Helen Wills Neuroscience Institute, University of California Berkeley, Berkeley, California 94720, USA.
J Nucl Med. 2011 Feb;52(2):173-9. doi: 10.2967/jnumed.110.082057. Epub 2011 Jan 13.
The availability of new PET ligands offers the potential to measure fibrillar β-amyloid in the brain. Nevertheless, physiologic information in the form of perfusion or metabolism may still be useful in differentiating causes of dementia during life. In this study, we investigated whether early (11)C-Pittsburgh compound B ((11)C-PIB) PET frames (perfusion (11)C-PIB [pPIB]) could provide information equivalent to blood flow and metabolism. First, we assessed the similarity of pPIB and (18)F-FDG PET images in a test cohort with various clinical diagnoses (n = 10), and then we validated the results in a cohort of patients with Alzheimer disease (AD) (n = 42; mean age ± SD, 66.6 ± 10.6 y; mean Mini-Mental State Examination [MMSE] score ± SD, 22.2 ± 6.0) or frontotemporal lobar degeneration (FTLD) (n = 31; age ± SD, 63.9 ± 7.1 y, mean MMSE score ± SD, 23.8 ± 6.7).
To identify the (11)C-PIB frames best representing perfusion, we ran on a test cohort an iterative algorithm, including generating normalized (cerebellar reference) perfusion pPIB images across variable frame ranges and calculating Pearson R values of the sum of these pPIB frames with the sum of all (18)F-FDG frames (cerebellar normalized) for all brain tissue voxels. Once this perfusion frame range was determined on the test cohort, it was then validated on an extended cohort and the power of pPIB in differential diagnosis was compared with (18)F-FDG by performing a logistic regression of regions-of-interest tracer measure (pPIB or (18)F-FDG) versus diagnosis.
A 7-min window, corresponding to minutes 1-8 (frames 5-15), produced the highest voxelwise correlation between (18)F-FDG and pPIB (R = 0.78 ± 0.05). This pPIB frame range was further validated on the extended AD and FTLD cohort across 12 regions of interest (R = 0.91 ± 0.09). A logistic model using pPIB was able to classify 90.5% of the AD and 83.9% of the FTLD patients correctly. Using (18)F-FDG, we correctly classified 88.1% of AD and 83.9% of FTLD patients. The temporal pole and temporal neocortex were significant discriminators (P < 0.05) in both models, whereas in the model with pPIB the frontal region was also significant.
The high correlation between pPIB and (18)F-FDG measures and their comparable performance in differential diagnosis are promising in providing functional information using (11)C-PIB PET data. This approach could be useful, obviating (18)F-FDG scans when longer-lived amyloid imaging agents become available.
探索早期（11）C-Pittsburgh 化合物 B（（11）C-PIB）正电子发射断层扫描（PET）帧（灌注（11）C-PIB [pPIB]）是否能提供与血流和代谢等同的信息。
为了识别最能代表灌注的（11）C-PIB 帧，我们在一个测试队列中运行了一个迭代算法，包括生成跨越不同帧范围的归一化（小脑参考）灌注 pPIB 图像，并计算所有脑区体素中这些 pPIB 帧与所有（18）F-FDG 帧（小脑归一化）之和的皮尔逊 R 值。一旦在测试队列中确定了这个灌注帧范围，我们就在一个扩展队列中验证了它，并通过对感兴趣区域示踪剂测量（pPIB 或（18）F-FDG）与诊断进行逻辑回归，比较了 pPIB 在鉴别诊断中的作用与（18）F-FDG 的作用。
7 分钟窗口（分钟 1-8，帧 5-15）产生了（18）F-FDG 和 pPIB 之间最高的体素相关性（R = 0.78 ± 0.05）。这个 pPIB 帧范围在扩展的 AD 和 FTLD 队列中在 12 个感兴趣区域进一步得到了验证（R = 0.91 ± 0.09）。使用 pPIB 的逻辑模型能够正确分类 90.5%的 AD 和 83.9%的 FTLD 患者。使用（18）F-FDG，我们正确分类了 88.1%的 AD 和 83.9%的 FTLD 患者。在这两个模型中，颞极和颞新皮层都是显著的判别因素（P < 0.05），而在使用 pPIB 的模型中，额叶区域也是显著的。
pPIB 与（18）F-FDG 测量值之间的高度相关性及其在鉴别诊断中的可比性能有望为使用（11）C-PIB PET 数据提供功能信息。当更长寿命的淀粉样蛋白成像剂变得可用时，这种方法可能很有用，可以避免进行（18）F-FDG 扫描。
