A 55-year-old woman was admitted with toxic shock syndrome. Her norepinephrine requirement was labile, fluctuating between 15 mcg/min and 30 mcg/min. Bedside echocardiogram showed a dilated inferior vena cava without respiratory variability, and a normal ejection fraction. On examination her extremities were cool and her urine output was marginal.
Epinephrine 4 mcg/min was added, with an improvement in blood pressure, perfusion, and urine output. She reported feeling better, and over the next hour the norepinephrine was weaned off. Her lactate increased from 2 mM to 6 mM. The ICU team was pleased with this epinephrine-induced rise in lactate, a positive prognostic sign. Sure enough, she continued to improve and was weaned off vasopressors entirely over the next 18 hours.
The critical care world is obsessed with fluid. Meanwhile, little attention has been given to the concepts of vasopressor responsiveness and vasopressor challenge. This is a missed opportunity, because vasopressor challenges may be done more safely and precisely than fluid challenges (1). To develop an approach to vasopressor responsiveness in septic shock, three concepts are helpful.
Foundational concept #1: Vasopressors don't always behave according to the textbook
Individual patients vary considerably in how they respond to vasopressors. Two factors in particular may contribute to this variability.
Endogenous sympathetic response
When we administer exogenous catecholamines to a patient, these are not operating in a vacuum: patients are equipped with their own adrenergic responses as well.
For example, imagine an elderly man and a young woman, both in septic shock from bacteremia. The young woman mounts a robust endogenous catecholamine response which defends her blood pressure. Thus, she presents with a MAP of 60 mm, heart rate 140 b/m, and a lactate of 8 mM (the lactic acid is due to endogenous epinephrine driving aerobic glycolysis). The older man has a weak catecholaminergic response, so he presents with a MAP of 45 mm, heart rate of 90 b/m, and a lactate of 2 mM. Of these two patients, the elderly man would probably respond better to exogenous catecholamines, because he is catecholamine deficient.
Variable end-organ sensitivity
The differences between various catecholamines results from the ratio of stimulation of alpha-receptors vs. beta-receptors (adjacent figure). However, patients vary in the responsiveness of their heart and vasculature. For example, reduced cardiac sensitivity to beta-agonists is common in advanced sepsis, due to receptor down-regulation. Refractory vasodilation is also well described.
Relative differences in end-organ sensitivity affect the way vasopressors function. For example, norepinephrine typically has predominantly vasoconstrictive effects with a small amount of isotropy. In a patient with reduced cardiac responsiveness to beta-agonist stimulation, this could cause the vasoconstrictive effects of norepinephrine to predominate even further. In this scenario, norepinephrine would have nearly the same effect as a pure vasoconstrictor such as phenylephrine:
Foundational concept #2: Norepinephrine and epinephrine are both evidence-based vasopressors for use in sepsis.
Although norepinephrine is typically first-line in septic shock, epinephrine is also a reasonable choice. Myburgh 2008 performed a RCT comparing norepinephrine vs. epinephrine in patients with shock, which found no difference in outcomes. Annane 2007 performed a RCT comparing epinephrine vs. the combination of norepinephrine and dobutamine, also finding similar outcomes. More recently, a RCT in pediatric sepsis found epinephrine to have a substantial mortality benefit compared to dopamine, further supporting the use of epinephrine in septic shock (Ventura 2015).
Groups of patients respond similarly to norepinephrine vs. epinephrine, but this doesn't mean that individual patients will have the same response to either drug (an assumption exemplifying the flaw of averages). On the contrary, it is likely that some patients respond better to norepinephrine, while other patients respond better to epinephrine.
Foundational concept #3: The epinephrine hyper-responsive patient
Average responsiveness to norepinephrine vs. epinephrine
The phenomenon of epinephrine hyper-responsiveness
Some patients seem to be especially responsive to epinephrine, with the following features: (3)
- They don't respond very well to norepinephrine, often requiring high doses of norepinephrine. Sometimes the norepinephrine dose will fluctuate widely, suggesting that the norepinephrine isn't very effective. While on norepinephrine, they may have poor perfusion.
- They are very responsive to epinephrine. For example, the image below shows a patient who was transitioned from 30 mcg/min of norepinephrine to 6 mcg/min of epinephrine. Following transition to epinephrine, these patients often have improved perfusion.
- There is often an increase in lactate levels following the initiation of epinephrine, which is a positive prognostic sign (discussed below)(4).
The physiology of epinephrine hyper-responsiveness is likely multifactorial, perhaps with varying contributions from the following factors.
Factor #1: High endogenous alpha-adrenergic tone? A patient who is already quite vasoconstricted may respond poorly to additional exogenous alpha-adrenergic stimulation.
Factor #2: Relative bradycardia? Occasional patients in septic shock are encountered who have inappropriately normal heart rates (e.g. 50-70 b/m), even despite being on norepinephrine. This could relate to a deficiency of endogenous beta-adrenergic tone, or perhaps an over-reactive carotid baroreceptor response to alpha-adrenergic stimulation (5). Regardless, such patients might benefit from the stronger chronotropic effects of epinephrine.
Factor #3: Occult systolic failure? An occasional septic patient has a normal ejection fraction, but after starting norepinephrine the ejection fraction drops. Why? Such patients have likely had systolic dysfunction all along. Initially their untreated vasodilation reduced their afterload, increasing their ejection fraction. Norepinephrine normalizes the afterload, thereby unmasking this systolic dysfunction.
Patients with occult systolic failure may experience less improvement in blood pressure in response to norepinephrine. In extreme cases, their cardiac output could even drop as the norepinephrine is titrated up (6):
Factor #4: Lactate deficiency? Although lactate is widely feared, it is actually a beneficial molecule that is used by the heart as bioenergetic fuel under stress. For example, exogenous lactate infusions have been shown to improve cardiac output in heart failure (Nalos 2014).
