IPMN on MRI: current high-risk stigmata often over-called malignancy in one series

Design

Single-centre retrospective diagnostic accuracy study

Population

418 patients with surgically confirmed IPMN, 173 malignant

Primary outcome

Accuracy of MRI high-risk stigmata for malignancy

Effect

Original criteria sensitivity 82.7%, specificity 66.9%; revised scale AUC 0.77 to about 0.80

This single-centre retrospective study reviewed preoperative MRI in 418 patients with surgically confirmed IPMN between 2011 and 2023, 173 of them malignant. It tested the international high-risk stigmata and then searched for better cut-offs for enhancing mural nodule size and main duct diameter.

The original high-risk stigmata had sensitivity of 82.7% and specificity of 66.9%. Obstructive jaundice and suspicious cytology were highly specific (about 99%), but an enhancing nodule of 5 mm or more and a main duct of 10 mm or more were much less so (80%). Raising thresholds improved specificity: at cut-offs of 12 mm for both nodule and duct, optimism-corrected sensitivity was 65% and specificity 82%. A six-point scale built on revised cut-offs improved AUC from 0.77 to about 0.80.

For reporting radiologists, the message is that a small enhancing nodule or a moderately dilated duct carries a real false-positive rate, and measurements near the thresholds matter. Raising cut-offs trades missed cancers for fewer unnecessary pancreatic resections, a balance that depends on patient age and fitness. The thresholds come from one centre without external validation, so current guideline criteria still apply.

• Measure and report enhancing mural nodule size and main pancreatic duct diameter precisely in every IPMN report.
• Flag obstructive jaundice and positive cytology as the most specific high-risk features.
• Note in the report when a nodule or duct is only just above the high-risk threshold.
• Discuss borderline high-risk IPMN in a multidisciplinary meeting, weighing patient age and fitness.
• Continue to apply current international guideline thresholds until revised cut-offs are validated.

Why it matters

A high-risk label on MRI often leads to pancreatic resection, an operation with real morbidity, so false positives carry a heavy cost.

Don't overread it

Single-centre derived thresholds without external validation should not replace guideline criteria.

The statistics, in plain English

Specificity of 67% means about one in three patients flagged as high risk had benign disease at surgery. Raising the cut-offs lifts specificity but lowers sensitivity, so more cancers would be missed. The cut-offs were derived and tested in the same centre, which tends to overstate how well they work elsewhere.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.