1. Steroids in critical illness Mark Daly Consultant Endocrinologist Exeter

2. This talk…  A challenging case  HPA axis insufficiency in traumatic brain injury  Diagnosing corticosteroid deficiency in the critical ill patient Different approaches Different approaches Controversies Controversies Understanding the pathophysiology Understanding the pathophysiology Towards a sensible consensus Towards a sensible consensus  Role of corticosteroid use in the ITU setting Concept of relative adrenal insufficiency Concept of relative adrenal insufficiency Failure of response to conventional vasopressor interventions Failure of response to conventional vasopressor interventions  Conclusions

3. LG, 55  48 yr F, presents with collapse, recent diarrhoeal illness  BP 70/40  39.5 o c  Swollen inflamed wrist  CRP 295, WCC 24, creatinine 295, bicarb 10  Na+ 107, K+ 6.2

4. LG, 55  Admitted to ITU  Baseline serum taken  Put on steroids  Endocrinology consult requested

5. LG, 55  Random cortisol 2  No ACTH  LH 55, FSH 72  PRL 450, TSH 2.9, FT4 18

6. LG, 55, PMH  Thyroid eye disease  Treated thyrotoxicosis  Recent uncontrolled hypertension  Manic depression  Anti-phospholipid syndrome

7. LG, 55  Differential diagnosis? Primary hypoadrenalism Primary hypoadrenalism Supported by U+Es, but other explanationsSupported by U+Es, but other explanations But recently hypertensive so NOT slow onsetBut recently hypertensive so NOT slow onset Secondary hypoadrenalism Secondary hypoadrenalism Why such striking preservation of other pituitary hormones?Why such striking preservation of other pituitary hormones? ?false cortisol result ?false cortisol result ?severe electrolyte loss ?severe electrolyte loss Acute total hypoadrenalism? Acute total hypoadrenalism?

8. LG, 55

9.  Clinical diagnosis of bilateral acute adrenal infarction confirmed by CT  Exacerbation of mania caused by higher dose steroids  Subsequent hyponatraemia caused by excessive water drinking (due to mania) then Addisonian crisis (omitting steroids)

10. HPA axis insufficiency in traumatic brain injury  TBI leading cause of death and disability in young adults  Autopsies have shown up to pituitary gland necrosis in one-third of patients

11. HPA axis insufficiency in traumatic brain injury  Beaumont Hospital Dublin  102 neuro-surgical survivors;  Broad-spectrum of AP testing, limited by risk of ITT  28.4% had at least one AP hormone deficiency  ~10% were ACTH deficient  Thus: Clear evidence of pituitary dysfunction as a result of acute injury persisting long-term Clear evidence of pituitary dysfunction as a result of acute injury persisting long-term  BUT: Is this an acute phenomenon? Is this an acute phenomenon?  J Clin Endo Metab 89:4929, 2004

12. HPA axis insufficiency in acute traumatic brain injury  50 patients with severe or moderate brain injury; 31 matched controls, median 12 days post injury  Glucagon stimulation test  Peak cortisol <450 in 16%  (80% had gonadotrophin deficiency with low sex steroids)  No reference to free cortisol or albumin levels  Thus in acute phase of brain injury there is evidence of pituitary dysfunction  Clin Endo (Oxf) 2004; 60:584-91

13. More difficult territory

14. Diagnosing adrenal/pituitary insufficiency in critically ill patients  Sometimes it feels like late-onset hypogonadism!  Why?  Because someone is trying to diagnose a disease without a clear pathogenesis and borderline results (and equally borderline treatment responses)

15. A review of physiology  The simple approach:

16. The simple approach

17. The complex approach

18. The clinical approach –in sickness and in health  Autonomic nervous system modulates CRH release by hypothalamus  CRH stimulates ACTH release from anterior pituitary  ACTH stimulates cortisol synthesis (plus aldosterone and DHEA)  Feedback of cortisol to ACTH and CRH  Vasopressin stimulates ACTH secretion in the presence of CRH

