From the editor — March 2026
Editor’s note — This is the online edition of the March 2026 newsletter. Comments are open at the bottom of the page for any reflections, counter-evidence, or questions on the featured papers.
The second edition of EUS & ERCP Monthly covers March 2026. The standout paper comes from Gastrointestinal Endoscopy — a prospective cohort with over four years of follow-up on EUS-guided radiofrequency ablation for BD-IPMNs with worrisome features. For the first time we have durable long-term local control data that directly challenges the current guideline position. A large propensity-matched Indian study settles an important practical question about gastric variceal obliteration, and a Markov model in Endoscopy makes the economic case for EUS-CDS as primary palliation for malignant biliary obstruction — questioning whether ERCP-BD should remain the reflexive first choice. Across all searches, 120 papers were reviewed; five are featured, eleven are listed.
— Dr Venkat Mahesh, MRCP, FEBGH — Education & Media Lead, UKIEUS
Top 5 featured papers — March 2026
#1 — EUS-RFA for branch-duct IPMNs with worrisome features: 98% local control at 4.1-year follow-up
Score 80/100 · Prospective cohort, 4.1yr follow-up
Barras J, Barthet M et al. 50 patients, 62 procedures. Gastrointestinal Endoscopy 2026. DOI: 10.1016/j.gie.2026.03.025 · PMID 41903818
Current guideline position. The Fukuoka/IAP guidelines classify EUS-guided ablation for BD-IPMNs as investigational and state it cannot be recommended at this time, citing theoretical risk of ablative agents entering the main pancreatic duct. The Barras 2026 paper provides the evidence that did not yet exist to update this position.
Fifty inoperable or surgery-declining patients (62 procedures, 58 lesions) were treated at a single tertiary centre in Marseille between 2015 and 2024, with mean follow-up of 4.1 years after RFA — 9 years from IPMN diagnosis. Technical success was 100%. Adverse events occurred in 27% of procedures, predominantly mild abdominal pain; post-procedural pancreatitis in three patients including one necrotising case; no procedure-related deaths. Local control was achieved in 98% of lesions: 17% showed complete radiological disappearance, 86% decreased in size. No treated lesion progressed to cancer. Four patients (8%) developed pancreatic cancer remote from the treated cyst — underlining that EUS-RFA addresses the focal lesion but not the pancreas-wide field-defect risk intrinsic to IPMN.
Contextualising evidence. Barthet 2019 prospective multicentre (PMID 30669161) was the first prospective multicentre EUS-RFA cohort — 17 PCNs (16 IPMNs), 71% significant response at 12 months, all 12 mural nodules resolved, AEs 10% falling to 3.5% after protocol modification. Cho 2024 ablation vs surgery propensity-matched n=310 in both groups (PMID 38588765): EUS ablation 10-year long-term morbidity 1.6% vs 33.5% surgery; new-onset diabetes 2.2% vs 22.8%; surgery complete resolution 100% vs 76.5% with relapse 0.3% vs 4.6%. OPTIMAL-IPMN 2026 multinational n=1728 resections (PMID 41874453): excellent post-resection 5-year survival, but 63% of resected specimens showed only low-grade dysplasia — the concern regarding overtreatment gives ablation a stronger argument.
Clinical bottom line. EUS-RFA provides durable local control in BD-IPMNs with worrisome features in patients unsuitable for or declining surgery, with the longest follow-up data published. This evidence now belongs in the informed consent conversation alongside surgical options — explicitly framed as investigational data challenging the current guideline position, including the 8% residual risk of remote pancreatic cancer that ablation cannot address.
#2 — EUS-guided coil and glue embolisation superior to endoscopic glue injection for gastric variceal obliteration
Score 76/100 · Propensity-matched cohort · Gastric varices
Puri R, Giri S et al. 442 patients, propensity-matched 71/group. Clinical Endoscopy 2026. DOI: 10.5946/ce.2025.344 · PMID 41956728
After propensity score matching (1:1), 4-week variceal obliteration was significantly higher with EUS-C+G: 95.8% vs 64.8% (p<0.001), using lower glue volume (1.5 vs 2.5 mL). Adverse events were more frequent with EUS-C+G (25.4% vs 8.5%, p=0.007), though all were mild and managed conservatively. Subgroup analyses showed superiority in isolated fundic varices (93.1% vs 50.0%) and secondary prophylaxis (95.0% vs 53.3%). Portal vein thrombosis and varix size ≥18 mm independently predicted incomplete obliteration with standard glue injection on multivariate analysis — both are identifiable pre-procedure and should inform the choice of technique.
