Fig. 2.

Inferring the MPN development. (A) Confronting JAK2^V617F (i) or CALR^m (ii) patient observations (black points) to the inferred evolution of the CF among progenitor cells. The dynamics of clonal expansion over time are obtained from 10,000 simulations of the model, sampling the parameters from the estimated posterior distributions. The bold line represents the median CF, computed over 10,000 simulated trajectories at each age, when the lines on both sides materialize 25, 50, 75, 90, 95 and 99% credibility intervals. The gradient of color indicates where the trajectories can be found; the darker, the higher probability. (B) Estimated joint posterior distributions of parameters Δ and λ, for JAK2^V617F (i) or CALR^m (ii). Black solid lines in the one-dimensional histograms correspond to the mean values. The darkest regions on the two-dimensional histograms correspond to those of higher probability density. (C) Schematic representation of the MPN development for JAK2^V617F and CALR^m malignant clones. i) JAK2^V617F mutation is found to be acquired earlier on average than CALR^m mutation (15 vs. 25 y old), and potentially during fetal life (with probability p[λ = 0 |D] = 0.24 when the same probability is estimated at zero for CALR^m). The histograms represent the distribution of parameter λ. ii) CALR^m is found to have a higher proliferative advantage at the stem cell level than JAK2^V617F. CALR^m malignant clone (in orange) will expand over time in a pool of WT HSCs (purple cells) at a higher rate than JAK2^V617F (in blue).