- Design
- Systematic review and bivariate meta-analysis of direct comparative studies
- Population
- 40 studies of patients with suspected pancreatic ductal adenocarcinoma having two or more imaging tests
- Primary outcome
- Diagnostic accuracy for detection and staging
- Effect
- Liver metastases sensitivity MRI 0.89 vs CT 0.60; resectability specificity CT 0.88 vs MRI 0.63
A PROSPERO-registered meta-analysis (European Radiology, 1 October 2026) pooled 40 studies in which patients with suspected pancreatic ductal adenocarcinoma had at least two imaging methods, with pathology as the reference where available.
CT, MRI and endoscopic ultrasound did not differ significantly in detecting the primary tumour. For liver metastases, MRI had higher sensitivity (0.89, 95% CI 0.67-0.97) and specificity (0.91, 0.76-0.97) than CT (0.60, 0.44-0.74; and 0.84, 0.70-0.93). CT outperformed MRI for vascular invasion and had higher specificity for resectability (0.88 vs 0.63). Endoscopic ultrasound was more sensitive than CT or PET/CT for nodes.
The techniques are complementary. Pancreatic-protocol CT remains the backbone for local staging and resectability; adding liver MRI before a potentially curative resection may find small metastases that would make the operation futile.
- Keep pancreatic-protocol CT as the main tool for vascular involvement and resectability.
- Consider liver MRI before potentially curative pancreatic resection to look for small metastases.
- Use endoscopic ultrasound when nodal status or tissue diagnosis will change management.
- State in the report which modality each staging judgement rests on.
Why it matters
CT missed about 4 in 10 liver metastases in pooled data, and each missed one may mean a major operation that cannot help.
Don't overread it
Studies varied in protocols and reference standards, and the confidence intervals for MRI are wide.
The statistics, in plain English
Sensitivity of 0.60 means CT detected about 60 of every 100 liver metastases present; MRI detected about 89. The wide MRI interval (0.67 to 0.97) means its true sensitivity could be lower.
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.

Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for radiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free