. 6
( 8)


ability transition pores (PTP) on the inner mitochondrial membrane. Opening of the mito-
chondrial PTP channels during early reperfusion (they remain closed during the ischemic
period) inhibits the mitochondrial membrane potential, uncoupling oxidative phosphor-
ylation, which being essential for ATP production, results in ATP depletion and cell death.
Large quantities of oxygen free radicals are generated on reperfusion of ischemic tissue. The
oxygen free radicals, if present in sufficient concentration, overwhelm endogenous scavenging
Chapter 11: Organ damage in cardiopulmonary bypass

mechanisms and cause further intracellular injury. Oxygen free radicals also exacerbate ara-
chidonic acid metabolism and the production of leukotrienes and thromboxanes, promoting
aggregation, transmigration and activation of neutrophils to further compound the injury
(see Figure 11.4). Neutrophils are the key final mediators of IRI by the production of toxic
chemicals generated during the metabolism of oxygen and by the secretion of proteolytic
enzymes released from granules stored in their cytoplasm.
During ischemia energy generation using high-energy phosphates (ATP) creates the
metabolite hypoxanthine, which has a tendency to accumulate. The enzyme xanthine dehy-
drogenase normally metabolizes hypoxanthine. Under conditions of ischemia followed by
rapid reperfusuion, xanthine dehydrogenase is converted to xanthine oxidase as a result of
the higher availability of oxygen. This oxidation results in molecular oxygen being converted
into highly reactive superoxide and hydroxyl radicals. Excessive nitric oxide produced during
reperfusion reacts with superoxide to produce the potent free radical peroxynitrite. These
radicals attack cell membrane lipids, proteins and DNA, causing further damage.



Generation of
oxygen free
Hypercontractility Activation of radicals
mitochondrial PTP

Phospholipase A2 activation

Decrease in ATP Leukotrienes and thromboxane A2

Neutrophil Neutrophil
local blood flow
attraction activation

Cell death
Myocardial contractile dysfunction

Figure 11.4 Mechanism of myocyte dysfunction and death on reperfusion of previously ischemic tissue.

Chapter 11: Organ damage in cardiopulmonary bypass



Figure 11.5 The role of endotoxins from gut translocation in the pathogenesis of the inflammatory response to
CPB. Lipopolysaccharides (LPS) bind to LPS-binding protein (LBP). This complex activates macrophages releasing
TNF and protein kinase.

The plasma levels of endotoxins increase during CPB. Endotoxins are lipopolysaccharides
(LPS) derived from the cell membranes of gram-negative bacteria. The source of the endo-
toxins is widely believed to primarily be the gastrointestinal tract. Reduction in blood flow
through the splanchnic circulation during CPB and the SIRS associated with CPB and aortic
cross-clamping may result in a breakdown of the mucosal barrier in the gastrointestinal tract
with consequent translocation of endogenous bacteria into the circulation.
The subsequent breakdown of bacterial cells releases LPS, which are bound to LPS-
binding protein (LBP). The LPS“LBP macromolecular complex is a highly potent activator
of macrophages, which release TNF and protein kinase, thus exacerbating SIRS further (see
Figure 11.5).

Therapeutic strategies
A number of strategies have been employed to ameliorate the extent of the SIRS seen during
• Pharmacological: using steroids prior to the onset of CPB, antioxidants and
proteolytic enzyme inhibitors. None of these interventions have had clinically
meaningful impact.
• Heparin-bonded circuitry can be used with the intention of reducing the degree of
complement activation, but has proven to be less effective in attenuating coagulation or
• Hemofiltration/ultrafiltration (convection and osmosis under hydrostatic pressure) has
been incorporated into the circuitry to remove low-molecular-weight substances from
plasma with the aim of reducing the circulating levels of proinflammatory mediators.
Current techniques have proven to be more effective in the pediatric than the adult
• Leukocyte-depleting filters incorporated in the CPB circuit, to reduce the number of
circulating activated white cells. Their value is presently unclear, but leukocyte
depletion may have a protective effect in reducing the severity of lung and myocardial
injury observed post-CPB. The most consistent benefit is found in higher risk
patients with pre-existing lung disease, ventricular dysfunction or those receiving
long CPB times.
Chapter 11: Organ damage in cardiopulmonary bypass

(ml/minute/per 100 g tissue)

CPB 2.3 l/(minute·m2)
O2 delivery



Sm a ll
a Brain Kidney Pancrea s Muscl e
bo w el


* Brain

* Kidney
* * *
50 Small bowel


* *
1.3 1.5 1.7 1.9 2.1 2.3
CPB flow rate
Figure 11.6 (a) Regional O2 delivery with onset of cardiopulmonary bypass. (b) Change in organ DO2 with changes
in flow rate (Q). Mean values ± SE *p <0.05 versus CPB at 2.3 l/minute/m2.

• Leukocytes in allogenic blood transfusions have important immunomodulatory effects
in the recipient. The use of leukocyte depleted stored blood has been shown to decrease
mortality in some patients who undergo CPB. This is predominantly due to a decrease
in non-cardiac causes of death, in particular multiorgan failure.
Chapter 11: Organ damage in cardiopulmonary bypass

CPB versus OPCAB
Off-pump CABG is an alternative technique for coronary revascularization, which is still
controversial and highly dependent on institutional preference. Clinical reports have shown
that oxidative stress and markers of inflammation (particularly IL-8, TNF and E-selectin) are
significantly reduced during OPCAB when compared to CABG performed on CPB. OPCAB
has also been shown to be associated with a reduction in blood transfusion. Full hepariniza-
tion may be avoided during OPCAB and this, together with the avoidance of hemodilution
by the priming volume required for CPB, may be more important in reducing transfusion
requirements than any advantageous effects from ameliorating SIRS.

Alterations in organ perfusion
The distribution of blood flow to organs is altered on transition from physiological circulation
to CPB and thus oxygen delivery also alters (see Figure 11.6a). Furthermore, tissue oxygen
delivery is influenced to a large extent by CPB flow rate (see figure 11.6b). Organ dysfunction
may, thus, be in part attributed to these changes in the regional distribution of blood flow and
the dependence of oxygen delivery on the maintenance of adequate CPB flow rates.

Gastrointestinal complications
Gastrointestinal (GI) complications are reported as occurring in 2“4% of patients follow-
ing cardiac surgery. There is a high associated mortality rate of about 30%; GI complications
account for about 15% of all cardiac surgical deaths.
GI complications usually present as GI bleeding, peritonitis or acute bowel obstruction
with abdominal distension. Perforation or ischemic bowel is a common finding. Often, the
earliest presenting sign is a progressive metabolic acidosis. Bleeding from the upper GI tract
accounts for almost 30% of all GI complications after cardiac surgery and is far more com-
mon than bleeding from the lower GI tract. Lower GI tract bleeding is usually associated with
bowel ischemia or pre-existing large bowel disease.
CPB causes profound reductions in blood flow in the splanchnic circulation and thus
leads to reduced perfusion to the GI tract and associated organs. Gastrointestinal mucosal
blood flow is reduced and remains reduced for several hours postoperatively. This may be
further exacerbated by the use of vasoconstrictors during CPB and by embolization of athero-
matous debris from the aorta or clot from the heart into the mesenteric circulation. Severe
intestinal ischemia may occur during CPB even when the indices of global body perfusion
remain normal. The release of a variety of vasoactive factors during CPB, such as vasopressin,
catecholamines and thromboxanes, lead to a redistribution of regional blood flow away from
the mucosa of the GI tract.
The combination of reduced splanchnic blood flow and the CPB-induced SIRS reduce the
efficacy of both the absorptive and barrier functions of the GI tract. The increase in gastroin-
testinal mucosal permeability results in the translocation of bacterial endotoxins from the GI
tract into the bloodstream, amplifying SIRS and subsequent further organ damage. Risk fac-
tors associated with adverse gastrointestinal (GI) outcome are summarized in Table 11.1.

