Nicole Bassoff writes the THURJ-web-exclusive column, “A Peer’s Review,” which highlights relevant trends and developments in research practice today. This is her first post in the weekly feature. Nicole is currently a freshman in Wigglesworth, a future resident of Adams House and a prospective MCB concentrator.

The use of animals in biological or medical research laboratories is often a source of unease and controversy. Unfortunately, the state of alternative methods such as computer modeling, cell culture, and other in vitro techniques is such that the most powerful progress in the understanding of human disease and the development of new drugs is still dependent upon the use of model organisms like mice, rats, and fish. While those in the medical and research community who find animal testing untenable will continue to pursue new methods, the fact remains that for many diseases and disorders, there are no other alternatives when it comes to providing relief and saving lives.

At the moment, then, the most pressing question for the medical research community is not whether we should use animal models, but when and how we should use them. In order to be both justified and valuable, the use of animals in research must be minimal, cost-effective, and, most importantly, predictive—that is, results achieved in model organisms must accurately and with some consistency predict results in humans. It stands to reason that simply by increasing the predictiveness of a model, one can reduce the number of animals needed and the time and cost of research. Though this may seem obvious, there are several recent cases in which researchers have been slow to re-evaluate their model organisms.

Consider sepsis, a potentially deadly condition caused when the immune system overreacts in response to infection, and the leading cause of death in intensive care units. For many years, mouse models of sepsis informed the research that led to the development of nearly 150 drugs—all of which failed in human trials. The reason for these inconsistencies between outcomes in mice and humans was finally explained by a study published last month in PNAS, which investigated the genes involved in the white blood cell response to sepsis, burns, and trauma and found that there were absolutely no similarities between the gene response patterns of humans and those of mice.

The implications of this were so shocking, and ran in such opposition to long-held confidence in the sepsis mouse model, that the study met much resistance and was rejected several times during its year-long pursuit of publication. In a recent New York Times article, lead author Ronald W. Davis commented on the reluctance of the scientific community to question the mouse model: “They are so ingrained in trying to cure mice that they forget we are trying to cure humans.” This serves as an important reminder for researchers that animal models can only go so far in replicating human disease and may not be feasible for use in some research.

In other instances, mouse models may fail because their use is not adjusted to keep up with developments in scientific understanding. One pressing example has been the under-use of mouse models to adequately investigate sex differences in drug response. In stress-related disorders, for example, women not only have a higher incidence than men, but they have also shown a lower response to treatment.

One cause of this may be the continued use of exclusively male animals in preclinical research, and a failure to examine sex as a factor during data analysis. Of course, this only compounds with the later discrepancies between enrollment of males and females in clinical trials. Given the growing body of biological information regarding differential sex response, drug developers would do well to begin the investigation of sex differences earlier in the process, through better use of mammalian models.

Despite cases of reluctance to change old or out-dated practices in animal research, some innovative scientists involved in translational research and drug development are realizing the potential of animal models in pre-clinical testing, with much success. In one recent and exciting example, highly accurate and predictive genetically engineered mouse models for acute promyelocytic leukemia (APL) paved the way for very successful clinical trials. APL is now a highly curable condition.

Here at Harvard, we are making great strides in animal modeling through the development of “humanized mice,” or mice containing human genes, cells, and tissues. At the Harvard University Center for AIDS Research, these mice are central to the acceleration of the development of an HIV vaccine.

Such advancements serve as evidence of what can be achieved through careful, humane, and regulated use of animals in research. It is likely that our reliance on these model organisms will continue for years to come, and in the absence of significant alternatives, we must be sure to constantly re-evaluate and innovate in the pursuit of efficient and productive animal research practices.