Screen-detected adenocarcinomas were smaller, slower and more often upper-lobe or pleural-based

Design

Retrospective analysis of a randomised screening trial cohort

Population

213 histologically confirmed screen-detected lung cancers from NELSON

Primary outcome

Association of CT features with adenocarcinoma vs other subtypes

Effect

Pleural attachment OR 6.67 (1.20-33.33); model AUC 0.75, specificity 29%

Using histologically confirmed lung cancers from the NELSON low-dose CT screening trial, investigators related standardised imaging features to histological subtype (European Radiology, 7 October 2026). Of 213 cancers, 67.5% were adenocarcinomas.

Adenocarcinomas were more often early stage (86.1% vs 47.8%), smaller (14.6 vs 17.3 mm) and in the right upper lobe. In adjusted analysis, pleural attachment raised the odds of adenocarcinoma (OR 6.67, 95% CI 1.20-33.33), larger size lowered them, and lower and middle lobe lesions were less likely to be adenocarcinoma. Longer volume doubling time was associated with adenocarcinoma. The model had an AUC of 0.75 but specificity of only 29%.

This helps explain why some screen-detected nodules grow slowly yet prove malignant. It does not allow histology to be predicted from imaging for an individual patient.

• Do not dismiss a slow-growing, pleural-based upper-lobe nodule in a screening participant as benign on growth alone.
• Expect faster growth and larger size in lower-lobe non-adenocarcinomas.
• Continue to manage screen-detected nodules by validated size and growth criteria.
• Do not use these features to predict histology in place of tissue.

Why it matters

It challenges the reflex reassurance that a slowly growing nodule is probably benign.

Don't overread it

With 29% specificity, the model cannot reliably tell adenocarcinoma from other subtypes in an individual.

The statistics, in plain English

The wide confidence interval for pleural attachment (1.20 to 33.33) reflects small numbers: the direction is clear, the size is not.

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.