Glucocorticoid Therapy in the Intensive Care Unit
http://www.100md.com
《新英格兰医药杂志》
Acute adrenal insufficiency, a rare cause of shock, is manifested as shock that is poorly responsive to fluid resuscitation and pressors, not unlike cardiogenic or septic shock. It is almost always associated with a history of supraphysiologic glucocorticoid administration or primary adrenal cortical disease. The best single test for the evaluation and diagnosis of this disorder is based on the ability of the adrenal glands to respond to a challenge with synthetic adrenocorticotropic hormone (250 μg of cosyntropin administered as an intravenous bolus, followed by the measurement of plasma cortisol levels 30 and 60 minutes later). Normally, the plasma cortisol level increases to more than 18.5 μg per deciliter (510.4 nmol per liter). A more useful number to keep in mind for the plasma cortisol level is 20 μg per deciliter (551.8 nmol per liter). In an insightful series of articles, Kehlet showed that among patients who were suspected of having adrenal insufficiency, those in whom a plasma cortisol level of 20 μg per deciliter could be generated were able to withstand the rigors of major surgical procedures without glucocorticoid supplementation.1
Patients with shock are generally treated in medical intensive care units (ICUs), in which the cosyntropin stimulation test is generally used to rule out the diagnosis of adrenal insufficiency. Since many more patients in ICUs appear to meet the biochemical criteria for adrenal compromise than would be expected, essentially all such patients end up being treated with large doses of intravenous glucocorticoids, usually the equivalent of 600 mg of hydrocortisone per day. If adrenal insufficiency is indeed the culprit causing their symptoms, glucocorticoids may be considered lifesaving. If not, glucocorticoid administration adds cost without benefit, for these agents are antianabolic steroids that result in decreased synthesis of new protein, attenuation of immunity, impairment of insulin action, loss of calcium, and other untoward effects. In a bedridden patient, this is a bad combination of effects that can lead to rapid clinical deterioration. It is unsettling that so many patients in ICUs meet the generally accepted biochemical criteria for adrenal compromise and thus receive glucocorticoids. Indeed, one might well ask whether the right test is being used to identify patients who truly require supplemental glucocorticoids in this circumstance.
A potentially important insight into this question is provided by Hamrahian et al. in this issue of the Journal (pages 1629–1638). These investigators report that plasma cortisol responses to synthetic cosyntropin frequently fall below the cutoff value of 20.0 μg per deciliter in critically ill patients who have hypoalbuminemia but not in matched critically ill patients with normal plasma albumin concentrations. They attribute the apparently low response to the fact that cortisol, which circulates primarily bound with high affinity to a specific binding globulin, cortisol-binding globulin, and with low affinity to albumin, is measured as "low" in patients with hypoproteinemic states (see Figure). Thus, alterations in the concentration of cortisol-binding protein would be expected to affect measurements of total plasma cortisol. A dramatic finding in this study, however, was that in patients with a low total cortisol response to a cosyntropin challenge, the circulating "free cortisol" concentration was not significantly different from that in patients with normal albumin concentrations. In other words, it appears that even though the total plasma cortisol response to a cosyntropin challenge is low in such patients, the response of free bioactive cortisol is appropriate for the situation, presumably because the feedback regulation remains intact.
Figure. Circulating Plasma Cortisol Levels in a Normal Person, a Stressed Person, and a Stressed Person with Hypoalbuminemia.
Approximately 10 percent of cortisol is present in the free (bioactive) state, unbound to protein; 20 percent is loosely bound to albumin; and 70 percent is tightly bound to cortisol-binding globulin. Binding to cortisol-binding globulin is nearly saturated at plasma cortisol levels in the range of 15 to 18 μg per deciliter (413.8 to 496.6 nmol per liter). With stress and hypoalbuminemia, the maximal total cortisol measurement will fall below that in persons with a normal albumin concentration and, often, below the traditional cutoff level for the cosyntropin stimulation test, even though the level of free, bioactive cortisol is the same in these two groups and is appropriate for the clinical situation. To convert values to nanomoles per liter, multiply by 27.59.
This finding was anticipated by earlier work that demonstrated the consistency of free cortisol concentrations in situations in which the concentrations of binding globulin vary widely.2 Hence, in critically ill patients in the ICU, the circulating free cortisol response to synthetic cosyntropin is a much more important clinical determination than the total cortisol response. As things stand, plasma free cortisol is not easy to measure, but a few years ago, it was not easy to measure plasma free thyroxine. Now, free thyroxine is the workhorse of the thyroid-function test panel. The ability to measure free cortisol, when translated into clinical practice, should prevent the treatment of many desperately ill patients with normal adrenal function who, because of hypoalbuminemia, meet the criteria for "stress-dose" glucocorticoid therapy. Refraining from treating these patients with glucocorticoids should facilitate their recovery and protect them from side effects, with the end results of better wound healing, healthier skin, improved immune responsiveness, lower blood sugar, and stronger bones.
