APACHE IV system helps to determine cholecystostomy role in elderly patients with acute cholecystitis

Cholecystostomy offers an alternative method for patients unt to undergo immediate cholecystectomy. Nevertheless, the role of cholecystostomy in the clinical management of high-risk surgical patients remains unclear. One of the main problems concerning the therapeutic effect in critically ill patients with acute cholecystitis is the lack of validated, well-established scoring systems to stratify the severity of patient disease states. APACHE IV scoring system was useful to estimate the hospital mortality for high-risk patients. We try to evaluate the performance of the APACHE IV scoring system in patients over 65 years of age with acute cholecystitis and the therapeutic effect of percutaneous cholecystostomy.

The tness of the APACHE IV score prediction is good with the area under the ROC curve of 0.894. The APACHE IV models were well-calibrated (with the Hosmer-Lemeshow statistic). Using the method of binary regression analysis, for the patients whose estimated mortality rate was more than 10%, cholecystostomy was an important factor for prognosis (P = 0.048). The estimated mortality of PC patients before and after operation was compared, which indicated that the estimated mortality after puncture was signi cantly decreased, either in the whole patient group (P = 0.004) or in the group with an estimated mortality greater than 10% (P = 0.008).

Conclusion
The APACHE IV scoring system showed that cholecystostomy was a safe and effective treatment for elderly high-risk patients with acute cholecystitis.

Background
Acute cholecystitis (AC) is a very heterogeneous disease and its life-threatening potential is strongly determined by the general status of the patient. In the aging subpopulation with medical comorbidities, immediate de nitive surgery may be hazardous. Percutaneous cholecystostomy (PC) offers an alternative method of management for patients un t to undergo immediate cholecystectomy because of severe sepsis or other underlying comorbidity. [1] Linet et al. [2] reported that the rate of PC markedly increased from 0.5% in 2005 to 12.2% in 2015 and was more common among the elderly. Smith et al. [3] also reported an increased use of PC in patients with AC over a 20-year period. Nevertheless, evidence for the role of gallbladder drainage as an effective alternative to early surgery remains poor, and there have been no randomized controlled trials on the issue [4] . Lu et al. [5] reported that the mortality rates were far higher in patients undergoing PC than in patients undergoing cholecystectomy. PC appeared to be of little bene t and not as safe as the Tokyo guidelines have suggested. [6] To date, it remains di cult to determine the role of PC in the clinical management of high-risk surgical patients with AC based on the currently available evidence. At present, the World Society of Emergency Surgery does not recommend PC among routine protocols for treatment of acute cholecystitis until better evidence becomes available. [7] Various de nitions are used to identify "high-risk patients", and the result also makes it more di cult to investigate the role of PC.
Outcome prediction is useful for therapeutic decision-making in critically ill patients. APACHE IV, the latest version of the APACHE (Acute Physiology and Chronic Health Evaluation) scoring system published in 2006, was developed to estimate the hospital mortality for adult ICU admissions. The APACHE IV score consists of several parts, including diagnosis, chronic health status, age, vital signs and physiologic measures, including blood gas analysis, urine output, blood sugar, WBC count, bilirubin level, and others; these are all not only accurate and reproducible but are also related to the prognosis of patients with AC [8] , especially elderly AC patients. The aim of this study was to evaluate the performance of the APACHE IV scoring system in patients over 65 years of age with AC and the therapeutic effect of percutaneous cholecystostomy.

