The public analysis code for Kuil et al. appears to test individual CosMx cells as independent observations using MAST and cell-level Wilcoxon tests, without accounting for the mouse or tissue core from which each cell originated. Because radiotherapy was assigned to mice and the experiment included only three mice per condition, this approach may substantially underestimate uncertainty and inflate the significance reported in Figure 4. The metadata code also indicates unequal representation of brain regions and source cores after two TMAs were excluded for low cell numbers, raising possible treatment–region or treatment–animal confounding. The authors are asked to provide the complete TMA-to-animal sample key and repeat the analyses using animal-level pseudobulk or appropriately specified hierarchical models. Clarification is also needed on whether individual microglial images or mouse-level summaries were used as the experimental units in Figures 2 and 5. In addition, animal-level reanalysis is required to determine which transcriptomic and morphometric conclusions remain supported.
