Why do animal models (sometimes) fail to mimic human sepsis?

OBJECTIVE: To describe potential mechanisms that may account for the observation that drugs that work in treating sepsis in animal models often fail in sepsis trials in patients. DATA SOURCE: MEDLINE searches were used to provide the key information. DATA
SUMMARY: Most animal studies are performed acutely in young healthy animals, whereas a significant percentage of the clinical population is elderly with many secondary complications (e.g., diabetes, systemic vascular disease, high blood pressure, immune suppression, cancer). Furthermore, unlike the acute challenge presented in most animal studies, many of the septic patients' clinical histories indicate a relatively slow onset of the disease. In many animal studies, intervention occurs before or during the very early stages of sepsis, when inflammatory cytokine levels are still rising and both organ damage and vascular leakage are minimal. In contrast, current treatment strategies are started when many (probably most) patients are switching from a proinflammatory cytokine response to an anti-inflammatory response and organ damage is already apparent. Patients are also generally receiving some form of supportive therapy (e.g., fluids, vasopressors, ventilators). Because these are seldom used in the animal model, their effect on a particular drug response is difficult to assess. In the animal model, a well-defined bacterial strain, endotoxin challenge, or, in the most complex case, cecal ligation puncture at a defined site is employed to bring about the onset of sepsis. Generally, bacterial proliferation can be controlled in these situations by selecting the appropriate antibiotic (if desired). In human sepsis, the pathogenic bacteria are often not known, mixed infections involving both Gram-negative and Gram-positive bacteria are common, and antibiotic treatment is incomplete or ineffective. Anti-inflammatory strategies that impair bacterial killing may be helpful in cases in which antibiotics were effective and harmful when they were not. Thus, an intervention in human sepsis is attempted at a later stage and under very different conditions than it is during efficacy testing in animal models.
CONCLUSIONS: Differences in the nature of the initiating agent causing sepsis and the lack of co-morbidities in the animal models probably contribute to some of the differences in animal studies and clinical trials in sepsis.