PET imaging for the evaluation of cerebral amyloid angiopathy: a systematic review

In the last years, the role of PET imaging in the assessment of cerebral amyloid angiopathy (CAA) is emerging. In this setting, some tracers have proven their utility for the evaluation of the disease (mainly 11C-Pittsburgh compound B [11C-PIB]), however, the value of other radiotracers has to be clarified. The aim of this systematic review is, therefore, to assess the role of PET imaging in the evaluation of CAA. A wide literature search of the PubMed/MEDLINE, Scopus, Embase, Web of Science and Cochrane library databases was made to find relevant published articles about the diagnostic performance of PET imaging for the evaluation of CAA. Quality assessment including the risk of bias and applicability concerns was carried out using QUADAS-2 evaluation. The comprehensive computer literature search revealed 651 articles. On reviewing the titles and abstracts, 622 articles were excluded because the reported data were not within the field of interest. Twenty-nine studies were included in the review. In general, PET imaging with amyloid tracers revealed its value for the assessment of CAA, for its differential diagnosis and a correlation with some clinico-pathological features. With less evidence, a role for 18F-fluorodeoxiglucose (18F-FDG) and tau tracers is starting to emerge. PET imaging demonstrated its utility for the assessment of CAA. In particular, amiloid tracers revealed higher retention in CAA patients, correlation with cerebral bleed, the ability to differentiate between CAA and other related conditions (such as Alzheimer's disease) and a correlation with some cerebrospinal fluid biomarkers.


Introduction
Cerebral amyloid angiopathy (CAA) is a neurological disorder caused by the deposition of β-amyloid in the walls of small and medium vessels of the cerebral cortex and leptomeninges [1,2]. Its development is associated with multiple risk factors such as aging, the presence of Alzheimer's disease (AD) and genetic mutations (apolipoprotein E [APOE] and amyloid-β protein precursor genes). In this setting, the prevalence of the disease is particularly evident in elderly patients [1,3,4].
The clinical manifestations of CAA can be really heterogeneous and can include a wide range of symptoms such as spontaneous lobar intracerebral hemorrhage (ICH), focal transient neurological episodes and cognitive impairment like dementia [1,[5][6][7]. In this scenario, ICH has central importance given its high recurrence rate estimated at more than 10% per year [1,2].
Assessment of CAA is challenging and definitive diagnosis can be made only by postmortem histopatological confirmation by autopsy [1,8]. In life, the disease is often recognized because of the presence of symptomatic and spontaneous ICH, preferentially affecting cortical-subcortical regions of occipital and posterior temporal lobes [2,9]. In clinical practice, magnetic resonance (MR) imaging is mandatory for the correct evaluation of the patients, leading to the development of Boston criteria, that demonstrated high accuracy for the diagnosis of CAA [1,8,10,11]. Notably, the presence of multiple strictly lobar microbleeds (MB) is one of the hallmark biomarkers for the disease within these criteria [11][12][13][14].
Recently, the role of positron emission tomography (PET) imaging in the correct assessment of CAA is emerging. In this setting, various tracer has demonstrated their ability to evaluate the presence of amyloid deposition, such as 11 C-Pittsburgh compound B ( 11 C-PIB) and 18 F-florbetapir [1,5,15]. However, in the last years the role of other radiotracers in the assessment of the disease is emerging. In particular, as mentioned, some tracers have the ability to bind to amiloyd plaques in the brain such as 11 C-PIB, 18 F-florbetapir, 18 F-florbetaben and 18 F-flutemetamol, while 18 F-T807 and 18 F-flortaucipir bind to tau protein aggregates. Furthermore, 18 F-FDG has the ability to evaluate the glucose metabolism of brain cells.
The purpose of this systematic review is to assess the role and the state of the art of PET imaging for the evaluation of CAA.
No beginning date limit was applied to the search and it was updated until 31 January 2022. Only articles in the English language were considered. Furthermore, pre-clinical studies, postmortem studies, conference proceedings, reviews, case reports, case series and editorials were excluded from the review. To expand our search, the references of the retrieved articles were also screened for additional papers.

