*3.2. Surgical Approach and Postoperative Outcomes*

As general management, the laparoscopic approach was the first choice for all patients of all ages. Open cholecystectomy was performed only when laparoscopy was not considered safe due to comorbidities or local conditions.

We noted a statistically significant difference between the age distribution for LC compared to OC and conversion: the mean age for LC is 55, while the mean age for OC and conversion is 68 (*p* < 0.001 for ANOVA test). However laparoscopic cholecystectomy was the most frequent procedure in all subgroups, with superior outcomes when compared to open surgery and conversion in terms of hospital stay and surgical and cardiovascular complications (*p* < 0.001).

Furthermore, the linear-by-linear association shows an increase in the conversion rate with age (*p* < 0.001). The frequencies of the conversion rate and the classic surgical approach were significantly higher in patients aged over 50.

Conversion to open was a surgical decision due to elective (lack of advancing in dissection and specimen removal, lack of critical view of safety—20 cases) or emergent causes (incontrollable hemorrhages—four cases; main bile duct lesion—one case, cholecysticduodenal fistula—one case). In the present study, we found no statistically significant differences between conversion and open cholecystectomy in terms of mortality, morbidity

and hospital stay. In the case of intraoperative main bile duct lesion, the conversion was imposed by the difficult dissection due to chronic inflammation of the cystic pedicle. The lesion was situated in the proximity of the cystic duct and was classified as minor according to the Mc Mahon Classification (<25% of main bile duct diameter) and was repaired by a T tube insertion. Large papillosphincterotomy was performed by endoscopic retrograde cholangio-pancreatography (ERCP) in the early postoperative period (3 days later) to allow faster recovery.

The classic approach of first intention was used in a total of 12 cases (one in group A, four in group B, five in group C, two in group D). The causes for open surgery were: increased local inflammation (gangrenous gallbladder, biliary peritonitis) in eight cases, the association of the main biliary duct lithiasis with mechanical jaundice ± angiocholitis (two cases) and a history of previous surgical interventions in the upper abdominal region (two cases) (Table 4).

**Table 4.** Surgical approach and postoperative outcomes in the 4 subgroups.


(1) Fisher's exact test; (2) ANOVA Linearity test; LC: laparoscopic cholecystectomy; OC: open cholecystectomy; Drain insertion was not a routine practice in our clinic for laparoscopic cholecystectomy; \* SD—standard deviation.

> The cases in which drainage of the subhepatic space was considered necessary were those cases with severe local inflammation, increased intraoperative bleeding or suspected lesion of the bile duct. The fact that the drain was used more often in the elderly is well correlated with the increased incidence of the moderate and severe forms with advanced age. Drainage was used in all cases with open surgery and conversion to open.

> The postoperative outcome was favorable in most cases for all age subgroups. No patients required re-surgery in the following 30 days. Surgical related complications were managed conservatory: hemorrhages (seven cases), bile leakage (nine cases), one septic intraperitoneal collection and one main bile duct lesion, classified as minor according to the Mc Mahon Classification solved by ERCP stenting. The procedure consisted of papillosphincterotomy, and a plastic material 7F stent of 10 cm length was introduced in the main bile duct to allow healing. The stent was removed after 3 months, with a favorable outcome. Surgical site infections were less common in laparoscopic cholecystectomy vs. open cholecystectomy and conversion (Table 5), and increased with age.


(1) Fisher's exact test; (2) ANOVA Linearity test; LC: laparoscopic cholecystectomy.

The rate of surgery related complications was significantly higher in patients over 50 years old (*p* = 0.045), which also proved to be a turning point for an increasing rate of conversion and open surgery. However, the comparative incidence did not differ significantly between patients aged from 50–64 years, 65–79 years and over 80 years. (6.3%, 9.09% and 5.8%, respectively).

