The ten barriers for translation of animal data on AKI to the clinical setting

Animals can be pretreated or treated at the time that AKI occurs; humans usually cannot With the exceptions of the perioperative setting (coronary artery bypass grafting and high risk vascular surgery, for example) and high-risk medication use (contrast-induced nephropathy), it is generally not possible to pretreat humans to prevent AKI. Furthermore, in humans, serum creatinine (SCr, our primary biochemical marker of AKI) lags changes in glomerular filtration rate, so by the time AKI is detected, the disease is well established [1]. While there is an effort among basic scientists to prove efficacy of novel therapeutics by administering the agent after the AKI insult, how this would translate to patients where “time zero” is indeterminate remains challenging.

AKI in animals is often due to a single insult; human AKI is usually multifactorial Animal models typically induce AKI with a single severe insult; in contrast, humans may have several acute insults that culminate in AKI. In addition, the single insult that leads to AKI in animal models is often ischemia–reperfusion injury (IRI), whereas sepsis is the most common cause of AKI in critically ill humans. Furthermore, not all humans appear to have the same susceptibility to a given insult. Efforts are ongoing to develop models to address what happens when acute injury occurs on a background of prior injury [2, 3], with the hope that this will address the issue of why some patients develop AKI and others do not, when faced with similar insults.

At its core, human AKI is still poorly understood, and human AKI pathophysiology cannot be inferred from rodent ischemia–reperfusion models Despite advances in disease epidemiology, human AKI remains poorly understood—with very little insight into what changes glomerular filtration rate (GFR) and when it happens. The scope of AKI in humans (defined by sCr and urine output) is a late injury event, and our understanding of molecular, cellular, and intrarenal hemodynamic events remains poorly defined. In particular, there is a major knowledge gap concerning the pathophysiology of human AKI—our assumptions are largely based on rodent IRI studies, which can be considered the equivalent of using Galen of Pergamon’s manual of anatomy based on dissection of monkeys and pigs to define human anatomy. The National Institutes of Health (NIH) has recently committed resources to studying human AKI by evaluating research biopsies using a variety of “omic” approaches [4].

Some common insults in humans cannot be replicated in small animals Although there are number of recent studies focused on models of myocardial infarction-induced [5] and cardiorenal AKI [6] in rodents (which may lead to a better understanding of these conditions), there is no simple rodent model of contrast-induced AKI, and no model of multifactorial ICU-AKI. The absence of better preclinical AKI models represents a major barrier to progress.

Small animals are inherently more resistant to sepsis Unlike humans undergoing surgery, rodents undergoing IRI do not require antibiotics to prevent infection. Along the same lines, rodents exposed to bacterial lipopolysaccharides (LPS) to induce a sepsis-like response require much higher doses than humans [7]. Until recently, many rodent models of sepsis did not include supportive care (e.g., fluids, antibiotics) to mimic the human condition. In fact, acute inflammatory stresses cause a characteristic set of genomic responses in humans, but changes in the expression of the murine orthologs of these human genes are close to random in rodents [8]. Thus, the injury and the host response that lead to AKI in these animal models are inherently different than what occurs in humans.

Young animals ≠ elderly humans in terms of disease susceptibility The expense of maintaining animal colonies for extended periods of time and the desire to complete scientific experiments in an efficient manner have led to the establishment of AKI models that use young animals that undergo a severe and homogenous injury—the predictable nature of the injury results in smaller sample sizes and reduced costs. However, in marked contrast, the humans at greatest risk for AKI are elderly individuals with multiple comorbidities who are often treated with multiple medications, some of which may increase the risk of AKI (such as angiotensin-converting enzyme inhibitors and angiotensin receptor blockers). Furthermore, humans appear to have variable susceptibility to injury.

Time and money are not on our side: rodent models of CKD require months for the animals to age, and old rodents and sheep (best animal models) are expensive Much work has been done to develop AKI models in older rodents with comorbidities including diabetes (though again the injury that results in diabetes mellitus is more homogeneous than the human disease entity). Along the same lines, elegant work has been done to elucidate renal hemodynamics and histopathology during sepsis in sheep AKI models [9, 10]. However, relatively few investigators have the resources to employ and further develop these types of models because they are so expensive compared to models that use young, inbred rodents.

Variable conditions during animal studies have led to low reproducibility even in animals—not surprisingly, the results do not translate to humans Finally, even animal models that use young, inbred rodents have been subject to inherent variability due to issues in lack of standardization of models across research groups, inadequate group size, differences in animal housing and chow, as well as seasonal differences [11]. Other challenges to animal studies have included a lack of blinding and power calculations, measures that are standard in human studies. Consequently, even in these models there has been limited reproducibility—so it is again not surprising that results do not translate to humans.