Hepatic dysfunction
Hepatic metabolism is reduced during CPB in conjunction with the reduction in splanchnic
blood flow. Hepatic blood flow has been reported to decrease by 19% after CPB is commenced.
Chapter 11: Organ damage in cardiopulmonary bypass

Table 11.1. Risk factors associated with adverse gastrointestinal (GI) outcome

Models for adverse GI outcome (n = 133): odds ratio (95% CI)
Variable Preoperative model Intraoperative model Both models
Increased preoperative total 2.4 (1.2“4.9) 2.5 (1.2“5.0)
bilirubin >1.2 mg/dl*
Combined cardiac procedures 2.9 (1.8“4.5) 2.0 (1.3“3.2)
Preoperative platelets 130 000/μl 2.9 (1.3“6.4) 2.8 (1.3“6.2)
Previous cardiovascular surgery 3.0 (1.9“4.7) 2.2 (1.4“3.4)
Preoperative EF <0.4* 1.9 (1.2“3.0) 1.7 (1.1“2.6)
Age >75 years 1.8 (1.1“2.9)
Preoperative PTT >37 seconds 2.0 (1.3“3.1) 1.7 (1.1“2.5)
Pharmacological cardiovascular 1.9 (1.3“2.9) 2.0 (1.3“3.2)
Intraoperative transfusion of PRBCs 1.9 (1.6“2.3) 1.7 (1.4“2.1)
Intraoperative circulatory failure 1.7 (1.1“2.6)
Aortic cross-clamp time 1.5 (1.0“2.3)
CI = confidence interval; EF = ejection fraction; PTT = partial thromboplastin time; PRBCs = packed red blood cells.
Missing data were included as low risk on justification.
Adapted from Multicenter Study of Perioperative Ischemia Research Group. Anesth Analg 2004; 98: 1610“17.

There may be a transient rise in the levels of hepatic enzymes measured in the blood, which
usually peaks early in the postoperative period. Clinically evident jaundice is only apparent in
a small number of patients, although bilirubin levels rise in about 20% of cases. Moderate or
severe degrees of hepatic dysfunction are rare and usually occur in concert with multiorgan
There is poor correlation between preoperative hepatic function and the risk of develop-
ing postoperative hepatic dysfunction. Consequences of hepatic dysfunction relevant to CPB
specifically include:
• impaired drug metabolism;
• reduced plasma protein concentrations leading to reduced plasma oncotic pressure and
alteration in the volume of distribution of drugs;
• impaired coagulation due to reduction in production of clotting factors; and
• impaired ability to generate heat and regulate temperature.
Cholecystitis, in the absence of gall stones, may also occur postoperatively in 0.2“0.5% of
all cardiac surgical patients. The gall bladder distends and the stasis of bile leads to inflamma-
tion of the gall bladder; it carries a mortality rate of 25“45% once diagnosed, despite aggres-
sive treatment.

Overt pancreatitis, characterized by a rise in serum amylase to over 1000 IU/l, occurs
in 0.1“1% of cases following cardiac surgery. Lesser degrees of pancreatic cellular injury
with mild elevations in serum amylase concentrations are, however, common. The etiology
Chapter 11: Organ damage in cardiopulmonary bypass

is probably related to perioperative reduction in splanchnic blood flow causing pancreatic
ischemia. Risk factors for developing postcardiac surgical pancreatitis are:
• prolonged CPB;
• perioperative hypotension;
• low postoperative cardiac output;
• hypothermia; and
• perioperative administration of large quantities of Ca2+.
Ca2+ administration, frequently used to treat intraoperative hypertension, has not clearly
been identified as an independent risk factor.
Uncomplicated pancreatitis carries a mortality of 5“10%, but cases that progress to
necrotizing pancreatitis or to the development of abscesses or a pseudocyst usually result
in death.
There is limited evidence that OPCAB surgery has significant benefits over cardiac sur-
gery with CPB in preventing GI complications

Pulmonary dysfunction
Cardiac surgery results in impairment of gas exchange for a variety of reasons. While
most patients will display subclinical functional changes, the incidence of post-CPB
acute respiratory disease syndrome (ARDS) is <2%. The mortality rate associated with
post-CPB ARDS, however, is >50%. The principal causes of postoperative respiratory
failure are:
• atelectasis;
• increase in lung water content as a result of:
· increased capillary permeability caused by SIRS
· impaired hemodynamics in the immediate postoperative period
· the additional fluid load during CPB;
• alterations in the production of surfactant, particularly during the period of lung
collapse during CPB and as a result of SIRS;
• transfusion-related acute lung injury (TRALI);
• altered chest wall mechanics resulting from sternotomy;
• decreased static and dynamic lung compliance;
• pneumothorax or hemothorax; and
• phrenic nerve injury impairing diaphragmatic function.
Although some of the factors listed above relate to cardiac surgery in general and are not
specific to CPB, the overall effect is the development of intrapulmonary shunts that cause
a mismatch between ventilation and perfusion. This manifests as a higher inspired oxygen
concentration being required to maintain an acceptable level of blood oxygenation. This mis-
match tends to resolve gradually postoperatively, but patients may require supportive meas-
ures such as the application of PEEP during mechanical ventilation or continuous positive
airway pressure (CPAP) when spontaneously breathing until resolution occurs. The main-
tenance of adequate tidal volumes without reaching excessive airway pressures during sup-
ported respiration may help to limit atelectasis. The administration of diuretics as an adjunct
to careful fluid balance may help to reduce interstitial lung water.
Lung injury may be more evident following cardiac surgery in patients with pre-existing
lung disease and in smokers.
Chapter 11: Organ damage in cardiopulmonary bypass

The therapeutic interventions investigated have had little or no effect on postoperative
lung function:
• steroid administration before CPB fails to prevent poor postoperative lung
• leukocyte depleting filters in the CPB circuitry show inconsistent effects on post-CPB
lung function;
• heparin-coated circuits and continuous hemofiltration on CPB improve pulmonary
vascular resistance and transpulmonary shunting in a transient and clinically insignifi-
cant way; and
• maintaining mechanical ventilation on CPB does not lead to any significant preserva-
tion of lung function.
There is conflicting evidence as to the benefits of avoiding CPB and electing for OPCAB in
terms of postoperative pulmonary dysfunction. However, there is a suggestion that patients
with chronic pulmonary disease benefit from OPCAB more in terms of preservation of lung
function than those with no significant pre-existing lung disease.

Myocardial dysfunction
The period of CPB during cardiac surgery can be divided into three phases:
• onset of CPB until application of the cross-clamp;
• the period of cross-clamping and cardioplegic or fibrillatory arrest; and
• the reperfusion period following removal of the cross-clamp and ultimately separation
from CPB.
During these periods the heart is subjected to injury from microemboli, the inflamma-
tory products of SIRS, regional hypoperfusion, complete ischemia and finally reperfusion
injury. The injurious effects incurred from these insults, together with the potential for inad-
equate myocardial protection and distension of the flaccid heart during the period of
cross-clamping, result in myocardial edema and reduced ventricular contractility, which
may continue into the postoperative period. Furthermore, if the heart is subject to excessive
preloading or high afterloading during weaning from CPB, left ventricular end-diastolic
volume, myocardial wall stress and oxygen consumption are all increased, further contributing
to deterioration in cardiac function.
OPCAB avoids the need for cross-clamping of the aorta and for cardioplegia, which are
both essential during CABG with CPB. In theory, this should minimize the risk of global
myocardial ischemia and myocardial stunning. In practice, the incidence of myocardial
infarction is similar following OPCAB and CABG with CPB. OPCAB, however, is associated
with a more rapid recovery of myocardial oxidative metabolism and this leads to more rapid
replenishment of myocardial high-energy phosphates such as ATP and so to better myocar-
dial function in the early postrevascularization phase.

Quality of life
The ultimate measure of the success of a medical intervention is its ability to improve the qual-
ity of life for patients. There is much evidence to show that both mental and physical health
are improved after cardiac surgery and furthermore studies comparing quality of life after
cardiac surgery with CPB or OPCAB yield similar results. This is a testament to the fact that
CPB provides a safe and effective means for performing cardiac operations.
Chapter 11: Organ damage in cardiopulmonary bypass

• Ngaage DL. Off-pump coronary artery bypass
Suggested Further Reading grafting: the myth, the logic and the science.
• Laffey JG, Boylan JF, Cheng DCH. Eur J Cardiothor Surg 2003; 23(4): 557“70.
The systemic inflammatory response to
• Sellke FW, DiMaio M, Caplan LR, et al.
cardiac surgery. Anesthesiology 2002; 97:
Comparing on-pump and off-pump
coronary artery bypass grafting. Circulation
• McSweeney ME, Garwood S, Levin J, et al. 2005; 111: 2858“64.
Adverse gastrointestinal complications
• Wan S, LeClerc J, Vincent J. Inflammatory
after cardiopulmonary bypass: can
response to cardiopulmonary bypass:
outcome be predicted from perioperative
mechanisms involved and possible therapeutic
risk factors? Anesth Analg 2004; 98:
strategies. Chest 1997; 112: 676“92.
• Yellon DM, Hausenloy DJ. Myocardial
• Ng CSH, Wan S, Yim APC, et al. Pulmonary
reperfusion injury. N Engl J Med 2007;
dysfunction after cardiac surgery. Chest
2002; 121: 1269“77.

