The quest for protection from ischemic (actually ischemia/reperfusion) injury is not new, but in the past several years, this concept has been approached somewhat more rigorously by investigators addressing important clinical problems, particularly heart attack, cardiopulmonary bypass, stroke, peripheral vascular embolism, and the preservation of organs for transplantation. Much of the recent work has focused on the rather remarkable observation that relatively short periods of ischemia sustained just prior to a more prolonged (i.e., clinically relevant) episode appear to have improved the tolerance to the latter in a number of organs, including the heart, brain, ¹ spinal cord, ² skeletal muscle, ³ retina, kidney, intestine, ^4–6 and liver. ^7–17

A particularly thoughtful, exciting, and potentially clinically applicable example of this approach is provided by Dr. Clavien and his colleagues at the Duke University Medical Center in this month’s issue of Annals of Surgery⁸ : a group of 24 patients undergoing hemihepatectomy alternatively received either ischemic preconditioning (10 min ischemia and 10–15 min reperfusion) or no preconditioning prior to a uniformly fixed period (30 min) of total inflow occlusion to facilitate hepatic transection. The patients subjected to ischemic preconditioning showed less evidence of hepatocellular injury (serum AST and ALT levels) and hepatic endothelial cell apoptosis (TUNEL staining and changes in morphology), but no significant differences in mortality nor need for intensive care or hospitalization. (The differences between the two groups in morbidity raise questions about the comparability of the two treatment groups in this nonrandomized, nonstratified study more than they suggest an effect of ischemic preconditioning on the occurrence of largely unrelated complications.) The apparent incremental benefit of ischemic preconditioning for that (unfortunately not stratified) subgroup of patients with preexisting hepatic steatosis, a population known to be particularly vulnerable to hepatic ischemia/reperfusion, is especially noteworthy. Therefore, notwithstanding its methodologic limitations, this study stands as an important step in the translation of this curious experimental observation to the benefit of our patients.

A broader and more fundamental question about ischemic preconditioning is, how does it work? Here, I believe, we have an excellent example of the principle proverbially illustrated by the strikingly disparate interpretations of the true nature of an elephant by a group of blind men, each one based upon the limited palpation of a different appendage thereof. Similarly, the somewhat remarkable salutary effects of ischemic preconditioning have been variously attributed to arachidonic acid metabolism (prostaglandins/leukotrienes), ^18,19 acidosis, ²⁰ calcium fluxes, ²¹ reactive oxygen species (“free radicals”), ^4,22–25 including the currently fashionable nitric oxide free radical, ²⁶ the synthesis of antioxidant enzymes, ²⁷ heat shock protein expression, ^28,29 the PARS pathway, ³⁰ apoptosis, ³¹ purine (adenosine) signaling, ³² kinins, ^33,34 glutamate receptors, endothelial adhesion molecule expression, ^5,35,36 and a host of intracellular signalling pathways, including tumor necrosis factor, ³⁷ protein kinase C, ³⁸ G-proteins, ceramide, ³⁹ and innumerable others.

Indeed, each of the above cited studies does suggest the association of one or more of the above mechanisms with the clinically relevant beneficial effects of ischemic preconditioning. But there are two major obstacles to overcome before we can attribute one or more these pathways to this salutary phenomenon. The first is what I perceive to be the single most common logical fallacy in the medical literature, what Aristotle called the post hoc ergo propter hoc fallacy, which simply means that association does not necessarily indicate causation. (I sought the appropriately illustrative analogy of this commonly ignored principle for years, until it was provided by the distinguished scientist, Bruce Ames, who, while giving the John E. Hoopes lecture at Johns Hopkins in 1996, pointed out that those deluded by this fallacy should also believe that the population of southern Florida is born Hispanic and dies Jewish.) Each investigator, (appropriately) focused on the study of a particular mechanism/pathway of tissue injury/protection applies his or her techniques and assays to the phenomenon of ischemic preconditioning, and, mirable dictu, they are significantly affected. Convinced as each of us are of the overwhelming importance of our own pet mechanism, we infer causality by a process that is literally irrational. (For years I have suffered from the (totally unbiased) delusion that ischemic preconditioning is largely related to free radical metabolism!) Here, Clavien and colleagues provide some circumstantial evidence for diminished hepatic apoptosis in response to ischemic preconditioning, and speculate about causality. Indeed, I suspect that each of the above-described prototypical investigators might have sought and seen changes in the parameters related to his or her pet pathway in this study, yet we have no idea from this approach whether these changes represent causes, effects, or epiphenomena.

