Endoscopic diagnosis of pancreatic cancer
Article information
Abstract
Purpose
Pancreatic cancer remains a major clinical challenge due to its high mortality rate and limited treatment options. Timely diagnosis is crucial for improving survival; however, its aggressive nature and anatomical location often delay detection. In this context, advanced endoscopic techniques have emerged as essential tools for evaluating pancreatic cancer.
Current Concepts
Endoscopic ultrasound (EUS) plays a key role by providing high-resolution images and enabling tissue acquisition through fine-needle aspiration or biopsy—both safe and effective methods for obtaining diagnostic specimens from pancreatic lesions. Adjunctive techniques, such as contrast-enhanced EUS and elastography, further enhance evaluation by assessing vascularity and tissue stiffness. Additionally, endoscopic retrograde cholangiopancreatography (ERCP) holds a critical role, particularly in patients with obstructive jaundice. ERCP facilitates delineation of pancreatic and bile duct anatomy and enables cytologic and histologic sampling via brush cytology or forceps biopsy. Although technologies such as cholangioscopy and probe-based confocal laser endomicroscopy have been introduced to increase diagnostic yield, their routine clinical use remains limited. Furthermore, emerging applications of artificial intelligence (AI) in both EUS and ERCP are being integrated to support image interpretation and reduce operator dependency.
Discussion and Conclusion
The synergistic use of EUS and ERCP has advanced the diagnostic strategy for pancreatic cancer by improving early detection and guiding treatment strategies. Ongoing improvements in imaging techniques, molecular diagnostic methods, and AI tools are expected to further enhance diagnostic accuracy, facilitating the development of personalized therapeutic approaches, ultimately leading to better patient outcomes.
Introduction
1. Background
Pancreatic cancer remains a formidable clinical challenge globally, characterized by a high mortality rate and limited therapeutic options [1]. In Korea, the incidence continues to rise, underscoring a critical need for early detection amid persistently limited prognostic improvements [2]. Numerous studies have demonstrated that timely diagnosis substantially improves survival and guides therapeutic planning; nevertheless, early detection remains challenging due to the tumor’s infiltrative nature and the pancreas’s complex anatomy [3]. Pancreatic ductal adenocarcinoma (PDAC) constitutes more than 90% of all pancreatic malignancies [4]. This review focuses on the diagnostic evaluation of PDAC. Endoscopic ultrasound (EUS) provides high-resolution imaging from a close vantage point, offers lesion characterization using contrast-enhanced EUS (CE-EUS) and elastography, and enables definitive tissue sampling through EUS-guided fine-needle aspiration or biopsy (EUS-FNAB). Endoscopic retrograde cholangiopancreatography (ERCP) visualizes pancreatic and biliary ducts fluoroscopically following contrast injection and permits cytologic or histologic sampling at strictures, thus combining diagnostic and therapeutic capabilities.
2. Objectives
This review aims to evaluate the diagnostic accuracy, recent technological advances, and guideline-based roles of EUS and ERCP; highlight the clinical value of EUS-FNAB; discuss the strengths and limitations of ERCP-based ductal evaluations; and explore future directions involving artificial intelligence (AI) and molecular diagnostics.
Endoscopic Ultrasound
1. Overview
EUS combines an endoscope with a high-frequency transducer, enabling precise imaging of the gastrointestinal wall, pancreas, and adjacent vasculature [5]. By positioning the probe in the stomach or duodenum, EUS circumvents gaseous interference that typically limits transabdominal ultrasound. Radial scanners provide comprehensive 360° anatomical surveys, whereas linear probes enable real-time tissue acquisition (Figure 1A, 1B). After fasting and sedation, a systematic evaluation is conducted: the pancreatic head is examined from the duodenal bulb and descending limb, whereas the body and tail are assessed via the stomach. FNAB is performed under direct ultrasound guidance when malignancy is suspected. Meta-analyses report sensitivities and specificities of 90% to 100% for EUS in detecting pancreatic malignancy, demonstrating superior sensitivity for lesions ≤2 cm compared with computed tomography (CT) or magnetic resonance imaging [6,7]. EUS also outperforms CT in assessing vascular invasion (sensitivity 86% vs. 58%) and nodal metastasis (58% vs. 24%), although evaluation of distant disease still requires complementary imaging modalities [8].
Linear array echoendoscope, radial array echoendoscope, and a fine-needle biopsy sample. All figures are provided by the author. (A) Linear array echoendoscope with a sector-shaped ultrasound scanning field (orange cone). (B) Radial array echoendoscope with a 360-degree ultrasound scanning field (orange disc). (C) Fine-needle biopsy sample obtained from the pancreas. The whitish core material (arrow) is visible within the collection tube.
2. Tissue diagnosis with EUS-FNAB
Histologic confirmation is frequently mandatory before initiating treatment. Introduced in 1991, EUS-FNAB remains the primary modality for sampling pancreatic masses [5]. Fine-needle aspiration (FNA) retrieves cytologic material, whereas fine-needle biopsy (FNB) utilizes core-cutting designs (e.g., Franseen, fork-tip) to preserve tissue architecture. Recent studies demonstrate higher sample adequacy and diagnostic accuracy for FNB, especially in cases of suspected malignancy [9,10]. Using 19-, 22-, or 25-gauge needles, the operator advances the needle under Doppler guidance, employing suction or capillary techniques, and deposits the specimen onto slides or into formalin (Figure 1C). Rapid on-site evaluation by a cytopathologist can improve diagnostic yield, though its availability remains limited domestically. Reported sensitivities for diagnosing solid pancreatic tumors range from 84% to 87% (specificity≈98%), achieving diagnostic accuracy of approximately 91%, even for lesions ≤10 mm [10–12]. Nodal FNAB achieves approximately 87% accuracy [13]. Additionally, specimens are increasingly employed for next-generation sequencing (NGS), with FNB achieving higher NGS adequacy rates (70%–91%) than FNA [14]. Overall complication rates remain low (0%–2.5%), with bleeding, perforation, infection, and rare needle-tract seeding (≈0.4%) reported [15–17].