Methods
This study was approved by the ethics review board of the Xuanwu Hospital of Capital Medical University of China. All patients signed informed consent. Consecutive patients admitted to our hospital (a 1200-bed urban tertiary-care referral center with a 12-bed surgical ICU) for the treatment of acute cholecystitis diagnosed by Tokyo guidelines diagnostic criteria [9] between January 2011 and December 2018 were retrospectively analyzed. The diagnosis of acute cholecystitis was based on clinical symptoms (right upper quadrant or epigastric pain or tenderness), leukocytosis or raised C-reactive protein (CRP) and at least one of the following sonographic ndings: distended gallbladder, gallbladder wall thickening > 3 mm or debris in the gallbladder. [10] Patients who were admitted with acute cholecystitis were identi ed by International Classi cation of Diseases (ICD)-9 diagnosis code. The exclusion criteria included (1) patients with acute pancreatitis or choledocholithiasis; (2) patients with malignancy of the bile duct, gallbladder, liver, or pancreas; (3) patients younger than 65 years old; (4) patients with missing values and the APACHE score could not be calculated. The data were collected from each patient on the day of admission to compute the APACHE IV scores, and the worst values for each parameter in the rst 24-hour period were used. Hospital mortality was de ned as the death of patients before discharge from the hospital. The decision to perform PC or cholecystectomy were made by the hepatobiliary surgeons based on their experience and the will for surgery on the part of the patients.
PC was performed by a hepatobiliary surgeon who used ultrasound guidance under local anesthesia.
First, a Chiba needle was introduced transhepatically into the gallbladder; second, the guidewire was placed, and the track was sequential dilated; then, the 7-Fr pigtail catheter was positioned with its tip in the gallbladder. The PC catheter was left open, connected to a collection bag and drained by gravity. After 2 weeks, if the patient's symptoms were relieved, the PC tube would be clipped. After 2 months, the PC tube can be preserved in situ or removed during cholecystectomy. Conservative treatment mainly included antibiotic treatment, fasting and nutritional support. Cholecystectomy was carried out according to conventional methods, 94.3% of them were performed by laparoscopy, 5.7% by laparotomy, no patients underwent subtotal cholecystectomy.

Statistical analysis
Data were analyzed using SPSS 21 software. Continuous variables are expressed as the mean ± standard deviation, and categorical variables are expressed as frequencies and percentages. The area under the ROC curve was used to compare the accuracy of the studied models. The t-test and Rank sum test were used for inter group comparisons, and P < 0.05 was considered statistically signi cant. We used the Hosmer-Lemeshow goodness-of-t test to evaluate the calibration prediction tness. Binary regression analysis was performed to identify predictors of mortality rate.

Results
A total of 862 consecutive patients with AC were enrolled in this study, of which 265 patients were excluded. Among the latter, 124 had choledocholithiasis, 108 had acute pancreatitis, 5 had cholangiocarcinoma, 3 had gallbladder cancer and 25 had incomplete APACHE IV data. The patients' origins before ICU admission were the emergency department (521/597, 87.30%) and the inpatient units The characteristics of the patients, along with a comparison of the survivors and nonsurvivors, are described in Table 1. Many details about patients, including leukocyte, transaminase, bilirubin, blood gas analysis and so on have been included in Apache IV score, so we do not list them in detail. The survivors were younger and had lower APACHE IV scores and lower risks of death than did the nonsurvivors. The discrimination of the APACHE IV score prediction was good with an AUC of 0.894 (95% CI, 0.849-0.930, Fig. 1). The Youden index was 0.62, the sensitivity was 78.26%, the speci city was 83.33%. The APACHE IV models were well-calibrated (with the Hosmer-Lemeshow statistic, chi-squared = 8.179, 8 degrees of freedom, and = 0.416); the data are shown in Fig. 2.  Table 2. When the estimated mortality rate was more than 10% or 12.5%, PC was an important factor in prognosis for patients with relatively severe illness (P = 0.048 and P = 0.049); however, cholecystectomy did not show relation with reduced mortality in the same population (P = 0.479 and P = 0.277). The estimated mortality of PC patients before and after operation was compared, which indicated that the estimated mortality after puncture was signi cantly decreased, either in the whole patient group (P = 0.004) or in the group with an estimated mortality greater than 10% (P = 0.008). The details are shown in Table 3.

