Article In Brief
New research suggests amyloid PET scans, specifically C11-Pittsburgh Compound B, or PIB, were more accurate in diagnosing Alzheimer's disease in the earliest stages of clinical disease than 18-Fluorodeoxyglucose PET.
Amyloid PET scans—C11-Pittsburgh Compound B, or PIB—were more accurate in diagnosing Alzheimer's disease (AD) in the earliest stages of clinical disease than 18-Fluorodeoxyglucose PET (FDG-PET), according to a study published online on November 20 in Annals of Neurology.
While FDG-PET shows hypometabolism in the temporoparietal regions in AD patients, PIB and other amyloid tracers measure amyloid burden in the brain.
Scientists at the University of California, San Francisco (UCSF) ordered both scans for the 100 patients who died and donated their brains for research. Investigators at UCSF's Memory and Aging Center were able to compare the results of the two scans to the final pathological diagnosis, and see how they matched up.
“Amyloid PET was significantly more sensitive in detecting AD neuropathology, especially in the early stages of the disease,” said Gil Rabinovici, MD, FAAN, the Edward Fein and Pearl Landrith distinguished professor in memory & aging, and senior author of the study.
“When you looked at people who were more impaired at the time of the initial scans there was less of a difference between the two scans.”
This study allowed the scientists to get at the clinical question: How well does the PET scan live up to what they find at autopsy? “They both performed very well, but the amyloid PET scan was more sensitive in the early stages. Early on, there is a clear advantage to PIB,” Dr. Rabinovici said.
“It is important to know that amyloid scans could offer a more accurate picture to help doctors diagnose AD,” added Dr. Rabinovici. “This is a real clinical population coming in for a diagnosis. It means that we should probably be offering amyloid-PET scans early on to determine whether someone has Alzheimer's, and rule it out if they have another form of dementia, or something else.”
UCSF scientists have access to a cyclotron to make the PIB radiotracer for amyloid testing, which is necessary because of the short half-life of the carbon-11 radioisotope (20 minutes). PIB is a research tool and was used in this study. However, three amyloid radiotracers labeled with fluorine-18 (110 min half-life) are more broadly available and have been approved by the Food and Drug Administration for clinical use.
Study Design, Findings
For the study, the scientists collected PIB and FDG PET data on 101 patients who had been seen early on in their disease process and had donated their brains to the ongoing research projects at the UCSF Alzheimer's Disease Research Center or at University of California, Davis. They all had scanning done at the Lawrence Berkeley National Laboratory.
The PET scans were sent to three blinded raters to visually interpret what they saw on the scans. The PIB PET was rated either positive for amyloid or negative. The FDG scans were rated as showing an Alzheimer's pattern or not. Then, they looked at the neuropathological diagnosis and compared it to the diagnoses made during their first visit to the center. There was an average of four years between the initial diagnosis and the pathological one determined from the patient's autopsy tissue.
Eighty-eight of the autopsies were performed at UCSF, two at the University of Pennsylvania, one at UCLA, and one at the Mayo Clinic in Jacksonville, FL. The neuropathologists were blinded to the PET findings but they had access to each patient's clinical history. The pathologists stained for amyloid-beta, hyperphosphorylated tau, alpha synuclein, and TDP-43, and rated whether the patient's cognitive impairment was due to a primary pathology of AD or a contributing pathology (a mixed picture). The average age of the patients at death—60 men and 41 women was 67-years-old, and their Mini-Mental State Examination was 21.9. (The test is scored from one to 30 and measures a decline in cognitive activity with 21 to 24 falling in the range of mild dementia.)
The majority of the patients had early age-of-onset disease. Three raters reviewed every PIB and FDG scan, with similar findings: the accuracy was 89 percent, 96 percent, and 100 percent for the PIB visual readouts, and 79 percent for FDG. There was full agreement among the raters for 84 of 101 PIB scans and 72 of 101 FDG scans.
At autopsy, 32 patients had a primary AD pathological diagnosis and 56 had a non-AD pathology that was primarily frontotemporal lobar degeneration (FTLD) and 13 had a mixed AD/FTLD pathology. The two scans were taken on the same day.
The PIB scans had higher sensitivity (96 percent) compared to the FDG scans (80 percent) for detecting intermediate-high AD neuropathologic change, which includes amyloid-beta plaques and neurofibrillary tangles. The accuracy rates for PIB were even higher—100 percent—when patients were in the early symptomatic stages at the initial exam. There was no difference in patients who were further along in the disease as indicated by a score greater than 0.5 on the Clinical Dementia Rating scale.