Cerebral morbidity in adult

cardiac surgery
12 David Cook

Neurological complications in adult cardiac surgery
Postoperative brain injury has been a focus of attention since the inception of cardiac surgery.
In the last decade, approximately 2000 English-language articles have been published in this
area. However, there has been only a modest decrease in the incidence of stroke or encepha-
lopathy and the syndrome of cognitive dysfunction may be more prevalent today than 20
years ago. In fact, McKhann, comparing stroke incidence in 1994 (2.9%) and in 2004 at Johns
Hopkins Hospital in Baltimore found a greater incidence in 2004 (4.5%) presumably because
of the greater level of pre-existing disease in the more recent group of patients presenting
for cardiac surgery. A great deal, however, has been learned about brain physiology and
mechanisms of injury in cardiac surgery; changes in practice have occurred that are probably
making neurological outcomes better than might be predicted, given the increasing age and
associated medical conditions prevalent in the surgical population.
In the 1980s and 1990s, a large number of physiological and clinical studies were con-
ducted that better characterized brain physiology and function during cardiac surgery and
cardiopulmonary bypass (CPB). The physiological variables that were investigated as possible
causes of perioperative brain injury included mean arterial pressure (MAP), body tempera-
ture, hematocrit (HCT), bypass pump flow rate, the use of pulsatile flow and CO2 manage-
ment. During this same period pharmacological and physiological interventions and changes
in surgical technique were also investigated with an eye to reducing neurological morbidity.
While many of those investigations can be faulted for being statistically underpowered, two
decades of research have not led to either a brain protectant drug or device that has become
part of our routine practice and substantially improved neurological outcomes. In fact, there
is little evidence that intraoperative physiological management itself is an independent deter-
minant of neurological outcome.

Cerebral physiology during cardiopulmonary bypass
The results of a large number of investigations can be best summarized by saying that in
adults, over the range of conditions in which nearly all CPB is conducted, the determinants of
cerebral blood flow and metabolism are the same as those under non-bypass conditions.
CPB may profoundly affect cerebral blood flow (CBF) and cerebral metabolic rate of
oxygen consumption (CMRO2), but these changes are qualitatively no different from those
that would occur under non-CPB conditions; they are simply quantitatively greater. During
CPB conducted above 27°C, which constitutes about 90% of adult surgery, brain physiology
is straightforward and predictable. However, when bypass is conducted under moderately to
profoundly hypothermic conditions some of these relationships change, primarily because of
the non-linearity of changes in CMRO2 and a relative cold-induced vasoparesis.
Cardiopulmonary Bypass, ed. S. Ghosh, F. Falter and D. J. Cook. Published by Cambridge University Press.
© Cambridge University Press 2009.
Chapter 12: Cerebral morbidity in cardiac surgery

The determinants of cerebral perfusion during cardiopulmonary bypass are, in order of
• mean arterial blood pressure;
• hematocrit;
• cerebral metabolism; and
• PaCO2.
The absence of pulsatility does not determine CBF nor does pump flow, independent of
its effect on MAP.
In the past there was considerable confusion over the effect of mean arterial pressure on
cerebral perfusion during CPB. This arose from the poor design of studies conducted in the
1980s and a failure to appreciate the profound effect that changes in HCT have on cerebral
blood flow. At least two very prominent studies from the 1980s concluded that CBF was inde-
pendent of MAP, to MAPs as low as 30“40 mmHg during CPB. This conclusion was based on
pooling very few measurements of CBF from large numbers of patients. Because the measure-
ments were conducted at multiple MAP, temperature, HCT and PCO2 conditions, in multiple
patients, a great deal of scatter was demonstrated in the data. When regression analysis relat-
ing MAP and CBF was performed on the widely scattered data, no relationship between MAP
and CBF could be identified. The study design was thus not adequate to test the hypothesis
and the conclusion was misleading.
The other primary source of confusion about the relationship between MAP and CBF
arose from a failure to appreciate the profound effect of HCT on CBF. A variety of investiga-
tions determined CBF before CPB. Then during CPB, at a significantly lower MAP (below
55 mmHg), CBF measurements were repeated and found to be nearly the same as CBF at a
higher MAP prior to CPB. These studies failed to take into account the fact that significant
hemodilution occurs during CPB and that this reduces blood viscosity and increases CBF.
This was well elucidated by Plöchl and Cook who randomized exposure to varying MAP in
dogs during CPB at 33°C and found that, while CBF was increased for any given degree of
hemodilution, CBF and cerebral oxygen delivery decreased when MAP fell below approxi-
mately 55 mmHg (see Figure 12.1).
After MAP and HCT, cerebral metabolism is a primary determinant of cerebral blood
flow. Over the temperature range where most adult bypass is conducted, 27°C to 37°C, there
is a clear relationship between temperature, CMRO2 and CBF. Below 25°C, the relationships
become much more complex. If all other variables (primarily MAP, HCT and CO2) are con-
trolled, a 10°C decrease in temperature reduces CMRO2 by about 50% and this is associated
with a 50% reduction in CBF.

Figure 12.1 Cerebral oxygen delivery (CDO2) and
cerebral oxygen consumption (CMRO2) versus mean
arterial pressure (MAP) during cardiopulmonary bypass
(CPB) at 33°C. Values for oxygen (on ordinate in ml/100 g
per minute) are the mean ± standard deviation (*P
<0.05 versus MAP of 60 mmHg by repeated-measure
analysis of variance followed by Student“Neuman“
Keuls test). (From Plöchl W, Cook DJ, Orszulak TA, et al.
Critical cerebral perfusion pressure during tepid heart
surgery in dogs. Ann Thorac Surg 1998; 66: 118“124, with

Chapter 12: Cerebral morbidity in cardiac surgery

PaCO2 is an independent determinant of CBF during bypass. However, during most adult
cardiac surgery the effect of PaCO2 is relatively small. If all other variables are controlled, every
1 torr increase or decrease in PaCO2 alters CBF approximately 3%. As such, between 32°C
and 37°C the maximal effect of CO2 on CBF is about 15%. The effect of CO2 and alpha-stat/
pH-stat strategies becomes increasingly relevant below 27°C, but it is a minor consideration
above 32°C.
Cerebral physiology during CPB has been somewhat difficult to determine because so
many variables are subject to change simultaneously and some of the physiological variables
interact. This is clear in the interactions of HCT, MAP and CBF. The same physiological link-
age of variables also leads to confusion about the effect of pump flow on CBF. Some literature
has reported that cerebral perfusion is dependent on pump flow. This misunderstanding arose
from the failure to appreciate that pump flow, like cardiac output, is a primary determinant
of mean arterial pressure. While decreases or increases in CBF may be seen when pump flow
is increased or decreased, this is really only clearly demonstrated below or near the autoregu-
latory threshold: above a MAP of approximately 55 mmHg, increases in pump flow do not
increase CBF, while below about 55 mmHg reductions in pump flow result in reductions in
MAP that then lead to reductions in CBF. This was well demonstrated in an animal study by
Sadahiro (see Figure 12.2).
The dependence of CBF on pump flow is seen only when pump flow is too low to generate
a MAP above the autoregulatory threshold. In humans and in animals, the independence of

Figure 12.2 Continuous monitoring of perfusion pressure and CBF during perfusion flow rates from 80 to 10 ml/
kg/minute. Arrows indicate the point at which the relationship between CBF and perfusion pressure was evaluated.
Black arrows show the presence of an autoregulatory response with CBF returning to its prior level after an initial
drop. White arrows show the loss of a vascular response. (From Sadahiro M, Haneda K, Mohri H. Experimental study
of cerebral autoregulation during cardiopulmonary bypass with or without pulsatile perfusion. J Thorac Cardiovasc
Surg 1994; 108: 446“454, with permission.)

