Diagnostic workup of the indeterminate biliary stricture — ERCP brushing, FISH polysomy, and cholangioscopy-guided biopsy
A biliary stricture found on ERCP cholangiogram or MRCP without a clear benign explanation is one of the most consequential diagnostic dilemmas in gastroenterology. Roughly two-thirds of such strictures ultimately prove malignant, but a substantial minority are benign mimics that imaging alone cannot reliably exclude — and the downstream decision (major hepatobiliary resection vs. surveillance) carries real morbidity in either direction. Tissue confirmation is therefore the pivot point of the entire workup.
On cholangiography, malignant strictures classically appear as irregular, asymmetric, "shouldered" narrowing — an abrupt transition with overhanging edges — accompanied by marked dilation of the proximal (upstream) ductal system as bile backs up behind the obstruction. Benign strictures more often taper smoothly and symmetrically into the narrowed segment, with less dramatic upstream dilation and, frequently, a longer, more gradual stenotic segment.
The problem is that these patterns overlap substantially. Chronic inflammatory strictures can look irregular and shouldered; early, infiltrative cholangiocarcinomas can look deceptively smooth and tapered, particularly when growing along the duct wall (periductal-infiltrating pattern) rather than as a discrete mass. Cross-sectional imaging (CT, MRI/MRCP) adds information — mass effect, vascular encasement, regional adenopathy, hepatic lobar atrophy — but even combined with cholangiographic morphology, imaging alone cannot achieve the certainty required before committing a patient to a major hepatic or pancreaticobiliary resection, or conversely to watchful surveillance of what could be an early, resectable cancer.
Malignant causes: perihilar cholangiocarcinoma (the most common site, arising at or near the biliary confluence — a Klatskin tumor), distal and intrahepatic cholangiocarcinoma, pancreatic ductal adenocarcinoma compressing or invading the distal common bile duct, ampullary carcinoma, gallbladder carcinoma with direct biliary invasion, and extrinsic compression from metastatic or locally advanced periportal lymphadenopathy.
Benign causes: primary sclerosing cholangitis (PSC), IgG4-related sclerosing cholangitis (a steroid-responsive mimic of both PSC and cholangiocarcinoma), post-surgical or anastomotic strictures (after cholecystectomy, liver transplantation, or bilioenteric anastomosis), chronic pancreatitis with distal common bile duct fibrosis, stone-related fibrotic strictures, and, less commonly, Mirizzi syndrome or portal biliopathy. PSC deserves special emphasis: patients with PSC already have chronically abnormal, beaded, irregular ducts at baseline, making superimposed cholangiocarcinoma — which arises in 10–15% of PSC patients over a lifetime — exceptionally difficult to distinguish from a background of benign fibrotic change.
Because imaging morphology cannot reliably separate malignant from benign strictures, professional society guidelines uniformly recommend tissue or cytologic sampling at the index ERCP whenever a stricture of uncertain etiology is encountered — rather than proceeding directly to resection on imaging alone, or dismissing a suspicious stricture without sampling. The stakes are asymmetric: a missed cancer allows disease to progress toward unresectability, while an unnecessary hepatectomy or Whipple procedure for benign disease carries a 30-day major morbidity rate in the 30–40% range and a mortality risk in specialized centers of roughly 1–5%.
Up to 15–20% of biliary strictures resected for suspected malignancy based on imaging alone turn out to be entirely benign on final surgical pathology. That statistic alone justifies the entire brush cytology and adjunct-sampling pathway described in the following stages — every percentage point of diagnostic sensitivity gained translates directly into avoided unnecessary major surgery or, conversely, earlier oncologic treatment.
Endoscopic retrograde brush cytology remains the first-line tissue sampling method for a newly identified biliary stricture, largely because it can be performed during the same ERCP session used to establish drainage. A cytology brush is threaded over a guidewire that has already been passed across the stricture, then mechanically dragged back and forth against the stenotic segment to shear off epithelial and, if present, malignant cells for cytologic analysis.
