Simplified quantification of PIB amyloid imaging PET studies

Introduction

We recently reported the first positron emission tomography (PET) studies of the [C-11]-labeled amyloid-binding radiotracer, Pittsburgh Compound-B (or PIB), performed in Uppsala, Sweden. These semi-quantitative studies showed greater PIB retention in Alzheimer's disease (AD) patients, relative to healthy control subjects (parietal; p=0.0002) that was consistent with amyloid deposition in AD. Quantitative PIB studies have been performed in Pittsburgh in three groups: AD, mild cognitive impairment (MCI), and healthy controls. The present study compares arterial and reference tissue measures of PIB retention.

Methods

PET studies were performed over 90 min (ECAT HR+, 14±2 mCi) in 16 patients (6 mild-moderate AD: 67±10 yrs; 10 MCI: 72±9 yrs) and 8 controls (65±16 yrs). Arterial blood was collected and metabolite-corrected input functions were determined. Magnetic resonance images were acquired for region-of-interest definition and atrophy correction of the PET data. Regional PIB distribution volume (DV) values were determined using the Logan analysis with either arterial (ART90) or cerebellar (CER90) data as input. PIB retention was assessed using the distribution volume ratio (DVR: regional DV measure normalized to the cerebellar reference DV). For a subset of subjects (5 AD, 5 MCI, 5 controls), analyses also were performed using carotid ROI data (CAR90) as input with arterial metabolite correction or the simplified reference tissue method (SRTM: DVR=BP+1, where BP is binding potential). Results were compared to the Logan ART90 DVR values.

Results

Greatest PIB retention was observed in AD posterior cingulate (PCG DVR: AD=2. 62±0.34, Cont=1.25±0.20). Heterogeneous PIB retention was observed in MCI subjects, which ranged from control to AD levels. Similar PIB retention was noted in cerebellum (n=24, CER DV=3.67±0.5). For the subset of 15 subjects, all methods yielded the same regional rank order of average PIB retention in AD subjects (see Table 1). The CAR90 DVRs agreed best with the ART90 DVRs (PCG: r²=0.98, slope = 0.95), with more bias observed in the CER90 DVR (PCG: r²=0.97, slope = 0.85). The SRTM DVR was most biased but strongly correlated with the ART90 values (PCG: r²=0.92, slope = 0.78). For a given region, the rank order of individual subject DVR values was maintained across the various Logan analyses, while deviation of this rank order was sometimes observed in the SRTM DVRs.

Table 1

Group	Region	Logan DVR ART90	Logan DVR CAR90	Logan DVR CER90	SRTM BP+1
CONTROL	Posterior Cingulate	1.24±0.22	1.30±0.27	1.22±0.17	1.42±0.26
	Parietal	1.29±0.26	1.30±0.26	1.26±0.20	1.33±0.25
	Lateral Temporal Cortex	1.19±0.18	1.16±0.19	1.18±0.13	1.13±0.12
	Cerebellum	3.38±0.34(DV)	7.53±0.71(DV)	−	−
AD	Posterior Cingulate	2.63±0.38	2.61±0.35	2.39±0.26	2.51±0.25
	Parietal	2.55±0.46	2.55±0.42	2.34±0.39	2.31±0.09
	Lateral Temporal Cortex	2.37±0.44	2.37±0.42	2.17±0.31	2.27±0.22
	Cerebellum	3.79±0.55(DV)	7.63±0.85(DV)	−	−

Conclusion

These results support the validity of image-based methods for the quantification of PIB retention in control, MCI, and mild to moderate AD subjects. The carotid method provided excellent results but its usefulness will require validation of radiolabeled metabolites from venous samples. Of the reference tissue methods, the Logan CER90 was less biased than SRTM. The performance of these methods in severe AD subjects and in longitudinal studies remains to be investigated.