the waitlist candidates with PSC experienced at least 1 episode of bacterial cholangitis before they were waitlisted and/or while they were on the waitlist. A significantly greater proportion of waitlist candidates from the University of Colorado–Denver had cholangitis at any point in time [59.8% (49/88) versus 40.2% (33/83) at the University of Pennsylvania, P = 0.04].
Table 3 displays the reasons for the waitlist removal of PSC candidates at the 2 centers. The median times on the waitlist were similar for the 2 centers (data not shown). The waitlist times were also similar for those with a history of cholangitis and those without one.
|Deceased donor orthotopic liver transplantation||67 (44.7)|
|Living donor liver transplantation||18 (12.0)|
|Medically unsuitable or too sick for transplantation||14 (9.3)b|
|Lost to follow-up||12 (8.0)|
|Transferred to another center||4 (2.7)|
|Patient request||1 (0.7)|
|Underwent transplantation at another center||8 (5.3)|
At the end of follow-up, 12.3% (21/171) of the waitlist candidates with PSC who had been listed since February 27, 2002 were still listed. During follow-up, 49.7% (85/171) underwent transplantation at 1 of the 2 centers, with 21.2% (18/85) being recipients of living donor liver transplants. Another 8 patients (4.7%) who were listed at the University of Pennsylvania underwent transplantation at another center. There was a nonsignificant increase in the odds of receiving a deceased donor transplant among patients with a history of cholangitis versus patients without one [35/82 (42.7%) and 32/89 (36.0%), P = 0.20].
During follow-up, 13 waitlist candidates underwent transplantation with exception points (9 at the University of Colorado–Denver and 4 at the University of Pennsylvania) for cholangiocarcinoma (n = 7); recurrent cholangitis (n = 3); hepatocellular carcinoma (n = 1); fatigue, bone disease, and a failure to thrive (n = 1); and colon dysplasia requiring colectomy in a patient with decompensated cirrhosis (n = 1).
Thirty waitlist candidates (17.5%) were removed from the waitlist for death or clinical deterioration, with 53.3% (16/30) dying while they were still waitlisted; the remaining 46.7% (14/30) were removed for clinical deterioration. None of these 30 patients had bacterial cholangitis immediately before delisting or as the inciting event for death or delisting.
The causes of waitlist removal for death or clinical deterioration are listed in Table 3. Notably, 46.7% (14/30) had biopsy-proven or presumed (on the basis of imaging or atypical cells on ERCP brushings) cholangiocarcinoma as the cause of removal for death or clinical deterioration. Only 16.7% (5/30) were removed because of infection/sepsis, and none were cholangitis-related.
Among the patients who received a deceased donor liver transplant, the median MELD scores were 22 (IQR = 15–26) for those with a history of cholangitis and 25 (IQR = 20–28) for those without cholangitis (P = 0.12). Among the patients who were removed from the waitlist for death or clinical deterioration, those with a history of cholangitis had a median MELD score of 27 (IQR = 11–31), whereas those without cholangitis had a median MELD score of 26 (IQR = 18–36, P = 0.74).
The waitlist time was significantly longer for patients who were removed for death or clinical deterioration (median = 341 days, IQR = 107–1064 days) versus patients who underwent transplantation (median = 151 days, IQR = 41–492 days, P = 0.01). At the time of removal, those who died or clinically deteriorated had numerically (but not statistically significantly) higher laboratory MELD scores [median = 26 (IQR = 18–33)] in comparison with those who underwent transplantation [median = 21 (IQR = 16–26), P = 0.06].
During follow-up, 14.6% (25/171) of the waitlist candidates developed cholangiocarcinoma; 44% (11/25) underwent transplantation, whereas the other 56% were removed for death or clinical deterioration. The risk of developing cholangiocarcinoma did not differ with a history of bacterial cholangitis [18.3% (15/82) with cholangitis developed cholangiocarcinoma, whereas 11.2% (10/89) without cholangitis did; P = 0.19].
Outcomes Based on a History of Cholangitis
We compared the crude risks of waitlist removal for death or clinical deterioration according to (1) a history of cholangitis before listing, (2) cholangitis on the waitlist, and (3) cholangitis before waitlisting and/or on the waitlist. Within each group, the risk of waitlist removal for death or clinical deterioration was similar when this was stratified by centers (data not shown).
Twelve of the 66 waitlist candidates with bacterial cholangitis before listing (18.2%) were removed for death or clinical deterioration, whereas 17.1% (18/105) without cholangitis before listing were removed (P = 0.86 for a comparison of the 2 groups).
