PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Nucl Med. Author manuscript; available in PMC 2011 September 2.
Published in final edited form as:
PMCID: PMC3166243
NIHMSID: NIHMS316830

Early 11C-PIB Frames and 18F-FDG PET Measures Are Comparable; A Study validated in a Cohort of AD and FTLD

Ara H. Rostomian, MPH,corresponding author1 Cindee Madison, MA,1,2 Gil D. Rabinovici, MD,1,3,4 and William J. Jagust, MD1,2,3,4

Abstract

The availability of new PET ligands offers the potential to measure fibrillar β-amyloid in the brain. Nevertheless, physiological 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 11C-PIB PET frames (perfusion, pPIB) can provide information equivalent to blood flow and metabolism by assessing the similarity of pPIB and 18F-FDG PET images first in a test cohort with various clinical diagnoses (N=10) and then validating the results on a cohort of Alzheimer’s disease (AD, N=42, age 66.6±10.6, MMSE 22.2±6.0) and frontotemporal lobar degeneration (FTLD, N=31, age 63.9±7.1, MMSE 23.8±6.7) patients.

METHODS

To identify the 11C-PIB frames best representing perfusion, an iterative algorithm was run on the test cohort. This included: (1) generating normalized (cerebellar reference) perfusion pPIB images across variable frame ranges, and (2) calculating Pearson’s R values of the sum of these pPIB frames with the sum of all 18F-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 to 18F-FDG by performing a logistic regression of ROI tracer measure (pPIB or 18F-FDG) versus diagnosis.

RESULTS

A seven-minute window, corresponding to minutes 1–8 (frame 5–15) produced the highest voxel-wise correlation between 18F-FDG and pPIB (R=0.78±0.05). This pPIB frame range was further validated on the extended AD and FTLD cohort across 12 ROIs (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 18F-FDG, 88.1% of AD and 83.9% of FTLD patients were classified correctly. The temporal pole and the temporal neocortex were significant discriminators (p<0.05) in both models, whereas in the model with pPIB the frontal region was also significant.

CONCLUSIONS

The high correlation between pPIB and 18F-FDG measures and their comparable performance in differential diagnosis is promising in providing functional information using 11C-PIB PET data. This could be a useful approach, obviating the need for 18F-FDG scans when longer-lived amyloid imaging agents become available

Keywords: Pittsburgh compound-B (11C-PIB), perfusion, 18F-Fluorodeoxyglucose (18F-FDG), Aβ-amyloid plaques, cerebral glucose metabolism

Positron Emission Tomography (PET) has been a useful tool in the study of clinical and basic aspects of dementia. For example, 18F-Fluorodeoxyglucose (18F-FDG) is especially useful in differentiating frontotemporal lobar degeneration (FTLD), often characterized by hypometabolism in frontal and temporal lobes, from Alzheimer’s disease (AD), marked by parietal and temporal hypometabolism that is related to dementia severity(13).

Pittsburgh Compound-B (11C-PIB)(4), a relatively new radiotracer that binds to fibrillar Aβ-amyloid plaques, has been used to characterize AD and differentiate it from FTLD, which is not characterized by β-amyloid plaques(57). The recent introduction of [18F] amyloid ligands makes this approach clinically viable(810). 11C-PIB PET can estimate regional Aβ-amyloid and is generally quantified by either a Binding Potential (BP) or Distribution Volume Ratio (DVR)(11) using the Logan reversible kinetic model and a cerebellar reference region, or by standardized uptake value ratios (SUVRs) which are tissue activity ratios at later times after injection similar to the approach taken with 18F-FDG (12). These approaches are reduced from more complex compartmental models that have been investigated using arterial input functions and dynamic PET data(13). In both 3-tissue, 6-parameter and 2-tissue, 4-parameter models, the term K1 represents transport across the blood-brain barrier which is proportional to flow and tracer extraction. In simplified models that utilize a reference tissue in place of an arterial input function (simplified reference tissue model, SRTM) the term R1 represents the ratio of K1 in the target to K1 in reference tissue (14). Thus, perfusion information is available from these models in the form of K1 information. Early PET frames have been used with other tracers to estimate K1 and thus, indirectly, perfusion (15).

