ERCP and EUS Procedures: Patient Selection and Complications

0:00Welcome to Board Pearls. This is episode two of seven of the ERCP and EUS Procedures chapter, in the Endoscopic Procedures module. In this episode we cover who belongs on the ERCP table: sphincter of Oddi dysfunction as a landmark trial rewrote it, the pregnant patient who has a radiation-free alternative, and the bleeding, perforation, and infection complications the whole workflow exists to prevent.

0:23The last episode was about getting into the duct safely once you had committed to the procedure. This one is about the decision that comes before that, and about the restraint it takes to decline the procedure when the objective evidence for benefit is absent.

0:38That same restraint, knowing when not to instrument the pancreas, is the whole story of sphincter of Oddi dysfunction in the current era, and a landmark trial rewrote it. The old framework split post-cholecystectomy biliary-type pain into three types by how much objective evidence came with it. Type one had it all, pain plus elevated liver enzymes plus a dilated bile duct. Type two had pain plus one objective abnormality, usually a transient enzyme bump during an attack. Type three was pain alone, with normal labs and a normal duct. That trial took the type three patients, the ones with pain and nothing else, and randomized them to biliary sphincterotomy, dual sphincterotomy, or a sham procedure, and the sham group did at least as well as the patients who got cut. Sphincterotomy bought them nothing, while exposing them to post-ERCP pancreatitis rates of five to fifteen percent. So type three is retired as a sphincterotomy indication. That is the central teaching point of the whole section.

1:40What that means at the bedside is precise. A post-cholecystectomy patient with biliary-type pain, normal liver enzymes, and a non-dilated duct is not an ERCP candidate, no matter how real the pain is. You reframe it as a functional pain disorder, now categorized under the Rome framework as a functional biliary or pancreatic sphincter disorder, or as centrally mediated pain. And you manage it with neuromodulators and behavioral therapy, not with a sphincterotomy. The same restraint applies to idiopathic recurrent acute pancreatitis when there is no objective ductal disease. Reflexive manometry and sphincterotomy there carry substantial PEP exposure for modest expected benefit, so the discipline is to not reach for the procedure.

2:28The other two types still have a role, and they divide by how much objective evidence backs the pain. Type one, with the full triad of pain, elevated enzymes, and a dilated duct, gets a biliary sphincterotomy without manometry. Manometry would only confirm what the biochemistry and imaging have already proven, that there is mechanical obstruction at the sphincter, so it adds risk without changing the decision. Type two, with pain and a single objective finding, sits in the middle. The probability of meaningful response is intermediate, so this is genuine shared decision-making. If you do proceed, the prophylaxis is non-negotiable, because sphincter manometry involves repeated pancreatic duct cannulation and is the single highest-PEP procedure in all of ERCP. Indomethacin, lactated Ringer, and a pancreatic duct stent, every time.

3:17And there is one finding here that will tempt you to over-call obstruction. A post-cholecystectomy patient with a bile duct of ten to twelve millimeters, and the reflex is to read that as pathologic dilation pointing to sphincter dysfunction. It is not, on its own. The gallbladder used to act as a reservoir, holding bile between meals while the sphincter stayed shut. Take the gallbladder out and the duct has to absorb that reservoir function, so it dilates modestly to accommodate continuous hepatic bile flow. A duct of ten to twelve millimeters in a post-cholecystectomy patient with normal enzymes is physiologic compensation, and by itself it does not justify ERCP.

3:56Pregnancy is where the threshold to do ERCP at all rises to its highest, and the reason is that you now have a second patient who absorbs every risk and gains none of the benefit. Diagnostic ERCP in pregnancy is simply not done. If you need to see the anatomy, you get an MRCP without contrast. And the contrast part matters. Gadolinium crosses the placenta and is avoided in pregnancy, so the question of whether to add gadolinium to optimize the images answers itself. If you need tissue or stone confirmation without therapy, EUS gives it to you with no radiation at all. ERCP is reserved for the situations where intervention genuinely cannot wait. Cholangitis. Choledocholithiasis with persistent obstruction. Severe gallstone pancreatitis with ongoing obstruction. A bile leak after a cholecystectomy injury.

4:46When you do have to proceed, every modification on the list traces back to protecting the fetus, and radiation is the central one. The principle is to deliver as little radiation as possible and as little as possible to the uterus. You shield the gravid uterus with lead. You run pulsed fluoroscopy at the lowest acceptable rate, four to eight pulses per second, instead of continuous imaging. You collimate to the smallest field. You lean on last-image-hold so you are not keeping the beam live just to look. The target is total fluoroscopy time under a minute when feasible. Stack those techniques and the cumulative fetal dose typically comes in under one millisievert per ERCP, which sits well below the fifty milligray threshold associated with measurable teratogenic risk. Timing follows the same fetal-risk logic. The first trimester is the most radiosensitive period and the highest teratogenic risk, and the third trimester brings premature labor and difficult positioning, so the second trimester is preferred whenever the procedure can be timed.

5:48Sedation in pregnancy runs on the same fetal-exposure reasoning. Propofol is preferred, fentanyl is acceptable in moderate doses, and midazolam is the one to avoid, especially in the first trimester. You will sometimes hear these described by the old FDA pregnancy letter categories, propofol as category B, fentanyl as category C, midazolam as category D. Keep in mind those letter grades were retired by the 2015 Pregnancy and Lactation Labeling Rule, which replaced them with narrative risk summaries, so the letters now survive only as board shorthand. Position the patient left-lateral so the gravid uterus does not compress the aorta and vena cava.

