In the case of the coupling steps in losartan, the residual Pd was around 1000 ppm when the trityl-protected losartan coupling product was crystallized directly from diethoxymethane (DEM). Filtration of the THF/DEM solution of the product through a short column of carbon, silica or chelating resin prior to crystallization still did not provide isolated material having satisfactorily low residual Pd levels. The solution to this problem was to increase the solubility of the Pd by complexing it with tributylphosphine. Thus by adding 10 mol equiv of tributylphosphine vs. Pd prior to crystallization, the residual Pd in the isolated product was reduced to less than 25 ppm. After the final deprotection and salt formation steps, the residual Pd was further reduced to 1-2 ppm.I have not worked on anything as interesting or famous as losartan, but I have been in similar situations where palladium content from a previous step has been an unfortunate surprise. You're always hoping that it gets filtered away, or washed away (ha ha!) or magically evaporates and coats the dryer walls in a lovely mirror that your operators can scrape off and give to you as a present for your wife. (What is most painful, I found, is that quantification of palladium is not something that can be done routinely by a process chemist, and so you're stuck nervously waiting for someone else's results.) Ultimately, filtration was our best bet (after a lot of painful experimentation.)
The above excerpt was published in 2004 (and one imagines the Losartan story is a few years older than that). I think that filtration/chelation technology has advanced further, for all of us who cringe a little bit at adding tributylphosphine right before a crystallization step.
1. King, A.O.; Yasuda, N. "Palladium-Catalyzed Cross-Coupling Reactions in the Synthesis of Pharmaceuticals." Topics in Organometallic Chemistry, 2004, 6, pp 205-245.