When both scans were consistent with the pathological finding—which occurred in 77 of the 101 study subjects—the sensitivity of the combined tests was 97 percent. When the scans were incongruent with the pathological findings, nine of the 24 patients had mixed pathologies. There was no difference in the specificity between the two types of scans.
Even in expert hands, a clinical diagnosis has limited accuracy of anywhere from 70 to 87 percent, Dr. Rabinovici pointed out, and scanning and other biomarkers have significantly helped increase the odds of coming up with the right diagnosis. In this study, the outcome was AD neuropathology or no AD neuropathology, he explained. So a false positive was scan positive PIB or FDG-AD pattern when no-low AD neuropathy was found. A false negative was PIB-negative or FDG-not suggestive of AD, when an intermediate-high AD neuropathy was found.
Dr. Rabinovici said that there are several explanations for the false positive and false negative reads. The false negatives were more common if the interval between the PET and death was longer, and a mixed pathology—in seven of nine false negative cases, comorbid FTLD was a primary cause and AD was a contributing factor. High amyloid burden with low Braak stage resulted in the majority of false positive PIB cases, whereas corticobasal degeneration was the primary pathology in five of nine cases with a false-positive FDG.
The Centers for Medicare & Medicaid Services (CMS) does not reimburse for amyloid PET technology outside of a research study, Dr. Rabinovici said. The agency made its decision in 2013, saying they believed there wasn't sufficient evidence that it offered an advantage over a clinical exam and neuropsychological testing.
Three years later, Dr. Rabinovici, and his colleagues launched the Imaging Dementia—Evidence for Amyloid Scanning study to collect amyloid PET scanning, diagnosis, and management data on over 18,000 patients seen by 964 dementia specialists in almost 600 medical centers and 350 imaging sites. The scientists were able to show that the results of the amyloid PET made an important difference in offering patients a more rigorous diagnosis and a plan for managing their illness. Findings from the study were published in 2019 in the Journal of the American Medical Association.
CMS reimburses for FDG PET, but has still not altered its 2013 decision on amyloid PET, said Dr. Rabinovici, adding: “Amyloid PET should be the gold standard for early-onset symptoms in younger people.”
“In the evaluation of patients with dementia, neurologists often obtain scans and other tests to try to determine what the likely brain disease is that is causing the cognitive decline. We need more studies of these tests obtained in living people who are followed to autopsy to evaluate accuracy,” said Bradford Dickerson, MD, FAAN, professor of neurology at Harvard Medical School and the Tommy Rickles chair in progressive aphasia research and director of the frontotemporal disorders unit and laboratory of neuroimaging at Massachusetts General Hospital.
“This study of a reasonably large sample of patients with a variety of forms of dementia showed that amyloid PET scans or FDG PET scans clearly are sensitive and specific for the likely pathology. When both are used, the accuracy approached 100 percent in this sample of patients, most of whom had either Alzheimer's disease or frontotemporal lobar degeneration. This is important new data that demonstrates that, in dementia subspecialty practice, a comprehensive evaluation which includes one or more PET scans enables a neurologist to accurately diagnose the disease causing a patient's dementia. The high level of diagnostic accuracy now possible provides a strong foundation for efforts to develop treatments of the future.”
“I think this careful study by this team of investigators is an important step forward in refining our clinical (antemortem, during life) diagnosis of dementia type,” added Sudha Seshadri, MD, FAAN, founding director of the Glenn Biggs Institute for Alzheimer's & Neurodegenerative Diseases at UT Health San Antonio.
“The lack of effective treatments has reduced payer enthusiasm for reimbursing tests to enable these diagnostic distinctions, but it is only through routine application of careful clinical classification—imaging by PET, cerebrospinal fluid/blood biomarkers and genetics, cognitive assessment and MRI—the last being most valuable for vascular burden characterization that we are going to be able to conduct more targeted (effective) trials and find treatments.”
“FDG PET predicts neurodegeneration, synaptic loss, and decreased metabolism regardless of etiology and we suspected that,” she said. “Unlike CSF/blood markers, FDG gives additional localizing value. How does this compare to MRI-based regional atrophy? Some studies suggest FDG PET is more sensitive.”
“The next big step would be comparing the value of amyloid/FDG and tau together,” she continued, “but that is probably best done with the emerging generation of new tau tracers with more specific binding as well as binding to 4R and 3R/4R.”