Fig. 3: Intra-tumoral macroheterogeneity identified by PACpAInt with whole tumor analysis. | Nature Communications

Fig. 3: Intra-tumoral macroheterogeneity identified by PACpAInt with whole tumor analysis.

From: Pacpaint: a histology-based deep learning model uncovers the extensive intratumor molecular heterogeneity of pancreatic adenocarcinoma

Fig. 3

For 77 cases defined as classical by RNAseq, all the histological slides containing tumor were digitized (n = 660) and classified by PACpAInt-B/C. Top center panel: The PACpAInt-B/C score estimating the “basalness” of each slide is represented on the Y axis while patients (1 to 77) are lined along the X axis. Each spot represents a slide. Cases with all their slides showing a low PACpAInt score (<0.2) were called “pure” classical compared to more heterogeneous tumors called “mixed” classical because at least one slide was predicted to be basal-like. Bottom center panel: Kaplan–Meyer analysis of overall survival comparing “pure” and “mixed” classical tumors (p value = 0.001538). ***p < 0.001; **p < 0.01; *p < 0.05; +p < 0.1; −p > 0.1. Left panel: Case identified to be “pure” classical by PACpAInt (green arrowhead on the central panel). Macroscopic images of the resection showing where the tissue was sampled and the corresponding histological aspect (scale bar = 200 μm). While the areas were spatially distant, the tumor morphology was homogeneous, featuring a gland-forming pattern and good differentiation across the whole tumor. Right panel: Case identified as a “mixed” Classic by PACpAInt (red arrowhead on the central panel). Macroscopic images of the resection showing where the tissue was sampled and the corresponding histological aspect. Here the morphology is highly heterogeneous with spatially distinct gland-forming and non-gland-forming areas. p values were computed using a two-sided log-rank test. Source data are provided as a Source Data file.

Back to article page