Chapter 12: Cerebral morbidity in cardiac surgery

Table 12.1. Effect of pump flow on cerebral perfusion

30 adult patients, CPB at 27°C High flow Low flow
Pump flow (l/minute/m2) 2.3 ± 0.1 1.2 ± 0.1
MAP (mmHg) 63 ± 9 62 ± 6
CBF (ml/100g/minute) 29 ± 7 30 ± 8
From Cook DJ, et al. Cardiothorac Anesth 1997; 11: 415“9 with permission.

CBF from pump flow between 1.2 and 2.3 l/minute/m2) at a stable MAP has been well shown
at 27°C (see Table 12.1).
Pulsatile flow appears to have no effect on cerebral blood flow during CPB independent of
any effect of pulsatility on MAP.

Intraoperative ischemia and physiological management
For strokes initiated in the operating room, watershed infarcts constitute the minority of cer-
ebral ischemic events. When available, neuroimaging usually demonstrates embolic events
and associated regional hypoperfusion. This is probably why, in spite of intensive clinical
and laboratory study, it has been difficult to show that physiological variables, such as tem-
perature or perfusion pressure, during CPB are independent determinants of neurological
outcome. For intraoperative strokes it is more likely that these variables modulate the severity of
injury that occurs subsequent to a cerebral embolic event. Because of the frequency of cerebral
ischemic events, physiological management remains a relevant part of practice even if it does
not prevent most strokes.

Effect of perfusion pressure
Through much of the 1980s the surgical, and some of the anesthesia literature, indicated
that the cerebral autoregulatory threshold was shifted leftward during bypass such that lower
mean arterial pressures (35“40 mmHg) were capable of maintaining normal cerebral blood
flow. This was incorrect. Although hemodilution associated with CPB increases cerebral
blood flow for any given mean arterial pressure, the autoregulatory curve still “breaks” at a
pressure of approximately 55 mmHg; below this level cerebral blood flow is compromised.
A combination of well-conducted laboratory investigations, the clinical investigation by
Gold and colleagues showing better composite cardiac and neurological outcomes at higher
mean arterial pressures, and a better understanding of cerebral autoregulation in the elderly,
patients with diabetes and hypertensives have led clinical practice to maintain MAPs above
55“60 mmHg during CPB. Rather than preventing watershed infarcts, this practice helps
to maintain cerebral perfusion in the presence of carotid, cerebral and penetrating vessel
disease and supports collateral flow and perfusion of the peri-ischemic region when embolic
events do occur.

Effect of temperature on neurological outcome
Given the profound effects of temperature on cerebral oxygen demand and the widely held
belief in the neuroprotective effect of hypothermia, it was reasonable to expect that absolute
CPB temperature would be identified as a primary determinant of cognitive outcome. How-
ever, this has not been the case in randomized or non-randomized trials. Although there are
Chapter 12: Cerebral morbidity in cardiac surgery

“multiple” publications investigating the effect of perioperative temperature management on
neurological outcomes, the weight of the evidence is far weaker than would be expected. This
is not to say that perioperative temperature management is unimportant, only that the best
evidence of an effect of a hypothermic management on neurological outcome is quite weak.
Whilst the debate over the relative benefits of hypothermic versus normothermic bypass
remain unresolved, more detailed examination of the influence of re-warming rate and post-
operative temperature in determining cognitive outcomes have produced interesting results.
Data from the 2001 Grigore study randomizing intraoperative temperature management
was reported again when a subset of that negative outcome trial was used to examine the effect
of re-warming speed on cognitive outcomes. The cognitive data was analyzed as a continuous
variable (better or worse) as well as a dichotomous variable (defect present or not). Univariate
analysis did not show a re-warming effect on cognitive outcome, nor did treatment of cogni-
tive outcome as a dichotomous variable in multivariate analysis. However, the authors con-
cluded that slow re-warming had a positive effect because analysis of cognitive outcome data
as a continuous variable was associated with greater improvement in cognitive performance
than conventional re-warming.
The simplest improvement in clinical practice relating to temperature management during
CPB followed the first documentation of cerebral hyperthermia in 1996. Cook et al. showed
that brain temperature was systematically underestimated during CPB and that cerebral tem-
perature can approach 40°C during re-warming, a period associated with a great number of
embolic events. From this observation, closer monitoring of nasopharyngeal and perfusate
temperature and prevention of hyperthermia during CPB have become a standard part of
intraoperative care.

Effect of glucose control
Maintaining blood glucose in the normal range is more a matter of not doing harm than actu-
ally doing something to prevent or reverse ischemic injury. The experimental stroke literature
clearly demonstrates that hyperglycemia, like hyperthermia, worsens neurological outcome
in the event of an ischemic insult. So while maintaining perioperative normoglycemia will not
independently determine the incidence of perioperative stroke, it is very likely to moderate its
severity when ischemia does occur.

Effect of other measures
Few other intraoperative interventions hypothesized to improve neurological outcome have
found their way into clinical practice. A range of different classes of drugs have been tried
including aprotinin, complement inhibitors, steroids, barbiturates, propofol, xenon, calcium
channel antagonists and magnesium, to name but a few. None has proved efficacious in a suf-
ficiently powered clinical trial.

Perioperative stroke
Stroke is one of the most devastating complications following adult cardiac surgery. Litera-
ture from the 1960s and 1970s indicates that the incidence of stroke was around 2“4%. When
one looks at very large populations, reports from the last 5 years demonstrate that this is
largely unchanged. The overwhelming risk factor for stroke is physiological age, particularly
manifest as atherosclerotic disease. In the 1990s a great deal of attention was paid to whether
physiological variables were responsible for neurological outcomes, but none of these studies
Chapter 12: Cerebral morbidity in cardiac surgery

produced compelling evidence. With the rapid expansion of echocardiography and transcranial
Doppler studies, attention shifted towards intraoperative embolization as the primary etiol-
ogy of perioperative stroke and cognitive dysfunction.
While difficult to measure, due to lack of a concurrent control group, improvements in
surgical technique have probably restricted the rise in incidence of stroke, which would have
been anticipated in the ageing population of patients, with more complex medical conditions,
presenting for cardiac surgery. Thus, even if overall stroke incidence in cardiac surgery is
relatively unchanged, at least it has not risen to the levels that outcome models would predict.
Recognition of the importance of embolic stroke has led to increased care in handling of the
ascending aorta. Transcranial Doppler, echocardiographic and neuroimaging data all point
to the ascending aorta as the primary cause of intraoperative stroke. Intraoperative imaging,
single application of the aortic clamp, femoral cannulation, all-arterial grafting and off-pump
techniques that eliminate ascending aorta instrumentation probably all reduce intraopera-
tive embolization. However, even with excellent surgical management of the ascending aorta
there still remains a substantial incidence of perioperative brain injury. This is best exempli-
fied in off-pump CABG during which the aorta is not manipulated.

Figure 12.3 Chronological distribution of the onset of postoperative stroke for on-pump and off-pump CABG.
(From Garrett K, Peel MHS, Sotiris C, et al. Chronological distribution of stroke after minimally invasive versus
conventional coronary artery bypass. J Am Coll Cardiol 2004; 43: 752“6, with permission.)

Chapter 12: Cerebral morbidity in cardiac surgery

Off-pump CABG (OPCAB) and stroke
Off-pump CABG evolved from minimally invasive surgery and a desire to eliminate any mor-
bidity associated with CPB. There was an expectation that elimination of CPB would dramati-
cally eliminate perioperative strokes. Interestingly, this has not been fully borne out. Stroke
rates vary greatly in cardiac surgical reports depending on the patient population and the
type of surgery; however, in single institution CABG surgery, aortic “no touch techniques” or
off-pump surgery seem to only moderately reduce stroke risk. In a compelling study of over
16 000 patients, Bucerius described a stroke rate of 3.9% in CABG with conventional bypass
versus 2.5% in the off-pump group. A similar effect of off-pump surgery is identified by Peel
and colleagues who in a study population of almost 3300 off-pump and 7300 on-pump CABG
found a stroke rate of 1.35% in off-pump and a 2.4% in on-pump CABG (see Figure 12.3).
This effect of eliminating aortic instrumentation, about a 1% decrease in stroke incidence,
is similar to what has been described in a large off-pump meta-analysis as well as the stroke
reduction identified with surgical management guided by epiaortic scanning. This moderate
but meaningful effect is important because it indicates that more than half of perioperative
strokes may not be related to intraoperative embolization from the aorta. The limited neu-
roimaging data available support this: brain imaging shows a 30% incidence of subclinical
cerebral ischemic events in OPCAB.
There are also data to suggest that strokes that occur with off-pump and on-pump CABG
have different timing.

