This is the first pediatric study to specifically evaluate patients undergoing ERCPs with underlying hemolytic disorders. We found that existing ERCP criteria perform poorly in patients with hemolytic diseases. The existing criteria overestimate their risk of choledocholithiasis, possibly resulting in unnecessary ERCPs and their associated risks and costs. Current criteria rely on pre-procedure laboratory and imaging findings; in our cohort, there were no differences in these pre-ERCP evaluations in patients with hemolytic diseases with a common bile duct stone removed at ERCP and those without. These patients' clinical presentations and laboratory values, especially bilirubin, can be clouded by disease-specific factors such as ongoing hemolysis, intrahepatic sickling with or without infarction (in sickle cell disease), and iron overload related to blood transfusions. Total bilirubin includes components of conjugated or direct and unconjugated or indirect bilirubin. Unconjugated bilirubin in the setting of hemolysis is expected to be elevated. Therefore, risk stratification for choledocholithiasis is more complicated than in the general pediatric population and continues to be challenging. The most recent ASGE guideline made a conscious effort to increase specificity to greater than 90% by using more stringent criteria (3).

The poor performance of bile duct size and identification of a stone on ultrasound is consistent with non-hemolytic pediatric data that our group has previously published, in which there was no significant difference in ultrasound findings between patients who had a stone removed during ERCP and those who did not (4). Ultrasound overall seems to have poor sensitivity in detecting choledocholithiasis but is frequently used given its accessibility and lack of radiation.

Few patients in our study underwent other peri-procedure evaluations such as endoscopic ultrasound, MRCP, or intraoperative cholangiography, as the use and timing of these modalities is center-dependent. Nevertheless, these may be very important modalities given the poor ability to predict choledocholithiasis through labs and other imaging modalities in patients with hemolytic diseases. For example, a prior study demonstrated that MRCP better predicts choledocholithiasis in children than transabdominal ultrasound or CT, though it is important to consider the timing of MRCP in relation to ERCP as stones may pass if there is delay (4). The delay between MRCP and ERCP is of particular relevance in pediatrics, as many pediatric patients require anesthesia to undergo MRCP and therefore MRCP is oftentimes not performed on the same day as the ERCP, potentially allowing time for the stone to pass prior to ERCP. Additionally, some pediatric centers lack the ability to perform sedated MRCP on weekends.

While very few children in our study underwent intraoperative cholangiogram (IOC) or endoscopic ultrasound (EUS) (n = 1 and 6, respectively), these modalities appeared to perform well; the patient who had a positive IOC had a stone removed during ERCP, and all three of the patients who had a positive EUS had a stone removed during ERCP (Table 3). However, these are very small samples. EUS has been recommended in select patients to diagnose choledocholithiasis with good specificity (9, 10). However, in the adult literature, EUS is thought to have higher sensitivity for detecting small stones (9, 10). While EUS is a relatively recent development in pediatric gastroenterology, it has been shown to be safe and have good technical outcomes in children (11–13). EUS is radiation-sparing and, in at least two studies in the pediatric population, has been shown to successfully prevent a subset of patients from undergoing non-therapeutic ERCP (12, 13). Regarding IOC, there is some data that performing IOC prior to planned ERCP can prevent non-therapeutic ERCP in a subset of patients who have preoperative labs and imaging that are falsely concerning for choledocholithiasis (14).

Overall, increased use of peri-procedure evaluations may decrease the incidence of non-therapeutic ERCPs in the hemolytic disease population, but more data is needed to determine acceptable sensitivity and specificity of these modalities in pediatric patients. Additionally, the potential costs and risk of additional time spent under sedation to perform these evaluations must be weighed clinically. This dilemma is of particular relevance in the hemolytic disease population. For example, when considering patients with sickle cell disease who have a history of stroke or acute chest syndrome, one must balance the desire to avoid anesthesia exposure with potential ERCP complications such as pancreatitis, which could also prompt an episode of acute chest syndrome.

Finally, while not statistically significant, the increased number of adverse events in patients with hemolytic diseases highlights the importance of careful patient selection for ERCP. In particular, the suggestion of increased bleeding in patients with hemolytic diseases is of interest. Our incidence of bleeding in 2.5% of patients (7/277) is slightly higher than the previously published incidence of 1.2% from the PEDI database, though we did not sub-categorize by severity of bleeding in this substudy and the difference between patients with and without hemolytic diseases was not significant (7). Sickle cell disease, the most common hemolytic disease in our patient population, is typically conceptualized as a prothrombotic state that predisposes patients to cerebral infarction and venous thromboembolism (15); however, there are some reports of spontaneous bleeding events in adults with sickle cell disease, mainly in the gastrointestinal tract (16). A plausible contributing factor to bleeding is heavy NSAID use due to chronic sickle cell-related pain, but this is unlikely to fully explain the bleeding risk in this population. Overall, further investigation is needed to determine which factors would better predict the need for ERCP in patients with hemolytic diseases.