Supporting evidence base. Samanta 2023 international multicentre propensity-matched (PMID 37269164, Liver Int) — 58 EUS-C+G vs 118 E-CYA: rebleeding 13.8% vs 39.1% (p<0.0001), reintervention 12.1% vs 50.4% (p<0.001). Jhajharia 2025 RCT (PMID 39293480, Endoscopy), the only published RCT, n=50: reinterventions 20.8% vs 53.8% (p=0.03), survival longer with EUS-C+G (p=0.04). Florencio de Mesquita meta-analysis (PMID 39389435, GIE 2025), 6 studies / 445 patients: lower rebleeding OR 0.22, lower reintervention OR 0.29, no difference in pulmonary embolism or mortality. Mohammadpour meta-analysis (PMID 41360391, J Gastroenterol Hepatol 2025), 9 studies / 579 patients: lower reintervention RR 0.32, higher obliteration RR 1.18, lower adverse events RR 0.55, lower rebleeding RR 0.36.
Guideline note. AGA 2023 recommends EUS-guided coil+glue as the preferred EUS strategy. AASLD 2024 lists ECI, TIPS, or retrograde obliteration as first-line — EUS-C+G is not yet the default over conventional ECI in international guidelines despite the consistent evidence base.
Clinical bottom line. EUS-C+G achieves substantially higher obliteration and fewer reinterventions than standard glue injection for gastric varices. Evidence is now consistent across one RCT and two 2025 meta-analyses. Portal vein thrombosis and varix ≥18 mm predict standard glue failure — identify these before deciding on approach.
#3 — EUS-CDS with LAMS cost-effective as primary palliation for malignant distal biliary obstruction
Score 75/100 · Markov modelling study · ERCP / Biliary drainage
Magahis PT, Sharaiha RZ et al. Markov model. Endoscopy 2026. DOI: 10.1055/a-2837-1405 · PMID 41887604
This is potentially another ERCP practice-change paper. The central question is whether ERCP-guided transpapillary biliary drainage should remain the reflexive first choice for malignant distal biliary obstruction. A state-transition Markov model comparing EUS-CDS with LAMS versus ERCP-BD with SEMS in a 70-year-old with locally advanced pancreatic cancer and CBD dilation >15 mm yields an ICER of $47,711/QALY — well below the $100,000/QALY threshold. EUS-CDS remains cost-effective unless ERCP-BD technical success exceeds 91%, reintervention falls below 11%, or post-procedural pancreatitis falls below 4% — benchmarks rarely achieved in real-world practice. In 74% of probabilistic sensitivity iterations, EUS-CDS was cost-effective.
Note: Modelling study — not a clinical trial. US cost structure does not translate directly to NHS. Directional argument supported by ASGE 2024 conditional recommendation for EUS-BD. Clinical survival data (Ciesielski, Cancers 2026; de Jong, Endoscopy 2025) were covered in the February 2026 edition of this series.
Clinical bottom line. In patients with MDBO and CBD dilation >15 mm, EUS-CDS with LAMS may be economically as well as clinically justified as a primary approach — not just rescue after failed ERCP. The case for EUS-CDS as first-line should be part of MDBO MDT discussions at units with established EUS-BD expertise, weighed against local audited ERCP outcomes for MDBO. Worth considering as a prospective UK-based study.
#4 — 22G Franseen needle outperforms 20G forward bevel for EUS-FNB of subepithelial lesions
Score 74/100 · Propensity-matched cohort · FNB needle comparison
Gadour E, Facciorusso A et al. 145 vs 145 propensity-matched. Digestive Diseases and Sciences 2026. DOI: 10.1007/s10620-026-09827-4 · PMID 41843039
Sample adequacy (98.6% vs 89.6%, p=0.01) and diagnostic accuracy (96.5% vs 88.9%, p=0.01) were significantly higher with the 22G Franseen needle. Specimen integrity scores were also significantly higher. Diagnostic sensitivity favoured Franseen (97.9% vs 88.5%, p=0.02); specificity 100% with both. Consistent across subgroup analyses by location and diagnosis. No adverse events in either group.
Caveat: gauge differed alongside tip design (22G vs 20G), making it difficult to isolate needle geometry as the sole variable.
Guideline alignment. ESGE 2025 Technical Review recommends end-cutting FNB needles (Franseen design) over reverse-bevel or FNA for solid pancreatic lesions; for SELs ≥20 mm, EUS-FNB or mucosal incision-assisted biopsy are equally recommended. ACG 2023 guideline on SELs recommends FNB alone or FNA with ROSE for SEL sampling — FNB increases accuracy (90% vs 52%) and reduces unnecessary surgical resections vs FNA without ROSE.
