Carotid Plaque PET Imaging and Cerebral Ischemic Disease: A Systematic Review and Meta-analysis

Abstract

Background and Purpose:

The clinical utility of PET imaging in evaluating carotid artery plaque vulnerability remains unclear. Two tracers of recent interest for carotid plaque imaging are ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) and ¹⁸F-sodium fluoride (¹⁸F-NaF). We performed a systematic review and meta-analysis evaluating the association between carotid artery ¹⁸F-FDG or ¹⁸F-NaF uptake and recent or future cerebral ischemic events.

Methods:

A systematic review of Ovid MEDLINE, Ovid EMBASE and the Cochrane library was conducted from inception to December 2017 for articles evaluating PET tracer uptake in recently symptomatic versus asymptomatic carotid arteries, and articles evaluating carotid uptake in relation to future ischemic events. Cerebral ischemic events were defined as ipsilateral strokes, transient ischemic attacks or amaurosis fugax. We quantitatively pooled studies by a random-effects model when 3 or more studies were amenable for analysis. We assessed the standardized mean difference between tracer uptake in the symptomatic versus asymptomatic carotid artery using Cohen’s d metric.

Results:

After screening 4,144 unique articles, 13 prospective cohort studies assessing carotid artery ¹⁸F-FDG uptake in patients with recent cerebral ischemia were eligible for review. Eleven cohorts of 290 subjects scanned with ¹⁸F-FDG were eligible for meta-analysis. We found that carotid arteries ipsilateral to recent ischemic events had significantly higher ¹⁸F-FDG uptake than asymptomatic arteries (Cohen’s d=0.492, CI=0.130–0.855, P=0.008) as well as significant heterogeneity (Cochran’s Q = 31.5, P = 0.0005; I² = 68.3%). Meta-regression was not performed due to the limited number of studies in the analysis. Only 2 studies investigating ¹⁸F-NaF PET imaging and another 2 articles investigating ischemic event recurrence were found.

Conclusions:

Recent ipsilateral cerebral ischemia may be associated with increased carotid ¹⁸F-FDG uptake on PET imaging regardless of degree of carotid stenosis, although significant heterogeneity was found and these results should be interpreted with caution. Emerging evidence suggests a similar association may be present with ¹⁸F-NaF plaque uptake. More studies are warranted to provide definitive conclusions on the utility of ¹⁸F-FDG or ¹⁸F-NaF in carotid plaque evaluation before investigating carotid PET as a diagnostic tool for cerebral ischemic events.

Introduction

Carotid artery atherosclerosis is responsible for approximately 10% of first-time ischemic strokes^1,2. In the evaluation of symptomatic carotid plaques, information provided by non-invasive imaging techniques can be used for more comprehensive assessment of plaque vulnerability beyond simple luminal stenosis measurements³. By providing insight into tissue metabolism, studies have shown that atherosclerotic plaque uptake of ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) on positron emission tomography (PET) is strongly correlated with well-established markers of inflammation such as hypermetabolic CD68 macrophages, cathepsin K, MMP-9 and IL-18 on histology as well as plaque angiogenesis and hypoxia⁴. The potential value of such imaging tracers in pinpointing recent culprit plaques is suggested by data demonstrating that symptomatic plaques show a gradual decline in macrophage density on histology for up to 180 days after the onset of symptoms^5,6. By contrast, ¹⁸F-sodium fluoride (¹⁸F-NaF) highlights vessel microcalcifications implicated in actively inflamed atherosclerotic lesions⁷ by binding to hydroxyapatite in vessel walls. Moreover, ¹⁸F-NaF uptake has been shown to correlate with generalized atherosclerotic risk factors such as age, hypertension, diabetes, male sex, smoking, hypercholesterolemia and prior cardiovascular events^8,9. Thus ¹⁸F-FDG and ¹⁸F-NaF PET studies have potential to identify active plaques and provide an assessment of plaque biology and, possibly, vulnerability.

