Cardiopulmonary bypass temperature and brain function


Dr M. Shaaban Ali


A debate has emerged in recently published studies about the optimum cardiopulmonary bypass temperature for good neurological outcome – warm vs. cold, i.e. normothermic vs. hypothermic. Although many comparative studies have been performed, the results of these studies are inconclusive and are difficult to interpret. Brain function has been studied in terms of neurological and neuropsychological outcome, protein S100β levels as a marker of brain damage, and cerebral oxygenation using jugular bulb oximetry and near-infrared spectroscopy. The studies produce no conclusive proof of the superiority of warm or cold cardiopulmonary bypass. However, it appears that any degree of bypass hypothermia (< 35 °C) may protect the brain. On the other hand, even a slight increase in bypass temperature to > 37 °C may cause marked brain injury.

The effect of cardiopulmonary bypass (CPB) temperature on brain function has been studied extensively. However, the results of the studies are difficult to interpret. Brain function has been investigated in terms of cerebral oxygenation, as monitored by jugular bulb oxygen saturation and near-infrared spectroscopy, serum markers of brain damage, e.g. protein S100β, and neurological and neuropsychological outcome. This article reviews the available literature examining the effects of CPB temperature on brain function.

Hypothermia has long been thought to provide a degree of cerebral protection during CPB. It provides this protection during periods of inadequate oxygen delivery by at least two mechanisms. Firstly, metabolic rate is directly related to temperature. Therefore, a low temperature increases cerebral tolerance to inadequate oxygen delivery. Secondly, a decrease in temperature (even by small amounts, e.g. to 34–35 °C) attenuates the release of glutamate and other excitatory amino acids from ischaemic neuronal cells. This phenomenon is thought to play an important role in hypothermic cerebral protection [1].

Ever since a publication by Lichtenstein et al. in 1991 [2] many clinicians have questioned the advantages of low-temperature CPB and have advocated warm or normothermic CPB (Table 1). The increased interest in using warm rather than cold CPB arises because it is possible to avoid prolonged rewarming, brain hyperthermia during rewarming, excessive postoperative bleeding and the ‘after-drop’– a decrease in patient temperature after the cessation of CPB. However, there is a potential risk of increased cerebral injury during warm CPB due to the lack of hypothermic brain protection.

Table 1.  Advantages and disadvantages of warm and cold cardiopulmonary bypass (CPB).
Warm cardiopulmonary bypassCold cardiopulmonary bypass
1 Decreased bypass duration due to shorter rewarming period [11, 13]1 Cerebral protection against ischaemia [48]
2 Avoidance of brain hyperthermia during prolonged rewarming [1, 9]2 Decreased oxygen requirements when flow to vital organs is decreased [49, 50]
3 Body temperature is maintained near normal3 Allows the use of lower flow rates during CPB [9]
4 Allows early tracheal extubation after CPB [13]4 Decreased anaesthetic requirements [51]
5 Less postoperative bleeding [52, 53] 
6 Rapid recovery of normal cardiac rhythm [54] 
1 Potential risk of increased cerebral damage [19]1 Impaired coagulation profile with increased bleeding [52]
2 Increased requirements for vasopressors during CPB to maintain mean arterial pressure > 50 mmHg [46]2 Prolonged rewarming with increased CPB duration [10, 11, 13]
3 Large doses of vasopressors may cause renal vasoconstriction [46]3 Risk of brain hyperthermia during rewarming [4]
4 Impaired cerebral oxygenation if rewarming is used early in CPB [29]4 Postoperative shivering due to incomplete rewarming and ‘after-drop’ in body temperature [55]
5 Uneven rewarming may lead to regional ischaemia [18] 
6 Altered drug metabolism [51] 
7 Hypothermia predisposes to arrhythmias [21] 

Temperature monitoring during cardiopulmonary bypass

The sites that can be used for monitoring patient temperature during CPB are listed in Table 2. Each of these monitoring sites has one or more limitations: tympanic membrane probes can perforate ear drums; oesophageal probes are influenced by ice or cold saline solutions used for myocardial protection; faeces can insulate rectal probes; the rate of urine production can affect bladder temperature. None of these sites has been shown to reflect cerebral temperature consistently and reliably.

