Endoscopic Sphincterotomy Increases the Risk of Pyogenic Liver Abscess: A Retrospective Study Using Real-World Data

Introduction

Endoscopic retrograde cholangiopancreatography (ERCP) is a procedure done with an endoscope that aims to diagnose hepatobiliary conditions, with the ability to intervene when required [¹,²]. The endoscope is inserted through the mouth into the second part of the duodenum. Once the sphincter of Oddi is visualized, another arm from the scope is inserted through the sphincter to gain access into the ampulla of Vater to reach for the common bile duct (CBD) or the pancreatic duct while being visualized under x-ray [³].

During ERCP, some patients undergo endoscopic sphincterotomy (ES), which involves cutting the sphincter of Oddi to open either the common bile duct or the pancreatic duct. ES serves several purposes, including the extraction of CBD stones and treating papillary stenosis and Sphincter of Oddi dysfunction [¹,²,³]. While this procedure has great benefits and is considered safe, various adverse events (AE) can occur, including bleeding, pancreatitis, cholangitis, and perforation [⁴]. However, one less common and overlooked AE is pyogenic liver abscess (PLA). PLA is an infected fluid collection in the liver that could be caused by biliary diseases, the spread of bacteria from blood or the GI tract, intrahepatic rupture of cholecystitis, or superinfection of necrotic tissue [⁵,⁶,⁷]. Gram-negative microorganisms, such as Escherichia coli, Klebsiella pneumoniae, and other anaerobic microorganisms, are the predominant organisms isolated from PLA [⁸].

A retrospective cohort study done in Taiwan showed an increased risk of PLA in patients undergoing ES to treat choledocholithiasis compared to patients who had ERCP without ES [⁹]. A case was reported in Portugal for a patient who developed PLA with sepsis with Escherichia coli, Streptococcus anginosus, and Enterococcus faecalis three days after he underwent ERCP with ES for choledocholithiasis [¹⁰]. In this study, we aim to evaluate the association between ES and PLA development.

Methods

Results

Discussion

In this retrospective cohort study, we found that ES was associated with an increased risk of post-ERCP PLA compared to ERCP without ES (RR: 1.37, 95% CI: 1.09–1.73; P = 0.007). This also corresponded to a higher incidence of sepsis with enteric organisms (RR: 1.37, 95% CI: 1.29–1.66; P = 0.001) and use of broad-spectrum antibiotics (RR: 1.38, 95% CI: 1.31–1.45; P < 0.001), thereby increasing the burden on the healthcare system. Despite the higher morbidity, we did not observe a statistically significant difference in the rates of procedural abscess drainage or mortality within one year following ERCP. To the best of our knowledge, this is the first study examining this association in the US and Western Hemisphere population, as earlier studies originated mainly from Taiwan.

Our findings align with and expand upon the existing literature regarding post-ERCP PLA risk. Prior studies have reported a similar association between ES and subsequent PLA. For instance, Peng et al. (2018) conducted a population-based cohort study in Taiwan and found that the incidence of PLA was significantly higher after ES than in those without ES (4.20 vs 0.94 per 1000 person-years), with an adjusted hazard ratio of 4.5, P-value < 0.001 [⁹]. They concluded that patients receiving ES have a markedly increased risk of liver abscess, which is consistent with our observation of increased risk in the ES group [⁹]. Another large retrospective cohort study from Taiwan by Wu et al. [¹¹] also demonstrated a higher cumulative incidence of PLA in patients who underwent sphincterotomy for choledocholithiasis compared to those who had ERCP without ES where on multivariate analysis, the ES increased the risk of PLA by an adjusted hazard ratio [aHR] of 1.49; 95% CI=1.12-1.98; p-value = 0.0058). These epidemiological findings reinforce earlier clinical reports, including multiple case reports [⁶,⁷,¹⁰] that had flagged PLA as a potential AE following ERCP with sphincterotomy. Notably, the magnitude of risk observed in our study (approximately 1.4-fold increase) is more modest than that reported in some of the Asian cohort studies, for instance, in Peng et al., who reported a fourfold or greater relative risk [⁹]. This discrepancy could stem from differences in study design (e.g., our analysis was limited to one-year outcomes and controlled via propensity matching, whereas Peng et al. examined longer-term hazard), variations in patient populations, or differences in baseline biliary disease severity. Additionally, geographic and practice differences – such as the microbiological spectrum (Klebsiella pneumoniae is a more common cause of PLA in East Asia [¹²]) or thresholds for performing sphincterotomy – might contribute to the varying effect sizes. Importantly, no prior research to our knowledge has shown a decreased risk of PLA with sphincterotomy; the trend across studies consistently supports our core finding that ES is a risk factor for PLA. Finally, while earlier studies primarily focused on the occurrence of PLA, our work adds nuance by examining outcomes like sepsis by gut derived organisms, PLA drainage and mortality. The lack of mortality difference we observed is in line with the notion that, if recognized early, PLA can be managed effectively [¹³,¹⁴].

