**2.3.1 [11C]Pittsburgh Compound B ([11C]PiB)**

The most well-characterized and studied radiopharmaceutical for Aβ pathology is [11C]Pittsburgh Compound B ([11C]PiB, (N-methyl-[11C])2-(4'-methylamino-phenyl)-6-OH-

Diagnosis of Dementia Using Nuclear Medicine Imaging Modalities 205

ε4) allele, a genotype strongly associated with AD, are more likely to have a [11C]PiB-

While [11C]PiB has primarily shown utility in differentiating AD patients from healthy controls, this tracer has also been assessed for the differential diagnosis of dementia. [11C]PiB binding patterns allow for great separation between AD and Frontotemporal Lobar Degeneration (FTLD) patients, as those with the frontotemporal and semantic forms of the latter rarely have Aβ plaques *post-mortem* (A. Drzezga, et al., 2008; Engler, et al., 2008; Rabinovici, et al., 2007; Rowe, et al., 2007). Contrastingly, due to the presence of Aβ deposits in a large proportion of patients with Lewy body dementia (DLB) (only 15% of cases represent "pure" DLB with no Aβ pathology), it is harder to distinguish DLB from AD solely based on [11C]PiB binding alone (McKeith, et al., 1996; Rowe, et al., 2007). In comparison to the distribution pattern typical of AD patients, DLB cortical [11C]PiB uptake has been noted to be more variable and lower in a majority of cases (Rowe, et al., 2007). While [11C]PiB cannot distinguish AD and DLB pathology with a high degree of accuracy, this technology could potentially be used to identify Aβ-negative and Aβ-positive DLB subsets. Whether this discrimination signifies differences in therapeutic options remains unknown at this time. In a study by *Rowe et al*., the presence of Aβ deposition (high Aβ load) was associated with a more rapid onset of the full DLB phenotype (Rowe, et al., 2007). [11C]PiB studies, thus, may be able to shed new insights into DLB pathophysiology. Healthy controls have been additional subjects of [11C]PiB studies because 10-30% of cognitively normal elderly people have Aβ plaques at levels comparable to those of AD patients *postmortem* (Aizenstein, et al., 2008; Price and Morris, 1999). Concomitantly, 10-30% of healthy controls with normal cognition show increased cortical [11C]PiB binding (W. E. Klunk, et al., 2004; Mintun, et al., 2006; Morris, et al., 2009; Rowe, et al., 2007). To date, this technology does not allow discrimination between healthy controls with high Aβ load and AD patients. The meaning of these false-positive scans is unclear; while low specificity is possible, high [11C]PiB uptake and thus Aβ load in cognitively normal people most likely represents either pre-clinical AD or "benign" pathology (Mintun, et al., 2006; Morris, et al., 2009; Rowe, et al.,

There is evidence to suggest a prodromal stage of AD does exist, in which Aβ deposition begins in a small subset of adults as a primary event (Jack, et al., 2009; Pike, et al., 2007; Rowe, et al., 2007). As these individuals reach the MCI phase, the amount of Aβ accumulation approximates the levels seen in AD patients (Jack, et al., 2009; Pike, et al., 2007). Upon conversion to AD, Aβ load either plateaus or progresses slowly. Consequently, clinical cognitive decline (or severity of dementia) does not correlate with Aβ load, as measured *in vivo* by [11C]PiB (Jack, et al., 2009; Rowe, et al., 2007). This time course for Aβ accumulation is further supported by a 2-year study in which a majority of 16 AD subjects did not show a significant change in [11C]PiB uptake from baseline, despite decreases in clinical cognitive parameters and temporoparietal metabolism (as measured by [18F]FDG-PET) (Engler, et al., 2006). The results from this study support the use of [11C]PiB as a potential early biomarker for AD, but not as an indicator of disease severity. Other measures, such as [18F]FDG-PET and tau PET imaging agents, which track later-developing biomarkers, correlate better with cognitive decline and thus can be used to assess

To further evidence [11C]PiB's ability to act as an early biomarker for AD, radiotracer studies have been performed in MCI patients. [11C]PiB distribution patterns in MCI patients are

positive scan than a negative one (A. Drzezga, et al., 2009; Storandt, et al., 2009).

2007).

neurodegeneration (Meyer, et al., 2011).

