OUP user menu

Carbon-11-Pittsburgh compound B positron emission tomography imaging of amyloid deposition in presenilin 1 mutation carriers

William D. Knight, Aren A. Okello, Natalie S. Ryan, Federico E. Turkheimer, Sofia Rodríguez Martinez de Llano, Paul Edison, Jane Douglas, Nick C. Fox, David J. Brooks, Martin N. Rossor
DOI: http://dx.doi.org/10.1093/brain/awq310 293-300 First published online: 17 November 2010

This article has a correction. Please see:

Summary

11Carbon-Pittsburgh compound B positron emission tomography studies have suggested early and prominent amyloid deposition in the striatum in presenilin 1 mutation carriers. This cross-sectional study examines the 11Carbon-Pittsburgh compound B positron emission tomography imaging profiles of presymptomatic and mildly affected (mini-mental state examination ≥20) carriers of seven presenilin 1 mutations, comparing them with groups of controls and symptomatic sporadic Alzheimer’s disease cases. Parametric ratio images representing 11Carbon-Pittsburgh compound B retention from 60 to 90 min were created using the pons as a reference region and nine regions of interest were studied. We confirmed that increased amyloid load may be detected in presymptomatic presenilin 1 mutation carriers with 11Carbon-Pittsburgh compound B positron emission tomography and that the pattern of retention is heterogeneous. Comparison of presenilin 1 and sporadic Alzheimer’s disease groups revealed significantly greater thalamic retention in the presenilin 1 group and significantly greater frontotemporal retention in the sporadic Alzheimer’s disease group. A few individuals with presenilin 1 mutations showed increased cerebellar 11Carbon-Pittsburgh compound B retention suggesting that this region may not be as suitable a reference region in familial Alzheimer’s disease.

  • Alzheimer’s disease
  • PET imaging
  • genetics

Introduction

11Carbon-Pittsburgh compound B positron emission tomography (11C-PiB PET) has, to date, been the most successful in vivo imaging marker of cerebral β-amyloid. 11C-PiB binds with high affinity to both fibrillar and vascular amyloid (Lockhart et al., 2005, 2007; Johnson et al., 2007) and discriminates clinically probable patients with Alzheimer’s disease from healthy controls. Sporadic Alzheimer’s disease cases show 2-fold increases in 11C-PiB retention in the cingulate and neocortical association areas known to be targeted by β-amyloid deposition (Klunk et al., 2004). 11C-PiB retention correlates with rates of cerebral atrophy in Alzheimer’s disease (Archer et al., 2006), with decreased levels of β-amyloid1–42 in the cerebrospinal fluid of demented and non-demented subjects (Fagan et al., 2006) and with post-mortem measurement of insoluble β-amyloid levels (Ikonomovic et al., 2008).

Patients with mild cognitive impairment who are ‘amyloid positive’ on 11C-PiB PET scanning are likely to have early Alzheimer’s disease (Wolk et al., 2009). Increased 11C-PiB retention is present in around 50% of amnestic mild cognitive impairment cases (Kemppainen et al., 2007) and this subgroup progresses more rapidly to Alzheimer’s disease than amyloid negative cases (Forsberg et al., 2008; Villemagne et al., 2008; Okello et al., 2009). 11C-PiB PET and MRI confer complementary information when used in combination (Jack et al., 2008).

While 11C-PiB PET localizes the distribution and density of fibrillar β-amyloid plaques, post-mortem studies have shown that plaque load correlates poorly with functional or cognitive impairment (Giannakopoulos et al., 2003). It seems likely that the density and distribution of soluble, neurotoxic β-amyloid oligomers are more relevant to neuronal dysfunction (Lesne et al., 2006), but these are not detected by thioflavin based radioligands such as Pittsburgh compound B. Additionally, 11C-PiB retention is elevated in as many as 40% of cognitively normal elderly controls (Mintun et al., 2006; Rowe et al., 2007; Aizenstein et al., 2008; Gomperts et al., 2008), in keeping with the finding that 30% of healthy people over 75 years of age have cerebral β-amyloid deposition at post-mortem (Price and Morris, 1999). Whether members of this group are/were destined to develop clinical Alzheimer’s disease is not yet certain. However, it now seems likely that such amyloid deposition is not a consequence of normal ageing, but rather that it represents preclinical Alzheimer’s disease (Villemagne et al., 2008).