3. Advances in EUS imaging
CE-EUS evaluates microvascularity using contrast agents such as SonoVue or Sonazoid, distinguishing hypoenhancing PDAC from hyperenhancing neuroendocrine tumors and heterogeneous inflammatory masses. Its accuracy for differentiating mucus plugs from mural nodules within cystic lesions approaches 94% (Figure 2) [18,19]. Strain elastography visually encodes tissue stiffness via color (blue=hard, red=soft). A meta-analysis showed excellent sensitivity (98%) but moderate specificity (63%) since chronic pancreatitis also increases tissue stiffness [20]. Shear-wave elastography potentially provides quantitative refinement [21].
Endoscopic ultrasound (EUS) images of a pancreatic head cyst. All photos are provided by the author. (A) Grayscale EUS image demonstrating a mass-forming lesion within the cystic lesion. (B) Contrast-enhanced EUS image showing non-enhancement of the lesion, suggestive of mucinous material rather than a solid tumor.
4. Artificial intelligence integration
Deep-learning models, particularly convolutional neural networks, have achieved sensitivities of 83% to 94% and area under the curve values around 0.93 for classifying pancreatic lesions on EUS images [22–24]. AI-assisted digital pathology of EUS-FNAB whole-slide images facilitates rapid assessment of specimen adequacy and molecular profiling [25]. Nevertheless, substantial challenges remain, including the need for large multicenter datasets, algorithm interpretability, regulatory approval, and reimbursement pathways [26].
Endoscopic Retrograde Cholangiopancreatography
1. Overview
Although magnetic resonance cholangiopancreatography and EUS have largely supplanted diagnostic ERCP, the procedure remains essential for simultaneous biliary drainage and tissue acquisition, especially in PDAC of the pancreatic head causing distal bile-duct obstruction [27,28]. However, post-ERCP pancreatitis (3%–10%) limits its purely diagnostic application, making ERCP reserved primarily for therapeutic indications or inconclusive cases following other diagnostic methods.
2. Tissue acquisition during ERCP
Brush cytology and forceps biopsy at ductal strictures yield sensitivities of approximately 45% and 48%, respectively, increasing to about 59% when combined; specificity remains high at around 99% [29]. Following cannulation and cholangiography, a guidewire traverses the stricture, enabling brushing and biopsy sampling, followed by placement of a stent or nasobiliary drainage catheter (Figure 3). Fluorescence in situ hybridization (FISH) further increases sensitivity up to 85% but may slightly reduce specificity; polysomy demonstrates a positive predictive value exceeding 80% [30]. NGS analysis of bile or pancreatic juice, detecting mutations in genes such as KRAS, TP53, and SMAD4, increases sensitivity to 77% to 83% while preserving specificity around 99% [31].
Endoscopic view and fluoroscopic images of cannulation and sample acquisition (all photos are provided by the author). (A) Endoscopic view showing cannulation of the ampulla followed by endoscopic sphincterotomy. (B) Fluoroscopic image demonstrating brush cytology being performed; the brush is indicated by the arrow. (C) Fluoroscopic image demonstrating forceps biopsy; the forceps are indicated by the arrow.
3. Direct visual systems: cholangioscopy and pancreatoscopy
Single-operator digital cholangioscopes offer high-resolution visualization and targeted biopsy, significantly improving diagnostic sensitivity to approximately 88% and specificity to about 95% for indeterminate biliary strictures (Figure 4) [32–35]. Pancreatoscopy aids in mapping intraductal papillary mucinous neoplasms and subtle ductal lesions, although it presents greater technical difficulty and a higher risk of pancreatitis. Emerging probe-based confocal laser endomicroscopy and narrow-band imaging further enhance real-time microstructural assessment but face barriers related to cost and specialized training [36].
Magnetic resonance cholangiopancreatography and fluoroscopic and cholangioscopic images of the bile duct. All photos are provided by the author. (A) Magnetic resonance cholangiopancreatography image showing a non-dependent filling defect in the bile duct (arrow). (B) Fluoroscopic image demonstrating a non-mobile filling defect (arrows) within the bile duct. (C) Cholangioscopic image revealing an impacted stone that appeared immobile during an endoscopic evaluation.
4. Artificial intelligence in ERCP
Deep-learning analysis of cholangioscopy images effectively differentiates benign from malignant strictures with approximately 90% accuracy. Prospective real-time models have reported sensitivities around 95% and specificities about 89% [37,38]. Broader clinical integration awaits standardized datasets, external validation, and regulatory approval.
Conclusion
EUS and ERCP complement cross-sectional imaging by providing high-resolution visualization and definitive tissue diagnosis of pancreatic cancer. EUS excels in detecting sub-centimeter lesions, staging locoregional disease, and obtaining cytologic or histologic confirmation. CE-EUS and elastography enhance differential diagnosis by assessing vascularity and stiffness. Although ERCP’s diagnostic role has diminished, it remains invaluable for ductal sampling, cholangioscopy-guided biopsy, and therapeutic drainage. Advances in AI, high-definition endoscopy, and molecular diagnostics—including FISH and NGS—have significantly enhanced sensitivity, specificity, and predictive accuracy. The integration of these advanced technologies promises more precise, personalized management of pancreatic cancer, facilitating earlier detection and tailored therapeutic strategies.
Notes
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
Funding
None.