Study selection
Two researchers (FD and DA) independently reviewed the titles and the abstracts of the retrieved articles. The full-text version of the remaining articles was then independently reviewed by the same authors, to determine their eligibility for inclusion. In addition to previously presented exclusion criteria, the presence of less than 8 patients affected by CAA was another criteria used to screen the articles, to avoid articles with small samples of patients. The quality assessment, including the risk of bias and applicability concerns was carried out using QUADAS-2 evaluation [16].

Data abstraction
For each included study, data concerning the basic study were collected (first author, year of publication, country of origin, type of study) and PET device used, number of patients evaluated, and number of patients affected by CAA. The main findings of the articles included in this review are reported in the Results.

Literature search
A total of 651 articles were extrapolated with the computer literature search and by reviewing the titles and abstracts, 622 of them were excluded because the reported data were not within the field of interest of this review. Twenty-nine articles were selected and retrieved in full-text version ; no additional studies were found screening the references of these articles (Fig. 1). Generally speaking, the quality assessment of these articles using QUADAS-2 underlined a low risk of bias (Fig. 2a) and low risks for applicability concerns (Fig. 2b).

Role of PET imaging for the assessment of CAA
Many studies have proven the ability of 11 C-PIB to concentrate in patients affected by CAA, demonstrating the usefulness of PET imaging for the assessment of the disease [17-22, 28, 31, 32, 37, 41, 42, 45]. Interestingly a single work revealed low specificity [21]. In this setting, a correlation between tracer uptake and the site of cerebral MB (CMB) was reported by some works [17,19,28,40]. Furthermore, the role of PET imaging with 11 C-PIB for the differential diagnosis between CAA and other conditions related to cerebral hemorrhage has been proven in some studies [18,28,32,37,42].
The ability of 11 C-PIB PET imaging to differentiate between CAA and AD was investigated by some works [17,18,20,22,41] In general higher tracer retention was associated with the presence of CAA and occipital regions were characterized by higher uptake compared to AD.
PET imaging with 18 F-florbetapir also revealed the ability to assess CAA [25,30,33,35,36]. In this setting, the correlation between tracer uptake and lobar ICH [25,30] and with some cerebrospinal fluid (CSF) biomarkers were reported [33]. A single work revealed also a possible trend for 18 F-florbetapir imaging to differentiate between AD and CAA-ICH [35]. Similarly, a new pharmacokinetic model has demonstrated the ability to differentiate probable CAA and deep ICH [38].
Three works investigated the role of 18 F-FDG imaging in CAA [23,39,40], revealing its capability to differentiate between CAA-related and CAA-unrelated CMB [23] and between CAA and AD [40]. Furthermore, the value of 18 F-FBB PET imaging was investigated by two works, demonstrating its ability to differentiate CAArelated inflammation (CAA-ri) from CAA and, with less evidences, from normal controls [26,27].
Some works used mixed radiotracers for the assessment of CAA [24,29,34,43,44] and in this setting a correlation between 18 F-florbetapir and 11 C-PIB uptakes was demonstrated [24]. Furthermore, a similar proportion of positive scan between 18 F-FBB and 11 C-PIB was reported, with a correlation between some clinicopathological features and 18 F-FBB positivity [29]. Both amyloid and tau deposition in CAA were evaluated by two different studies with the use of 11 C-PIB and 18 F-T807 [43] and 11 C-PIB and 18 F-flortaucipir [44], reporting that PET tau positivity was correlated with some clinicopathological features. Lastly, a single work used three tracer ( 11 C-PIB, 18 F-FBB and 18 F-flutemetamol) reporting a correlation between PET positivity and the pattern of MB presentation [34].