BTA), a 6-hydroxyl-substituted benzothiazole aniline; a team led by Mathis and Klunk developed this 11C-labeled neutral derivative of Thioflavin-T amyloid dye at the University of Pittsburgh in 2002 (W. E. Klunk, et al., 2001; Mathis, et al., 2002). Pre-clinical studies of [11C]PiB have demonstrated that this radiotracer has excellent brain penetration (approximately 7% injected dose per gram at 2 minutes post injection) and initial brain uptake, rapid clearance from normal brain tissue, high binding affinity to Aβ plaques (Ki=0.87±0.18 nM), and moderate lipophilicity (W. E. Klunk, et al., 2001; Mathis, et al., 2002; Mathis, et al., 2003). As a carbon-11 labeled ligand, [11C]PiB has the advantage of delivering a lower radiation burden (whole-body effective dose conversion factor for a 70-kg adult= 4.74±0.8 μSv/MBq; typical administered dose= 489±61 MBq (13±1.65 mCi); effective dose= 2.37±0.53 mSv). Furthermore, [11C]PiB is able to reach a maximum effect size more rapidly than 18F-labeled radiopharmaceuticals, leading to shorter imaging times (50-70 minutes as the optimal time-window) (McNamee, et al., 2009; O'Keefe, et al., 2009; Scheinin, et al., 2007). Of particular interest, *in vitro* autoradiography studies have confirmed that [11C]PiB binds to aggregated, fibrillar Aβ deposits in the cortex, striatum, and cerebral vessel walls, but not to amorphous, cerebellar Aβ deposits (Greenberg, et al., 2008; W. E. Klunk, et al., 2003; Lockhart, et al., 2007). At nanomolar concentrations, however, [11C]PiB does not bind to free soluble amyloid, NFTs, or Lewy bodies (Fodero-Tavoletti, et al., 2007; W. E. Klunk, et al., 2003; Lockhart, et al., 2007).

In 2004, Klunk *et al.* published the results from their first in human proof-of-mechanism study (W. E. Klunk, et al., 2004). With 16 mild AD patients and 9 healthy controls, this study supported the use of [11C]PiB uptake patterns for the reliable discrimination between the two diagnostic groups. AD patients had a twofold increase of [11C]PiB uptake over healthy controls, as measured by regional and average neocortical standard uptake value ratios (SUVRs) with the cerebellum as the reference region, specifically in the frontal, temporal, parietal, and lateral temporal cortices, portions of the occipital cortex, and the striatum. This pattern of [11C]PiB uptake is consistent with the distribution of Aβ plaques in previous *post-mortem* studies (Thal, et al., 2002). Three AD patients, however, had [11C]PiB retention levels in the range typical for healthy controls. Contrastingly, all healthy controls, except one, showed little or no cortical [11C]PiB binding. Non-specific binding was present in white matter areas, but at identical levels in healthy controls and AD patients. This study additionally cited a significant inverse correlation between [11C]PiB uptake and cerebral metabolic rate, especially in the parietal cortex, but only when healthy controls and AD patients were pooled together. Despite this significant trend, [11C]PiB uptake levels provided higher discriminatory power than cerebral metabolic rate in differentiating AD patients from healthy controls.

Since this groundbreaking study, the number of investigations using [11C]PiB in humans, as well as research applications for this radiopharmaceutical, have greatly proliferated. Numerous studies have been dedicated to determining the relationship between [11C]PiB retention and other AD biomarkers, as well as clinical features/parameters. For example, [11C]PiB uptake negatively correlates with CSF Aβ1-42 levels in AD patients and healthy controls (Fagan, et al., 2007; Forsberg, et al., 2010). Furthermore, tracer binding is directly proportional to CSF tau in MCI patients, but not AD patients, cerebral atrophy as measured by MRI in AD subjects (although NFTs correlate better), and temporoparietal hypometabolism as measured by [18F]FDG-PET (Forsberg, et al., 2008; W. E. Klunk, et al., 2004; Storandt, et al., 2009). In addition, subjects with at least one Apolipoprotein E ε4 (ApoE