Several recent studies have used PET imaging to identify culprit plaques following stroke or transient ischemic attacks and to predict future cerebral ischemic events. However, these studies have generally been small and sometimes shown conflicting results making it difficult to draw definite conclusions on the utility of PET imaging in symptomatic carotid artery evaluation. Given the preliminary level of evidence available in the literature, we did not intend to assess carotid PET as a diagnostic tool for cerebral ischemic events. Rather, we performed a systematic review and meta-analysis of the medical literature for studies that examined carotid plaque PET tracer uptake in relationship to recent prior or future cerebral ischemic events to estimate the effect size of carotid SUV uptake across symptomatic and asymptomatic arteries. We hypothesized that ¹⁸F-FDG and ¹⁸F-NaF uptake in carotid arteries ipsilateral to a recent ischemic event is higher than uptake in the contralateral artery.

Methods

The authors declare that all supporting data are available within the article and its online supplementary files. This study was performed in accordance with the guidelines set forth by the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) group¹⁰ and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement¹¹.

Data Sources and Searches

An experienced medical librarian performed comprehensive literature searches in the electronic databases Ovid MEDLINE, Ovid EMBASE and The Cochrane Library (Wiley) to identify relevant published studies from database inception through December 5^th 2017. There were no language, publication date, or article type restrictions on the search strategy. The full search strategy for Ovid MEDLINE is available in supplement. Subject headings and keywords were then adapted for the other databases. Additional records were identified by employing the “Cited by” and “View references” features in Scopus. The full PRISMA flow diagram outlining the study selection process is available in supplement (Supplemental Figure I).

Study Selection

We included studies evaluating the association of ipsilateral carotid plaque ¹⁸F-FDG or ¹⁸F-NaF uptake on PET examinations in patients with recent and/or future ischemic events. Specific inclusion criteria were (a) studies of adult subjects (aged >18 years); (b) studies of patients with acute cerebral infarction, transient ischemic stroke (TIA) or retinal embolism/amaurosis fugax; (c) studies of patients who underwent PET scanning within 180 days of an index ischemic cerebral event; (d) studies measuring PET tracer uptake in atherosclerotic lesions within arterial stenosis or plaques in the distal common carotid or proximal internal carotid artery (tracer uptake within plaques or artery walls will be used interchangeably); (e) studies correlating carotid plaque uptake with at least one of the following outcome measures: recent ischemic events, recurrence of ischemic events in patients who were recently symptomatic, future ischemic events in asymptomatic patients; and (f) studies including ≥5 subjects to avoid the inclusion of case reports or very small case series. Patients with infarctions that were likely due to cardiogenic embolic sources or secondary non-atherosclerotic etiologies such as vasculitis, infective endocarditis or reversible cerebral vasoconstriction syndrome were excluded from this study.

We chose to include only peer-reviewed journal articles rather than conference abstracts in order to include studies which provide sufficient detail for systematic review and meta-analysis. Only original studies were considered for inclusion. If data from a single patient cohort was published more than once, the single paper with the largest sample size was included to minimize analysis of duplicate study samples. If necessary, we contacted corresponding authors for clarification during the data extraction process.

Data Extraction

Two investigators independently screened all references produced by the literature search against predefined inclusion/exclusion criteria, then examined the shortlisted articles in full to determine eligibility. Discrepancies were resolved by consensus or 3^rd reviewer. After eligibility was determined, an investigator then extracted data such as sample size, inclusion and exclusion criteria, patient demographics and medical treatment, degree of stenosis studied, symptom-to-scan time, and all relevant reported PET tracer uptake metrics.

PET tracer uptake metrics

PET tracer uptake has been reported in a variety of metrics (Supplemental Table I). Standardized uptake values (SUV) are commonly used due to standardization to total body injected activity¹². By demarcating regions or volumes of interest (ROI/VOI) around a carotid artery/plaque (Figure 1), maximum SUV (SUV_max) or mean SUV (SUV_mean) across all voxels of a section can be calculated. We also collected data on the target/tissue-to-background ratio (TBR) when made available by authors, which is calculated by dividing the SUV of a tissue of interest by that of a venous blood pool.

FIGURE. 1. ¹⁸F-FDG PET/MR Image.

showing increased uptake of FDG in a right carotid artery plaque causing less than 50% stenosis (white arrow).

Risk of Bias Assessment

Since PRISMA guidelines on risk of bias assessment are more suited for assessing randomized controlled than cohort studies, we opted to assess the screened articles for risk of bias by adapting a previously published questionnaire¹³ for bias assessment in the imaging of carotid disease (Supplemental Table II).

