Not sure

It's thought that multicellular organisms have codon 
usage patterns which reflect gene-to-gene variation 
in nucleotide composition rather than selection on 
synonymous codons

Unicellular organisms seem to show more selection
for abundant tRNAs.

This is not always the case, though.
(I also don't really trust inferences made for all unicellular organisms based on E.coli and 
S.cerevisiae, which have been selected for many generations to replicate as quickly as possible.)

The pattern of Drosophila resembles that of humans, 
but the highly expressed genes have higher G+C content.