Table 2.  Temperature measurement sites used by Stone et al.[3] for comparison with direct, invasive measurements in the cerebral cortex.
Tympanic membrane
Sole of the foot
Pulmonary artery

On theoretical grounds, the temperature in the jugular venous bulb should be a good substitute for direct brain temperature measurement, as blood leaving the brain capillaries will be in thermal equilibrium with nearby brain tissue. Therefore, the temperature of the jugular bulb should provide an accurate estimate of brain temperature during CPB, except that during cooling or rewarming there is likely to be a bias towards the temperature of the better-perfused regions of the brain. Regrettably, Stone et al. [3] did not include the jugular bulb as a measurement site in their comparisons with direct measurements of brain temperature. During rewarming from hypothermic CPB (27 °C), jugular bulb temperature has been shown to be higher than nasopharyngeal temperature [4, 5]. In addition, Johnson et al. [6] found that the mean [range] temperature of the arterial line of the bypass machine was 3.6[2.8–4.2] °C higher than the nasopharyngeal temperature in hypothermic patients. This is to be expected during rewarming because the brain is one of the better-perfused organs of the body [7]; its temperature will therefore increase more rapidly than most other tissues during rewarming and decrease more rapidly during cooling. Therefore, if temperature control is based on nasopharyngeal temperature, the brain may be become hyperthermic during rewarming unless the perfusate temperature is limited to 37 °C. On the other hand, during neurosurgical procedures using mild hypothermia (33–34 °C), and when temperature differences within the body are minimal, jugular bulb temperature is similar to core temperature as monitored by oesophageal and pulmonary artery probes [8]. Changes in brain temperature did not parallel tympanic membrane temperature in Stone's study, possibly because the authors did not directly visualise the outer ear and could therefore not be sure that the presence of significant amounts of wax was leading to erroneous measurements [3]. The jugular bulb is likely to be the most effective site for estimating brain temperature, but should be compared with direct measurement of cerebral cortex temperature when the latter can be obtained in an ethically acceptable scientific study.

Effects of cardiopulmonary bypass temperature on neurological outcome in adults

Many investigators have assessed the impact of CPB temperature on cerebral outcome in cardiac surgery patients and conflicting conclusions have been reached (Table 3). This may in part be due to different study designs, differences in the definition of ‘warm’ CPB or different methods used to determine adverse neurological outcome. Indeed, the definition of warm CPB is quite different in different studies, ranging from 33 to 37 °C. More recently, the use of what is called ‘tepid’ bypass temperatures (31–34 °C) has appeared in the literature [1, 9, 10]. Only studies of patients undergoing coronary artery bypass grafting are listed in Table 3. Attempts were made to perform a power calculation for neurological outcome in some studies [11, 12] but not in others [13–17]. The studies listed in Table 3 are heterogeneous: most (but not all) were randomised; bypass temperatures ranged from 33 to 37 °C for warm bypass and 23–32 °C for cold bypass; some used early rewarming in the warm CPB group, others did not. The incidence of stroke was used as the primary outcome in seven large studies and in three small studies; assorted tests of neurological and neuropsychological function were used in the other studies and as a supplementary comparison in four of the large studies. Stroke is usually due to particulate embolism from sources such as an aortic atherosclerotic plaque, a diseased valve or other debris from the operative field.

Table 3.  Neurological outcomes in studies comparing warm and cold cardiopulmonary bypass (CPB).
StudyCommentWarm CPBCold CPBSignificanceOtherneurologicaloutcomes
Temp.nIncidence of strokeTemp.nIncidenceof stroke
  • *

    Studies using early rewarming.

  • CT, Computerised tomography. Temp., temperature.