BTA), a 6-hydroxyl-substituted benzothiazole aniline; a team led by Mathis and Klunk developed this 11C-labeled neutral derivative of Thioflavin-T amyloid dye at the University of Pittsburgh in 2002 (W. E. Klunk, et al., 2001; Mathis, et al., 2002). Pre-clinical studies of [11C]PiB have demonstrated that this radiotracer has excellent brain penetration (approximately 7% injected dose per gram at 2 minutes post injection) and initial brain uptake, rapid clearance from normal brain tissue, high binding affinity to Aβ plaques (Ki=0.87±0.18 nM), and moderate lipophilicity (W. E. Klunk, et al., 2001; Mathis, et al., 2002; Mathis, et al., 2003). As a carbon-11 labeled ligand, [11C]PiB has the advantage of delivering a lower radiation burden (whole-body effective dose conversion factor for a 70-kg adult= 4.74±0.8 μSv/MBq; typical administered dose= 489±61 MBq (13±1.65 mCi); effective dose= 2.37±0.53 mSv). Furthermore, [11C]PiB is able to reach a maximum effect size more rapidly than 18F-labeled radiopharmaceuticals, leading to shorter imaging times (50-70 minutes as the optimal time-window) (McNamee, et al., 2009; O'Keefe, et al., 2009; Scheinin, et al., 2007). Of particular interest, *in vitro* autoradiography studies have confirmed that [11C]PiB binds to aggregated, fibrillar Aβ deposits in the cortex, striatum, and cerebral vessel walls, but not to amorphous, cerebellar Aβ deposits (Greenberg, et al., 2008; W. E. Klunk, et al., 2003; Lockhart, et al., 2007). At nanomolar concentrations, however, [11C]PiB does not bind to free soluble amyloid, NFTs, or Lewy bodies (Fodero-Tavoletti, et al., 2007; W. E. Klunk, et

In 2004, Klunk *et al.* published the results from their first in human proof-of-mechanism study (W. E. Klunk, et al., 2004). With 16 mild AD patients and 9 healthy controls, this study supported the use of [11C]PiB uptake patterns for the reliable discrimination between the two diagnostic groups. AD patients had a twofold increase of [11C]PiB uptake over healthy controls, as measured by regional and average neocortical standard uptake value ratios (SUVRs) with the cerebellum as the reference region, specifically in the frontal, temporal, parietal, and lateral temporal cortices, portions of the occipital cortex, and the striatum. This pattern of [11C]PiB uptake is consistent with the distribution of Aβ plaques in previous *post-mortem* studies (Thal, et al., 2002). Three AD patients, however, had [11C]PiB retention levels in the range typical for healthy controls. Contrastingly, all healthy controls, except one, showed little or no cortical [11C]PiB binding. Non-specific binding was present in white matter areas, but at identical levels in healthy controls and AD patients. This study additionally cited a significant inverse correlation between [11C]PiB uptake and cerebral metabolic rate, especially in the parietal cortex, but only when healthy controls and AD patients were pooled together. Despite this significant trend, [11C]PiB uptake levels provided higher discriminatory power than cerebral metabolic rate in differentiating AD

Since this groundbreaking study, the number of investigations using [11C]PiB in humans, as well as research applications for this radiopharmaceutical, have greatly proliferated. Numerous studies have been dedicated to determining the relationship between [11C]PiB retention and other AD biomarkers, as well as clinical features/parameters. For example, [11C]PiB uptake negatively correlates with CSF Aβ1-42 levels in AD patients and healthy controls (Fagan, et al., 2007; Forsberg, et al., 2010). Furthermore, tracer binding is directly proportional to CSF tau in MCI patients, but not AD patients, cerebral atrophy as measured by MRI in AD subjects (although NFTs correlate better), and temporoparietal hypometabolism as measured by [18F]FDG-PET (Forsberg, et al., 2008; W. E. Klunk, et al., 2004; Storandt, et al., 2009). In addition, subjects with at least one Apolipoprotein E ε4 (ApoE

al., 2003; Lockhart, et al., 2007).

patients from healthy controls.

ε4) allele, a genotype strongly associated with AD, are more likely to have a [11C]PiBpositive scan than a negative one (A. Drzezga, et al., 2009; Storandt, et al., 2009).

