Brain death imaging tests are unreliable (INDex study)
In making the high-stakes decision, exam and perfusion tests often don't align
To help make the diagnosis of brain death, specialized brain imaging tests are endorsed in many U.S. hospitals’ policies for patients who cannot complete an apnea test or other key aspects of the exam.
Qualitative CT perfusion, CT angiography, and conventional (invasive) 4-vessel cerebral angiography are generally considered valid ancillary tests to supplement incomplete brain death exams.
However, recognized experts on brain death have acknowledged that these tests have not been adequately tested to determine how reliable they are when used to pronounce death by neurologic criteria (brain death).
A new study suggests that in fact, these tests probably lack the necessary precision to make this high-stakes decision.
The INDex Study
Between 2017 and 2021, 282 consecutive adults with deep coma (GCS 3) were prospectively enrolled and followed at 15 adult intensive care units across Canada.
No more than 2 hours before their brain death exams, enrolled patients underwent whole-brain CT perfusion along with CT angiography (which was derived from the CT perfusion images).
These images were read by neuroradiologists blinded to the clinical information and neurologic exam findings.
After their exams, 204 of the 282 patients were pronounced dead by neurologic criteria (brain dead), using standard criteria (absence of sedatives or other confounders, absence of brainstem reflexes, positive apnea test).
However, on imaging, many of the clinically brain-dead patients had cerebral blood flow, and many patients with brainstem reflexes (i.e., who were not brain dead) had essentially absent perfusion.
For example, qualitative brainstem CT perfusion classified 20 patients as brain-dead who did not fulfill clinical criteria for death by neurologic criteria. Its specificity (accuracy at ruling in) was low at 74%, while its sensitivity (accuracy at ruling out) was 98.5%.
A high-stakes confirmatory test should have the reverse pattern in its performance: a very high specificity at the cost of lower sensitivity.
Whole-brain perfusion had a higher specificity at 93.6%, which is arguably still inadequate for this purpose.
CT angiography performed worse, maxing out at about 90% specificity and sensitivity (influenced by which contrast phase was chosen for interpretation).
Caveats
There are several cautions to consider before interpreting or extrapolating the results more broadly.
Authors report that “some centers experienced challenges using the CT perfusion protocol, leading to the exclusion of several patients from the final analysis due to a lack of interpretable images.”
They acknowledge that using CT perfusion imaging to reconstruct CT angiogram images may have produced different results than using a dedicated CT angiogram.
However, it seems unlikely that any imaging protocol or specialized center would be able to achieve the performance characteristics most people would consider necessary to declare neurologic death with certainty. And if such a protocol could be created, it’s unlikely it could be replicated widely at the required threshold of consistent accuracy.
Discussion
INDex is one of the few studies that have carefully examined the actual performance of imaging tests in the diagnosis of death by neurologic criteria (brain death).
The results in this cohort weren’t confidence-inspiring, either from the perspective of the physician responsible for pronouncing brain death or the family affected by the brain death decision.
It is striking that these tests are and have been endorsed as alternate or supporting methods for confirming brain death in hospitals’ policies for many years without thorough evaluation and validation. This is acknowledged in a widely cited 2023 systematic review.
They have been used as the final deciding criterion to declare the legal death of human beings despite their unknown accuracy. Infrequently, to be sure, but certainly on the order of thousands of times over the years.
The take-home here isn’t “these tests don’t work,” but rather the more complicated and unsurprising truth that they are complex, difficult to interpret, and often don’t align perfectly with the findings at the bedside.
Although not studied in INDex, 4-vessel invasive cerebral angiography also has significant limitations. While considered the gold standard for its very high specificity (i.e., a very low false-positive rate when performed correctly), invasive angiography has a ~7% false-negative rate, when blood flow is seen in larger cerebral vessels without perfusing the microvasculature of a clinically brain-dead patient.
At most centers, these imaging tests are rarely performed as confirmatory tests for brain death, and it’s safe to assume most radiologists lack strong experience in their interpretation.
When it’s impossible to complete the apnea test or other key parts of the brain death exam, imaging is recommended as the next step. But this study suggests that when used to identify brain death, imaging tests provide only an illusion of certainty (when positive), or worsen uncertainty (when negative), with only a “pretty good” chance in either case of actually being “right.”
The problem lies not in the tests but in the binary construct of brain death, which was conceived and promulgated over 50 years ago in response to two of that era’s new challenges: inadequate ICU capacity to support the growing number of catastrophically brain-injured patients, and an insufficient supply of organs for the growing number of eligible transplant recipients.
Brain death was originally proclaimed to be an irreversible condition rapidly leading to death with no hope of recovery of even partial brain function. These assertions turned out not to always be true.
So it’s not surprising that advances in brain imaging would eventually reveal even greater complexity and ontological weakness in the concept of a binary state of being either neurologically “dead” or “alive.”
What this implies for policy and clinical practice will be the subject for later posts, not to mention innumerable uncomfortable conversations among those responsible for maintaining the increasingly tenuous scientific, ethical, and political consensus regarding the concept of brain death.
Like most things in medicine, the easy patients are easy: prolonged down time without resuscitation, massive midbrain ICH, massive brain trauma etc…
It is the patients with less “obvious” causes that always give me pause. Personally, giving the patient time and performing serial exams often helps the care team and the family come to terms with the diagnosis. Nuclear scintigraphy for confirmation can be useful AFTER exam due to its visual nature. This topic continues to challenge me especially in younger patients with anoxia or severe non-traumatic neurologic injuries. Even then, the concept where someone has vital signs yet is “dead” is very challenging to most lay people and some medical professionals.
If I recall correctly, of the three tests in the article only conventional angiography is in the most recent American Association of Neurology brain death guidelines. Our practice is nuclear flow testing. I would also add that for patients with brain stem reflexes, they wouldn’t be sent for brain death testing confirmatory imaging - so my interpretation of that scan wouldn’t be “consistent with brain death” but rather “very poor neurologic prognosis”. This article does confirm that CT angio based studies aren’t accurate enough - something already outlined in the AAN and World Brain Death Project most recent updates (Greer, JAMA 2020).