Talan, Jamie
July 21, 2016
BERLIN — When interpreting brain
dopamine transporter SPECT scans for signs of Parkinson's disease, which is
more reliable — the trained eye of a nuclear physician or a software program
that uses automated analyses to assess regions of interest (ROI)?
Generally, visual interpretations
by experienced physicians were accurate and matched with the automated
assessments of ROI analysis, according to the findings presented here in June
at the International Congress of Parkinson's Disease and Movement Disorders.
But there were were discrepancies
in the interpretation of 12 of 120 (10 percent) dopamine transporter SPECT
scans reviewed by the two physicians, study author Elina Mäkinen, MD, of the
Division of Clinical Neurosciences at Turku University Hospital and University
of Turku in Finland, told Neurology Today.
In nine cases, the physicians
suggested the scans were abnormal while the automated software said they were
not. The three other cases were thought to be normal on visual inspection while
the software said the scans were abnormal.
STUDY FINDINGS
The study investigators looked closely
at the measures of the tracers, finding that the dozen scans had 17.6 percent
lower mean striatal tracer binding compared to normal scans (p=0.003)
and 62.7 percent higher binding compared to abnormal scans (p<0.0001).
These patients were older compared to patients with non-discrepant normal
findings (72.6 vs. 62.4 years, p=0.023), and after a minimum clinical
follow-up of 4.5 years, none of them developed neurodegenerative dopaminergic
parkinsonism; one case was lost to follow-up.
These cases could possibly fall on
the borderline of an abnormal scan, said Dr. Mäkinen. That the patients with
discrepant imaging findings did not develop degenerative parkinsonism
syndromes, she said, calls for a “cautious interpretation in these discrepant
cases.”
“Should the discrepant cases be
labeled as normal or abnormal?” Dr. Mäkinen asked. “This is an important issue
as SPECT appears to heavily impact the clinical diagnosis. Abnormal cases
easily get a diagnosis of Parkinson's disease.”
Based on these findings, she
added: “It seems that some cases should probably be interpreted as normal,
based on the automated analysis, even if the visual analysis interprets it as a
slightly abnormal finding.”
Dr. Mäkinen explained that the
visual analysis could be affected by the age-related dopaminergic changes,
whereas the automated analysis can include age-corrections. “The visual
analysis seems also to be very much training dependent, and it's always
subjective,” she said.
But she added that the visual
analysis is important in some of the discrepant cases where the automated
methods interpret falsely that the scan is abnormal. “It seems that cases with
atypical reduction patterns in striatal tracer binding can be analyzed more
accurately by the visual analysis,” she said. “The visual analysis also allows
room for clinical interpretation. In addition, the automated method can always
include artifacts and errors.
“Generally, a great majority of
cases can be correctly diagnosed by visual analysis alone, whereas the value of
the semi-quantitative analysis is emphasized in borderline cases that appear to
show mild uptake defects. This has been observed in previous studies.”
Visual interpretation by an
expert, which is usually a nuclear medicine physician, is generally accepted as
the preferred method, she added. There are several semi-quantitative automated
methods available, but the use of these methods and to what extent these should
be used (not at all, equal or supplementary to the visual analysis, or even as
a primary method of analysis) varies between different centers,” Dr. Mäkinen
said.
The problem with the earliest
automated programs is that they lacked reference values for what defines normal
and parkinsonism, and the regions of interest were drawn by hand, Dr. Mäkinen
explained. Only recently have large scale SPECT studies of healthy controls
versus parkinsonism patients become available, she said, and these values are
now used in evaluating if a scan is abnormal or not.
EXPERTS: ON VISUAL INSPECTION VERSUS IMAGING SOFTWARE TO INTERPRET DAT SCANS
Neurology Today Conference Reporters in 2016!
EXPERTS COMMENT
Patrick Hickey, DO, assistant
professor of neurology at Duke University Medical Center, who was not involved
with the study, said that other studies have shown that there is variability
between raters, as well. And two scans on the same person could have
differences in the uptake of the tracer. Of course, there is also the question:
What is normal and what is abnormal?
“It is an important issue,” he
said. “This is not just a research tool. If you are going to tell people that
they have a chronic neurodegenerative disorder you want to be correct.”
Robert A. Hauser, MD, professor of
neurology, molecular pharmacology and physiology and director of the University
of South Florida Byrd Parkinson's Disease and Movement Disorders Center of
Excellence and the Parkinson's Disease Clinical Trial Center, agreed.
Dr. Hauser has done a number of
studies on these tracers, noting that the scans are federally approved as an
adjunct to a detailed history and clinical exam and are ordered only when a
clinical exam is unclear.
He said that the scan is typically
normal in people with essential tremor and in patients who are taking dopamine
receptor blockers that can mimic some of the symptoms of PD.
“So what does the automated
software bring to the table?” Dr. Hauser asked. “Is it better? Is it additive?”
Should we be using both visual inspection and software?”
“An automated system may be more
accurate but overall, this study suggests that if you use both it is better,”
he added. “The DAT scan is not 100 percent by either method.”
There is growing interest in using
the scan to evaluate disease progression in clinical trials, but he noted that
“there is too much noise in the scan results to use it to monitor an individual
patient's disease progression over time.
Whether you are attending a
meeting or reading about it remotely, you'll be able to access daily, concise
peer-reviewed reports from major neurology conferences via email blasts that
are optimized for web and mobile viewing.
The Neurology Today
Conference Reporters will highlight only those papers and presentations deemed
noteworthy by the Neurology Today editors. All reports are chosen,
vetted, and reviewed by our team of neurologist reviewers. Each report will
include commentary and analysis from independent experts. Look for expanded
coverage and perspectives in the print edition of Neurology Today.
Look for Neurology Today
Conference Reporters for these upcoming neurology meetings in 2016:
* Alzheimer's Association
International Conference, Toronto, July 23-28
* Congress of the European
Committee for Treatment and Research in Multiple Sclerosis, London, September
14-17
* Child Neurology Society Annual
Meeting, Vancouver, October 7-10
* American Epilepsy Society Annual
Meeting, Houston, December 2-6
Missed a meeting? See our 2016
coverage to date by linking to the Conference Reporter tab at NeuroTodayOnline.com.
LINK UP FOR MORE INFORMATION:
•. International Congress of
Parkinson's Disease and Movement Disorders Abstract 1242: Mäkinen E, Joutsa J,
Johansson J, et al. Visual vs. automated analysis of [123] FP-CIT SPECT scans
in patients with parkinsonism http://www.mdsabstracts.org/abstract/visual-vs-automated-analysis-of-123ifp-cit-spect-scans-in-patients-with-parkinsonism/.
http://journals.lww.com/neurotodayonline/Fulltext/2016/07210/News_from_the_International_Congress_of.6.aspx
No comments:
Post a Comment