sign in sign up
<<
Home , Adrenalectomy, Adrenal insufficiency

Adrenal Insufficiency in High-Risk Surgical ICU Patients*

Emanuel P. Rivers, MD, MPH, FCCP; Mario Gaspari, MD; George Abi Saad, MD; Mark Mlynarek, PharmD; John Fath, MD; H. Matilda Horst, MD; and Jacobo Wortsman, MD

Study objectives: To examine the incidence and response to treatment of adrenal insufficiency (AI) in high-risk postoperative patients. Design: Prospective observational case series. Setting: Large urban tertiary-care surgical ICU (SICU). Participants: Adults > 55 years of age who required vasopressor therapy after adequate volume resuscitation in the immediate postoperative period. Interventions: Each patient underwent a cosyntropin (ACTH) stimulation test; at the discretion of the clinical team, some patients were empirically given (100 mg IV q8h for three doses) before serum values became available. Measurements: Adrenal dysfunction (AD), defined as serum cortisol < 20 ␮g/dL at all time points, with ⌬cortisol (60 min post-ACTH minus baseline) of < 9 ␮g/dL; functional hypoadrenalism (FH), defined as serum cortisol < 30 ␮g/dL at all time points or ⌬cortisol (60 min post-ACTH minus baseline) < 9 ␮g/dL; and AI, as the presence of either AD or FH. Results: One hundred four patients were enrolled with a mean age (SD) of 65.2 ؎ 16.9 years. AI (AD plus FH) was found in 34 of 104 patients (32.7%): AD was found in 9 patients (8.7%), FH in 25 patients (24%), and normal adrenal function in 70 patients (67.3%). The absolute eosinophil count was significantly higher in the combined AD and FH groups compared with the group with normal adrenal function (p < 0.05). Forty-six of 104 patients (44.2%) received hydrocortisone; 29 (63%) could be weaned from treatment with vasopressors within 24 h. This beneficial effect of hydrocortisone reached statistical significance in the FH group when compared with untreated Mortality was also .(0.083 ؍ patients (p < 0.031); a similar trend was seen in the AD group (p lower in the hydrocortisone-treated AI patients (5 of 23 [21%] vs 5 of 11 [45%] in those not receiving hydrocortisone; p < 0.01). Conclusion: There is a high incidence of AI among SICU patients > 55 years of age with postoperative requiring vasopressors. There is also a significant association between hydrocortisone replacement therapy, resolution of vasopressor requirements, and improved survival. (CHEST 2001; 119:889–896) Key words: adrenal disorders; adrenal dysfunction; ; adrenal insufficiency; critically ill; ; functional hypoadrenalism; postoperative period; ; septic ; surgical patient; systemic inflammatory response syndrome Abbreviations: ACTH ϭ corticotropin; AD ϭ adrenal dysfunction; AI ϭ adrenal insufficiency; FH ϭ functional hypoadren- alism; SICU ϭ surgical ICU

drenal insufficiency (AI) is overall rare, with an phase represents a significant risk factor for AI. This A incidence of Ͻ 0.01% in the general population.1 is thought to result from marked stimulation of the However, up to 28% of seriously ill patients are often neurohumoral hypothalamic-pituitary-adrenal axis, found to have occult or unrecognized AI.2–4 It has which may fail to respond to the combination of been hypothesized that the postoperative surgical underlying disease, surgical procedure, and postop- erative homeostatic adaptation. *From the Departments of (Drs. Rivers, Gaspari, and In the setting of critical illness, the failure of an Horst) and Pharmacy (Dr. Mlynarek), Henry Ford Hospital, Case Western Reserve University, Detroit, MI; Wayne State University appropriate neurohumoral response with insufficient (Dr. Fath), Grace Hospital, Detroit, MI; Department of Medicine cortisol release can lead to the clinical picture of (Dr. Wortsman), Southern Illinois University, Springfield, IL; and American University of Beirut (Dr. Saad), Beirut, Lebanon. vasopressor-dependent refractory hypotension. This Manuscript received July 1, 1999; revision accepted August is characterized by elevated cardiac output and 14, 2000. decreased systemic vascular resistance.3–5 An addi- Correspondence to: Emanuel P. Rivers, MD, MPH, FCCP, Henry Ford Hospital, Department of Surgery, 2799 W. Grand Blvd, tional risk factor for AI in the surgical ICU (SICU) is Detroit, MI 48202; e-mail: [email protected] represented by age, inasmuch as the incidence

