Targeted temperature management in intensive care – Do we let nature take its course?

Abstract

Should we aim to intervene and control fever in the critically ill patient? The answer is not straightforward and there is certainly no universal agreement on the subject. This article aims to discuss whether we should over-ride nature and disallow it to take it’s course, particularly where it appears that this evolutionary response to invading pathogens is actually becoming harmful to the patient. Also discussed here are the physiology of temperature control and the scope of our current understanding of the impact of fever in patients manifesting systemic inflammatory response syndrome (SIRS) and sepsis in ICU, the possible interventions to combat fever (both physical and pharmacological) and the evidence for anti-pyretic drug therapy. The final section examines the potential role of targeted temperature management in the management of sepsis / SIRS in the critically ill.

Fever as a response to infection has been preserved by evolution throughout many disparate species. In addition, fever was recognised as a sign of disease centuries before the advent of germ theory and recent studies have shown that such signs described in 200AD are still relevant today.^1,2

Given the highly preserved nature of the febrile response to infection, it is reasonable to suggest that fever provides a net survival benefit to a species. However, despite various studies examining the effects of fever and antipyretic therapy in intensive care, it is not clear whether fever provides a net benefit or harm amongst septic intensive care patients.^3,4

There is good evidence for the role of targeted temperature management (TTM) in improving neurological outcome in patients, post-cardiac arrest and there is an emerging role in traumatic brain injury (TBI).^5,6 The hypothesised mechanism is to reduce the cerebral rate of metabolic demand for oxygen (CMRO₂) in order to prevent secondary brain injury. An interesting study by Schortgen et al. studies the role of TTM in sepsis and suggests a potential role in non-neurological critically unwell patients. ⁷

The aim of this article is to briefly discuss the physiology of temperature control and to examine current understanding of the impact of fever in patients manifesting systemic inflammatory response syndrome (SIRS) and sepsis in ICU. The role of physical anti-pyretic treatment interventions will be discussed, as well as evidence for anti-pyretic drug therapy (the pharmacology of which is without the scope of this section). The final section examines the potential role of TTM in the treatment of SIRS / sepsis in the critically ill.

Physiology of fever

Thermoregulation

In humans thermoregulation occurs within the temperature control centre in the pre-optic area of the hypothalamus. Fever and hyper-pyrexia must be clearly distinguished. Fever is defined as a body temperature persistently elevated above the normal of around 37.1 ℃, however it seems there is no agreement on the upper limit for normal temperature (values between 37.5 ℃ and 38.3 ℃ being quoted). Hyper-pyrexia is defined as a persistent temperature >40 ℃. It is stated that brain and multi-organ damage can occur if body temperature exceeds 40 ℃ for even minimal time periods. ⁸ Fever is normally the result of stimulation by elevated levels of prostaglandin E2 (PGE2) in response to various pyrogenic cytokines such as IL-1, IL-6 and tumour necrosis factor (TNF), interferon and macrophage inflammatory proteins. Bacterial lipopolysaccharides act as superantigens and are also potent initiators. ⁹ These compounds migrate within the circulation to the circumventricular organs of the brain, where they bind with endothelial receptors on vessel walls, or interact with local microglial cells. When these cytokine factors bind, the arachidonic acid pathway is activated.

Toll like receptors (TLRs)

Exogenous pyrogens such as endotoxin also stimulate fever via TLRs that span cell membranes of many immune cells, such as macrophages, dendritic cells, lymphocytes, neutrophils and mast cells to name but a few. ¹⁰ There are some 13 sub-classes of TLR, all geared to recognise pathogen specific molecules (PSMs), leaving host molecules largely untouched (see list below). These PSMs have remained unchanged through evolution, constituting one of the most rudimentary parts of the immune system. TLR activation leads production of many of the pyrogenic cytokines mentioned above.

