Creamer, Andrew William; (2023) Developing and validating risk prediction approaches for the management of screen-detected pulmonary nodules. Doctoral thesis (Ph.D), UCL (University College London).

Text Creamer_10174263_THESIS_post_viva-merged_ REDACTED.pdf Access restricted to UCL open access staff until 1 August 2025. Download (8MB)

Abstract

Pulmonary nodules are small focal opacities within the lung parenchyma. Whilst the majority represent a benign process, a small proportion represent early lung cancer. Whilst previously encountered as incidental findings on thoracic imaging, they are now a key feature of low-dose CTs (LDCT) performed for lung cancer screening. Lung cancer screening with LDCT has been demonstrated to reduce lung cancer-specific mortality by 20-24%. However, effective screening demands the ability to distinguish the small proportion of malignant nodules from the high background prevalence of benign nodules, whilst minimising both false negatives and unnecessary interventions for changes which ultimately prove benign. This thesis reports the development and validation of risk-prediction approaches to screen detected pulmonary nodules using data from the SUMMIT study. The first part of this thesis describes risk prediction approaches at baseline CT. Solid nodules <100mm3 conferred no increased risk of malignancy compared to participants with no nodules. Both a 5mm diameter and 80mm3 volume threshold achieved a negative predictive value of >99.5%, but the volumetric threshold was able to encompass a further 20% of participants, significantly and safely reducing the need for further surveillance. The second part investigates malignancy prediction in nodules at subsequent screening rounds. Volumetric, but not diameter stability, at 12 months was sufficient to exclude malignancy in solid nodules, and a volumetric approach to assess growth achieved better sensitivity, specificity, NPV and PPV than a diameter-based approach. New solid nodules 30-200mm3 appearing at annual CTs were more likely to represent malignancy than equivalently sized nodules at baseline CT, whereas new nodules appearing within the shorter timeframe of a 3-month interval CT had a <1% chance of malignancy. Finally, I analysed investigations performed following referral of high-risk nodules for multi-disciplinary assessment, and the rate of recurrence and metachronous lung cancer developing following radical treatment of screen-detected lung cancer. The crude rate of recurrence was 9.1 per 100 patients per year, and of metachronous lung cancer diagnosis 2.1 per 100 patients per year. The results presented in this thesis contribute to our understanding of the risk assessment and management of screen detected nodules, and it is my hope that resulting publications are considered in future nodule management guidelines and help shape the future of screening.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item