Thread by @XiaoFu90

Super excited to share our latest paper in @NatureEcoEvo doi.org/10.1038/s41559… - Spatial patterns of tumour growth impact clonal diversification in a computational model and the TRACERx Renal study.

Very grateful for the amazing support and guidance by @PaulBatesBMM, @TurajlicLab, @kevlitchfield, @eriksahailab, the great collaboration with @YueZhao97 and team, and fantastic research atmosphere and opportunities offered by @TheCrick.

Please see a tweetorial below. 😃

Our understanding of clinically relevant evolutionary dynamics and trajectories remain limited. Thus, we established a computational modelling framework to investigate the spatial and temporal features of clonal diversification. nature.com/articles/s4155…

Outputs from the model are compared to analysis of tumour data in the TRACERx Renal study (doi.org/10.1016/j.cell… , doi.org/10.1038/s41559…).

We observe that the mode of tumour growth, predominantly at the surface (“Surface Growth”) or uniform throughout the volume (“Volume Growth”, impacts the extent of subclonal diversification.
nature.com/articles/s4155…

Surface growth led to more extensive subclonal diversification, corresponding to highly branched tumour evolution, while volume growth resulted either in tumours with limited evidence of clonal diversification or in tumours with early fixation of a fit subclone.

Interestingly, these modes of evolution correspond to evolutionary subtypes identified in ccRCC tumours (doi.org/10.1016/j.cell…).

We next studied the spatial features of clonal diversity. We found that hotspots of clonal diversity (“Microdiversity hotspots”) are increasingly frequent towards the tumour edge.
nature.com/articles/s4155…

The cumulative probability distribution can be approximated by a power law distribution, in both simulations and actual tumours.
nature.com/articles/s4155…

Interestingly, different modes of growth in the model showed different values of power law exponents and corresponded to subsets of tumours mapped to varying clinical outcomes and rates of progression.

Incorporation of central necrosis enhanced the extent of clonal diversification and led to the emergence of fit clones and enrichment of additional microdiversity hotspots at the tumour centre.
nature.com/articles/s4155…

This observation reflected ongoing evolution and selection of fitter clone, facilitated by competition between death and birth events, at the tumour centre, in keeping with our recent finding (doi.org/10.1038/s41559…).

Parallel mutation events with limited clonal expansion in the TRACERx Renal study showed distinct spatial patterns in different ccRCCs, suggesting that ongoing convergent evolution in the same gene could operate at varying locations in a tumour.

nature.com/articles/s4155…

(tweetorial continued)
We hypothesized that the observed distinct patterns of parallel mutation events could be attributed to the proliferation and the recent subclone births.We returned to modelling and found diverse spatial patterns of recent subclone births.

Surface Growth with necrosis led to an enrichment of recent subclone births either at the centre or near the edge. Intriguingly, representative tumours were characterised by such this pattern.
nature.com/articles/s4155…

We then investigated temporal features. Interestingly, Surface Growth models showed more extensive clonal diversification over time with a late-stage collapse of diversity under some conditions, while Volume Growth models showed limited diversification.

These observations reconciled with previous observation of a non-monotonic relationship between the tumour size and clonal diversity in the TRACERx Renal Study.

Lastly, can the model be used to suggest early predictive features for future evolutionary trajectories?
In Surface Growth models budding structures appear to precede extensive subclone diversification, concomitant with a reduction and subsequent recovery in surface circularity.

Interestingly, 16 sub-7cm tumours had radiologically apparent budding structures on the surface, with some already exhibiting higher clonal diversity. Thank you to radiomics team @c_messiou @Prof_DM_Koh and others @RM_Imaging @ICR_London @royalmarsden

nature.com/articles/s4155…

Summary – simulated distinct patterns of clone diversification map to diverse evolutionary subtypes characterised in ccRCCs; temporal features reconciled with the non-monotonic relationship between tumour size and diversity; budding could be predictive of diversification. 😃

Source code for our computational model is publicly available on:
@github (github.com/iamfuxiao/tumo… and github.com/FrancisCrickIn… )
@figshare (crick.figshare.com/articles/softw… )

A HUGE THANK YOU to the incredible team.
Wonderful collaboration with @CharlesSwanton, @uclcancer, and others!
THANK YOU to the funders @CR_UK, @wellcometrust, @royalmarsden.