Because the parameter estimate for pdi_AvSev.ct is about 1/10,000th the size of the standard error and, therefore, if, in the whole population the true effect was 0, you would still get test statistics larger than this almost every time.
As Stephen said in a comment (+1) this could be rounding from a very high value of p, but, in any case, there is pretty much zero evidence that this variable has any linear relation to CAPS_PDI_Cond25_CV.
This is, essentially, like getting 501 heads in 1000 tosses of a coin (or maybe more like 5001 out of 10,000 -- I'm just illustrating, I didn't do the calculations).
Since you haven't told us what these variables are, or what your dataset is, we can't really say more than that.
Note that there could be a non-linear relationship.