So, the the noradrenaline is at double strength, and vasopressin is at 2.4 units per hour. Antibiotics are given, corticosteroids have been thrown in, and the lungs are so full that you won't consider another fluid bolus. You've just congratulated yourself on identifying some decreased cardiac contractility, and levosimendan has increased the cardiac index to some reasonable number, but the mean arterial pressure just won't stay up. Lactate remains high, extremities remain mottled, urine output has dwindled to zero. In short, you're screwed. You remind yourself that this condition has a mortality rate of over 50% even in developed countries, and prepare for a difficult discussion with the family.
But before that, is there anything else you can do to prevent the heart from stopping? The following list of therapies falls into the category of exotic, experimental, controversial, or plain wacky. It has minimal exam relevance beyond the a list of "things to consider" as the sixth line of therapy in refractory sepsis. The time-poor exam candidate can safely limit their reading to skimming through the headings.
What do you mean by "super-refractory" septic shock?
To define the statement better than the grim case vignette offered in the introduction, one may wish to use some sort of consensus definition from the literature. Fortunately, somebody (Bassi et al, 2013) has trawled the literature and found such a definition for us. It seems that a noradrenaline dose in excess of 0.5 μcg/kg/min is generally held to be "high dose" - that corresponds to an infusion rate of around 35 ml/hr (standard dilution) for a 70kg patient.
The 0.5 µcg/kg/min seems like a purely arbitrary cut off. Is this really a physiologically relevant threshold? Probably, nothing much different happens at 34 ml/hr which will not be happening at 45 or 55 ml/hr. However, there is some science behind this threshold, and most of the trials listed by Bassi et al use it as their enrolment cut-off. In fact, Benbenishty et al (2011) were able to correlate vasopressor dose with the risk of death using an AUROC model. At 0.5 µcg/kg/min, the AUROC was 0.85 and the sensitivity and specificity for the likelihood of mortality was 96% and 76%, respectively. At 90 days, people receiving 1.0 µcg/kg/min (or around 35ml/hr of "double strength" noradrenaline) had a mortality of 83% as per Brown et al (2011). This cutoff value is not completely blind to the use of other vasopressors (eg. adrenaline, vasopressin, et cetera) because most trials use "noradrenaline-equivalents" to convert doses of other agents into a total pressor infusion dose rate.
Correction of metabolic acidosis
One might consider exogenous buffer therapy if the patient has a pH under 7.15, which is thought to be the threshold below which catecholamine receptors are particularly insensitive, and where myocardial depression by metabolic acidosis begins to play a role. Different thresholds are quoted in the literature (UpToDate recommends a pH of 7.10). The best literature summary on this topic can be found at the end of the 2004 article by Cariou et al. In brief, the main point of that section is that no, acidosis is probably not the main player in septic haemodynamics. Consider: ketoacidosis patients frequently turn up with a pH of 6.8, but they never have such severe haemodynamic instability as the septic public. Even so, while acidosis might play a minor role, correcting severe acidosis is worth a shot if nothing else is working.
Cruel rat studies (Fox et al, 1992) discovered that in conditions of metabolic acidosis, α-1 receptor responses are blunted. The correction of acidosis by bicarbonate should therefore improve the catecholamine receptor sensitivity, and in turn improve the vasoplegia. Sodium bicarbonate should also have a series of other positive effects, such as volume resuscitation (it is hyperosmolar and it has plenty of sodium in it).
Does it work? Bicarbonate certainly has not had much haemodynamic effect in clinical studies. Cooper et al (1990) had given bicarbonate to patients in whom the pH wa as low as 6.9, and did not find any haemodynamic improvement. That said, these authors did not correct ionised calcium, allowing it to drift dangerously low (0.87 mmol/L) which might explain some of the effect failure. Who can say what this might have looked like if the calcium replacement was concurrent.
This equimolar mixture of sodium carbonate (Na2CO3) and sodium bicarbonate is no longer on the market in Australia, but one should probably be aware of it in case it comes up in an exam at some stage. It can be said to have a similar effect to sodium bicarbonate, with the exception of the CO2 production (it is supposed to generate less of it by about a third).
Discussed elsewhere (see the chapter on buffer therapies ), THAM is an option for correction of acidosis in situations where additional sodium bicarbonate is for some reason undesirable. One might find oneself in such a situation if one has already overdosed the patient with bicarbonate, and now the patient has a serum sodium of 155 mmol/L. Just as an example.
