Quantification of amyloid PET scans feasible, consistent with visual analysis in AD
Click Here to Manage Email Alerts
Key takeaways:
- Researchers analyzed more than 10,000 PET scans of patients from the IDEAS study.
- Measurement of Centiloids from analysis and visual scans had high concordance.
For clinicians who treat patients with Alzheimer’s disease, quantification of amyloid positron emission tomography was feasible and possessed a high degree of concordance with visual analysis, according to a presentation at CTAD.
“Little is known about the feasibility of quantifying real-world amyloid-PET scans,” Renaud La Joie, PhD, assistant professor of neurology at the Weill Institute for Neurosciences at the University of California, San Francisco, and colleagues wrote. “Major challenges lie in scans being acquired on various scanners without standardization of acquisition or reconstruction protocols, and without MRI to preprocess data.”
La Joie and colleagues sought to examine heterogeneous amyloid-PET scans from IDEAS, a large-scale study of patients with cognitive impairment, to quantify these scans in Centiloids and assess any association with visual reads performed at each center.
Their study included 18,295 Medicare beneficiaries with mild cognitive impairment or dementia who underwent PET with FDA-approved radiotracers at 343 facilities. Sites shared 10,700 raw scans that identified either positive or negative visual interpretations by local radiologists or nuclear medicine physicians.
Scans were processed using a PET-only pipeline, which derives Centiloids based on a cortex-to cerebellum tissue ratio in template space. A total of 9,958 scans were successfully pre-processed with default parameters and another 403 scans were included following manual intervention. Of the 10,361 patients with qualified scans, median age was 75 years (51% female; 88% white; 63% at mild clinical stages; median Mini Mental State Exam score, 26). Participants were scanned with [18F]Florbetapir (n = 6,699), [18F]Florbetaben (n = 3,033) or [18F]Flutemetamol (n = 629).
The researchers reported that Centiloids demonstrated an expected bimodal distribution and differentiated visually positive (median = 74 CL) from negative (median = -2 CL) scans (AUROC = 0.913). A threshold of 24.6 CL reached the highest concordance with visual reads (kappa = 0.715, 86% agreement). Discordant cases (visual positive but CL < 24.6 or vice versa, 14% of all scans) were distributed around the threshold and were more common when scan analyzers reported lower analysis confidence (discordance rates = 10%, 30% and 41% for high, moderate and low confidence, respectively). PET acquisition generally followed tracer-specific FDA recommendations for injected dose and acquisition time, but major variability was observed.
“Greater standardization of PET acquisition than is currently required by FDA labels will be needed to reliably quantify amyloid PET in clinical practice,” La Joie and colleagues wrote.