This is greatly complicated by the second major problem in understanding this phenomenon: the complex interrelationship of these mechanisms. For example, those who study reactive oxygen species delight in pointing out that each step in the pathway of arachidonic acid metabolism (both the cyclooxygenase pathway that generates prostaglandins and the lipoxygenase pathway that produces leukotrienes) is a free radical-mediated lipid peroxidation; therefore, the discussion of whether the events triggered by this pathway are oxidant- or arachidonate-mediated are academic and largely moot. With regard to the present study, we cannot discriminate the meaning of the appearance of perivascular apoptosis as a seminal causative mechanism from a nonspecific end response to a totally disparate injury pathway. While it is a politically comfortable cliché to say that the mechanism is undoubtedly multifactorial and that the various pathways interact, can we be more specific? Of all those pathways apparently affected by the preconditioning stimulus, do they act in series or in parallel? Oxidant-mediated reperfusion injury proceeds by at least one linear, series-coupled pathway (greatly oversimplified here): ischemia → proteolytic activation of xanthine oxidase from xanthine dehydrogenase in the endothelium → superoxide generation by xanthine oxidase at reperfusion → a characteristic free radical chain reaction → endothelial surface selectin and integrin expression by a yet to be defined mechanism that involves platelet activating factor → leukocyte rolling, sticking, and diapedesis → microvascular inflammation and injury → secondary compromise of nutrient perfusion and inflammation → parenchymal organ dysfunction. While this linear sequence interacts with and branches into other pathways, its fundamental linearity has provided a conceptual framework that has facilitated fundamental progress and also led to successful clinical trials ^40–42 (also S. Grossman et al, unpublished data, 1995) over the past 20 years. With respect to ischemic preconditioning, if these multiple pathways act in parallel, are the pathways additive, antagonistic, or synergistic? Are the events triggered fundamentally at the transcriptional or posttranslational level? (The short time course almost certainly indicates the latter.) And in which of the multiple hepatic cell types does the trigger mechanism reside? Considering the generalized nature of the phenomenon, I suspect, as do the authors, that it is the sinusoidal endothelium. Because it is both triggered by ischemia/reperfusion and protects against ischemia/reperfusion, I suggest that it is related to the above mechanism of ischemia/reperfusion, at least to a substantial degree.

Why does this matter? In one sense, if ischemic preconditioning is beneficial, let’s just apply it and not worry too much about its mechanism of action. After all, the beneficial effects of aspirin were experienced by our patients long before we understood how it worked. Fair enough. But aside from the purely academic, theoretical argument that understanding trumps ignorance, here we have a very real and practical opportunity to initiate this protective response pharmacologically or even genetically rather than by an approach as crude (and of limited applicability to other clinical situations) as a Pringle maneuver. Indeed most, albeit not all, of the logarithmic growth in overall therapeutic efficacy seen in the past decade has been due to our increased understanding of fundamental mechanisms, and of discriminating those that function critically as initiators, control points, or final common pathways.

By all means, then, let us aggressively pursue the exciting empirical observation that ischemic preconditioning appears to protect the liver (and other organs) from ischemia/reperfusion into larger and more elaborately designed (formally randomized, stratified, multicenter, perhaps partially blinded) clinical trials, for which this study provides an important initiative and rationale. Moreover, let us to continue to observe and measure those parameters that we believe might be related to and therefore provide leads as to mechanism. But we must also proceed in carefully designed laboratory experiments, to specifically perturb precisely defined systems (such as hepatic endothelial cells in culture) in ways that define the putative causative role of the biochemical/pathophysiologic pathways that determine the response to such an insult. Such studies must be more than mechanistically descriptive (which is necessarily a major limitation of clinical studies), but strategically designed to discriminate cause from effect from epiphenomenon. We need to know not just what happens, but how it happens. I suspect that such truly mechanistic studies will provide the basis for more elegant and less traumatic salutary interventions than the peremptory administration of the “hair of the dog” that is about to bite you.