Computed Biological Relations among Five Select Treatment-Related Organ/Tissue Toxicities

Drug
toxicity presents a major challenge in drug development and
patient care. We set to build upon previous works regarding select
drug-induced toxicities to find common patterns in the mode of action
of the drugs associated with these toxicities. In particular, we focused
on five disparate organ toxicities, peripheral neuropathy (PN), rhabdomyolysis
(RM), Stevens–Johnson syndrome/toxic epidermal necrosis (SJS/TEN),
lung injury (LI), and heart contraction-related cardiotoxicity (CT),
and identified biological commonalities between and among the toxicities
in terms of pharmacological targets and nearest neighbors (indirect
effects) using the hyper-geometric test and a distance metric of Spearman
correlation. There were 20 significant protein targets associated
with two toxicities and 0 protein targets associated with three or
more toxicities. Per Spearman distance, PN was closest to SJS/TEN
compared to other pairs, whereas the pairs involving RM were more
different than others excluding RM. The significant targets associated
with RM outnumbered those associated with every one of the other four
toxicities. Enrichment analysis of drug targets that are expressed
in corresponding organ/tissues determined proteins that should be
avoided in drug discovery. The identified biological patterns emerging
from the mode of action of these drugs are statistically associated
with these serious toxicities and could potentially be used as predictors
for new drug candidates. The predictive power and usefulness of these
biological patterns will increase with the database of these five
toxicities. Furthermore, extension of our approach to all severe adverse
reactions will produce useful biological commonalities for reference
in drug discovery and development.