Effects of Hemorrhagic Shock on Adrenal Response in a Rat Model

Abstract

Introduction:

There is a documented association between critically ill patients who are in refractory shock and adrenal insufficiency. The underlying pathophysiology may be related to ischemia, necrosis, reperfusion, or resuscitative dilution. We hypothesize this blunted adrenal response is due to ischemia and necrosis of the adrenal parenchyma.

Methods:

Thirty Sprague-Dawley rats were intravascularly catheterized and hemorrhagic shock induced to a mean arterial pressure of 65 mm Hg. After 4 hours of hypotension, fluid resuscitation was initiated with a crystalloid solution (Lactated Ringers). A control group underwent catheterization without hemorrhage. Serum corticosterone levels were measured and adrenal glands harvested for histologic evaluation of hemorrhagic necrosis.

Results:

Baseline corticosterone was 30.8 ng/mL in control animals and 35.3 ng/mL in hemorrhagic animals (P = 0.10). One hour after hemorrhage, corticosterone was maximally stimulated at 406.2 ng/mL and in control animals was 35.0 ng/mL (P = 0.0001). In experimental animals after 4 hours of hypovolemia, corticosterone dropped to 308.9 ng/mL (P = 0.0001). At 6 hours, corticosterone levels dropped to 149.0 ng/mL in experimental animals (P = 0.0001). Adrenal microscopy showed 1.5+ hemorrhagic necrosis in experimental animals compared with 0.0+ in controls (P = 0.004).

Conclusion:

Our model suggests that ischemia and necrosis of the adrenal glands may be responsible for the adrenal insufficiency seen in patients with hemorrhagic shock. Further research may enable clinicians to discern earlier which patients will benefit from adrenal corticoid replacement.

There is a documented association between critically ill patients who are in refractory shock and adrenal insufficiency. Thirty rats underwent hemorrhagic shock, measurement of corticosterone, and histologic evaluation of the adrenal glands. Our model suggests that ischemia and necrosis of the adrenal glands may be responsible for the adrenal insufficiency seen in patients with hemorrhagic shock.

Patients with refractory shock may have relative adrenal insufficiency. Correction of glucocorticoid levels often allows for normalization of blood pressure and adequate resuscitation. It is well documented that patients with a normal hypothalamic-pituitary-adrenal axis develop elevated circulating serum cortisol levels in times of stress or severe illness.^1–4 While it is thought that the general incidence of adrenal insufficiency is low, relative adrenal insufficiency has a high incidence among the severely ill and trauma patients.^5–7 Steroids were initially started for treatment of sepsis in the 1950s. Increasing use of steroids for sepsis was seen after Schumer reported reduction in mortality with treatment using methylprednisolone or dexamethasone.⁸ Since his landmark investigation, several large multicenter trials have indicated an increase in morbidity and mortality with the use of steroids for treatment of sepsis.^9,10 Consequently, the use of steroids for treatment of sepsis was not advocated in the ICU. While these studies used large dosages of glucocorticoid for treatment, more recently, simple replacement dosages of glucocorticoid levels have been studied. In 2002, Annane et al published a prospective trial using low-dose glucocorticoid replacement for the treatment of sepsis, showing a reduction in mortality with no increase in adverse events.¹¹ Increasing use of steroid replacement for treatment of sepsis in the ICU has led critical care physicians to use steroids in other disease processes, such as trauma and hemorrhagic shock. While treatment modalities for refractory shock due to adrenal insufficiency have progressed, the causality remains elusive. The underlying pathophysiology may be related to ischemia, necrosis, reperfusion, or resuscitative dilution. This study was designed to delineate which of these processes is truly involved in the development of adrenal insufficiency after hemorrhagic shock.