Does it work? In humans, who can say. Certainly a 1997 isolated rat heart study demonstrated improved contractility and diastolic relaxation. Kraut et al (2016) in a recent review of lactic acidosis was unable to identify any good human studies to support its use. "Examination of these therapies in humans is warranted", he concluded.
If death is near, why not some THAM? It is not associated with any concerns about intracellular acidosis worsening because of some mythical CO2 migration, and its main side effect is hypoglycaemia (in a stress-hyperglycaemic ICU patient this seems trivial). Unlike bicarbonate, it will not cause an intracellular shift of potassium. It may even decrease PaCO2 if there is respiratory acidosis (by manipulating the edges of the Henderson-Hasselbach equation). With all of that said in support of it, we really don't know what the hell happens in severely septic acidotic patients who receive THAM. For those who intend to use it anyway, a detailed protocol exists (Nahas et al, 1998)
You've probably never even heard of this stuff, and with good reason. It is not available commercially. On paper, it sounds like an elegant solution to lactic acidosis. This substance activates mitochondrial pyruvate dehydrogenase, reversing its activity and thereby converting lactate into pyruvate.
Why would one want to get rid of lactate, is it not just a marker of severity? Well. Severe lactic acidosis - at least in animal studies (Kimmoun et al, 2015) - seems to have some unpleasant haemodynamic consequences. These are largely related to the changes in the way intracellular calcium interacts with the contractile proteins, and with the effect of lactate on opening ATP-sensitive potassium channels on vascular smooth muscle. Ergo, getting rid of lactate seemed like a laudable goal.
Vary et al (1987) had this idea long ago, and found it effective in lowering lactate among septic animals. These data could not be replicated by Stacpoole et al (1992), who described the resulting changes in biochemistry as "statistically significant but clinically unimportant".
Improved diastolic filling with heart rate reduction
The septic patient is tachycardic. Let alone the crusty elderly person with LV hypertrophy, even a young supple heart cannot be said to fill effectively at a heart rate of 150. So, the idea of slowing down the heart rate and increasing diastolic filling time to improve stroke volume has come up several times in the history of critical care literature. Certainly diastolic dysfunction (Sanfilippo et al, 2015) and tachycardia (Leibovici et al, 2007) have both been associated with increased mortality in sepsis. Sounds like something we should deal with.
Esmolol and other β-blockers
Sanfilippo et al (2015) performed the most recent meta-analysis of this topic. In short, there was not enough prospective data to perform a proper meta-analysis. However, ten promising studies were identified. Five of these focused on the use of esmolol, and one in particular was a trial with surprisingly positive results. Morelli et al (2013) randomised 77 patients to receive esmolol (aiming for a heart rate between 80 and 94) and another 77 to remain tachycardic. The esmolol group had massively improved mortality ( 49.4% in the esmolol group vs 80.5% in the control group), as well as decreased fluid requirements and improvements in a variety of Swan Ganz catheter variables. This rings alarm bells, because no sepsis control group should ever have 80% mortality - what sort of half-arsed measures did their "routine care" consist of? In any case, the trial was not designed to detect a mortality difference (they just happened to have reported on it). Furthermore, the results were probably affected by the fact that 49.4% of the esmolol group and 40.3% of control patients received levosimendan.
Apart from small trials with dodgy methodology, retrospective audits and case series also exist which have related the chronic use of β-blockers with an improved mortality in sepsis (Christensen et al, 2011). The results of the ESMOSEPSIS trial is currently recruiting participants, and will hopefully be published at some stage in 2017. In response to the Sanfilippo , the Maybauers (2015) issued a warning to enthusiastic esmolol users, pointing out the fact that safety and efficacy have yet to be confirmed by good quality trials.
Ivabradine is a "funny current (If ) channel inhibitor. It only does one thing, which is decrease the sinus node rate. If you believe in heart rate reduction for sepsis but are put off by the potentially myocardium-depressing effects of beta blockers, ivabradine seems like a logical choice. It is a very "clean" drug, with few other effects, and The MODIfY trial is supposedly currently recruiting, although the last time its site was updated was 2010. The trial protocol published in 2011 (Nuding et al) offers a good discussion of its expected effects in sepsis (in brief, they are the same as the effects of beta-blockers).