Singh et al.[56]Retrospective37 °C2585 1.0%25–30 °C16051.3%ns
Gaudino et al.[57]Not randomised≥34 °C1602 0.9%≥28 °C13851.0%nsMore extended brain damage in the CT scans of the warm CPB patients
Warm Heart investigators [21]Randomised33–37 °C860 1.6%25–30 °C8721.5%ns
Martin et al.[19]*Randomised≥35 °C493 3.1%≤28 °C5081.0%< 0.02Total neurological events: warm, 4.5% vs. cold 1.4%, p < 0.005
Rashid et al.[58]Randomised35 °C137 1.4%28 °C1442.8%ns
Engelman et al.[10]Randomised32–36 °C191 3.7%23–28 °C1006.0%nsNo significant difference in the infarct size or in the severity of neurological impairment between groups
Grigore et al.[59]Randomised35.5–36.5 °C117 1.7%28–30 °C1101.8%nsAdverse neurological outcome in 14.7% in the warm and 13.4% in the cold group. Corresponding figures for neurocognitive deficits were 39.3% and 39.1%)
Nandate et al.[60]Retrospective36–37 °C12827–28 °C122No significant difference in persistent postoperative neurological dysfunction (warm, 2.3% vs. cold, 4.1%.
Khatri et al.[20]Randomised35.5–36.5 °C11528–30 °C111Lower depression scores (p = 0.039) and lower anxiety levels (p = 0.008) in warm CPB patients
MacLean et al.[12]Randomised>34 °C78 3.0%≤28 °C776.0%nsNo significant difference in neurological and neuropsychological function
Grimm et al.[61]Randomised37 °C7232 °C72No significant difference in neuropsychological tests and subclinical cognitive function
Mora et al.[18]*Randomised≥35 °C6810.3%28 °C700%= 0.006No significant difference in neuropsychological outcome
Tönz et al.[13]Retrospective36 °C37 3.0%28 °C430%ns 
Shaaban Ali et al.[14]Randomised34 °C3028 °C30No significant difference in Mini-Mental State examination and protein S100β level
Plourde et al.[11]Randomised36 °C3028 °C24No significant difference in neuropsychological tests
Buschbeck et al.[15]Randomised36.5 °C1530 °C15No significant difference in neuropsychological tests
Graham et al.[16]Randomised34 °C1528 °C15No significant difference in neuropsychological tests
Wong et al.[17]Randomised≥34 °C1828 °C16No significant difference in neuropsychological tests
Sapin-Leduc et al.[62]Randomised35 °C1528 °C16No significant difference in neuropsychological tests

Two studies reported a significant difference in stroke incidence favouring cold CPB. However, one of them [18] simply used a subset of the patients in the other study [19]. On the other hand, one study [20] found significantly less depression and anxiety in those undergoing warm CPB. However, the differences were generally small and, with the large number of non-significant results, it seems that there is no consistent, clear advantage to cold CPB in terms of neurological protection.

In the study by Martin et al. [19] the incidence of total neurological events and peri-operative strokes was significantly higher in the warm group than in the cold group. Neurological events included peri-operative stroke, peri-operative encephalopathy and delayed cerebrovascular accidents. By contrast, the Warm Heart Investigators (WHI) reported no differences between warm and cold CPB [21]. Differences in these two studies that may have affected the results include the difference in CPB technique and in the in-patient populations studied. Martin et al. [19] maintained CPB temperature at > 35 °C, whereas in the WHI investigators' study, CPB temperature ranged from 33 to 37 °C. Also, in Martin et al.'s study, the duration of CPB was longer in warm CPB patients, and high-risk patients, such as diabetics and those undergoing redo surgery, were included. These factors might explain the poorer neurological outcome in Martin et al.'s study.

Why doesn't cold cardiopulmonary bypass offer more brain protection?

The failure of cold CPB to offer a clear advantage over warm bypass in terms of brain protection may be explained by the following:

  • • Most of the embolic load to the brain occurs during aortic clamping and off-clamping, i.e. at the beginning and end of bypass – during these periods the body temperature is similar during both warm and cold CPB [12].
  • • During rewarming from cold CPB, the brain may be exposed to extended periods of hyperthermia that exacerbate brain injury [18, 22, 23]. Similarly, continuous warming used to maintain systemic temperature > 35 °C in both Martin et al.'s [19] and Mora et al.'s [18] studies may involve the risk of exposing the brain to periods of hyperthermia.
  • • Even mild temperature increases to above normal may be markedly deleterious; for example, the volume of cerebral infarction increases substantially at 39 °C compared with lower temperatures [23, 24]. Conversely, mild degrees of brain cooling (2–4 °C) confer dramatic protection from ischaemic brain injury secondary to a decrease in extracellular glutamate levels [24–26]. This may account for the lack of increased risk of neurological injury in warm CPB studies using temperatures 2–4 °C below than 37 °C [12, 21]. However, strict normothermia with continuous warming is associated with increased incidence and severity of cerebral injury [18, 19].