Several mechanisms might explain why ES increases the likelihood of PLA. Physiologically, an intact sphincter of Oddi serves as a barrier between the duodenum and the biliary system. When a sphincterotomy is performed, this barrier is compromised, potentially allowing duodenal contents – including bacteria – to reflux into the bile ducts more freely [¹⁵]. This duodeno-biliary reflux can lead to ascending bacterial colonization of the biliary tree and, in susceptible individuals, the seeding of infection in the liver parenchyma. In essence, ES creates an open conduit for enteric microbes to access the intrahepatic biliary ducts, providing a direct path for infection that can result in liver abscess formation [¹⁵,⁹]. The bacterial ascent theory is further supported by the fact that ES increases the risk of acute cholangitis as well [¹⁶,¹⁷]. In a Swedish population-based study, Langerth et al found a significantly increased risk of acute cholangitis following ES (HR: 36, P <0.001), and noted that most of these episodes developed during the first four years following ES [¹⁸].

Additionally, the act of performing an ES (and ERCP in general) may introduce bacteria or cause transient bacteremia. Minor mucosal trauma or papillary edema from the cut could facilitate bacterial translocation into the bloodstream or bile ducts during the procedure, thereby increasing the risk of hepatic seeding. The organisms involved in post-ERCP abscesses are typically enteric Gram-negative flora and anaerobes, supporting this mechanism. Common isolates from PLA include Escherichia coli, Klebsiella pneumoniae, Streptococcus anginosus, and Enterococcus species, which are all gut-derived microbes [¹⁹,²⁰,¹¹]. The predominance of these organisms in PLA following ERCP/ES strongly implicates the gut as the source and is consistent with bacterial reflux or translocation after sphincterotomy. Another contributing factor is the alteration of biliary flow dynamics after ES. By widening the biliary outflow tract, ES reduces the tone that normally prevents bacteria from ascending, especially in the presence of biliary stasis [²¹]. Pre-existing biliary pathologies – such as choledocholithiasis, strictures, or malignancy – can compound this risk. This mechanistic understanding – duodenal bacterial entry combined with biliary stasis & injury – explains why ES, while facilitating bile drainage, also facilitates biliary and hepatic infections.

Given the increased risk of serious infection (PLA, cholangitis, and sepsis) associated with ES, several clinical precautions should be considered. While routine antibiotic prophylaxis for ERCP remains debated [²²], our findings suggest it may be beneficial in high-risk cases, such as patients with biliary obstruction, cholangitis, diabetes, or immunosuppression. Peri-procedural antibiotics could reduce PLA risk, but should be balanced against antibiotic stewardship concerns. Endoscopists should carefully assess the necessity of ES and consider alternatives, such as endoscopic papillary balloon dilatation, when feasible. Ensuring complete biliary clearance and using stents in cases of incomplete drainage can further reduce the infection risk [²³].

The management of PLA generally requires both antimicrobial therapy and source control. According to Rismiller et al., only a small proportion of patients (4 out of 64 in their study) were successfully treated with antibiotics alone, whereas the majority required percutaneous drainage (PD) or surgery for effective management [²⁴]. Similarly, Lo et al. identified that patients with PLA secondary to endoscopic interventions are at a higher risk of therapy failure (Odds ratio: 3.22, P-value: 0.01) when treated with antibiotics alone, emphasizing the need for early PD in these cases [²⁵]. On the other hand, Du et al. reported a 100% success rate in treating small liver abscesses (<5 cm in diameter) with antibiotics alone (53 of 125 cases in their retrospective study) [²⁶]. In summary, clinicians should implement preventive, procedural, and post-ERCP strategies to mitigate infection risk and ensure appropriate & timely treatment.

Our study has several limitations, mainly due to its retrospective design and reliance on the TriNetX database, EHRs, and ICD-10 codes, which may have introduced misclassification errors [²⁷]. While PSM helped balance confounders, residual confounding remains possible because factors such as ERCP urgency and biliary stone characteristics were not captured. Additionally, patients undergoing ES may have had more complex pathology, inherently increasing PLA risk. The retrospective nature of the study prevents establishing causation, and selection bias may have influenced outcomes. Variability in practice patterns across centers may affect generalizability, and our one-year follow-up may not have captured late PLA cases. Furthermore, we lacked granular data on PLA management. Despite these limitations, our large dataset and robust matching enhance validity. Furthermore, this is the first study that evaluates this important association in the US.

Future prospective studies are needed to confirm these results under controlled conditions. Future research should focus on confirming the ES-PLA association and identifying ways to reduce this risk. Prospective studies and trials are needed to determine whether administering prophylactic antibiotics lowers PLA incidence in ES patients and whether its benefit outweighs its risk.

Conclusions

Our study demonstrated that endoscopic sphincterotomy (ES) during ERCP was associated with an increased risk of pyogenic liver abscess (PLA), sepsis with enteric organisms, and the need for broad-spectrum antibiotics. However, it did not significantly impact mortality or the need for invasive abscess drainage. These findings reinforce the importance of recognizing PLA as a potential complication of ES and highlight the need for preventive strategies, including careful patient selection, procedural optimization, and enhanced post-ERCP care.