As far as published human data is concerned, there is little at the moment. De Santis et al (2014) offer a case study of three patients, in whom ivabradine was used to control heart rate in the contest of sepsis and multi-organ system failure following cardiac surgery. In short, all that can be said is that it is well tolerated, and produces a reliable heart rate reduction.
Catecholamines in refractory septic shock
Noradrenaline is not working. Is there anything else in the catecholamine range of molecules?
Many of us will become somewhat anxious when the nurses ask whether they can load up four times the usual dose of noradrenaline into the infusion bags so that they don't have to keep hanging them up every hour. Indeed, much of the literature on alternative vasopressors discusses things in terms of their "noradrenaline-sparing" effect. However, it is important to reflect on this behaviour. Is there an irrational fear of high dose noradrenaline among intensive care professionals?
Katsaragakis et al (2006) retrospectively reviewed data from 1999 - 2002 to determine whether high dose noradrenaline was really as bad as it is made out to be. Patients included in this review were receiving over 4 μcg/kg/min of noradrenaline, which corresponds to 280ml/hr of the standard dilution. Twelve such patients were found. The survival rate was 33%. "Safe and effective" was how the authors described this insane noradrenaline dose. Members of the same group presented this data as a poster in 2004; the implications of their conclusion is that people are being noradrenaline pansies and under-utilising a drug which can (and should) be pushed to higher doses before resorting to second line agents. Having no access to the full text of the article, I can only assume that the authors would have mentioned lost limbs in their abstract if these were a major feature among survivors.
In summary, if one were inclined to increase the rate of their noradrenaline infusion into these stratospheric ranges, one would be able to draw upon some of this supporting evidence. However, it strays far from accepted practice, and may be criticised harshly by the amputees.
Phenylephrine is a synthetic catecholamine best known to the non-ICU public as a stupidly ineffective oral decongestant. It has been studied in the context of sepsis by Morelli et al (2008) and Jain et al (2010). The main attraction of this substance is its extremely high α-1 receptor selectivity. As such, it constricts mainly large arteries rather than terminal arterioles. This has positive implication for splanchnic blood flow (full of terminal arterioles). Adding phenylephrine to noradrenaline should prevent gut ischaemia in severe sepsis, if you are otherwise going to be using tons of noradrenaline - so the hypothesis goes.
Unfortunately, these hypothetical benefits have failed to materialise. Morelli et al could not demonstrate any difference in regional haemodynamics between noradrenaline and phenylephrine. Jain et al also found little difference between the two drugs, with the exception of a lower heart rate with phenylephrine. Given what is known about diastolic dysfunction in sepsis, this can be seen as an advantage.
In short, adding phenylephrine to noradrenaline has little benefit apart from situations where tachycardia or ß-receptor stimulation is for whatever reason undesirable. One can conceive of several such situations, for instance sepsis in combination with HOCM, or Takotsubo cardiomyopathy, or severe coronary artery disease. If dynamic LVOT obstruction is the cause of one's haemodynamic instability, phenylephrine may solve the problem.
Exotic non-catecholamine vasopressors
With advancing shock, worsening acidosis, climbing heart rate lactate and glucose, one may be forgiven for becoming disappointed with catecholamine vasopressors. This disappointment is reflected in the international trend towards "decatecholaminisation", the movement characterised by an increased interest in vasopressin analogs and β-blockade. The non-catecholamine drugs also have the advantage of an additive effect in combination with noradrenaline, and can be expected to reduce the noradrenaline dose. A decreasing noradrenaline dose has many positive effects, not the least of which being the positive psychological effect on the intensivist.
Methylene blue is a dirty drug with numerous effects, among which is the inhibition of induceable nitric oxide synthase and guanylate cyclase. Now, nitric oxide synthase inhibition has a long rich history in the sepsis literature, and most of it is bad news. For example, a 2004 trial of stupidly named "Agent 546C88" (Lopez et al) was terminated prematurely because of increased mortality in the treatment group. However, methylene blue does not seem to have the same negative stigma.