METHODS

Approval from the Institutional Animal Care and Use Committee at Eastern Virginia Medical School (#04-019) was obtained prior to undertaking this study. Female adult Sprague-Dawley rats weighing 250 to 300 g underwent initial anesthesia with an intraperitoneal injection of ketamine/xylazine (37 mg/kg ketamine and 7 mg/kg xylazine). After satisfactory depth of anesthesia, the animal was prepared for surgery with hair clipping and skin prep using 10% povidone iodine. Intravenous access was obtained via jugular vein and carotid artery cannulation using PE-50 tubing. Intravenous anesthesia using 0.5 mL ketamine (100 mg/mL) and 0.08 mL of xylazine (20 mg/mL) in 9.5 mL of 2.5% dextrose-0.45% sodium chloride was administered via syringe pump (Harvard Apparatus, Holliston, MA) estimated at 40 to 50 μL/min, and adjusted using physiographic monitoring data. The carotid artery cannula was connected to a pressure transducer (Harvard Apparatus) for monitoring of heart rate as well as phasic and mean arterial blood pressure. Rectal temperatures were recorded and body temperature maintained by use of a warming pad (Harvard Apparatus).

Fifteen control animals underwent anesthesia and intravascular cannulation only. Another 30 animals underwent hemorrhage from the carotid artery cannula, induced to a mean arterial pressure of 65 mm Hg. An initial hemorrhage of 15% of total blood volume was performed over a 5-minute time interval, using 54 mL/kg of body weight for estimation of rat total blood volume. Further hemorrhage or replacement of shed blood was performed to maintain the mean arterial pressure at 65 mm Hg. After 4 hours of ischemia, resuscitation was initiated with crystalloid solution (Lactated Ringers) and continued for 2 hours of reperfusion. Serum corticosterone levels were drawn: 1) prior to hemorrhage, 2) 1 hour into ischemia, 3) prior to resuscitation, and 4) at the end of reperfusion. Serum corticosterone levels were determined using a commercially available ELISA kit (Diagnostic Systems Laboratories, Webster, TX). Adrenal glands were harvested and placed in formalin for evaluation of hemorrhagic necrosis by a pathologist blinded to treatment. Hematoxylin and eosin staining was used to evaluate tissue samples. Animals were killed using a bolus of pentobarbital intravenously at a dose of 100 mg/kg.

Statistical analysis was performed using commercially available statistics software (MedCalc, Mariakerke, Belgium) with the alpha value set at 0.05 for statistical significance. Statistical means were used for comparisons between groups using χ² analysis for differences in serum corticosterone. The nonparametric Kruskal-Wallis and Dunn tests were used to compare tissue necrosis scores.

RESULTS

Unlike cortisol in humans, the rat stress hormone is corticosterone. Serum corticosterone levels were similar for both the control group and experimental group at baseline. Mean baseline corticosterone was 30.8 ng/mL in control animals and 35.3 ng/mL in hemorrhagic animals (P = 0.10). At the 1-hour mark, hemorrhaged animal corticosterone was maximally stimulated at 406.2 ng/mL, while control animal corticosterone levels remained stable at 35 ng/mL (P = 0.0001). After 4 hours of ischemia, just prior to resuscitation, corticosterone levels fell to 308.9 ng/mL for experimental animals and 28.7 ng/mL for controls (P = 0.0001). After 2 hours of reperfusion, experimental animals continued to have a fall in serum corticosterone to 149.0 ng/mL, while controls remained stable at 32 ng/mL (P = 0.0001) (Fig. 1).

FIGURE 1. Corticosterone levels are from an ELISA run on serum. One-hour levels were drawn at 1 hour of ischemia, 4-hour levels at 4 hours of ischemia, and 6-hour levels were drawn after 2 hours of reperfusion.

A pathologist blinded to the study groups processed histology specimens. Specimens were ranked according to the amount of hemorrhagic necrosis present, so that comparisons could be made between groups. A scale of 1 to 4 was used with 1 representing 0% to 25% necrosis of adrenal tissue. A score of 2 represented 25% to 50% necrosis, a 3 represented 50% to 75% necrosis, and a 4 represented a specimen with >75% necrosis (Fig. 2). The statistical mean of hemorrhagic necrosis seen for the experimental group was ranked at 1.5 and ranged from 1 to 3 (Fig. 3). In comparison, controls exhibited no hemorrhagic necrosis (P = 0.004).

FIGURE 2. Histologic score for determination of tissue necrosis.

FIGURE 3. Adrenal gland at 10× and 100× magnification showing hemorrhagic necrosis.