19. The clinical approach –in sickness and in health  IL-1, IL-6, MIF* and TNF- alpha promote high corticosteroid levels  IL-6 receptors seen on pituitary corticotrophs and adrenal cortical cells  Effect in addition to classical pathway activation  Impaired clearance of steroid (esp renal/hepatic disease  Steroid in excess of Cushing’s AND less suppressible  (best seen in neuro- surgical patients where 24mg dexamethasone does not suppress endogenous production)  BUT individual variation in response to stress  *(MIF – macrophage migratory inhibitory factor)

20. So:  Strong, unsuppressible cortisol secretion in stress suggests this is an adaptive response  So inadequate response suggests failure and indication for high-dose steroids?  AND resistance to gluco-corticoid action may be present despite a strong response?

21. Evidence for HPA axis failure in critical illness  Published data cover very heterogeneous groups Many with severe sepsis Many with severe sepsis

22. Hypoadrenalism in septic shock Study Patient nos Criteria Relative insufficiency 132Inc<25019 222Inc<20075 321Peak<50024 420Inc<20045 522 Base<1000+inc<20 0 55 659B<69061 7189Inc<24854 845Inc<24833 968B<50032 J Clin Endo Metab 2006,91:3725-45

23. Hypoadrenalism in septic shock  Conclusions from small studies - small increments more likely to die (but sample of 13 v 6)  BUT higher baseline values more likely to die in larger studies  Higher cortisol:DHEAs ratio predicted death  BUT are we simply dealing with severity of illness markers?

24. Assessing HPA function in critical illness

25. Problems of assessing HPA function in acute illness  Total versus free cortisol Transcortin low capacity, high affinity Transcortin low capacity, high affinity Transcortin may fall in acute illness Transcortin may fall in acute illness Thus total but not necessarily free cortisol may fall Thus total but not necessarily free cortisol may fall Can use calculated correction factor, serum cortisol divided by the transcortin concentrations Can use calculated correction factor, serum cortisol divided by the transcortin concentrations BUT: BUT:

26. Problems of assessing HPA function in acute illness  Total versus free cortisol Transcortin low capacity, high affinity Transcortin low capacity, high affinity Albumin high capacity, low affinity Albumin high capacity, low affinity At higher concentrations more will be albumin bound (when transcortin is saturated) At higher concentrations more will be albumin bound (when transcortin is saturated) Thus when albumin falls in the stressed individual, the effect on total cortisol is disproportionate Thus when albumin falls in the stressed individual, the effect on total cortisol is disproportionate

28. Is this a significant effect?  66 critically ill patients  7-10 fold increase in free cortisol conc  AND:  40% of the hypoproteinaemic patients failed SST if total cortisol was used as the marker c/w free cortisol  NEJM 350:1629-38

29. Should we use the SST?

30. GR, 78 yrs old  Collapse post-hip replacement, severe headache  BP 80/50  Na+115  Diagnosed SIADH  Fluid-restricted

31. GR, 78 yrs old  Further collapse  Endo SpR reviewed  Random cortisol 125  SST rise to 490  Given steroids  Developed 3 rd nerve palsy

32. GR, 78 yrs old

33. Clinical conclusion  SST is not valid in acute onset of secondary hypoadrenalism

34. Returning to more common critical illness  Several studies have advocated RISE or increment in cortisol as key factor  Then used as justification for steroid use  Returning to complication of CBG/albumin levels, one proposal: Interpret in context of albumin Interpret in context of albumin If albumin > 25 g/l, peak 25 g/l, peak <20 mcg/dl (550)= deficiency

35. The Coolens method  U2 x K (1 + N) + U[1 + N + K(G – T)] – T = 0, where T is cortisol, G is CBG, U is unbound cortisol, where T is cortisol, G is CBG, U is unbound cortisol, K is the affinity of CBG for cortisol at 37 C K is the affinity of CBG for cortisol at 37 C N = albumin bound: free cortisol ratio N = albumin bound: free cortisol ratio  The value of N would be expected to change with altered concentrations of plasma albumin  Countered by investigating the distribution of cortisol (600 nmol/l) in varying concentrations of purified human serum albumin solutions using equilibrium dialysis.  Use experimentally derived values of N to further calculate free cortisol, thus compensating for variations in plasma albumin.  J Clin Endocrinol Metab. 2006 Jan;91(1):105-14  J Steroid Biochem. 1987 Feb;26(2):197-202