Familial, autosomal dominant Alzheimer’s disease is characterized by accumulation of β-amyloid, and provides a potential model for studying the sporadic condition. In contrast to sporadic Alzheimer’s disease, clinical disease onset can be predicted in familial Alzheimer’s disease so it could be used to study preclinical disease; however, an obstacle to this is the rarity (<1% of all cases) of pathogenic familial Alzheimer’s disease mutations (Campion et al., 1999). The most common are those of the presenilin 1 (PSEN1) gene on chromosome 14, with over 170 mutations described (http://www.molgen.ua.ac.be/ADMutations). Previous studies of familial Alzheimer’s disease have suggested patterns of amyloid deposition that differ from those seen in sporadic Alzheimer’s disease in that prominent striatal (Klunk et al., 2007; Koivunen et al., 2008; Remes et al., 2008) or cerebellar (Verkkoniemi et al., 2001) 11C-PiB retention can be present. The latter finding presents a potential problem—the cerebellum is a popular reference region of non-specific tracer binding in sporadic Alzheimer’s disease amyloid imaging studies as it is a region usually held to be relatively free of plaque amyloid. In this study, we aimed to characterize the 11C-PiB PET profiles of seven PSEN1 mutation carriers at the preclinical and early stages of their illness and compare them to those of healthy controls and individuals with sporadic Alzheimer’s disease. To achieve this, we have used early cross-sectional data from a larger prospective, longitudinal 11C-PiB PET cohort study. We have also addressed the issue of appropriate reference regions, investigating the pons as an alternative to the cerebellum, an approach previously employed by Klunk et al. (2007).

Materials and methods

Table 1 details subject demographics. Seven presymptomatic or mildly affected (Mini-Mental State Examination ≥20) PSEN1 mutation carriers, 10 patients with sporadic Alzheimer’s disease and 10 healthy controls (including four members of autosomal dominant familial Alzheimer’s disease families with negative predictive tests for the causative mutation) underwent MRI and 11C-PiB PET imaging. The PSEN1 mutations studied were Y115C (Cruts et al., 1998), M139V (Clark et al., 1995), M146I (Jorgensen et al., 1996), L171P (Ramirez-Duenas et al., 1998), E184D (Yasuda et al., 1997), R278I (Godbolt et al., 2004) and intron 4 (Tysoe et al., 1998; Janssen et al., 2002). Exclusion criteria comprised a current or recent history of drug or alcohol abuse/dependence, pregnancy, inability to undergo PET scanning (the most common reason envisaged was an inability to tolerate the process of reclining supine for 90 min) and a history of cancer within the last five years (except any non-melanoma skin cancer or prostate cancer).

View this table:
Table 1

Subject demographics

Sex
GroupFMMean age (range)Mean MMSE (range)
Controls7347.7 (25–66)NA
Sporadic Alzheimer’s disease4661.9 (51–69)21.7 (12–26)
PSEN1—affected2051.5 (40–63)22 (21–23)
PSEN1—PMC4134.6 (31–40)28.7 (27–30)
PSEN1—total6138.8 (31–63)27.2 (21–30)
  • MMSE = Mini-Mental State Examination; NA = not applicable; PMC = presymptomatic mutation carriers.

Presymptomatic mutation carriers were, on average, 7.2 years younger than the documented mean age at symptom onset for their family. Mean Mini-Mental State Examination (Folstein et al., 1975) score for the PSEN1 group was 27.2 years (range 21–30 years). Subjects were classified as symptomatic if cognitive symptoms were reported by the subject, corroborated on interview with a close informant, and confirmed by performance <5th percentile in at least one domain on formal neuropsychological testing. Whole brain and regional uptake from a single, baseline 11C-PiB PET scan were examined. The study received approval from the ethics committee of the National Hospital for Neurology and Neurosurgery and Institute of Neurology, and the Hammersmith Hospitals Trust. Permission to administer radiation was granted by the Administration of Radioactive Substances Advisory Committee (ARSAC) UK.

