Evaluation of Prediction Models for Identifying Malignancy in Pulmonary Nodules Detected via Low-Dose Computed Tomography

Abstract

IMPORTANCE Malignancy prediction models based on participant-related characteristics and imaging parameters from low-dose computed tomography (CT) may improve decision-making regarding nodule management and diagnosis in lung cancer screening. OBJECTIVE To externally validate 5 malignancy prediction models that were developed in screening settings, compared with 3 models that were developed in clinical settings, in terms of discrimination and absolute risk calibration among participants in the German Lung Cancer Screening Intervention trial. DESIGN, SETTING, AND PARTICIPANTS In this population-based diagnostic study, malignancy probabilities were estimated by applying 8 prediction models to data from 1159 participants in the intervention arm of the Lung Cancer Screening Intervention trial, a randomized clinical trial conducted from October 23, 2007, to April 30, 2016, with ongoing follow-up. This analysis considers end points up to 1 year after individuals' last screening visit. Inclusion criteria for participants were at least 1 noncalcified pulmonary nodule detected on any of 5 annual screening visits, receiving a lung cancer diagnosis within the active screening phase of the Lung Cancer Screening Intervention trial, and an unequivocal identification of the malignant nodules. Data analysis was performed from

Lung cancer screening implementation: Complexities and priorities

“…Nodule management algorithms incorporating risk prediction modelling and longitudinal volumetric analysis have been developed 4,144–149 . Nodule risk assessment using a published risk prediction model at the baseline LDCT has been shown to reduce the number of participants that require interval assessment compared to NELSON or LungRADS criteria 150 .…”

Management of screen-detected lung nodules: A Canadian partnership against cancer guidance document

“…47 External validation studies showed that although malignancy risk prediction models using mean diameter and volume have similar discrimination, the calibration of the models using nodule volume is not as good as models using mean 2-D diameter measurement. 48 A preliminary study in Canada also suggests the PanCan nodule malignancy risk calculator that uses mean diameter measurement may be more efficient in triaging screenees to a diagnostic pathway than the EU-NELSON volumetric protocol after the baseline screening LDCT with a significantly higher positive predictive value. 49 A randomized trial comparing the clinical utility of volumetric measurement with volume doubling time versus 2-D diameter change to determine the timing of the next imaging study or biopsy and cancer detection rate investigation is needed.…”

“…The prediction tool (https://brocku.ca/lung-cancer-screening-and-riskprediction/risk-calculators/) has been validated in several studies in different settings by different research groups. 48,[89][90][91][92][93][94] The PanCan risk calculator has been recommended for use in some settings by the British Thoracic Society Guidelines for the Investigation and Management of Nodules and the American College of Radiology's Lung Imaging Reporting and Data System (Lung-RADS). 32,51,52 One of the greatest strengths of the PanCan nodule malignancy risk calculator is identification of individuals with very low risk nodules who may have the next scheduled screening in two years instead of annually.…”

“…The PanCan model has been widely validated. 48,[89][90][91][92][93][94] One comparative study found that the PanCan model outperformed American College of Radiology Lung CT Screening Reporting & Data System (Lung-RADS) and the National Comprehensive Cancer Network (NCCN) guidelines. 93 The British Thoracic Society Guidelines on the Investigation and Management of Pulmonary Nodules and the Lung-RADS both recommend estimating cancer risk using the PanCan model in some circumstances.…”

Lung Cancer Screening by Low-Dose Computed Tomography: Part 2 – Key Elements for Programmatic Implementation of Lung Cancer Screening

“…Besides the need for being able to evaluate an individual's CT images over time, images obtained during screening may also provide important indications of an individual's risk of developing detectable lung cancer in the next few years [32][33][34], and may become an important component of risk algorithms to determine individually optimized screening intervals, e. g. with one-year intervals only for those at highest risk and longer (e. g. two-year) intervals for others. Finally, in view of improving the specificity of nodule detection, the archiving of screening scans is also important for monitoring the quality of the radiologic screening process, and for further research into possible improvements of algorithms for the radiologic detection of pulmonary malignancies [35][36][37]. As the program starts, artificial intelligence (AI) will not play a role in nodule management, since high-quality annotated and/or segmented datasets are not widely enough available.…”