Implications of ischemic penumbra for the diagnosis of brain death:
Apnea testing may induce rather than diagnose death 

Criticism

Opposing the 30-year old rationale behind the clinical diagnosis of brain death, it is now widely accepted that terminal depolarization that triggers tissue damage is only achieved at flow levels lower than 10 to 15 ml.100g^-1.min^-1 whereas synaptic activity is reversibly suppressed when brain blood flow (BBF) falls between 15 and 35 ml.100g^-1.min^-1 - the range of ischemic penumbra ^{7, 8, 9}. Due to the gradual progression of cerebral edema it is clearly conceivable that slowly progressive intracranial hypertension (ICH) may keep BBF within the range of ischemic penumbra for several hours or even days by proportionally decreasing brain perfusion pressure (BPP) in cases of severe brain damage. Hence, similarly to the periphery of focal ischemia, global reductions of BBF to the penumbra range should also render the whole brain tissue still recoverable although synaptically inactive for several hours or a few days. Such situation is here referred to as Global Ischemic Penumbra (GIP).

Analogously, partial occlusions of basilar artery or posterior fossa expansive lesions may selectively subject the brain stem to sublethal although inactivating circulatory deficits, and the nervous tissue may remain viable for longer than those periods advised for a reliable diagnosis of brain stem death. Such condition is designated here as Regional Ischemic Penumbra (RIP). As neurological function evaluated by clinical examination, apnea testing and electrographic methods (EEG, recording of evoked potentials) are dependent on synaptic activity, comatose patients sustaining such incomplete circulatory deficit, either to the whole brain or solely to brain stem, may be clinically and electrographically misdiagnosed as brain-dead or brain-stem dead.

Some clinical data are clearly consistent with GIP The study by Schrader and co-workers ⁴ provides clear evidence of preserved secretory functions of hypothalamus and hypophysis for many hours in 6 clinically brain dead patients despite absent vascular images on cerebral angiography. As no tissue function can be sustained without blood supply, the primary conclusion to be taken from that study is that the sensitivity of angiography is limited and the test does not prove that BBF to the functionally silent structures (such as cerebral cortex and brain stem) is below 10-15mL.100g^-1.min^-1.

In addition, there are other indications to be considered. If the secretory functions of hypothalamus remain active, whole BBF is necessarily above 10-15mL.100g^-1.min^-1 in patients with diffuse ICH. Because terminal depolarization of all brain cells is dependent on inactivation of the same enzyme - Na+, K+ ATPase, all cellular elements are to reach terminal depolarization at the same threshold. Therefore, if the BBF level is enough to sustain some specialized functions that do not require much energy in patients as those studied by Schrader and co-workers ⁴, it is also sufficiently high to prevent terminal depolarization (necrosis) of all brain cells (not only those of hypothalamus), although it may be not enough to maintain specialized functions dependent on intense synaptic activity. The concept of differential thresholds for functional inactivation is supported by data obtained from studies comparing the glucose consumption required by diverse brain structures during their normal specialized functional activity: under normal conditions the metabolic rate of hypothalamus is about one-third of that observed in cerebral cortex ¹⁰.

In addition, the case-report by Kosteljanetz and co-workers ¹¹ demonstrated that the images of brain vessels may continue visible for several days on consecutive angiograms after the "clinical diagnosis" of brain death. Taken together, the data from these two studies ^{4, 11} indicate that these thresholds may be related according to the following order: synaptic inactivation (± 35mL.100g^-1.min^-1) > sensitivity of angiography > inactivation of secretory functions > terminal depolarization (Na+, K+ - ATPase inactivation, or 10-15mL.100g^-1.min^-1).

Accordingly, since intracranial blood vessels may continue visible on repeated cerebral angiograms for days after fulfilling of current diagnostic criteria for brain death ¹¹, some patients with ICH may sustain BBF levels within the range of ischemic penumbra for quite longer than currently advised for observation of absent clinical responsiveness during brain death diagnosis. It is therefore obvious that patients with diffuse ICH with evidences of preserved hypothalamic function, such as control of body temperature or production/release of humoral factors to the blood stream, should not be considered brain-dead (figure 1). Such analysis provides further support to previous criticism by others, who have pointed that requiring exclusion of hypothermia for diagnosis of brain death ^{1, 2, 3} is not in agreement with basic neurophysiologic knowledge, since the absence of hypothermia generally indicates the continuation of neurologically mediated temperature homeostasis 12. This and other paradoxical guidelines render the current criteria for diagnosis of brain death incoherent and confused 12. Likewise, irreversible damage to the whole brain is more likely but still uncertain when hypothalamic function becomes inactive.