One of the major concerns with this class of drugs is the potential for pulmonary vasoconstriction. Kirov et al (2001) did not find this to be a problem in their small open-label trial. The drug was given as a loading bolus (1.0mg/kg) followed by an infusion of 0.5 mg/kg/hr for 4 hours. A total dose of about 7mg/kg is thought to be the daily maximum (adverse efects such as methaemoglobinaemia tend to develop beyond that dose range). Everybody seems to give a short course as described above; nobody has ever studied infusions going for 24 hours or longer.
So, how good is it? El Adawy et al (2016) found that haemodynamic goals were achieved faster with methylene blue as compared to vasopressin. Kwok et al (2006) tried to perform a systematic review but was frustrated by the lack of high-grade evidence. Ten years later, Hosseinian et al (2016) had exactly the same problem. Good quality trial evidence for the use of methylene blue is hard to find, likely at least in part because of the fact that it is a relatively old cheap drug and nobody is going to make money from an increase in its use (in fact, its cheaper than noradrenaline). Most of the observational studies throw methylene blue at the patient in the desperate hours prior to death, which completely obliterates any chances of a positive effect being observed. In spite of this, a clear signal from the literature is that it increases systemic vascular resistance and acts to decrease the requirements of other vasopressors.
Terlipressin is a synthetic analogue of arginine vasopressin, a longer 12-amino-acid molecule with a longer half life which is owed to delayed metabolism. It is also supposed to have an increased selectivity for the V1 receptor, which suggests that it may have a greater vasopressor effect with fever side-effects than vasopressin.
The ide of using terlipressin instead of noradrenaline is not new. Albanese et al (2005) gave 1mg boluses of it to patient with septic shock, and compared it to noradrenaline infusion. Both drugs were equally effective in improving blood pressure, but terlipressin "pressored" at the expense of cardiac index, which dropped markedly. That probably was not clinically significant in hyperdynamic sepsis patients (the CI fell from a mean 5.1 to a mean of 4.2, which is still damn good) but the concern was that in patients with depressed myocardia, the contractility disadvantage would translate into worse tissue perfusion. In answer to these concerns, the DOBUPRESS study (Morelli et al, 2008) paired terlipressin boluses with dobutamine and found that a relatively high dose of dobutamine (20 µcg/kg/min) was required to counteract the cardiodepressant effects. The subsequent TERLIVAP trial (Morelli et al, 2009) compared terlipressin (as infusion) with noradrenaline (at 15 µcg/min) and vasopressin (at 0.03 units/kg/hr); the drugs all had equivalent efficacy and the infusion seemed to be a better administration strategy.
All this demonstrates that terlipressin is probably safe in septic shock, and is a valid alternative or adjunct to noradrenaline. But none of these studies included patients in truly refractory shock. Rodríguez-Núñez et al (2006) tested it in septic children whose noradrenaline doses were around 2.0 µcg/kg/min, and found that it halved the noradrenaline requirement. Seven of the sixteen children survived, and of the survivors four went on to have vasopressor-related limb amputations. That seems like a major problem, but it is unclear whether terlipressin killed those limbs, or the DIC-related gangrene, or whether they would have fallen off anyway with such a massive noradrenaline dose.
In short, terlipressin seems like a reasonable rescue agent. It is unclear whether there is any point in adding it to vasopressin in "super-refractory" septic shock, or whether there is any benefit in replacing vasopressin with terlipressin. At this stage, all that can be said is that it has vasopressin-like noradrenaline sparing effects, and that it seems safer as an infusion
This drug is a V1a-selective vasopressin receptor agonist, with potent vasopressor activity. Maybauer et al (2014) infused it into infected sheep, with positive effects. Among them, apart from maintaining the blood pressure selepressin prevented fluid accumulation by selectively ignoring the antidiuretic V2 receptors. If one is in favour of a dry approach to sepsis, this drug is an ideal replacement for vasopressin. Marks and Pascual in their 2014 editorial for CCM also brought up its effects on vascular leakyness, suggesting that vasopression is inferior to selepressin because it permits more interstitial fluid accumulation.
Currently, there is little human evidence for the efficacy of selepressin, or its comparison with vasopressin and noradrenaline. However, a large mult-center trial (SEPSIS-ACT) has recently been launched to investigate the role of selepressin. It is sponsored by the manufacturer (Ferring Pharmaceuticals) and should therefore be anticipated with a mixture of excitement and contempt.