Discussion
As the prevalence of acute cholecystitis increases with the increasing elderly population, we pay more attention both to the old patients with AC. One of the main problems concerning the role of PC is the lack of validated, well-established scoring systems to stratify the severity of patient disease states and different de nitions are used to identify "high-risk patients". The authors of the Tokyo guidelines suggested a staging system using mild, moderate and severe, depending upon the degree of local in ammation and the patients' conditions [11][12][13] . However, a retrospective series failed to nd any signi cant bene t from the application of the Tokyo guidelines [14] . Amirthalingam et al. concluded that the Tokyo guidelines 2013 may be too restrictive for patients with moderate and severe acute cholecystitis, and more attention should be paid to patient comorbidities during clinical decisionmaking [15] . Yacoub et al. tried to develop a score to stratify patients with gangrenous cholecystitis but did not pay much attention to the comorbidities of the patients; their score cannot be applied to patients who do not desire surgery [16] . We propose the APACHE IV score as a good tool to stratify and compare elderly patients with AC in clinical trials, and our study results indicate that this is possible [8] . For most hospitals, there are no conditions or data to establish their own evaluation criteria, therefore it is convenient and accurate to apply APACHE IV scoring in their clinical wards.
Although laparoscopic cholecystectomy is considered the gold standard treatment for patients with AC, it remains di cult to make de nitive recommendations regarding treatment by PC or cholecystectomy in elderly or critically ill patients. A recent review reported a particularly detailed examination of 53 papers regarding cholecystostomy as an option in acute cholecystitis, and suggested that PC was bene cial in high-risk patient groups, predominantly as a bridging therapy, allowing safer elective cholecystectomy once the patient has recovered from the acute illness [17] . Our data agree with this suggestion, as the procedure can stabilize the patient so that appropriate therapeutic planning can be achieved. There are reports that, in patients with acute cholecystitis who presented to the hospital ≥ 72 hours after symptom onset and did not respond to nonoperative treatment for 48 hours, PC with delayed laparoscopic cholecystectomy produced better outcomes and fewer complications than did emergency laparoscopic cholecystectomy. [18] Our experience partly con rmed this argument. In this group of data, the number of patients with predicting high mortality in both the PC group and the cholecystectomy group is too small, and the comparison between the two surgical methods is not statistically accurate, so we did not make this comparison. More data are wanted to clarity this question.
In a randomized controlled trial by Hatzidakis, [19] the e cacy of PC was compared with that of conservative management; the authors suggested that PC should not be performed as the rst choice in patients with acute cholecystitis, since it did not produce lower mortality rates than did conservative treatment. They also suggested that PC should be appealing to patients not showing clinical improvement following 3 days of conservative treatment, as well as to critically ill intensive care unit patients [19] . Their results were in harmony with ours. For patients in mild or moderate condition, we did not nd PC superior to conservative treatment. One reason for the indistinguishability is that the death rates were very low; on the other hand, conservative treatment did have a good effect. We suggest more precise indications for PC with the estimated mortality rate more than 10%. Of these eligible patients, PC was associated with reduced mortality. Early recognition and intervention are required in these patients, because of the rapid progression of acute cholecystitis to gangrene and perforation. We believe our indications are more concise and suitable for clinical work.
In previous studies of the safety of PC, procedure-related complications were only about 6.25%, relatively low considering candidates are usually high-risk patients. In our study, we found procedure-related complications in elderly patients with AC were only 7.16%, comparable to those of previous reports. [20] Although urgent cholecystectomy (within 72 hours from symptom onset) has been proposed as a de nitive treatment for symptomatic gallbladder disease, [21] routine cholecystectomy in an emergency setting can be challenging. Rather, gallbladder puncture is a better choice for surgeons with less experience. PC is a comparatively safe and effective procedure for the treatment of not only elderly highrisk patients with serious comorbidities but also for patients with contraindications for the general anesthesia required for cholecystectomy. [22] Our study has two major limitations. First, although this is the rst external validation of the APACHE IV admission prognostic model in elderly patients with AC, it is a nonrandomized and retrospective study.
Second, this research was done in China. Among the 597 patients, 480 (80.40%) received conservative managements rather that the surgical management that they should have had according to the international guidelines. This was mainly due to patients' unwillingness to operate and lack of experienced surgeons and this may have impacted on the patient prognosis. The study in itself has limitation in the external validity and its purpose is to compare PC to conservative treatment, excluding the patients who underwent cholecystectomy. Further studies will be required to elucidate this point.

Conclusion
The APACHE IV scoring system showed that PC puncture was a safe and effective treatment for high-risk elderly patients with acute cholecystitis. The conservative treatment was acceptable for the milder forms of the disease, but the most severe forms need a more aggressive treatment. More precise systems to predict mortality in AC patients are welcome as they can help therapeutic decision-making and enrollment of patients in clinical trials. Receiver operating characteristic curve for the APACHE IV score of the overall patients. The AUC was 0.878 (95% CI, 0.780 -0.932).