Very excited to present my first article, "Mechanistic modeling of metastatic relapse in early-stage breast cancer (eBC) to investigate the biological impact of prognostic biomarkers" !
(supervision @SBenzekry)
#cancer #mathonco #mechanisticModeling #nlme
doi.org/10.1016/j.cmpb…
Our model is (voluntarily) simple and describes the metastatic process through 2 mathematical parameters: alpha (growth) and mu (dissemination), to be confronted to distant metastasis-free survival (DMFS) data
We studied 3 clinical datasets, 2 datasets with routine data (Bergonie, n = 591 and AP-HM, n = 167) and the latter from public databases (IPC, n= 676).
A main original feature is to use the statistical framework of mixed-effects modeling to describe time-to-event (censored) data.
Using bootstrap, we rigorously demonstrated parametric identifiability
A major point was to develop a mechanistic-based covariate selection method.
Starting from univariable analysis, the selection
procedure iteratively test all models with one parameter less, keeping at each step the one with minimal BIC.
The model was able to accurately reproduce DMFS curves stratified by the major biological parameters in eBC.
For prediction, calibration was excellent, but discrimination performances were modest (c-indices of 0.63, 0.67 and 0.72).
These results also emphasized UPA/PAI-1 (only present in the latter data) as having significant predictive power of relapse.
A great thanks to @SBenzekry and X. Muracciole for their supervision and to the other coauthors ( F. Bertucci, P. Finetti and G. Mac Grogan) for their help and guidance.
This work was funded by an @Inria - @Inserm PhD grant and in collaboration with @ap_hm and @paoli_calmettes
The Compo team (team.inria.fr/compo) is part of @inria_sophia, @crcm_marseille and @univamu
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.