This study describes and quantifies postmortem amyloid pathology in a subject, clinically diagnosed with DLB and possible AD, who had a negative [C-11]PiB PET scan 17 months before death, and compares the results to region-matched analyses in a subject with clinically severe AD and a positive [C-11]PiB PET scan 10 months before death.
Cortical Aβ plaques were detected in both cases, however plaques in the [C-11]PiB(−) case were infrequent, primarily diffuse in morphology, and were labeled weakly with thioflavin S, X-34 and 6-CN-PiB. Although some cortical regions in the [C-11]PiB(−) case had focal areas of moderate and even high frequencies of Bielschowsky positive neuritic plaques, these lesions are not representative of the entire region’s pathology because semi-quantitative assessments of plaque frequencies by a standard neuropathological assessment were performed in a 100× microscopy field placed over the areas of the highest plaque density.
Biochemical analyses revealed that [H-3]PiB binding levels and Aβ1–42 concentrations in the [C-11]PiB(−) case were consistently lower than values determined in matched regions from the [C-11]PiB(+) case. Aβ1–40 concentrations were on average lower than the Aβ1–42 levels by a factor of approximately five in the [C-11]PiB(−) case and by a factor of approximately 7 in the [C-11]PiB(+) case. The two cases had comparable Aβ1–40 levels in several cortical regions including middle temporal, inferior temporal and occipital cortex. In other brain regions, Aβ1–40 concentration was lower in the [C-11]PiB(−) case. [H-3]PiB binding did not correlate with Aβ1–42 or Aβ1–40 levels in the [C-11]PiB(−) case; however, these correlations were significant in the [C-11]PiB(+) case. Similarly, in vivo [C-11]PiB DVR measures correlated with both Aβ1–42 and Aβ1–40 in the [C-11]PiB(+) case, and with Aβ1–42, but not Aβ1–40, in the [C-11]PiB(−) case. Histologically, 6-CN-PiB plaque load correlated with Aβ plaque load in the [C-11]PiB(+) case, but not in the [C-11]PiB(−) case. Collectively, these observations indicate that lack of significant [C-11]PiB retention in the [C-11]PiB(−) case is likely due to a combination of low Aβ1–42 levels and low levels of β-sheet (i.e., fibrillar) content of the Aβ deposits in this case (β-sheet structure is required for [H-3]PiB binding and 6-CN-PiB labeling in vitro and [C-11]PiB retention in vivo).
Diffuse Aβ deposits may represent an early state of Aβ aggregation, or contain truncated Aβ forms that assemble into fibrils less readily. For example, plaques in the cerebellum are composed primarily of the 17–42aa p3 fragment of Aβ [25
] which does not bind [H-3]PiB or [C-11]PiB [16
]. To determine if Aβ plaques in the [C-11]PiB(−) case contained proteolytically modified Aβ, we performed immunohistochemistry using an array of antibodies targeting the N-terminus (6E10), mid-portion (4G8), and C-terminus (Aβx-40 and Aβx-42; see ) of the Aβ peptide to gain insight into the type of Aβ present in the plaques from the [C-11]PiB (−) case. Comparably low densities of Aβ plaques were detected in the [C-11]PiB(−) case with all antibodies used in this study (), suggesting that [C-11]PiB PET negativity was likely not the result of an unusual profile of proteolytically modified Aβ species. We obtained similar results using an antibody against the N3pE modification of Aβ (N-terminal truncation and cyclization of glutamate-3), a significant finding in view of the observations of Maeda et al. [30
], who suggested that in Aβ plaque-depositing transgenic APP mice, this particular modification of Aβ was critical for [C-11]PiB binding.
Several previous in vivo-postmortem PiB correlation studies are relevant to the interpretation of our findings (), and two studies of PiB binding and neuropathology in postmortem tissue without correlative in vivo [C-11]PiB PET data yield additional insights [27
]. Bacskai et al. [2
] reported a correlation between high postmortem levels of Aβx-42 (3,700–4,700 pmol/g) and [H-3]PiB binding (325–560 pmol/g) in a [C-11]PiB(+) DLB case. Histopathological analysis of this case showed that PiB fluorescence was intense in CAA-containing blood vessels and rare cored plaques but was weak in diffuse plaques. Bacskai and colleagues concluded that extensive CAA was the main contributor to the positive in vivo [C-11]PiB PET signal in that DLB case. We reported [C-11]PiB PET to postmortem correlations in a typical Braak stage VI AD case that had relatively little CAA (the [C-11]PiB(+) case referred to in this study) and compared it to postmortem samples from other typical AD brains [16
]. We demonstrated a strong correlation of the in vivo [C-11]PiB PET signal with measures of Aβ pathology postmortem, and demonstrated that Aβ42 concentration drove the correlation with in vivo [C-11]PiB data [16
]. These results were corroborated in the autopsy analysis of the first subject scanned with [C-11]PiB PET [17
Cairns and colleagues [6
] described the neuropathology of a subject with normal cognition and only background [C-11]PiB retention when PET scanned, but who progressed to very mild clinical AD (CDR = 0.5) a year later. Although Aβ deposits were detected histologically at autopsy (30 months after the in vivo [C-11]PiB PET scan), the low frequencies of neuritic plaques were consistent with a diagnosis of only possible AD by CERAD criteria, and there was only a low probability that clinical symptoms were caused by AD pathology according to the NIA/Reagan Institute criteria [6
]. In contrast, biochemical analyses in the Cairns [C-11]PiB(−) case showed high levels of postmortem Aβ1–42 (up to 1,785 pmol/g) determined by the same ELISA method reported here, well above the highest value we observed in our [C-11]PiB(−) case. Furthermore, the Aβ plaque load (up to 5.4% area) in the Cairns [C-11]PiB(−) case also was higher than in the [C-11]PiB(−) case described in our study. Postmortem [H-3]PiB binding in several brain areas from the Cairns [C-11]PiB(−) case, again performed by the same methods reported here, was much higher than in the [C-11]PiB(−) case reported here and could conceivably have produced a weakly-positive [C-11]PiB PET signal in vivo had the PET-to-death interval been shorter. In fact even 30 months prior to death, the in vivo [C-11]PiB PET signal reported by Cairns et al. approached the global cutoff for [C-11]PiB-positivity established by the Washington University group [6
Postmortem pathological correlates of in vivo [C-11]PiB retention were also examined in studies of two [C-11]PiB(−) cases of Creutzfeldt-Jakob disease [49
], three cases of Parkinson’s dementia (2 [C-11]PiB(+), 1 [C-11]PiB(−) [5
], a [C-11]PiB(+) case of DLB [19
], and in four [C-11]PiB(−) nondemented cases and in one nondemented and one demented [C-11]PiB(+) case from the Baltimore Longitudinal Study of Aging [46
]. In addition to these autopsy studies summarized in , Leinonen and colleagues [26
] reported a correlation between neuropathology and in vivo [C-11]PiB retention in biopsies from ten cases of suspected normal pressure hydrocephalus (NPH). Correlations were observed in 9/10 cases, but diffuse cortical Aβ plaques were found in the frontal biopsy sample from one NPH case with dementia who was subsequently determined to be [C-11]PiB(−) [26
]. The small amount of tissue available prevented more detailed histological or biochemical assays in that biopsy study.
