Subscribe
Share
Share
Embed
The final episode turns EUS from a guide for therapy into a diagnostician and a palliator, and it runs on one habit: reason from substrate to answer. The five alternating wall layers plus echotexture place a subepithelial lesion and narrow the differential before a needle moves, so a hypoechoic layer-four mass is a GIST until tissue proves otherwise. The needle choice then turns on a single distinction, cells versus architecture, because lymphoma, GIST, and autoimmune pancreatitis all live in structure a core preserves and cytology destroys. Finally the same probe that finds the tumor treats its pain: alcohol neurolysis for cancer, a reversible steroid block for benign disease, with complications that read straight off the interrupted sympathetic outflow.
Topics covered
Key decisions
This is an AI-generated podcast, and some pronunciations may be imperfect. Thank you for your understanding, and we hope you enjoyed this content.
Study the full chapter on Board Pearls, with practice questions, tables and primary-guideline references: ERCP and EUS Procedures
Read this episode: boardpearls.com/gi/episodes/ercp-and-eus-procedures-ep7
Questions or feedback: hello@boardpearls.com.
The most complete gastroenterology board-review podcast on the web, and it is entirely free. This is the GI series from Board Pearls: board review built around clinical reasoning, not recall. Every episode takes one high-yield topic and works it the way you would on rounds: a case to anchor it, the framework that sorts the differential, and the specific decisions the exam rewards. No trivia, no textbook dictation, just the thinking that earns the point.
Over 100 episodes span the full blueprint across nine modules (esophagus, stomach and duodenum, small bowel, colon, pelvic floor, liver, pancreas and biliary, endoscopy, and the cross-cutting topics), grouped by chapter and built from the guidelines and pivotal trials the boards are written from: ACG, AGA, AASLD, and ASGE.
Pair it with the written curriculum, question bank, and AI tutor at boardpearls.com. Questions or feedback: hello@boardpearls.com.
Welcome to Board Pearls. This is episode seven of seven of the ERCP and EUS Procedures chapter, in the Endoscopic Procedures module. In this episode we cover what EUS does beyond the ducts: reading the gut-wall layers to identify a subepithelial lesion, choosing the needle that answers the diagnostic question, and using the celiac plexus to control pancreatic cancer pain.
Start with the wall itself. When the endoscope finds a smooth bulge under intact mucosa, the surface tells you almost nothing. The whole question is what sits underneath, and EUS answers it by resolving the gut wall into five alternating bands. The reason the layers alternate is acoustic. Tissue interfaces reflect sound and read bright, or hyperechoic, while the bulk of muscle and deep mucosa scatter less and read dark, or hypoechoic.
So when a clinician reads the wall, they read it as a stack of bright and dark, from the lumen outward. The first band is the superficial mucosa and the interface echo, and it reads bright. Just deep to it the deep mucosa reads dark. Then the submucosa, dense with connective tissue interfaces, reads bright again. Below that the muscularis propria, a solid sheet of muscle, reads dark. And the outermost serosa reads bright. Bright, dark, bright, dark, bright, from inside out.
That alternating pattern is not trivia. It tells you where a lesion lives, and where it lives narrows the differential before a needle ever moves. A lesion that arises from the submucosa, the third layer, behaves differently from one rooted in the muscularis propria, the fourth. The echotexture inside the lesion then sharpens the read further.
Take the classic case. A hypoechoic, homogeneous mass arising from layer four, the muscularis propria. That is the textbook gastrointestinal stromal tumor, the GIST. It is dark because it is cellular and uniform, and it sits in the fourth layer because that is where the interstitial cells of Cajal that give rise to it actually reside. The location is the diagnosis until proven otherwise.
But notice the trap, because the muscularis propria houses two tumors that look the same. Leiomyoma is also a layer-four hypoechoic mass, smooth and homogeneous, and on the image it is indistinguishable from GIST. The temptation, when a small benign-looking mass sits in the muscle, is to call it a leiomyoma and move on. You cannot. The echo image does not separate these two, and the separation matters because GIST has malignant potential and leiomyoma does not.
What separates them is immunohistochemistry on actual tissue. GIST stains positive for CD117, also called KIT, and positive for DOG1. Leiomyoma stains positive for desmin and smooth muscle actin and is negative for CD117. So a layer-four hypoechoic mass is a tissue question, not an imaging question, and the only honest answer comes from staining.
That is also why size in this layer drives the decision to sample. Any muscularis-propria lesion two centimeters or larger warrants a tissue diagnosis with biopsy and consideration for resection. The reason is that within the fourth layer, increasing size correlates with malignant behavior. A small lesion may be watched. Once it crosses two centimeters in the muscle, the risk has shifted enough that you need to know what it is.