Warm vs. cold cardiopulmonary bypass and protein S100β levels

Protein S100β is a potentially useful early marker of brain damage. It has been used in attempts to differentiate between the benefits and adverse effects of different bypass temperatures [14, 27, 28]. Gao et al.[27] and Tonninger et al.[28] found similar protein S100β levels after cold and warm bypass using the same CPB temperatures (32 °C and 37 °C, respectively). In addition, no significant differences were observed in the protein S100β level between warm (34 °C) and cold CPB (28 °C) in a similar study [14].

Warm vs. cold cardiopulmonary bypass in terms of cerebral oxygenation

Jugular bulb oximetry

In 1994, Cook et al.[29] studied cerebral venous saturation in patients randomly allocated to warm (37 °C) or cold (27 °C) cardiopulmonary bypass. Jugular bulb oxygen saturation (Sjo2) ≤ 50% occurred in the first 40 min of warm CPB in 54% of patients and during rewarming from cold CPB in 12% of patients (Table 4). In addition, arterial jugular oxygen content difference during CPB was greater in the warm group than in the cold group, which is consistent with higher oxygen consumption during normothermia. Similar results were reported by Okano et al.[30]: Sjo2 < 50% within 40 min of the start of warm CPB (> 35 °C), but no change in Sjo2 in patients with mild hypothermic bypass (32 °C). However, the number of patients showing desaturation in each group was not mentioned. Kadoi et al.[31] found more frequent episodes of Sjo2 desaturation during the early part of warm CPB (> 35 °C) than during rewarming from cold CPB.

Table 4.  Number of patients with jugular bulb oxygen saturation (Sjo2) < 50% during warm and cold cardiopulmonary bypass in published studies.
 Warm cardiopulmonary bypassCold cardiopulmonary bypass
Bypass temperature; °CNo. of patients with Sjo2 < 50%Bypass temperature; °CNo. of patients with Sjo2 < 50%
Cook [29]37 °C14/2627 °C3/26
Kadoi et al.[31]35 °C10/1530 °C5/15
Shaaban Ali et al.[33]34 °C3/3028 °C5/30

Jugular venous desaturation at the beginning of warm CPB may be due to haemodilution, a decrease in mean arterial pressure, a decreased prebypass temperature being increased at the start of CPB, and increased embolic load due to aortic manipulation during this period. Croughwell et al. observed that desaturation was associated with a greater decrease in mean arterial pressure [32]. In addition, they suggested that impaired cerebral oxygenation with neuropsychological dysfunction after CPB may be due to cerebral emboli. These emboli may result in subclinical neurological damage without evidence of stroke [32]. However, our group [33] found that more patients in the cold CPB group showed evidence of desaturation than those in the warm CPB group (Table 4), which may be attributable to the avoidance of early rewarming and the use of bypass temperatures as low as 34 °C in the warm group.

Near-infrared spectroscopy

Amory et al.[34] used near-infrared spectroscopy to compare cerebral oxygenation between warm (35 °C) and cold (30 °C) CPB, and found minimal differences between the two groups. They found that total haemoglobin concentration was significantly lower in the cold bypass group at 10 min after the start of CPB, and both before and 10 min after aortic cross-clamping. Haemoglobin concentrations showed little change during CPB without any significant difference between warm and cold groups [31, 34] in a study using a device capable of measuring regional cerebral oxygen saturation (rSO2) during warm (> 35 °C) and cold (30 °C) CPB. These authors found that rSO2 was stable throughout the peri-operative period in the cold bypass group, but decreased during warm CPB when compared with pre-bypass values. The decrease in rSO2 during warm CPB in Kadoi et al.'s study [31] can be attributed to continuous rewarming from the start of CPB to maintain the nasopharyngeal temperature above 35 °C.

