The Centiloid scale is a standardized scoring system that converts amyloid PET brain scan results into a universal 0-to-100 metric, regardless of which radioactive tracer or scanning method was used. Introduced in 2015, it solved a frustrating problem: different research groups scanning for amyloid plaques in the brain were producing numbers that could not be directly compared. The scale anchors 0 to young, cognitively healthy adults (essentially no amyloid) and 100 to patients with typical mild-to-moderate Alzheimer’s disease, giving clinicians and researchers a shared language for talking about how much amyloid is in someone’s brain.
Why Alzheimer’s Research Needed a Common Yardstick
Amyloid PET imaging works by injecting a radioactive tracer that binds to amyloid-beta plaques in the brain. The scanner then measures how much tracer has accumulated in cortical regions compared to a reference region, producing a ratio. The problem is that several different tracers exist, each binding to amyloid with different affinities and producing different ratio numbers. A value of 1.5 with one tracer might mean something entirely different from 1.5 with another. On top of that, different software pipelines, reference regions, and analysis methods could produce different numbers even from the same scan. Comparing results across clinical trials or pooling data from multiple research centers was, in practical terms, unreliable.1PubMed Central. A practical overview of the use of amyloid-PET Centiloid values in clinical trials and research
The Centiloid Project addressed this by defining a “standard” method for analyzing PiB PET data (PiB being the original and most widely studied amyloid tracer) and then providing a mathematical framework for converting any other tracer or analysis pipeline onto the same 0-to-100 scale.2PubMed Central. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET Think of it like converting temperatures between Fahrenheit and Celsius: the underlying measurement is the same, but the numbers are rescaled so everyone is speaking the same language. Once a tracer has been calibrated to the Centiloid scale, its output can be compared head-to-head with any other calibrated tracer. That comparability has been essential for multi-site drug trials and for understanding how amyloid accumulates across large, diverse populations.
What the Numbers Actually Mean
A Centiloid value of 0 represents the average amyloid burden in young, cognitively unimpaired adults under 45, while 100 represents the average in patients diagnosed with mild-to-moderate Alzheimer’s disease. Values can go above 100 in people with especially heavy plaque loads, and they can dip slightly below 0 due to normal measurement noise. The scale is continuous, so a score of 50 does not mean “halfway to Alzheimer’s” in any clinical sense; it means the amyloid burden falls midway between those two anchor groups.
The real clinical utility comes from specific threshold zones that researchers have validated against autopsy findings and cognitive outcomes. A large individual-participant meta-analysis pooling data from nearly 49,000 people identified a single statistical cutoff for amyloid positivity at 18 Centiloids. But the same analysis found that scans could be confidently classified as negative below 11 Centiloids and positive above 26 Centiloids, with a gray zone in between where interpretation is less certain.3JAMA. Amyloid PET Quantitation and Centiloid Thresholds in the Diagnosis of Alzheimer Disease: An Individual Participant Data Meta-Analysis The AMYPAD consortium, a European research group focused on amyloid PET, reached a similar conclusion: below 10 Centiloids, amyloid pathology can be excluded with high confidence; above 30, pathological amounts are present.4PubMed Central. Centiloid recommendations for clinical context-of-use from the AMYPAD consortium
What Autopsy Studies Tell Us About These Thresholds
The most convincing validation of any brain imaging measure comes from comparing it with what is actually found in the brain at autopsy. Studies that matched Centiloid values from PET scans taken before death with postmortem neuropathology have been reassuring. One study found that a threshold around 20 Centiloids had excellent accuracy for detecting moderate or frequent amyloid plaque density, while values below 10 were optimal for ruling out the presence of neuritic plaques entirely.5PubMed Central. Comparison of amyloid PET measured in Centiloid units with neuropathological findings in Alzheimer’s disease People who ultimately received a confirmed Alzheimer’s diagnosis at autopsy had a median Centiloid value of about 88, with 94% scoring above 45.
An earlier neuropathological correlation study found that Centiloid values between 15 and 20 were highly sensitive for ruling out amyloid plaques, while a threshold of 35 had the highest specificity for confirming moderate or frequent plaques.6Journal of Nuclear Medicine. Correlation of Amyloid PET in Centiloid units with neuropathological findings in Alzheimer’s disease The fact that these autopsy-validated thresholds cluster in the same general neighborhood as the statistically derived cutoffs from living participants gives the Centiloid scale a solid biological grounding. The numbers are not just statistical abstractions; they correspond to real quantities of plaque in the brain.
The Intermediate Zone and Why It Matters
If you look at the thresholds discussed above, you will notice a gap. Below about 10 to 12 Centiloids, you are almost certainly amyloid-negative. Above about 26 to 30, you are almost certainly amyloid-positive. But what about the range in between? This intermediate zone, roughly 10 to 30 Centiloids, is one of the most actively debated areas in the field.
