Introduction

Death is said to occur when there is an irreversible loss of the integrative unity of the organism as a whole.1, 2 Brain death (BD) is accepted in most countries of the world as death of the patient;3 this is because it purportedly shows that the supreme regulator of the body is dead, and therefore all that is left is a disintegrated corpse.1, 2 It is said that BD is fundamentally a clinical diagnosis made at the bedside.4, 5, 6, 7 Using medically standardized tests that examine functions of the brainstem, one can diagnose the irreversible state of BD at the bedside. Ancillary or confirmatory radiological or electrophysiological testing is not required unless there are confounding factors interfering with the clinical bedside tests.4, 6, 7, 8

The American Academy of Neurology writes that to diagnose BD there must be ‘exclusion of complicating medical conditions that may confound clinical assessment.’7 They further write that ‘the clinical examination of the brainstem includes testing of the brainstem reflexes, determination of the patient's ability to breath spontaneously, and the evaluation of motor responses (of the limbs) to pain…All clinical tests are needed to declare BD and are likely equally essential. (One should not prioritize individual brainstem tests). A confirmatory test is needed for patients in whom specific components of clinical testing cannot be reliably evaluated.’9 Similarly, the Canadian Neuro-Critical Care Group wrote that there should be ‘no movements…arising from the brain…no confounding factors for the application of clinical criteria…’; coma includes there being ‘no spontaneous or elicited movements’, and apnea testing is ‘to ensure that an adequate stimulus is presented to the respiratory center (in the medulla).’10 The more recent Canadian Forum wrote that minimum clinical criteria that must be present in ‘brain arrest’ include ‘deep unresponsive coma with bilateral absence of motor responses…absent respiratory effort on the apnea test; absent confounding factors’; indeed, to do an ancillary test one must document ‘deep unresponsive coma.’4 High cervical spinal cord injury would be a confounding factor.11 Other authors have written that the clinical testing to diagnose BD is done to show ‘cerebral unresponsivity’;5 ‘absence of the brainstem functions’12 and ‘loss of brainstem function; and signifies that breath as an essential element of life has vanished from man;’13 or ‘loss of the breath of life.’14

We present a case and review the literature to argue that bedside testing cannot diagnose BD, because the cervical spinal cord is often injured and dysfunctional after cerebellar herniation, and therefore is a confounding factor. A test that can differentiate lower brainstem irreversible loss of function from upper cervical spinal cord loss of function should be required, if the criterion of BD and the diagnostic standards for it are to be taken seriously.

Case report

An 11-year-old boy had a history of months of morning vomiting, headaches, lethargy, weight loss, and weeks of ataxia and nystagmus. He presented with acute onset of coma, and was intubated, hyperventilated, given mannitol, started on an infusion of hypertonic saline and had an urgent external ventricular drain inserted. Computed-tomographic scan before the ventricular drain showed acute hydrocephalus with a posterior fossa mass. Magnetic resonance imaging (MRI) after the ventricular drain showed severe cerebellar tonsillar herniation with displacement of the upper cervical spinal cord (Figure 1a), and injury of the first cervical spinal cord segment (Figure 1b). Repeated clinical testing was compatible with BD; however, because of the MRI cervical spinal cord findings we did a 99mTc-ethyl cysteinate dimer planar radionuclide blood flow test that documented no uptake in the cerebrum or cerebellum (Figure 2). The family consented to organ donation, which was done.

Figure 1
figure 1

Magnetic resonance imaging (MRI) scans of the brain showing cerebellar tonsillar herniation through the foramen magnum into the cervical spinal canal with (a) sagittal T1-weighted image showing displacement of the upper cervical spinal cord, and (b) axial T2-weighted image through the level of the first cervical spinal segment showing hyperintense signal within the substance of the spinal cord consistent with injury, and associated anterior displacement by the cerebellar tonsils. The MRI scan did not image below this level.

Figure 2
figure 2

The 99mTc-ethyl cysteinate dimer brain blood flow study showing lack of uptake of tracer in (a) the cerebral hemispheres on the static anterior view, and (b and c) cerebral and cerebellar hemispheres on the static lateral views.