While [11C]PiB has primarily shown utility in differentiating AD patients from healthy controls, this tracer has also been assessed for the differential diagnosis of dementia. [11C]PiB binding patterns allow for great separation between AD and Frontotemporal Lobar Degeneration (FTLD) patients, as those with the frontotemporal and semantic forms of the latter rarely have Aβ plaques *post-mortem* (A. Drzezga, et al., 2008; Engler, et al., 2008; Rabinovici, et al., 2007; Rowe, et al., 2007). Contrastingly, due to the presence of Aβ deposits in a large proportion of patients with Lewy body dementia (DLB) (only 15% of cases represent "pure" DLB with no Aβ pathology), it is harder to distinguish DLB from AD solely based on [11C]PiB binding alone (McKeith, et al., 1996; Rowe, et al., 2007). In comparison to the distribution pattern typical of AD patients, DLB cortical [11C]PiB uptake has been noted to be more variable and lower in a majority of cases (Rowe, et al., 2007). While [11C]PiB cannot distinguish AD and DLB pathology with a high degree of accuracy, this technology could potentially be used to identify Aβ-negative and Aβ-positive DLB subsets. Whether this discrimination signifies differences in therapeutic options remains unknown at this time. In a study by *Rowe et al*., the presence of Aβ deposition (high Aβ load) was associated with a more rapid onset of the full DLB phenotype (Rowe, et al., 2007). [11C]PiB studies, thus, may be able to shed new insights into DLB pathophysiology. Healthy controls have been additional subjects of [11C]PiB studies because 10-30% of cognitively normal elderly people have Aβ plaques at levels comparable to those of AD patients *postmortem* (Aizenstein, et al., 2008; Price and Morris, 1999). Concomitantly, 10-30% of healthy controls with normal cognition show increased cortical [11C]PiB binding (W. E. Klunk, et al., 2004; Mintun, et al., 2006; Morris, et al., 2009; Rowe, et al., 2007). To date, this technology does not allow discrimination between healthy controls with high Aβ load and AD patients. The meaning of these false-positive scans is unclear; while low specificity is possible, high [11C]PiB uptake and thus Aβ load in cognitively normal people most likely represents either pre-clinical AD or "benign" pathology (Mintun, et al., 2006; Morris, et al., 2009; Rowe, et al., 2007).

There is evidence to suggest a prodromal stage of AD does exist, in which Aβ deposition begins in a small subset of adults as a primary event (Jack, et al., 2009; Pike, et al., 2007; Rowe, et al., 2007). As these individuals reach the MCI phase, the amount of Aβ accumulation approximates the levels seen in AD patients (Jack, et al., 2009; Pike, et al., 2007). Upon conversion to AD, Aβ load either plateaus or progresses slowly. Consequently, clinical cognitive decline (or severity of dementia) does not correlate with Aβ load, as measured *in vivo* by [11C]PiB (Jack, et al., 2009; Rowe, et al., 2007). This time course for Aβ accumulation is further supported by a 2-year study in which a majority of 16 AD subjects did not show a significant change in [11C]PiB uptake from baseline, despite decreases in clinical cognitive parameters and temporoparietal metabolism (as measured by [18F]FDG-PET) (Engler, et al., 2006). The results from this study support the use of [11C]PiB as a potential early biomarker for AD, but not as an indicator of disease severity. Other measures, such as [18F]FDG-PET and tau PET imaging agents, which track later-developing biomarkers, correlate better with cognitive decline and thus can be used to assess neurodegeneration (Meyer, et al., 2011).

To further evidence [11C]PiB's ability to act as an early biomarker for AD, radiotracer studies have been performed in MCI patients. [11C]PiB distribution patterns in MCI patients are

Diagnosis of Dementia Using Nuclear Medicine Imaging Modalities 207

this proof-of-concept study. After 80 minutes post injection, most regions of the neocortex showed a large difference in SUVRs (with the cerebellum as the reference region) between AD patients and healthy controls, except in the medial temporal cortex, which is more prone to NFT buildup than amyloid deposition, and the occipital cortex. This spatial distribution of [18F]flutemetamol uptake in AD patients closely resembles that typically seen in its parent molecule, [11C]PiB. Interestingly however, non-specific binding in white matter was more pronounced, but not statistically significant, in healthy controls injected with [18F]flutemetamol in comparison to when using [11C]PiB (Fodero-Tavoletti, et al., 2009). While this study showed that [18F]flutemetamol PET imaging can be used to differentiate AD patients and healthy controls, 2 AD patients had particular regional SUVRs within the range seen in healthy controls. These results are comparable to previous [11C]PiB studies, in which 10-20% of clinically diagnosed AD patients did not show high cortical tracer uptake (W. E. Klunk, et al., 2004). Conversely, one healthy control had cortical SUVRs in line with those seen in AD patients. One possible explanation is that high white matter binding led to increased cortical values. The proportion of [18F]flutemetamol-positive healthy controls in this study, however, is comparable to the 10-30% of elderly healthy controls who have

increased [11C]PiB brain uptake at levels indistinguishable from AD patients.

misclassifications of the scans by the visual readers.

from this trial are pending (GEHC, Accessed 2011).

