I think that the critical finding is the gelatinous appearing intranuclear inclusion, since the density of a large nucleolus or an atypical nucleus ( in case of high grade urothelial carcinoma) is different. The presence of inflammation and degenerative changes make the whole picture more difficult.
Not only BK but also CMV can contribute to epithelial cells that look like cancer cells. The infected cells have an altered morphology of the nucleus, cytoplasm and over all cell shape. This is because of the viral inclusion bodies which are present and can grow over time eventually become salient. It is these viral changes that differentiate the two conditions. The inclusion bodies can be found in the nucleus but also in the cytoplasm for CMV. The typical stains used are PAP and H&E. The n. inclusion bodies are basophilic. The chromatin can be altered in appearance (granular or blebs). This alter chromatin can have the “ground glass” look or "gelatin like" look. (See above) (It looks better the sooner you fix it). In some cases the chromatin may have light and dark areas. Also a space or clear halo may surround it. If the cells are from the urothelium the overall shape of the n. may be irregular and the cell body may display an atypical shape. The difference between CMV and Bk is the "owl eye" or "birds eye" look of the inclusion. Which also is not found in cancer. Antibodies for immunocytochemistry are available for these viruses particularly CMV . So if there is confusion, the viruses can be stained for in another prep. Even if the cells don't look as good as you would wish. As noted above, inflammation and degenerative changes complicate the picture, but as these types of cells are often found in immune compromised patients, as you know, the history will help with what to look for. The better your sample is collected, fixed and stained the clearer these changes will be. Check with Micro to see their fresh preps of viral cultures even unstained these can help.