Figure 3.
Spatial clustering reveals differences in tumor topology that associate with COO and TME abundance. (A-C) The spatial distribution corresponding to B cells were examined in reactive lymph nodes, and in NGCB/GCB tumor environments (A, B, and C, respectively). By visual examination, heterogeneous spatial arrangements of tumor differed by COO classification (GCB/NGCB). The Clark-Evans aggregation index quantifies the level of spatial regularity (index > 1), or clustering (index < 1), and was applied to the B-cell topology classes in reactive lymph node (RLN) and DLBCL (GCB and NGCB). (D) The Clark-Evans index was standardized and compared against the RLN. This indicated the B-cell topology classes in non-GCB cases were significantly more irregularly clustered compared with RLN (Tukey P = .03), whereas GCB had spatial regularity similar to RLN (Tukey P = .47). There was marginal difference comparing GCB with NGCB (Tukey P = .058). This indicates that the spatial distribution of the malignant B cells in GCB tumors more closely resembles the B-cell architecture of normal follicles, whereas malignant B cells in non-GCB tumors were more dispersed. A further multivariate linear model identified that NGCB tumors were significantly more clustered compared with GCB cases after adjusting for each Chapuy molecular signature (C1-C5), and IPI (P = .034, estimate = −0.1974; 95% CI, -0.374 to -0.012). Representative ROIs of highest spatially organized RLN (left, Clark-Evans Index = 1.16), most organized GCB case (middle, Clark-Evans Index = 1.29), and least spatially organized in NGCB (right, Clark-Evans Index = 0.54).

Spatial clustering reveals differences in tumor topology that associate with COO and TME abundance. (A-C) The spatial distribution corresponding to B cells were examined in reactive lymph nodes, and in NGCB/GCB tumor environments (A, B, and C, respectively). By visual examination, heterogeneous spatial arrangements of tumor differed by COO classification (GCB/NGCB). The Clark-Evans aggregation index quantifies the level of spatial regularity (index > 1), or clustering (index < 1), and was applied to the B-cell topology classes in reactive lymph node (RLN) and DLBCL (GCB and NGCB). (D) The Clark-Evans index was standardized and compared against the RLN. This indicated the B-cell topology classes in non-GCB cases were significantly more irregularly clustered compared with RLN (Tukey P = .03), whereas GCB had spatial regularity similar to RLN (Tukey P = .47). There was marginal difference comparing GCB with NGCB (Tukey P = .058). This indicates that the spatial distribution of the malignant B cells in GCB tumors more closely resembles the B-cell architecture of normal follicles, whereas malignant B cells in non-GCB tumors were more dispersed. A further multivariate linear model identified that NGCB tumors were significantly more clustered compared with GCB cases after adjusting for each Chapuy molecular signature (C1-C5), and IPI (P = .034, estimate = −0.1974; 95% CI, -0.374 to -0.012). Representative ROIs of highest spatially organized RLN (left, Clark-Evans Index = 1.16), most organized GCB case (middle, Clark-Evans Index = 1.29), and least spatially organized in NGCB (right, Clark-Evans Index = 0.54).

Close Modal

or Create an Account

Close Modal
Close Modal