Interview with Dr. Eidelberg at AAN

From James Beck Ph.D., Director of Research Programs

At the American Academy of Neurology (AAN) annual meeting in Toronto, I had the opportunity to sit down and chat with David Eidelberg, M.D., this year’s PDF-AAN Movement Disorder Research Awardee. Dr. Eidelberg is a Professor of Neurology at New York University School of Medicine and Director of both the Center of Neurosciences at the Feinstein Institute for Medical Research and of the Movement Disorders and Functional Neuroimaging Center at North Shore Long Island Jewish Health Systems. He has been hailed as an innovative thinker who has developed new ideas to advance our understanding of Parkinson’s disease (PD) and movement disorders in general.

In particular, Dr. Eidelberg (pictured at right) is known for his work with neuroimaging. In a recent study, he and his colleagues used one neuroimaging method – FDG-PET scans—combined with their own computerized image classification program to accurately differentiate people living with classic Parkinson’s from those living with related disorders, or parkinsonisms, such as progressive supranuclear palsy (PSP) and multiple system atrophy (MSA).

How does this work and is it a real option for solving cases of difficult diagnosis? Is the cost prohibitive? As we find out below, the technique may not be ready for maintstream use, but it is helpful in teaching us more about Parkinson’s.

Q1: Your team’s recently published study in Lancet Neurology, the culmination of many years of work, used a special form of glucose to measure brain metabolism via a type of imaging called FDG-PET in order to differentiate PD from other related disorders like PSP and MSA. Can you briefly explain how FDG-PET works and discuss your method of classifying classic Parkinson’s versus PSP and MSA?

A: The name “FDG-PET” simply refers to the two parts that comprise the imaging process—a tracer (FDG) and an imager (PET). For the first part, FDG, it is important to understand that the living brain, even at rest, consumes glucose (sugar) as its energy source. In our study, we took advantage of this fact by ‘tagging’ glucose with a harmless, FDA-approved radioactive molecule to see what parts of the brain were more active than others. The resulting compound is called 18F-fluorodeoxyglucose or FDG for short. Part two, PET, simply refers to the specialized imaging apparatus, or positron emission tomography machine, that is used to measure the location of the FDG as it gets utilized in the brain. By studying the images we take, we can then observe how metabolically active are the various regions of person’s brain.

Administering the FDG is very straightforward. The person undergoing imaging is injected with a small amount of FDG through an IV line and then is able to move around freely for about 25 minutes while the compound circulates in his or her body. The PET scan itself is also relatively brief and lasts about 20 minutes. Results are ready within a few minutes of completing the scan.

If the scan shows high FDG uptake in a particular brain region, this reflects tissue with high metabolic activity, whereas low FDG uptake reflects tissue with low metabolic demand, due to loss of synaptic activity (loss of input signals from other brain regions, compromised cell function, or even neuronal death).

We have found that the brains of people living with neurodegenerative diseases, including idiopathic Parkinson’s disease and atypical parkinsonian syndromes such as MSA and PSP, are associated with distinct patterns of altered regional metabolism that are indicators of highly specific pathological change. We can observe the activity of each of these abnormal patterns of brain metabolism in an individual’s FDG PET to measure his or her disease severity. This is important, because in clinical practice, people who actually have MSA and PSP can be misdiagnosed as having PD, particularly early in the clinical course.

While the interpretation of the results of a FDG-PET scan can currently be made by someone who is highly trained to read these studies, the problem is that there are more of these machines available than are physicians trained to interpret the results. Therefore, in our recent study, we developed a fully automated computer algorithm to differentiate accurately between PD, MSA and PSP. In addition, this approach allows for computing the probabilities of each of the three disorders in a given parkinsonian individual based on the expression of the disease patterns in his or her brain.

By comparing the disease probabilities in each case with the imaging criteria for each disease possibility, the person could then be given a tentative pattern-based diagnosis. This automated approach remains experimental and we intend to be able to test and validate it in a larger, multi-center study before it would be widely available for use.

Q2: How do the sensitivity and specificity and predictive rates of your method compare to a clinician’s diagnosis of PD, or even diagnostic approaches for other diseases?