Statistical Analysis

Studies that quantitatively reported tracer uptake for symptomatic and asymptomatic carotid arteries were used for meta-analysis. Reported means and standard deviations for symptomatic and asymptomatic carotid arteries in each study were used to calculate Cohen’s effect size (d), which is the standardized mean difference (i.e., mean difference between symptomatic and asymptomatic carotid arteries divided by the pooled standard deviation). It should be noted, however, that a true side-by-side “paired” comparison of SUV values between symptomatic and asymptomatic arteries within the same patient was not possible because the individual studies did not provide the standard deviation of the differences (in SUV) between the symptomatic and asymptomatic arteries (required to compute Cohen’s effect size for paired data). As a result, we assumed independence of the symptomatic and asymptomatic arteries within a patient (i.e., to be conservative) and calculated the effect size as the mean difference between symptomatic and asymptomatic arteries divided by the pooled standard deviation. The individual study effect sizes were then combined into a pooled effect size. In studies where the median and range (or interquartile range) were reported instead of the mean and standard deviation, the median was used instead of the mean and the standard deviation was estimated from the range/interquartile range. Due to the significant variety in reported outcomes, degrees of stenosis studied, scan protocoling and analysis, and symptom-to-scan times, a random-effects (DerSimonian-Laird) model was used to pool the effect sizes when a minimum of 3 or more studies were available to be pooled. Additionally, we performed 5 subgroup analyses in order to adjust for the two most consistently reported patient covariates: degree of stenosis and radiotracer circulating time. Heterogeneity was measured using Cochran’s Q and a p-value ≤ 0.20 was used to indicate the presence of heterogeneity. Heterogeneity was also tested using an inconsistency measure (the I² statistic), and a I² percentage ≥ 50% was used to indicate the presence of heterogeneity. Publication bias was statistically tested with the Begg–Mazumdar rank-correlation test and Egger’s test. Meta-regression (to further explore sources of heterogeneity between the individual study effect sizes) was not performed due to the small number of studies in the analysis. Meta-analysis was conducted with the use of StatsDirect statistical software (Version 3.1.20) (7/18/2018 StatsDirect Ltd, Cheshire, England).

Results

A total of 4,144 abstracts were screened, of which 30 studies potentially met our inclusion criteria. Ultimately, 13 studies^14–26 were found to be eligible for review (Table 1). One study²⁷ was found to meet the inclusion criteria but was not included in analysis due to an overlap in sample size with a more recent and larger study²⁰.

TABLE 1.

Overview of Included Studies

	First Author and Year	PET Tracer	Number of Subjects	Mean Age	% Males	Ischemic Events included	Patient Medical Therapy at enrollment			Stenosis % studied	Symptom-to-scan time range (days)
							Statins	Anti-hypertensives	Anti-platelets
1	Rudd 200214	18F-FDG	8	63.5	75 (6/8)	TIA	N/A	N/A	N/A	70–99%	105†
2	Davies 200515	18F-FDG	12	71	83.3 (10/12)	TIA	All	Some	All	65–99%	64±43‡
3	Graebe 201016	18F-FDG	33	68	75.8 (25/33)	Stroke/TIA/amaurosis fugax	All	N/A	All	Any%	<90
4	Moustafa 201017	18F-FDG	16	70.5	87.5 (14/16)	Minor stroke/TIA/amaurosis fugax	14/16*	7/16*	All*	50–99%	<90
5	Kwee 201118	18F-FDG	50	67.8	N/A	non-debilitating stroke/TIA	45/50	N/A	N/A	<70%	9–95
6	Saito 201319	18F-FDG	25	68	88 (22/25)	Stroke/TIA/amaurosis fugax	N/A	N/A	N/A	>70%	<180
7	Chroinin 201420	18F-FDG	61	71.4	72.1 (44/61)	Stroke (≤3 Rankin score)/TIA/amaurosis fugax	54/61*	N/A	51/61*	50–99%	<7
8	Kim 201421	18F-FDG	21	66.3	85.7 (18/21)	Stroke	Some*	N/A	All*	Any%	5.8 (4.3–17.9)†
9	Shaikh 201422	18F-FDG	29	67.1	75.9 (22/29)	Stroke/TIA/amaurosis fugax	All	12/29	All	50–100%	21–35
10	Skagen 201523	18F-FDG	36	67.9	72.2 (26/36)	Minor stroke/TIA/amaurosis fugax	27/36	22/36	N/A	70–99%	<30
11	Hyafil 201624	18F-FDG	18	70	37 (NA)	Cryptogenic stroke	6/18	N/A	7/18	<50%	<7
12	Quirce 201625	18F-FDG, 18F-NaF	9	N/A	88.9 (8/9)	Stroke/TIA	Yes*	N/A	N/A	Any%	<10
13	Vesey 201726	18F-FDG, 18F-NaF	18	71.7	61.5 (11/18)	Minor stroke/TIA	17/18	Many	17/18	>50% males, >70% females	N/A