6:28That brings us to the complications the whole workflow exists to prevent, and they divide into two mechanisms that each predict their own management. The first is hemostasis, and the second is biofilm. PEP the last episode covered, running about five percent overall and far higher in high-risk patients, prevented by the prophylaxis stack. The next is bleeding.

6:52Post-sphincterotomy bleeding happens in one to two percent of cases and climbs with anticoagulation, because the sphincterotomy leaves a cut surface that has to clot. There are five classic risk factors, and the value of holding them is that each one names a distinct reason the cut will not stop bleeding. Coagulopathy and anticoagulation within three days both impair clotting at the cut surface itself. Pre-procedure ascending cholangitis means inflamed, edematous, friable tissue that bleeds more readily. Low endoscopist case volume tracks with technical inefficiency and longer cautery times. And bleeding during the procedure is the most direct predictor of all, because intraprocedural bleeding reveals a vascular anatomy at the papilla that is going to keep bleeding. Run a case list of patients all booked for sphincterotomy, and the one you worry about is the one carrying these factors, the coagulopathic or anticoagulated patient above all.

7:45That patient also tells you the alternative. In cirrhosis or coagulopathy, endoscopic papillary balloon dilation is often safer than a full sphincterotomy for stone extraction. The logic falls straight out of the bleeding mechanism. A balloon stretches the papillary opening open without creating a cut surface. So it preserves the sphincter and avoids the bleeding bed that coagulopathic mucosa cannot control, while still widening the orifice enough to pull the stone. For larger stones, you can extend that idea with a limited sphincterotomy followed by large-balloon dilation. The unifying idea is that you trade the cut for a stretch in exactly the patient whose problem is that cuts bleed.

8:25Perforation rounds out the mechanical complications, under one percent overall. It is more common with precut or needle-knife work, because those cuts are made without the anatomic guidance of an established cannulated tract. It is classified by the Stapfer system, one through four. Stapfer one is a free wall duodenal perforation, the most serious. Stapfer two is periampullary. Stapfer three is a ductal injury. Stapfer four is just retroperitoneal air on imaging with no clinical correlate, the most benign. Management scales with the type, and the gradient from one to four roughly tracks how surgical the problem is. And cholangitis after ERCP is its own avoidable failure, because it reflects an obstructed system that was only partly drained. The prevention is to achieve complete drainage at the index procedure rather than leaving a partially decompressed duct behind a stent, with antibiotic prophylaxis when full drainage cannot be assured.

9:24The second mechanism is entirely different, and it lives in the scope itself. The high-profile outbreaks of carbapenem-resistant organisms, including the KPC-producing Klebsiella strains, were traced to the duodenoscope. Specifically to the elevator mechanism, the little hinged part at the tip that swings the cannula up toward the papilla and makes ERCP possible at all. That elevator is a complex moving part full of tiny crevices and channels, and biofilm forms in those spaces and survives high-level disinfection, persisting from one patient to the next. So the very design feature that lets you cannulate is the feature that resists reliable cleaning, which is why this became a device problem rather than a technique problem.

10:06Once you see that the reservoir is the elevator, the FDA-driven solution follows directly. Eliminate the persistent reservoir. That means disposable elevator caps and disposable elevators. It means redesigned scopes that are easier to clean. And it means fully disposable single-use duodenoscopes such as the EXALT Model D and the aScope Duodeno. Single-use components give biofilm nothing to colonize. The same mechanism explains a result that surprises people. In a non-outbreak unit, doubling up on high-level disinfection or adding ethylene oxide sterilization confers no incremental benefit per the current FDA and multisociety guidance. The reason is precise. Neither one reaches the interior of a non-disposable elevator any more reliably than standard high-level disinfection does, so you gain no infection control while adding reprocessing time and scope wear. Layered on top is traceability. You document the scope serial number, the patient identifier, the reprocessing staff, and the date. That way any future cluster can be traced back to a specific scope-and-patient pairing.

11:13Pull the selection-and-complication half together. You decline the procedure when the objective evidence is absent, in type three sphincter dysfunction where a sham did as well as a cut, and in the pregnant patient who has a radiation-free alternative in MRCP or EUS. And you respect the two failure mechanisms the procedure can produce, the cut that bleeds and the elevator that harbors biofilm. Each one names the patient to protect and the device choice that protects them: the balloon instead of the cut for the coagulopathic patient, and the single-use component instead of the reprocessed one for infection control. Restraint and mechanism, not steps.

11:51That settles how to get into the duct safely and who belongs on the table. The next episode turns to what you do once you are in front of an abnormal duct. The obstructed and indeterminate biliary tree. How you get tissue out of a stricture that hides its malignancy, how you choose a stent for an obstruction, and how you resect an ampullary lesion through the scope.

12:10For the full chapter, the practice vignettes, and the topic-tagged question bank, head to board pearls dot com. You'll find the rest of the series on Apple Podcasts, Spotify, or wherever you listen to podcasts. That brings us to the end of episode two of seven of chapter twenty nine, and I'll see you in the next one.