**2.3.3 [18F]Florbetaben ([18F]BAY94-9172 or [18F]AV-1/ZK)** 

Based on the positive Phase I results, [18F]flutemetamol continued to be investigated in a clinical Phase II capacity (Vandenberghe, et al., 2010). Twenty-seven patients with clinically probable AD, 20 patients with amnestic MCI, 15 elderly healthy controls, and 10 younger healthy controls were used to determine the efficacy of blinded visual assessments of [18F]flutemetamol scans as well as to directly measure [18F]flutemetamol against its parent molecule [11C]PiB in terms of its discriminatory power. Researchers found that mean SUVRs in the frontal cortex, lateral temporal cortex, parietal cortex, anterior/posterior cingulate, and striatum were significantly higher in AD patients than in the elderly healthy controls. These results are consistent with the Phase I clinical study for [18F]flutemetamol. Based on blinded visual assessments of [18F]flutemetamol scans, 25 of 27 scans from AD subjects and 1 of 15 scans from the elderly healthy controls were PET-positive, corresponding to a sensitivity of 93.1% and a specificity of 93.3% against the clinical standard of truth. For MCI patients, 9 of 20 subjects were assigned to the high tracer uptake category. The proportion of [18F]flutemetamol-positive scans for MCI patients is comparable to that reported for [11C]PiB (Forsberg, et al., 2008). Additionally, investigators found that the test-retest variability ranged from 1 to 4%, which is lower than that reported for [11C]PiB. Most important to the validation of this radiotracer is that both visually and quantitatively, [18F]flutemetamol uptake was highly concordant with that of [11C]PiB for both AD and MCI patients. However, non-specific binding was greater with [18F]flutemetamol. Regardless, as was seen in Phase I clinical studies, high white matter uptake did not lead to any

Before clinical application, flutemetamol PET imaging needs to be tested against histopathology findings at autopsy. Thus, GE Healthcare is currently organizing and recruiting for an ongoing [18F]flutemetamol Phase III clinical study (Clinical Trial NCT01165554) that will include patients willing to undergo *post-mortem* studies. Results

[18F]Florbetaben ([18F]BAY94-9172, previously [18F]AV-1/ZK, *trans*-4-(N-methyl-amino)-4'- [2-[2-(2-[18F]fluoro-ethoxy)-ethoxy]-ethoxy]-stilbene) is a fluorinated polyethylene glycol

dichotomous, with one subset of MCI patients showing abundant "AD-like" neocortical binding ([11C]PiB-positive) and the other subset showing low, non-specific binding ([11C]PiB-negative) (Forsberg, et al., 2008; Okello, et al., 2009; Pike, et al., 2007; Rowe, et al., 2007). As only 40-60% of MCI patients progress to AD, longitudinal studies are needed to determine if this bimodal distribution pattern of [11C]PiB uptake accurately predicts those who will convert to AD (Forsberg, et al., 2008; Kukull, et al., 1990; Okello, et al., 2009; Petersen, et al., 2009). In a study by Forsberg *et al.*, 7 out of 21 tested MCI patients converted to AD after 8.1±6.0 months (Forsberg, et al., 2008). Interestingly, there were detectable group differences between the 7 MCI converters and the 14 non-converters. MCI converters were shown to have lower levels of CSF Aβ1-42 and MMSE test scores compared to non-converters. Additionally, MCI converters were more likely to be ApoE ε4 carriers (85%) than were nonconverters (57%). Most importantly, MCI converters had high [11C]PiB uptake in the frontal, parietal, and temporal cortices and the posterior cingulum, similar to levels in AD patients. Contrastingly, MCI non-converters had significantly lower cortical [11C]PiB retention, indistinguishable from healthy controls. These promising results demonstrate the prognostic value of [11C]PiB for predicting which MCI patients will progress to AD.

[11C]PiB has now been used in a large number of subjects, consistently showing high sensitivity and specificity in detecting cerebral amyloid deposition *in vivo* with high intraand inter-reader agreement (W. E. Klunk and Mathis, 2008). Due to the short physical halflife of carbon-11 (20.4 minutes), however, [11C]PiB is limited in clinical availability. As a result, Aβ tracers that are radiolabeled with fluorine-18, a radioisotope with a considerably longer half-life (109.4 minutes) than carbon-11, have been developed. Fluorine-18 labeled Aβ PET tracers do not require on-site cyclotrons for their production, thus allowing for a more widespread distribution of this imaging technology.