Materials and methods

This case prompted us to question whether cervical spinal cord injury may be a common confounding factor in diagnosing BD. As MRI is only rarely done in suspected BD, cervical spinal cord involvement would rarely be identified in usual practice. We searched MEDLINE and PubMed from 1965 to 2008 with any combination of the following search terms: intracranial hypertension, quadriplegia, spinal cord injuries, spinal cord compression, tentorial herniation, uncal herniation and brain herniation. All abstracts were reviewed, and potentially relevant publications retrieved. Reference lists of relevant publications were also reviewed and potentially relevant publications retrieved. Any report mentioning quadriplegia, spinal cord injury or spinal cord compression in the setting of intracranial hypertension or brain herniation was considered potentially relevant. We also searched MEDLINE and PubMed from 1965 to 2008 for BD and pathology, and retrieved relevant publications. Any report of spinal examination postmortem was considered potentially relevant, retrieved, and the reference lists were also reviewed.

Results

We identified 12 cases reported of brain herniation during meningitis that resulted in quadriplegia (Table 1).15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 These reports have remarkable similarities: a patient with an altered level of consciousness has a lumbar puncture, and shortly thereafter has a respiratory arrest followed by high cervical spinal cord quadriplegia with variable partial later recovery. Two cases died and autopsy showed infarction of the upper cervical cord without evidence of arachnoiditis.19, 24 Three cases had MRI and this showed swelling or compression of the upper cervical spinal cord.23, 24, 25 Several of the reports specifically mention that the findings were compatible with anterior spinal artery compression resulting in high cervical spinal cord injury.15, 17, 19

Table 1 Case reports of meningitis with acute brain herniation and quadriplegia from cervical spinal cord injury15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25

We identified nine cases reported of brain herniation without meningitis that resulted in quadriplegia (Table 2).26, 27, 28 Six cases had sudden brain herniation and a respiratory arrest, four of these shortly after a lumbar puncture. These cases at autopsy had upper cervical spinal cord necrosis (four) or tense spinal dura (two) most likely due to ‘compression of the spinal cord and possibly compromise of its vascular supply by the large amounts of displaced cerebellar tissue.’26 Three cases had sudden brain herniation from acute hydrocephalus and later variable partial recovery.27, 28 These three cases at MRI had evidence of cord edema or infarction, and the pathophysiology was hypothesized to be due to arterial (anterior spinal artery) insufficiency because of brain herniation (the vessels ‘descend through the foramen magnum after taking off from the vertebral arteries intracranially’28), and direct spinal cord compression with local ischemia and venous obstruction.27, 28

Table 2 Reports of non-meningitis cases of acute brain herniation and quadriplegia from cervical spinal cord injury26, 27, 28

There were some case series that reported outcomes after brain herniation (Table 3).29, 30, 31, 32, 33, 34, 35 Unfortunately, none of these specifically commented on whether the outcomes in survivors were due to spinal cord injury. These series show that there is a high mortality after brain herniation, and that 25–50% of survivors are left with severe sequelae, including ‘spastic limbs.’29, 30, 31, 32, 33, 34, 35 In one series of traumatic transtentorial herniation, cardiac arrest, flaccidity or bilateral fixed pupils after the herniation predicted worse neurological outcome.33

Table 3 Case series of brain herniation describing outcomes suggestive of possible cervical spinal cord injury29, 30, 31, 32, 33, 34, 35

