Fig. 5: Growth modes impact the temporal features of clonal diversification.
a, Number of subclones as a function of the diameter of a 2D tumour slice in in silico tumours under surface growth and volume growth, for n = 50 simulations with \(p_{\mathrm{driver}} = 6 \times {10^{-4}}\) for each condition. b, KDE with respect to the number of subclones and the diameter of a 2D tumour slice in in silico tumours under volume growth (i) and surface growth (ii). Each KDE plot is based on 250 simulations (50 per condition) under five conditions with \(p_{\mathrm{driver}} = 2 \times 10^{ - 4},4 \times 10^{ - 4},6 \times 10^{ - 4},8 \times 10^{ - 4}\, {\mathrm{and}}\,1 \times 10^{ - 3}\). c, Vector maps of evolutionary flows over time with respective to the number of subclones and the average number of drivers accumulated among tumour voxels in surface growth models without (i) or with (ii) the implementation of necrosis. n = 50 simulations with \(p_{\mathrm{driver}} = 6 \times 10^{ - 4}\) are shown. d, The spatial patterns of parallel mutations in PBRM1 (upper, distinct events in different colours) and microdiversity (lower) over time in a representative in silico tumour under surface growth. The arrows indicate budding structures preceding subclonal expansion. e, Time evolution with respect to the number of subclones and the mean fitness of the tumour slice, along with the diameter of tumour slice, contour circularity and average number of drivers accumulated, in surface growth models. n = 50 simulations with \(p_{\mathrm{driver}} = 6 \times 10^{ - 4}\) are shown. f, Axial image in the corticomedullary contrast phase of a representative case (K523) showing budding structure on the tumour surface (red arrow). Outlines in red reflect the tumour contour giving volumetric tumour coverage. g, Maps of tumour regions with the number of subclones colour coded in a representative case. Hues from red to purple to blue reflect decreasing number of subclones. ‘Low’ and ‘High’ reflect one and four subclones, respectively. h, The number of subclones as a function of ultimate tumour size in the TRACERx Renal study, overlaid with KDE based on simulated data. Tumours with size up to 7 cm and with radiologically evident budding structures are highlighted (orange). Contours reflect 90% probability density based on in silico tumours under surface growth (red) and volume growth (blue) in Fig. 6b.