PET imaging with 11 C-PIB
As mentioned, several studies have investigated the role of PET imaging with 11 C-PIB for the assessment of CAA [17-22, 28, 31, 32, 37, 41, 42, 45] demonstrating in general the capability of this tracer to be retained in patients affected by the disease. Correlation with CMB and differential diagnosis of ICH First, Dierksen et al. [17] demonstrated higher 11 C-PIB retention in CAA compared to control subjects and a strong correlation between amyloid deposition and CMB, in particular for patients with high-CMB counts. Similarly, Gurol et al. [19] confirmed this correlation with new bleeds, reporting that increased tracer retention characterized sites of future bleeds and a higher risk of incidental bleeds. Again, Chang et al. [41] reported higher SUV values in CMB area of patients with CAA compared with those of AD or healthy subjects. In this setting Ly et al. [18] reported increased 11 C-PIB uptake in patients with CAA-related hemorrhage (CAAH) and higher binding of tracer in patients with probable CAAH compared to patients with possible CAAH. Four different studies by the group of Tsai et al. [28,32,37,42] reported higher 11 C-PIB retention in CAA-ICH patients compared to non-CAA-ICH patients (hypertensive and mixed ICH) and a correlation between lobar lacune counts and SUV values. Furthermore, higher tracer uptake in patients with high-degree enlarged centrum semiovale perivascular spaces (ECSPVS) compared to low-degree patients was reported.
Interestingly a study by Baron et al. [21] did not report differences in terms 11 C-PIB uptake between CAA patients and healthy controls. In this work, 11 C-PIB PET imaging revealed low specificity for CAA diagnosis, due to the frequent presence of high tracer uptake in the healthy elderly, reflecting incipient AD. However, a negative 11 C-PIB was able to rule out CAA with excellent sensitivity.

Differential diagnosis between CAA and AD
First Farid et al. [22] in a study with early and late phase imaging reported different 11 C-PIB uptake for occipital and posterior cingulate cortex between AD and CAA, with lower whole cortex to occipital ratio and occipital/posterior cingulate ratio in CAA patients. Similarly, Dierksen et al. [17] reported an elevated occipital-to-global ratio for CAA patients compared to AD patients. In this scenario Chang et al. [41] reported lower global cortical 11 C-PIB uptake for CAA patients compared to AD, however, tracer uptake in occipital regions was higher in CAA compared to AD patients. In contrast, AD subjects had higher lateral temporal lobe deposition of tracer. Similarly, Ly et al. [18] reported that 11 C-PIB uptake in occipital-global neocortical and frontal-global neocortical regions was somewhat different between CAAH and AD patients. Interestingly Gurol et al. [20] reported similar high 11 C-PIB retention between CAA and AD patients, but a correlation between tracer uptake and white matter hyperintensities (WMH) in CAA patients.

Miscellany
Gokcal et al. [45] reported that 11 C-PIB uptake was correlated with vascular reactivity, consistent with their hypothesis of the mediating role of this reactivity between tracer uptake and WMH.
Lastly, a singular work on patients with hereditary autosomal dominant forms of CAA was proposed by Schultz et al. [31], reporting high 11 C-PIB uptake in these patients and an inverse correlation between uptake and Aβ40 levels in CSF.

Correlation between PET imaging and ICH
First, the group by Raposo et al. [25,26] reported greater cortical 18 F-florbetapir retention for CAA patients than deep ICH patients and a higher ratio of positive PET scan for lobar ICH than deep ICH. Furthermore, among CAA patients the highest uptake was present in the occipital lobe and a value of 1.18 for global SUVR (standardized uptake value ratio with cerebellum as reference) was obtained. Interestingly, in patients with lobar ICH the ratio of positive scan decreased in concordance with the probability of CAA diagnosis and PET positivity was independently correlated with ECSPVS.
More recently Planton et al. [36] investigated whether amyloid burden was increased in the ICH-affected emisphere in patients with asymptomatic CAA-ICH, reporting no differences between the two emispheres.

Miscellany
In their analysis of CSF biomarkers in CAA, Banerjee et al. [33] reported that half of CAA patients had a positive 18 F-florbetapir PET scan. Furthermore, these patients had lower Aβ42, higher t-tau, higher p-tau, NFL and neurogranin. compared to patients with negative PET scans were demonstrated.
Planton et al. [35] reported higher global retention of 18 F-florbetapir in mild cognitive impairment (MCI)-AD patients compared to CAA subjects, however, no differences were reported for relative regional uptake. Nevertheless, a trend for increased occipital/global ratio in CAA-ICH patients was reported.