We found only two case series of BD that specifically commented on the pathological findings in the spinal cord (Table 4).36, 37, 38 The Cerebral Survival Study, the only prospective study of BD ever reported, found that upper cervical spinal cord damage was present in 71/127 (56%) of autopsies.36, 37 In total, 54 cases had ‘localized edema, necrosis, infarction or hemorrhage at the cervicomedullary junction,’ and in another 17 cases ‘acute and chronic neuronal changes, and glial alteration were present in the upper cervical spinal cord.’36, 37 The authors wrote that ‘because this is the location of tonsillar herniations and the boundary of cerebral and spinal circulation, it is susceptible to the vascular changes produced by the cut-off of the vertebral blood supply which caused ‘demarcating reaction’ consisting of localized edema and laceration associated with petechial hemorrhages in the substance of the spinal cord.’36, 37 Similarly, Schneider and Matakas38 found in a consecutive series of 15 cases ‘if the organism is kept alive, it reacts in all cases identically…demarcation develops in the anterior pituitary lobe, in the upper cervical segments of the spinal cord.’ There was hemorrhagic necrosis of C1-C3/4 shown by necrosis in 11 cases, and discoloration and softening in 4 cases.38 These authors wrote that the cervical cord is ‘supplied by arteries, arising from the intracranial portion of the vertebral arteries…it corresponds to the marginal area of an ischemic infarction.’38 Other possible etiologies that were hypothesized included ‘hindrance of venous drainage’ and ‘cuff of necrotic cerebellar tissue compressing the spinal cord.’38 Both pathological series note that other areas of the spinal cord can be affected by a different pathological process; specifically, necrotic cerebellar tissue that had sedimented, causing inflammatory reaction in the marginal areas of spinal white matter of various parts of the spinal cord.36, 37, 38 Schroder did not describe detailed spinal cord pathology but in his series wrote, ‘these [brain pathology] alterations decreased…caudal (lower medulla, first cervical segment),’ implying some changes in the upper cervical spinal cord.39

Table 4 Pathology series of brain death that include examination of the spinal cord36, 37, 38

Discussion

Brain death is said to be fundamentally a clinical diagnosis at the bedside.4, 5, 6, 7 Only when confounding factors make the clinical examination of brainstem functions unreliable is an ancillary test required.4, 5, 6, 7, 8, 9, 10, 11 We report a case, and review the literature, to make the suggestion that upper cervical spinal cord injury is a common result of brain herniation, and a confounding factor in the clinical examination for BD. If unresponsive coma could be partly accounted for by the lack of ability to move the limbs, and if apnea could be attributed to absent respiratory muscle function, both due to upper cervical spinal cord injury, then clinical testing for BD is unreliable.11 Many case reports in the setting of sudden brain herniation both with and without meningitis show that permanent or partially reversible cervical spinal cord injury can occur, and the clinical, autopsy and MRI findings suggest this is due to direct compressive injury to the spinal cord, its arterial supply and its venous drainage.15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28 Furthermore, the autopsy series of BD show that high cervical spinal cord injury is a common finding (56–100% of cases), and most likely due to compression of the upper cervical spinal cord and its blood supply during cerebellar herniation.36, 37, 38 This is compatible with early pathological reports of BD in other languages that found necrosis of the upper cervical segments of spinal cord,40, 41 and with early cat and baboon animal models of BD showing demarcation at the C1/C2 cord segments.42

If cervical spinal cord injury is common after cerebellar herniation, one could argue that there should be more reports of this phenomenon. There are several possible reasons to believe that this is not a sound argument. First, death usually occurs so quickly after herniation that quadriplegia is likely not recognized.19, 28 Second, when quadriplegia is suspected after herniation, it is likely attributed to concomitant brainstem insult and followed by death.19, 28 Third, severe sequelae in survivors of herniation are common, and likely attributed to other brain or brainstem injuries, without investigation of the cervical spinal cord.29, 30, 31, 32, 33, 34, 35 Fourth, in the setting of suspected BD, the spinal cord is rarely investigated or suspected to be affected when testing for limb movement to pain or apnea.4, 5, 6, 7, 8, 9, 10, 11 A final reason is that most patients are imaged with a CT scan of the brain, and this is often not adequate to show cerebellar herniation, and not inclusive of the upper cervical spinal cord.43, 44 To adequately show tonsillar herniation an MRI scan is required, and this is usually not done due to the hemodynamically unstable state of the patient.43, 44, 45

