Reply to: Yang et al, Clinical Features of Recurrent Acute Pancreatitis
We appreciate Dr Li and colleagues’ interest in the findings of our study. They raised 2 main issues: First was regarding our definition of recurrent acute pancreatitis (RAP) and second our conclusion that RAP is associated with better outcomes.1
We defined RAP as an episode of acute pancreatitis occurring more than 2 months after the previous event with a pain-free interval between episodes. Dr Li et al proposed that the interval between the current and previous episode should be 3 months instead of 2.1 They maintained that smoldering symptoms and chronic elevation of pancreatic enzymes might take longer than 2 months to resolve.
To our knowledge, there is no consensus definition for RAP, and the length of pain-free interval needed between each episode of acute pancreatitis remains a debated subject.2–4 We believe that the most important criterion in defining RAP is the resolution of symptoms between episodes. All patients in our study with RAP reported resolution of symptoms for at least 2 months prior to the recurrent episode. Furthermore, patients with smoldering symptoms from the previous episode were excluded. We believe our exclusion criteria were sufficient to rule out patients with smoldering symptoms from their prior episode.
Dr Li et al reported significantly worse outcomes in patients who experienced RAP than observed in our study. Patients with RAP in Dr Li and colleagues’ center did poorly with respect to all outcomes. The mortality rate (5.88%), incidence of necrosis (72%), incidence of multisystem organ failure (44%), and percentage of patients needing intervention (47%) among RAP were strikingly higher than those in our population and most studies of pancreatitis in general.1 Intuitively, their data seem to suggest that RAP actually runs a much more severe course than the sentinel episode of pancreatitis. However, there are several aspects in their study that deserve closer attention.
Hyperlipidemia was the most common etiology of RAP in Dr Li and colleagues’ study population, accounting for 47% of patients with RAP. This proportion is significantly higher than that in our study population and other studies reported in the literature.4,5 As they stated, there is emerging evidence that patients with hypertriglyceridemia may experience a more severe course.6 As such, their data seem to support an association between severity of pancreatitis and a specific etiology rather than a causal relationship between severity of disease and recurrence.
Based on their data, there is a 72% chance that patients with RAP will suffer from pancreatic necrosis, and almost half of them will require an intervention. This strongly raises the question of selection bias for sicker patients in their cohort.
In our study, 7% of patients with RAP developed new areas of necrosis. We were careful to distinguish between those with new-onset and previously existing necrosis. This was to eliminate the chance of overestimating the rate of necrosis attributable to RAP. If we did not make this distinction, the rate of necrosis in our population would have been 25% instead of the 7%, which would have been a gross overestimation of rate of necrosis among patients with RAP.
Another possible explanation for the striking difference in outcomes is how organ failure was defined. We adhered to the modified Marshall Scoring System to define organ failure because the Revised Atlanta Classification of Acute Pancreatitis uses the system.7 Reported incidence of acute kidney injury, shock, and acute respiratory distress syndrome can depend on how organ failure was defined and may account for the observed differences in the incidence of organ failure.
We appreciate the interest and thoughtful contribution of Dr Li and colleagues’ data to this topic.