Among the 44 waitlist candidates with bacterial cholangitis after listing, 9.1% (4/44) were removed from the waitlist for death or clinical deterioration, whereas 21.2% (24/113) without cholangitis were (P = 0.07 for a comparison of the 2 groups; there were 157 waitlist candidates in all with at least 1 MELD update).
Overall, 14.6% (12/82) of the waitlist candidates who ever had an episode of cholangitis were removed for death or clinical deterioration, whereas 20.2% (18/89) without a history of cholangitis were removed (P = 0.34 for a comparison of the 2 groups). Similar unadjusted results were obtained in time-dependent competing-risk analyses (P = 0.26 for competing-risk survival curves; Fig. 1).
Waitlist candidates with a history of cholangitis at any point in time were numerically more likely to undergo transplantation during follow-up [59.8% (49/82) versus 49.4% (44/89)]; however, this did not reach statistical significance (P = 0.18). Similar results were obtained after the exclusion of the 8 patients undergoing transplantation at another center.
In univariate competing-risk Cox regression models, a history of cholangitis before waitlisting and/or on the waitlist and male sex were associated with a decreased hazard of removal for death or clinical deterioration, whereas white race, older age, and a higher laboratory MELD score at listing were associated with an increased hazard (a history of hepatic decompensation was nonsignificant and was excluded from the final model; Table 4). In multivariate competing-risk models, a history of cholangitis was associated with a significantly decreased hazard of removal for death or clinical deterioration [subhazard ratio = 0.67, 95% confidence interval (CI) = 0.65–70, P < 0.001], whereas white race and increased age were associated with an increased risk.
|Variable||Univariate Subhazard Ratio (95% CI)||Multivariate Subhazard Ratio (95% CI)||P Value|
|Cholangitis ever||0.66 (0.53–0.84)||0.67 (0.65–0.70)||<0.001|
|Male sex||0.90 (0.85–0.96)||0.84 (0.61–1.16)||0.29|
|White race||1.88 (1.67–2.12)||2.11 (1.83–2.43)||<0.001|
|Laboratory MELD score at listinga||1.03 (1.02–1.04)||1.02 (0.99–1.06)||0.18|
|Age at listingb||1.35 (1.02–1.80)||1.37 (1.05–1.79)||0.02|
In multivariate competing-risk Cox models evaluating each of the bacterial cholangitis covariates separately, a history of cholangitis before waitlisting was not associated with an increased risk of waitlist removal for death or clinical deterioration (subhazard ratio = 1.00, 95% CI = 0.60–1.66, P = 0.99), whereas cholangitis on the waitlist was associated with a decreased risk (subhazard ratio = 0.39, 95% CI = 0.16–0.93, P = 0.03).
Using 10 years of data from 2 large transplant centers, this study is the first to evaluate the risk of waitlist mortality associated with bacterial cholangitis in waitlist candidates with PSC. During a 10-year period, nearly half of the waitlist candidates with PSC experienced at least 1 episode of bacterial cholangitis. A history of bacterial cholangitis was not associated with an increased risk of waitlist removal for death or clinical deterioration. Most importantly, none of the 30 patients removed for death or clinical deterioration had known cholangitis as the inciting event. The data call into question the prevailing belief that PSC patients with bacterial cholangitis are at increased risk for waitlist mortality and thus merit exception points and/or referral for living donor transplantation solely on the basis of the risk of waitlist mortality.
PSC is a progressive disease that has a significant impact on patient morbidity and mortality. Early natural history studies demonstrated that nearly 75% of PSC patients were symptomatic at the time of diagnosis. The median survival from the time of diagnosis was 11.9 years. With earlier diagnosis due to an increased awareness of the disease, the availability of noninvasive imaging of the biliary tree, the increased use of routine liver enzyme evaluation in asymptomatic patients, and the recognition of the association between IBD and PSC, more asymptomatic cases are being detected. Despite this, a significant disease burden remains, with approximately 5% of all liver transplants and 14% of living donor liver transplants being performed in patients with PSC.[6, 7]
Using UNOS data for all waitlist transplant candidates, we previously showed that 13.6% of waitlist candidates with PSC from 2002 to 2009 were removed from the waitlist for death or clinical deterioration. These data, however, did not allow the evaluation of bacterial cholangitis in waitlist candidates, and this necessitated the analysis of patient-level data.