A correlation between metabolism and cerebral perfusion has been long established in normal older individuals and those with degenerative dementias, likely because reduced metabolic demand is coupled to reduced perfusion (16). Values of K1 obtained with PIB have also been found to represent the unidirectional influx of the 11C-PIB tracer in the brain and related to perfusion (17). We would therefore expect to find a correlation between perfusion data from early PIB frames, which we term perfusion PIB (pPIB) and metabolic data from 18F-FDG. Thus 11C-PIB PET can potentially provide two kinds of information: perfusion pPIB, obtained from early time frames which gives a measure comparable to regional metabolism, and DVR or BP, obtained from later time frames, which gives a measure of regional fibrillar Aβ-amyloid plaques.

In this report, we specifically examined two etiologies of dementia, Alzheimer’s disease (AD) and frontotemporal lobar dementia (FTLD) because of their different pathological substrates, clinical features, and metabolic patterns as revealed with 18F-FDG. Research suggests 18F-FDG PET may approach the accuracy of clinical diagnosis and may provide additional information over that obtained with a clinical exam (18, 19). These two disorders thus provide a method for the evaluation of pPIB as a potential clinical tool that could have application to a variety of different dementing illnesses at different stages. For example, combing pPIB with amyloid measurements could be particularly useful in the clinical evaluation of patients with mild cognitive impairment (MCI) in situations where PIB scans might be borderline or equivocal, or in situations when amyloid imaging is negative but functional information might be informative. In addition, FDG scans provide information about disease stage and symptom severity(20), while PIB scans are not strongly related to symptoms either cross-sectionally or longitudinally(21, 22). Thus early frames could be useful in assessing, for example, a treatment response in a clinical trial. The approach of utilizing early PET frames of an amyloid imaging ligand in place of an 18F-FDG scan could become particularly useful with an [18F]-labeled amyloid radiotracer, since such tracers would require subjects to return for a second day of scanning. We therefore performed this study in order to investigate how early 11C-PIB scan data compared to FDG-PET data in a group of subjects with AD and FTLD, since these two groups have been best established as the dementia diagnoses most easily differentiated with FDG-PET.

Materials and Methods

We used a small testing cohort of subjects with a variety of dementias to determine the optimal timeframe representing perfusion from the PIB data. This timeframe was then validated on a larger cohort of AD and FTLD patients. Subsequently, a logistic regression model was used to test the utility of pPIB compared to FDG scans at discriminating these two distinct groups. Finally, a correlation was run between patient’s MMSE and relevant regions in the 18F-FDG and pPIB scans for the AD and FTLD patient’s to determine relationships between cognition and these tracers.

Subjects and Image Acquisition

All subjects were recruited from the University of California San Francisco Memory and Aging Center (UCSF MAC) (7). Patients were diagnosed based on standard research criteria for AD(23) and FTLD(24).

Two cohorts were used in this study: the test cohort (N=10) and the validation cohort (N=73). Characteristics of the validation cohort are shown in Table 1. Subjects in the test cohort had a unique set of clinical diagnoses compared to those in the validation cohort: two subjects had Amyloid Angiopathy (AA), three had corticobasal syndrome, two had Dementia with Lewy Bodies (DLB), two had Mild Cognitive Impairment, and one had Primary Progressive Aphasia (PPA). The average age of the test cohort was 66.4±7.3, with an average education of 17.9±3.9 years and a Mini-Mental State Examination (MMSE) average score of 28±1.8. There were six PIB positive and 4 PIB negative subjects, as determined by visual interpretation.

TABLE 1
Validation Cohort Characteristics

All subjects underwent 11C-PIB and 18F-FDG PET scans at Lawrence Berkeley National Laboratory on a Siemens ECAT EXACT HR PET scanner in three-dimensional acquisition mode. 11C-PIB tracer was synthesized in Lawrence Berkeley National Laboratory’s Biomedical Isotope Facility based on a previously published protocol(25). The 18F-FDG tracer was purchased from a vendor (IBA Molecular Gilroy Pharmacy, Gilroy, CA). Approximately 550 MBq of 11C-PIB was injected as a bolus into an antecubital vein. Dynamic frame acquisition was obtained with the following sequence: 4×25 seconds, 8×30 seconds, 9×60 seconds, 2×180 seconds, 8×300 seconds, and 3×600 seconds, for a total of 90 minutes. Two hours after the 11C-PIB injections, the patients were injected with approximately 370 MBq of 18F-FDG. Six emission frames of five minutes each were acquired beginning 30 minutes after the tracer injection.