After the stricture is traversed with a guidewire under fluoroscopic guidance, a dedicated cytology brush catheter is advanced over the wire until its bristled tip straddles the stenotic segment. The endoscopist then performs repeated to-and-fro passes — advancing and withdrawing the brush across the full length of the stricture with firm contact against the ductal wall — with most protocols recommending a minimum of 10 back-and-forth strokes to maximize cellular yield, since cellularity correlates with the number of passes up to a point of diminishing returns.
Once brushing is complete, the brush is retracted into its protective sheath or catheter before the entire assembly is withdrawn from the working channel — a maneuver intended to minimize the theoretical risk of tumor cell seeding along the endoscope or biliary tree. The bristles are then rolled directly onto glass slides for conventional smear cytology, or agitated into a liquid-based cytology (LBC) fixative vial, depending on institutional preference; LBC preparations tend to reduce obscuring blood and mucus and allow residual fluid to be used for ancillary testing such as FISH.
Cellular yield rises with additional brushing passes because each stroke shears a fresh layer of superficial epithelium and any exposed tumor cells into the bristles. However, the relationship is not linear: after roughly 8–12 passes, the readily exfoliable superficial cell population is largely exhausted, and additional strokes mainly re-sample the same denuded surface, so cellularity and downstream diagnostic sensitivity plateau rather than continuing to climb. This ceiling is a direct consequence of the underlying tumor biology described in Stage 4 — brush cytology can only ever sample what the tumor surface sheds, and densely desmoplastic cholangiocarcinomas and pancreatic cancers shed comparatively few cells no matter how vigorously the stricture is brushed.
Brush cytology adds minimal incremental risk to an ERCP already being performed for stricture evaluation and biliary drainage, though the underlying procedure carries its own risks of post-ERCP pancreatitis (roughly 3–10%), cholangitis, bleeding, and perforation. Specimen quality can be degraded by prior biliary stenting (reactive epithelial changes from a pre-existing stent are a major confounder discussed in Stage 4), by blood or bile obscuring the smear, by an inadequately traversed stricture that the brush cannot fully cross, and by tortuous or very tight strictures that limit the number of effective passes achievable before the wire or brush must be repositioned.
Because brush cytology alone under-detects malignancy, the same ERCP session is increasingly used to acquire adjunct specimens that interrogate the stricture through entirely different biological and visual lenses: fluorescence in situ hybridization for chromosomal aneuploidy, direct cholangioscopic visualization with targeted biopsy, and, where cholangioscopy is unavailable, blind fluoroscopy-guided intraductal forceps biopsy.
Fluorescence in situ hybridization applies a panel of fluorescently labeled DNA probes — originally developed as the UroVysion assay for urothelial carcinoma and subsequently adapted to biliary cytology — to the same brush cytology specimen. The panel typically targets the centromeric regions of chromosomes 3, 7, and 17, along with the 9p21 locus harboring the CDKN2A/p16 tumor suppressor gene. Malignant biliary epithelial cells frequently display aneuploidy: gains ("polysomy") of two or more of the chromosome 3/7/17 probes in a subset of cells, and/or homozygous loss of the 9p21 signal, patterns rarely seen in reactive or benign epithelium.
FISH is particularly valuable as an adjunct rather than a standalone test — it is run on residual liquid-based cytology material from the same brushing, requiring no additional sampling pass — and it materially increases sensitivity for malignancy detection when added to routine cytology, especially in the historically difficult PSC population, where chronic inflammatory atypia otherwise confounds conventional cytologic interpretation.
In PSC patients, where a stricture may reflect either fibro-inflammatory disease or superimposed cholangiocarcinoma, conventional brush cytology is especially insensitive because chronic inflammation itself produces reactive atypia that mimics malignancy. Adding FISH polysomy analysis meaningfully improves detection of true malignant transformation in exactly this population, where the diagnostic stakes of a missed cancer are highest.
