Interpreting surgical associations in primary sclerosing cholangitis: Caution in clinical messaging

Primary sclerosing cholangitis (PSC) remains one of the greatest unmet needs in modern hepatology, given its ill-defined etiology, critical absence of medical therapy, and the fact that transplantation is still the only life-extending intervention for patients. Moreover, clinical trials in PSC continue to be challenging, due to varying rates of liver disease progression, the effects of concomitant inflammatory bowel disease (IBD) and associated treatments, and a wealth of competing risk events. The latter include biliary and colorectal cancers that give way to surgical interventions, which themselves carry inherent risks. Reciprocally, as abnormal mucosal immune responses contribute to the development of PSC, it has long been questioned whether and how the presence of IBD, alongside the timing and type of colonic resection, impacts the clinical course of immune-mediated liver disease. In this issue, Leung et al.1 report on hepatopancreaticobiliary cancer risk and disease trajectories across a population-based, provincial cohort of patients from Canada. The overarching goal was to model colectomy and cholecystectomy as explicit intermediate disease states, and to account for time-dependency in surgical exposure. The study cohort consisted of ~500 patients diagnosed between 2002 and 2018, followed up for at least 3 years, and encompassed a host of clinically relevant events: colorectal cancer, colonic resection, hepatobiliary cancer, cholecystectomy, transplantation, non–PSC-related death, and PSC-related death. In so doing, the investigators validated several previous findings. Firstly, the majority of individuals presented with IBD many years (7.7 y) prior to their onset of liver disease. This contrasts with the minority in whom PSC was diagnosed prior to IBD, who tended to be of younger age (33 y compared with 47 y), with intestinal manifestations being identified, on average, within 12 months. During the study period, nearly 50 transition states were identified from the point of study entry, representing movement from the point of diagnosis to experiencing one or more clinical events. An important observation for future patient counseling is the fact that 40% of individuals remained event-free from the index date to the end of follow-up; whereas a single transition event from the point of study entry was observed for colectomy (11%), liver transplantation (9.6%), and death (7.1%). In turn, the 5-year and 10-year probabilities of experiencing more than one event were 4.3% and 13.3%, respectively. A controversial finding was the fact that post-colectomy and post-cholecystectomy states were associated with higher transition intensity ratios (TIRs) to mortality (3.08 and 3.85, respectively). Of note, the authors acknowledge that these elevated rates are unlikely to be causally related to the procedures per se and may in fact relate to the indications for surgical intervention. Most importantly, a sensitivity analysis using a time-dependent Fine and Gray model did not detect a corresponding increase in post-surgical mortality, further supporting this interpretation. Additionally, in the few individuals who had a colectomy at or before PSC–IBD diagnosis, or who had a colectomy during follow-up, 55% and 71% did not transition to other states, with 75% being alive at the end of study follow-up. From a clinical perspective, this nuance is essential. Without careful interpretation, the study could be misconstrued as evidence that colectomy or cholecystectomy shortens life expectancy in PSC–IBD. Such a conclusion would be inappropriate and potentially harmful. Patients should not defer, be discouraged, or be denied an indicated surgery out of fear that the procedure itself worsens prognosis. In many cases, withholding surgery would likely pose a greater risk than the intervention. The heterogeneity of colectomy type is also relevant. Several studies have shown that IBD classifications and surgical subtype matters (Figure 1), with proctocolectomy and permanent ileostomy being associated with improved event-free survival in PSC–IBD patients;2 whereas the presence of IBD, and/or a colectomy with restorative ileal pouch anal anastomosis is associated with poorer quality of life, no protective value from a liver perspective, and perhaps even a heightened risk of adverse PSC-related outcomes.2–4 These observations extend to liver transplant recipients, wherein IBD, particularly active colitis, is a risk factor for recurrent disease and graft loss;5–7 whereas colectomy prior to transplantation is protective8 when an ileostomy is retained and a restorative procedure avoided.9 Thus, grouping all colectomies into a single state may obscure important biological and clinical differences related to IBD activity, inflammatory burden, restorative compared to non-restorative procedures, and the timing relative to transplantation.FIGURE 1: Inflammatory bowel disease, colectomy, and the clinical course of PSC. Infographic summarizing contemporary literature on how the presence and subtype of inflammatory bowel disease, and the effects of colectomy (timing and type), affect the clinical course of primary sclerosing cholangitis pre-transplantation and post-transplantation. Values shown reflect hazard ratios from referenced studies unless otherwise indicated. Abbreviations: IBD, inflammatory bowel disease; IPAA, ileal pouch anal anastomosis; LT, liver transplantation; PSC, primary sclerosing cholangitis.Post-cholecystectomy data are also difficult to interpret, although 2 mechanisms can be proposed for a protective effect of the gallbladder in PSC: a buffer of increased bile duct pressure and regulation of bile acid composition. However, regarding PSC specifically, the obstruction is typically upstream. Moreover, bile acid composition is expected to normalize after cholecystectomy due to compensatory mechanisms. Nevertheless, previous work by Cazzagon et al.10 demonstrated that post-cholecystectomy, individuals with PSC manifest a more severely affected biliary tree. Cholecystectomy itself is a common intervention among PSC patients, likely reflecting cholestasis-related gallstone formation, increased cholecystitis risk, and the higher prevalence of gallbladder polyps. Together, these data suggest that the need for cholecystectomy may serve as a surrogate marker of biliary disease severity more broadly, rather than an independent determinant of clinical outcomes. Currently, there are no other studies available describing the associations of cholecystectomy on long-term solid clinical endpoints, and Leung et al. are to be commended for such a thorough evaluation as relates to PSC-specific outcomes. Finally, as the authors also note, the time-homogeneous Markov model may not incorporate prior disease duration nor cumulative inflammatory exposure into the transition hazard. Although surgery is modeled as an intermediate state, the analysis does not fully capture the pre-surgical trajectory. Given that most surgeries occurred after PSC–IBD diagnosis, the reported TIRs likely reflect heterogeneity in disease course rather than independent surgical effects. In summary, while the multistate model provides valuable prognostic insight, the associations between surgical states and mortality require cautious clinical interpretation. These findings should refine risk stratification and disease understanding, but must not discourage appropriately indicated surgical management in patients with PSC–IBD.