A: In the population of people with Parkinson’s that we studied, the image-based diagnoses are both sensitive and specific – meaning very accurate – when compared to the final clinical diagnoses obtained by movement disorders specialists who followed the same individuals for several years. Our imaging approach was designed as a confirmatory diagnostic test to assist clinicians in specialty practices.

We therefore focused on something called the positive predictive value (PPV), a measure of the accuracy of classification using the early scan versus the later clinical ascertainment by a movement disorder specialist. Indeed, our automated classification approach achieved a PPV of 98% for diagnosing PD, 97% for MSA, and 91% for PSP. More importantly, the diagnostic accuracy of our approach remained excellent (> 90%) even in people in the early stages of parkinsonism with very short symptom durations (i.e., >94%) of the technique. We note that our approach also had reasonable sensitivity (~85%), which is a little lower than seen with dopamine imaging methods like FDOPA PET and βCIT-SPECT. It is important to appreciate that the latter imaging methods are often used as screening tools to distinguish people with presynaptic nigrostriatal defects from healthy subjects –with very few false negatives. These methods, however, have limited specificity in that they don’t do a good job in separating PD from atypical look-a-like conditions, because both have presynaptic nigrostriatal dopamine abnormalities.

Q3: Given the cost of this imaging to people with Parkinson’s (and parkinsonisms) and insurance companies, do you think it will be practical to use such techniques beyond academic medicine, e.g., would imaging techniques such FDG-SPECT or BOLD-MRI offer viable approaches too?

A: Our study demonstrated that our approach is most useful to assist in differential diagnosis of individuals with uncertain parkinsonism. Previous research has shown that such individuals are not infrequently misdiagnosed (~25%), especially at early stages. Indeed, such diagnostic errors can be costly because of the expense and morbidity (complications) of attendant management decisions, as can happen when people with atypical parkinsonian syndromes are referred for invasive procedures such as DBS. By the same token, incorrect early diagnosis can lead to problems in the conduct and interpretation of clinical trials of potential disease modifying agents. So when applied correctly, this technique could help lower the costs of incorrect treatments for some people.

Regarding the costs of FDG-PET, it is important to bear in mind that this FDA-approved imaging modality is now widely available (most Americans live within 70 miles of a clinical PET instrument) and is routinely reimbursed for other neurological indications including epilepsy, brain tumor and dementia. Moreover, the costs (and invasiveness) of this procedure have declined steadily over the past five years and is now in the range of $1500-2000.

Lastly, it is important to realize that pattern-based diagnostic algorithms such as ours do not specifically require FDG-PET. Indeed, we and others have shown recently how this method can be easily applied to other imaging techniques, i.e., perfusion imaging of the brain with SPECT and with functional MRI. While FDG-PET is currently the clinical “gold standard” for resting state metabolic imaging, it is likely that these (and similar) techniques will ultimately be used for pattern-based diagnosis on a broader clinical scale.

Q4) There seems to be a real push, as evidenced in the abstracts presented here at AAN in Toronto, to identify potential anatomic differences (volumetric magnetic resonance imaging (MRI) or diffusion trace imaging (DTI)) in the brains of people living with Parkinson’s. Where do you feel that technology stands and what promise do you think it holds for the future?

A: While these new anatomical approaches can be used to identify significant brain abnormalities, it is not clear how “diagnostic” they will prove to be on the individual subject level. Progressive changes in local tissue volume (voxel-based morphometry, VBM) or pathway microstructure (DTI) occur in PD as a reflection of ongoing neurodegeneration. But how large are these regional changes at early stages of disease, and how quickly do they evolve over time? Also, it is important to understand how these structural abnormalities relate to changes in brain function – both regionally and at the circuit level. Indeed, such changes can be extensive and are likely to be present before symptoms actually appear (see our recent paper mentioned above). In the end, complementary approaches employing specialized MRI scans to elucidate brain structure and anatomy, in conjunction with neurochemistry/metabolism imaging to reveal brain function, may provide the most effective strategy to identify robust imaging biomarkers for Parkinson’s.

PDF thanks Dr. Eidelberg for his time. Do you have questions or thoughts on his research? Please post your comments and we’ll try to answer them to the best of our ability.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>