The data presented here demonstrate that in both unadjusted and adjusted analyses, bacterial cholangitis in waitlist candidates with PSC was not associated with an increased risk of mortality or clinical deterioration. This is underscored by the fact that the causes of death or clinical deterioration were cholangiocarcinoma, portal hypertension, noncholangitis sepsis, and other nonhepatic conditions but not bacterial cholangitis. These data echo early PSC data from the Mayo Clinic's series from the 1970s. Thirteen of 39 PSC patients (33.3%) who were followed prospectively died, with 11 of the 13 deaths due to liver failure, 1 due to cholangiocarcinoma, and 1 due to a complication of colon cancer. Similar to our data, no deaths were directly related to bacterial cholangitis. More recent data from Norway, Sweden, Finland, and Denmark demonstrated that among 255 waitlist patients with PSC followed over an 11-year period, 32 (12.5%) were removed for death or clinical deterioration. None of the waitlist removals were related to bacterial cholangitis. Our analysis, the first to specifically evaluate the association between bacterial cholangitis and mortality, validates these earlier studies.
Although the risk of cholangiocarcinoma in our series was higher than the risk shown in previously published Scandinavian data, when one accounts for tumors diagnosed from explants or during the transplant operation, the risk of cholangiocarcinoma in our cohort was not substantially greater than that previously reported. The higher rate of cholangiocarcinoma on the waitlist seen in our cohort may be related to improved noninvasive imaging techniques (MRI and magnetic resonance cholangiopancreatography) in comparison with the Scandinavian data from 1990 to 2000 and/or to the increased waiting times in the United States, which may result in an increased risk of developing and/or detecting cholangiocarcinoma. Although there is no specific protocol for the surveillance of cholangiocarcinoma at each center, clinicians typically perform CT or MRI scans along with carbohydrate antigen 19–9 tests yearly or more frequently if they are clinically indicated. All the transplants for cholangiocarcinoma with exception points were performed at the University of Colorado, which follows the Mayo cholangiocarcinoma transplantation protocol.
An additional notable finding is that infections were rarely a direct cause of waitlist removal for death or clinical deterioration; this is unlike what is seen in patients with other forms of end-stage liver disease. One potential explanation is that at least on the basis of radiological imaging before waitlisting, a substantial proportion of our patients did not have radiographic evidence of cirrhosis. Although MRI and CT scans are not 100% sensitive or specific for this diagnosis, the data suggest that not all PSC patients waitlisted for transplantation may have cirrhosis, and this may explain the different waitlist risk profile seen in this patient population.
The risk of mortality associated with bacterial cholangitis in PSC patients and its impact on organ allocation are important because of the limited supply of transplantable organs. Because of the concern that patients with PSC and bacterial cholangitis potentially have an increased risk of waitlist mortality, transplant centers frequently apply for exception points for these patients. In fact, since February 27, 2002, 337 PSC patients have applied for MELD exception points for bacterial cholangitis, with the vast majority of these patients being granted exception points and ultimately undergoing transplantation, even though nearly three-fourths did not meet published recommendations for exception points for PSC and bacterial cholangitis. Additionally, waitlist candidates with PSC are significantly more likely to receive a living donor transplant than patients with other forms of end-stage liver disease, with recurrent cholangitis frequently being the indication. Although our data do not speak to the utility or necessity of transplantation due to the morbidity and impaired quality of life related to bacterial cholangitis, they do demonstrate that bacterial cholangitis alone is not associated with increased mortality.
Because of the nonsignificant differences in the odds of transplantation between patients with a history of cholangitis and patients without one, the decreased risk of mortality in patients with a history of cholangitis is not solely due to an increased rate of transplantation in this cohort due to exception points, living donor transplantation, and/or increased MELD scores and thus increased prioritization for deceased donor organs. Furthermore, the MELD scores at transplantation and removal were not different in patients with cholangitis versus patients without cholangitis, and this suggests that cholangitis alone does not result in a significantly increased MELD score that would change waitlist outcomes.
Our observations have several limitations. We combined data from 2 transplant centers for this analysis. The sample size for this study was small and included only 171 waitlist candidates with PSC. The 2 centers are situated in UNOS regions with above-average MELD scores at transplantation, so the data may not be generalizable to all waitlist candidates with PSC in all UNOS regions. The data were retrospective in nature, and we were reliant on accurate documentation in the medical records for our data. Further prospective, multicenter data are needed to confirm our findings. The demographic distributions of PSC candidates at the 2 centers were similar. Although patients at the University of Pennsylvania were more likely to have a history of IBD before listing, this should not account for our findings. The laboratory MELD scores at listing at the University of Pennsylvania were slightly higher; however, we adjusted for MELD scores in our final models. The distributions of portal hypertensive complications were similar, however. We did note that a significantly greater proportion of waitlist candidates at the University of Colorado–Denver had cholangitis on the waitlist. Although the reason for this is unclear (perhaps provider differences in the use of prophylactic antibiotics), the risk of adverse waitlist outcomes did not differ at the centers, so this was not a substantial limitation.