The other layers fill in the rest by their echotexture. A bright mass in the submucosa, the third layer, is a lipoma, fat reflecting brightly, typically incidental and benign, and usually diagnosed by appearance alone. A black, echo-free mass in the third layer is a duplication cyst, anechoic because it is fluid with no internal interfaces to reflect sound. A neuroendocrine tumor sits in the second or third layer and reads dark. A granular cell tumor is typically dark in the third layer. Pancreatic rest is heterogeneous and often crosses several layers, which is itself the clue.
So the layer plus the echotexture gives you a working diagnosis, and the working diagnosis tells you whether you even need a needle. A bright submucosal lipoma usually needs nothing. A dark layer-four mass at or above two centimeters needs tissue. And once you decide to sample, the question shifts from where the lesion is to what answer the tissue has to provide, because that determines which needle you reach for.
The needle choice turns on a single distinction: cells versus architecture. A fine-needle aspiration needle uses suction to pull cells into a hollow lumen, and it gives you cytology, cells in suspension. A fine-needle biopsy needle uses a specialized cutting tip, the Franseen tip on the Acquire or the fork tip on the SharkCore, to shear off an intact core of tissue. That core preserves architecture, the arrangement of cells, glands, and stroma, not just the cells themselves.
Whether that distinction matters depends entirely on the diagnosis you are chasing. Pancreatic adenocarcinoma can be diagnosed on cytology alone, because the malignant cells themselves carry enough atypia to make the call. For that question, cells are enough. But several diagnoses are invisible to cytology because they live in the architecture, and those are the ones that demand a core.
Lymphoma is the clearest example. You cannot diagnose lymphoma from loose cells, because the diagnosis depends on the follicular pattern and on immunohistochemistry and flow cytometry performed on intact tissue. GIST needs the CD117 stain we already met, and that stain reads on tissue. Autoimmune pancreatitis needs histology to show its signature, which is storiform fibrosis, obliterative phlebitis, and IgG4-positive plasma cells numbering more than ten per high-power field. None of those features survives as a cell suspension. They are structural, so they require a structure.
That is why fine-needle biopsy gives higher diagnostic yield than aspiration for solid pancreatic masses, lymphoma, GIST, and autoimmune pancreatitis. The mechanism is simply that the core preserves what cytology destroys. So the rule reduces to a question you ask before the procedure. Does the diagnosis live in the cells, or in how the cells are arranged? If it lives in the arrangement, you bring a biopsy needle.
This also resolves the role of rapid on-site evaluation, ROSE, the cytopathologist in the room confirming that a pass has enough cells before you stop. ROSE earned its place in the aspiration era, when immediate feedback on cellularity prevented a nondiagnostic procedure and let the operator know when to keep going. With modern biopsy needles, the core is diagnostic on its own, regardless of whether anyone is standing at the microscope. So ROSE is largely obviated by fine-needle biopsy and is no longer mandatory in most settings.
Cyst fluid follows the same logic of going to the right substrate for the answer. When the lesion is a pancreatic cyst rather than a solid mass, the panel is the fluid: carcinoembryonic antigen, glucose, amylase, and cytology. The key thresholds track the epithelium that lines the cyst. A CEA above one hundred ninety-two nanograms per milliliter and a glucose below fifty milligrams per deciliter favor a mucinous cyst. The glucose drop has a mechanism worth holding. Mucinous epithelium consumes glucose as it metabolizes, so the fluid runs low. A pseudocyst, lined by no true epithelium, instead retains the glucose that diffused in from serum. High amylase, by contrast, points to ductal communication, which fits a pseudocyst or a branch-duct intraductal papillary mucinous neoplasm.
Molecular markers refine the picture when CEA and glucose are equivocal. Mucinous cysts and IPMNs frequently carry KRAS and GNAS mutations, with GNAS nearly exclusive to IPMN, while serous cystadenomas carry VHL mutations. These are not standard everywhere, but the principle is the same as the needle principle: the test you order should match the cell behavior you are trying to confirm.
That moves us outside diagnosis entirely, to the one place where EUS reaches into the celiac plexus not to sample it but to treat through it. The patient is someone with unresectable pancreatic cancer whose pain has outrun what opioids can safely do. They are on escalating morphine equivalents, still in pain, and now constipated, sedated, and confused from the dose itself. The drug is failing on both ends. That is the patient for celiac plexus neurolysis.