TLR and their antigen activators:

TLR-1 - bacterial lipoprotein and peptidoglycans

TLR-2 - bacterial peptidoglycans

TLR-3 - double stranded RNA

TLR-4 - lipopolysaccharides

TLR-5 - bacterial flagella

TLR-6 - bacterial lipoprotein

TLR-7 - single stranded RNA

TLR-8 - single stranded RNA

Heat shock proteins (HSPs)

The febrile state induces transcription of HSPs such as HSP70. ¹¹ These proteins are synthesised in response to thermal stress and have three protective roles: they play a part in vascular smooth muscle tone in response to fever (vasodilatation), thought to occur via induction of nitric oxide synthase; they help to maintain intracellular protein integrity during febrile states, overseeing correct folding and unfolding during transcription; they help to maintain intracellular steroid receptor and transcription factor structure.

Scope of practice in ITU

Fever is one of the most commonly treated phenomena in medicine, and is particularly prevalent on ITU. In this section, we will present some arguments for and against the treatment of fever in sepsis; are we right to intervene?

Negative effects of fever

The febrile response may well be an effective defence preserved by evolution, but it is clear that it also comes at a cost. It significantly increases metabolic demand and thus oxygen consumption, normally accompanied by a corresponding increase in cardiac output. What is unclear is the extent of the damage this may be causing patients who are already receiving physiological support, or have pre-existing disease (most ITU patients from the outset). In a septic shock model of pneumonia in sheep, paracetamol and external cooling were shown to improve respiratory function and prolong survival time. ¹² In an alternative murine model of pneumonia, the presence of fever was associated with decreased survival and worsened pulmonary vascular injury, irrespective of enhanced elimination of pathogens. ¹³ This could be a result of the additive effect of infection, coupled with the additional inflammatory ‘collateral damage’ to the lung and other organs, explaining the increased duration of mechanical ventilation seen in one observational study by Netzer et al. ¹⁴

Positive effects of fever

Potentially beneficial effects of fever include slower replication of bacteria, enhanced activity of antibiotics, increased mobility of leukocytes with enhanced phagocytosis, decreased endotoxin effects, increased proliferation of T-cells, amplified immune response and transcription of HSPs.^15,16 In another paper, it was observed that there might be a link between febrile state in early ARDS and improved survival, although this was a secondary outcome measure. ¹⁷

In vitro studies and animal models of sepsis have shown fever to slow the replication of bacteria and reduce the bacterial load in murine models of peritonitis. ¹⁸ Induced hyperthermia was associated with lower mortality in rat models of faecal peritonitis. In the same rat model, hyperthermic preconditioning also seemed to prevent a drop in CD4+ lymphocytes and may also be associated with an enhanced host immune response. ¹⁹

Too hot; too cold or just right (the Goldilocks phenomenon)

A recent review demonstrated the impact of fever on ICU patients is unclear, but it did demonstrate that hypothermia in the setting of severe sepsis was associated with a higher mortality. ²⁰ This is in agreement with Cheng et al. who showed that the absence of fever in septic patients was an independent predictor of mortality. It is unclear whether these findings are as a result of the beneficial effects of fever or selection of a more elderly, more unwell population.

Pharmacological intervention or not?

The FACE study showed that amongst patients with sepsis, there was no significant association of fever with 28-day mortality but the administration of paracetamol or NSAIDs to control fever was independently associated with increased 28-day mortality. ²¹

Within the last five years, three meta-analyses have been conducted examining the effects of antipyretic treatment on critically ill patients. Jefferies et al. found that there were insufficient randomised controlled trials (RCTs) to allow elucidation of the effect of antipyresis on ICU patients with suspected infection. ²² Notably, patients with acute brain injury undergoing some form of physical cooling were excluded from this analysis. Hammond and Boyle included physical cooling methods but did not exclude those with neurological injury. ²³ They were unable to show a significant mortality benefit in treated patients. Finally, Niven et al. selected trials in patients without neurological injury but studied fever as opposed to sepsis. They also did not show a significant benefit of antipyretic therapy in their patient population.