CHEST / 119/3/MARCH, 2001 889 among patients Ͼ 55 years old is 2.5 times that of formed in all patients.9,10 Immediately after collection of baseline their younger counterpart.6 serum cortisol, ACTH, 0.25 mg, was administered as an IV bolus for 2 min. Blood for serum cortisol determination was obtained again at It is possible that the outcome of postoperative 30 and 60 min after ACTH injection. Blood was collected in sterile patients could be improved if the effect of a single siliconized glass tubes containing ethylenediaminetetra-acetic acid factor, namely AI, were isolated for evaluation. To and sent to the immunoassay ligand laboratory for processing. Sera that end, this study examined the incidence and were separated and frozen at Ϫ20°C until assayed. Cortisol was outcome of diagnosis and treatment of AI in a determined with a commercially available chemiluminescent immu- high-risk population of postoperative vasopressor- noassay kit (Access Immunoassay System; Sanofi Diagnostics Pas- Ͼ teur, Inc; Chaska, MN). In normal subjects, serum cortisol concen- dependent patients with age 55 years. trations range from 5 to 20 ␮g/dL (138 to 552 nmol/L), being higher at 7 am to 9 am. The response to ACTH stimulation in nonstressed healthy subjects is an increase of Ն 50% in serum cortisol concen- Ն ␮ Materials and Methods tration or a rise of 9 g/dL by 60 min after ACTH injection. Patients were treated at the discretion of the clinical management team before cortisol measurements were available with hydrocorti- Study Design sone (100 mg IV q8h for three doses) administered immediately after the ACTH test. This study was approved by the Henry Ford Health Systems Institutional Review Board for Human Research. This was a prospective, nonoutcome, observational convenience case study Outcome Measurements of adult patients in a large, urban, tertiary-care SICU during a In critically ill patients, adrenal dysfunction (AD) has been 2-year period (from 1995 to 1997). defined as the presence of random serum cortisol Ͻ 20 ␮g/dL (Ͻ 552 nmol/L).4,6,9–13 An additional category, functional hypo- Patient Selection adrenalism (FH) has been defined as the combination of random serum cortisol Ն 20 ␮g/dL, and a serum cortisol level at 60 min Ͼ Consecutive postoperative patients 55 years of age who after ACTH stimulation of Ͻ 30 ␮g/dL or ⌬cortisol (60-min experienced hypotension requiring vasopressor therapy after concentration minus baseline) of Յ 9 ␮g/dL.14–19 In the present adequate volume resuscitation within 24 h of SICU admission investigation, AD was defined as the finding of serum cortisol were enrolled in this study. All patients initially underwent a fluid Ͻ 20 ␮g/dL in any of the blood samples (before and after ACTH) challenge that consisted of a 500-mL bolus infusion of crystalloid plus a ⌬cortisol after ACTH of Յ 9 ␮g/dL. FH was defined by a solution given IV for 5 min. Cardiac output and pulmonary cortisol level Ͻ 30 ␮g/dL in any of the blood samples (before and capillary wedge pressure were measured before and at 5 min and after ACTH) or ⌬cortisol Յ 9 ␮g/dL. AI was defined as the Ͼ 10 min after fluid challenge. A rise in cardiac output of 20% presence of either AD or FH. A positive hemodynamic response Ͼ was taken to indicate , a fall 20% indicated was defined as cessation of the need for vasopressor therapy to Ͻ hypervolemia, and changes of 20% indicated euvolemia. Pa- maintain a mean arterial pressure Ͼ 65 to 70 mm Hg within 24 h tients were considered vasopressor dependent if these agents of the first hydrocortisone dose or within 24 h of the ACTH were required to maintain the mean arterial pressure at levels stimulation test in patients not treated with hydrocortisone. Ͼ 65 to 70 mm Hg. The vasopressor agents and corresponding dose ranges were norepinephrine (3 to 40 ␮g/min), phenyleph- Statistical Methods rine (30 to 300 ␮g/min), epinephrine (2 to 100 ␮g/min), or ␮ dopamine (6 to 30 g/kg/min). Criteria for exclusion from the Continuous variables were compared using a Student’s two- study were HIV , known preexisting adrenal disease or sample t test unless variances were unequal; in the latter case, adrenalectomy, administration of etomidate, administration of Welch’s two-sample t test was used. Categorical variables were during surgery, or administration of steroids within the 3 analyzed with the ␹2 test. Cell counts were compared with the months previous to admission.7,8 Fisher’s Exact Test. Analysis of variance was performed with Tukey’s method of multiple comparisons. The Kruskal-Wallis test Study Protocol and the Wilcoxon nonparametric test were used to identify differences in the ranks of data among the three groups studied. A corticotropin (ACTH or cosyntropin) stimulation test (Cor- Statistical significance was defined as a p Ͻ 0.05 and an ␣ of trosyn; Organon Pharmaceuticals; West Orange, NJ) was per- 0.017. All results are presented as the mean Ϯ SD.