The definition of cholangitis was based on physician documentation in the medical record without definitive corroborating documentation of bacteremia or ERCP. We chose to define cholangitis with physician documentation based on expert opinion and clinical judgment for several reasons. First, the diagnosis of bacterial cholangitis is difficult in the setting of PSC. Traditional laboratory abnormalities such as hyperbilirubinemia and elevated alkaline phosphatase levels are common at the baseline in these patients. Second, findings such as worsening jaundice, fever, and right upper quadrant pain or tenderness in the absence of another infection may signify bacterial or nonbacterial cholangitis in this patient population, and retrospectively defining an episode of bacterial cholangitis requires the use of previous opinions of treating physicians. Lastly, even with data on blood cultures or ERCP procedure reports (taken in the setting of a dominant stricture), there is no single finding that is confirmatory for bacterial cholangitis in PSC, and the diagnosis (or presumed diagnosis) is reliant on the clinician's judgment. Although there might have been misclassification such that patients were misclassified as having cholangitis in the absence of such an infection, we would not expect this to change our results. Only if a significantly greater proportion of the patients who did not die were misclassified as having bacterial cholangitis (which would lead to an underestimation of the risk of mortality from cholangitis) would we expect our results to show an increased risk of mortality in patients with cholangitis, and this is extremely unlikely. Additionally, this difficulty in definitively diagnosing cholangitis in this population also speaks to the challenge of granting exception points or using this as a basis for increased prioritization. It may in fact be that only those patients with bacteremia and/or septic complications are at greater risk of mortality, and this not only supports the published consensus recommendations for exception points in PSC but also forces us to more stringently define cholangitis in this population to better treat and risk-stratify PSC patients. Because of the retrospective nature of the study, we could not specifically determine which episodes of cholangitis were associated with bacteremia with or without sepsis. However, no waitlist removals for death or clinical deterioration were directly due to cholangitis, and this suggests that even in the subset of patients with cholangitis complicated by bacteremia, the risk of waitlist mortality may not be significantly increased. Further prospective, multicenter data are needed, however, to confirm this.
We were missing data on 10 patients and excluded them from the final multivariate model. However, because the univariate and multivariate hazard ratios were similar and the hazard ratio was significantly less than 1, this should not have affected our results. Waitlist candidates with bacterial cholangitis may have received a living donor liver transplant and/or received MELD exception points to decrease their probability of dying on the waitlist and thus increase the likelihood of transplantation. However, as shown, only 11% of the waitlist candidates with PSC received a living donor transplant, and 9% received exception points. Eleven of the 18 living donor recipients had a history of cholangitis, whereas only 3 of the 13 transplant recipients with MELD exception points were granted exceptions for recurrent cholangitis. This demonstrates that the referral of waitlist candidates with PSC and cholangitis for living donor liver transplantation and the granting of MELD exception points do not explain the findings. Even under a worst-case scenario in which all living donor recipients and exception point recipients died, the outcomes of patients with cholangitis would be the same as the outcomes of noncholangitis patients and not worse. We defined cirrhosis on the basis of radiographic imaging because few patients had undergone liver biopsy, but this is not 100% sensitive. Despite this, because of the substantial number of patients without radiographic evidence of cirrhosis, it is still likely that a sizable proportion of the patients did in fact not have cirrhosis. Lastly, we were unable to accurately determine whether patients were on suppressive antibiotic therapy due to a history of recurrent cholangitis. Although this may have decreased the risk of complications from cholangitis and/or prevented the development of noncholangitis complications, this is unlikely to fully account for our results.
In summary, we have shown that waitlist transplant candidates with PSC and bacterial cholangitis do not have an increased risk of waitlist mortality in comparison with patients without cholangitis. Over a 10-year period, no patients died on the waitlist or were removed for clinical deterioration as a result of bacterial cholangitis. The data call into question the systematic granting of exception points for patients with PSC and bacterial cholangitis on the basis of increased waitlist mortality. Multicenter patient-level data are needed to validate these findings and develop predictors of waitlist mortality to better risk-stratify the waitlist candidates with PSC in greatest need of exception points and preemptive living donor liver transplantation.
- 4Model for End-Stage Liver Disease (MELD) exception guidelines: results and recommendations from the MELD Exception Study Group and Conference (MESSAGE) for the approval of patients who need liver transplantation with diseases not considered by the standard MELD formula. Liver Transpl 2006;12(suppl 3):S128–S136., , , , , , et al.
- 11A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509., .
- 17Second infections independently increase mortality in hospitalized patients with cirrhosis: the North American Consortium for the Study of End-Stage Liver Disease (NACSELD) experience. Hepatology 2012;56:2328–2335., , , , , , et al.