Apr 17 2014

Predicting Recovery from Coma

I have been following the literature on using newer technologies (PET, fMRI, and quantitative EEG) to evaluate the brain activity of patients who appear unresponsive, loosely referred to as coma, or more generally disorders of consciousness. A new study, which I will get to below, adds an interesting element to the research.

For background, disorders of consciousness result from brain injury or disease that affects either the brainstem (needed for arousal) and/or both brain hemispheres. Level of consciousness is a continuum from drowsy to brain dead, but for the purposes of this discussion I am going to focus on three categories of chronic impairment of consciousness.

The first is called a minimally conscious state. In this condition patients are barely able to interact with their environment. They are mostly unresponsive, non-verbal, cannot care for themselves, but may respond to stimulation or show some interaction with their environment.

The second is the persistent vegetative state, in which patients may open and move their eyes, experience sleep-wake cycles, and move non-purposefully but they show absolutely no response to stimuli, purposeful movement, or interaction with their environment.

The third category is the locked in syndrome. Patients in this state are awake and conscious, but brain damage has rendered them unable to move or speak. They may consciously move their eyes or blink, but are essentially paralyzed below the eyes. Their exact abilities depend upon the nature of their brain injury.

The question that Steven Laureys and other researchers have been addressing is this – can we use functional imaging techniques to improve our ability to sort comatose patients into their proper category. The current standard is to use a detailed neurological exam. However, the exam might miss “covert consciousness,” or subtle signs of minimal consciousness, or even consciousness with no outward manifestation. In other words, we can’t really see on exam what is going on inside someone’s brain, we can only infer from the exam, which may be obscured if the patient is paralyzed. PET, fMRI, and EEG, however, look directly at brain function.

Laureys demonstrated that using fMRI scan about 40% of patients who were clinically diagnosed as vegetative were actually minimally conscious.  This was a little surprising, but understandable for the above reasons.

My question at the time – why should we care? Obviously this is a scientifically fascinating question, and we can never tell where new advances will lead, but what are the current clinical implications of making such a distinction? In short, are their any implications for either treatment or prognosis between being vegetative and minimally conscious? The short answer at the time was, we don’t know, but probably not much.

One other interesting aspect of this original research is that patients who were comatose secondary to trauma were more likely to have covert consciousness than those who were comatose due to diffuse anoxia – lack of oxygen to the entire brain. This also makes sense, as trauma can cause patchy damage, include paralysis, without necessarily impairing every part of the brain capable of contributing to consciousness. Anoxia, however, tends to take out the whole brain.

Now we come to the new study, the obvious follow up to the previous studies – can PET or fMRI scans of comatose patients predict how they will do over time?

Laureys and his fellow researchers looked at patients who were, by exam, either vegetative (also called unresponsive wakefulness syndrome) or minimally conscious. They excluded those with ambiguous exams. The remaining they scanned with PET and/or fMRI. Some patients were moving too much to get an fMRI scan (movement artifact would make the scans impossible to interpret). They also looked at 4 locked in patients just to confirm that their techniques would identify the consciousness. They found:

We included 41 patients with unresponsive wakefulness syndrome, four with locked-in syndrome, and 81 in a minimally conscious state (48=traumatic, 78=non-traumatic; 110=chronic, 16=subacute). 18F-FDG PET had high sensitivity for identification of patients in a minimally conscious state (93%, 95% CI 85–98) and high congruence (85%, 77–90) with behavioural CRS–R scores. The active fMRI method was less sensitive at diagnosis of a minimally conscious state (45%, 30–61) and had lower overall congruence with behavioural scores (63%, 51–73) than PET imaging.

So PET scanning was accurate when compared to clinical exam, and was very sensitive to detecting minimal consciousness. Again they found that about a third of patients who were clinically vegetative had signs of minimal consciousness when scanned. Of course, as the researchers admit, there is no gold standard of consciousness and therefore we cannot know if the clinical exams in such cases are false negatives or if the scans are false positives. We also cannot know how much awareness there actually is in the patient, only that their brains respond to stimuli.