(104 ؍ Table 1—Surgical Diagnosis in Hypotensive SICU Patients (n

Emergent Operations Elective Operations Neurosurgical Operations Perforated viscus 23 Nephrectomy 2 Subarachnoid hemorrhage 6 Trauma 18 Whipple 2 Ischemic bowel 15 Resection laryngeal tumor 1 GI 12 1 Necrotizing infection 8 Gangrenous cholecystitis 5 Intra-abdominal abscess 5 Vascular emergency 4 Toxic megacolon 2

890 Clinical Investigations in Critical Care Table 2—Serum Cortisol in Hypotensive SICU Patients and Subgroups by Cortisol Levels and Response to ACTH

Normal Variables All Patients AD FH Response Patients, No. 104 9 25 70 % of total 100 8.7 24 67.3 Baseline 29.9 Ϯ 24.4* 11.2 Ϯ 6.9 16.3 Ϯ 5.2 37.2 Ϯ 26.6 30 min† 40.9 Ϯ 28.9 14.9 Ϯ 6.2 22.6 Ϯ 2.4 50.3 Ϯ 30.6 60 min† 46.3 Ϯ 32.8 14.7 Ϯ 6.2 24.2 Ϯ 2.9 57.8 Ϯ 33.9 *Values are mean Ϯ SD in micrograms per deciliter. †Time after injection of ACTH, 0.25 mg IV bolus.

Results groups, and the number of lymphocytes did not differ among groups. However, the number of eo- A total of 104 patients were enrolled in the study; the mean age was 65.2 Ϯ 16.9 years, and surgical sinophils, relative and absolute, was significantly diagnoses are listed in Table 1. Adrenal function was higher in the AD and FH groups compared with abnormal in 34 of 104 of all patients (32.7%); 9 normal ACTH responders (Table 4). patients (8.7%) had AD and 25 (24%) had FH (Table Forty-six of 104 patients (44.2%) received hydro- 2). All patients enrolled fulfilled criteria for the at a dose representing physiologic replace- systemic inflammatory response syndrome and were ment in this setting. Twenty-nine of the treated classified as having severe sepsis or .20 patients (63%) could be weaned off of vasopressor Seventy patients (67.3%) exhibited normal baseline therapy within 24 h of the first hydrocortisone dose. serum cortisol (Ͼ 20 ␮g/dL) and response to the Comparison of the effect of hydrocortisone within ACTH stimulation test. There were no significant each group shows that in the FH group, hydrocorti- differences between groups for any of the following sone therapy was associated with significantly higher variables: age, sex, temperature, heart rate, BP, rate of success in the withdrawal of vasopressor central venous pressure, pulmonary capillary wedge therapy (p Ͻ 0.031; Table 5). A similar trend was pressure, and cardiac index (Table 3). The amount of noted in the AD group, but because of the small fluid given before the administration of vasopressor number of patients, the difference did not reach therapy was similar in all groups. statistical significance (p ϭ 0.083; Table 5). Of note, Laboratory data also showed lack of significant among the hydrocortisone-treated normal ACTH differences among groups in sodium, potassium, responders, those who could be weaned from vaso- chloride, glucose, calcium, magnesium, and phos- pressor therapy within 24 h had significantly lower phorous (Table 4). Leukocytosis with a left shift baseline serum cortisol levels as compared with (increased bands) was represented equally in all those who continued requiring vasopressor agents

Table 3—Demographics, Vital Signs, and Hemodynamic Data of Hypotensive SICU Patients*

Normal Variables All Patients AD FH Response Patients, No. 104 9 25 70 Age, yr 65.2 Ϯ 16.9 66.1 Ϯ 9.8 60.4 Ϯ 21.1 66.8 Ϯ 16.8 Sex, No. (%) Male 63 (61) 6 (66) 15 (60) 42 (60) Female 41 (39) 3 (33) 10 (40) 28 (40) Temperature, °C 37.4 Ϯ 0.9 37.8 Ϯ 0.8 37.5 Ϯ 0.9 37.5 Ϯ 1.0 Heart rate, beats/min 103.0 Ϯ 18.9 106.2 Ϯ 20.9 107.0 Ϯ 3 101.2 Ϯ 17.3 Systolic BP, mm Hg 117.8 Ϯ 22.2 116.2 Ϯ 22.2 114.7 Ϯ 16.8 119.1 Ϯ 23.9 Diastolic BP, mm Hg 60.6 Ϯ 13.2 68.9 Ϯ 11.3 68.6 Ϯ 13.5 61.6 Ϯ 13.4 Mean arterial pressure, mm Hg 78.3 Ϯ 16.2 76.9 Ϯ 12.4 76.7 Ϯ 13.9 79.1 Ϯ 17.6 Fluid volume for resuscitation, L 13.3 Ϯ 8.6 13.0 Ϯ 6.1 13.5 Ϯ 7.0 13.3 Ϯ 9.5 Fluid challenge response Euvolemic Euvolemic Euvolemic Euvolemic Central venous pressure, mm Hg 12.5 Ϯ 5.2 11.9 Ϯ 4.9 11.7 Ϯ 5.11 12.81 Ϯ 5.3 PCWP, mm Hg 17.1 Ϯ 6.2 16.8 Ϯ 6.6 17.0 Ϯ 7.5 17.2 Ϯ 6.02 Cardiac index, L/min/m2 3.6 Ϯ 1.2 3.7 Ϯ 0.8 3.9 Ϯ 1.2 3.6 Ϯ 1.27 *Values are expressed as mean Ϯ SD unless otherwise indicated; PCWP ϭ pulmonary capillary wedge pressure.

