IMMUcan (SPECTA NCT02834884) is a European initiative to profile the tumor microenvironment (TME) for a better understanding of immune-tumor interactions. Here, we explored the association between distinct molecular phenotypes and spatial TME patterns in the prospective neoadjuvant IMMUcan TNBC cohort. From a preliminary cohort of 132 patients, matched baseline RNA-seq and multiplex immune fluorescence (mIF) data were available for 66 cases. The mIF panel included CD8, PD1, PD-L1, granzyme B (GB), Ki67 and CK markers. Spatial TME patterns were defined by a graph-based approach detecting densely populated regions of tumor cells and their immune neighbors. TNBC molecular subtypes were derived from RNA-seq as described by Bareche et al. Area Under the Curve (AUC) was used to evaluate the accuracy of spatial patterns to predict TNBC subtypes. A total of eight distinct clusters were identified across the 66 samples, each exhibiting a specific spatial distribution of mIF markers. Two of the clusters showed high performance in predicting immunomodulatory phenotype (AUC: 0.72, 0.71, respectively). These clusters presented elevated densities of CD8+, CD8+/GB+, and CD8+/Ki67+ cells, consistent with CD8+ effector T cells. In addition, a cluster characterized by tumor cells correlated with the luminal androgen-receptor phenotype (AUC: 0.91). The basal-like phenotype was represented by a cluster exhibiting high levels of Ki67+ tumor cells (AUC: 0.61). A distinct cluster displaying an intermediate proportion of Ki67+ tumor cells was observed as well, representing the mesenchymal subtype (AUC: 0.69). These preliminary analyses revealed the presence of informative spatial patterns populating mIF data, linked to the distribution of immune/tumor markers within the TME of TNBC. Of note, these spatial patterns were associated with distinct RNA-seq TNBC subtypes. These findings suggest the predictive power of mIF markers as a potential surrogate to discern TNBC heterogeneity. Consequently, these observations, if confirmed by further validations, could facilitate the implementation of treatment strategies tailored to the TNBC molecular subtypes.
Visual Attention Models are usually tested using collections of natural images that have intentionally salient objects and obvious context information. On the other hand, in the literature, few algorithms have considered datasets with non-context information to modeling attention. Moreover, Visual Attention Models haven't been well-measured considering both contextless and context-awareness environments. In this paper, we compare some well-known Bottom Up visual attention models performance using contextless and context aware datasets, using the Pearson Correlation Coefficient as a method to assess the efficiency of each Visual Attention Model in terms of accuracy and eye fixations predictions. The best algorithm outperforms the others by reaching 59,1% and 43,8% of correlation with ground truth information in the contextless and context awareness datasets respectively.