Two postmortem PiB-labeling studies without associated in vivo [C-11]PiB PET data also contribute to the characterization of PiB PET as a technique to detect Aβ plaque pathology in vivo. Using autoradiography, Lockhart and colleagues [27
] reported that [H-3]PiB labels classic and diffuse plaques, CAA, and NFT. PiB labeling of NFT is not supported by other studies and may be due to the low resolution of the autoradiographic technique employed or to the binding of PiB to aggregated Aβ associated with extracellular tangles [16
]. Rosen et al. [42
] studied a series of 10 AD autopsies and largely confirmed previous findings in nine of these cases. One end stage (Braak stage VI) AD case was unusual, with “copious dense-cored and diffuse” plaques and “significant large vessel and capillary CAA” detected using Aβ immunohistochemistry, Congo red, thioflavin T and silver staining [42
]. The most unusual feature of this case was extremely high postmortem levels of both Aβx-42 (>9,000 pmol/g) and Aβx-40 (>26,000 pmol/g), with a predominance of Aβx-40, yet extremely low [H-3]PiB binding (<50 pmol/g). In addition, there was an “unusual distribution of low- and high-molecular weight Aβ oligomers, as well as a distinct pattern of N- and C-terminally truncated Aβ peptides in both the soluble and insoluble cortical extracts” [42
]. Although no in vivo [C-11]PiB PET data was available, the Rosen case suggests that insoluble Aβ40 may be present in high quantities without significant [H-3]PiB binding. This idea is supported by the results of our current analysis which demonstrate that several cortical regions in the [C-11]PiB(−) case had levels of Aβ1–40 which were comparable to those in the [C-11]PiB(+) case. In the caudate of the current [C-11]PiB(+) case, high Aβ1–42 levels and very low Aβ1–40 levels (45-fold difference) coupled with high [C-11]PiB retention and high [H-3]PiB binding suggest that Aβ1–42 is a good substrate for [C-11]PiB binding in human brain. Since Aβ1–42 is usually the prevalent Aβ species in human brain, its mass alone will likely dictate that this longer Aβ species will be dominant in determining [C-11]PiB retention in vivo.
The clinical relevance of Aβ pathology in [C-11]PiB(−) cases is unclear. In the Cairns et al. case [6
], the [C-11]PiB(−) subject was cognitively normal (CDR = 0) when the [C-11]PiB PET scan was obtained, suggesting that Aβ pathology, if present, was without clinical effect. In the present [C-11]PiB(−) case, the existence of another pathology (Lewy bodies and threads) that can alone contribute to dementia prevents an unambiguous determination of whether the [C-11]PiB(−) Aβ deposits contributed to cognitive dysfunction of this patient. Similarly, the Leinonen [C-11]PiB(−) biopsy case had clinically diagnosed NPH [26
]. However, in the Rosen case with high insoluble Aβ1–40 and low [H-3]PiB binding, only typical AD pathology was noted [42
]. Thus, although the in vivo [C-11]PiB PET status of the Rosen case is unknown, this may have been a [C-11]PiB(−) case in which large amounts of an atypical Aβ species contributed to the dementia.
In summary, we report histological and biochemical evidence of Aβ pathology in a cognitively impaired individual with clinical diagnosis of DLB and no detectable [C-11]PiB retention 17 months before death. These observations, along with the postmortem findings of Cairns et al. [6
] and biopsy findings of Leinonen et al. [26
] suggest that in vivo [C-11]PiB PET may not be 100% sensitive for the presence of histologically detectable Aβ even if the latter were to be determined at the time of the in vivo scan. As a corollary, even weakly-positive [C-11]PiB PET scans will likely be associated with substantial Aβ pathology (i.e., high specificity of in vivo [C-11]PiB PET for Aβ deposits). However, no accurate determinations of pathology level thresholds necessary to elicit a positive PiB PET signal can be made using single or small numbers of cases. Furthermore, long imaging-to-death intervals make it unclear what level of pathology developed after the [C-11]PiB scan and prior to death. The clinical relevance of amyloid deposits if present at time of imaging and not detectable by [C-11]PiB PET is poorly understood and requires further investigation in large numbers of [C-11]PiB PET autopsy cases with different clinical diagnoses and shorter imaging-to-death intervals.