Timing of cardiac surgery-related stroke
Given the low incidence of perioperative stroke, the vast majority of clinical studies have been
retrospective to attain study populations of sufficient size. They typically identify a discharge
code for stroke and have rarely identified the timing of an adverse cerebral event. However, one
of the most important publications on perioperative stroke in cardiac surgery demonstrated
that more than 50% of perioperative strokes occurred postoperatively (see Figure 12.4). This
has been confirmed in at least three subsequent investigations from other institutions. In one
retrospective study of 10 573 patients, nearly 73% of strokes occurred after the patient had
woken from surgery without neurological deficit. The implications of this observation have
not been fully appreciated and are important because the etiology and prevention of early and
delayed stroke are likely to be different. As such, current bias towards intraoperative interven-
tions would have an impact on less than 50% of the strokes observed in practice.
Using multivariate analysis in a study population of 1172, Zingone and colleagues found
that early strokes were significantly more frequent in patients with ascending aortic athero-
sclerosis while delayed strokes were most strongly predicted by patient age. Reinforcing this
was their observation that aortic scanning and changing surgical technique had far greater
impact on early stroke than on delayed stroke. A conclusion of this study was that for the
majority of late strokes a plausible mechanism, different from aortogenic embolism, could
be identified. The most prominent of mechanisms were postoperative CPR and atrial fibrilla-
tion (AF). Comparison of stroke data in on-pump and off-pump CABG is also supportive of
precipitating events. Peel and colleagues looked at stroke timing in on-pump and off-pump
CABG patients and found that on-pump CABG was associated with earlier events than off-
pump CABG.
The origin of delayed strokes may also be embolic, but relatively little research has
gone into investigating their etiology or prevention. Apart from rarer interventions such as
Chapter 12: Cerebral morbidity in cardiac surgery

Figure 12.4 Number of strokes detected immediately after surgery (early strokes) and after initial uneventful
neurological recovery (delayed strokes) by day neurological event was detected. Note: postoperative day 0 refers
to day of surgery, which begins after arrival in intensive care unit. (Hogue CW Jr, Murphy SF, Schechtman KB,
Dávila-Román VG. Risk factors for early or delayed stroke after cardiac surgery. Circulation 1999; 100: 642“647, © 1999
American Heart Association, Inc.)

ventricular assist devices or cardiac arrest with CPR, postoperative stroke might result from
aortic plaques, which become unstable at the time of surgery, or embolism of cardiac thrombi
related to sludging in the left atrium due to AF or generalized left atrial enlargement.
In the geriatric general cardiology population, AF has a high incidence of associated stroke
with clot formation in the left atrium, occurring early after the onset of AF. In cardiac surgical
patients, the overall incidence of AF is about 25% and in some populations the incidence of
new onset AF can be as high as 60%. Data from a variety of sources indicate that postopera-
tive AF may be responsible for at least 30% of late strokes. In studies separating early and late
strokes, multivariate analysis consistently shows that postoperative AF is an independent pre-
dictor of late stroke and has been reported to be associated with a six-fold increase in stroke
risk. The mean time of postoperative occurrence has typically been identified as postoperative
day 3 or 4. In addition to thromboembolic risk from an atrium in fibrillation, postoperative
stroke may also result from thrombus formation associated with regional wall motion abnor-
malities or from thrombus originating on left heart suture lines. This is more likely in low
cardiac output states. Prophylactic therapy with antiplatelet drugs, such as aspirin, postopera-
tively may thus improve neurological as well as cardiac outcomes in certain groups of cardiac
surgical patients (see Figure 12.5).
Chapter 12: Cerebral morbidity in cardiac surgery

Figure 12.5 Fatal and non-fatal ischemic outcomes amongst patients who received aspirin within the first
48 hours and patients who did not. The number of patients at risk varied with the type of outcome, since outcomes
occurring within 48 hours after surgery were excluded from the analysis. A total of 73 patients had multiple causes
of death. (Modified with permission from Mangano DT, et al. N Engl J Med 2002; 347: 1309“17, figure 1.)

Stroke risk in the general population
History of a prior cerebral ischemic event is one of the most powerful predictors of periopera-
tive stroke in cardiac surgery. In the study by Hogue and coworkers, prior stroke increases the
risk of an early perioperative stroke by almost 12-fold and the risk of a delayed stroke by nearly
28-fold, which suggests that delayed stroke may be more related to patient-intrinsic risk fac-
tors than specific intraoperative events (see Table 12.2).
The general population has a background incidence of stroke risk factors, in particular:
• Hypertension;
• atrial fibrillation;
• diabetes; and
• prior stroke.
The practice of cardiology concentrates the highest risk general population patients under
its care and often refers the worst of those to cardiac surgery. As such, the risk of stroke and
renal disease is progressively distilled from general internal medicine to cardiology and then
into the cardiac surgical population.
The American Heart Association provides population statistics for major cardiovascu-
lar disorders and the 2008 report identifies an overall stroke prevalence of 2.6% in the gen-
eral population. The prevalence of stroke in 60“79 year-olds is about 6.3% and close to 13%
in those aged over 80. Even more specific for understanding neurological injury in cardiac
surgery is the incidence of annual hospital stroke admissions for diabetic and non-diabetic
patients (see Figure 12.6), which shows that for diabetic patients over age 65 their annual
likelihood of hospital admission for stroke is 3“7%. This population data places the incidence
of perioperative stroke in an important light. It indicates that the likelihood of periopera-
tive stroke (generally thought of as 2“4% in moderate-risk patients) is nearly identical to the
Chapter 12: Cerebral morbidity in cardiac surgery

Table 12.2. Strokes classified depending on whether the neurological deficit was identified either
immediately after surgery (early events) or after initial uneventful neurological recovery (delayed events)

Variable Odds ratio (95% CI) P
Early strokes
History of stroke 11.6 <0.001
Female sex 6.9 0.004
Ascending aorta atherosclerosis 2.0 0.004
Cardiopulmonary bypass time 1.1 0.005
Delayed strokes
History of stroke 27.6 <0.0001
Diabetes 2.8 0.008
Female sex 2.4 0.028
Low cardiac output syndrome and atrial fibrillation 1.7 0.033
Ascending aorta atherosclerosis 1.4 0.047
Odds ratio reflects risk of stroke with increase in a single level in the aortic scan results. The odds of an increase of
2 levels (e.g., normal to moderate/severe) is the square of the reported odds ratio.
From Hogue CW, et al. Circulation 1999; 100: 642“7, table 6 with permission.

Figure 12.6 Age- and sex-
specific annual admission rates for
cerebrovascular disease (CVD) in
patients with and without diabe-
tes in the general UK population.
Admission rate (%)

Diabetic patients
F (Modified with permission from
5 Currie CJ, Morgan, CL, Gill L, Stott
NCH, Peters JR. Stroke 1997; 28:
1142“6, © 1997 American Heart
Association, Inc., figure 1.)