PET imaging with 18 F-FDG
First, Samuraki et al. [23] investigated the role of 18 F-FDG for the assessment of CAA in patients with probable AD, reporting that patients with CAA-related CMB had hypometabolism mainly in left temporal lobe and in bilateral insular gyri. Conversely, patients with CAA-unrelated CMBs had reduced tracer uptake mainly in the right putamen and right cerebellum. Furthermore, a positive correlation between Mini-Mental State Examination (MMSE) and verbal memory score with 18 F-FDG uptake were recognized.
More recently two different works by the group Bergeret et al. [39,40] investigated the possible role of 18 F-FDG PET imaging to differentiate CAA from AD. Lower occipital/ posterior cingulate SUVR ratio in CAA patients compared to AD subjects was demonstrated. Furthermore, with the exception of the anterior cingulate and medial prefrontal cortex, patients with CAA had global cortical significant glucose hypometabolism, however, significant only in the posterior areas,

PET imaging with 18 F-FBB and 18 F-flutemetamol
The role of 18 F-FBB in CAA was evaluated in two studies by the group Renard et al. [26,27]. They reported general higher tracer uptake for CAA-ri patients compared to normal control, in particular in the occipital lobe with a posterior to anterior trend, however, without significant difference between the lobes. Furthermore, higher SUVR of the pons in CAA-ri patients compared to CAA subjects was reported. When considering pons as the reference standard, a correlation between Aβ40 and SUVR was underlined.
More recently, Papanastasiou et al. [38] demonstrated the role of a new pharmacokinetic model with PET/MR able to separately detect impaired haemodynamic and amiloyd load in patients with probable CAA compared to patients with deep ICH. These findings were also reproducible with a reduced time acquisition and underlined an overlapping in perfusion deficits and amyloid burden in patients with CAA-related ICH.

PET imaging with mixed radiotracers
First, the role of 18 F-florbetapir and 11 C-PIB in the assessment of CAA was evaluated by Gurol et al. [24], demonstrating a strong correlation between the uptake of the two tracers. Furthermore, mean global cortical 18 F-florbetapir uptake and mean occipital SUVR were higher for CAA patients compared to patients with deep hypertensive ICH.
Jang et al. [29] used both 18 F-FBB and 11 C-PIB to evaluate patients with probable CAA with a positive scan percentage of 66.7% for 11 C-PIB and 66% for 18 F-FBB. Interestingly, patients with positive amyloid PET (Aβ +) 1 3 had a higher frequency of APOE ɛ4 carriers, more CMB, a higher frequency of cortical superficial siderosis and worse performances in cognitive tests. Furthermore, occipital/ global SUVR was higher for Aβ + CAA patients in comparison to Aβ + AD patients.
An evaluation of patients with cerebral small vessels disease (SVD), including also CAA subjects, was made by Tsai et al. [43] with 11 C-PIB and 18 F-T807 to assess both amyloid and tau cortical deposition. They reported that 33.3% of CAA patients had a positive 18 F-T807 scan and patients with a positive tau scan (both AD and CAA) had higher triggering receptor expressed on myeloid cell 2 (TREM2) plasma levels; furthermore, these levels correlated with MMSE score. Both amyloid and tau evaluation was also performed by Schoemaker et al. [44]. General high global retention of 11 C-PIB and increased 18 F-flortaucipir retention for amnestic CAA patients compared to non-amnestic forms were reported. Furthermore, patients with positive 18 F-flortaucipir PET scans had lower performances in the memory domain.
Lastly, Jung et al. [34] used three tracers ( 11 C-PIB, 18 F-FBB and 18 F-flutemetamol) to evaluate patients with CAA, revealing that the frequency of PET positivity was correlated with the pattern of MB presentation.
In conclusion, generally speaking PET imaging demonstrated its utility for the assessment of CAA. In particular, β-amiloid tracers revealed higher retention in CAA patients, correlation with CMB, the ability to differentiate between CAA and other related conditions (such as AD) and a correlation with some CSF biomarkers. The role of 18 F-FDG imaging for the differential diagnosis of CAA and its correlation with cognitive performances are arising, however, with initial evidence. Similarly, the use of tau PET imaging in CAA is starting to emerge.