CHEST / 119/3/MARCH, 2001 891 Table 4—Blood Chemistry and Hematologic Data of Hypotensive SICU Patients*

Normal Variables All Patients AD FH Response Patients, No. (%) 104 9 (8.7) 25 (24.0) 70 (67.3) Sodium, mmol/L 138 Ϯ 6.0 140 Ϯ 5.8 138 Ϯ 5.6 138 Ϯ 6.2 Potassium, mmol/L 4.3 Ϯ 0.7 4.4 Ϯ 0.6 4.4 Ϯ 0.7 4.3 Ϯ 0.6 Chloride, mmol/L 106 Ϯ 8.0 108 Ϯ 8.2 107 Ϯ 7.1 105 Ϯ 8.2 Bicarbonate, mmol/L 22 Ϯ 4.8 21 Ϯ 6.1 22 Ϯ 4.4 22 Ϯ 4.8 Blood urea nitrogen, mg/dL 37 Ϯ 28.8 30 Ϯ 17.6 36 Ϯ 24.3 39 Ϯ 31.4 Creatinine, mg/dL 1.9 Ϯ 1.7 1.5 Ϯ 0.9 1.9 Ϯ 1.5 1.9 Ϯ 1.8 Glucose, mg/dL 168 Ϯ 83 212 Ϯ 98 147 Ϯ 83 172 Ϯ 79 Calcium, mg/dL 7.6 Ϯ 0.8 7.7 Ϯ 0.7 7.6 Ϯ 0.8 7.6 Ϯ 0.8 Magnesium, mg/dL 1.7 Ϯ 0.3 1.7 Ϯ 0.3 1.6 Ϯ 0.6 1.7 Ϯ 0.4 Phosphorous, mg/dL 4.0 Ϯ 1.8 3.9 Ϯ 1.9 4.1 Ϯ 0.9 4 Ϯ 2.0 WBC count, 103/␮L 13.7 Ϯ 6.9 16.3 Ϯ 5.4 13.7 Ϯ 7.5 13.3 Ϯ 6.9 Neutrophils, % 66 Ϯ 24.0 60 Ϯ 24.3 60 Ϯ 26.3 69 Ϯ 23.2 Bands, % 25.8 Ϯ 22.3 22 Ϯ 26.8 26 Ϯ 21.8 26 Ϯ 22.1 Lymphocytes, % 7.6 Ϯ 7.1 6 Ϯ 5.1 8 Ϯ 7.3 7 Ϯ 7.4 Eosinophils, % 1.5 Ϯ 2.8 3.5 Ϯ 6.4† 2.5 Ϯ 2.5† 0.9 Ϯ 1.4 Absolute eosinophils, 103/␮L 222 Ϯ 504 635 Ϯ 1350† 308 Ϯ 376† 123 Ϯ 202 *Values are expressed as mean Ϯ SD unless otherwise indicated. †Significant difference (p Ͻ 0.05) compared with normal response group.

(p Ͻ 0.001; Table 6), although the post-ACTH 30- release of corticotropin-releasing factor from the min and 60-min serum cortisol levels were similar induces an increase in ACTH secre- (Table 6). tion by the anterior , overriding the Overall mortality in the entire patient population was normal diurnal pattern of ACTH and cortisol secretion. 40% (42 of 104 patients). Among the patients with AI, The responds to ACTH by increasing the mortality rate was 29% (10 of 34). Mortality was cortisol secretion, but prolonged elevation of serum significantly lower in the hydrocortisone-treated group, cortisol triggers a negative feedback inhibition loop that 5 of 23 patients (21%), than in the untreated group, 5 results in subsequent decreases in ACTH and cortisol Ͻ of 11 patients (45%; p 0.01). release.23 Previous studies have defined criteria for normal adrenal function at rest, as well as in the stressed state, allowing the identification of patients at Discussion risk for the development of .24,25 When Cortisol is a major response hormone that conducted on ambulatory healthy subjects, such studies has metabolic, catabolic, anti-inflammatory, and va- have concluded that a normal response is an ACTH- soactive properties on cardiac muscle and the pe- stimulated serum cortisol level Ն 20 ␮g/dL.14,24,26,27 ripheral vasculature. Thus, cortisol mediates mainte- Patients with normal hypothalamic-pituitary-adrenal nance of peripheral vasomotor tone by facilitating axis function are found to consistently have elevated catecholamine-induced vasoconstriction and has a circulating cortisol concentrations during periods of permissive effect on the synthesis of catecholamines stress or serious illness.28–31 Total serum cortisol is and vasoactive peptides.21,22 Cortisol also has inotro- increased, from 2 to 10 times the upper limit of normal, pic effects and modulates free water distribution and there is a loss of diurnal variation.32–35 The failure within the vascular compartment.5 In response to to reach these levels in the stressed state has lead to the external or internal stress, the neuronally stimulated diagnosis of AI.