People in this range may have subtle amyloid pathology that has not yet reached the levels typically seen in symptomatic Alzheimer’s disease. Research on cerebrospinal fluid biomarkers identified an inflection point around 12 Centiloids that appears to mark the transition from no pathology to subtle pathology, with a second inflection around 30 Centiloids marking established pathology. The AMYPAD consortium noted that values falling in the intermediate range are associated with an increased risk of disease progression, even though they do not yet meet the threshold for full amyloid positivity.4PubMed Central. Centiloid recommendations for clinical context-of-use from the AMYPAD consortium For a person with mild memory concerns and a Centiloid score of, say, 18, the picture is genuinely ambiguous, and a repeat scan after a year or two might be needed to see whether amyloid is accumulating.
This matters practically because the intermediate zone is exactly where treatment-eligibility decisions get complicated. A group of global dementia experts in 2024 collectively considered a 24-to-30 Centiloid threshold as a practical cutoff for initiating anti-amyloid therapy in patients already diagnosed with mild cognitive impairment or mild Alzheimer’s dementia.7PubMed Central. Expert opinion on Centiloid thresholds suitable for initiating anti-amyloid therapy Someone scoring 15 Centiloids with mild symptoms might not qualify for treatment under current frameworks, even though they could be on a trajectory toward higher levels.
How the Scale Is Used in Drug Trials
The Centiloid scale has become central to how anti-amyloid drugs are tested and evaluated. Clinical trials use it both to select patients with enough amyloid to benefit from treatment and to measure whether the drug is actually clearing plaques. In the TRAILBLAZER-ALZ 2 trial of donanemab, one of the newer anti-amyloid antibodies, nearly all participants in the lowest posttreatment amyloid groups achieved clearance below about 24 Centiloids by 76 weeks. Critically, lower posttreatment Centiloid values were strongly correlated with slower cognitive and functional decline.8JAMA Neurology. Posttreatment Amyloid Levels and Clinical Outcomes Following Donanemab for Early Symptomatic Alzheimer Disease That correlation gives researchers a measurable target: clear the amyloid to below a specific Centiloid value, and you can expect better outcomes.
Earlier dosing studies of donanemab showed that single and multiple doses produced rapid and sustained amyloid plaque reduction, with no significant reaccumulation observed out to 72 weeks.9Journal of Prevention of Alzheimer’s Disease. Randomized Controlled Trial to Assess Safety, Pharmacokinetics, and Pharmacodynamics of Donanemab in Early Alzheimer Disease Without Centiloid values to standardize those plaque measurements across trial sites and scanning equipment, interpreting whether the drug was working would have been far more complicated. The scale essentially turned amyloid clearance into a trackable, reproducible endpoint rather than a fuzzy estimate.
Tracking Amyloid Accumulation Over Time
Beyond treatment trials, the Centiloid scale also helps researchers understand how amyloid builds up naturally over years. One longitudinal study found that people who started with a positive amyloid scan accumulated roughly 3 to 11 additional Centiloids per year, while those who started amyloid-negative showed essentially no change.10PubMed Central. Defining a Centiloid scale threshold predicting long-term progression to dementia in patients attending the memory clinic This asymmetry suggests that once amyloid accumulation crosses a certain threshold, it tends to accelerate, while people below that threshold remain stable for extended periods.
The prognostic value of combining Centiloid values with other clinical measures is also strong. One study found that combining a patient’s age, a blood-based tau biomarker (p-tau181), Centiloid score, a verbal fluency test, and hippocampal volume produced high accuracy for predicting who would transition to a worse cognitive stage.11PubMed Central. Predicting cognitive stage transition using p-tau181, Centiloid, and other measures The Centiloid value on its own provides a piece of the puzzle, but its predictive power improves considerably when paired with cognitive testing and other biomarkers.
Research comparing amyloid levels with cognitive scores in people with mild cognitive impairment reinforces this. People with both high amyloid (above roughly 44 Centiloids) and lower cognitive test performance had the fastest rate of decline and the highest risk of progressing to dementia within three years, while those with low amyloid and preserved cognition almost never progressed.12Journal of Nuclear Medicine. Predicting progression from mild cognitive impairment to dementia in Alzheimer’s disease The Centiloid number, in other words, means different things depending on the clinical context it sits within.
How Blood Tests Are Starting to Map Onto Centiloid Values
Amyloid PET scans are expensive, require specialized facilities, and expose patients to radiation. A major push in Alzheimer’s research has been to develop blood-based biomarkers that can approximate what a PET scan would show. The Centiloid scale has become the benchmark against which these blood tests are validated.
