Background
[11C]-PIB is a novel PET tracer which binds to amyloid plaques in Alzheimer's disease (AD). Using Spectral analysis or Logan analysis with arterial input functions it has been reported that [11C]-PIB PET can distinguish AD patients from healthy controls. For the wider use of PIB PET as a diagnostic marker a non invasive method of analysis is essential. Pathological studies have shown that PIB binds to fibrillar but not to amorphous amyloid. The cerebellum is devoid of fibrillar amyloid in AD and so should provide a reference brain area.
Objective
To determine whether measurement of regional cerebral[11C]-PIB binding using “target to cerebellar ratios” (RATIO) gives comparable findings to spectral analysis (SA) - dependent on an arterial plasma input function) in AD. To correlate regional cortical [11C]-PIB uptake obtained with these methods with performance on recognition memory tests in AD patients and to assess test retest variability.
Methods
12 early AD patients (MMSE scores 15–26) and 9 healthy age-matched control subjects underwent [11C]-PIB PET. All subjects had online arterial sampling and parametric images of the Impulse response function at 75 min were generated using spectral analysis. All subjects were then analysed using a RATIO method where cerebellar grey matter was defined as the tissue reference and ratio images were generated from 60–90 min. Object maps were created by segmenting individual MRIs and spatially transforming the grey matter images into standard stereotaxic MNI space on which a probabilistic atlas was superimposed. [11C]-PIB uptake was correlated with memory tests using ANOVA. 5 AD patients underwent repeat [11C]-PIB PET within 6 weeks of their first scan.
Results
Using spectral analysis, the AD patients showed significantly raised [11C]-PIB binding in frontal (2.3x), temporal (2.1x), and parietal association (2.4x) cortex along with cingulate (2.5x) and occipital cortex (1.8x). Amyloid binding was also detected in striatum (2.4x) but not in thalamus or cerebellum. Using the “RATIO” method there were similar increases in [11C]-PIB binding in frontal (2.1x), temporal (1.9x), and parietal association (2.1x) cortex, cingulate (2.3x) and occipital (1.8x) cortex, and striatum (2.1x) compared to cerebellum. All these regional binding increases were statistically significant, p<0.001. There was no significant increase in regional cerebral [11C]-PIB binding in controls compared to AD or normal cerebellum. The test-retest variability was 6–8% across cortical regions for the RATIO method and 8–10% for spectral analysis. MMSE scores showed a trend towards correlation with cortical amyloid load (p=0.08 RATIO; .09 SA) with both approaches. Performances on the short Warrington Recognition Memory Test for Faces (RMTF) and Words (RMTW) were both significantly correlated (p<0.001) with 11C-PIB binding in cortical association areas using both analytical approaches.
Conclusion
Our findings demonstrate that “Target to cerebellum” 60–90’ ratios provide a robust and non invasive method for analyzing r[11C]-PIB etention in AD patients and give comparable results to the more invasive spectral analysis. Test–retest variability is slightly better with “Target to cerebellum” ratio method and both methods demonstrate that amyloid load correlates with memory performance in Alzheimer's disease.