Table 5—Response to Vasopressor Therapy Withdrawal in Hypotensive SICU Patients*

Group Patients, No. HC Treated, %† Untreated, % p Value AD 9 6/7 (86) 0/2 (0) 0.083 FH 25 13/16 (81) 3/9 (33) 0.031 Normal response 70 10/23 (43.5) 16/47 (33.3) 0.057 *Values are expressed as No./total (%) unless otherwise indicated; a positive response indicates not vasopressor dependent within 24 h. †Received hydrocortisone (HC), 100 mg IV bolus q8h for three doses.

892 Clinical Investigations in Critical Care Table 6—Serum Cortisol Concentrations in screening of all ICU patients becomes impracti- Hydrocortisone-Treated SICU Patients With a Normal cal.34,43 Nevertheless, selected groups, such as post- Response to ACTH operative surgical patients, may be at significantly Serum Cortisol, ␮g/dL higher risk for AI because of the combined burden represented by the initial pathologic insult, surgical Patient No. Baseline 30 min 60 min procedure, and course in the postoperative ICU. Age Vasopressor-dependent has been shown to be an additional risk factor for AI, 1 132 124 130 inasmuch as patients Ͼ 55 years old have a threefold 2 32 41 180 3 57 106 126 increase in the risk for AI in the SICU compared 4324248with younger patients.6,44 Moreover, the clinical 5504246picture of vasopressor-dependent refractory hypo- 6304240tension with elevated cardiac output and decreased 7 66 86 148 systemic vascular resistance that is frequently ob- 8364036 3,5 9214046served in the SICU has also been reported in AI. 10 89 98 110 Thus, the risk factors for the postoperative period of 11 21 29 30 advanced age and persistent hypotension after ade- 12 29 44 43 quate volume resuscitation, provide the rationale for 13 38 54 57 defining the patients included in this study as being Mean Ϯ SD 49 Ϯ 32* 61 Ϯ 31 80 Ϯ 51 Weaned from vasopressors at high risk for development of AD. 14 18 30 31 The pathogenesis of AI in the high-risk patient is 15 18 30 28 complex, and concepts on this area continue to 16 27 35 33 evolve. In the absence of ACTH levels to confirm the 17 10 36 49 diagnosis, the inability of the AD group to increase 18 21 94 0 19 24 37 44 cortisol levels after ACTH suggests primary adrenal 20 20 34 60 insufficiency. Likewise, the partial response in the 21 20 33 34 FH group suggests secondary AI. Regarding the 22 22 29 31 patients who had a normal response to ACTH and 23 24 168 105 who received hydrocortisone with positive hemody- Mean Ϯ SD 20 Ϯ 4* 53 Ϯ 45 46 Ϯ 24 namic response, the initial baseline cortisol levels Ͻ *Significant difference (p 0.01) were actually low for a stressed state. The ability of patients in this latter group to increase relatively low baseline concentrations (20 Ϯ 4 ␮g/dL) to levels In 1855 Thomas Addison first described a syn- consistent with a normal response to ACTH also drome of “languor, debility and remarkable feeble- supports a secondary adrenal disorder. ness of the heart” caused by “failure of the suprare- This syndrome of “transient ACTH deficiency in nal glands” that is now recognized as AI.36 It can be critical illness or secondary AI” has been previously either primary (failure of adrenal gland) or secondary described.2,3 Patients in that category exhibit re- (failure of hypothalamic or pituitary stimulation of sponses to ACTH that are remarkably similar to the adrenal gland) and is defined as a relative or those of the current study patients, with similar absolute deficiency in and mineralo- restoration of hemodynamic stability after replace- corticoid availability.7,37,38 The classic signs and ment therapy with hydrocortisone.2,3 The mecha- symptoms of AI are hypotension, , nism for the production of this postoperative disor- hyperkalemia, hypercalcemia, , metabolic der is complex. The systemic inflammatory response acidosis, and eosinophilia.30 Such classic findings were syndrome, which frequently accompanies the post- not a distinguishing feature in this study, perhaps operative course, results in the release of cytokines because of the combination of underlying disease and such as tumor necrosis factor-␣,45 which in vitro prior therapeutic intervention that obscured the clinical suppresses the pituitary response to the hypotha- picture. However, although the clinical diagnosis of AI lamic corticotropin-releasing hormone46 and the re- could not be entertained in this population, the study lease of cortisol from ACTH-stimulated adrenal did, nevertheless, confirm previous findings of eosino- cells.47,48 Because the systemic inflammatory re- philia as a marker of AI.39,40 sponse is frequently transient in the postsurgical The incidence of AI is low in the general popula- phase,49 it would explain at least partly the temporary tion (Ͻ 0.01%),1,41,42 but as occult or unrecognized nature of the adrenal disorders in this patient popu- AI it has higher prevalence among seriously ill lation.2,44 patients (0.1 to 28%).2–4,13,28 Because of the wide Interestingly, 10 of 23 empirically treated patients range of prevalence in the latter population, routine with a normal baseline cortisol and ACTH test

