When reviewing a paper, you subconsciously have an expectation of how good it is. If you know previous work by the authors and liked that work, you expect to find good work in the paper under review (and you are much more disappointed if this expectation is not fulfilled). This can affect the way the paper is read and hence the review as well. Similarly, the institution names of the authors induce some expectation of quality if the institution is known to the reviewer. Double-blind review is supposed to minimize this effect, but there are always some hints left in the paper.
Now a western reviewer may be less likely to know for the majority of Asian institutions how well-reputed they are. This means that the authors do not get a subconscious bonus here. This is normally not a big deal if the paper is good enough, but for journals with low acceptance rates, this can be problematic for papers that are close to the borderline of acceptance.
Likewise, language may be a big deal here. For instance, experienced US-American reviewers may have seen all the usual writing errors that French, Italian, and German native speakers typically make, and because of this these do not obstruct their understanding of the paper content. But for the typical errors made by Asians, this may be different, and a reviewer who is not certain about the meaning of some statements in the paper is just less likely to write a very supportive review. Local customs in structuring the paper and building up a compelling argument may also lead to a similar effect.