ATII is an octapeptide hormone, cleaved from angiotensiongen by actions of renin and ACE. It has a circulating half-life of approximately 30 seconds, and in the tissues it can last as long as 15 to 30 minutes. It is rapidly degraded to angiotensin-III Acting on Gq-protein-coupled angiotensin receptors, it affects a totally separate vasopressor pathway and is therefore synergistic with noradrenaline vasopressin and methylene blue.
Possible specific indications for the use of ATII include severe ARDS (where ACE levels are low because of pulmonary parenchymal destruction, i.e. there is not enough lung to convert angiotensin-I to angiotensin-II) and to counteract premorbid ACE-inhibitor treatment. These are straightforward: angiotensin is genuinely absent in the first case, and its receptors are disabled in the second. It also has no inotropic or chonotropic properties, which makes it a perfect agent for preserving a nice long diastolic filling time.
As far as sepsis goes, apart from being used empirically as a potent vasopressor angiotensin is also useful to counteract the sepsis-associated downregulation of the angiotensin-II receptor. This downregulation does not seem to be a part of any sort of adaptive response; in fact it is completely counterproductive. As the ATII receptors in the adrenal medlla are downregulated, so the release of catecholamines and aldosterone diminishes, which contributes to the hypotension of sepsis. Bucher et al (2010) who discovered this phenomenon in rats culd not explain why it happens, only that it seems to be associated with elevated nitric oxide levels.
Angiotensin-II has seen successful use in humans. In a pilot study (the ATHOS trial - Chawla et al, 2014) angiotensin-II was infused at a dose of 2 to 10 ng/kg/min. This infusion reduced the noradrenaline dose from around 30 μcg/min down to around 7 µcg/min. Surprisingly, 20% of the patients were exqusitely sensitive to angotensin-II and became hypertensive even at minimal doses. ATHOS was not powered to detect a mortality difference, nor will be the ATHOS-3 trial (currently recruiting to a goal of 315 patients).
In short, if angiotensin-II were available locally, its uses would be in severe shock with ARDS, in patients affected by their previous ACE-I use and as a third (fourth?) agent to add for horrific refractory vasoplegia.
Ionised calcium as a vasopressor and inotrope
Ionised hypocalcemia is a well-acknowledged cause of hypotension in critically ill patients. Desai et al (1988) described this as an "association" because a causal relationship could not be inferred from their study, but certainly patients with low calcium levels tended to require more vasopressors.
The article by Stanislaw Jankowski and J.L Vincent (1994) offers a detailed exploration of the cardiovascular effects of calcium. In short, hypocalcemic animal (and human patients) respond well to calcium administration - their blood pressure and cardiac output improve. In turn, normocalcemic organisms benefit less, with only a slight improvement in their cardiovascular performance. Correction of hypocalcemia leads to significant improvement of haemodynamics only if the ionised calcium level was severely decreased (i.e. down by one third, to about 0.9-0.8 mmol/L).
With specific reference to shock, there is little supportive data. Gary Zaloga (2000) listed a series of contemporary studes which did not show any benefit in terms of survival, and quoted some animal evidence of increased mortality (apparently, the increased serum calcium sends a really powerful pro-apoptotic signal to near-death injured tissues). Ishibashi et al (2015) compared retrospective data and found that septic patients became more unstable after their calcium was corrected. J.L Vincent (1995) instead found that arterial pressure increased, but the effect was only sustained for about one hour.
In short, correction of hypocalcemia is at least transiently helpful, but induction of hypercalcemia is probably completely pointless.
High dose insulin in septic shock
The use of glucose insulin and potassium as an infusion has been extrapolated from the experience of toxicologists in treating β-blocker overdose. Theoretically, the use of insulin to "force" more glucose into the myocardium should somehow enhance its contractility, and the activation of c-AMP second messenger system should bypass the catecholamine receptors which have been disabled by severe acidosis. Like methylene blue, this strategy has seen more use in the setting of post-bypass vasoplegia.
In 2006, Hamdulay et al published a report of two cases where septic shock with myocardial depression was treated with glucose insulin and potassium. The dose rate was 1.5 units/kg/hour, similar to the ß-blocker overdose protocol (the patient ends up getting about 100 units per hour). Apart from the cardiotonic effects, insulin has a variety of immunomodulatory effects, including decrease in the circulating levels of IL-6 and TNF-α.