Lactate is mostly generated aerobically as a result of beta-2 adrenergic stimulation. Epinephrine (which stimulates beta-1 and beta-2 receptors) causes lactate production, whereas norepinephrine does not. Although it has traditionally been believed that lactate production from epinephrine was an adverse side-effect, this could actually be a strength of epinephrine:
Among patients with septic shock, an elevation of lactate levels in response to starting epinephrine correlates with survival (Omar 2011, Wutrich 2010). This raises the possibility that some patients may have a relative deficiency of endogenous epinephrine, causing a relative lactate deficiency.
Evaluating epinephrine responsiveness: The epinephrine challenge
Baseline hemodynamics and echocardiography provide static hemodynamic variables which usually cannot predict epinephrine responsiveness (7). Thus, the only way to be certain of how a patient will respond to epinephrine is to administer epinephrine. One potential exception might be a patient with marked tachycardia and a hyperkinetic ventricle.
The best way to determine epinephrine responsiveness may be to simply start a low dose epinephrine infusion (e.g. 4 mcg/min)(8). The patient's response to epinephrine may be judged based on clinical variables (table below). Although measuring cardiac output could also be performed, this is not my usual clinical practice (9).
Algorithm for vasopressor titration
- Epinephrine is trialed relatively early (before the patient is frankly “failing” norepinephrine).
- Many algorithms involve sequential addition of different vasopressors. However, it may also be useful to down-titrate vasopressors to which the patient responds poorly.
- Epinephrine and norepinephrine are both acceptable, evidence-based approaches to hemodynamic support in septic shock.
- Individual patient responsiveness to vasopressors is variable and unpredictable.
- Some patients respond better to epinephrine than norepinephrine.
- For patients who are not responding well to norepinephrine, it is reasonable to empirically trial a low dose of epinephrine (“epinephrine challenge”).
- Alternative viewpoint on phenylephrine infusions (PulmCrit)
- Why we fail at hemodynamics: Cognitive errors (PulmCrit)
- Vasopressor basics (EMCrit)
- I must admit that I've never been entirely satisfied with the performance of a fluid challenge. The fluid often runs in more slowly than I would have hoped, by the time I re-assess the patient further time has elapsed, and it is often a bit murky whether the patient has responded. Vasopressor challenge is faster and easier: simply increase or decrease the vasopressor and observe the results. If it is unclear whether the patient responded, repeat it. Or use a higher dose of vasopressor. By paying attention to the patient's physiology during serial adjustments to the vasopressor, and by having the courage to sometimes make large changes in vasopressor dose suddenly, it is possible to get a very good idea of how the patient is responding to vasopressor.
- Brown 2013 also suggested that in general, 1 mcg/min epinephrine is equivalent to 1 mcg/min norepinephrine.
- The fact that some patients respond much better than others to dobutamine was previously explored here. So the concept that some patients respond better to beta-adrenergic stimulation than others, and that these beta-adrenergic responders tend to do well, is nothing new – this has been established in the literature for years.
- Note that it is very unclear what the role of cycling lactate is for a patient on an epinephrine infusion. My guess is that it would probably be better not to check it, but rather to rely on clinical endpoints (e.g. urine output, etc.). Perhaps it might be worth keeping an eye on it to make sure there wasn't a profound lactic acidosis from the epinephrine, but overall cycling lactate in this context seems to cause more confusion than benefit.
- Morelli 2008 showed among patients with septic shock that the average heart rate decreased in response to norepinephrine, exactly the same response as was seen with phenylephrine. Thus it is possible that in selected patients norepinephrine could produce a suboptimal heart rate.
- The response to norepinephrine is complex; for example, there is often also an increase in preload due to venoconstriction. Thus, it is possible for patients to experience an increased cardiac output even if there is a decrease in ejection fraction (due to the effect of increased preload overcoming the effect of increased after load). Patients who experience a frank reduction in cardiac output due to norepinephrine are probably a small subset of patients with the most severe occult systolic dysfunction.
- At first, I thought epinephrine responsiveness was simply a matter of systolic heart function (i.e. patients with low ejection fraction would respond better to the inotropic effects of epinephrine). However, epinephrine hyper-responsiveness may also be observed in patients with a normal ejection fraction. This is consistent with evidence that dobutamine responsiveness couldn't be predicted based on baseline hemodynamic profiles (discussed previously here).
- Could we assess epinephrine responsiveness by just giving a push-dose of epinephrine. Maybe, I don't know. The physiologic effect of a bolus may not be exactly the same as the effect of an infusion, because with an infusion the body has more time to re-equilibrate to the effects of the epinephrine. Until more evidence is available, the best way to predict the effect of an epinephrine infusion is to try an epinephrine infusion.
- Reasons that I don't monitor cardiac output are as follows:  Most forms of cardiac output monitoring aren't accurate enough to detect small changes in cardiac output (e.g. on order of 10-15%)  Patients who have a significant increase in cardiac output from epinephrine should also experience an increase in blood pressure, based on the relationship Bp=SVR*CO. Low-dose epinephrine shouldn't really affect the SVR much, so the Bp ought to track roughly with the cardiac output.  Clinical indicators of perfusion can further help assess whether the cardiac output is improved. It is possible that skin and urine markers of tissue perfusion could actually be more important than whole-body cardiac output.  Not measuring cardiac output allows this challenge to be performed easily by a nurse without fancy equipment.
- Early vasopressin initiation was explored previously here, and is also weakly supported by the VANISH trial results (which have been presented at the Intensive Care Society meeting but not yet published). This will be discussed further after the VANISH trial is published in full. Overall this remains an area of equipoise, so there is a lot of variation regarding when different folks chose to add vasopressin.