The mechanism is direct. Under EUS guidance, the plexus is identified anterior and lateral to the celiac axis. Absolute alcohol mixed with bupivacaine is then injected into the celiac ganglia or the surrounding plexus. The alcohol denatures axonal proteins and destroys the nerve, while the bupivacaine gives earlier, shorter blockade on top. Destroying those visceral afferents interrupts the pain signal traveling from the upper abdomen, and that is what buys the patient relief and lets you back off the opioids.
There are two ways to deliver it, and the difference is targeting. Central neurolysis injects the alcohol around the celiac trunk and aorta to bathe the whole plexus through the perivascular fat. Direct ganglion-targeted neurolysis instead places smaller-volume injections into each visible ganglion. When the ganglia are clearly seen on EUS, the direct approach has shown better analgesia in some series, because it concentrates the agent at the actual target rather than dispersing it through fat. So the rule of thumb is straightforward: if you can see the ganglia, hit them directly.
Now the distinction the boards lean on, and it is a distinction of indication, not technique. Everything just described is for cancer. For benign chronic pancreatitis pain, you do not perform neurolysis. You perform a celiac plexus block, bupivacaine plus a corticosteroid like triamcinolone, which is reversible and does not destroy the nerve. The reason is the time horizon. In cancer, life expectancy is short and durable analgesia is the priority, so permanent destruction is acceptable. In benign chronic pain, the patient may live for decades. The destroyed nerve regenerates and gives poor long-term relief anyway, and irreversible neurolysis risks lasting deafferentation. So the steroid block reduces inflammation, gives shorter-term relief, and leaves the nerve intact.
Hold that as a single rule. Cancer gets alcohol, which destroys. Benign disease gets bupivacaine and steroid, which calms and reverses. Seeing visible ganglia does not change that. It changes how you inject in the cancer patient, not whether the chronic-pancreatitis patient should get alcohol at all. They should not.
The complications follow from one piece of neuroanatomy: the celiac plexus carries the sympathetic outflow to the upper gut and its vasculature. Block that outflow and the consequences are predictable. Serious adverse events are genuinely rare, on the order of two to six per thousand. What is common is a set of transient, minor effects, and those come straight from removing sympathetic tone.
The most common is transient diarrhea, in roughly four to fifteen percent. The mechanism is unopposed parasympathetic activity once the sympathetic brake is gone, which accelerates colonic transit and stimulates secretion. It typically resolves within twenty-four to seventy-two hours as the autonomic balance recalibrates, and loperamide helps. Next is orthostatic hypotension, also common, reflecting blunted sympathetic vasomotor tone after the plexus is interrupted; it lasts hours to a couple of days and is blunted by intravenous prehydration. And a transient pain flare in the first twenty-four hours is recognized and managed with a short course of opioid escalation before the durable relief sets in.
So picture the patient six hours out. They are light-headed on standing, with a few loose stools and a flare of the baseline pain. But they are afebrile and stable supine, with stable hemoglobin and no neurologic deficit. That whole constellation is the expected autonomic aftermath, not a catastrophe. The teaching move is to recognize it as physiology. You treat it with hydration, loperamide, and reassurance, rather than chasing a complication that is not there.
What you do screen for are the rare serious events that look different. Retroperitoneal abscess and hematoma are rare. A hematoma announces itself with sustained back or flank pain, hemodynamic instability beyond simple orthostasis, and a falling hemoglobin. That is nothing like the stable picture just described. And paraplegia, rare but devastating, comes from inadvertent injection into a spinal radicular artery, because the artery of Adamkiewicz can arise near the celiac axis. Alcohol entering that vessel produces spinal cord infarction. It presents with motor and sensory deficits in the legs, not with light-headedness and loose stools. Careful technique and small-volume injection lower the risk but cannot eliminate it.
Step back and the whole episode runs on one habit of mind. EUS lets you reason from substrate to answer. The layer and echotexture place the lesion and narrow the differential, so you bring the right needle. The diagnostic question, cells or architecture, picks aspiration versus biopsy. The fluid chemistry reads the epithelium that made the cyst.
And when the target is pain rather than tissue, the same anatomy the needle reaches becomes the thing you treat. The agent is matched to whether the disease is malignant or benign, and the complications read directly off the sympathetic outflow you just interrupted. Match the tool to the substrate, and the answer follows.
That brings us to the end of the ERCP and EUS chapter, and the next chapter steps out of the duodenum and into the colon. Chapter thirty is Colonoscopy Practice and Quality. There the organizing question shifts from what a single lesion is to how well a whole exam performs. That means colorectal cancer screening, detection metrics like the adenoma detection rate, and the rules for polyp resection and surveillance intervals.