Highly read on www.pnas.org in February

In a systematic evaluation of how well mice mimic human inflammatory responses, the popular disease model receives poor marks.

A large multi-institute group co-led by Ronald Tompkins of Harvard Medical School in Boston, Massachusetts, studied genome-wide gene expression in blood samples from 35 healthy people and 411 people with burns, severe blunt trauma or sepsis. They also used public databases to gather human data for other acute inflammatory stresses. Shifts in gene expression as a result of stress were highly similar across the different groups of patients, but correlated poorly with those in a mouse model. Such disparities could lead to misleading results in studies of disease or in the development of therapies, the authors say. The quality of a disease model should be assessed not just for the appearance of disease but also for its biomolecular fidelity, they suggest.

Proc. Natl Acad. Sci. 110, 3507–3512 (2013)