SpaceANOVA: Spatial Co-occurrence Analysis of Cell Types in Multiplex Imaging Data Using Point Process and Functional ANOVA

Abstract

Multiplex imaging platforms have enabled the identification of the spatial organization of different types of cells in complex tissue or the tumor microenvironment. Exploring the potential variations in the spatial co-occurrence or colocalization of different cell types across distinct tissue or disease classes can provide significant pathological insights, paving the way for intervention strategies. However, the existing methods in this context either rely on stringent statistical assumptions or suffer from a lack of generalizability. We present a highly powerful method to study differential spatial co-occurrence of cell types across multiple tissue or disease groups, based on the theories of the Poisson point process and functional analysis of variance. Notably, the method accommodates multiple images per subject and addresses the problem of missing tissue regions, commonly encountered due to data-collection complexities. We demonstrate the superior statistical power and robustness of the method in comparison with existing approaches through realistic simulation studies. Furthermore, we apply the method to three real data sets on different diseases collected using different imaging platforms. In particular, one of these data sets reveals novel insights into the spatial characteristics of various types of colorectal adenoma.

Keywords: IMC; MIBI; R package; co-localization; colorectal adenoma; differential study; multiplex immunofluorescence.

Figure 1

Workflow of the proposed method. Assumptions of spatial point process are used to summarize the level of spatial co-occurrence of pairs of cell types in each image of the subjects from different groups in the form of g function and compared either using a univariate or multivariate functional ANOVA approach. Although only two groups are depicted in the figure, the method can handle any number of groups, i.e., G ≥ 2.

Figure 2

(A) Individual subject-level g function (averaged over images) of every subject in two groups (in the color black) and the group mean function (in the color red), for the pair of cell types: (Treg, Th). (B) Similar plot for the pair: (Tc, Th). (C,D) Cellular organization of representative images from two subjects from the groups SSA and TA/VA, respectively, where different colors correspond to different cell types.

Figure 3

For cell type pairs (beta, beta) and (delta, delta), the image-level and subject-level (averaged over images) g functions of every subject in two groups (in the color black), and the group mean functions (in the color red) are displayed. The top panel corresponds to the subject-level and the bottom panel corresponds to the image-level functions.

Figure 4

(A) Individual subject-level g function (averaged over images) of every subject in two groups (in the color black) and the group mean function (in the color red), for the pair of cell types: (alpha, delta). (B) Similar plot for the pair: (Tc, beta). (C,D) Cellular organization of representative images from two subjects from the groups nondiabetic and onset, respectively, where different colors correspond to different cell types.

Figure 5

(A,B) Power comparison of the methods in simulation setups based on the mixed Poisson process and Neyman–Scott cluster process described in Sections Simulation Based On Mixed Poisson Process and Simulation Based On Neyman–Scott Cluster Process, respectively. The horizontal red-dotted line in every subfigure corresponds to the level α = 0.05.

Figure 6

Power comparison of the methods in the simulation setup described in Section Simulation in the Presence Of Missing Tissue Regions, involving missing tissue regions or holes. The horizontal red-dotted line in every subfigure corresponds to the level α = 0.05.