FDG, fluorodeoxyglucose; NaF, sodium fluoride; N/A, data not available.

^*At the time of PET scan as opposed to time of enrollment,

^†median and interquartile ration,

^‡mean and standard deviation

Qualitative Assessment and Study Characteristics

All 13 selected articles were prospective cohort studies. Five studies were conducted in the United Kingdom^{14,15,17,22,26}, and 1 each in Ireland²⁰, Denmark¹⁶, Germany²⁴, Japan¹⁹, Republic of Korea²¹, the Netherlands¹⁸, Norway²³ and Spain²⁵. A wide range of carotid stenosis severity was studied amongst the reviewed literature with 3 studies investigating arteries regardless of degree of stenosis^16,21,25, 3 studies including patients with ≥50% stenosis^17,20,22, and 3 including patients with ≥70% stenosis^14,19,23. Four studies examined a mixture of nonstenotic and stenotic arteries^15,18,24,26.

The mean age of patients across the included articles was similar and ranged from 63.5 to 71.7. All but one study¹⁵ had a preponderance of males with percentage males ranging from 37% to 88.9%, with an average of 73.9% males across all studies. Symptom-to-scan time varied between 3 days and 6 months across the studies. One study scanned patients awaiting carotid endarterectomy but did not specify a timeline or symptom-to-scan time²⁶. Tracer dosing and scan protocols varied widely amongst the included articles (Supplemental Table III). All but two studies^18,26 reported a single metric of interest. For the studies that included two metrics, we collected data on max TBR_max rather than TBR_mean in one article¹⁸, and mean SUV_max in another²⁶ in line with the most common metrics in our meta-analysis (Supplemental Table I).

Amongst the 13 included articles, 2 investigated carotid ¹⁸F-FDG uptake in relation to recurrence^20,21, and another 2 also analyzed carotid ¹⁸F-NaF uptake^25,26. The patient characteristics of these cohorts did not differ considerably from the main ¹⁸F-FDG cohort but because only 2 cohorts were found in each group, quantitative analysis was not carried out. With regards to ¹⁸F-NaF cohort, these two articles investigated a total of 27 subjects with ¹⁸F-FDG and ¹⁸F-NaF. Quirce et al.²⁵ showed that carotid ¹⁸F-NaF uptake was overall higher in symptomatic plaques (2.12±0.44 vs 1.85±0.46) but the results did not reach significance amongst the either ¹⁸F-FDG or ¹⁸F-NaF cohorts due to the relatively small sample size (n=9). Vesey et al.²⁶ on the other hand found a significantly higher ¹⁸F-NaF uptake in symptomatic carotid plaques than contralateral plaques (2.42, IQR=2.24–3.24 vs. 1.97, IQR =1.78–2.74), but did not find the same with ¹⁸F-FDG.

Risk of Bias Assessment

Our risk of bias questionnaire (Supplemental Table II) revealed that all 13 studies reported their inclusion criteria clearly, but two did not specify exclusion criteria^19,22. Eight of the 13 included studies evaluated carotid plaque uptake in relation to recent or future ischemic events as their primary objective^{14,15,18,19,23,24,26}. Given the semi-quantitative nature of uptake measurement, some studies showed risk of outcome ascertainment bias. Only 7 studies described some form of blinding to patient characteristics or stroke laterality in the image analysis process^{15,18,20,21,23,24,26}. Furthermore, only 2 studies carried out inter-rater reliability analysis^24,26. Four studies provided individual patient uptake data^14,15,17,25. Six studies in total collected potentially confounding factors^{16,20,21,23,24,26}. Loss to follow-up was not reported explicitly in the a majority of the included studies with the exception of 2 articles, of which one reported complete follow-up²⁰ and another reported a loss of 3 patients in the ¹⁸F-FDG and 4 in the ¹⁸F-NaF arms of the study²⁶.