35“44 45“54 55“64 65“74 >75
Age (years)

annual risk in the general population. This is not to say that cardiac surgery isn™t responsible
for strokes; however, it is helpful to consider that the perioperative experience may be pre-
cipitating the existing stroke risk. If we think of cardiac surgery as precipitating pre-existing,
population-based risk, very important parallels with neurological and renal dysfunction fol-
lowing cardiac surgery are evident.
In the general population, those who have had transient ischemic attacks (TIAs) have
an approximately 50% chance of stroke within the next 6 months and an approximately
10% risk of another stroke within 2 years. In fact, the predictive value of prior stroke on all
major adverse cardiovascular outcomes is profound, particularly in low-income patients
(see Table 12.3). Table 12.3 shows the high incidence of recurrent stroke and mortality from
Chapter 12: Cerebral morbidity in cardiac surgery

Table 12.3. Cumulative occurrence of secondary events in Medicare sample by cohort and type of secondary

Stroke cohort (n = 1518) type of secondary event (%)
Time since initially
identified event, y AMI Stroke OVD
0.5 0.7 3.7 0
1 1.6 6.2 0
2 3.4 10.8 0.3
3 5.1 12.2 1.8
AMI = acute myocardial infarction; OVD = other vascular death.
Modified with permission from Vickrey BG, et al. Stroke 2002; 33: 901“6, table 4.

cardiovascular causes in those who have had a prior stroke. These are the population-based
risks independent of hospitalization, interventional cardiology or cardiac surgery.
While cardiac surgeons are reluctant to take patients to the operating room with recent
stroke (primarily because of hemorrhagic risk), the clinical implications of a history of cere-
brovascular events are underappreciated. A Cleveland Clinic study analyzed 126 patients with
prior stroke undergoing cardiac surgery. They demonstrated that 17% of patients who had had
a stroke within 3 months before surgery had a new perioperative stroke while those with prior
stroke more than 3 months before surgery had a 12% incidence of new perioperative stroke. In
patients with perioperative strokes, those with a more recent history of stroke appeared to be
more sensitive to perioperative hypotension than those with more remote events, suggesting
persistent cerebral vulnerability of these patients to hemodynamic instability.
Placing cardiac surgical stroke in the context of the stroke risk in the general population
strongly suggests that:
• the most profound determinant of neurological outcome is the patient™s previous
medical history rather than specific intraoperative event; and
• a large proportion of adverse outcomes may result from the precipitation or triggering
of patient risk factors by events in the perioperative period.

Neurocognitive outcomes in cardiac surgery
If we think of perioperative stoke as being intimately related to pre-existing population-based
risk, similar observations can be made for postoperative cognitive change. Postcardiac surgi-
cal cognitive decline has been an area of intense interest for the last 10 years. Depending on
the assessment tools, study timing and design, the incidence of early postoperative cognitive
change is 35“85% with longer term dysfunction seen in up to 10“30% of patients. It has also
been suggested that the perioperative cardiac surgical experience was responsible for wors-
ened cognitive status 5 years following surgery.
Because cognitive change is far more frequent than stroke it became the endpoint of many
outcomes trials in the last several years. However, no intervention has resulted in a clinically
meaningful reduction in its incidence. The most likely intervention to reduce cognitive injury
should be off-pump surgery because exposure to CPB and instrumentation of the ascending
aorta is eliminated. However, randomized studies with sufficient numbers of patients have
failed to show a clinically meaningful effect. The trial by Van Dijk randomized patients to
CABG on or off pump. At 3-month follow up, 21% of off-pump and 29% of on-pump patients
Chapter 12: Cerebral morbidity in cardiac surgery

Table 12.4. Randomized 142 off-pump, 139 on-pump CABG cognitive assessment at 3 and 12 months

Incidence of
cognitive decline (%) 3 months (P = 0.15) 12 months (P = 0.69)
Off-pump 21 30.8
On-pump 29 33.6
From Van Dijk D, et al. JAMA 2002; 287(11): 1405“12 with permission.

Table 12.5. Mean changes in z-scores for coronary artery bypass graft patients and PCI controls for the eight
cognitive domains

Baseline to 3 months: 3 to 12 months:
Domain CABG vs. controls CABG vs. controls
Verbal memory P >0.017 ns
Visual memory ns ns
Language ns ns
Attention ns ns
Visuoconstruction ns ns
Psychomotor ns ns
Motor speed ns ns
Executive ns ns
From Selnes OA, et al. Ann Thorac Surg 2003; 75(5): 1377“84 with permission.

showed cognitive decline, while at 1 year follow up decline had occurred in 31% of off-pump
and 34% of on-pump patients (see Table 12.4).
Most studies have concluded that off-pump CABG has better cognitive outcomes, but the
differences between groups appear to be so small as to be of limited clinical importance.
Several other observations place cognitive change following cardiac surgery in a perspec-
tive outside of perioperative injury. Firstly, there is an incidence of cognitive change after
major non-cardiac surgery. While of a lower incidence, the character of the cognitive change
is the same and in these other types of surgery there is no CPB and little or no risk of cerebral
ischemia. Secondly, long-term cognitive outcomes in percutaneous coronary intervention
(PCI) patients appear to be no different from those in patients undergoing CABG (see Table
12.5). This is important because CABG and PCI patients are very closely matched for patient
(or population) risk factors of age, atherosclerosis, diabetes and hypertension. If cognitive
outcomes at 1 year are the same in these PCI and CABG populations, it suggests that the
longer term cognitive changes described following cardiac surgery are probably an expression
of chronic brain changes related to those comorbidities rather than the cardiac surgery itself.
Cook and colleagues used diffusion MRI and cognitive testing to document perioperative
cerebral ischemia in cardiac surgical patients. They found that approximately 30% of patients
showed ischemic changes indicative of postoperative cerebral embolization; however, there
was no relationship between perioperative ischemic events and either in-hospital or postdis-
charge cognitive dysfunction (see Figure 12.7). The incidence of cognitive dysfunction was
exactly the same whether or not patients had experienced a cerebral ischemic event. Fur-
thermore, MRI data was suggestive that the cognitive decline was more a function of chronic
Chapter 12: Cerebral morbidity in cardiac surgery

Figure 12.7 Incidence of cognitive decline in cardiac
surgery patients with and without acute cerebral
ischemia ( N = ischemia; n = no ischemia.), P <0.05.
(From Cook DJ, et al. Post cardiac surgical cognitive
impairment in the aged using diffusion-weighted
magnetic resonance imaging. Ann Thorac Surg 2007; 83:
1389“95, figure 2, with permission.)


Women Men
Prevalence (%)




70“74 75“79 80“84 85+ 65“69 70“74 75“79 80“84 85+
Age (years)

No Prior Stroke Prior Stroke

Figure 12.8 Prevalence of MRI infarct by sex, age and prior stroke. Association with age was significant at
P < 0.0001 in men and women without prior stroke; sex association was not significant in those without prevalent
stroke (P > 0.1). Neither sex nor age associations were significant in those with prior stroke. (From Price TR, et al.
Stroke 1997; 28: 1158“64, figure 1, with permission.)

ischemic microvascular disease than of a perioperative event. This would be consistent with
growing neurology literature on chronic organic brain disease, cognition and occult cerebral
ischemia in the general elderly population.
Two large studies have used neuroimaging to identify the incidence of occult cer-
ebral ischemia in older cardiology patients: The Cardiovascular Health Survey conducted
more than 3000 cerebral MRIs in community cardiology patients and the Rotterdam Scan
Study carried out more than 1000 MRIs. These studies found that approximately 20“40% of
the general population over age 60 have evidence of occult, small-vessel cerebral infarction
(see Figure 12.8). This appears to be the result of penetrating vessel disease due to chronic
Chapter 12: Cerebral morbidity in cardiac surgery

hypertension, diabetes and atherosclerotic disease. Separate studies have linked this type of
chronic small vessel ischemia to vascular dementia and cognitive change. As such, the cogni-
tive changes seen following cardiac surgery may be the manifestation of underlying chronic
brain disease unmasked in the perioperative period by drugs, metabolic changes, sleep dep-
rivation and environmental changes rather than an expression of a perioperative insult. From
this perspective it would also follow that cognitive declines seen 5 years after surgery are the
evolution of underlying chronic brain disease rather than the evolution of an insult occurring
in the perioperative period.

