Sometimes Coma Is Not a Coma

Abstract & Commentary

By Andrew Goldfine, MD

Assistant Professor of Neurology, Weill Cornell Medical College

Dr. Goldfine reports no financial relationships relevant to this field of study.

Synopsis: Novel protocols with functional MRI may allow clinicians to determine if some unresponsive patients are able to hear, understand, and respond to questions.

Source: Naci L, Owen AM. Making every word count for nonre-sponsive patients. JAMA Neurol 2013 Aug 12; doi:10.1001/jamaneurol.2013.3686 [Epub ahead of print].

Vegetative state (vs) and minimally conscious state (MCS) are behaviorally defined as no interaction and minimal interaction with the environment, respectively. Of increasing concern is that patients with these syndromes may be misdiagnosed and actually have higher levels of consciousness than they demonstrate on exam, due possibly to fluctuating level of arousal or disproportionate damage to motor systems. Proper diagnosis is essential, as patients in MCS have a higher likelihood of recovery to independence than those in VS, and patients with full consciousness theoretically could communicate through brain-computer interfaces.

To bypass damaged motor systems, investigators have recently made use of brain imaging tools including functional magnetic resonance imaging (fMRI) and electroencephalography (EEG), with the typical output being change in brain activity to a command. In JAMA Neurology, Naci and Owen report taking this approach one step further by developing a fMRI system to allow patients to communicate. This same research group previously reported a case of a VS subject who communicated through fMRI (imagine playing tennis for yes and navigating around your house for no), but here they report on an experimental paradigm that is more natural for a patient to perform.

In this study, subjects had fMRI performed while they listened to a voice say the words "yes" or "no" in alternate blocks of 22 seconds. These target words were interspersed with the numbers one through nine as distractors. At the beginning of each block, subjects were asked a question (e.g., "Is your name Steven?") and then told to attend to the word that answered the question.

To analyze the data, the authors compared brain activity during "yes" and "no" blocks. To narrow down where to look for a response, they first exposed the subjects to the same stimuli, but instead asked them to count the target words (yes/no) or just relax. The comparison of "count" to "relax" blocks revealed brain regions involved in attention, which were then used in the communication runs as "regions of interest" for analysis. This process allowed the authors to have subject-specific brain regions of interest, essential as these subjects had widespread brain injury and years of potential plasticity for recovery.

The authors report on one MCS and one VS subject who demonstrated evidence of communication — increased activation of attention-related brain regions during presentations of the correct answer (yes/no). In the MCS subject, the results were positive two out of four times, while in the VS subject results were positive four out of four times (two at one visit and two 5 months later). They do not report on other subjects they tested who may have had negative results.

Commentary

This study offers a novel approach for detection of consciousness with some advantages and disadvantages to other approaches. Compared to the more standard task of motor imagery (e.g., playing tennis), this approach is more natural (attend to the word yes for a yes response) and does not require an intact motor imagery network (though like all existing approaches, does require language, thereby excluding subjects with aphasia). Their approach also has the advantage of subject-specific regions of interest, rather than requiring activation in brain regions developed from studies of healthy subjects. The primary disadvantage of this technique is that it uses fMRI, which is an inconvenient means of communicating with a potentially locked-in subject (though a positive result could justify further study with a bedside technology such as EEG).

The clinical implications of this and related studies are still not clear. This study and all others use convenience samples (highly chosen subjects), but to determine the prevalence of patients with positive responses as well as the prognostic significance, we need large-scale studies with randomly chosen subjects. Larger studies are also needed to know who is most likely to have a positive response, e.g., based on type of injury, clinical EEG, or imaging findings. Without this guidance, widespread implementation of these techniques may result in many false-positive or inconclusive results that could misguide therapeutic interventions.