Meta-analysis Results

Eleven of the 13 studies meeting the inclusion criteria were amenable for ¹⁸F-FDG meta-analysis^{14–20,22–26}. Two studies were not included because one study¹⁹ reported uptake as either positive or negative based on a predefined threshold, and another²¹ study did not provide asymptomatic artery uptake for comparison. A total of 290 subjects were analyzed for 266 symptomatic and 220 asymptomatic arteries using a random-effects model (Table 2 and Figure 2). Because three studies^19,21,23 compared plaques from symptomatic patients to plaques from asymptomatic controls, carotid artery uptake values from all studies were conservatively analyzed as independent even when they originated from the same subject. We found that carotid ¹⁸F-FDG uptake was significantly higher in recently symptomatic carotid arteries than asymptomatic ones (Cohen’s d = 0.492, CI = 0.130 – 0.855, P = 0.008). When a fixed-effects model was used, the results were similar. We found no significant publication bias (Kendall’s tau [Begg-Mazumdar test] = 0.454, P = 0.0602; Egger’s bias = 2.32, P = 0.103) but found significant heterogeneity amongst the analyzed literature (Cochran’s Q = 31.5, P = 0.0005; I² = 68.3%) (Supplemental Figure II and Table IV).

TABLE 2.

¹⁸F-FDG Studies Results

	First Author and Year	Number of Subjects	Selected Metric of Interest	Symptomatic Arteries (N)	Mean Symptomatic Value (SD)	Asymptomatic Arteries (N)	Mean Asymptomatic Value (SD)
1	Rudd 200214	8	Net FDG accumulation rate	8	7.95 (0.58)	6	6.21 (0.96)
2	Davies 200515	12	Plaque-to-normal artery uptake ratio	12	1.705 (0.44)	6	1.145 (0.25)
3	Graebe 201016	33	Mean SUVmax†	34	2.086 (1.76–2.78)	21	1.847 (1.65–2.13)
4	Moustafa 201017	16	PV-corrected SUVmean	16	3.723 (0.97)	16	4.257 (2.03)
5	Kwee 201118	50	Max TBRmax*	50	1.46 (0.05)	50	1.44 (0.06)
7	Chroinin 201420	61	Mean SUVmax	61	2.483 (0.616)	56	2.341 (0.478)
9	Shaikh 201422	29	Patlak maximum dynamic FDG uptake‡	29	9.7 (7.1–12.2)	5	8.6 (6.8–11.9)
10	Skagen 201523	36	Mean SUVmax‡	18	1.75 (1.26–2.04)	18	1.43 (1.15–2.38)
11	Hyafil 201624	18	Mean SUVmax	18	2.5 (0.6)	18	2.5 (0.8)
12	Quirce 201625	9	Max TBRmax	9	1.85 (0.37)	9	1.81 (0.53)
13	Vesey 201726	18	Mean SUVmax	11	2.21 (0.72)	15	2.24 (0.74)
	Total	290		266	Total	220

FDG, fluorodeoxyglucose; SUV, standardized uptake value; TBR, target-to-background ratio; SD, standard deviation.

^*Standard error reported,

^†median and inter-quartile range reported,

^‡median and range reported

FIGURE. 2. Effect size forest plot.

illustrates the association between carotid ¹⁸F-FDG uptake and recent ipsilateral ischemic events. Meta-analysis was carried out using a random-effects model. Squares represent point estimates of effect size (Cohen’s d). The size of the squares is proportional to the inverse of the variance of the estimate. The diamond represents the pooled estimate and horizontal lines represent the 95% confidence interval of each study.

When subgroup analysis was carried out (Supplemental Figures III–VII), we found that ¹⁸F-FDG uptake remained significantly higher in ipsilateral carotid arteries of patients with >70% stenosis (Cohen’s d = 1.645, CI = 1.000 – 2.289, P < 0.0001) and >50% stenosis (Cohen’s d = 0.737, CI = 0.143 – 1.331, P = 0.0151). Our results were not significant when we pooled studies of <70% stenosis (P = 0.894). Only one study investigated <50% stenosis, and so no analysis on this subgroup could be carried out. Analyzing studies that utilized 2 or more hours of circulating time approached significance (P = 0.06), and so did studies that used less than 2 hours of circulating time (P = 0.09).

