Upon first reading the ATHOS-3 trial, I was pleasantly optimistic. Who wouldn't be interested in a shiny new vasopressor? The trial didn't prove much, but it was intriguing. However, it was alarming to hear that the FDA has approved angiotensin II based on it. Precious little evidence is available about this drug. With angiotensin II arriving at hospitals soon, some cautions are in order.
Caution #5: Angiotensin II has pro-inflammatory activity
Septic shock results from a cytokine storm involving several pro-inflammatory cytokines (perhaps most prominently interleukin-1, tumor necrosis factor, and interleukin-6). Angiotensin II has pro-inflammatory properties, which include increasing interleukin-6 levels:
Sepsis usually causes a hypercoagulable state, for example increasing the risk of DVT/PE. At a microvascular level, tiny thromboses may contribute to tissue damage. A pro-thrombotic drug is undesirable in sepsis.
Angiotensin II increases thrombin formation and impairs thrombolysis (Celi 2010, Dielis 2007). The ATHOS-3 manuscript didn’t report an increase in thrombotic risk when various types of thrombosis were examined separately. However, the FDA package insert reports an increase in the combination of venous plus arterial thrombotic events (13% vs 5%, p=0.02):
Caution #3: Blockade of the renin-angiotensin-aldosterone system improves survival in animal models.
Based on several mechanisms (including #4-5 above), some investigators have suggested that blocking angiotensin II could be beneficial in sepsis. Animal models confirm that blockade of the renin-angiotensin-aldosterone system may improve survival (Laesser 2004, Hirano 2014)(3). This has led to a bizarre situation where different clinical trials are investigating the efficacy of both angiotensin II and also angiotensin II blockers in sepsis:
The fact that opposite hypotheses are being investigated reveals weakness in the basic science behind using angiotensin II for septic shock. If there were definitive pre-clinical evidence supporting the use of angiotensin II, then it would be inconceivable to perform a clinical trial of an angiotensin II antagonist.
Caution #2: Limited safety data.
FDA approval was based largely on the ATHOS-3 trial, which cannot establish safety:
- This study exposed only 163 patients to angiotensin II, which is underpowered to detect rare adverse events. For example, regardless of your interpretation of the thrombotic risks discussed above, this trial doesn't resolve the issue.
- Adverse events involving the skin were more common in the angiotensin II group (13.5% vs. 7%). This may be a signal of angiotensin II causing digital ischemia.
- Study inclusion required that the patient must have high-output shock based on either a central venous oxygen saturation >70% plus CVP > 8 mm or a cardiac index >2.3. Angiotensin II would be predicted to be less safe among patients not meeting these criteria.
- The dose of angiotensin II was allowed to increase up to 200 ng/kg/min over the first three hours, but subsequently decreased to 0-40 ng/kg/min (figure below). Higher doses over the first three hours were used to prove efficacy, whereas safety was achieved by dropping the infusion dose to a lower level after three hours. This bait-and-switch scheme leaves it unclear whether it is possible to find a dose which is both effective and safe for >3 hours.
The primary endpoint in ATHOS-3 was a composite of either achieving a MAP rise of 10 mm or a MAP >75 mm within three hours. Angiotensin II has been known to increase blood pressure for over fifty years. Therefore, choosing a blood pressure endpoint was an easy goal to reach. This composite endpoint seems engineered to guarantee the study a “positive” result (e.g. MAP >75 mm is an artificially high target). When the blood pressure graph is formatted properly, there is no clinically meaningful difference:
What is missing from the ATHOS-II trial is improvement in any meaningful patient-centered outcome (e.g. survival, renal function, ICU length of stay). The study protocol did record several additional endpoints including serum creatinine, urine output, time on the ventilator, full SOFA score breakdown by organ, and ICU length of stay (Chawla 2017). Unfortunately these endpoints weren't reported, perhaps because they didn't reflect positively on angiotensin II. It is tantalizing that angiotensin II improved the cardiovascular component of the SOFA score despite making zero difference in the composite SOFA score – arithmetic requires that angiotensin II must have impaired the function of some other organ(s) not reported in the manuscript (1).
Vasoconstriction is a tricky business. Some vasoconstriction is needed to establish a reasonable pressure, but excessive or nonselective vasoconstriction can choke off tissue perfusion. The fact that a vasoconstrictor increases blood pressure doesn’t guarantee that it will help patients. For example, 546C88 (a nitric oxide synthesis inhibitor) caused vasoconstriction and shock resolution in a 312-patient MCRCT eerily similar to ATHOS-II (Bakker 2004, abstract below). This seemed like a terrific drug, until it was proven to increase mortality in a larger RCT (Lopez 2004).