The objection that not all parts of the brain must die to characterize brain death ¹³ does not apply to the present criticism. Raising such argument clearly implies in missing the point here put forward: sustained hypothalamic function in a patient in coma, apnea and absent cephalic reflexes secondary to diffuse ICH indicates that BBF is high enough to preserve the vitality of the remaining brain tissue, which is therefore not necrotic, only synaptically silent, regardless of whether vascular images can be demonstrated on cerebral angiograms. In other words, preserved temperature homeostasis is a simple, but eloquent clinical indication of the GIP situation in these patients.

The concept of GIP offers a theoretical model to support statistical evidences that recovery of patients fulfilling criteria for "diagnosis" of brain death cannot be ruled out through clinical studies, no matter what N patients is involved ¹⁴. Although rarely, some patients may reach maximal accumulation of edema shortly after BBF is reduced to less than 35 ml.100g^-1.min^-1. As shown in figure 1 (case 3), provided that brain edema subsides quickly enough from then on, such patients may regain consciousness. Unfortunately, however, that would not be the case for most of them due to the discontinuation of ADH production by hypothalamic nuclei. During the next few days, deterioration of cardiovascular function secondary to conditions such diabetes insipidus would ordinarily lead to further impairment of BBF to levels lower than 10-15 ml.100g^-1.min^-1, and eventually to cardiac arrest even prior to establishment or irreversible brain damage (case 2).

The GIP hypothesis similarly invalidates the rationale for apnea testing. Being dependent on synaptic function to remain active, the respiratory center is not expected to respond to increased levels of carbon dioxide (PCO₂) under such partial circulatory deficit. On the contrary, as hypercarbia induces a marked rise in intracranial pressure related to cerebrovascular dilatation, with a corresponding decrease in BPP 15, apnea testing is expected to further reduce blood supply to the brain stem. While it is true that blood vessels are less responsive to changes in PCO2 following traumatic brain injury (TBI), some response persists even in deeply comatose patients 16. When intracranial compliance is reduced, even the smallest increase in brain blood volume may critically worsen ICH 16, 17.

Most importantly, the detrimental effects of apnea testing on cardiac function strikingly oppose the most elementary postulate for care of patients with severe TBI and ICH, for their absolute intolerance to hypotension ^{18, 19}. Despite rigorous observance of standard criteria (ventilation with 100% oxygen for 10 minutes followed by passive flow of oxygen at 6 L/min. through the endotracheal or tracheotomy tube) and careful monitoring of vital signs, apnea testing induced severe hypotension in up to 23 out of 70 evaluations of deeply comatose patients (33%), in addition to 4 opportunities (6%) when the patients required prophylactic intervention (administration of epinephrine or dopamine hydrochloride), comprising 39% of cases altogether ²⁰. Even cases of reversible and fatal asystole occasionally occurred during apnea testing²⁰. The most likely mechanisms of these complications are the effects of severe respiratory acidosis inevitably associated with apnea testing ²⁰ on cardiovascular function ²¹.

For unknown reasons, BBF is more severely impaired by arterial hypotension than by aggravation of ICH when BPP is similarly affected by these factors in severe TBI patients ^{22, 23}. In these individuals, the threshold of BPP bellow which BBF is largely reduced may be as high as 80 or 90 mm Hg, and even modest and brief reductions in arterial pressure may result in devastating brain ischemia ²³. Indeed, if BPP to the traumatized brain is reduced below critical levels, collapse of intracranial vessels ensues and BBF cannot be restarted by rising BPP to levels as high as 100 mm Hg ^{22, 23, 24}. This persistent vascular collapse may be due to establishment of tension forces between luminal surfaces of intracranial vessels combined to ICH.

Accordingly, by reviewing current treatment protocols and risk factors studied cooperatively (The Traumatic Coma Data Bank), Chesnut ¹⁹ pointed hypotension as the single most important determinant of poor prognosis in TBI patients. Hypotension (one or more recordings of a systolic blood pressure < 90 mm Hg) decreased the rate of good or moderate recovery from 64% to 40% of cases when registered on admission (early hypotension), to 20% when recorded in intensive care unit (ICU) (late hypotension), and to only 15% when observed in both occasions. Correspondingly, the occurrence of persistent vegetative state or death increased from 17% in normotensive TBI patients to 47% in those patients presenting early hypotension, to 66% in those patients with late hypotension, and to 77% of cases when early and late hypotension were consecutively registered ^{18, 19}. Hypoxia, which occurs in about 13% of patients during apnea testing despite the preventive measures ^{20, 25-27}, largely aggravates the condition of TBI patients when associated with hypotension ^{18, 19}.

The above data clearly indicate that apnea testing may induce potentially irreversible collapse of intracranial vessels by simultaneously worsening ICH and inducing hypotension in ‘clinically brain-dead’ patients with critically reduced (yet within the limits of GIP) BBF levels. Therefore, apnea testing may induce rather than diagnose irreversible brain damage, and the results of all ‘confirmatory’ tests subsequently carried out may reflect nothing but the detrimental effects of such apnea-induced intracranial circulatory arrest.