Identification of optimal pPIB timeframe

The six 18F-FDG PET frames were realigned and summed using SPM8 and segmented into three tissue classes: gray-matter, white-matter, and non-brain probability images. This segmentation was done to improve the gray matter mask of the template cerebellum. The image was then warped to the Montreal Neurological Institute (MNI) standard space 15O-H2O PET template. A cerebellum reference region in MNI space was created by combining all the cerebellum regions from the Automated Anatomical Labeling (AAL) atlas (26). This cerebellum region was masked with each individual’s PET-defined gray-matter mask (threshold at 0.0), and the mean value was used to intensity normalize the volume on a voxel-wise basis.

Using data from the ten subjects in the testing cohort we iteratively generated potential perfusion 11C-PIB (pPIB) images using the following processing approach. As an initial step, the 11C-PIB frames were aligned using a two-pass approach, first to the sixth frame, then to the mean using SPM8. Because the first five frames have very little signal, they were summed, aligned to the mean, and the parameters were then applied to the individual frames. At this stage we employed an iterative algorithm to generate potential pPIB data that represented the sums of different frame ranges.

The algorithm did the following:

  1. Sum frames X–Y from the PIB scan (X and Y represent distinct time points in the scan protocol, e.g. X=frame1, Y= frame 6).
  2. The summed image was segmented into gray-matter, white-matter, and non-brain probability images. The gray-matter image was thresholded at 0.0 to generate a gray-matter mask.
  3. The summed image was warped to the (MNI) standard space PET template (SPM8), and the derived warp parameters were applied to the gray-matter mask.
  4. The same AAL template cerebellum reference region used with 18F-FDG was again used, and masked with the individual subject’s pPIB PET-derived gray-matter mask.
  5. The mean value from this masked-cerebellum reference region was used to intensity normalize the whole brain on a voxel-wise basis.
  6. A voxel-wise correlation (Pearson’s R) was calculated between the pPIB (sum of frames X to Y) and the single 18F-FDG metabolism image for all brain tissue voxels greater than zero in the pPIB image. These correlations were used to define the optimal pPIB frame range and start time.

Validation of optimal pPIB timeframe

For subjects in the validation cohort, 18F-FDG and 11C-PIB (pPIB) images were processed identically to the test cohort except that only the frames from the optimal pPIB, were summed. This resulted in two images for each subject in the validation cohort, an 18F-FDG measuring metabolism, and a perfusion 11C-PIB (pPIB).

A within subject voxel-wise correlation (Pearson’s R) was calculated for the optimal pPIB image and the single 18F-FDG metabolism image for all brain tissue voxels greater than zero in the pPIB image. In addition, a set of regions of interest (ROIs) were selected to represent the regions that are potentially most involved in AD and FTLD30–33. The software WFU Pickatlas (27) was used to define the ROIs using the Automated Anatomical Labeling (AAL) atlas (26). Twelve regions in the left and right hemispheres of six unique ROIs were selected and used to run within subject correlations (Pearson’s R) to further validate regional similarities between pPIB and 18F-FDG. These ROIs are listed in Table 2 and depicted in Figure 1.

Figure 1
Regions of interest (ROIs) used in the correlations and logistic regression models. The red regions represent the collective ROIs that are affected by AD and the blue regions represent the collective ROIs that are affected by FTLD.
TABLE 2
Definition of ROIS used to extract mean values from 18F-FDG and pPIB images used in correlations and the logistic regression model for each subject.

Statistical Analysis

Regression Models

In order to asses the similarity of pPIB to FDG in a clinically relevant situation, a classification was done using logistic regression to compare the ability of each tracer to discriminate between AD and FTLD. Software used was STATA 10 for Mac (StataCorp. 2007. Stata Statistical Software: Release 10. College Station, TX: StataCorp LP). The logistic regression model, regressing diagnosis (AD or FTLD) versus ROI radiotracer concentration (the original 12 bilateral ROIs were combined resulting in six whole brain ROIs) was performed once on pPIB data and once on FDG data. The classification power using each radiotracer was computed and compared. The goal was not to assess diagnostic accuracy for these diseases, but to compare the performance of the pPIB and 18F-FDG data in classification.