Patients were approached by the investigators and oral and written information provided prior to consent. All subjects and controls underwent 11C-PiB PET imaging at the Hammersmith Hospital using a Siemens ECAT EXACT HR+ scanner in 3D mode as described previously (Edison et al., 2007). Each scan participant received a bolus administration of up to 370 MBq 11C-PiB intravenously followed by a 90 min dynamic PET scan. 11C-PiB was manufactured by Hammersmith Imanet, GE Healthcare and a 10 min transmission scan was performed prior to the injection of 11C-PiB.

All mutation carriers and controls had anatomical T1-weighted, volumetric MRI to allow co-alignment of parametric PET images of 11C-PiB binding, permitting anatomical localization of regions of interest. Where possible, MRI was performed on the same day as the PET scanning, although an interval between the two scans of up to three months was deemed acceptable.

Positron emission tomography image analysis

Parametric ratio images of 11C-PiB retention were first created using the pons as a reference region of non-specific tracer retention. The use of the pons rather than cerebellum as a reference region in PET studies is not novel (Minoshima et al., 1995; Klunk et al., 2007) and is supported by data suggesting it is a relatively amyloid-free region (Kyriakides et al., 1994; Thal et al., 2002). The late (60–90 min) summation images of tracer uptake were co-registered to each subject’s MRI using SPM2 (Statistical Parametric Mapping; Wellcome department of Imaging Neuroscience, University College London, UK) software. Each late summation image was then normalized into Montreal Neurological Institute space using individual MRIs as a template. Activity in the pons was sampled using a region of interest placed manually on the corresponding normalized MRI image, allowing the calculation of a pontine uptake value and subsequent creation of the target region:pons ratio image. Using an individualized anatomical atlas, created using analyse software and an in-house probabilistic brain atlas (Hammers et al 2003), we sampled 11C-PiB uptake for each subject in the anterior and posterior cingulate, the thalamus, striatum and the frontal, temporal, parietal, occipital and cerebellar cortices. Mean regional ratios of 11C-PiB by group are detailed in Table 2. In order to allow comparison with studies that have used the cerebellum as a reference region we also calculated regional ratios relative to the cerebellum—these are given in Supplementary Table 3.

View this table:
Table 2

Mean regional 11C-PiB uptake ratios by group (Pontine reference region)

11C-PiB uptake ratio (SD)
Regions of interestControlsSporadic Alzheimer’s diseasePSEN1
Anterior cingulate0.69 (0.08)1.44 (0.18)1.24 (0.22)
Posterior cingulate0.72 (0.05)1.42 (0.18)1.24 (0.24)
Thalamus0.68 (0.07)0.88 (0.11)1.09 (0.14),*
Striatum0.71 (0.06)1.25 (0.11)1.20 (0.35)
Frontal0.67 (0.08)1.30 (0.13)1.08 (0.17),*
Temporal0.68 (0.07)1.18 (0.14)0.99 (0.14),*
Parietal0.67 (0.07)1.27 (0.15)1.09 (0.17)
Occipital0.70 (0.08)1.12 (0.18)1.04 (0.14)
Cerebellum0.62 (0.07)0.63 (0.07)0.75 (0.17)
  • P < 0.05 versus controls; *P <0.05 versus sporadic Alzheimer’s disease.

Statistical analysis

Statistical interrogation was performed using Statistical Package for the Social Sciences (release 17.0, SPSS Inc.). Mean regional 11C-PiB region : pons ratios were analysed to detect significant differences. Groups were compared using the Mann–Whitney test. P-values were corrected for the total number of comparisons using the Hochberg Correction combined with the P-plot estimation of the number of null hypotheses in the set (Turkheimer et al., 2001). See Supplementary material for data tables.

Results

Across all regions sampled, mean region : pons 11C-PiB uptake ratios were significantly increased in the sporadic Alzheimer’s disease group when compared with controls apart from the cerebellum.

When PSEN1 and control groups were compared, all region : pons ratios with the exception of the cerebellum (P = 0.133) showed significant increases in the former that survived correction for multiple comparisons. When regional comparisons were made using the cerebellum as the reference region, the temporal and striatal 11C-PiB uptake ratios were no longer significantly different between the groups, after correction for multiple comparisons, but 11C-PiB uptake in all other regions was significantly raised in PSEN1 compared with controls.

