Percutaneous treatment of gallbladder disease

Acute calculous cholecystitis in high-risk patients

Despite the dearth of high-quality data, the use of PC in patients with acute calculous cholecystitis (ACC) at high risk for surgery has become commonplace. In reviews of surgical databases, patients who receive PC are demonstrably older and have greater medical comorbidity than those receiving LC. ^,

A systematic review of cohort studies examining PC for ACC identified 53 studies examining the question; however, differences in outcome reporting, biases in control selection, and the significant heterogeneity of study populations made it impossible to draw conclusive recommendations on clear indications in high-risk surgical patients. The review confirmed that mortality rates after PC are high (15.4% vs. 4.5% with cholecystectomy), likely indicating selection biases favoring the use of PC in patients with higher comorbidity. Although the high mortality rate is likely inflated by the higher mortality in early cohorts, it remains elevated in contemporary cohorts, reflecting the high-risk population in whom PC is performed.

Only three randomized studies examining the use of PC in acute cholecystitis have been published; however, they ask different questions, and have significant methodologic limitations. Hatzidakis and colleagues randomly assigned 123 high-risk patients (Acute Physiology and Chronic Health Evaluation [APACHE] score > 12) with ultrasound-proven ACC or acute acalculous cholecystitis (AAC) to ultrasound-guided PC versus conservative medical therapy (63 and 60 patients, respectively; see Chapter 34 ). In this trial, ultrasound-guided PC was unsuccessful in 5% of patients who had to undergo computed tomography (CT)-guided PC. The authors were unable to demonstrate a difference in symptom resolution after 3 days or in 30-day mortality and concluded that PC was indicated if symptoms failed to resolve after 3 days of conservative medical management. The study, however, had significant methodologic flaws. There was no pre-planned power analysis, no measures taken to conceal allocation, and no attempt to blind the investigators, the patients, or the care providers. Moreover, an evolving procedural learning curve within the study period made PC appear more morbid than it likely is in more experienced hands. The conclusion of this trial (i.e., that PC should be delayed until more conservative measures have failed to improve symptoms within 3 days) is counter to the current thinking, which promotes the use of early PC if it is to be used at all. This early use of PC is supported by observational cohorts where delayed PC is associated with more bleeding complications and an increased length of stay.

The second randomized trial by Akyürek et al. randomly assigned 70 patients with ultrasound-proven AAC to PC followed by early LC 3 to 4 days later or to conservative management with delayed LC 8 weeks after recovery (33 patients). In this study, PC followed by early LC decreased hospital length of stay and costs compared with conservative management with delayed LC. There was no difference in the proportions of conversion to open cholecystectomy at operation. The authors did not define inclusion or exclusion criteria, and no definition was provided for “a high-risk” surgical patient. The study was also not adequately powered to detect meaningful differences in the outcomes it compared. Furthermore, the trial did not state how randomization was performed and lacked blinding or allocation concealment. Of note, the analysis was performed strictly per-protocol and there was a notable 13% drop-out rate. Such methodologic shortcomings make it difficult to make recommendations based on this study.

A Cochrane Collaboration Systematic Review on the use of PC in high-risk patients with ACC identified the aforementioned two randomized trials, ^, grading their quality as very low and their risk of bias as very high. The group was unable to generate any recommendation on the usefulness of PC compared with conservative therapy, calling for higher-quality randomized trials.

The largest multicenter randomized trial to date, the CHOCOLATE (Acute Cholecystitis in High-Risk Surgical Patients: Percutaneous Cholecystostomy Versus Laparoscopic Cholecystectomy) trial provides us with the highest-level evidence. It randomized high-risk patients with ACC, defined as patients with an APACHE II score between 7 and 15, to PC versus LC. The trial excluded critically ill patients (APACHE II score > 15), those not considered to be surgical candidates, and patients with delayed presentations (>7 days from symptom onset). In this population, as in the literature at large, patients undergoing PC had a rapid improvement in their symptoms, but the incidence of major complications was greater in the PC group at an interim analysis (12% vs. 65%, respectively; P < .001). Healthcare utilization and cost were also greater in the PC group. The major morbidity increase in the PC group was driven by reinterventions. Importantly, the 65% morbidity rate far exceeds what would be considered standard for PC. Nevertheless, it is clear from this study that patients who are physiologically able to tolerate a cholecystectomy should be offered surgery rather than PC. As a result of this trial and other data, the 2018 Tokyo Guidelines on the management of cholecystitis now recommends a cholecystectomy even in patients with a Tokyo grade II or III acute cholecystitis if they have a Charlson Comorbidity Index less than 6 , American Society of Anesthesiologists (ASA) class of 2 or less, good functional status, favorable organ system failure (defined as cardiovascular or renal organ system failure rapidly reversible during admission and before laparoscopic cholecystectomy for acute cholecystitis), and no negative predictive factors. This is in contrast to the 2013 guidelines, which recommended PC more liberally for acute cholecystitis grades II and III.

In conclusion, despite the absence of convincing high-level evidence, and based mainly on retrospective unmatched cohorts, PC is frequently performed in patients with ACC who do not respond to medical therapy and who are deemed unfit for surgery because of age or medical comorbidity. The evidence suggests that LC should be offered to patients with acute cholecystitis who can tolerate surgery.

Acute calculous cholecystitis with delayed presentation

PC, followed by interval LC, has been used in the specific context of ACC late in the course of the disease, hoping that conversion to open cholecystectomy would be less than during LC at the index presentation. This was the subject of a retrospective cohort study of patients with an ASA score of 1 or 2 presenting with ACC after more than 72 hours of symptoms who did not respond to nonoperative treatment for 48 hours. It compared early PC with PC and delayed LC 4 weeks later (48 patients) to LC (43 patients). The authors found a lower frequency of conversion to open surgery, shorter hospital stay, and fewer total complications with PC and delayed LC (40% vs. 19%; P = .029 in the early LC vs. early PC and delayed LC group). The main methodologic limitation of the study was selection bias: It is unclear how patients in the early cholecystectomy group were selected, and no attempt was made to adjust for baseline differences between the groups. Zehetner and colleagues analyzed patients with ultrasound-confirmed ACC presenting more than 72 hours after symptom onset. They compared 23 patients treated with PC with those treated by LC. The authors pair-matched patients for age, sex, race, body mass index, diabetes, and sepsis. Contrary to the previous study, the authors demonstrated increased length of hospital stay (LOS) with PC and no difference in major morbidity; however, there was a significant proportion of conversion to open surgery in the non-PC group (17%) and a trend toward higher 30-day mortality in the PC group (13%). The deaths in the PC group were because of advanced cancer, illustrating the significant selection bias that went into treatment attribution, a bias that was not accounted for by matching on basic demographic characteristics in a very small sample.