The HEAT trial (Permissive Hyperthermia Through Avoidance of Paracetamol in Known or Suspected Infection in ICU) is a multi-centre, randomised, double-blind placebo controlled trial currently underway in New Zealand. The trial compares intravenous paracetamol with 5% dextrose to control fever in 700 patients being treated for infection (surgical prophylaxis excluded) and having an axillary temperature of >38 ℃. Physical cooling methods will be added for temperatures over 39.5 ℃. The primary endpoint will be the number of ‘alive ICU-free days’, up to day 28 of the study. There are also various secondary outcomes to be taken into consideration. ²⁴ The results should add much needed data to the enigma of fever control in sepsis.

As many as 64% of ICU patients receive paracetamol; we are all using it, making it one of the most commonly prescribed drugs in Hospital. Yet, so little is actually known about the association between paracetamol administration and patient outcomes. Previous studies have demonstrated that antipyretic therapy was neither beneficial nor harmful in critically ill adults without neurological injury. A multicenter, retrospective observational study of 17,110 patients was conducted by Suzuki et al. ²⁵ They examined whether intravenous (IV) paracetamol influenced the outcome of their ICU patients – the primary outcome being in-hospital mortality and the secondary outcome, ICU mortality. Their results showed that administration of paracetamol appeared to be independently associated with reduced in-hospital mortality and time to death, after adjustment for multiple potential confounders and propensity scoring. This was broadly consistent across surgical and medical patients and was observed in febrile and non-febrile patients. Crucially however, this association was attenuated in the sub-group of medical patients with infection as the admitting diagnosis and lower illness severity scores pointing to potential indication bias. Along with this, it was not possible to control the interventions or decisions clinicians made (other cooling measures / other drugs, etc.) amongst the four ICUs.

It is possible that there is a ‘happy medium’ of febrile response permitting optimum / enhanced pathogen clearance, but avoiding the more deleterious effects of high fever. This may of course depend upon inter-individual patient co-morbidities and other cofounding variables such as the severity of illness according to physiological scoring, the pattern of lactate clearance, the level of multi-organ support and the response to other interventions, etc. It is here however, within this ‘happy medium’, where the potential role for TTM may lie.

Targeted temperature management

TTM post-cardiac arrest 36 ℃

Evolving from its predecessor therapeutic hypothermia, TTM has many established beneficial indications within modern medicine. The most notable of these is in the setting of post-cardiac arrest. Several landmark trials have established the potential beneficial effects of cooling on overall neurological outcome. ²⁶ Within these trials, it was noted that avoidance of hyperthermic states was of particular importance in terms of overall outcome and survival.^27,28 To date, these suggestions and observations have yet to make it to post-resuscitation care guidelines.

TTM in brain injury

In other areas of care, TTM is less well established. There have been multiple studies involving TBI showing mixed results, the largest of which showed no difference in mortality between normothermic and hypothermic groups. ²⁹ The study did however demonstrate decreased episodes of intracranial hypertension in the hypothermic group, which is why some neuro-ICU centres still use this as part of routine practice in this cohort of patients.

TTM in sepsis

Aside from the work in TBI and post-cardiac arrest patients, temperature control in sepsis is gaining a vast amount of interest. As briefly alluded to above, there has been an array of studies looking at the possible benefits and side effects of fever itself, as well as looking at the benefits and potential harms of anti-pyretic therapy.

