Several studies plot the density (or effect size) of eQTLs by distance to gene TSS; such as:

As you can see, these desnities are highly peaked around the TSS, but are heavy tailed. The reason to go further than 100kb or 250kb is to capture long-range eQTLs that can arise from chromatin looping -- i.e., bringing opposite sides of topologically associating domains (TADs) into close proximity. (Sensitivity)

A reason not to go further than 100kb or 250kb is power; by incorporating more genes (or more SNPs) into your analysis, with insufficient data you necessarily lose power to detect what you're looking for, and run the risk of false-negatives (and potentially false-positives as well) by diluting true signal. (Specificity)