There is a certain amount of cortisol which is required for the organism to survive critical illness. A critically ill organism which is producing too little cortisol is thus said to have a "relative" adrenal insufficiency. This contrasts with absolute adrenal insufficiency, which the failure to produce a normal amount of cortisol in a normal state of relatively good health. Question 8 from the second paper of 2004 vaguely touched on this topic as one of the causes of adrenal insufficiency in the critically ill.
Theoretical foundations of relative adrenal insufficiency
Who came up with this concept? Well. In 1924, J.M. Scott injected dead streptococci into adrenalectomised rats, and noted that they tended to die more horribly than his controls. In fact, he noted that "A dose of killed streptococci or staphylococci can be obtained that is invariably fatal to adrenalectomized rats, before hypertrophy of cortical accessories, but never kills control rats." This ancient paper is a becautiful read; each rat is described carefully, and its final days are subjected to careful biographic scrutiny, eg "At 1.06 p.m. injected 2 cc. of streptococcus suspension (31 billion per cc.). 2.30 p.m. Quiet. Found dead at 5 p.m."
So perhaps removing both adrenal glands was somewhat crude. However, it was obvious that cortisol must play a role in the response of the organism to sepsis.
The interest in the role of steroids in the management of infectious diseases persisted thoughout the twentieth century. Adrenal glands were being variably blamed for increased susceptibility to infections, or praised for their ability to induce defervescence and symptomatic relief.
A flurry of research activity ensued. Thousands of mice and rabbits were killed. Positive sentiment rose. In 1966, from atop a pile of dead animals Robson and Cluff declared that "hydrocortisone administered with penicillin prevents death not prevented by penicillin alone".
The concept of relative adrenal insufficiency in septic patients and the idea that its management might improve outcomes was ultimately solidified in the late 1970s, when Sibbald et al reported on the responses of septic patients to synthetic ACTH. A group of septic patients was discovered who had suprisingly low-normal cortisol levels in spite of their raging sepsis, and who did not respond to synthetic ACTH. The authors concluded that "patients with severe sepsis who are not responding adequately to standard therapy should be suspected of having adrenocortical insufficiency and treated accordingly."
More recent studies continue to support the idea that a certain group of sepsis patients simply fail to mount a satisfactory adrenocortical response. It seems that septic shock patients who are unable to increase their serum cortisol above 90μg/L in a short synacthen test are more likely to die.
Recognition of relative adrenal insufficiency
Though people seem to agree that some septic patients have too little cortisol, there is a widespread disagreement of how much cortisol is enough, and even more disagreement as to how one identifies such patients. Things are made more difficult by the fact that septic ICU patients do not demonstrate the classical features of hypoadrenalism, such as hyperkalemia and hyponatremia.
Patients identified by high vasopressor dose
One typically resorts to using vasopressor responsiveness to recognise these patients, partly because the use of steroids promises to restore vasopressor responsiveness. The CORTICUS trial enrolled patients who required on average 0.5μg/kg/min of noradrenaline. That would be about 35ml/hr for a 70kg person, with the routine 6mg/100ml dilution. In contrast, the French in 2002 had patients running on 75ml/hr of noradrenaline. Of the two, only the French study had demonstrated any sort of survival benefit from steroids; and in the opinion of most people, the CORTICUS trial essentially crushed this study. If one were to draw some sort of arbitrary conclusions about the "typical" noradrenaline requirement of a relatively adrenally insufficient patient, one would be forced to conclude that it would be high, and perhaps above 35ml/hr.
The unreliability of the cortisol assay
If one is going to base one's decisions on a blood test, one ought to have some confidence in the accuracy of the test. Even this is debatable when it comes to cortisol. An article by a selection of powerful CICM figures (Jeremy Cohen, Bala Venkatesh, et al 2006) weighed in on the utility of the commonly used cortisol assays in diagnosis of relative adrenal insufficiency. They tested three commonly used commerial immunoassays comparing them with high pressure liquid chromatography (which is too expensive for everyday use). The assays gave wildly diverging levels, disagreeing with the chromatograph by a shocking range of -62% to 770%. All three assays reached diagnostic concordance only in 44% of patients. This, plainly, will not do.
The unreliability of the Short Synacthen Test
There has been some interest in carrying out tests which stimulate the adrenal glands artificially, and record their response. Theoretically, the glands of a septic patient with "relative adrenal insufficiency" will not respond to ACTH. Annane et al published a 2005 paper in which the levels of cortisol were assayed after an overnight metyrapone stimulation test. On the basis of their findings, they have decided that any septic patient with a serum cortisol change of less than 90μg/L is likely to have relative adrenal insufficiency.
Having identified "responders" in this way, we might expect them to "respond" to steroid treatment with something positive, like increased survival. Unfortunately, this does not seem to be the case. Results of the CORTICUS trial published in 2008 did not find any mortality difference between responders and non-responders. The ACTH challange simply failed to discriminate between people whose survival depends on steroid supplements, and those who will be just fine without them.
The disagreement about normal and low dose ACTH tests
One might argue that 250μg is an insanely supraphysiological dose of ACTH. The pituitary never makes that much ACTH, one might complain. How are we to trust the adrenal glands which respond to this stimulus? Perhaps they respond to such vigorous prodding, but when it comes to more "normal" stimulation, they might just sit there uselessly and secrete nothing. In performing the 250μg ACTH test, one fools oneself into believing that the adrenal glands are working normally, and thus one might miss many genuinely "relatively adrenally insufficient" patients.
With this in mind, clever people have made attempts to reduce the ACTH dose to detect more patients in need of steroids. Specifically, two ACTH doses (1μg and 250μg) were tested in septic patients. As expected about 50% of people who responded to the high-dose test failed to respond to the low-dose test. "The low-dose test identified a subgroup of patients in septic
shock with inadequate adrenal reserve", the authors gloated.
However, what did this really mean? Thus far there is no evidence that the use of the low-dose test improves survival by revealing hidden hypoadrenalism among septic patients. In fact, there's no evidence that ACTH testing in general adds anything to the process of managing sepsis. In an intelligent response to all the gibberish published on this subject, the Surviving Sepsis people have recommended (with a 2B-level of commitment) that we stop wasting time with the ACTH challange, and just use steroids in patients who are refractory to fluids and vasopressors.
Management of relative adrenal insufficiency
Let us say that, having navigated the evidence-sparse quagmire of cortisol testing, one has somehow convinced oneself that one's patient is in need of "stress dose" steroids.
The rationale behind the administration of steroids in shocked patients, as well as the evidence in support of this practice, is discussed in the chapter concerning the use of corticosterids in sepsis.