injury and mortality. Crit Care Med 2006;
Suggested Further Reading 34(12): 2979“83.
• Chertow GM, Levy EM, Hammermeister
• McKhann GM, Grega MA, Borowicz LM Jr,
KE, et al. Independent association between
et al. Stroke and encephalopathy after
acute renal failure and mortality following
cardiac surgery: an update. Stroke 2006;
cardiac surgery. Am J Med 1998; 104(4):
37(2): 562“71.
• Mehta RL, Cantarovich F, Shaw A, et al.
• Cook DJ. Neurologic effects of
Pharmacologic approaches for volume
cardiopulmonary bypass. In Gravlee GP, ed.
excess in acute kidney injury (AKI). Int J
Cardiopulmonary Bypass: Principles and
Artif Organs 2008; 31(2): 127“44.
Practice, 3rd ed. Philadelphia: Wolters
Kluwer Health/Lippincott Williams & • Peel GK, Stamou SC, Dullum MK, et al.
Wilkins; 2008: 376“408. Chronologic distribution of stroke after
minimally invasive versus conventional
• Cook DJ, Oliver WC, Jr, Orszulak TA, et al.
coronary artery bypass. J Am Coll Cardiol
Cardiopulmonary bypass temperature,
2004; 43(5): 752“6.
hematocrit, and cerebral oxygen delivery in
humans. Ann Thorac Surg 1995; 60(6): • Plöchl W, Cook DJ. Quantification and
1671“7. distribution of cerebral emboli during
cardiopulmonary bypass in the swine: the
• Cook DJ, Orszulak TA, Daly RC, et al.
impact of PaCO2. Anesthesiology 1999;
Cerebral hyperthermia during
90(1): 183“90.
cardiopulmonary bypass in adults. J Thorac
Cardiovasc Surg 1996; 111: 268“9. • Pl–chl W, Cook DJ, Orszulak TA, et al.
Critical cerebral perfusion pressure during
• Cook DJ, Proper JA, Orszulak TA, et al.
tepid heart surgery in dogs. Ann Thorac
Effect of pump flow rate on cerebral blood
Surg 1998; 66: 118“24.
flow during hypothermic cardiopulmonary
bypass in adults. J Cardiothor Vasc Anesth • Sadahiro M, Haneda K, Mohri H.
1997; 11: 415“9. Experimental study of cerebral
autoregulation during cardiopulmonary
• Gold JP, Charlson ME, Williams-Russo P,
bypass with or without pulsatile perfusion.
et al. Improvement of outcomes after
J Thorac Cardiovasc Surg 1994; 108:
coronary artery bypass. A randomized trial
comparing intraoperative high versus low
mean arterial pressure. J Thorac Cardiovasc • Shaw PJ, Bates D, Cartlidge NEF, et al.
Surg 1995; 110: 1302“11. An analysis of factors predisposing to
neurological injury in patients undergoing
• Hix JK, Thakar CV, Katz EM, et al. Effect of
coronary bypass operations. Q J Med 1989;
off-pump coronary artery bypass graft
72: 633“46.
surgery on postoperative acute kidney

Acute kidney injury (AKI)
Robert C. Albright

The incidence of perioperative renal injury appears to be increasing in the ever more
complex elderly population presenting for cardiac surgery. More often than not acute
kidney injury (AKI) is associated with well-defined risk factors that precede the surgi-
cal event. AKI complicates cardiovascular surgery in as many as 30% of all procedures,
leading to dramatically worse outcomes, including increased mortality and substantial
financial cost.
The incidence of AKI requiring dialysis among patients who undergo coronary artery
bypass grafting alone is roughly 1%. However, when valve surgery or coronary artery bypass
grafting and valve surgery occur concomitantly, the risks of AKI requiring dialysis are 1.7
and 3.3%, respectively. The risks for AKI rise substantially with the severity of chronic kidney
disease (CKD), which afflicts as many as 30 million people in the USA.

Lack of a universally accepted definition of the syndrome of acute renal failure has hampered
the study, understanding, management and prevention of this disastrous complication. The
complexities of the accompanying fluid, electrolyte, acid“base and azotemic solute accumula-
tion have led to approximately 50 different diagnostic criteria for acute renal failure to be cited
in the literature.
A subgroup of intensivists and critical care nephrologists has formed the AKI Outcomes
and Quality Study Group. This group has agreed upon a new definition of acute renal injury,
which replaces “acute renal failure” and enhances the recent Risk, Injury, Failure, Loss, End
stage (“RIFLE”) criteria. Acute kidney injury has replaced the previous term of acute renal
An abrupt decline in kidney function over less than 48 hours as defined by an increase in
serum creatinine of 0.3 mg/dl (greater than 25 μmol/l) or a 50% increase over baseline accom-
panying a decreased urine output of less than 0.5 ml/kg/hour is generally accepted as a defini-
tion of AKI.
• Stage I “ As an increase in serum creatinine of greater than 0.3 mg/dl or greater
than 150% increase in baseline and urine output of 0.5 ml/kg/hour for
6 hours.
• Stage II “ A serum creatinine increase of greater than 200“300% over baseline with
concomitant decreased urine output to <0.5 ml/kg/hour for the past 12 hours.
• Stage III “ An increase in serum creatinine greater than 300% over baseline, or an
absolute level of greater than 4 mg/dl, accompanying a urine output of <0.3 ml/kg/
hour) for the past 12 hours. Also included within the stage III definition would be
any patient who requires renal replacement therapy.
Cardiopulmanory Bypass, ed. S. Ghosh, F. Fatter and D. J. Cook. Published by Cambridge University Press.
© Cambridge University Press 2009.
Chapter 13: Acute kidney injury (AKI)

However, even this new definition is limited by the utilization of increasing creatinine as
the serum marker of decreased renal function (decreased glomerular filtration rate, GFR).
Increased serum creatinine is well known to lag significantly behind the development of acute
injury, and is confounded by its dependence on tubular secretion and relationship to muscle
mass and catabolism. Biomarkers of AKI, including serum and urine neutrophil gelatinase-
associated lipocalin (NGAL), serum cystatin C, interleukin 17 (IL-17) and kidney injury
marker-1 (KIM-1), are relatively new markers of kidney injury and poor kidney function.
NGAL has been shown to be an excellent predictor of AKI in the pediatric population, spe-
cifically in the cardiovascular surgical population. Serum and urinary increases in NGAL
preceded increase in creatinine by over 2 days. Importantly, among pediatric cardiovascular
surgical patients with AKI, NGAL is very specific as well. Unfortunately, NGAL and the other
currently studied newer biomarkers in general are not specific enough in the adult population
to be of use as yet.

Outcomes associated with AKI
It is difficult to overstate the negative clinical impact AKI portends when occurring in associa-
tion with cardiovascular surgery. Any AKI occurring in the perioperative period carries an
accompanying mortality rate of 15“30%, increasing substantially to at least 50% when dialysis
is required. In fact, an adjusted covariant-independent observation of an eight-fold increase
in death rate has been reported among a large cohort of cardiovascular surgical patients (see
Table 13.1).
Even slight decreases in GFR imply an increased mortality risk. A 30% decrease in GFR
found during the perioperative period is associated with a 6% overall morality over the sub-
sequent year, as compared with 0.4% mortality without an accompanying AKI. A relative
increased mortality risk of four- to five-fold with any increase in serum creatinine has addi-
tionally been reported among patients followed for 1 year.
When dialysis is required for AKI, recovery of renal function sufficient to discontinue
chronic dialysis occurs in less than half of these patients. This obviously leads to a dramatic
decrease in quality of life and longevity (20% mortality rate per year).
The cause of death associated with AKI is most often infection. In fact, approximately 58%
of patients with AKI requiring perioperative dialysis in the cardiovascular surgery arena have
a diagnosis of sepsis as compared to 3.3% of those without AKI. Whether the sepsis was the
cause or result of the AKI is not determined in these studies.
The risks for bleeding, wound complications and nutritional compromise are also
increased among patients with AKI.

Table 13.1. Influence of renal dysfunction and AKI on the incidence (%) of mortality and duration of intensive
care unit and hospital stay (LOS) (in days) after coronary revascularization

Mortality (%) ICU LOS (days) Hospital LOS (days)
Normal renal 0.9 3.1 10.6
Renal dysfunction 19.0 6.5 18.2
AKI 63.0 14.9 28.8
Adapted from Mangano CM, Diamonstone LS, Ramsey JG, et al. Renal dysfunction after myocardial
revascularization: risk factors, adverse outcomes and hospital utilization. Ann Intern Med 1998; 128: 194“203.