But another knotty question remains: How do we know that these patients with normal "free cortisol" levels do not have glucocorticoid resistance? If they do, might they not require glucocorticoid supplementation even if they have a normal free cortisol response to synthetic cosyntropin? Might they, if you will, have partial adrenal insufficiency? This argument has been around for a long time. The question may be illuminated, however, by a wonderful experiment of nature: the well-characterized syndrome of glucocorticoid resistance associated with mutations in the glucocorticoid receptor.3 Even in the absence of stress, patients with such mutations have high circulating levels of cortisol, measurable corticosyntropin, and somewhat hypertrophied adrenal glands. They have no signs of glucocorticoid excess. More to the point, there is no evidence that patients with this syndrome fare any worse under stressful conditions than their normal counterparts do. Their hypothalamic–pituitary–adrenal axis adjusts to their resistant status. A patient in the ICU who has plasma free cortisol levels that are appropriate for the clinical situation should have a similar response.
But what are the "appropriate" levels for a given clinical situation? This is mostly unexplored territory. Appropriate data will have to be gathered. Only after such data are obtained can the idea of general or localized glucocorticoid resistance be fruitfully addressed. The necessary experiment is straightforward. Patients should be stratified, perhaps according to the Acute Physiology and Chronic Health Evaluation (APACHE) score, plasma free cortisol measured, and the range, mean, and confidence intervals determined. On the basis of the results, patients with "free cortisol" levels in the lowest 10 percent for the population would be given supplemental glucocorticoids. The remainder would not. Of these patients, those who were not doing as well as expected would be randomly assigned to receive glucocorticoid therapy or placebo. The outcome measures are obvious.
Hospital demographics are changing. The average length of stay is shorter, the level of acuity of illness is higher, and the proportion of hospitalized patients who are in ICUs is steadily increasing. It is clear that many of the laboratory tests developed for use in more stable hospitalized patients may no longer apply. It may be better to consider "normal" to be situational — or even existential. The best care for patients in the ICU will depend on the development of a new normative data base for the critically ill. The article by Hamrahian et al. prods us in this direction. As Hippocrates pointed out, the first task of the physician is to remove from the regimen all things that might impede recovery. This concept holds promise for improving the care of the seriously ill.
Source Information
From the Division of Endocrinology, Diabetes, and Clinical Nutrition, Oregon Health and Sciences University, Portland.
References
Kehlet H. A rational approach to dosage and preparation of parenteral glucocorticoid substitution therapy during surgical procedures: a short review. Acta Anaesthesiol Scand 1975;19:260-264.
Coolens J-L, Van Baelen H, Heyns W. Clinical use of an unbound plasma cortisol as calculated from total cortisol and corticosteroid-binding globulin. J Steroid Biochem 1987;26:197-202.
Chrousos GP, Detera-Wadleigh S, Karl M. Syndromes of glucocorticoid resistance. Ann Intern Med 1993;119:1113-1124.(Lynn Loriaux, M.D.)
Patients with shock are generally treated in medical intensive care units (ICUs), in which the cosyntropin stimulation test is generally used to rule out the diagnosis of adrenal insufficiency. Since many more patients in ICUs appear to meet the biochemical criteria for adrenal compromise than would be expected, essentially all such patients end up being treated with large doses of intravenous glucocorticoids, usually the equivalent of 600 mg of hydrocortisone per day. If adrenal insufficiency is indeed the culprit causing their symptoms, glucocorticoids may be considered lifesaving. If not, glucocorticoid administration adds cost without benefit, for these agents are antianabolic steroids that result in decreased synthesis of new protein, attenuation of immunity, impairment of insulin action, loss of calcium, and other untoward effects. In a bedridden patient, this is a bad combination of effects that can lead to rapid clinical deterioration. It is unsettling that so many patients in ICUs meet the generally accepted biochemical criteria for adrenal compromise and thus receive glucocorticoids. Indeed, one might well ask whether the right test is being used to identify patients who truly require supplemental glucocorticoids in this circumstance.
A potentially important insight into this question is provided by Hamrahian et al. in this issue of the Journal (pages 1629–1638). These investigators report that plasma cortisol responses to synthetic cosyntropin frequently fall below the cutoff value of 20.0 μg per deciliter in critically ill patients who have hypoalbuminemia but not in matched critically ill patients with normal plasma albumin concentrations. They attribute the apparently low response to the fact that cortisol, which circulates primarily bound with high affinity to a specific binding globulin, cortisol-binding globulin, and with low affinity to albumin, is measured as "low" in patients with hypoproteinemic states (see Figure). Thus, alterations in the concentration of cortisol-binding protein would be expected to affect measurements of total plasma cortisol. A dramatic finding in this study, however, was that in patients with a low total cortisol response to a cosyntropin challenge, the circulating "free cortisol" concentration was not significantly different from that in patients with normal albumin concentrations. In other words, it appears that even though the total plasma cortisol response to a cosyntropin challenge is low in such patients, the response of free bioactive cortisol is appropriate for the situation, presumably because the feedback regulation remains intact.
