Evaluation of the POSSUM, P-POSSUM and E-PASS scores in the surgical treatment of hilar cholangiocarcinoma

In the last 20 years surgical treatment of hilar cholangiocarcinoma has mainly evolved because of the enhanced appreciation of tumor characteristics and improvements in preoperative imaging [1]. However, due to complex biliary and hepatic resections, this surgical procedure is still considered to be one of the most challenging procedures faced by hepatobiliary surgeons. Postoperative morbidity and mortality rates remain in a high range (14 to 76% and 0 to 19%, respectively) even in reports from high-volume centers [1‐5]. The accurate prediction of outcomes after a high-risk procedure such as surgical treatment of hilar cholangiocarcinoma can early detect postoperative complications, allow improved treatment planning and increase the precision of individual prognosis.

Many surgical risk scoring systems have been devised, but the Physiologic and Operative Severity Score for the enUmeration of Mortality and Morbidity (POSSUM) model by Copeland et al. [6] was recommended as the most appropriate for general surgery [7, 8]. This model, utilizing scores relating to twelve physiological and six operative variables, was developed to predict in-hospital mortality and morbidity postoperatively. However, POSSUM was then reported as over predicting postoperative mortality, particularly in patients at low risk. This led to a revision: the Portsmouth modification (P-POSSUM) by Whiteley et al. [9] Another surgical risk scoring system that has been validated in hepatobiliary surgery worldwide is the Estimation of Physiologic Ability and Surgical Stress (E-PASS) [10‐12]. This system comprises a preoperative risk score (PRS), a surgical stress score (SSS) and a comprehensive risk score (CRS) that is calculated from both the PRS and SSS.

Surgical risk scoring systems were initially designed for large populations and populations with different pathologies. However, later they were also applied to patients with a single diagnosis or one type of operation. It is unclear whether these two surgical risk scoring systems are useful for predicting morbidity and mortality in high-risk surgery such as surgical treatment of hilar cholangiocarcinoma. Hellmann et al.[13] was the first and only study to evaluate POSSUM for the surgical treatment of cholangiocarcinoma, and they found that the model overestimated postoperative morbidity and mortality. However, owing to its heterogeneity in selection design, this study considered only a limited and unclear number of operations for hilar cholangiocarcinoma over a 13-year period. The aim of the present study was to evaluate the POSSUM and E-PASS score in surgery for hilar cholangiocarcinoma, a most high-risk context in which to our knowledge, they have never been applied.

Postoperative complications and death may result depending on three major factors: the quality of the surgical team, the patient’s physiological status and the degree of surgical stress [16]. Where the quality of a surgical team in one hospital has remained stable for a certain period, surgical risk scoring systems could be applied to assess the risk of complications and death by quantification of the patient’s physiological status and the surgical stress applied. Several surgical groups have used POSSUM [17], P-POSSUM [18] and E-PASS [12, 19] successfully to perform comparative audit in hepato-biliary-pancreatic surgery. Therefore, these two scoring systems were chosen in the present study.

This is the first time surgical risk scoring systems were applied to a specific hilar cholangiocarcinoma surgical patient population. With a postoperative morbidity rate of 52.0% and an in-hospital mortality rate of 10.0%, our institution lies within the accepted range of complications after surgical treatment of hilar cholangiocarcinoma [1]. When the χ 2 test was used to compare actual morbidity and mortality rates with estimated ones, there was no significant lack of fit (P >0.05), indicating that the POSSUM and E-PASS scoring systems accurately estimate the outcomes. They also yielded an AUC value exceeding 0.7, suggesting their utility in predicting morbidity and mortality after surgery for hilar cholangiocarcinoma. However, improvements are still needed in the future because none of these scoring systems yielded an AUC value exceeding 0.9 for operations with all different levels of severity. Previous meta-analysis [20] and some reports [10] revealed that the POSSUM and E-PASS scoring systems failed to offer a significant predictive value for morbidity and mortality after hepatobiliary surgery. The main reason for the different findings may be that surgery of hilar cholangiocarcinoma is a more complex and severe operation than other hepatobiliary procedures. It has a higher operative severity score in POSSUM and a higher surgical stress score in E-PASS and therefore results in a higher risk prediction. Because the potential for morbidity and mortality is greater after this operation, surgical risk scoring systems would demonstrate a more accurate predictive value. We evaluated the corresponding results if only the patients underwent major and major plus operations were included. POSSUM indeed had more accuracy in predicting postoperative morbidity after major and major plus + operations. Similar findings have been observed in other studies, where POSSUM has a significantly more accurate predictive value for higher acuity procedures, such as pancreaticoduodenectomy, than for other pancreatic surgeries [17, 21]. However, no additional value was found for POSSUM, P-POSSUM and E-PASS scores in predicting in-hospital mortality after major and major plus operations. Firstly, operative type was not a factor multivariate associated with postoperative complications in our study, therefore, it remains unclear whether the type of operation influences the validity of the scores. Secondly, some independent factors for morbidity and mortality, such as operation type, intraoperative blood loss and preoperative hemoglobin are scored in POSSUM, P-POSSUM and E-PASS. Multivariate predictors for hepatobiliary surgery may differ from those in POSSUM and E-PASS scoring systems [22]. Based on the findings of our multivariate analysis, preoperative serum albumin, aspartate aminotransferase, and the Bismuth classification are independent factors associated with postoperative morbidity or in-hospital mortality but are included neither in E-PASS nor in POSSUM systems. Therefore, if researchers would like to improve the AUC value of these surgical scoring systems for hilar cholangiocarcinoma in the future, these factors might be added as new parameters in revised models.

Among the three surgical risk scoring systems employed in the present study, none of them showed a higher AUC value in predicting in-hospital mortality than the others. The advantage of the E-PASS scoring system is the relative ease with which data are acquired. This is favorable to the POSSUM or P-POSSUM score, which requires 18 different variables compared with the nine variables needed for the E-PASS score [23]. Furthermore, the POSSUM was generated only for surgical auditing and not for surgical decision making. However, the application of E-PASS has a potential role not only in surgical auditing but also in surgical decision making both between and within individual practice [24]. In our institution, we have developed preoperative management scenarios in our pancreato-biliary surgical practice. For example, patients with a high comprehensive risk score (CRS) are provided additional preoperative interventions such as enteric tube feedings, hyperalimentation, antibiotics, and biliary stenting, to improve preoperative parameters. This is often indicated, particularly when patients present with malignant obstructive jaundice, comorbid cardiac or respiratory illness, diabetes, or malnutrition.

There are some limitations to the present study. Firstly, since hilar cholangiocarcinoma is an uncommon neoplasm, the mortality rate corresponds to only ten patients, resulting in a relatively small group (100 patients) available for analysis over a long period of time (2003 to 2012). Operative technique and postoperative care have been developed during the past 10 years and treatment strategies are evolving. Secondly, in constructing the E-PASS model postoperative complications were only included when medical or interventional treatment had been carried out and mild complications were not regarded to be the same as severe ones [16]. However, POSSUM and P-POSSUM use a different definition and analysis of complication [6] which may affect the comparison of predictive value of scoring systems in our study.

The present study shows that POSSUM, P-POSSUM and E-PASS scores effectively predict morbidity and mortality in surgical treatment of hilar cholangiocarcinoma. However, improvements are still needed in the future because none of these scoring systems yielded an AUC value exceeding 0.99 for operations with all different levels of severity. Only POSSUM had more accuracy in predicting postoperative morbidity after operations with higher severity.