When PSEN1 and sporadic Alzheimer’s disease groups were compared, thalamus (PSEN1 > sporadic Alzheimer’s disease), frontal and temporal regions (sporadic Alzheimer’s disease > PSEN1) demonstrated significant differences that survived multiple comparison corrections. When the same comparison was made using the cerebellum as the reference region (Supplementary Table 3) the sporadic group had significantly greater 11C-PiB retention ratios (relative to cerebellum) in parietal and occipital as well as frontal and temporal regions.

Two mutation carriers (E184D and intron 4 mutations) had higher striatal 11C-PiB retention ratios than the mean for the sporadic Alzheimer’s disease group (1.71 and 1.54, respectively). Figures 1 and 2 demonstrate regional variation in11C-PiB retention for PSEN1 and sporadic Alzheimer’s disease groups when compared with controls. Supplementary Figs 4 and 5 show equivalent scatter plots where cerebellum is used as the reference region. Table 2 shows mean regional 11C-PiB uptake ratios by group and Fig. 3 shows three transaxial and parasagittal 11C-PiB scans with different binding patterns in three separate PSEN1 mutation carriers.

Figure 1

Univariate scatter plot showing pontine 11C-PiB retention ratios for PSEN1 subjects (f1–f7) versus controls (c1–c10). f2 and f4 = affected/symptomatic PSEN1 mutations carriers. ROI = region of interest.

Figure 2

Univariate scatter plot showing pontine 11C-PiB retention ratios for subjects with sporadic Alzheimer’s disease (s1–s10) versus controls (c1–c10). ROI = region of interest.

Figure 3

Transaxial and parasagittal 11C-PiB images (pontine reference region) in PSEN1 mutation carriers showing heterogeneity of 11C-PiB binding pattern. (A) Increased cerebellar binding in PSEN1 M146I presymptomatic mutation carriers. (B) Increased striatal binding in an affected PSEN1 E184D mutation carrier (f2 in Fig. 1). (C) Typical Alzheimer’s disease-like pattern in an affected PSEN1 R278I mutation carrier (f4 in Fig. 1).

Discussion

At a group level, we observed significantly higher global 11C-PiB uptake in established sporadic Alzheimer’s disease cases compared with the PSEN1 group, where the majority were presymptomatic and around seven years before the anticipated onset of symptoms. However, we have confirmed previous findings (Klunk et al., 2007; Remes et al., 2008) that it is possible to detect significantly elevated 11C-PiB retention in the brains of individuals destined to develop clinical Alzheimer’s disease before symptoms arise. Structural and metabolic changes are already known to occur early in patients with familial Alzheimer’s disease (Matsushita et al., 2002; Ridha et al., 2006) as they are in sporadic Alzheimer’s disease (Minoshima et al., 1997; Desikan et al., 2008) and mild cognitive impairment (Kemppainen et al., 2007). In sporadic Alzheimer’s disease, the earliest 11C-PiB retention may occur in the retrosplenial cortices, particularly the precuneus (Mintun et al., 2006) although early retention has been noted in the striatum as well as the frontal, parietal, temporal and occipital cortices (Klunk et al., 2004).

The rarity of familial Alzheimer’s disease mutations means that rather less is known about their associated 11C-PiB PET profiles. One study of the PSEN1 C410Y and PSEN1 A426P mutations showed elevated 11C-PiB retention beginning in the striatum of presymptomatic and affected mutation carriers, with neocortical areas involved later (Klunk et al., 2007). A similar pattern has been observed in those with early-onset disease associated with atypical clinical (spastic paraparesis) and neuropathological (cotton wool plaques) features (Koivunen et al., 2008) and in amyloid precursor protein (APP) gene duplication cases (Remes et al., 2008). Such studies raise important issues about the neuropathological profile of early-onset forms of Alzheimer’s disease, although the extent to which they are representative remains unknown, particularly as other investigators have found a more typical, sporadic Alzheimer’s disease-like 11C-PiB profile in symptomatic APP mutation-carriers (Theuns et al., 2006). Our study included two individuals (E184D and intron 4 mutations) with higher striatal 11C-PiB retention ratios than the mean for the sporadic Alzheimer’s disease group (1.71 and 1.54, respectively) although, at a group level, the clear striatal prominence seen in previous reports was not evident. This finding serves to underline the value of studying a number of different mutations—it permits the observations that striatal prominence is not universal and that amyloid imaging profiles, even amongst the mutations of a single gene, may be varied. The two PSEN1 mutations studied by Klunk and colleagues (2007) were in exons 11 and 12, unlike our mutations, none of which were situated beyond exon 8. This may have contributed to the different profiles observed, just as clinical phenotypic heterogeneity is known to be associated with varied mutation position (Larner and Doran, 2006).

