New Scoring System for Prediction of Surgical Difficulty During Laparoscopic Cholecystectomy After Percutaneous Transhepatic Gallbladder Drainage

Abstract Background The surgical difficulty of laparoscopic cholecystectomy (LC) after percutaneous transhepatic gallbladder drainage (PTGBD) remains unknown. This study aimed to establish a scoring system (SS) to predict the necessity of a bailout procedure during LC after PTGBD and to evaluate the relationship between SS and perioperative complications. Methods We retrospectively studied 70 patients who underwent LC after PTGBD. Preoperative factors potentially predictive of the need for the bailout procedure were analyzed. The SS included significantly predictive factors, with their cutoff values determined by receiver operating characteristic curves. Patients were assigned a score of 1 when exhibiting only one of these abnormalities. We compared the perioperative factors between three groups with scores of 0, 1, or 2. The SS was applied to another series of 65 patients for validation. We compared the score‐2 patient perioperative factors between LC with the bailout procedure and open cholecystectomy from the beginning (OC). Results Independent predictors were time until PTGBD after symptom onset and the maximal wall gallbladder thickness (cutoff values: 3 days and 10 mm, respectively). The high‐score group was significantly associated with bile duct injury (BDI). The sensitivity and specificity of our SS were 75.0% and 98.1% in validation, respectively. The score‐2 OC and laparoscopic subtotal cholecystectomy (LSC) groups had no BDI. Conclusions The SS using time until PTGBD after symptom onset and gallbladder wall thickness for predicting the need for the bailout procedure correctly predicted the need. The scores might be associated with the risk of BDI, and LSC or OC might be a better choice for score‐2 patients.


| INTRODUC TI ON
Laparoscopic cholecystectomy (LC) has become a standard procedure for benign diseases of the gallbladder (GB) worldwide. 1 Severe inflammation of GB and its surroundings increases both the difficulty of complete LC and the frequency of postoperative complications. 2 Bile duct injury (BDI) is known to occur in a certain proportion of cases, and the prognoses of patients who suffer vasculo-biliary injury (VBI) in particular are poor. 3 Therefore, it is very important to take prudent steps to prevent complications. 4 The Tokyo Guidelines 2018 (TG18) propose management bundles for acute cholecystitis (AC) and cholangitis.
When LC for AC is difficult, not only open conversion but also laparoscopic subtotal cholecystectomy (LSC) with the fenestrating or reconstituting and fundus first technique, called the bailout procedure, can be chosen to prevent BDI according to the intraoperative findings. 4 On the other hand, early surgery for AC cannot be performed for all surgically high-risk patients. Percutaneous transhepatic GB drainage (PTGBD) should be considered the first alternative to cholecystectomy in surgically high-risk patients with AC because several studies have described PTGBD as less invasive and having a lower risk of adverse events than cholecystectomy. 5 However, the degree of surgical difficulty during LC after PTGBD is unknown, and no report has provided scientific evidence of the conditions supporting use of the bailout procedure during LC after PTGBD.
This study aimed to establish a scoring system (SS) to predict the necessity of the bailout procedure during LC after PTGBD and to evaluate the relationship between the SS and perioperative complication.

| PATIENTS AND ME THODS
The medical records of a series of 178 consecutive patients who  ostomy for choledocholithiasis, one patient who had undergone   open cholecystectomy with choledochojejunostomy for choledocholithiasis, one patient who had undergone LC 7 years after PTGBD, and six patients with unknown data due to PTGBD in other hospitals were excluded. The remaining 65 patients were divided into 53 patients who had undergone pure LC without the bailout procedure and 12 patients who had undergone LC with the bailout procedure. The 65 patients were examined to assess the accuracy of our SS (Figure 1).
Most of PTGBD procedures were performed by physicians in our hospital. The main indication of PTGBD in our hospital was Grade II (moderate) or III (severe) AC according to TG18 5 when the patients could not withstand surgery and were AC refractory to antibiotics.
Measurement of the maximum GB wall thickness was performed on the axial or coronal plane of noncontrast or contrast-enhanced computed tomography just before PTGBD because most of the images of abdominal ultrasonography just before or during PTGBD had not been stored.
The following 14 preoperative factors of these patients were analyzed to predict the necessity of the bailout procedure during LC in univariate and multivariate analysis: age, gender, body mass index, procedure (single-port or multi-port LC), American Society of Anesthesiologists physical status classification, age-adjusted Charlson comorbidity index (CCI), 6,7 anticoagulant therapy, past history of upper abdominal surgery, time until PTGBD after symptom onset, time until surgery after PTGBD, maximal GB wall thickness, maximal diameter of impacted stone in the GB, maximal white blood cell (WBC) count in the peripheral blood, and maximal serum value of C-reactive protein. The cutoff score of CCI was determined based on a past report. 8 The SS was designed by using the significant predictive factors, the cutoff values of which were determined by a receiver operating characteristic (ROC) curve. Patients were assigned to a score of 2 if they had two significant factors predictive of both abnormalities for the respective cutoff values, a score of 1 if they only had one factor predictive of these abnormalities, and a score of 0 if neither abnormality was present.
We then compared the 18 perioperative factors, which are the previous