However, one needs to behold this exotic therapy within the "what would the coroner say?" framework of decision-making. Consider: if levosimendan is available and the patient has sepsis-associated cardiac depression, then why not use this more conventional inotrope instead? What could compel you to attack the patient with a hundred units of insulin per hour? One can conceive of a situation where all possible inotropes have been employed, with persisting severe global systolic dysfunction. In such a scenario, high dose insulin may be the last-ditch pharmacological alternative to mechanical haemodynamic support.
Techniques to control dysregulated inflammation
Control of ridiculous brain-frying fever is a well-accepted strategy in sepsis. Apart from this protective effect, lowering the temperature has a vasoconstrictor effect and should therefore decrease vasopressor requirements. Schortgen et al (2012) trialled external cooling to normothermia (aiming for 36.5-37.0° C) and found that this strategy indeed decreased noradrenaline doses by up to 50%. Chang et al (2013) demonstrated improved mortality in septic rats with mild therapeutic hypothermia (34.0°C), prompting calls for human studies. An excellent review of the contemporary literature is afforded by the 2011 Masters of Science dissertation by Karen Luo, available online from the University of Ottawa. The rats in Luo's study were cooled to 21°C a'la deep hypothermic circulatory arrest, with impressive immunomodulatory effects.
The counter-argument is that fever is a protective antimicrobial strategy which has evolved over millions of years, and that artificially depressing this adaptive response is probably unhelpful. Moreover, cooling has a cardio-depressant effect, and this might be counterproductive in septic cardiomyopathy. Hypothermia in the post cardiac arrest setting is also known to depress immune function, particularly neutrophil and monocyte activity (indeed that is one of its benefits in global ischaemic-hypoxic reperfusion injury). So perhaps it is not uniformly helpful (i.e. the pursuit of an improved MAP is not worth all the other problems).
There is also the idea of artificially adjusting the temperature so as to hurt the microbes. Theoretically, there is some temperature range at which the cocci are going to have some sort of reproductive disadvantage or impaired defences. However, one might have to consider the possibility that the cocci are much more resilient to temperature changes than the poor mammal they are infesting. They will probably thrive at a temperature range which is rapidly fatal for the patient.
In short, one can make an evidence-based recommendation for the pursuit of a normal temperature in a patient with high fever (reduce metabolic requirements, protect organs, etc etc) but therapeutic hypothermia remains experimental, even though there is both animal and human evidence of a strongly attenuated inflammatory response with temperatures in the 33-34° C range.
High-flux haemodialysis and haemofiltration
Still on the topic of removing cytokines an endotoxin via some sort of extracorporial blood purification therapy, the use of super-high doses of CVVHDF (eg. 50-100ml/kg/hr) has been proposed in the late 1990s-ear;y 2000s. This is covered with more gusto in the chapter on evidence against the use of high-volume haemofiltration. In brief, the biological rationale is plausible (remove the inflammatory badness, relieve the severity of SIRS). A good representative article comes from the 3rd ADQI conference (2005), where a workshop headed by Bellomo addressed the use of haemofiltration and haemoperfusion in septic shock. That workshop rode the crest of a great wave of interest for this therapy. "Promising but untested" was the consensus opinion.
Subsequent years whittled away the enthusiasm, as with many things in intensive care medicine. No mortality benefit was discovered, and the application of this therapy is not without its dangers (for one, your filter sucks out all your noradrenaline). These days, interest has focused on "high cut-off membranes" (Day et al, 2016) which remove everything smaller than albumin, for instance myeloma light chains. For sepsis, this strategy appears the more promising the poorer the quality of the evidence. For instance, a retrospective case-control study found a mortality benefit (Chelazzi et al, 2016), but there were only sixteen patients, and all of them had renal failure 9so... did they improve because of cytokine clearance, or because you treated the uraemia?) Even veterans of exotic dialysis stratgies (Forni, Ricci, Ronco - 2015) have become disappointed. Their article makes excuses for the failure of this technique, eg. "despite our wishes, perhaps the most plausible explanation is that HVHF may be ineffective at providing adequate mediator clearance at the cellular level rather than in the circulation."
In summary, one should be at least dimly aware of the potential use of CRRT in sepsis (particularly membranes with unusually high porosity and prescriptions with unusually high dose). One should also be aware that it has the potential to make the patient substantially less stable, and may not be associated with any sort of improvement in outcome. It would be difficult to make a straight-faced argument in favour of this technique.