The Fisher's linear discriminant analysis was performed to identify the risk factors significantly related to surgical complications. The highest correlation was found with systemic comorbidities: diabetes (r = 0.813) and chronic bronchopneumopathy (r = 0.502) and CCI (r = 0.381, but with no significant increase in discrimination power). Among the local factors, the severity of inflammation and the presence of gangrenous cholecystitis had the most significant predictive power (r = 0.288), followed by fibrinogen (r = 0.348), and TG13/TG18 severity forms (r = 0.218).

Severe cardiovascular complications encountered in the study group were: acute myocardial infarction (nine cases), stroke (seven cases) and malign arterial hypertension (two cases). In total, three out of four causes of death were cardiovascular acute events. Only one patient died of sepsis: a diabetic patient aged 57 with a severe form of acute cholecystitis. The incidence of severe cardiovascular postoperative complications increased with age (ANOVA test for linearity: *p* < 0.001; Mantel-Haenszel test for trend: *p* < 0.001). There were no statistically significant differences between the incidence of cardiovascular complications in groups B, C and D (*p* = 0.344).

## *3.3. Multivariate Analysis of Risk Factors for Open Surgery and Conversion*

In order to describe the preoperative and intraoperatory patients' characteristics which determined the applied surgical procedure (LC = Laparoscopic Cholecystectomy, Conversion or OC = Open Cholecystectomy), we have used the stepwise variant of Fisher's linear discriminant analysis. The discrimination between the classes is based on the two Canonical Discriminant Function described in Table 6. The Canonical Discriminant Function is displayed in standardized form in order to allow the comparison of the importance of each variable.

**Table 6.** Standardized Canonical Discriminant Functions for (LC, OC, Conversion).


The variables significantly correlated with Standardized Canonical Discriminant Function F1 are gangrenous cholecystitis (r = 0.807), leukocytes (r = 0.650), fibrinogen, and severity form classified by TG 13/18. The variables significantly correlated with Standardized Canonical Discriminant Function F2 are total bilirubin (r = 0.637), CCI (r = 0.531) and high aspartate transaminase (AST) and alanine transaminase (ALT) (r = 0.351), previous history of stroke (r = 0.296), diabetes (r = 0.223) and cardiovascular disease (r = 0.236). The parameters not included in the definition of F1 and F2 are clinically significant, but they do not add a supplementary increase in the discrimination power.

F1 could be labeled as the score of inflammatory risk (higher values of leukocytes, the presence of severe inflammation and higher age imply high values if F1), and F2 could be labeled as the score of comorbidities (CCI and associated pathologies). Main bile duct complications, such as lithiasis, angiocholitis, and Mirizzi Syndrome (characterized by increased bilirubin), but also increased inflammation with a secondary increase in bilirubin, are also associated with Function 2.

Figure 3 suggests the following simple interpretation: small and moderate values of F1 and F2 (near zero) generally characterize the laparoscopic approach; positive values of F1 (severe inflammation and sepsis) and negative values of F2 generally characterize the open approach; and positive values of F1 and F2 (association with severe inflammation and comorbidities/main bile duct complications) generally characterize conversion.

**Figure 3.** Patient representation (*n* = 333) in the space (F1, F2) of Unstandardized Canonical Discriminant Functions between (LC, OC, Conversion). (Specification: functions at Group Centroids are: (−0.203, −0.004) for LC (laparoscopic cholecystectomy); (1.739, −0.964) for OC (open cholecystectomy) and (1.564, 0.508) for Conversion). Wilks' lambda computed for the two canonical functions are significant (test of the two functions: chi-square (df = 10) 111.08, *p* < 0.001; test of second function: chi-square (df = 4) = 17.05, *p* = 0.002). Cross-validation of the model: 81.4% of cross-validated cases are correctly classified. The relative dispersion of patients with conversion to open surgery indicates that other factors, such as surgical experience or particular intraoperative findings, may be involved.