Chapter 13: Acute kidney injury (AKI)

Risk factors for AKI
Generally, the risk factors for developing AKI can be separated into those that are patient
related versus those that are procedure related. Patient-related factors are predominant,
once again emphasizing the overwhelming consequences of the ageing population with their
concomitant increased burden of chronic illness. The most important patient-related issue
predicting AKI is pre-existing chronic kidney disease. There is an overall 10“20% risk of AKI
requiring dialysis among cardiac surgical patients with a serum creatinine preoperatively of
2“4 mg/dl, and the risk of requiring dialysis increases to nearly 28% with a preoperative serum
creatinine of greater than 4 mg/dl.
The proportional impact of pre-existing subclinical renal insufficiency is extremely well
illustrated by the decade-old study by Chertow. In a study of 43 000 patients, Chertow used
multivariate analysis to identify independent risk factors for dialysis in cardiac surgical
patients. A fraction of his data is presented in Table 13.2. Of greatest importance is the
profound effect of moderately reduced creatinine clearance (CrCl) on the likelihood of
postoperative dialysis. In this study of over 40 000 patients, approximately 60% had a CrCl
less than 80 ml/minute. As such, 60% of the population has an odds ratio for dialysis that
is equal to or greater than the increase in odds ratio seen with the risk factor of prior heart
surgery. The weight of the numbers is astonishing. Anesthetists and surgeons think of prior
heart surgery as a profound risk factor, but Chertow demonstrates that moderate decreases
in preoperative CrCl is as potent a predictor of postoperative dialysis. Even more impor-
tant, there are five times as many patients with subclinical renal insufficiency as those who
undergo redo operations. Given the “normal” declines in CrCl seen as adults age from 65 to
80 years, the importance of this risk factor cannot be overestimated.

Table 13.2. Laboratory evaluation of acute kidney injury

Prerenal Intrinsic Renal Postrenal
BUN/Cr ratio >20 10“20 10“20
Urine specific gravity >1.020 ˜1.010 >1.010 early, <1.0101 late
Uosmol (mosmol/kg) >350 ˜300 >400 early, ˜300 late
U Na (mEq/l) <20 >30 <20 early, >40 late
FE Na (%) <1* >2“3 <1 early, >3 late
U Cr/P Cr ratio ≥40 ¤20 >40 early, ¤20 late
Urine microscopy Normal, hyaline casts ATN: dark granular casts,
hyaline casts, renal epithelial
GN: RBCs, dysmorphic RBCs
(>20%), RBC casts, WBC/WBC
casts, proteinuria
AIN: urine eosinophilia, WBC,
WBC casts, hyaline casts
(consider CES)
AIN = acute interstitial nephritis; ATN = acute tubular necrosis; BUN = blood urea nitrogen; CES = cholesterol
emboli syndrome; Cr = creatinine; FE Na = fractional excretion of sodium (calculated as: U Na/P Na — P Cr/U Cr —
100); GN = glomerulonephritis; P Cr = plasma creatinine; P Na = plasma sodium; U Cr = urinary creatinine; U Na =
urinary sodium; Uosmol = urinary osmolality.
*Falsely low FE Na seen occasionally with acute GN, radiocontrast nephropathy, rhabdomyolysis.

Chapter 13: Acute kidney injury (AKI)

Other associated risk factors from a patient perspective include pre-existing diabetes mel-
litus, female gender, increasing age, preoperative congestive heart failure, peripheral vascular
disease, preoperative balloon pump requirements, chronic obstructive pulmonary disease,
emergency surgery, anemia and, although somewhat controversial, decreased serum ferritin
Whether on-pump versus off-pump cardiac surgery may afford increased risk for AKI has
been recently evaluated. Overall, a propensity matched trial has found that on-pump CABG
carries a risk of AKI of approximately 2.6% versus 1.2% when surgery is performed off-pump.
Difficulties in weaning from CPB and postoperative intra-aortic balloon pump (IABP) are
intuitive additional risks for AKI.
No single etiological factor is responsible for the development of postoperative ARF, but a
number of related factors probably interact to contribute to cause renal injury.

Etiology of AKI
In this context, the principal etiological factors are reduction in renal blood flow during CPB,
the mediators generated by the systemic inflammatory response syndrome (SIRS) accompa-
nying CPB and the translocation of endotoxins from the gastrointestinal tract.
Under normal circumstances blood flow to the kidney remains constant despite variations
in blood pressure in the range from 80 to 200 mmHg; the kidney thus autoregulates its
blood supply. The kidney receives approximately 20% of the total cardiac output (about
1 l/minute). Oxygen delivery thus exceeds 80 ml/minute/100 g tissue. The distribution of
blood flow within the kidney is not uniform, with the cortex receiving more than 90% of
total blood flow.
Oxygen consumption, however, is less than 10% of total body utilization, and thus there
is a low arterio-venous oxygen content difference (1.5 ml oxygen per 100 ml blood). The low
oxygen extraction by the kidney suggests that supply exceeds demand and that there should
be an adequate oxygen reserve. However, the kidney is highly sensitive to reduction in per-
fusion, with AKI being a frequent complication of hypotension. The sensitivity of the kidney
to damage as a result of hypoperfusion, despite its low overall oxygen consumption, is related
to the physiological gradient of intrarenal oxygenation. Within the kidney the cortex and
medulla have widely disparate blood flows and patterns of oxygen extraction.
Although a high percentage of blood goes to the cortex (about 5 ml/minute/g), the cortex
extracts only about 18% of total oxygen delivered to it. On the other hand, the medullary
region has a far smaller blood flow (0.03 ml/minute/g), but has a far greater extraction (about
79% of the delivered oxygen) as a result of the high oxygen requirement for tubular reabsorp-
tion of sodium and chloride ions.
Medullary oxygenation is normally strictly balanced by a series of control mechanisms,
which match regional oxygen supply and consumption. Failure of these controls renders the
outer medullary region susceptible to acute or repeated episodes of hypoxic injury, which may
lead to acute tubular necrosis (ATN).

Hypoxia and renal damage
The differing requirements of cortex and medulla for blood flow and oxygen result in an
oxygen tension in the cortex of about 50 mmHg higher than that of the inner medulla. This
explains why renal tubules are extremely vulnerable to hypoxic injury and why ATN can be
induced by as little as a 40“50% decrease in renal blood flow.
Chapter 13: Acute kidney injury (AKI)

SIRS and endotoxins
Many of the inflammatory mediators generated by the SIRS associated with CPB are poten-
tially damaging to the kidney. Injury by these mediators results from direct cellular effects,
their ability to directly cause vasoconstriction and so impair blood flow, and by their effects
on endothelial function in general. For example, during CPB the production of nitric oxide,
a smooth muscle relaxant produced by endothelial cells, is reduced and the production of
endothelin-1, a potent vasoconstrictor, is increased.
Whilst endotoxins may be directly nephrotoxic, the associated inflammatory responses to
circulating endotoxins, in particular the generation of the proinflammatory cytokines such as
TNF, mediate further renal damage.

Prevention of acute kidney injury
General measures that have been proven to prevent AKI include adequate preoperative
hydration, avoidance of nephrotoxins (particularly radiocontrast dye) and optimization of
hemodynamic parameters.
Unfortunately, pharmacological interventions as a whole have been disappointing in their
ability to prevent AKI. Diuretics, including loop and mannitol types, have been evaluated
in multiple trials. Most of these have not been controlled, or randomized. Treatment with
diuretics is often initiated in response to decrease in urinary flow or clinical signs of vol-
ume overload; these situations may be harbingers of/or associated with AKI, but may also
be due to decreased effective circulating volume, poor cardiac output or, rarely, urinary tract
obstruction. Clearly, diuretics would not be expected to be successful therapeutic interven-
tions specifically for AKI in these situations. Electrolyte imbalance, metabolic alkalosis and
renal tubular damage are all associated with diuretic use, leading to concerns about increasing
clinical risk with their use.
Studies that have evaluated preoperative use of loop diuretics have shown no benefit in
their ability to prevent, correct or shorten the duration of AKI. A non-randomized trial of
immediate postoperative utilization of a “cocktail” of mannitol and furosemide compared
to furosemide alone among patients deemed high risk for AKI (creatinine >2 mg/dl) dem-
onstrated a 50% decrease in risk of permanent dialysis. However, this study did not control
for severity of illness, perioperative events, IABP use or administration of vasopressor drugs.
Small series have suggested a benefit from utilization of mannitol in the pediatric population
at a dose of 0.5 g/kg body weight. It has been proposed that this is due to mannitol™s non-
diuretic “free radical scavenger” effect.
Dopamine, utilized at its “renal dose” of 1“3 μg/kg/minute, has long been widely accepted
as an agent for prevention and treatment of AKI. However, recent randomized trials have dis-
proven its benefit. Whether it may have a role in truly diuretic-resistant situations of volume
overload due to its proximal renal tubular natriuretic effects remains controversial.
Fenoldapam is a more specific dopaminergic agent that has been proposed as an alterna-


. 6
( 8)