Figure. Circulating Plasma Cortisol Levels in a Normal Person, a Stressed Person, and a Stressed Person with Hypoalbuminemia.
Approximately 10 percent of cortisol is present in the free (bioactive) state, unbound to protein; 20 percent is loosely bound to albumin; and 70 percent is tightly bound to cortisol-binding globulin. Binding to cortisol-binding globulin is nearly saturated at plasma cortisol levels in the range of 15 to 18 μg per deciliter (413.8 to 496.6 nmol per liter). With stress and hypoalbuminemia, the maximal total cortisol measurement will fall below that in persons with a normal albumin concentration and, often, below the traditional cutoff level for the cosyntropin stimulation test, even though the level of free, bioactive cortisol is the same in these two groups and is appropriate for the clinical situation. To convert values to nanomoles per liter, multiply by 27.59.
This finding was anticipated by earlier work that demonstrated the consistency of free cortisol concentrations in situations in which the concentrations of binding globulin vary widely.2 Hence, in critically ill patients in the ICU, the circulating free cortisol response to synthetic cosyntropin is a much more important clinical determination than the total cortisol response. As things stand, plasma free cortisol is not easy to measure, but a few years ago, it was not easy to measure plasma free thyroxine. Now, free thyroxine is the workhorse of the thyroid-function test panel. The ability to measure free cortisol, when translated into clinical practice, should prevent the treatment of many desperately ill patients with normal adrenal function who, because of hypoalbuminemia, meet the criteria for "stress-dose" glucocorticoid therapy. Refraining from treating these patients with glucocorticoids should facilitate their recovery and protect them from side effects, with the end results of better wound healing, healthier skin, improved immune responsiveness, lower blood sugar, and stronger bones.
But another knotty question remains: How do we know that these patients with normal "free cortisol" levels do not have glucocorticoid resistance? If they do, might they not require glucocorticoid supplementation even if they have a normal free cortisol response to synthetic cosyntropin? Might they, if you will, have partial adrenal insufficiency? This argument has been around for a long time. The question may be illuminated, however, by a wonderful experiment of nature: the well-characterized syndrome of glucocorticoid resistance associated with mutations in the glucocorticoid receptor.3 Even in the absence of stress, patients with such mutations have high circulating levels of cortisol, measurable corticosyntropin, and somewhat hypertrophied adrenal glands. They have no signs of glucocorticoid excess. More to the point, there is no evidence that patients with this syndrome fare any worse under stressful conditions than their normal counterparts do. Their hypothalamic–pituitary–adrenal axis adjusts to their resistant status. A patient in the ICU who has plasma free cortisol levels that are appropriate for the clinical situation should have a similar response.
But what are the "appropriate" levels for a given clinical situation? This is mostly unexplored territory. Appropriate data will have to be gathered. Only after such data are obtained can the idea of general or localized glucocorticoid resistance be fruitfully addressed. The necessary experiment is straightforward. Patients should be stratified, perhaps according to the Acute Physiology and Chronic Health Evaluation (APACHE) score, plasma free cortisol measured, and the range, mean, and confidence intervals determined. On the basis of the results, patients with "free cortisol" levels in the lowest 10 percent for the population would be given supplemental glucocorticoids. The remainder would not. Of these patients, those who were not doing as well as expected would be randomly assigned to receive glucocorticoid therapy or placebo. The outcome measures are obvious.
Hospital demographics are changing. The average length of stay is shorter, the level of acuity of illness is higher, and the proportion of hospitalized patients who are in ICUs is steadily increasing. It is clear that many of the laboratory tests developed for use in more stable hospitalized patients may no longer apply. It may be better to consider "normal" to be situational — or even existential. The best care for patients in the ICU will depend on the development of a new normative data base for the critically ill. The article by Hamrahian et al. prods us in this direction. As Hippocrates pointed out, the first task of the physician is to remove from the regimen all things that might impede recovery. This concept holds promise for improving the care of the seriously ill.
Source Information
From the Division of Endocrinology, Diabetes, and Clinical Nutrition, Oregon Health and Sciences University, Portland.
References
Kehlet H. A rational approach to dosage and preparation of parenteral glucocorticoid substitution therapy during surgical procedures: a short review. Acta Anaesthesiol Scand 1975;19:260-264.
Coolens J-L, Van Baelen H, Heyns W. Clinical use of an unbound plasma cortisol as calculated from total cortisol and corticosteroid-binding globulin. J Steroid Biochem 1987;26:197-202.
Chrousos GP, Detera-Wadleigh S, Karl M. Syndromes of glucocorticoid resistance. Ann Intern Med 1993;119:1113-1124.(Lynn Loriaux, M.D.)