It is established that gradient separations cause differential loss of white
cell types; one would also want to avoid washing, since this could also lead
to differential cell loss.
In clinical hematology systems, sample volume is measured; samples are lysed
but not washed. I'm not sure that there's any published paper examining
diferential cell loss due to lysing agents, because that would require that
a whole-blood counting method be used to determine counts without lysing,
and I'm not aware that any large-scale study comparing whole-blood and
lysing methods has been done. Several non-lysing whole blood counting
methods exist; the oldest is that developed by Adams and Kamentsky in the
1970's using acridine orange in isotonic saline, which gives counts of
lymphocyts, monocytes, and granulocytes, but does not discriminate
eosinophils and basophils. The Block Engineering technology of the
mid-1970's used a mix of fluorescent dyes on glutaraldehyde-fixed cells to
produce a five-part differential, but originally required two- or three-beam
excitation. More recently, (mid 1980's) Leon Terstappen, Mike Loken et al,
then at B-D, described methods using combinations of dyes (Thiazole orange
and LDS751) and monoclonals to provide multipart differentials including
retic counts. Various modifications, permutations and combinations of the
above methods are workable; which one is best for you depends on which cell
types you need to count.
-Howard