Coupled plasma filtration-adsorption
This technique consists of passing the patient's blood though a resin filter which non-selectively adsorbs inflammatory mediators and endotoxin. Ronco et al did this to ten patients in 2002. The plasma-filtered patients had improved haemodynamics, which the authors attributed to improved leukocyte responsiveness to lipopolysaccharide. Formica et al (2007) reviewed the contemporary evidence behind this practice and decided that it has merit "without renal indications", meaning they proposed to use it in sepsis purely for the clearance of these sepsis-associated proinflammatory molecules. Berlot et al (2012) also wrote favourably about it, suggesting that the limited studies are interesting enough to merit routine (early) use and calling for a randomised trial.
Such a trial was published by Livigni et al in 2014. The authors randomised 330 patients in 18 ICUs to either receive CPFA or ...not receive it. The intervention consisted of five daily sessions of 10 hours per day. CPFA did not reduce mortality in patients with septic shock, nor did it positively affect other important clinical outcomes like organ failures or ICU stay. Only in an analysis of the a priori "high dose" subgroup, a trend towards improved mortality was seen.
Given that this is the only large trial examining this technique, defending its routine use in refractory sepsis might be difficult in a forum of EBM-savvy peers. Additionally, logistic barriers exist. Not everybody has these CPFA cartridges laying around in their ICU store room. According to a quick Google search, the only company manufacturing such consumables is the Italian division of Medtronic (called Bellco).
Activated Protein C
The failure of independent investigators to verify the findings of company-sponsored trials had killed and buried drotrecogin-alpha. The recombinant stuff has been taken off the market in 2011 ("fraud", they shouted) but what most people don't realise is that the human-derived Protein C is still available. The theoretical benefit is related to the prevention of DIC and its progression to microvascular thrombosis which is probably a major mechanism of sepsis-induced organ dysfunction.
Even though recent articles (eg. Griffin et al, 2015) continue to extol the virtues of human and murine APCs, its use does not seem to be supported by evidence. A recent RCT (Pappalardo, 2016) used human Protein C zymogen to treat sepsis in severely shocked patients (APACHE II > 25). The study was ceased due to evidence of harm: ICU mortality was 79 % in the protein C zymogen group vs 39 % in the placebo group.
In summary, it is probably worth knowing about even though it has been largely discredited. Some still recall anecdotes and personal experiences with the drug, where catastrophic sepsis simply "melted away". These days, the place formerly occupied by Xigris will probably soon be occupied by recombinant thrombomodulin.
Thrombomodulin is a membrane protein expressed on the surface of vascular endothelial cells. it binds to thrombin, forming a 1:1 complex and acting as an anticoagulant. It also activate Protein C. The 2015 article by Takayuki Ikezoe offers an excellent discussion of the coagulation cascade pathways involved, and it has the added credibility of originating in Japan where this stuff has been available since 2008. To summarise, the objective of administering the soluble recombinant thrombomodulin (rTM) is to prevent DIC, or at least reduce its severity and ensuing organ system damage.
Kato et al (2013) reports on the Japanese experience with rTM in the form of a retrospective cohort study. This was a small coghort, but it did demonstrate reduced DIC scores and relative safety (i.e. no catastrophic bleeding episodes). J.L Vincent et al (2013) trialled rTM outside of Japan in an multi centre RCT, randomising 750 patients. All meaningful outcome measures were the same between groups. Only in post-hoc analysis did some benefit emerge (apparently if you already have at least one organ dysfunction, rTM is more beneficial). This and two other trials were scraped together and remixed in a 2015 meta-analysis by Yamakawa et al. With the data combined and properly massaged, a non-significant reduction in mortality was discovered. Some sort of "negative slope" was found in association with increasing mortality, prompting the authors to conclude that rTM must have some sort of hidden benefit in the super-sick severe sepsis patient group.
All that can be said about rTM on the basis of these data is that it is safe. Among the studied patients, risk of bleeding with rTM was no greater than in the control group. The combination of rTM and antithrombin may be better than either agent alone (Yasuda et al, 2016). Ultimately, larger trials are needed, but without the deep pockets of Eli Lilly the going may be slow.