Results

Identification of optimal pPIB timeframe

Figure 2 shows the results of voxel-wise correlation of 18F-FDG with pPIB for various frame ranges. The correlation data show a dip in the initial frames and for short cumulative sums, a prolonged plateau for cumulative sums of around 6 to 10 minutes, and then a gradual decrease for cumulative sums greater than 10 minutes. Within the plateaus, the peaks of the highest correlations are found for summed frames starting at time 0.5 minutes to 1.5 minutes. A seven-minute window, corresponding to minutes 1–8 (starred), is where perfusion measured by the pPIB PET scans has the highest correlation with metabolism, measured with 18F-FDG, for this test cohort (0.78±0.05).

Figure 2
Relationship between pPIB and FDG for different frame start times and duration. The Y axis is the correlation between the two modalities. Each panel reflects a different start time, and the x-axis indicates the different durations of data acquisition. ...

Following selection of this optimal frame range, we evaluated the relationship between our pPIB measure and the estimate of R1 (K1/K1’) using a simplified reference tissue model and a cerebellar input function (14). We used 26 ROIs from the MNI atlas masked with a PET-derived gray-matter mask to determine both R1 and the pPIB value for the ROI. The correlation across all 10 subjects was 0.86, indicating a high concordance of our measure with the model’s regional estimate of tracer influx.

Validation of optimal pPIB timeframe

The optimal pPIB frame range (minute 1–8) as found in the testing cohort also results in a very high voxel-wise correlation in the validation cohort. The mean Pearson’s R for correlation of pPIB and 18F-FDG tracer values across the selected ROIs (twelve ROIs with left and right hemispheres considered separately) is 0.91±0.09, and voxel-wise is 0.80±0.07.

Figure 3 gives example pPIB and 18F-FDG scans for the two diagnostic groups (AD and FTLD). Representative subjects from each group were chosen to show examples of pPIB and 18F-FDG scans that had both high and low correlation, along with corresponding scatter plots for the 12 ROIs.

Figure 3
Example 18F-FDG and pPIB PET scan images of AD and FTLD subjects that had high and low correlations across 12 ROIs. The ROI values resulting in correlations associated with each image are shown in the scatter plots with 18F-FDG on the x-axis and the pPIB ...

The results of independently running the logistic regression models on each of the radiotracers, 18F-FDG or pPIB, is shown in Table 3. Diagnosis was the dependent variable, and the mean value of six ROIS served as independent variables. For every 0.01 unit increase in the pPIB value in the lateral frontal and temporal pole, a subject has higher odds of being diagnosed as AD as opposed to FTLD. Similarly, for every incremental increase of the pPIB value in the temporal neocortex region, a subject has lower odds of being diagnosed as AD, which is expected since this region is more hypometabolic in AD than FTLD. The same pattern and interpretation for the odds ratios are reflected when using 18F-FDG as the radiotracer and performing the logistic regression. The only exception is the lateral frontal region is not significant as a discriminating variable for 18F-FDG.

TABLE 3
Logistic Regression Results and Classification

Table 3 also shows that the odds ratios for the temporal pole and temporal neocortex ROIs, which are significant in both models, share directionality and act similarly in discriminating AD from FTLD. However, pPIB is slightly more accurate than 18F-FDG in correctly classifying AD patients (38 vs 37 out of 42 patients correctly classified). Both models perform equally in terms of classifying FTLD patients (26 out of 31 patients correctly classified).

In order to explore effects of confounding factors of sex, age, education and cognitive status measured by Mini-Mental State Examination (MMSE), these factors were included in a second iteration of the logistic regression model. For the 18F-FDG data temporal pole and temporal neocortex ROIs still remained significant with odds ratios in the same direction as before, 1.2 (p=0.01) and 0.8 (p=0.01), however age was also found to be a significant discriminator with an odds ratio of 1.18 (p=0.007). For the pPIB data the lateral frontal ROI showed a trend (p=0.08) but the temporal pole and temporal neocortex as well as age remained significant with the same direction in their odds ratios respectively, 1.4 (p=0.003), 0.6 (p=0.007), 1.2 (p=0.01).