| Statistical methods
Continuous data were expressed as the median and range, and compared by the Mann-Whitney U-test between two groups and by the Kruskal-Wallis test among three groups. Categorical data were compared by the chi-square test. Univariate factors predictive of the need for the bailout procedure (P < .05) were entered into a logistic regression model to identify the independent

| Characteristics of the clinical patients
The characteristics of the 70 patients are summarized in Table 1

| Univariate and multivariate predictors of the conversion
Significant univariate predictors of the necessity of the bailout procedure during LC after PTGBD were time until PTGBD after symptom onset (P = .001), the maximal GB wall thickness (P = .001), and the maximal WBC count in the peripheral blood (P = .031) ( Table 1).

| Determination of cutoff values for the independent predictors
The results of the ROC curve of time until PTGBD after symptom onset and the maximal GB wall thickness are shown in Figure 2

| Univariate analysis of the relationship between patient characteristics and the SS
Two variables were used in the design of the SS. In brief, patients were assigned a score of 2 if they had both a longer time until PTGBD after symptom onset (≥3 days) and a thicker GB wall (≥10 mm), a score of 1 when exhibiting only one of these abnormalities, and a score of 0 if neither abnormality was present (

| Validation of the SS for prediction of the bailout procedure
The frequency distribution of the necessity of the bailout procedure during LC according to the time until PTGBD after symptom onset and the maximal GB wall thickness was examined in another series of 65 patients. Table 5 shows the patient characteristics of the 65 patients; of these, 27, 28, and 10 patients had scores of 0, 1, and 2, respectively. Three patients with a score of <2 underwent the bailout procedure during LC. The reasons for conversion were as follows: difficulty in dissection at Calot's triangle in four patients, difficulty in dissection for adhesion to the duodenum and/or transverse colon in two patients, and BDI in one patient. One of the patients who underwent LSC was diagnosed with GB cancer by histopathological examination after surgery. This patient underwent the additional liver bed resection and biliary resection and reconstruction. In univariate analysis, the GB wall in the LC with the bailout procedure group was significantly thicker than the GB wall in the pure LC without the bailout procedure group (P = .003). The percentage of patients who required the bailout procedure during LC was 75.0% in patients with a score of 2 (P < .0001). The SS was designed using these two variables. By comparing patients with a score of 2 with those with a score of 1 and less, the sensitivity and specificity of our SS for prediction of the bailout procedure were 75.0% (=9/12) and 98.1% (=52/53), respectively (Table 5).

| D ISCUSS I ON
In this study we proposed a new SS to predict the necessity of the bailout procedure during LC after PTGBD, by using signifi-  are less distinct and bleeding is more common due to repetitive inflammation episodes. 13 Therefore, a thickened GB wall might present a more challenging laparoscopic dissection.

| CON CLUS IONS
We proposed a new SS using the significant factors of time until PTGBD after symptom onset and the maximal GB wall thickness to predict the necessity of the bailout procedure during LC after PTGBD. Our SS for predicting the necessity of a bailout procedure correctly predicted the need and might be associated with a risk of BDI. Therefore, it might be better to select the flexibility of LSC or OC for patients with a score of 2 after PTGBD.

D I SCLOS U R E
Funding: The authors received no funding support for this article.
Conflict of interest: The authors declare no conflicts of interest for this article.