Single-operator cholangioscopy systems (such as the SpyGlass DS platform) pass a dedicated miniature scope with its own optical and working channels directly into the bile duct alongside or in place of the standard duodenoscope-based approach, allowing the endoscopist to directly visualize the mucosal surface of the stricture rather than inferring its character from fluoroscopic silhouette alone. Visual criteria suggestive of malignancy include irregular, friable, or nodular mucosa, dilated and tortuous surface ("tumor") vessels, and infiltrative mass lesions — features entirely invisible on standard cholangiography.
Crucially, cholangioscopy allows targeted forceps biopsy under direct vision: rather than blindly sampling wherever the guidewire happens to cross the stricture, the endoscopist can direct the biopsy forceps precisely at the most visually suspicious mucosal focus. This targeted approach measurably improves diagnostic yield over blind brushing or blind biopsy alone, and is of particular value for indeterminate hilar (perihilar) strictures, where the complex ductal anatomy and eccentric tumor growth patterns make blind sampling especially unreliable.
Where cholangioscopy is unavailable, small-caliber biopsy forceps can be advanced over a guidewire and passed across the stricture under fluoroscopic guidance alone, obtaining a histologic core rather than a cytologic smear. Blind forceps biopsy adds incremental sensitivity to brushing when combined, but because the biopsy site cannot be visually confirmed, yield is lower than that of visually targeted cholangioscopic biopsy, and multiple bites (typically several per procedure) are generally taken to offset this blind-sampling limitation.
Despite decades of technical refinement, brush cytology alone remains a notoriously insensitive test for biliary malignancy. Pooled sensitivity across large series clusters around 40–60%, with a commonly cited figure near 45%, even though specificity remains excellent at roughly 95–100% — when cytology does call a specimen malignant, that call is almost always correct; the problem is how often a true cancer is missed rather than misclassified.
The central biological obstacle is the growth pattern of most biliary and pancreatic malignancies: cholangiocarcinoma and pancreatic ductal adenocarcinoma both provoke an intense desmoplastic (fibrotic) stromal reaction, in which malignant glandular cells are embedded within dense collagenous tissue rather than proliferating as a loose, exfoliation-prone mass. A brush dragged across the ductal surface can only recover cells that are already loosely attached to the epithelial surface — and in a densely desmoplastic tumor, relatively few malignant cells achieve that superficial, exfoliable position. The result is a specimen that may be markedly hypocellular for malignant cells even when the underlying tumor burden is substantial.
Additional contributors compound this core limitation: strictures that grow in a submucosal or periductal-infiltrating pattern (rather than an exophytic, luminal mass) present even less exfoliable surface to the brush; sampling error occurs when the brush fails to fully traverse a tight or tortuous stricture; and prior biliary stenting — extremely common in this patient population, since drainage is usually established at the same or a prior ERCP — induces reactive epithelial atypia from chronic mechanical irritation and inflammation that closely mimics the cytologic appearance of true malignant atypia.
Biliary cytology results are conventionally reported in one of four tiers, mirroring the structure used across most exfoliative and fine-needle cytology systems:
• Benign/negative — no atypical cells identified; reassuring but does not fully exclude malignancy given the test's limited sensitivity • Atypical (indeterminate) — cytologic changes beyond what is expected for reactive/reparative change, but insufficient in degree or quantity to call suspicious or malignant • Suspicious for malignancy — cellular features strongly favor malignancy but fall just short of the strict criteria for a definitive positive call • Positive/malignant — unequivocal cytologic features of malignancy present
In practice, "atypical" and "suspicious" results are common and clinically vexing — they neither confirm nor exclude cancer, and in a population where imaging already raised concern for malignancy, an indeterminate cytology result often changes management little on its own, driving the push toward adjunct FISH and cholangioscopic biopsy described in Stage 3.
