a b s t r a c tObjectives: Although exponential growth is assumed for lung cancer, this has never been quantified in vivo. Aim of this study was to evaluate and quantify growth patterns of lung cancers detected in the Dutch-Belgian low-dose computed tomography (CT) lung cancer screening trial (NELSON), in order to elucidate the development and progression of early lung cancer. Materials and methods: Solid lung nodules found at â„3 CT examinations before lung cancer diagnosis were included. Lung cancer volume (V) growth curves were fitted with a single exponential, expressed as V = V 1 exp(t/ ), with t time from baseline (days), V 1 estimated baseline volume (mm 3 ), and estimated time constant. The R 2 coefficient of determination was used to evaluate goodness of fit. Overall volumedoubling time for the individual lung cancer is given by * log(2). Results: Forty-seven lung cancers in 46 participants were included. Forty participants were male (87.0%); mean age was 61.7 years (standard deviation, 6.2 years). Median nodule size at baseline was 99.5 mm 3 (IQR: 46.8-261.8 mm 3 ). Nodules were followed for a median of 770 days (inter-quartile range: 383-1102 days) before lung cancer diagnosis. One cancer (2.1%) was diagnosed after six CT examinations, six cancers (12.8%) were diagnosed after five CTs, 14 (29.8%) after four CTs, and 26 cancers (55.3%) after three CTs. Lung cancer growth could be described by an exponential function with excellent goodness of fit (R 2 0.98). Median overall volume-doubling time was 348 days (inter-quartile range: 222-492 days). Conclusion: This study based on CT lung cancer screening provides in vivo evidence that growth of cancerous small-to-intermediate sized lung nodules detected at low-dose CT lung cancer screening can be described by an exponential function such as volume-doubling time.© 2017 Elsevier B.V. All rights reserved.