In APP duplication cases, abundant intravascular amyloid, particularly in the leptomeningeal vessels, may contribute to increased 11C-PiB signal (Remes et al., 2008) and it is possible that retention in our subjects was similarly influenced. Without neuropathological examination, the amount of cerebrovascular amyloid in our subjects, along with any possible influence on 11C-PiB signal, cannot be precisely known. However, we can say that our study did not include any APP duplications and that, for PSEN1 mutations, significant cerebral amyloid angiopathy is more frequently associated with those beyond codon 200 (Mann et al., 2001), for example, L282V (Dermaut et al., 2001), E280G (O'Riordan et al., 2002) and G217D (Takao et al., 2002). As five out of six coding region mutations in our study affected regions between codons 100 and 200, we suggest that cerebrovascular amyloid alone is unlikely to account for our findings.

We observed marked 11C-PiB retention in the thalamus, a pattern peculiar to the PSEN1 group. The thalamus is not a structure traditionally associated with a heavy amyloid burden, although thalamic β-amyloid deposition has previously been described in a post-mortem study of PSEN1 mutation carriers (Lippa et al., 1996). Specific mutations, such as M146I, have since been associated with unusually high levels of β-amyloid pathology in the thalamus and mesencephalon (Gustafson et al., 1998) raising the possibility that such a pattern is codon- or mutation specific. Post-mortem histopathological analyses have previously demonstrated immunopositive, diffuse β-amyloid plaques in the thalamus of a PSEN1 M139V carrier (Larner and du Plessis, 2003) and small increases in thalamic 11C-PiB have been noted (Kemppainen et al., 2006).

The cerebellum has long been used as a reference region for non-specific tracer binding in 11C-PiB PET imaging as, in the sporadic Alzheimer’s disease brain, it does not classically accumulate amyloid or retain 11C-PiB (Svedberg et al., 2009). In accord with this we found that the subjects with sporadic Alzheimer’s disease did not have raised 11C-PiB retention—with almost identical cerebellum : pons ratios when compared with the controls (Fig. 2). However 11C-PiB retention does occur in the cerebellum in familial Alzheimer’s disease although there is considerable variability between subjects (Fig. 1); this suggests, in line with previous studies (Klunk et al., 2007), that the cerebellum may not be an appropriate reference region in familial Alzheimer’s disease and that the pons (as used here) is a suitable alternative. In order to allow comparison with other studies in Alzheimer’s disease, we subsequently calculated ratios with the cerebellum as the reference region. The respective pontine and cerebellar ratios highlighted abnormal 11C-PiB uptake in varying, though overlapping regions. Full interpretation of these differences would require pathological correlation beyond the current scope of this study. Our results also suggest that the distribution of tissue pathology (at least in terms of plaque amyloid) associated with PSEN1 mutations may be more varied than previously thought. Finally these findings raise intriguing questions about why these different patterns of amyloid binding occur and what their clinical and pathological implications may be.

Funding

This work was partly undertaken at University College London Hospitals/University College London who received a proportion of funding from the Department of Health’s National Institute for Health Research (NIHR) Biomedical Research Centres funding scheme. The Dementia Research Centre is an Alzheimer’s Research Trust Coordination Centre. The Medical Research Council UK also funded this work.

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

N.C.F. is an MRC Senior Clinical Fellow, N.C.F. and M.N.R. are NIHR Senior Investigators. Prof. D.J.B. is Senior Neurologist at Medical Diagnostics, GE Healthcare PLC.

Abbreviations
11C-PiB PET
11Carbon-Pittsburgh compound B positron emission tomography
PSEN1
presenilin 1

References

View Abstract