DISCUSSION

Prior retrospective analysis of adrenal insufficiency at our institution led to the initiation of this important project.^12,13 The specific pathophysiology of adrenal insufficiency with regards to hemorrhagic shock is unknown. Possible reasons include ischemia, necrosis, and resuscitative dilution of cortisol or reperfusion injury to the gland. Wang et al suggest that liver impairment may play a role in the pathogenesis of adrenal insufficiency after trauma and hemorrhage.¹⁴ Prior to our investigation, this was the only study in animals that attempted to define the etiology of adrenal insufficiency after hemorrhage. However, adrenal insufficiency associated with sepsis has been evaluated extensively. Plasma from septic patients, along with TNF-α seems to impair the ability of the adrenal gland to synthesize cortisol.^15–18 Animal models of septic shock have shown that a combination of glucocorticoid resistance and reduced synthesis are causes of adrenal insufficiency in this setting.¹⁹

We think that the 4-hour corticosterone levels are the critical data points of this study. These demonstrate that adrenal function becomes impaired during hypotension. This is important because it eliminates reperfusion injury and resuscitative dilution as possible etiologies for adrenal insufficiency. The findings of hemorrhagic necrosis on histology lend support to the theory that it is ischemia and necrosis of the gland that leads to adrenal insufficiency.

The relatively mild amount of necrosis seen in our model may be a function of the severity of hypotension. Other studies using hemorrhage in rats have had more profound levels of hypotension, but for only limited time periods.^14,20 Our model was created to mimic the partially compensated hemodynamically labile patient. The limited amount of hemorrhagic necrosis seen on histology in this study elucidates that adrenal gland function can be impaired with minimal insult. The ability of the adrenal glands to recover is demonstrated by patients who respond to corticoid replacement, and after bridging their acute illness, do not require lifelong steroid therapy.

Our model suggests that ischemia and necrosis of the adrenal glands may be responsible for the adrenal insufficiency seen in patients with hemorrhagic shock. Further benchtop research will better define the role of adrenal glucocorticoid replacement in the hemorrhagic shock state.

Discussions

Dr. Lewis M. Flint, Jr. (Tampa, Florida): The interest in adrenal hormones as a means of modulating responses to stress is long standing, and this report attempts to further clarify our understanding of this matter. As in most studies that make a genuine contribution, more questions than answers are found.

This is a timely contribution because interest in adrenal cortical hormone supplementation or replacement has been rekindled by studies that have demonstrated reduced pain and nausea in patients receiving perioperative cortisone. Moreover, reduced blood levels of adrenal cortical hormones and/or reduced adrenal responsiveness have been observed in patients after cardiopulmonary bypass and in patients with septic shock, multiple trauma, traumatic shock, and traumatic brain injury. Hormone replacement has been associated with reduced vasopressor needs and shorter periods of hemodynamic instability in patients with septic shock. Beneficial effects of hormone administration on other outcomes and in other disease states are far less clear.

The experiments reported here show increased blood levels of corticosterone after shock which then decrease over time, and this decrease is associated with histologic findings of ischemic injury to the adrenal glands. I have the following questions for the authors:

First, what are the functional correlates of the observed blood levels? For example, are the responses normal at all time points or does the response pattern suggest abnormal capacity to produce corticosterone?

The blood levels of corticosterone decreased over time in all animals, including the controls. During the course of the experiment, there was an 18% reduction in the control animals and a 24% reduction in the experimental animals. What does this mean?

What do you make of the histologic findings?

Finally, what happens to the hormone levels if adrenal ischemia is produced by a non-shock mechanism?

Finally, I would like to comment that most laboratory abnormalities in patients who are in shock or have been in shock such as metabolic acidosis, abnormal prothrombin times, and thrombocytopenia really represent symptoms, not diseases. Whether resuscitation strategies should be changed to respond to these abnormalities is a challenging research question. Studies to address this question are difficult to design and execute. Only through careful laboratory investigations are we likely to get meaningful guidance to assist with these problems. I would, therefore, congratulate these authors for their efforts and urge them to continue this work.

Dr. Gregory D. Rushing (Norfolk, Virginia): Regarding question 1 dealing with the functional correlation of corticosterone levels, we did see a correlation at the 4-hour time point during ischemia. We felt that this represented a statistically significant drop in the production of corticosterone from the gland, which should not be the case. The adrenal gland should be maximally stimulated if you are still under stress from hypovolemia.