Comparison of angiotensin II vs. ascorbic acid/thiamine/hydrocortisone
Treatment with ascorbate/thiamine/hydrocortisone is controversial, commonly criticized as lacking evidentiary support. However, there is even less evidence to support angiotensin II. The basic science evidence supporting ascorbic acid is also considerably more robust than the rationale for angiotensin II (more on this here).
Predictions about angiotensin II based on the performance of its next-door neighbor, vasopressin.
Vasopressin and angiotensin II share numerous commonalities:
- Non-catecholamine vasoconstrictors.
- Preferential vasoconstriction of the efferent arteriole of the kidney, potentially increasing glomerular filtration rate and urine output.
- Endogenous neurohormones which may be deficient in septic shock.
- Sometimes effective at low, quasi-physiologic levels.
How useful is vasopressin in septic shock? In the past, I was very enthusiastic about vasopressin based on theory and evidence that it could improve renal function (e.g. prior posts here, here, and here). However, over time this enthusiasm was tempered by growing awareness of problems from excessive vasoconstriction that result from combining vasopressin plus norepinephrine. In some cases, patients developed necrotic toes possibly related to this combination. Currently, my practice patterns have shifted towards reserving vasopressin for patients who are refractory to catecholamine pressors.
That said, there does seem to be a subset of patients who are refractory to norepinephrine and epinephrine, yet respond nicely to vasopressin. These patients often do beautifully with a combination of vasopressin plus epinephrine (“vepinephrine”), with inotropy from epinephrine preventing an excessive drop in the cardiac output:
Overall, vasopressin is a useful pressor to have in the proverbial toolbox. However, most patients respond nicely to norepinephrine. For such patients, vasopressin is unnecessary and could pose risks of digital ischemia.
Vasopressin's track record should temper our enthusiasm about angiotensin II. In particular, simultaneous use of three vasoconstrictors (norepinephrine, vasopressin, and angiotensin II) could increase risks of digital or intestinal ischemia. However, as with vasopressin, there might be a subgroup of patients who respond unusually well to angiotensin II.
Comparison of angiotensin II vs. phenylephrine
Phenylephrine is often feared, due to concern that it may cause excessive vasoconstriction and drop the cardiac output. This fear isn't really supported by available evidence, which suggests that phenylephrine is rather similar to norepinephrine (2).
For the many people who avoid phenylephrine due to concerns about dropping cardiac output, it is logical to share the same concern for other pure vasoconstrictors (vasopressin and angiotensin II). It is physiologically illogical to hate phenylephrine while loving angiotensin II.
- Angiotensin II has pro-inflammatory effects, causing increased levels of interleukin-6.
- Angiotensin II has pro-coagulant effects, which could increase microvascular thrombosis and the risk of DVT/PE.
- Angiotensin II inhibition improved survival in an animal model of sepsis, suggesting that excessive angiotensin II activity can be harmful.
- Angiotensin II hasn't been shown to improve any outcome other than a minor increase in blood pressure. Full stop.
- ATHOS-3 is underpowered to detect uncommon adverse events.
- The rationale for using angiotensin II is very similar to the rationale for using vasopressin. Although vasopressin is occasionally quite useful, it hasn't proven to be a wonder drug.
- EMCrit interview with Mink Chawla: Strongly recommended, with an excellent discussion focusing on the rationale for using angiotensin II. This post is intended to be considered in combination with the EMCrit podcast, to provide different perspectives. Read the ATHOS-3 manuscript, consider the pro- and con- arguments, read the Bottom Line Review of the study, and see what you think. The truth is out there.
- Vasopressor challenge concept (PulmCrit). As our menu of vasopressors expands, it becomes harder to figure out which one to choose. When in doubt, a reasonable approach is to try test doses of different agents. If the patient responds strongly to a small dose, that suggests that it's helping (e.g. if 0.03 U/min of vasopressin bumps the blood pressure a lot, that patient probably had a vasopressin deficiency). Alternatively, if the patient requires a crushingly high dose to see any measurable effect, then that drug probably isn't working very well. The ultimate goal is to support the patient's pressure and perfusion with a minimum total medication dose (noting that excessive doses of any inotrope or pressor can be toxic). Please note that this is different from the traditional approach of adding agents sequentially in a rigid fashion until the patient is on max doses of every imaginable drug simultaneously – which isn't a particularly elegant strategy.
- Full breakdown of the SOFA score over time would be required to sort this out, but I can't find it published in either the manuscript or the supplemental appendix.
- I find phenylephrine useful in patients with atrial fibrillation (doesn't speed up the heart rate) or poor IV access (minimal risk of extravasation). I have gotten some patients through septic shock with peripheral phenylephrine (patients who merely required a minimal dose of vasopressor and didn't want to get a central line. Yes, I realize this runs against the guidelines, but it's OK provided that you monitor the patient and the patient is clearly responding well to therapy).
- A recent study found that irbesartan reduced inflammation and cardiac dysfunction in a mouse sepsis model (Yousif 2017).