Bypass temperature and the brain during paediatric cardiac surgery

Neurological deficit is common after paediatric cardiac surgery with hypothermic CPB, with an incidence of 2–25%[35]. Fallon et al.[36] found that 5% of children had a neurological event in the immediate postoperative period. Several paediatric centres have assessed neurological outcome after hypothermic CPB (moderate or profound), but there are a few studies of warm CPB in this age group. Corno et al.[37] used a CPB temperature of 37 °C and a high bypass flow of 3.4 l.m−2.min−1 in 77 children undergoing open-heart surgery without any reported neurological complications. However, no control group was included. In a pilot study [38], we studied cerebral oxygenation and protein S100β levels during warm and cold CPB in children and found no significant difference in the levels between groups. However, cerebral oxygenation in terms of tissue oxygen index was significantly impaired during rewarming from cold CPB. Also, five patients were desaturated (tissue oxygen index < 50%) during rewarming in the cold CPB group compared to two of the warm CPB patients. Others have found no significant differences in serum protein S100β and neurone-specific enolase between CPB conducted at temperatures of 25 °C vs. 36 °C in children during repair of congenital heart disease [39]. These studies suggest that warm CPB causes little brain damage in children undergoing open heart surgery, but further studies of short-term and long-term neurological outcome are warranted.

However, a slower rewarming rate with lower peak temperatures during CPB has been associated with better neurocognitive outcome and improvement in Sjo2 in adults [40–45].

Grigore et al. found that slow rewarming, maintaining no more than a 2 °C difference between nasopharyngeal and CPB perfusate temperature, was associated with better cognitive performance 6 weeks after surgery than conventional rewarming that maintained a 4–6 °C gradient between nasopharyngeal and CPB perfusate temperature [41]. Whether similar results will be obtained in children undergoing open heart surgery remains to be seen.

Patients who were warmed rapidly had a greater decrease in Sjo2 than those who were warmed slowly [40, 42, 43]. Sapire et al. found that during rewarming there was a marked increase in anaerobic metabolism [40]. In addition, van der Linden et al.[44] and von Knobelsdorff et al.[45] found that mean blood flow velocity measured in the middle cerebral artery by transcranial Doppler ultrasonography was increased when rewarming was started but was also associated with Sjo2 desaturation. These findings suggest that the increase in cerebral blood flow was inadequate to meet the increasing cerebral metabolic oxygen demand during rewarming, and that this imbalance was reflected by Sjo2 desaturation.

Summary of the effects of warm vs. cold cardioplumonary bypass

Neurological studies after warm and cold CPB have failed to demonstrate that cold CPB provides a greater protective effect than warm CPB either in terms of neurological outcome or protein S100β release. On the other hand, more patients show early Sjo2 desaturation during warm than cold CPB. This paradox can be explained by the following:

  • • Poor cerebral oxygenation in some of the warm CPB studies can be attributed to continuous warming from the beginning of bypass to maintain bypass temperature > 35 °C or strictly at 37 °C [30, 31, 46]. This finding is supported by poor neurological outcomes in warm CPB studies that used continuous rewarming to keep bypass temperature > 35 °C [18, 19]. This is in contrast to the lack of any significant differences in outcome between warm and cold CPB in those studies not employing this technique [12, 21], allowing bypass temperatures to decrease to as low as 34 °C.
  • • During rewarming from cold CPB, the brain can be exposed to periods of hyperthermia. This rewarming period is associated with impaired cerebral oxygenation [40]. These periods of hyperthermia have been claimed by Murkin [47] to be responsible for 50–80% of the neuropsychological dysfunction seen after cardiac surgery. Therefore, better rewarming strategies should be used to avoid this damage. Slow rewarming has been shown to decrease cerebral injury [41] and to avoid Sjo2 desaturation [42].
  • • The most significant issue is the lack of accurate monitoring and control of brain temperature during cardiac surgery, especially during rewarming. The site that most closely reflects brain temperature is the jugular bulb. The addition of a temperature sensor to all jugular bulb catheters might allow more accurate monitoring.


It is clear that there is no consistent advantage to cold CPB over warm or normothermic CPB. However, despite a great deal of research, the neurological safety of so-called ‘normothermic’ CPB has yet to be conclusively demonstrated. In addition, any drift in bypass temperature to 2–3 °C below 37 °C seems to offer comparable brain protection to formal hypothermia (25–30 °C). It is worth noting that the currently used temperature monitoring sites may grossly underestimate actual brain temperature. Excessive rewarming can expose the brain to periods of hyperthermia that may cause an imbalance between cerebral oxygen supply and demand, and may therefore exacerbate neuronal injury. Therefore, rewarming strategies that can avoid brain hyperthermia should be explored. In addition, sites for monitoring temperature that best reflect brain temperature, e.g. the jugular bulb, should be used during cardiac surgery.