CHEST / 119/3/MARCH, 2001 893 showed a positive hemodynamic response. The of this study, post hoc analysis of mortality data mechanism for this beneficial therapeutic response is suggests survival benefit in diagnosing and treating unclear, but could represent a masked form of AI this disorder. That impression notwithstanding, it from diagnostic limitations. The total measured must be stated again that this study was not designed plasma cortisol represents the sum of cortisol bound with mortality as a primary outcome, and this post to cortisol-binding globulin and albumin, and free hoc analysis reflects only crude mortality data with cortisol. It is the latter fraction that is physiologically no determination of attributable causes. Future stud- active, and free cortisol levels are significantly af- ies are needed to determine survival benefits of fected by changes in cortisol-binding globulin and treatment. Secondly, as plasma ACTH levels were albumin. The concentration of these proteins could not obtained, we cannot clearly delineate whether fluctuate as a result of endogenous and exogenous our findings represent a primary or secondary AI. hormones, organ dysfunction (such as disease), or changes in the volume of distribution secondary to fluid resuscitation. These alterations, which fre- quently accompany the postoperative surgical phase, Conclusion would explain, at least partly, why some patients may Although the incidence of AI is higher than nor- respond to cortisol replacement in spite of having mal in critically ill patients, the incidence in surgical 5 appropriate total cortisol levels. In addition, a de- ICU patients is even higher when restricting the ranged interaction between catecholamines, adren- evaluation to patients with the risk factors of age ergic receptors, and would lead to Ͼ 55 years and postoperative hypotension requiring adrenergic hyporesponsiveness. This desensitization ␣ ␤ vasopressors after adequate volume resuscitation. In or down-regulation of - and -adrenergic receptors that setting, the laboratory examination and hemo- may result in vascular hyporesponsiveness and myo- dynamic profile do not assist in the diagnosis of AI, 50,51 cardial depression, at appropriate total cortisol except for the presence of eosinophilia. Administra- levels. Such conditions would reverse with physio- tion of hydrocortisone replacement therapy in those 52,53 logic doses of hydrocortisone. patients is significantly associated with resolution of There are few outcome studies regarding the vasopressor dependency within 24 h; hydrocortisone effect of glucocorticoid replacement therapy in pa- treatment was also associated with a trend toward tients with AI in this setting, although it is known improved survival among patients with AI. that the disorder may have life-threatening conse- quences.38 In the present study, the overall mortality ACKNOWLEDGMENT: The authors thank Julie Massura, MS, rate of abnormal responders was significantly lower and Gary Chase, PhD, for their biostatistical expertise and among those receiving hydrocortisone (21%) than in Alexandria Muzzin and Julie Ressler for their contributions to the untreated patients (45%). A number of case series article. studies have described hemodynamic improvement after cortisol replacement therapy that followed di- 3,11,28,54 55 agnosis of AI. Bollaert et al showed reversal References of severe late septic shock (improved hemodynam- 1 Mason AS, Meade TW, Lee JA, et al. Epidemiological and ics) and a beneficial effect on mortality with hydro- clinical picture of Addison’s disease. Lancet 1968; 2:744–747 cortisone therapy. As confirmed in this study, an 2 Kidess AI, Caplan RH, Reynertson RH, et al. Transient important therapeutic benefit of early diagnosis and corticotropin deficiency in critical illness. Mayo Clin Proc treatment is the potential to decrease the need for 1993; 68:435–441 administering high doses of vasopressor therapy with 3 Baldwin WA, Allo M. Occult hypoadrenalism in critically ill 56 patients. Arch Surg 1993; 128:673–676 its possible negative consequences. 4 Jurney TH, Cockrell JL Jr, Lindberg JS, et al. Spectrum of serum cortisol response to ACTH in ICU patients: correlation with degree of illness and mortality. Chest 1987; 92:292–295 Limitations 5 Dorin RI, Kearns PJ. High output circulatory failure in acute The present study provides evidence for the exis- adrenal insufficiency. Crit Care Med 1988; 16:296–297 6 Barquist E, Kirton O. Adrenal insufficiency in the surgical tence of glucocorticoid-responsive shock states in the intensive care unit patient. J Trauma 1997; 42:27–31 severely ill ICU patient, and it also confirms that the 7 Berstein W, Aduen J, Chernow B. Adrenal disease in the standard criteria for abnormal adrenal response may critically ill patient. Crit Care Med 1995:1127–1137 not be appropriate in this population; however, 8 Harrison BD, Rees LH, Cayton RM, et al. Recovery of questions still remain unanswered. The first, and hypothalamo-pituitary-adrenal function in asthmatics whose oral steroids have been stopped or reduced. Clin Endocrinol most clinically relevant, is whether diagnosis and 1982; 17:109–118 treatment of these disorders has an effect on mor- 9 Oelkers W. Dose-response aspects in the clinical assessment tality. Although mortality was not a primary outcome of the hypothalamo-pituitary-adrenal axis, and the low-dose