36. Medication – a complicating factor  Potential effects on binding but also synthesis  Etomidate is a imidazole used as an anaesthetic to facilitate intubation  Causes reversible inhibition of 11 β-hydroxylase  Associated with impaired HPA axis function even after single injection  This has confounded some of the larger studies  Either should be abandoned or given with steroid cover

37. Relative adrenal insufficiency  Patients without risk factors for adrenal dysfunction  critical illness related corticosteroid insufficiency (CIRCI). ( inappropriate steroid activity given a patient's severity of illness)  Serum cortisol levels FELT to be inadequate  299 patients with septic shock Non-responders<250 increment Non-responders<250 increment Non-responders benefited from corticosteroid Non-responders benefited from corticosteroid 68 of the non-responders had etomidate 68 of the non-responders had etomidate ? No published response re benefit for non-etomidate, non- responders ? No published response re benefit for non-etomidate, non- responders JAMA 288:862-871 JAMA 288:862-871

38. Glucocorticoid therapy during acute illness  CORTICUS study   (Corticosteroid Therapy of Septic Shock)   Multicentre, international, double-blind RCT.   primary end point was 28-day all-cause mortality in "nonresponders" (defined as a change of ≤ 9 mcg/dL in cortisol after a 250-mcg SST   Secondary end points mortality, organ failure resolution, and safety.

39. CORTICUS study   Powered on 800 patients to detect a 10% difference in mortality.   500 participants  50 mg of hydrocortisone every 6 hours for 5 days with a tapering dose over the next 6 days v placebo

40. CORTICUS study results  no differences in these baseline characteristics / severity of illness  For no outcome end point was there a difference with use of corticosteroids.  All-cause mortality was similar (34% corticosteroids vs 31% placebo).  Mortality rates also did not vary based on responder status.  nonresponders tended to have higher mortality overall.  Rates of superinfection were higher in those given corticosteroids (34% vs 27%, P =.099).  The frequency of hospital-acquired new sepsis was also higher in those randomized to steroids.  Hyperglycemia was also more common on study day 1 in persons treated with corticosteroids.  (report of American Chest meeting)

41. Corticus – a word of caution  Clinical trials register: Study Type: Interventional Study Design: Treatment, Randomized, Double-Blind, Placebo Control, Parallel Assignment, Safety/Efficacy Study Study Type: Interventional Study Design: Treatment, Randomized, Double-Blind, Placebo Control, Parallel Assignment, Safety/Efficacy Study  BUT:  Adrenal function in sepsis: the retrospective Corticus cohort study  Crit Care Med. 2007 Apr;35(4):1012-8.  Annane is listed as last author and also on trials database

42. A last resort?  “The decision to administer steroids in in sepsis cannot be based on markers of adrenal function, rather treatment should be considered in septic patients with vasopressor refractory hypotension”  Ann Pharmacother 2007;41:1456-65

43. Conclusions  Management of hypoadrenalism in critically ill patients with identified aetiology is straightforward There remains a clear role for considering the individual patient and using clinical acumen There remains a clear role for considering the individual patient and using clinical acumen

44. Conclusions  Total cortisol levels may be misleading in critical illness Due to alterations in CBG and albumin levels Due to alterations in CBG and albumin levels  This can be corrected for either by testing free cortisol OR the Coolens method

45. Conclusions  Drugs previously commonly used in anaesthesia impair cortisol synthesis and genuinely compromise adrenal function  Other drugs can affect CBG levels and total cortisol measurements

46. Conclusions  The evidence for blanket use of moderate high dose steroids in critical illness has NOT been made  Some use pragmatic approach – i.e. if all else fails

47. Finally…..  Thus use clinical acumen and consider the aetiology  Interpret function in the light of altered physiology  Be aware of limitations of interventional studies