One could argue that some of the spinal cord injury cases could be due to spinal arachnoiditis with vasculitis in the setting of meningitis. There are several cases reported of spinal cord injury in meningitis due to this mechanism, proven at laminectomy or myelogram.46, 47, 48, 49, 50, 51, 52 These cases are different from the ones reported here. The cases reported usually occurred later in the meningitis course, were not associated with sudden respiratory arrest, were not shortly after a lumbar puncture, and involved lower parts of the spinal cord from C5 and lower.46, 47, 48, 49, 50, 51, 52 We found three cases of high cervical spinal cord injury in meningitis likely due to arachnoiditis: a 24-week premature neonate with meningitis who at 7 days of treatment became flaccid below the neck despite being awake and having spontaneous movements; a 33-week premature neonate with meningitis who at 4 days of treatment became flaccid and areflexic despite grimacing to pain, and autopsy showed extensive necrosis of the cervical spinal cord with exudates and vasculitis; and a neonate with meningitis who at 4 days of treatment was noted to be areflexic with minimal movement of the upper extremities and no movements of the lower extremities but was extubated 3 days later.50 These cases do not resemble the cases discussed here of sudden herniation with respiratory arrest and quadriplegia, and they do not explain the findings in the non-meningitis cases. Banks and McCartney46 in 1942 noted this when they described a variant of meningococcal meningitis that they called ‘focal encephalomyelitis,’ where there was sudden fatal collapse and pathological changes ‘usually about the basal ganglia, midbrain, medulla, or upper cord…postmortem no satisfactory cause of death may be found in these cases on macroscopic examination, since the meningitis may have almost cleared up…it is only on careful histological examination…’.

It could also be argued that some of the cases of spinal cord injury may be due to the hypoxemia and ischemia of the respiratory arrest event. This mechanism would be expected to affect mostly the thoracic and lumbar spinal cord because the border zone of arterial circulation is in the thoracic cord.38, 50 We note that, even if some cases of high cervical spinal cord injury are due to this mechanism, this still results in a confounding factor for the clinical examination for BD.

There are important implications of this discussion for the diagnosis of BD. First, we suggest that to clinically examine the brainstem functions of motor responses and respiratory drive, one must first know that upper cervical spinal cord injury is not a confounding factor. This may prove to be difficult because the sensitivity and specificity of MRI of the cervical spinal cord for this purpose is not known, and obtaining an MRI on all these potentially unstable and ventilated patients is difficult to manage and arrange. We are not aware of other tests, which could be used to prove that the upper cervical spinal cord is functioning. Whatever test is chosen, this makes the diagnosis of BD fundamentally not a clinical diagnosis until upper cervical spinal cord injury can be ruled out. It is interesting that this point is similar to the one made regarding the fact that brainstem damage is a confounding factor for the examination of cerebral functions in patients being assessed for BD.53

Second, if cervical spinal cord injury cannot be ruled out or is in fact present, another option may be to do an ancillary test to confirm that brainstem function is irreversibly lost.54 The absence of blood flow to the brainstem would likely show that the brainstem has died, regardless of cervical spinal cord function. This is problematic for several reasons. No radionuclide test, including HMPAO-SPECT, is known to diagnose brainstem death in isolation.6, 8, 55 High resolution HMPAO-SPECT may be capable of defining brainstem blood flow,56 but larger numbers of cases would be required to validate this test. In a recent review, it was found that for clinically confirmed BD the specificity of HMPAO-SPECT was 12/12 (100%) (95% CI 78.4–100%), and for contrast angiography confirmed BD the specificity of HMPAO-SPECT could not be estimated, as there were no patients without clinical BD having both tests.57 A bilateral four-vessel internal carotid and vertebral artery angiogram may be required to show lack of blood flow to the brain and brainstem. To do this on every case of suspected BD would be difficult, and in addition, the angiogram is invasive and has the potential to cause vasospasm and worsen the brain injury if BD has not yet occurred.58 Moreover, we are not aware of a study showing that an angiogram can prove brain circulatory arrest and rule out residual brain blood flow enough to sustain some brain viability. Indeed, it has been shown that electroencephalographic activity can remain in cases with clinical BD and absent flow on four-vessel angiography; this suggests that the angiogram did not rule out residual blood flow sustaining viability of parts of the brain.59, 60 Another study found that 13/43 (30%) of clinically brain-dead patients with absent conventional angiographic brain blood flow had persistent cerebral perfusion by computed-tomographic angiography.61

In conclusion, this review suggests a major limitation in our ability to diagnose BD clinically at the bedside. Further expert discussion is needed to resolve this difficult issue.