Introduction
Amyloid-β and the AD continuum
Amyloid PET
Clinical utility of amyloid PET
Global multi-centre studies adopting amyloid PET
Challenges of amyloid PET visual assessment across the clinical spectrum
Aims of this state-of-the-art review
Quantitative measures for clinical assessment of amyloid burden
Standardised uptake value ratio
Centiloid scaling
Tracer | Variance (CL SD) young controls | Variance ratio (tracer SD/PiB SD) | Slope (tracer SUVR to PiB SUVR) | Intercept | R2 | CL equation CL = |
---|---|---|---|---|---|---|
12.0 | 4.6 | 0.54 | 0.5 | 0.89 | 175.4*SUVRfbp–182.3 | |
5.4 | 1.54 | 0.78 | 0.2 | 0.95 | 121.4*SUVRflute–121.2 | |
6.8 | 1.96 | 0.61 | 0.4 | 0.96 | 153.4*SUVRfbb–154.9 | |
3.5 | n/a | n/a | n/a | n/a | 93.7*SUVRpib–94.6 |
Implementation
Z-scores
Implementation
Aβ load
Aβ index
Implementation
AMYQ
Comparison of quantitative measures for assessing brain amyloid
Metric | Units | Basis of measure | Utility/widespread use | Validation | Imaging needs | Strengths | Weaknesses |
---|---|---|---|---|---|---|---|
SUVr | Ratio | Ratio of tracer uptake between a target and reference region | Widely implemented through CE/FDA-approved software | Static PET Structural MRI, although SUVr can be calculated PET only from template ROIs | Easy to calculate across multiple regions Available through CE/FDA-approved software Widely validated against other measures on a variety of clinical populations | Dependent on tracer, reference/target region, and analytical implementation | |
CL | Centiloids (0–100), unbounded | Mean amyloid deposition of young healthy controls (0) to typical AD patients (100) | Increasingly widely used in research and clinical settings, available through CE/FDA-approved software | Against SUVr, and including test-retest [74] Positivity threshold validation [146] | Static PET Structural MRI recommended Needs to be calibrated via [11C]PiB or a surrogate reference tracer | Universal, tracer independent metric available through CE/FDA-approved software Widely validated against other measures on a variety of clinical populations Easily interpreted | Currently only validated for global/whole brain ROIs rather than regional — not as sensitive to focal uptake as regional measures |
Z-score | Standard deviations | Difference from mean of a cognitively healthy population | Widely implemented through CE/FDA-approved software | Static PET | Well known and widely used metric available through CE/FDA-approved software Easy to calculate across multiple regions, easily interpreted | Reliant on accurate SUVr measurements Dependent on reference/target region and analytical implementation Requires a normative reference database | |
Aβ load | % | Global Aβ burden | Not widely used | Against SUVr [98] | Static [18F]florbetapir PET Structural MRI | Larger effect sizes than SUVr — increased power in clinical trials Easily interpreted as a % | Unavailable through CE/FDA-approved software Tracer specific ([18F]florbetapir), although work is ongoing for other tracers Not yet widely validated Assumes spatially harmonised pattern of amyloid accumulation according to the maximum carrying capacity of each region |
Aβ index | −1, 1 | Global Aβ burden/specific binding | Not widely used | Static [18F]florbetapir and [18F]flutemetamol [18F]florbetaben work is ongoing [180] | Does not require an MRI Interchangeable across [18F]florbetapir and [18F]flutemetamol PET Independent of reference and target regions | Unavailable through CE/FDA-approved software, although planned to be incorporated as part of Hermes Medical Solutions’ BRASS software Not yet widely validated or implemented | |
AMYQ | 0–100, unbounded | Global Aβ burden | Not widely used | Against CL and neuropathology [100] | Static PET | Does not require an MRI Interchangeable across tracers Independent of reference and target regions | Unavailable through CE/FDA-approved software Not yet widely validated or implemented |
Regulatory approved tools and research methods for amyloid PET quantification
Why is amyloid PET quantification valuable and clinically beneficial?
Quantification in clinical practice
What research is still required to validate amyloid PET quantification? | What studies are in place to perform this validation? |
---|---|
Technical validation | |
Measure agreement among quantification and visual read across cohorts to assess robustness across populations | |
Evaluate the utility and robustness of longitudinal quantification measures | Systematic review (PROSPERO ID: CRD42021254695) updating previous work by Schmidt et al., from 2015 [134] |
Calculate the impact of data harmonisation on global CL quantification | Ongoing work presented at AAIC 2020: “Harmonization of Amyloid PET Scans Minimizes the Impact of Reconstruction Parameters on Centiloid Values” [103] |
Assess CL stability as a function of pipeline design, reference region selection, cortical target, and image resolution. Provide optimal pipeline for multi-centre studies | Ongoing work presented at AAIC 2021: “Evaluating robustness of the Centiloid scale against variations in amyloid PET image resolution” [194] |
Compare static acquisition derived metrics with full quantitation derived from dual-time window dynamic imaging | |
Routine clinical use (diagnostic settings) | |
Determine clinical utility of amyloid PET quantification using a randomised-controlled trial design | Primary outcome of the DPMS [50] |
Formally test if and when quantification approaches support visual assessment of difficult cases | Secondary outcome of the DPMS [50] |
Assess the value of regional visual read and quantification in routine clinical settings | Tertiary outcome of the DPMS [50] |
Scientific and clinical trial settings | |
Assess value of quantification to improve risk stratification and individualised disease trajectory in the earliest stages of AD | Primary outcome of the PNHS [51] |