The harmful effects of apnea testing become evident by observing how it changes the outcome of patients in coma and apnea, with absent cephalic reflexes for at least 6 hours. When apnea is simply defined as "lack of any observed effort of the patient to override the respirator or by a ventilatory effort or movement other than that induced by the respirator" as in The NINCDS Collaborative Study of Brain Death, that investigated 503 cases of coma and apnea ^{28, 29}, imminent cardiac arrest does not occur in 7% of these cases. Contrastingly, only when apnea testing is implemented (according to current guidelines proposed to prevent hypoxia) as in 4 different studies ^{20, 30-32} encompassing altogether 152 of these cases, the association with imminent cardiac arrest is shown to be uniformly perfect, while no patient is excluded from the diagnosis of brain death. Obviously, at least some patients should pass apnea testing if its reliability and inoffensiveness were indisputable. It is therefore clearly conceivable that apnea testing may either hasten (case 2) or set (case 3) the clinical outcome to irreversible brain damage (figure 2).

Nevertheless, the paucity of cases excluded from the diagnosis of brain death is not surprising. Physicians have been even advised not to consider ‘respiratory-like movements’ as signs of brain stem vitality, and even quite complex motor responses of the limbs elicited within a few minutes from terminal removal of the ventilator or apnea testing have been regarded as agonic cord reflexes ³³. Brain death itself, presumed according to clinical and EEG criteria, seems to be the only evidence provided to support such capacity of spinal cord neurons ³³.

According to Walker and co-workers ²⁹, when apnea testing is not implemented, the correlation with imminent asystole may be rendered perfect by requiring cephalic reflexes to be absent for at least 48 hours. However, comatose patients cannot be considered brain-dead even by meeting such requirement. As one can anticipate by considering the concept of GIP, cardiac arrest may ensue as a result of hypothalamic failure regardless of irreversible brain damage. That is because cardiac arrest may originate from detrimental effects of metabolic dearrangements on myocardium that may take place more easily at any time from inactivation of hypothalamic and pituitary secretory functions, even at BBF levels above the threshold of tissue damage. Hypotension secondary to diabetes insipidus may ultimately induce collapse of intracranial vessels and catastrophic brain ischemia. Therefore, hypothalamic failure may cause rather than result from irreversible brain damage (figure 1, case-example number 2).

Such sequence of pathophysiologic events further supports previous arguments against misconceptions underlying current diagnostic criteria for brain death. Evidently, showing that patients who clinically display no active synaptic-dependent brain functions would (either inevitably or probably) evolve to cardiac arrest within a short period of time demonstrates that they are dying, but says nothing about whether they are dead ^{12, 34}. The case-example number 2 (figure 1) fulfills current diagnostic criteria for brain death from 18 hours after TBI, although irreversible brain damage establishes only at 31-hour survival. As a fundamental issue, during that interval, the patient is dying, not dead. Evidently, a dying patient may turn recoverable, provided that a better understanding of the pathophysiologic mechanisms underlying their conditions is reached and efficient therapeutic tools are subsequently developed (same case-example, figure 4). Dead patients will never recover, no matter how much progress is attained in therapeutics.

Ironically, according to the GIP concept here put forward, current therapeutic protocols intended for preserving transplantable organs from the consequences of hypothalamic failure (such as vasopressin and dopamine administrations) may also prevent the establishment of irreversible brain damage. By preventing hypotension and consequent collapse of intracranial circulation, the treatment may provide time for regression of brain edema and consequent recovery to normal BPP. As Shewmon ¹⁴ has emphasized, "…few if any patients are maintained sufficiently long for any potential rare recoverer to become manifest."

Different studies ^{29, 35, 36} indicated that although approximately 12 hours of continuous mechanical ventilation is the minimal time required for the characteristics of a respirator brain to develop, near-maximal occurrence of respirator brains requires 24-48 hours of apnea ²⁹. Since about 1 hour of terminal cell depolarization is required for establishment of irreversible damage to the brain tissue ⁹, for current guidelines to be valid all patients fulfilling such diagnostic criteria had to reach BBF levels below 10-15 ml.100g^-1.min^-1 within no more than 5 hours from the moment when synapse-dependent brain functions become silent, so that a total of 6 hours of observation (no more than 5 hours of GIP plus 1 hour of terminal neuronal depolarization) would be indeed enough to characterize irreversibility, and therefore nearly all autopsied cases should present features of respirator brain after 48 hrs of sustained cardiovascular function and silent synaptic function.