The single hardest interpretive problem in biliary cytology is distinguishing reactive atypia — driven by chronic inflammation, indwelling stents, prior instrumentation, or biliary stasis — from genuinely malignant atypia. Both can produce nuclear enlargement, irregular nuclear contours, and hyperchromasia; cytopathologists must weigh the degree, uniformity, and combination of these features (nuclear-to-cytoplasmic ratio, chromatin texture, nucleolar prominence, loss of normal cellular architecture and polarity) rather than relying on any single criterion.
A patient with a long-indwelling biliary stent placed for presumed benign disease can develop florid reactive epithelial atypia that closely resembles malignancy on a brush specimen — a major source of false-positive "suspicious" calls. Conversely, a well-differentiated cholangiocarcinoma can shed cells that look deceptively bland, contributing to false-negative "benign" calls. This dual pitfall is precisely why cytology is interpreted in the context of the full clinical and imaging picture, and why adjunct FISH and targeted biopsy add independent, complementary information rather than simply repeating the same test.
No single sampling technique is sensitive enough to confidently exclude malignancy on its own. The clinical solution is additive: combining brush cytology, FISH polysomy analysis, and cholangioscopy-guided (or blind) biopsy exploits the fact that each modality fails to detect a partially different, only partially overlapping subset of cancers — so their sensitivities compound rather than merely duplicate one another.
Brush cytology detects malignant cells that are cytologically overt and superficially exfoliable; FISH detects chromosomal aneuploidy even in cells that look bland or scant on routine cytology, catching a portion of the "cytology-negative but genuinely malignant" cases; cholangioscopy-guided biopsy recovers histologic tissue from visually targeted, often submucosal or deeper foci that a passive brush never contacts. Because these three techniques interrogate different biological signals — morphology, genomic copy number, and direct visual/architectural assessment — a cancer missed by one modality is frequently caught by another, and the union of positive results across all three modalities yields a substantially higher overall sensitivity than any single test, commonly cited as exceeding 80% in combined-modality series (with a reported range of roughly 70–90% depending on stricture location, operator experience, and which specific combination of adjuncts is used), while specificity remains high because each individual component is itself highly specific.
A positive result on any component of the combined workup — malignant or strongly suspicious cytology, FISH polysomy, or a cholangioscopically biopsy-confirmed tumor — is generally sufficient to proceed toward oncologic staging (cross-sectional imaging, staging laparoscopy where indicated) and definitive resection planning, since specificity across these modalities is high and false positives are uncommon.
A negative or indeterminate combined workup is interpreted in context: in a patient with low pretest clinical suspicion (no discrete mass, stable imaging, no red-flag features such as weight loss or rising bilirubin trajectory), a thoroughly negative combined evaluation can support a strategy of close surveillance — repeat imaging and, if the stricture persists or changes, repeat ERCP with re-sampling — rather than proceeding directly to major surgery. In a patient with high pretest suspicion (a discrete mass, vascular encasement, or a PSC patient with a rapidly evolving dominant stricture), a negative sampling result carries less reassurance given the known sensitivity ceiling of even the combined approach, and multidisciplinary discussion — sometimes including empiric resection — may still be warranted.
The evolution from single-pass brush cytology (~45% pooled sensitivity) to a combined brushing-plus-FISH-plus-cholangioscopy strategy (>80% sensitivity, specificity preserved near 95%+) represents one of the more concrete, guideline-relevant diagnostic advances in pancreaticobiliary endoscopy over the past two decades. It does not eliminate diagnostic uncertainty entirely — no achievable combination of endoscopic sampling reaches 100% sensitivity given the underlying tumor biology — but it substantially narrows the population of patients who either undergo unnecessary major resection for benign disease or experience diagnostic delay for a treatable cancer.
The practical message for the indeterminate biliary stricture: obtain adequate brushings (10+ passes), send adjunct FISH whenever available — especially in PSC — and pursue cholangioscopy-guided biopsy for indeterminate hilar strictures or when initial sampling is non-diagnostic, before committing to either major resection or long-term surveillance based on a single insensitive test.