Fig. 2. CN sample level clustering and clinical association.
a CNC composition by patient sample for Cohort A. Hierarchical clustering of the CNC composition separated samples into four “subtypes” or CNSTs (top stacked bars). The next three stacked bars annotate these samples by their progression-free survival status at 36, 24, and 9 months; followed by tumor microenvironment signature status (TME-classifier); double-hit/triple-hit (DH/TH) status by FISH and DLBCL COO status by NanoString. R/R relapse samples, N/A undetermined. b Visualization of ROIs from four samples from the four CNSTs highlighted in Fig. 2a. Top: locations of individual (primary) cell types. Bottom: individual cells colored by their CNC assignment. c Kaplan–Meier survival plot based on PFS for dual pos mac-rich CNC by median split, log-rank test p value is shown.