#### *3.4. Multivariate Analysis of Risk Factors for Adverse Outcome in the Eldery*

The incidence of acute cardiovascular events in the early postoperative period increases statistically significantly in patients with ASA ≥ 3, and that of deaths in ASA ≥ 4 (*p* = 0.001). When the correlations between the severe forms of acute cholecystitis and the occurrence of complications were analyzed, statistical analysis showed that severe forms with organ/system dysfunction correlated with the incidence of severe complications and deaths, for all age groups.

Regarding the type of operation, the incidence of cardiovascular complications is significantly higher in the case of the open approach and conversion in comparison with laparoscopic cholecystectomy. However, conversion and open surgery were chosen in severe forms, with necrotic gallbladder, pericholecystic plastron or biliary peritonitis. Multivariate analysis of preoperative and intraoperative risk factors shows that the incidence of severe cardiovascular complications and deaths correlates best with the severity of the septic process and inflammation (gangrenous cholecystitis, fibrinogen > 400 mg/dL and Grade III cholecystitis according to TG13/18 severity forms), and among comorbidities, with a previous history of stroke, chronic renal failure and diabetes (Table 7).


**Table 7.** The Fisher's linear discriminant analysis for cardiovascular severe complications and mortality.

#### **4. Discussion**

Increased technical experience with laparoscopic cholecystectomy favorably affected outcomes over time [26]. Together with the important achievements in intensive care, more patients, initially considered at risk, can benefit from the important advantages of minimally invasive surgery. The present contraindications for laparoscopic cholecystectomy are few, and they may be classified as absolute (uncorrected coagulopathy, high anesthetic and surgical risk, gallbladder carcinoma) or relative. The latter includes either general conditions (end-stage liver disease) or local findings (previous surgery in the upper abdominal region, calcified gallbladder, cholecysto-enteric fistula, Mirizzi's syndrome) [27]. Age and severe inflammatory forms, such as gangrenous and emphysematous cholecystitis, are no longer considered unsuitable for laparoscopy [28]. In the present study, we analyzed the factors that influence the surgical decision the most. We found that severe local inflammation as well as a high CCI and high values of total bilirubin could favor open surgery or conversion. Other unquantifiable factors such as local anatomy, tissue friability, or surgeon's experience may play a significant role in the decision to convert to open.

Hyperbilirubinemia significantly increases the likelihood of finding common duct stones in patients with acute cholecystitis, but it also occurs in patients with acute cholecystitis without common duct stones. In these cases, the increase in value is mild and it returns to normal values quickly after resolving the septic process. The significance of bilirubin in acute cholecystitis and other intraperitoneal infections was also investigated by other authors [29–36]. Hyperbilirubinemia in acute abdominal infections is caused either by the excessive production of bilirubin or by altered clearance. Both mechanisms lead to bilirubin accumulation and play a role in the hyperbilirubinemia observed in patients with appendicular perforation. Patients in severe sepsis express proinflammatory cytokines, with cholestasis triggered by nitric oxide, by blocking bilirubin conjugation and elimination at the hepatocellular and intraductal level [32]. Common pathogens of the biliary and digestive wall, such as Escherichia coli and Bacteroides fragilis, were supposed to interfere with hepatocyte microcirculation, inducing sinusoidal lesions [35]. In addition, Escherichia coli infection has been shown to induce hemolysis of normal erythrocytes. This results in increased bilirubin loading in infected individuals, a process that promotes hyperbilirubinemia [34–36].

There are concerns about using the laparoscopic approach in patients with respiratory and cardiovascular comorbidities due to the metabolic effects of the induced pneumoperitoneum. This loss of reserve capacity is the single most important factor that decreases the elderly patient's ability to tolerate operations. The proper management of fluid and electrolyte replacement, respiratory management to prevent atelectasis and pneumonia, and monitoring for possible cardiac complications are necessary to minimize the risk of systemic complications in the perioperative period [2,3,37,38].
