Added Value of Amyloid PET Quantification with the Centiloid Scale in Clinical Practice

BACKGROUND AND PURPOSE: The objectives for amyloid brain PET/CT and PET/MRI fusion images are to determine the agreement between visual interpretations of amyloid PET and to compare the results of visual interpretation with quantitative analysis as measured by the Centiloid (CL) scale.

MATERIALS AND METHODS: One hundred sixty-seven clinical amyloid brain PET/CT scans were reviewed by 3 readers blinded to the original reports. Readers interpreted the amyloid PET scans as negative (0), positive (1), or indeterminate (2) for amyloid deposition. For quantitative analysis, 2 additional readers analyzed amyloid PET images using MIMNeuro to generate CL scores and regional standardized uptake ratios. Reader agreement was determined for visual and quantitative assessment. Using positive scan cutoffs of CL ≥ 40, we determined visual assessment-versus-quantitative assessment for scans consistently interpreted as positive by all 3 readers and for each reader.

RESULTS: Fifty-three scans (31.74%) were rated amyloid-positive by all readers, while 62 scans (37.13%) were rated as amyloid-negative by all readers. The remaining 52 scans had inconsistent ratings with an agreement rate of 69.46%. Most (99/167; 59.28%) scans had CL scores above the CL-positive cutoff (CL ≥ 40), and 47/167 (28.14%) were CL-negative scans (CL <10). The lowest CL score to achieve visual positivity among all 3 readers was 57, while the lowest CL score to receive at least 1 indeterminate score was 11. The readers had a high degree of interreader reliability when rating scans as either positive (κ = 0.62) or negative (κ = 0.66) but were inconsistent when rating scans as indeterminate (κ = 0.17). Optimal cut-points were CL <3.6 for consistent negative and CL ≥28.8 for consistent visual positivity.

CONCLUSIONS: Given the emergence of antiamyloid monoclonal therapies for early-stage Alzheimer disease, reliable detection of amyloidosis is critical for patient care. This study suggests that visual reads of amyloid PET may be insensitive when amyloidosis is milder but are spread across multiple regions (CL ranging from 40 to 59). Quantification of amyloid PET using the CL scale may help guide treatment of patients with mild amyloidosis who are under consideration for antiamyloid disease-modifying therapeutics.