- Design
- Systematic review and diagnostic meta-analysis (20 studies; 12 pooled)
- Population
- Patients with histologically proven renal cell carcinoma who had pre-operative MRI
- Primary outcome
- Accuracy for WHO/ISUP high-grade disease
- Effect
- ADC sensitivity 0.84 (95% CI 0.77 to 0.89), specificity 0.57 (0.51 to 0.63), AUC 0.71; radiomics AUC 0.90
This meta-analysis looked at whether quantitative MRI can predict the WHO/ISUP nuclear grade of renal cell carcinoma before surgery. It included 20 studies, with pooled analysis possible in 12.
Seven studies using apparent diffusion coefficient (ADC) found a pooled sensitivity of 0.84 (95% CI 0.77 to 0.89) and specificity of 0.57 (95% CI 0.51 to 0.63) for high grade, with a summary AUC of 0.71. Low-grade tumours had higher ADC values than high-grade (mean difference 0.21 x 10^-3 mm²/s). Five radiomics or deep learning studies gave sensitivity 0.79 and specificity 0.86, AUC 0.90, but varied more between studies.
The authors say standardised protocols, external validation and decision-impact studies are needed before clinical use. For now, ADC supports a rough impression of grade and is not a substitute for histology.
- Report ADC for renal masses, but do not infer grade from it on its own.
- Expect many false positives for high grade with ADC (specificity 0.57).
- Treat radiomics results as research until validated outside the development centre.
- Keep biopsy or surgical histology as the grading reference.
Why it matters
Pre-operative grade could help triage biopsy and surveillance, but only if imaging can be trusted for it.
Don't overread it
Heterogeneous studies, small pooled sets for radiomics and no evidence yet that imaging grade changes management.
The statistics, in plain English
A specificity of 0.57 means that for roughly four in ten low-grade tumours, ADC would still flag high grade. AUC 0.90 for radiomics looks strong, but models tested on the data they were built on tend to overestimate, which is why external validation matters.
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