The paradigm they used was similar to previous studies – asking patients to imagine themselves either playing tennis or walking around their home, and then seeing if the pattern of brain activity that results matches the task and is reproducible.

To follow up these results, and to provide information about the validity of using such scans to examine brain activity, they followed up the patients for 12 months. The big question was whether or not the scans would predict the outcome. They found:

8F-FDG PET correctly predicted outcome in 75 of 102 patients (74%, 64–81), and fMRI in 36 of 65 patients (56%, 43–67). 13 of 42 (32%) of the behaviourally unresponsive patients (ie, diagnosed as unresponsive with CRS–R) showed brain activity compatible with (minimal) consciousness (ie, activity associated with consciousness, but diminished compared with fully conscious individuals) on at least one neuroimaging test; 69% of these (9 of 13) patients subsequently recovered consciousness.

Again PET scanning was useful, while fMRI was essentially a coin flip. But, the key finding is that there were 13 patients in the study who were clinically vegetative but by at least one imaging technique had signs of consciousness. Of those 13 patients, 9 of them 12 months later had signs of consciousness on exam. (“Recovered consciousness” does not mean they woke up, it just means they were minimally conscious or better.)

Of those 9 who improved, 5 were the result of trauma, which means trauma patients were more likely to recover, again consistent with prior studies.

For the PET scan specifically:

We obtained outcome data for 102 of 112 (91%) patients. Demographic and clinical data were not significantly different between patients with outcome data and those lost to follow-up (data not shown) PET imaging correctly predicted 75 of 102 (74%) known outcomes. 51 of 76 (67%) of the patients diagnosed as being in a minimally conscious states by PET remained conscious on follow-up (24% had died). 24 of 26 (92%) diagnosed as unresponsive wakefulness syndrome by PET were unconscious (35%) or dead (56%) on follow-up. Imaging diagnosis correlated with outcome (p<0·0001).

Using functional imaging to enhance the assessment of patients in a comatose state has two possible prognostic uses. The obvious one is to predict who has a chance of making some recovery. PET but not fMRI was found to be useful in this study.

The other utility is to predict who has no chance of recovery. PET was particularly useful in this type of prediction. Knowing when a patient has essentially no chance of recovery is extremely useful to families who have to make decisions about care. No one wants to feel they are giving up too soon, and it can be agonizing to hold onto a loved one in a coma in the faint hope they will make a meaningful recovery.

Families also have to wrestle with the uncertainty of how much recovery will be made, and if the end result will be preferable to death. Would their loved one want to live in an extremely neurologically impaired state? How impaired, exactly, will they be?

If a PET scan can deliver some certainty, even if that certainty is that the patient will never recover, that is better than the agonizing uncertainty.


This study is a very useful follow up to previous studies looking at the uses of fMRI and PET scans as functional scans of brain activity in patients with severely impaired consciousness.

As a result of this study the authors propose a new category of coma, one that would be between vegetative (unresponsive wakefulness syndrome) and minimally conscious – non-behavioural minimally conscious state. This would define those who are clinically vegetative but have signs of minimal consciousness on functional imaging.

The current study shows there are implications for this distinction, as patients in the non-behavioural minimally conscious state were more likely to improve clinically than those in the vegetative state.

Because recovery was still modest in these patients, the practical utility of predicting an improved chance for recovery is not clear. The authors also point out that the techniques used are likely to prove difficult to apply outside a specialist research facility.

The more immediate utility would be, in my opinion, predicting which patients have no chance of recovery, allowing families to have closure and confidence in their decision to withdraw care.

However, as research continues into treatments for brain damage, such as stem-cell therapy and neural prosthetics, it may become more important in the future to make these subtle distinctions. Functional imaging may one day be used to tell us which patients are candidates for having computer chips implanted in their brains and which are not.

2 responses so far