There is currently scant work investigating the role of TTM in sepsis. The majority of the evidence comes from the Shortgen study mentioned above. In this RCT, patients with septic shock were either randomised to a non-cooling arm or a TTM arm aiming for normothermia. This demonstrated a decrease in early vasopressor usage in the first 48 h (most marked at 12 h) and a decrease in early mortality (days 0–14) in the TTM arm. From the outset this holds promise, however the mortality benefits were not demonstrable later on in the treatment time-course. By days 15–28 and at hospital discharge, there was no survival benefit. Looking in more detail at the study, there is limited information on the use of anti-pyretics in both arms, and relatively small volumes of fluid resuscitation were given to all subjects. This raises the possibility that many were relatively under-filled. This is of particular poignancy in the non-cooled arm, as these subjects were more likely to be hypovolaemic secondary to extra insensible losses associated with fever. Other positive findings of the study include earlier reversal of septic shock in the TTM arm, and relatively few adverse effects of TTM within the sepsis setting (accepting so many other confounding variables). But, the fact that late mortality was the same in both arms must highlight the need to ‘tread with care’, here.

Harm from TTM

But what of the potential harm that TTM could cause. In general, being hypothermic has detrimental effects upon patients and in sepsis is associated with poorer outcomes. This is postulated to be secondary to alteration of metabolic pathways, increased risk of coagulopathies, raised infection rates and cardiac dysrhythmias. These effects have been observed with regularity, particularly as TTM is becoming far more widespread in its usage in critical care. There are two large meta-analyses worth noting; both were performed in post-cardiac arrest patients.^30,31 One looked at bleeding risk and the other looked at all risks associated with TTM. Both meta-analyses came to the same conclusion; on balance, there was no increased clinical risk posed to patients from the negative effects of TTM. A point to note however is that this sub-set of patients have entirely different pathophysiology to septic patients. For example, those with sepsis have an increased risk of significant coagulopathy as well as a significant likelihood of developing further immunocompromise. Finally, Zhang conducted a systematic review and trial sequential analysis of randomized controlled trials looking at antipyretic therapy in the setting of sepsis. Their conclusion was that there was no overall benefit in controlling temperature in patients with already established sepsis. ³²

Cooling methods

There are various considerations to be made when deciding how to actively manage temperature. These come in the form of cost, availability of equipment, practicality, associated morbidity, internal or external cooling.

Many papers have examined internal and external methods, with several key outcomes. The majority of them suggest no difference in hospital mortality and neurological outcome between the two. ³³ Interestingly, there may also be little difference in morbidity / adverse event rates between traditional cooling methods (ice packs or hydrogel pads) vs. the more advanced regulated cooling blankets. ³⁴ Simple cooling techniques such as ice packs are associated with increased rates of overcooling, rebound hyperthermia, re-warming related hypoglycaemia and hypotension. ³⁵ Of note there is some evidence to suggest that intravascular cooling is associated with a decreased time to achieve a desired hypothermic temperature, as well as increased stability at a target temperature when compared to extra-vascular cooling. However, this decreased time does not show any increased survival benefit or improved neurological outcome. ³⁶

From a practicality point of view, there are benefits and drawbacks to each method. Intravascular cooling may permit tighter more rapid temperature control, but requires available skilled personnel to insert large invasive lines and carries an infection / bleeding risk. External cooling on the other hand can be performed by nursing staff, is non-invasive and quick, but can take longer to achieve target and be difficult to apply initially limiting further access to the patient. One caveat to this is that most of the studies cooled patients to hypothermia, rather than TTM and avoidance of hyperthermia (i.e. 36 ℃). Further trials are needed here.

Conclusion

It is clear that targeted temperature control has its benefits in certain patient sub-groups, exemplified by its usage in prevention of secondary brain injury from neurotrauma and cardiac arrest. However, in a similar parallel the ‘great fluid debate’, there is no steadfast guidance as to its utilisation in sepsis with persistent febrile states. Our experience with a protocolised approach to cooling in sepsis for refractory febrile states now spans many more cases. Our choice of temperature comes with little evidence base in this setting. It is similar to that used for post-cardiac arrest cooling and would permit relative metabolic ‘slow-down’. Published data demonstrates relative safety; therefore this seemed a sensible compromise, however we could equally have chosen normothermia. What is certain is that there is a vast amount of scope for further investigation into the role of TTM in sepsis, perhaps with tighter RCTs.