Over the past two decades, our understanding of intercellular communication between glia has fundamentally switched from the idea of a syncytium to the recognition that glial cells might in fact organize as networks. In particular, astrocytes, the main type of glial cells in the cortex, can propagate calcium signals from one cell to the other through gap junctions. The reported speed and extent of propagation of these intercellular calcium signals however can largely vary. Of course, this variability in the propagation patterns may reflect different intracellular properties (biochemical, signaling). But experimental evidence also suggests that the way astrocytes connect to each other in the network (topology) varies depending on the brain region. Such different topologies may already bring forth, by themselves, different modes of intercellular calcium propagation. Here, we explore this possibility using a biophysically realistic model of large (i.e. >1000 cells) tridimensional astrocyte networks. In our networks, each astrocyte houses an individual model for intracellular calcium and IP3 dynamics and exchanges IP3 with connected astrocytes through gap junctions. Intensive numerical simulations of the model for different network connectivities revealed that the major classes of observed propagations can be emulated by a mere variation of the connection topology (i.e. keeping intracellular parameters unchanged). In particular our study indicates that calcium wave propagation is favored when the connections between astrocytes are mainly restricted to small inter-cell distances. This result is significant since, at constant number of cell-cell connections, space-constrained topologies exhibit large mean-shortest path. As a consequence, we obtain the non-trivial result that propagation is improved when the mean-shortest path of the network is large. Altogether, our findings provide theoretical support to the experimental observation that the spatial arrangement of astrocyte networks in the brain could bear some level of organization with deep implications on the regulation of network activity.