Hyperthermia delayed by 24 hours aggravates neuronal damage in rat hippocampus following global ischemia

Clinical retrospective and prospective analyses have documented a high frequency of fever on the days following acute strokes ^[1,2] and cardiac arrests. ^[3] In some of these studies, febrile occurrences have been associated with worsening of neurologic disability scores and with increased mortality. On the basis of these observations, fever has been considered empirically to confer a poor prognosis as regards outcome in patients with brain damage. It is generally agreed that fever, irrespective of its genesis, should be lowered to avoid possible deleterious consequences. However, clear experimental evidence has been lacking to substantiate that moderate hyperthermia is a direct and independent aggravating risk factor, and not merely an epiphenomenon of other complicating conditions (e.g., pulmonary or urinary tract infections, sepsis, or pulmonary embolism from deep venous thrombosis).

Although most experimental studies of temperature modulation of ischemic brain damage have focused upon the neuroprotective effects of hypothermia, there have been a few studies of the detrimental consequences of moderate hyperthermia. The available data demonstrate an acceleration and marked extension of neurologic lesions when mild elevations of cranial temperature are superimposed upon an ischemic insult. ^[4-7] Experimental studies of hypothermia have emphasized the importance of timing of the thermal modulation on the maturation of the resulting neurologic lesions. For example, the benefit of hypothermia is significant at the moment of the ischemic insult ^[8] but inconsistent and limited afterwards. ^[9-12] To our knowledge, however, there have been no experimental efforts specifically designed to substantiate a detrimental effect of elevated body temperature on the day following ischemic brain injury. The present study investigated the role of a moderate, transient, and delayed rise in body temperature 1 day following a 5- or 7-minute global ischemic insult. We sought to ascertain whether the imposed hyperthermic challenge aggravates the extent of neurologic damage induced by the prior insult.

Methods.

Results.

(Table 1) summarizes the physiological variables. Arterial blood gases, plasma arterial glucose, and mean arterial pressure prior to and following ischemia were generally within the normal range and were similar in all groups studied (one-way ANOVA, see Table 1).

Except for the delayed hyperthermic period, mean values for brain temperature were kept constant between 36.3 and 37 degrees C and for rectal temperature between 36.5 and 37.2 degrees C.

(Figure 1 and Figure 2) display rectal temperature time courses during the delayed hyperthermic period (24 hours after the ischemic insult); values were determined every 10 minutes for 3 hours. The mean value for normothermic animals was 38.0 +/- 0.5 degrees C and for hyperthermic animals was 39.6 +/- 0.1 degrees C. This difference was highly significant in both series (repeated-measures ANOVA, see Figure 1 and Figure 2). Hyperthermic animals showed an initial 5- to 10- minute temperature peak that overshot the thermostatically controlled range and coincided with increased motor activity caused by the animals' attempts to escape the hyperthermic chamber.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Rectal temperature measurements in rats of the 5-minute ischemia series during the 3-hour period of delayed hyperthermia (or normothermia) imposed 24 hours following the ischemic insult or sham procedure. Circles denote rats with ischemia + delayed normothermia; squares denote rats with ischemia + delayed hyperthermia; triangles denote rats with sham ischemia + delayed hyperthermia. Error bars represent means +/- SEM. Repeated-measures ANOVA revealed highly significant differences between the delayed-normothermia and the delayed-hyperthermia groups (p < 0.0001), but no differences between the two hyperthermic groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Rectal temperature measurements in rats of the 7-minute ischemia series during the 3-hour period of delayed hyperthermia (or normothermia) imposed 24 hours following the ischemic insult or sham procedure. Symbols are as defined in Figure 1. Error bars represent means +/- SEM. Repeated-measures ANOVA revealed highly significant differences between the delayed-normothermia and the delayed-hyperthermia groups (p < 0.0001), but no differences between the two hyperthermic groups except at the final two time points, at which post hoc Bonferroni tests revealed significant differences (p < 0.05) between the ischemia-hyperthermia and sham-hyperthermia groups (denoted by *).

Animals subjected to an ischemic insult showed signs of CA1 neuronal injury, with scattered clusters of irregularly shaped, dark, shrunken pyramidal cells containing pyknotic nuclei and eosinophilic cytoplasm, surrounded by several swollen reactive astrocytes and macrophages. Figure 3 shows high-power photomicrographs of the middle subsector of hippocampus in rats with 8-day survival following 7-minute global ischemia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Paraffin-embedded brain sections stained with hematoxylin and eosin from the middle CA1 hippocampal subsector, 8 days after 7-minute global ischemia. Original magnification x 1,200. (A) Rat without delayed hyperthermia. Most pyramidal neurons appear normal, and only scattered ischemic neurons are apparent. (B) Rat with delayed hyperthermia. Virtually all pyramidal neurons show the dark, shrunken appearance of classic ischemic cell change.

As expected, 7 minutes of global ischemia produced many more ischemic neurons throughout the subsectors of the hippocampus CA1 area than did the 5-minute insult (Table 2). Animals undergoing ischemia plus delayed hyperthermia showed the highest frequency of ischemic neurons (see Table 2). In contrast, virtually no damage was found in animals submitted to delayed hyperthermia alone (see Table 2). In the latter animals, pyramidal neurons appeared unremarkable, and swollen or reactive astrocytes and macrophages were uncommonly observed.

