IPT imfilter() is a linear spatial filter (convolution or correlation). On the other hand, histeq() applies a nonlinear value transformation to the image globally; it is not a spatial filter at all. You can't use one to do the job of the other.
At arm's length, adapthisteq() is similar to histeq(), but it operates blockwise on local information and the blockwise calculations are combined via interpolation. Similarly, there really isn't any relation to a linear spatial filter, nor is one used internally.
So disregarding what qualifies as "enhancement", if the goal is to effect a value transformation similar to that provided by histeq()/adapthisteq(), then no, I don't see any sensible way to use imfilter() to accomplish the goal.