With regard to question 2 in which you asked about control versus experimental animals, we did see a drop in the corticosterone levels of controls during the ischemic phase. However, it rebounded back to baseline during the resuscitative phase. I don’t really have a good answer for this. In the experimental animals, however, we did show that there was a statistically significant drop in their hormone levels after ischemia and resuscitation.

Question 3 asked about our histologic findings. I think that this observation of only 30% hemorrhage showing a functional decrease in the adrenal gland demonstrates that it is very sensitive to insult, unlike some other glands in the body such as the pancreas, which require massive amounts of destruction before you reach hormone depletion.

And finally, question 4 regarding non-shock mechanisms of adrenal insufficiency. While shock and septic shock have been associated with adrenal insufficiency and it is clearly linked, we still have to discover how adrenal insufficiency is related to other forms of critical illnesses. Traumatic brain injury, multiple trauma such as the ones that you mentioned, have not really been studied in the literature. There are animal models for burn scalded skin syndrome as well as multiple trauma injuries to which we could apply this particular topic for future study.

Dr. R. Neal Garrison (Louisville, Kentucky): ATLS defines shock as inadequate tissue perfusion. Multiple studies have shown that blood or some carrier of oxygen or energy is needed to be included in the resuscitation regimen to adequately return baseline blood flow to the splanchnic circulation. It is of interest that the one study that you commented on, Dr. Wang’s study, uses Ringer’s Lactate as the sole resuscitation solution, as do you.

So my questions are: Why did you not return the blood as we do clinically? Secondly, what indicators of adequacy of tissue perfusion to other organs or to the splanchnic circulation might you use or did you use to determine adequacy of resuscitation?

Dr. Gregory D. Rushing (Norfolk, Virginia): During the hemorrhagic phase, I did use shed blood in order to maintain a blood pressure of 60 to 65 mm Hg. However, during the reperfusion phase, I did not. I solely used a crystalloid solution. The thought for this was that in ATLS we initially give 2 L of crystalloid solution; and if that does not return blood pressure to normal, then we can use colloid solution. And I was able to return these animals to not quite baseline with just a crystalloid solution. Regarding other indicators of tissue perfusion, we looked at lactate levels as well and found that these were decreased. However, I did not present that data here today.

Dr. John A. Morris, Jr. (Nashville, Tennessee): I would like to tie in some clinical observations. We have just finished looking at a group of about 75 patients with adrenal insufficiency out of about 3000 trauma admissions. We found evidence of this disease much earlier than we had anticipated… in the 1st 48 hours post injury. I think you are right on with the ischemia model. So great work. Keep trucking.

Dr. Gerard M. Doherty (Ann Arbor, Michigan): Did you measure ACTH levels in these animals? You haven’t necessarily localized the problems to the adrenal. You demonstrated some damage to the adrenal gland, although not as much as I would have expected necessary to cause this much loss of function. Is the pituitary really the site of failure?

Dr. Gregory D. Rushing (Norfolk, Virginia): No, I did not use adrenal corticotropic hormone measurements in this study. The ELISA for serum corticosterone, which is the stress hormone in the rat, is a commercially available kit that was easy to use in the application of our model. However, I did not look at ACTH.

Dr. Steven E. Wolf (Fort Sam Houston, Texas): Excellent study. In looking at this, what you basically developed is a model of hemorrhagic necrosis associated with hypotension. This was associated with high levels of cortisol secretion. You wonder if the real culprit isn’t high levels of cortisol secretion, not necessarily the hypotension.

So to test that, you could take a model where you replace the hydrocortisone before intervention so you could shut off the whole axis to see if that would be the answer to part of the questions that the adrenal gland is dying because it is having to secrete too much stuff. That may be one of the reasons. If that is the case, that would support your clinical extension of this, maybe you need to replace the glucocorticoids and that wouldn’t happen.

Dr. Gregory D. Rushing (Norfolk, Virginia): I agree. And I thank you for that insight. This is definitely a good idea for study. We are currently trying to get through our Institutional Review Board a study in which we can implement low-dose replacement corticosteroid therapy in our trauma patients. However, maybe an animal model would be the first step to ensure patient safety because these patients may be at higher risk for infections or other disease processes if we start treating them with replacement doses.