Finally, in order to investigate whether the two tracers provided similar information about dementia severity, we examined correlations between MMSE scores and both pPIB and 18F-FDG values in the temporal neocortex region in AD patients, and the lateral frontal region in FTLD patients. In the AD patients MMSE was significantly correlated with both 18F-FDG (Pearson’s R = 0.49, p=0.001) and pPIB (Pearson’s R = 0.41, p =0.007). Both tracers in the lateral frontal region of FTLD patients also showed strong correlations to MMSE, 18F-FDG (Pearson’s R 0.59, p = 0.0004), pPIB (Pearson’s R = 0.56, p=0.001).

Discussion

The test cohort showed a high correlation between perfusion, measured by minutes 1 – 8 (frames 5–15) of a 11C-PIB PET scan, and metabolism, measured by 18F-FDG PET. This time frame was validated on a separate larger cohort, and yielded similarly high correlations. Validating the results on a cohort with diagnoses distinct from the test cohort provides further support that the time-window used to define the optimal perfusion image was not dependent on diagnosis, suggesting wider applicability of this methodology. Furthermore, application to a clinical situation suggests that these early PIB frames have similar discriminative power to 18F-FDG images.

While minutes one to eight, corresponding to frames 5–15, were chosen in this study as the optimal frame range, as depicted by a peak in our correlation graph, small variations in this range had similarly high correlations. In fact, the middle frame ranges, after the first noisy minute (frames 1–5) up to minutes 7,8,9 (frames 14,15, and 16) show the highest voxel-wise correlations in the test cohort with a mean Pearson’s R of 0.78±0.05. Thus, the high correlation observed in our choice of minutes one to eight is not necessarily unique to this frame range.

It should also be noted that shape of the results shown in figure 2 seem to be driven by two competing factors, perfusion in the early time frames and Aβ-amyloid binding in latter frames. The initial frames of PIB are not very well correlated with 18F-FDG, possibly due to the noisy characteristic of the initial PIB frames, non-uniform delivery of the tracer, and small sampling windows. In later frames, 11C-PIB binds to Aβ and thus the curve peak declines as a result of low correlation between binding (tissue bound 11C-PIB) and metabolism (18F-FDG). Representative time-activity curves (see figure 4) demonstrate this binding behavior concretely.

Figure 4
11C-PIB Time-activity curves for a patient with Alzheimer’s disease (AD) and frontotemporal lobar degeneration (FTLD) in a target brain region (Posterior cingulate cortex) and reference region (cerebellum). Early times following injection are ...

The strongest evidence of high correlation is in the side-by-side images of pPIB and 18F-FDG depicted in Figure 3. While pPIB images may be noisier, the similarity between the patterns of intensity across the two images is striking. This pattern holds up even for subjects whose Pearson’s R is in the lower range found in this study.

The results of the logistic regression models further demonstrate the utility of these early PIB frames. Both pPIB and 18F-FDG performed similarly in correctly classifying subjects into AD and FTLD diagnosis groups, as the odds ratios obtained from each model are in the same direction and the same ROIs are statistically significant discriminators. This classification is consistent with other results showing significant differences in hypometabolism in these regions when comparing patients with AD and FTLD(3, 7, 28).

One interesting finding from the logistic regression is the better performance of pPIB as a diagnostic discriminator in the lateral frontal region in comparison to 18F-FDG. This may reflect the fact that, especially in young patients with AD, hypometabolism (and brain atrophy) is seen in the frontal lobes(7). Thus this region alone may not always be the best at differentiating the two conditions. Since the subjects in our study were young, it is possible that hypometabolism in this ROI was not a strong discriminator. The reason that the pPIB was a more effective discriminator could be that these early frames contain some information about Aβ-binding, raising the ROI values more in AD than the FTLD subjects(7). Also it should be noted that age is a significant discriminator in the model that accounts for confounders and this is due to the fact that the AD cohort is slightly older than the FTLD cohort.

