As far as I know, immunohistochemistry is a valid method of determining regions of a specific protein expression, by capitalizing on the specific antibody-antigen binding.

So, finding spike protein expression in the heart does prove that the spike protein reached the heart, assuming no other sources of spike protein exist and assuming they used a proper spike protein-targeting antibody. (I just realized this paper did not provide their methods section.)

However, the possibilities of human error always exist. And it's bad practice that the paper did not provide a control tissue staining that shows no such spike protein expression.