Clinical bottom line. For subepithelial lesion sampling, the 22G Franseen needle provides meaningfully higher diagnostic accuracy and specimen integrity than the 20G forward bevel. Franseen design should be the default for SEL tissue acquisition where available. This study does not address the role of tissue processing post sampling or the cytopathologist role — see our EUS FNAB abstracts at BSG Live 2026.
#5 — EUS-GBD with LAMS: 100% technical and 94.6% clinical success at one year in frail non-surgical candidates
Score 70/100 · Prospective cohort, 366-day follow-up
Fox T, Weilert F et al. 37 patients, median follow-up 366 days. ANZ Journal of Surgery 2026. DOI: 10.1111/ans.70629 · PMID 41860060
37 consecutive patients (mean age 74.9 years; Charlson 5.7, CFS 4.2, ASA 3) unsuitable for surgery. EUS-GBD technically successful in all; clinical success 94.6%. Two reinterventions (5.4%) for LAMS occlusion. Three deaths: two from sepsis despite intervention, one unrelated. No other adverse events at median 366-day follow-up. The 1-year follow-up duration distinguishes this from most prior cohorts reporting only 30- or 90-day outcomes.
Contextualising evidence. DRAC 1 RCT — Teoh AYB et al., Gut 2020 (PMID 32165407) — the only published RCT, 80 patients: EUS-GBD 1-year AEs 25.6% vs 77.5% (p<0.001), recurrent cholecystitis 2.6% vs 20% (p=0.029), reinterventions 2.6% vs 30% (p=0.001), readmissions 15.4% vs 50% (p=0.002); clinical success and 30-day mortality equivalent. Baron TH et al. expert panel, Ann Surg 2025 (PMID 40255177) — multidisciplinary consensus: EUS-GBD with LAMS recommended for never-surgical candidates if eligible for monitored anaesthesia or GA, institutional expertise available, minimal intervening ascites. ASGE 2024 guideline (PMID 39078360): conditional recommendation for EUS-GBD; lower recurrent cholecystitis OR 0.11, lower reintervention OR 0.14 vs PT-GBD.
Clinical bottom line. EUS-GBD with LAMS is highly effective and safe at one year in frail, genuinely high-risk surgical patients. At units with EUS-GBD expertise, this supports positioning it as the preferred drainage approach in never-surgical candidates — with explicit MDT discussion of local expertise and patient fitness for anaesthesia. In the UK setting, availability of an EUSist with appropriate expertise and ability to accommodate the patient urgently/emergently must be weighed against an established radiology team with quick percutaneous drainage capability.
Also published — March 2026
1. [ERCP] Fully covered metal stents as risk factor for acute cholecystitis after ERCP — shorter stent length is the key modifiable predictor — multicentre 365 patients. AC in 8.8%. Shorter stents 40–60 mm: 10.4% cholecystitis vs 1.5% for longer stents (p=0.02). Gallbladder stones (OR 4.2) and narrower CBD diameter independently predicted cholecystitis. Favour longer stent selection in patients with gallbladder stones and narrow CBD. Eur J Gastroenterol Hepatol 2026. DOI: 10.1097/MEG.0000000000003192 · PMID 41925049
2. [ERCP] Emergency EUS-guided ERCP for common bile duct stones: shorter stay, lower costs, fewer short-term complications versus elective approach — 215 patients. Both groups: 100% EUS technical success, comparable ERCP clinical success (96.3% vs 94.8%). Emergency pathway significantly reduced hospitalisation and costs with fewer post-ERCP complications. Requires same-session EUS capability. Dig Dis Sci 2026. DOI: 10.1007/s10620-026-09852-3 · PMID 41896483
3. [ERCP] Difficult biliary cannulation in compensated liver cirrhosis: 45.5% incidence, 12.6% complication rate — prospective predictors study of 127 patients. DBC in 45.5% at first ERCP. Complications 12.6% (8.7% pancreatitis). Independent predictors: type 2/3 papillae, duodenal diverticulum, precut use, older age. All four are identifiable pre-procedure — a practical pre-ERCP risk checklist for cirrhotic patients. Sci Rep 2026. DOI: 10.1038/s41598-026-41040-1 · PMID 41857076
4. [ERCP training] Integrating ERCP within UK higher surgical training: structured JAG/JETS model — 365 procedures over 33 months, >90% native papilla cannulation rates. UK authors: Latif J, Bhatti I, Dennison A (University Hospitals of Leicester and Derby). Structured ERCP training embedded within HPB higher surgical training, evaluated using JAG/JETS e-portfolio data. Cannulation exceeded 90% early in the structured pathway. Transferable model for units seeking to integrate ERCP competence within surgical rotations. Surg Laparosc Endosc Percutan Tech 2026. DOI: 10.1097/SLE.0000000000001459 · PMID 41919796
