The notion of disease-associated single-nucleotide polymorphisms (da-SNP), as determined in genome-wide association studies (GWAS), is relevant for many complex pathologies, including cancers. It appeared that da-SNPs are not only markers of causal genetic variation but may contribute to the disease development through an influence on gene expression levels. We argue that understanding this possible functional role of da-SNPs requires to consider their embedding in the tridimensional (3D) multi-scale organization of the human genome. We then focus on the potential impact of da-SNPs on chromatin loops and recently observed topologically associating domains (TADs). We show that for some diseases and cancer types, da-SNPs are over-represented in the borders of these topological domains, in a way that cannot be explained by an increased exon density. This analysis of the distribution of da-SNPs within the 3D genome organization suggests candidate loci for further experimental investigation of the mechanisms underlying genetic susceptibility to diseases, in particular cancer. Recently developed techniques of chromosome conformation capture combine chemical crosslinking and sequencing to identify genomic loci contacting each other in vivo. They have shown that the mammalian genome displays three main architectural features at the large-scale level (supranucleosomal level, beyond the kb scale), nested in a hierarchical way (Figure 2): chromatin loops, topologically associating domains (TADs) of larger size exhibiting more internal contacts than contacts between domains [34, 35], and a segregation in active and inactive compartments [36].