Preoperative prediction of severe cholecystitis (SC) is of great significance, as SC is associated with extensive use of antibiotics and longer duration of hospital stay [10]. Early prediction of SC may lead to timely specific therapies, such as early percutaneous cholecystostomy that may effectively reduce the complication rate and the duration of hospital stay of SC patients [11]. SC is associated with more adverse clinical features than ACC. Patients of SC tend to suffer from a variety of complications, such as damage to the main biliary ducts, ligation of aberrant hepatic ducts, and injury to the right hepatic artery during surgery. Once SC is detected before the operation, it is essential for doctors to select appropriate medical measures to avoid related complications and reduce the mortality rate [12, 13]. Imaging techniques, such as abdominal ultrasound and computerized tomography (CT) scanning are routinely used to diagnose ACC. However, it seems that these methods are ineffective for the detection of SC. In the present study, we demonstrated that the application of GBWT, the percentage of neutrophils and the WBC could respectively differentiate APC and AGC from ACC. Therefore, these three indexes could be used to determine surgical priority, improving diagnostic accuracy, and rate, and predicting patients’ risk of progressing from chronic to severe cholecystitis.
In previous studies, neutrophil to lymphocyte ratio (NLR) was widely used to differentiate SC from ACC, as NLR is related to inflammatory responses [14]. Additionally, some indexes of BRE, including ALT, ALP and AST were considered as risk factors affecting cholecystitis-associated mortality [15]. Based on these previous studies, we hypothesized that BRE indexes may contain key information for the diagnosis of different types of cholecystitis. Therefore, we tried to select some indexes from BRE to precisely differentiate two kinds of severe cholecystitis (APC and AGC) from ACC. We detected significant differences between two kinds of SC and ACC by one-way ANOVA. Specifically, APC and ACC significantly differed in terms of GBWT and the percentage of neutrophils, while AGC and ACC differed in terms of ALP, WBC and the percentage of neutrophils. No significant difference was detected for all other indexes. Therefore, GBWT, ALP, the percentage of neutrophils and WBC were selected as our candidate indexes while the rest of indexes including ALT, AST, FIB, GGT, PT and TB were excluded from the analysis.
ROC curve and odds ratio (OR) based on Chi-square test are widely used to evaluate the quality of a prediction model. To evaluate the value of GBWT, ALP, the percentage of neutrophils and WBC for the prediction of three kinds of SC, we created according ROC curves and calculated ORs [16].
As the significant difference in GBWT was only detected between APC and ACC, we believed this index may be optimal for the differentiation of APC from ACC. The analysis based on ROC curve showed that GBWT was effective in the differentiation of APC from ACC with the cut-off value of 5.5 mm, with the GBWT value over 5.5 mm, suggesting the diagnosis of APC. Previous study showed that purulent cholecystitis patients are likely to suffer from severe inflammation and increase of gallbladder drainage [17]. This implies that the gallbladder may be filled with stagnated or purulent contents due to the cholecystitis-induced inflammation and abscess on the gallbladder wall during the cholecystitis. As a result, the GBWT would increase, which is consistent with our findings.
Neutrophils are considered a marker of acute inflammation, as they are recruited to the sites of inflammation [18, 19]. As a result, the percentage of neutrophils would increase in severe inflammation. White blood cells (leukocytes) are also associated with a variety of inflammatory processes. For example, Ryder et al. found that leukocyte counts were associated with inflammation induced by obesity [20]. Bakhtiary et al. discovered a positive correlation between the level of leukocytes and inflammation. Their research showed that the depletion of leukocytes caused the reduction of inflammatory response [21]. Based on these results and our statistical data, we speculated that the percentage of neutrophil and WBC may be used to predict AGC. The analysis based on ROC curves showed that the percentage of neutrophils and WBC were both effective in the differentiations of AGC from ACC with the cut-off values of 79.75% and 11.1*109 /L. The percentage of neutrophils was also effective in differentiating APC from ACC with the cut-off value of 80.5%. To increase the accuracy of our diagnosis, our strategy was to distinguish cases of AGC and APC from ACC roughly using the percentage of neutrophils (cut-off value 79.75%), and then to diagnose AGC and APC using the unique indexes WBC and GBWT, respectively.
Although significant differences were detected between AGC and ACC in the levels of ALP, this index was removed from our study as it was proved to be ineffective to differentiate AGC from ACC, as indicated by the analysis of the ROC curves.
Previous studies always applied a simple index (e.g. NLR, WBC counting) to establish a prediction model for a certain SC, such as AGC [5, 22]. In contrast, we comprehensively applied various BRE indexes and a traditional computed tomography (CT) index, GBWT, to evaluate patients’ situations and precisely differentiated these patients into three categories, including two kinds of severe cholecystitis and acute exacerbation of chronic cholecystitis. This ability to precisely differentiate cholecystitis patients into three specific categories is very important, as it would allow clinicians to select the most suitable treatment options based on the concrete characteristics of each kind of cholecystitis, especially severe cholecystitis.
We report relatively low mortality rate in AGC and APC patients (5.26 and 3.51%, respectively). This can be related to the high conversion rate of laparoscopy in our study that may prevent postoperative complications caused by injury of adjacent organs (such as common bile duct, hepatic artery, portal vein, duodenum, etc.) during the operation. Therefore, in case of complicated gallbladder surgery in the elderly, timely conversion can ensure the safety of surgery and reduce mortality. Patients who died in this article were all emergency operation patients, with an average age of 80 years, and had severe basic diseases (such as lung infection, heart failure, kidney failure, etc.) before the operation.
Moreover, the average hospital stay was 9.32 days. 35.65% of patients stayed in ICU after surgery, with an average of 5.68 days and an average age of 74 years. Among them, 80.49% were emergency patients, and most of them were accompanied by basic diseases (such as hypertension, pulmonary infection, renal insufficiency, etc.) before the surgery.
There are some limitations in our study. Firstly, all patients were from the same hospital, which may decrease the representativeness of cases in our study to some extents. In addition, the number of cases in our study (114) was not sufficient. In particular, sample size of the APC group was only eight patients. Further studies with larger sample sizes are needed to enlarge the number of APC cases to further prove the reliability of GBWT as a predictive index for APC patients. Thirdly, we did not collect data about C reactive protein, an important index which is widely applied to evaluate inflammation [23, 24]. Future studies may include this index to modify our predictive models for cholecystitis.