894 Clinical Investigations in Critical Care adrenocorticotropin test. Eur J Endocrinol 1996; 135:27–33 asone and corticotropin-releasing hormone. J Clin Endocrinol 10 Oelkers W. Adrenal insufficiency. N Engl J Med 1996; Metab 1993; 77:151–156 335:1206–1212 32 Harris MJ, Baker RT, McRoberts JW, et al. The adrenal 11 Claussen MS, Landercasper J, Cogbill TH. Acute adrenal response to trauma, operation and cosyntropin stimulation. insufficiency presenting as shock after trauma and surgery: Surg Gynecol Obstet 1990; 170:513–516 three cases and review of the literature. J Trauma 1992; 33 Jarek MJ, Legare EJ, McDermott MT, et al. Endocrine 32:94–100 profiles for outcome prediction from the intensive care unit. 12 McKee JI, Finlay WE. Cortisol replacement in severely Crit Care Med 1993; 21:543–550 stressed patients [letter]. Lancet 1983; 1:484 34 Drucker D, Shandling M. Variable adrenocortical function in 13 Soni A, Pepper GM, Wyrwinski PM, et al. Adrenal insuffi- acute medical illness. Crit Care Med 1985; 13:477–479 ciency occurring during septic shock: incidence, outcome, 35 Lamberts SW, Bruining HA, de Jong FH. and relationship to peripheral cytokine levels. Am J Med therapy in severe illness. N Engl J Med 1997; 337:1285–1292 36 Addison T. On the constitutional and local effects of diseases 1995; 98:266–271 of the supra-renal capsules. Med Classics 1938; 2:244–280 14 Grinspoon SK, Biller BM. Clinical review 62: laboratory 37 Vesely D. Recognizing and managing acute adrenal insuffi- assessment of adrenal insufficiency. J Clin Endocrinol Metab ciency. J Crit Illness 1988; 3:101–111 1994; 79:923–931 38 Chin R. Adrenal crisis. Crit Care Clin 1991; 7:23–42 15 Dluhy RG, Himathongkam T, Greenfield M. Rapid ACTH 39 Hudson G. The marrow reserve of eosinophils. Br J Haematol test with plasma levels: improved diagnostic 1964; 10:122–130 discrimination. Ann Intern Med 1974; 80:693–696 40 Angelis M, Yu M, Takanishi D, et al. Eosinophilia as a marker 16 Moran JL, Chapman MJ, O’Fathartaigh MS, et al. Hypocor- of adrenal insufficiency in the surgical intensive care unit. tisolaemia and adrenocortical responsiveness at onset of J Am Coll Surg 1996; 183:589–596 septic shock. Intensive Care Med 1994; 20:489–495 41 Rusnak RA. Adrenal and pituitary emergencies. Emerg Med 17 Rothwell PM, Udwadia ZF, Jackson EA, et al. Plasma cortisol Clin North Am 1989; 7:903–925 levels in patients with septic shock. Crit Care Med 1991; 42 Quiney NF, Durkin MA. Adrenocortical failure in intensive 19:589–590 care. BMJ 1995; 310:1253–1254 18 Rothwell PM, Udwadia ZF, Lawler PG. Cortisol response to 43 Span LF, Hermus AR, Bartelink AK, et al. Adrenocortical corticotropin and survival in septic shock. Lancet 1991; function: an indicator of severity of disease and survival in 337:582–583 chronic critically ill patients. Intensive Care Med 1992; 19 Annane D, Sebille V, Troche G, et al. A 3-level prognostic 18:93–96 classification in septic shock based on cortisol levels and 44 Merry WH, Caplan RH, Wickus GG, et al. Postoperative cortisol response to corticotropin. JAMA 2000; 283:1038– acute adrenal failure caused by transient corticotropin defi- 1045 ciency. Surgery 1994; 116:1095–1100 20 American College of Chest Physicians/Society of Critical Care 45 Pinsky MR, Vincent JL, Deviere J, et al. Serum cytokine Medicine Consensus Conference. Definitions for sepsis and levels in human septic shock: relation to multiple-system organ failure and guidelines for the use of innovative thera- organ failure and mortality. Chest 1993; 103:565–575 pies in sepsis. Crit Care Med 1992; 20:864–874 46 Gaillard RC, Turnill D, Sappino P, et al. Tumor necrosis 21 Bernton EW, Long JB, Holaday JW. Opioids and neuropep- factor alpha inhibits the hormonal response of the pituitary tides: mechanisms in circulatory shock. Fed Proc 1985; gland to hypothalamic releasing factors. 