Targeted anti-cytokine therapy
As sepsis is nowadays defined as a "dysregulated host response", the regulatory system of such a response should make an attractive drug target. Indeed, there was extensive interest in cytokine antagonist therapies in the 1990s; trials included TNF-α monoclonal antibodies, antiendotoxin antibodies, platelet activating factor antagonists and numerous others. A sombre article from 1999 (E.Abraham) discusses why all of these trials came to nothing. In summary, it seems the main problem the author has is that the trials enrolled septic patients, instead of patients with specific cytokine excess. If a high TNF-α level was found in the septic patient, then that patient should have been enrolled in the trial of TNF-α antagonist - so the author argues. In either case, monoclonal antibodies to cytokines exist and are familiar today (eg. infliximab for TNF-α) but interest in their use for septic shock has died out.
Exotic antimicrobial strategies
Clindamycin has a well established role to play in toxic shock syndrome, where it inhibits bacterial protein synthesis and therefore prevents the generation of superantigen. Its role in sepsis in a broader sense is not well established. One might make the argument that in horrific refractory septic shock with an unknown organism, one needs to at least consider the possibility that toxic shock is present. In that context, giving a few doses of clindamycin begins to look appealing. At worst, it will do nothing (as a drug it is relatively harmless).
Like clindamycin, IVIG is best known for its ability to neutralise superantigens in the context of toxic shock syndromes, classically due to S.pyogenes (Darenberg et al, 2003). In other forms of sepsis, there is no superantigen to bind and so the effect of polyclonal IVIG will be diminished. Conceivably, some of the immunoglobulin molecules infused into the patient will bind to the circulating bacteria and thereby contribute to the resolution of sepsis, but this is far from an accepted use for this expensive substance, which costs upwards of AU$50 per gram.
Having said that, there really seems to be some benefit. Alejandrija et al (2002) performed a meta-analysis of polyclonal IVIG in septic shock. A significant reduction in mortality was found. Sure, these trials may have inadvertantly included some patients with toxic shock syndrome, but the data are still promising. There may be more to this effect than the consequences of infusing it as a hyperoncotic resuscitation fluid. In any case, locally the storage and distribution of IVIG is controlled by the Australian Red Cross, for whom it is a precious and carefully guarded treasure. If you ring them and ask for IVIG for the management of uncomplicated sepsis, they will laugh at you.
Mechanical haemodynamic support
VA ECMO as a circulatory assist in septic shock
Yes, people have done this. It seems like cheating, but the ultimate response to a failing circulatory system is to replace the failing motor with a reliable external alternative. Brechot et al (2013) reported retrospectively on the experiences of their single centre from 2008-2011. During that time, 14 patients underwent VA ECMO for septic shock (all had severe cardiac dysfunction as the main problem, rather than vasoplegia). All appeared doomed (mean SOFA scores were 18). There was a surprising 71% survival to discharge, which suggests that this strategy is not without merit. Indications for VA ECMO (apart from an LVEF < 25%) were sustained hypotension in the face of high dose vasopressors and inotropes (dobutamine > 20 μcg/kg/min and noradrenaline > 1.0 µcg/kg/min). Patients were excluded if their cardiac index was preserved. Brechot et al made no mention of levosimendan or high dose insulin and presumably did not use them.
So, one might say this study is not particularly illuminating, as it mainly confirms what we already know (i.e. that VA ECMO is helpful if your LV is dysfunctional). What, one might ask, is the outcome if you throw ECMO at patients with mainly vasoplegic shock? Another, more recent series looked at what might happen to those patients. Falk et al (2019), working from the Karolinska facility, were able to enrol 37 septic patients into their case cohort. Selection was guided by vasopressor requirements: a Vasoactive Inotropic Score greater than 50 was required to enter, and they were all approximatelky as sick as Brechot's patients (average SOFA was 16). Overall, total survival at 46 month follow-up was 59.5%. In-hospital survival was 90% for those who had mainly LV failure (n=20), and only 64.7% for those who had mainly distributive shock (n=17). The investigators were actually encouraged by these numbers, as they had expected a much higher mortality on the basis of the SAPS-3 scores (83% of these patients were supposed to be dead according to the estimated mortality rate based on their SAPS-3 scores). "Veno-arterial ECMO could be considered, not only for patients with myocardial dysfunction but also for distributive septic shock patients with preserved myocardial function", the authors cheered.