Dynamic Chiropractic - November 30, 1998, Volume 16, Issue 25|
The Current Roles of SPECT and PET in the Diagnosis of Brain Injury
Arthur Croft, DC, MS, FACO
More than a decade ago, when a CAD (cervical acceleration/deceleration) victim complained of memory difficulties, insomnia, irritability and other cognitive and emotional problems, I reluctantly made the diagnosis of mild traumatic brain injury (MTBI). Reluctantly, I say, because such a condition was not widely accepted as resulting from an injury without head strike or loss of consciousness. Of course, a thing can be self-evident without any hard evidence of it being real. Nevertheless, walking into a courtroom to support such a diagnosis back then was dicey at best. With no literature to support you, and a court of scientific opinion leering against you, the whole enterprise was held together by mere chutzpa. I was actually surprised at the occasional successes I did have in that arena.
In the early 1980s, we clinicians were offered new hope in the form of an electrodiagnostic tool called brain mapping. In keeping with the rich tradition in health care of providing quite unnecessary multiple synonyms for a thing, it has come to be known also as topographical brain mapping, quantitative EEG, BEAM (brain electrical activity mapping) and perhaps one or two others. I quickly embraced this new tool. It showed promise experimentally, and seemed, clinically, to conform the brain injuries we believed were present. In the absence of any other diagnostic tool we were, essentially, barking up the only tree.
Over the years, ongoing research has failed to reward us with the answers we had hoped for. In recent years, I've studied electrodiagnostics at Harvard (pronounce that with the lower jaw protruded), and the consensus now is that while the components of brain mapping remain valid, the actual mapping of that information has not proven to be particularly helpful. More importantly, it seems clear that some threshold level of actual brain damage or dysfunction must exist before reaching the level of sensitivity of these tests. And, quite inconveniently, the majority of MTBI never quite reach that level.
For a while, our hope returned to imaging studies. Emergent and rapidly developing technologies like MRI held some promise, but soon we realized that it too would reveal lesions no smaller than a millimeter or so. Even the more egregious of lesions seen surgically and experimentally following CAD -- the diffuse axonal injury (DAI) -- are not reliably found on even the best machines. Would we be forced to retreat to the time honored but fallible SWAG (scientific wild ass guess) method in diagnosing or patients' conditions? ‘ department in the form of SPECT (single photon emission computed tomography) and PET (positron emission tomography). The former measures regional cerebral blood flow; the latter maps glucose uptake/metabolism in the brain. Without belaboring the issue, let me just point out that SPECT has been shown to be much more sensitive to brain injuries than either CT or MRI in several studies. This should come as no surprise since the test examines a part of physiology which is much larger (i.e., region circulation or metabolism) than the minuscule lesions which frustrate the resolving power of MRI or CT.
Most recently, Otte et al.1 reported success using SPECT and PET in CAD injury patients. I spoke with one of the authors (Ettlin) of that study in Switzerland at a conference he organized there, and he assured me that the test showed great promise in delineating areas of hypoperfusion of the brain. Yet several questions remain: 1) Is there a window of time beyond which the diagnostic sensitivity of these test fades? 2) What is the sensitivity, specificity and (positive and negative) predictive values of these tests in CAD-related MTBI? 3) For whom should the tests be ordered?
In a recent edition of Neurology, Bicik et al.2 published the results of a study in which they compared a group of 13 subjects with late whiplash (a common synonym for chronic whiplash) to 16 control subjects. They compared the results of MRI, SPECT, PET, radiography and psychometric/neurocognitive tests (e.g., the Beck Depression Inventory, or BDI) between groups. However, the controls did not get SPECT or MRI, and comparisons from a database of normative data were used instead.
Significant abnormalities were noted in neurocognitive scores in the areas of working memory (54%) and divided attention (46%) in the patient group. A significant correlation between BDI score and PET was found in the frontopolar region, but not in other areas where PET was grossly abnormal. The greatest deviations on PET scanning were found in the right putamen, followed by the left putamen and frontopolar and temporal cortex. Correlation between PET, SPECT and MRI was noted primarily on the basis of cortical thinning (i.e., where cortical thinning was present, abnormalities were seen on PET and SPECT). The authors did not find that PET allowed reliable diagnosis of metabolic disturbances for individual patients, and they did not recommend either SPECT or PET as diagnostic tools in "routine examinations" of patients with late whiplash. They concluded that they could not determine whether any structural damage was responsible for the abnormalities, or whether depression itself -- known to be able to reduce brain metabolism in the areas noted to be abnormal on PET scanning -- might have been responsible for the findings.
The choice of wording ("routine examinations") in the conclusion section is perhaps unfortunate given the propensity for third-party payers to deny reimbursement for tests or therapies on the basis of any report that provides a plausible pretext for doing so. ‘ Voyvodic et al.3 last year regarding MRI, but it's doubtful that any clinician would seriously consider radionuclide tests or MRI "routine." On the other hand, as demonstrated by both of these studies, each appears to have merit in selected cases, but neither paper provides much help in attempting to determine which cases those might be.