1990; 44:290–299 127:101–106 22 Svedmyr N. Action of corticosteroids on ␤-adrenergic recep- 47 Jaattela M, Ilvesmaki V, Voutilainen R, et al. Tumor necrosis tors: clinical aspects. Am Rev Respir Dis 1990; 141(suppl): factor as a potent inhibitor of adrenocorticotropin-induced S31–S38 cortisol production and steroidogenic P450 enzyme gene 23 Burke CW. Adrenocortical insufficiency. Clin Endocrinol expression in cultured human fetal adrenal cells. Endocrinol- Metab 1985; 14:947–976 ogy 1991; 128:623–629 24 Mohler JL, Flueck JA, McRoberts JW. Adrenal insufficiency 48 Spangelo BL, Judd AM, Call GB, et al. Role of the cytokines following unilateral adrenalectomy: a case report. J Urol 1986; in the hypothalamic-pituitary-adrenal and gonadal axes. Neu- 135:554–546 roimmunomodulation 1995; 2:299–312 25 Hagg E, Asplund K, Lithner F. Value of basal plasma cortisol 49 Talmor M, Hydo L, Barie PS. Relationship of systemic assays in the assessment of pituitary-adrenal insufficiency. inflammatory response syndrome to organ dysfunction, Clin Endocrinol 1987; 26:221–226 length of stay, and mortality in critical surgical illness: effect 26 Knowlton A. Adrenal insufficiency in the intensive care of intensive care unit resuscitation. Arch Surg 1999; 134: setting. J Intensive Care Med 1989; 4:1127–1137 81–87 27 Speckart PF, Nicoloff JT, Bethune JE. Screening for adreno- 50 Silverman HJ, Penaranda R, Orens JB, et al. Impaired cortical insufficiency with cosyntropin (synthetic ACTH). ␤-adrenergic receptor stimulation of cyclic adenosine mono- Arch Intern Med 1971; 128:761–763 phosphate in human septic shock: association with myocardial 28 Sibbald WJ, Short A, Cohen MP, et al. Variations in adreno- hyporesponsiveness to catecholamines. Crit Care Med 1993; cortical responsiveness during severe bacterial : 21:31–39 unrecognized adrenocortical insufficiency in severe bacterial 51 McMillan M, Chernow B, Roth BL. Hepatic ␣ 1-adrenergic infections. Ann Surg 1977; 186:29–33 receptor alteration in a rat model of chronic sepsis. Circ 29 Melby JC. Drug spotlight program: systemic corticosteroid Shock 1986; 19:185–193 therapy: pharmacology and endocrinologic considerations. 52 Walker BR, Williams BC. Corticosteroids and vascular tone: Ann Intern Med 1974; 81:505–512 mapping the messenger maze [editorial]. Clin Sci 1992; 30 Leshin M. Acute adrenal insufficiency: recognition, manage- 82:597–605 ment, and prevention. Urol Clin North Am 1982; 9:229–235 53 Saito T, Takanashi M, Gallagher E, et al. Corticosteroid effect 31 Reincke M, Allolio B, Wurth G, et al. The hypothalamic- on early ␤-adrenergic down-regulation during circulatory pituitary-adrenal axis in critical illness: response to dexameth- shock: hemodynamic study and ␤-adrenergic receptor assay.

CHEST / 119/3/MARCH, 2001 895 Intensive Care Med 1995; 21:204–210 septic shock with supraphysiologic doses of hydrocortisone. 54 Alford WC Jr, Meador CK, Mihalevich J, et al. Acute adrenal Crit Care Med 1998; 26:645–650 insufficiency following cardiac surgical procedures. J Thorac 56 Meier-Hellmann A, Reinhart K. Effects of catecholamines on Cardiovasc Surg 1979; 78:489–493 regional perfusion and oxygenation in critically ill patients. 55 Bollaert PE, Charpentier C, Levy B, et al. Reversal of late Acta Anaesthesiol Scand (Suppl) 1995; 107:239–248

896 Clinical Investigations in Critical Care