As depicted in Table 3 the classification ability of both radiotracers pPIB and 18F-FDG are very similar. In computing these classification percentages it should be noted that since multivariate normality assumption is violated we have used logistic regression instead of discriminant function analysis for classification. Due to the use of logistic regression and the absence of a cross-validation approach, our classification accuracy is likely to be overly optimistic. However, the important fact is that despite the possible inflation of these estimates they are very comparable for the two tracers, suggesting that similar information is provided by pPIB and 18F-FDG. We also note that, while we did not exclude subjects with cerebrovascular risk factors, it is possible that ischemia could alter the pPIB data in subjects with extensive vascular disease, leading to another question about the full generalizeability of the results.

Further supporting the comparability between the pPIB and 18F-FDG images, we found that correlations between pPIB and MMSE, and between 18F-FDG and MMSE were similar for relevant ROIs in the FTLD and AD patients. It has been shown that cerebral glucose metabolism measured by 18F-FDG is correlated with cognitive status quantified by MMSE but such correlation is absent between MMSE and cerebral amyloid burden measured by PIB in patients with AD(21). Thus, these early PIB frames provide information distinct from the information provided in later frames.

The future of amyloid imaging is likely to include a number of radiotracers that will be labeled with [18F] and distributed widely(810). The approach that we have taken here could be directly translated to these other compounds and thus find clinical application. The brain penetration and first-pass extraction of these [18F] tracers will be important determinants of how useful this approach is. One of the most likely potential uses of these agents could be in aiding in the differential diagnosis of dementia. It is possible that in many cases the presence or absence of beta-amyloid binding will be adequate to determine whether or not an individual has AD or whether an alternative diagnosis, such as FTLD, should be considered. However, information about physiological change – metabolism or perfusion – may be helpful in both diagnosis and staging of disease. In this situation, the use of early image frames could provide this sort of information without the necessity of an additional patient visit or higher exposure to radioactivity.

Conclusion

The use of early frames of 11C-PIB data provides information that reflects perfusion (pPIB) and which is therefore related to metabolism. This relationship was demonstrated empirically through correlations between pPIB and 18F-FDG images, which showed high correlations with each other and with cognitive status. The approach of using early frames of amyloid imaging data as a proxy for physiological information related to blood flow and metabolism could be particularly useful with the advent of [18F] labeled amyloid imaging agents.

Acknowledgments

This research was supported in part by grants AG027859, AG034570, AG031861 P01-AG1972403, and P50-AG023501 from the National Institute of Health and ZEN08-87090 from the Alzheimer's Association. Support was also received from NIRG-07-59422 from the John Douglas French Alzheimer’s Foundation, and the State of California Department of Health Services Alzheimer's Disease Research Center of California 04-33516. We thank Adi Alkalay for her continuous support with updating and organizing the clinical data.