(Table 2) shows ischemic-neuron counts in hippocampal CA1 subsectors of the two animal series, 8 days after the ischemic insult. Essentially no ischemic neuron cells were counted in sham-operated animals, whereas a variable number were present in all animals subjected to ischemic procedures. Compared with normothermic ischemic values, hyperthermic rats of the first series (5-minute ischemia) showed mean increases in ischemic-neuron counts of 10.5-fold in the lateral, 5.8-fold in the middle, and 4.5-fold in the medial CA1 subsectors. Likewise, ischemic hyperthermic rats of the second series (7-minute ischemia) showed mean increases of 2.6-fold in the lateral, 2.6-fold in the middle, and 2.7-fold in the medial CA1 subsectors compared with normothermic ischemic rats. These intergroup differences did not reach statistical significance in the first series but were highly significant for the three subsectors in the second series (one-way ANOVA followed by Bonferroni procedure, see Table 2).

A gradient of damage was observed from the lateral (least injured) to the medial (most injured) CA1 subsectors of animals undergoing ischemia. These data emphasize the aggravating effect induced by delayed hyperthermia in the previously damaged hippocampus but not in the hippocampus of normal rats.

Discussion.

The data of this study provide direct evidence that delayed hyperthermia aggravates global ischemic pathology. Animals subjected to 5-minute ischemia exhibited small numbers of ischemic cells in the hippocampus, and rats receiving a 7-minute ischemic insult showed moderate numbers of ischemic neurons. These histopathologic findings were exacerbated when comparable animals underwent a moderate and transient elevation of body temperature 24 hours later. These results extend previous experimental studies of the effects of early hyperthermia on histopathologic outcome ^[5-8] and also corroborate clinical observations correlating poor prognosis with the occurrence of fever in convalescent brain-injury patients. ^[1,2]

The pathophysiologic mechanisms underlying this phenomenon remain to be defined, although we can speculate that mechanisms previously identified in the process of ischemic damage may play a role in this process. In fact, recent studies have found that both the postischemic formation of free radicals and the release of glutamate into the extracellular space are temperature-dependent processes. Studies in our laboratory have demonstrated that hyperthermic ischemia leads to a more massive surge in extracellular glutamate levels during the intraischemic period ^[13] and to an accentuation of hydroxyl free radicals during subsequent reperfusion. ^[14] Moreover, several studies have demonstrated that these two processes may be interrelated: glutamatergic mechanisms may be a trigger for hydroxyl radical production during reperfusion. ^[15-19] Indeed, several biochemical pathways might mediate this interrelationship. As a consequence of increased intracellular calcium levels produced by NMDA receptor activation, oxygen free radicals could be generated through the release of arachidonic acid. ^[20] NMDA receptor activation has also been associated with activation of nitric oxide synthase and the generation of nitric oxide. ^[21] Nitric oxide may react with superoxide anion to form peroxynitrite, ^[22] whose decomposition produces a strong oxidant with reactivity similar to that of hydroxyl radical. In addition to these NMDA receptor-dependent mechanisms, free-radical formation may involve other phenomena known to occur following brain ischemia-for example, oxidative deamination during the synthesis of catecholamines ^[23] and neutrophil activation. ^[24] It is likely that several of these pathophysiologic mechanisms may be involved, and it is conceivable that other mechanisms not directly related to the initial injury may also be implicated.

In this regard, investigators studying extreme hyperthermia sufficient to induce heat stroke point out that systemic and local hemodynamic alterations contribute importantly to the resulting central nervous system deterioration. ^[25,26] According to these workers, direct thermal injury to the brain results in congestion of cerebral vessels, cerebral edema, and other neurologic damage. Both cerebral edema and cerebral vascular congestion can induce intracranial hypertension. In response to heat stress, increased central venous pressure and peripheral vasodilation might induce a decrease in mean arterial pressure. Both intracranial hypertension and decreased mean arterial pressure might eventually lead to a local or generalized reduction in cerebral perfusion pressure. Finally, the reduction of cerebral perfusion pressure below the autoregulatory level would itself give rise to cerebral ischemia.

The seeming complexity of the pathophysiology of this condition highlights certain difficulties we encountered since the initial design of this experimental model. To address the direct effect of hyperthermia, we wished to avoid the excessive heat conduction to the brain observed in heat stroke as well as an interference with biochemical mechanisms involved in the primary ischemic insult. The latter consideration weighed against the utilization of immunologic methods to induce fever, and the former consideration urged us to adopt a model of moderate environmentally imposed hyperthermia finely servo-controlled by the animal's body temperature. This system was successful in producing a moderate and consistent elevation in body temperature for a shortlived period (+/- 1.6 degrees C for +/- 3 hours), 24 hours after the primary ischemic insult.

In summary, our data provide the first experimental evidence that delayed hyperthermia aggravates ischemic lesions in vulnerable regions of the central nervous system. The results of the present study suggest that fever must not be viewed as a mere indicator of poor prognosis or as an innocent epiphenomenon complicating the course of brain damage. On the contrary, our results argue strongly for the concept of a direct and independent thermal injury that exacerbates the sequelae of a previous insult. Further investigations of this phenomenon would be useful in allowing extrapolations from the experimentally controlled situation to the clinical setting.

Acknowledgments

The authors thank Susan Kraydieh for assistance with histological procedures.