However, the histopathological findings observed in cases of coma and apnea autopsied during The NINCDS Collaborative Study of Brain Death ²⁹ are not consistent with the time frame of pathophysiological events implied by current guidelines for the diagnosis of brain death, regardless of whether no periods of observation ³ or a 6-hour period of clinical monitoring ² is adopted. Walker and co-workers ²⁹ still found only about 40% of respirator brains among patients with at least 48 hours of sustained heart beating with absent supraspinal synaptic activity.

Prolonged GIP secondary to ICH may consistently account for the unexpectedly low percentage of "respirator brains" among these autopsied cases. As figure 3 clearly illustrates, rather than decreasing to values below 10-15 ml.100g^-1.min^-1 within the 6-hour period of observation (case 1) BBF levels may remain within the limits of GIP (cases 2 and 3) after fulfilling of current diagnostic criteria for brain death, except for apnea testing, which may induce irreversible collapse of intracranial vessels, thereby changing the percentual values observed. It is possible that these results (about 40% of respirator brains) were only obtained because apnea was defined without apnea testing during The NINCDS Collaborative Study of Brain Death ^{28, 29}. In other words, apnea-induced collapse of intracranial vessels cannot have influenced the histopatological outcome of the 226 autopsied patients of that study. Thus, after 48 hours of sustained cardiovascular function, autopsied brains may display variable histopathological appearance: "respirator brain" features may be fully developed (case 1), unapparent or less clear (case 2). Likewise, provided that the state of GIP is long enough, in some patients developing cardiac complications related to hypothalamic failure, brain damage may not be established even by the time when asystole ensues (case 3). Therefore, the data of Walker and co-workers ²⁹ support the assumption that irreversible brain damage is not established in most patients by the time when the diagnosis of brain death is currently proclaimed, and rather leads to consideration of GIP

Recovery from misdiagnosed brain death would be induced rather than spontaneously achieved, provided that normal BPP is reestablished before BBF remains lower than 10-15 ml.100g^-1.min^-1 for long enough as to determine irreversible damage (about one hour or less ⁹). To date, moderate hypothermia is the only therapeutic tool capable of promoting regression of brain edema, rather than only transiently reducing ICH. During body cooling to 33ºC, intracranial pressure (ICP) is simultaneously normalized and such moderate levels of hypothermia largely reduces brain edema when sustained for 12 hours ³⁷. As suggested in preliminary reports, cooling to 33ºC for 24 hr from 16 hr after severe TBI may induce good recovery even when fixed and dilated pupils and/or deep coma (GCS = 3) are observed ³⁸. Moderate hypothermia may increase the rate of good recovery by 10 times at 6 month survival ³⁸.

Besides inducing recirculation through rapid normalization of ICP followed by regression of cerebral edema ³⁷, moderate hypothermia also blocks the progression of cellular damage to irreversible necrosis ³⁹, even when initiated several hours after a severe ischemic insult ^40-42. Such effect is probably permanent provided that no episodes of hyperthermia follows the hypothermic treatment ⁴³. In addition to neutralizing the benefits of hypothermia to the ischemic brain, hyperthermia sustained for only 2 hours induces as much as 40% of additional accumulation of edema ⁴⁴, dramatically increases ICP ^{45, 37}, reduces functional recovery ⁴⁶, aggravates neuronal death ⁴³, and may trigger chronic neurodegeneration ⁴³. The concept of penumbra has been broadened to include also the nervous tissue affected by transient impairment of ion membrane gradients and energy metabolites ⁴⁸ and therefore all the effects induced by intra- or postischemic hyperthermia are of exceptional relevance in cases of ICH leading to ischemic penumbra (Figure 4). Accordingly, Jones and co-workers ⁴⁹ found pyrexia (in addition to hypotension and hypoxia) as a significant predictor of poor outcome in severe TBI patients controlled for age, admission pupillary response, and injury severity score.

Similarly to the value of hypothermia in severe TBI, thrombolysis may be life-saving in critical cases of basilar artery occlusion (BAO). In accordance with previous indications favouring thrombolytic therapy in severe BAO ^{50, 51}, good recovery following intra-arterial administration of urokinase to a deeply comatose patient requiring mechanical ventilation and sustaining absent cephalic reflexes for several hours was recently reported ⁵². Such outcome clearly contrasts with the uniform progression to death observed in patients with coma requiring ventilatory assistance and who were not treated with thrombolytic agents ⁵³, and strongly suggests that despite angiographic evidences of basilar artery oclusion, an unknown percentage of them may sustain enough blood supply to the brain stem as to prevent irreversible damage (> 10-15 ml.100g^-1.min^-1). RIP may be sustained either through the partially occluded artery or from collateral vessels. Evidently, because hypercapnia-induced rise in ICP and/or occasional hypotension may establish irreversible damage by further decreasing blood supply to the brain stem, apnea testing is not ethically admissible in these patients who should rather undergo intra-arterial thrombolysis.