5. [ERCP] PEP prevention adherence across 26 ERCP centres in Slovakia and Czechia: significant gaps between guideline and practice. Only 53.9% give rectal NSAIDs universally; 26.9% use aggressive hydration per protocol; 46.2% place prophylactic pancreatic stents during difficult cannulation; 53.8% track PEP rates. Gaps likely representative of wider European practice including the UK. J Gastrointest Liver Dis 2026. DOI: 10.15403/jgld-6680 · PMID 41894708
6. [EUS diagnostic] FNB significantly outperforms FNA for pancreatic serous cystic neoplasm diagnosis: OR 4.96 in 77 patients. Diagnostic yield: FNB 44.68% vs FNA 14.00% (p<0.01). Franseen needle, cyst size ≥17 mm, and higher pass count were independent predictors of success. J Clin Med 2026. DOI: 10.3390/jcm15062438 · PMID 41899360
7. [Review] Endoscopic management of biliary stricture in primary sclerosing cholangitis: comprehensive review covering cholangioscopy, CE-EUS, NGS, balloon dilation, and complication prevention. Covers ERCP role in PSC, limitations of brush cytology and FISH, advanced approaches (cholangioscopy, CE-EUS, confocal endomicroscopy), NGS and methylation markers for CCA, balloon dilation as first-line, and best-practice prophylaxis. Liver Transpl 2026. DOI: 10.1097/LVT.0000000000000866 · PMID 41879306
8. [Cholangioscopy] Cholangioscopy-guided laser lithotripsy for forgotten biliary stents with large stentoliths: 100% success in 8 consecutive patients — with video. Stents in place for median 9 years (range 5–20); stentoliths up to 42 mm. Single-session clearance in 87.5%. One mild post-ERCP pancreatitis; no perforations or mortality. Dig Dis Sci 2026. DOI: 10.1007/s10620-026-09848-z · PMID 41863757
9. [AI / Radiomics] PNET-PRISM: multicentre-validated radiomics nomogram for non-invasive PanNET grading — AUC 0.87, 100% sensitivity, rescued 52% of non-diagnostic EUS-FNA. CT radiomics across 407 patients, three cohorts. M-DLR score stratified progression-free survival (HR 2.05). Decision-support adjunct when biopsy is unavailable or non-diagnostic. Insights Imaging 2026. DOI: 10.1186/s13244-026-02250-3 · PMID 41874882
10. [EUS-TA / Genomics] Tumour cell count thresholds for EUS-TA specimens in pancreatic cancer comprehensive genomic profiling — National Cancer Center Tokyo, 114 patients. Thresholds of ≥2300 tumour cells (PPV 92.9%) and ≥6000 total cells (PPV 86.7%) identified for adequate DNA yield for CGP. Sensitivity ~50%. Practical guidance for centres establishing CGP pathways. J Hepatobiliary Pancreat Sci 2026. DOI: 10.1002/jhbp.70106 · PMID 41917794
11. [ERCP] Antifungal treatment for Candida in bile cultures during ERCP: not associated with improved survival or reduced invasive candidiasis — 197 inpatients. No significant difference in death or invasive candidiasis at one year (45.1% vs 45.9%). Supports a conservative stewardship-oriented approach: Candida in bile cultures does not warrant routine antifungal treatment in the absence of invasive disease. J Fungi (Basel) 2026. DOI: 10.3390/jof12030208 · PMID 41893140
Discussion thread — March 2026 issue. Comments are open below. The Barras EUS-RFA cohort and the Magahis cost-effectiveness model are both genuinely practice-questioning papers — would value the UK perspective on either, particularly from units already running pancreatic ablation or first-line EUS-CDS pathways.
— VM
© UKIEUS 2026. Editorial commentary by Dr Venkat Mahesh, MRCP, FEBGH. Paper titles and abstracts cited from PubMed with DOI attribution; no copyrighted full text reproduced. Primary literature identified via PubMed systematic search; broader clinical context verified through structured evidence synthesis; all references independently confirmed via PubMed prior to inclusion. For educational use only. www.ukieus.co.uk
Comments are open below. First-time commenters are moderated before publication; subsequent comments from the same email publish automatically. Name and email are required (pseudonymous handles are welcome), email is never displayed publicly, and the UKIEUS Privacy Notice explains how your data is handled. By posting a comment you confirm you are willing to be contacted by email about replies to your comment. Patient-identifiable information must not be posted; please anonymise all clinical details in line with GMC guidance.