I found a few potential problems in the Bicik et al.2 study and its conclusions. In their manipulation of data from SPECT and PET, they used statistical parametric mapping, which is a way of smoothing the data and then comparing on the basis of standard deviation. Their methods were quite complex and it was difficult to really determine exactly what they were doing. One wonders whether this is an accepted or validated method of analysis. Perhaps as a backup radiologists might also have provided a blind interpretation of them (necessary, of course, only if this is the standard method).
Another potential problem is that the patients were said to have head or neck pain or both. Apparently, then, some had only neck pain. In light of the previous findings of Radanov et al.4, i.e., that cognitive or psychological abnormalities can be a manifestation of pain alone, is it possible that some of these patients had no brain injury? Then there is question number one from above: is there a time window outside of which the diagnostic sensitivity diminishes? I suspect there may be. These patients averaged 27 months post injury.
A question that begs answering is the relationship between depression and hypoperfusion in the cortical areas that were abnormal in the study. Yes, some studies have shown that depressed patients sometimes have such hypoperfusion, but the authors of those studies did not inquire as to the history of MTBI. How many of the participants in those studies actually had a prior brain injury? And perhaps the finding of hypoperfusion is common to many disorders. If so, does that finding then become unhelpful? By way of example, the erythrocyte sedimentation rate (ESR) is elevated in practically every inflammatory disorder (and many others). A veritable Swiss Army knife of lab tests, it is an extremely non™specific one. However, in light of the overall clinical picture, it can still provide relevant information. The relationship between 1. cognitive disturbances (which result in elevated scores on some depression instruments) seen following CAD injury without brain injury; 2. those resulting from MTBI in CAD injury patients; 3. the depression commonly resulting from chronic pain; and 4. hypoperfusion seen in those who are depressed (and have not had CAD injury or MTBI in the recent past, and do not have chronic pain) will confound the interpretation of this data and require further study. In other words, perhaps hypoperfusion is seen in persons who are merely depressed as well as in persons who have had a brain injury, but with a divergent pathophysiology.
Another potential problem here is the size of the sample. With so‘ study lacks sufficient power for analysis, and the potential for a type II error looms ominously. Here they might conclude that no statistically significant correlation exists when, indeed, there is a correlation.
Appearing as an editorial in the same issue as the Bicik et al. study was a piece by Alexander.5 In a seemingly valedictory role, the author briefly reviews the relative lack of evidence against an actual brain injury occurring after whiplash. He cites Gennarelli's work with primates, for example, in which DAI was produced more easily with coronal accelerations of the brain (i.e., vs. the sagittal acceleration of CAD trauma). Translation of the work to humans is difficult, he notes.
Of course, those authors did produce lesions in animals accelerated in the sagittal plane. Moreover, he did not mention other work by Gennarelli et al., in which human and baboon skulls were filled with silicone material and accelerated. This study provides insight into the earlier work and helps to "translate" the primate work. Alexander also failed to mention other work (e.g., Liu et al.6) which balances the argument in favor of a true injury. Aside from these oversights, his arguments are mostly reasonable.
Alexander notes that there are two unanswered questions with SPECT: 1. Because SPECT abnormalities are not incompatible with complete functional recovery, are there SPECT findings that reliably predict failure to recover or delay in recovery? 2. In patients who fail to recover, does the SPECT provide diagnostic closure; can it serve as a pathognomonic indicator of neurologic disability? Based on what we currently know, there are no definitive answers to these questions.
He reviews the findings of Bicik et al. and he factors in some of the earlier work4 (suggesting that psychological/cognitive complaints improved in parallel with somatic complaints) and concludes that for some reason -- prolonged pain, perhaps -- some patients develop cognitive symptoms in the absence of true brain injury. Instead of psychosomatic illness, he notes, you have somatopsychic illness. He points to the fact that the PET and SPECT abnormalities seen in the Bicik et al. study are also seen in depression and concluded that attempts to "prove" that patients have brain damage through the use of SPECT, PET or evoked potentials are fundamentally misguided.
Alexander seizes the scientific high ground when he characterizes the foray into the uncertain world of SPECT and PET in search of an elusive diagnosis -- or rather confirmation of the diagnosis -- of brain damage as a bit of a misadventure. He rather peremptorily dismisses these technologies in this regard, in a way reminiscent of the early pundits who cautioned that CT was interesting but probably would have no clinical value and would primarily be useful as a research tool. A few fairly amnesic among these pundits actually made the same predictions about MRI only a few years‘ definitive evidence that SPECT or PET can answer his two questions, although that certainly does not imply that they will never do so.
It is just as easy to suffer from excess rationality as it is to see everything with an eye of faith. I think PET and SPECT scanning will be useful in a subset of patients suspected to have MTBI. The cruel world of the courtroom requires proof and, yes, closure is also important for many patients. Knowing that their problems are more than a figment of the imagination is a critical part of the healing process. My questions need answers, and those will probably be forthcoming in future research. In the meantime, clinicians must be guided by the overall map provided by the clinical compass.
Arthur Croft, DC, MS, FACO