Plasma pTau217, a form of phosphorylated tau protein measurable in a blood draw, shows a strong correlation with amyloid PET Centiloid values. One large study reported a correlation of 0.76 between plasma pTau217 and amyloid Centiloids across the full study population.13eBioMedicine. Detection and staging of Alzheimer’s disease by plasma pTau217 on a high throughput immunoassay platform Another study found that models incorporating pTau217 ratios could predict brain amyloid levels up to about 89 Centiloids, with high diagnostic accuracy across a wide Centiloid range. Combining pTau217 with another blood marker, the amyloid-beta 42/40 ratio, improved sensitivity for detecting low levels of amyloid, which could reduce the number of unnecessary PET scans.14PubMed Central. Plasma pTau217 predicts continuous brain amyloid levels in preclinical and early Alzheimer’s disease
The practical implication is that a blood test might eventually serve as a screening step: patients whose blood results suggest low amyloid could skip the PET scan entirely, while those with ambiguous or high results would proceed to imaging for precise Centiloid quantification. This tiered approach could dramatically widen access to Alzheimer’s diagnostics, especially in areas without PET scanning facilities.
The Connection Between Amyloid and Tau
Alzheimer’s disease involves two hallmark proteins: amyloid-beta, which forms plaques outside neurons, and tau, which forms tangles inside them. The Centiloid scale measures amyloid, but researchers have been keenly interested in how amyloid Centiloid values relate to tau pathology. Higher Centiloid values are associated with increased tau accumulation, especially in people who are positive for both amyloid and tau pathology. This relationship is strongest in early Alzheimer’s-related brain regions, supporting the “amyloid cascade” model in which amyloid buildup precedes and facilitates tau spread.15PubMed Central. Associations of 18F-RO-948 tau PET with fluid AD biomarkers, Centiloid, and cognition in early AD continuum
This connection has implications for how we interpret a Centiloid score. A person with a Centiloid value of 60 who also has significant tau on a separate PET scan is likely further along in the disease process than someone with the same Centiloid value but little tau. The Centiloid number captures one dimension of disease biology; pairing it with tau imaging and cognitive assessments provides a much fuller picture of where someone stands on the Alzheimer’s continuum.
Technical Variability That Clinicians Should Know About
The Centiloid scale was designed to eliminate variability between tracers and methods, but it has not eliminated all of it. A head-to-head comparison of five different software platforms used to calculate Centiloid values found that while overall agreement was excellent (with an intraclass correlation above 0.96), some platforms produced systematically higher values and others produced slightly lower ones. The limits of agreement reached plus or minus 30 Centiloids in some pairings.16PubMed Central. Inter-software variability of Centiloid quantification in amyloid PET That is a large spread when clinical decisions hinge on whether someone is above or below a threshold in the 20-to-30 range.
The choice of reference region in the brain also matters. Different reference regions can shift Centiloid values by meaningful amounts. One study found that using the pons as a reference region led to an average underestimation of about 23% compared to other reference regions, though the pons actually produced the best test-retest repeatability.17Alzheimer’s & Dementia. Robustness of the Centiloid scale to target and reference regions calculated with a commercial software using [18F]flutemetamol PET images In practice, this means that a Centiloid value generated by one processing pipeline at one center may not be directly comparable to a value from a different pipeline at another center, even though the whole point of the scale was comparability.
The practical recommendation from multiple research groups is straightforward: for any individual patient being monitored over time, use the same software, the same reference region, and ideally the same scanner throughout.18PubMed Central. The Centiloid Scale in Amyloid PET Imaging: Current Role in Alzheimer’s Disease Diagnosis, Treatment Planning, and Monitoring During Anti-Amyloid Therapy Switching platforms between scans could make it look like amyloid has changed when it has not, or mask a real change. For large multi-site trials, harmonization procedures exist but require careful calibration. The scale reduced inter-method variability enormously compared to the raw ratios it replaced; it just has not eliminated variability entirely, and near critical thresholds, those residual differences still matter.
When Amyloid PET Centiloids Do Not Tell the Whole Story
A high Centiloid value does not automatically mean Alzheimer’s disease. Amyloid plaques are common in cognitively normal older adults; by some estimates, a substantial fraction of people over 80 have enough amyloid to produce a positive PET scan. Some of these individuals will never develop dementia in their lifetime. The Centiloid number tells you about one protein in the brain, not about the full constellation of pathology, neurodegeneration, vascular disease, inflammation, and cognitive reserve that determines whether and when someone develops symptoms.
Conversely, a low Centiloid value does not guarantee cognitive health. Other forms of dementia, including frontotemporal dementia, Lewy body dementia, and vascular dementia, involve different pathologies that amyloid PET would not detect. A person with progressive memory loss and a Centiloid score of 5 probably does not have Alzheimer’s disease as the primary cause, which itself is diagnostically useful information. The scale’s value in differential diagnosis lies partly in what a negative result rules out, not just in what a positive result confirms.
As anti-amyloid therapies become more available and blood-based screening tests mature, the Centiloid scale is likely to become a routine part of Alzheimer’s clinical care rather than a tool confined to research settings. But the field is still working through which thresholds to use for which decisions, how to handle the intermediate zone, and how to ensure that different clinical sites produce comparable numbers. The 0-to-100 framework provided a crucial common language; the ongoing work is about making that language precise enough for the decisions it is now being asked to support.