References

1. Herholz K. FDG PET and differential diagnosis of dementia. Alzheimer Dis Assoc Disord. 1995;9:6–16. [PubMed]
2. Hoffman JM, Welsh-Bohmer KA, Hanson M, et al. FDG PET imaging in patients with pathologically verified dementia. J Nucl Med. 2000;41:1920–1928. [PubMed]
3. Foster NL, Heidebrink JL, Clark CM, et al. FDG-PET improves accuracy in distinguishing frontotemporal dementia and Alzheimer's disease. Brain. 2007;130:2616–2635. [PubMed]
4. Klunk WE, Engler H, Nordberg A, et al. Imaging brain amyloid in Alzheimer's disease with Pittsburgh Compound-B. Ann Neurol. 2004;55:306–319. [PubMed]
5. Jagust W. Mapping brain beta-amyloid. Curr Opin Neurol. 2009;22:356–361. [PMC free article] [PubMed]
6. Nordberg A. PET imaging of amyloid in Alzheimer's disease. Lancet Neurol. 2004;3:519–527. [PubMed]
7. Rabinovici GD, Furst AJ, O'Neil JP, et al. 11C-PIB PET imaging in Alzheimer disease and frontotemporal lobar degeneration. Neurology. 2007;68:1205–1212. [PubMed]
8. Wong DF, Rosenberg PB, Zhou Y, et al. In vivo imaging of amyloid deposition in Alzheimer disease using the radioligand 18F-AV-45 (flobetapir F 18) J Nucl Med. 2010;51:913–920. [PMC free article] [PubMed]
9. Rowe CC, Ackerman U, Browne W, et al. Imaging of amyloid beta in Alzheimer's disease with 18F-BAY94-9172, a novel PET tracer: proof of mechanism. Lancet Neurol. 2008;7:129–135. [PubMed]
10. Vandenberghe R, Van Laere K, Ivanoiu A, et al. 18F-flutemetamol amyloid imaging in Alzheimer disease and mild cognitive impairment: a phase 2 trial. Ann Neurol. 2010;68:319–329. [PubMed]
11. Logan J, Fowler JS, Volkow ND, Wang GJ, Ding YS, Alexoff DL. Distribution volume ratios without blood sampling from graphical analysis of PET data. J Cereb Blood Flow Metab. 1996;16:834–840. [PubMed]
12. Lopresti BJ, Klunk WE, Mathis CA, et al. Simplified quantification of Pittsburgh Compound B amyloid imaging PET studies: a comparative analysis. J Nucl Med. 2005;46:1959–1972. [PubMed]
13. Price JC, Klunk WE, Lopresti BJ, et al. Kinetic modeling of amyloid binding in humans using PET imaging and Pittsburgh Compound-B. J Cereb Blood Flow Metab. 2005;25:1528–1547. [PubMed]
14. Lammertsma AA, Hume SP. Simplified reference tissue model for PET receptor studies. Neuroimage. 1996;4:153–158. [PubMed]
15. Koeppe RA, Gilman S, Joshi A, et al. 11C-DTBZ and 18F-FDG PET measures in differentiating dementias. J Nucl Med. 2005;46:936–944. [PubMed]
16. Silverman DH. Brain 18F-FDG PET in the diagnosis of neurodegenerative dementias: comparison with perfusion SPECT and with clinical evaluations lacking nuclear imaging. J Nucl Med. 2004;45:594–607. [PubMed]
17. Blomquist G, Engler H, Nordberg A, et al. Unidirectional Influx and Net Accumulation of PIB. Open Neuroimag J. 2008;2:114–125. [PMC free article] [PubMed]
18. Silverman DH, Small GW, Chang CY, et al. Positron emission tomography in evaluation of dementia: Regional brain metabolism and long-term outcome. JAMA. 2001;286:2120–2127. [PubMed]
19. Jagust W, Reed B, Mungas D, Ellis W, Decarli C. What does fluorodeoxyglucose PET imaging add to a clinical diagnosis of dementia? Neurology. 2007;69:871–877. [PubMed]
20. Landau SM, Harvey D, Madison CM, et al. Associations between cognitive, functional, and FDG-PET measures of decline in AD and MCI. Neurobiol Aging. 2009 [PMC free article] [PubMed]
21. Furst AJ, Rabinovici GD, Rostomian AH, et al. Cognition, glucose metabolism and amyloid burden in Alzheimer's disease. Neurobiol Aging. 2010 [PMC free article] [PubMed]
22. Jack CR, Jr, Lowe VJ, Weigand SD, et al. Serial PIB and MRI in normal, mild cognitive impairment and Alzheimer's disease: implications for sequence of pathological events in Alzheimer's disease. Brain. 2009;132:1355–1365. [PMC free article] [PubMed]
23. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology. 1984;34:939–944. [PubMed]
24. Neary D, Snowden JS, Gustafson L, et al. Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. Neurology. 1998;51:1546–1554. [PubMed]
25. Mathis CA, Wang Y, Holt DP, Huang GF, Debnath ML, Klunk WE. Synthesis and evaluation of 11C-labeled 6-substituted 2-arylbenzothiazoles as amyloid imaging agents. J Med Chem. 2003;46:2740–2754. [PubMed]
26. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage. 2002;15:273–289. [PubMed]
27. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage. 2003;19:1233–1239. [PubMed]
28. Diehl-Schmid J, Grimmer T, Drzezga A, et al. Decline of cerebral glucose metabolism in frontotemporal dementia: a longitudinal 18F-FDG-PET-study. Neurobiol Aging. 2007;28:42–50. [PubMed]