Ischemic Event Recurrence Studies

Only 2 studies assessed recurrence of ischemic events after measurement of ¹⁸F-FDG plaque uptake^20,21. No studies investigating first time cerebral ischemic events in relation to baseline asymptomatic carotid artery tracer uptake were found. Chroinin et al.²⁰ followed a cohort of 61 subjects with no loss to follow-up for 90 days after the index event, finding that ¹⁸F-FDG uptake was associated with 90-day recurrence (odds ratio per unit increase in max SUV_max of 5.76, CI=1.48–22.39, p=0.01 after adjustment). Kim et al.²¹ also assessed recurrence using a series of diffusion-weighted imaging (DWI) scans a median of 6.4 days apart after an ischemic event in 21 patients and found that those with early recurrent ischemic lesions on repeat DWI had a significantly higher ipsilateral uptake than those that did not (3.07±0.79 vs. 2.17±0.68, p=0.013).

Discussion

In the assessment of carotid artery plaques, stenosis severity measurements play an important role in management and treatment guidelines^28–30. However, emerging evidence suggests that non-invasive modalities can characterize and assess carotid atherosclerotic lesion vulnerability³¹. In this systematic review and meta-analysis, we found that patients who suffered recent cerebral ischemic events had significantly higher ¹⁸F-FDG uptake in the ipsilateral carotid artery. We also found early evidence suggesting that ¹⁸F-NaF may similarly identify symptomatic plaques but that further studies are needed to arrive at more definitive conclusions about the role of ¹⁸F-NaF in characterizing carotid plaque.

We also found some important limitations of the existing body of PET literature of symptomatic carotid disease. Firstly, we noted considerable inconsistency in study methods and outcome reporting (Supplemental Table V), which likely contributed heavily to the heterogeneity found in our results. Unfortunately, due to the small number of studies in the meta-analysis overall, conducting a formal meta-regression to statistically evaluate potential study-level factors that could contribute to the observed heterogeneity in the individual study effects sizes was not possible. Specifically, with only 11 studies in total in the ¹⁸F-FDG meta-analysis, and with a small number of all 13 studies falling into categorical levels of study-level covariates of interest (e.g., degree of stenosis, scan protocolling and analysis, etc.), conducting a formal meta-regression analysis would be grossly underpowered for detecting significant heterogeneity related to the potential study-level covariates of interest. Therefore, our results should be interpreted with caution. Future studies may benefit from a more standardized approach as suggested by several studies^32–34 such as allowing for at least 2 hours of tracer circulating time, injecting an adequate dose of ¹⁸F-FDG to measure uptake in atherosclerosis, and avoiding scanning patients with blood glucose levels above a certain threshold. Additionally, due to considerable overlap between the symptomatic and asymptomatic uptake values as well as the technical variability of PET scans, definitions of threshold values for clinical use have proven difficult to establish. Our literature review revealed some uptake thresholds defined based on histological analysis of inflammation³⁵, and others based on uptake in healthy subjects^36,37. Furthermore, this review highlights the lack of consensus on uptake reporting despite an evident trend towards using TBR for carotid plaque imaging given evidence suggesting that TBR correlates better with inflammation on histology than SUV³⁸. Nevertheless, we recommend that studies report all available metrics to allow systematic analysis and validation of these metrics.

Another limitation of the existing literature, especially in the case of ¹⁸F-FDG, is the failure to account for several patient variables including patient risk factors and comorbidities, and patient medical therapy prior to scanning. Because these variables were not routinely reported in the literature, they could not be fully accounted for in our analysis. For example, several studies^39,40 have found that statin therapy lowers plaque ¹⁸F-FDG uptake, with one study noting a significant reduction in arterial inflammation within 4 weeks⁴¹. As such, we were limited in our ability to determine the magnitude for any effect modification statins, for example, may have had on our results. Similarly, no literature was found describing the impact statins on ¹⁸F-NaF uptake.

Though our study shows that the role of ¹⁸F-FDG as a marker of arterial inflammation has been relatively well-studied, understanding of the mechanism of ¹⁸F-NaF uptake in carotid plaques is based on relatively recent experimental data. ¹⁸F-NaF, commonly used in clinical practice to identify bone metastases, binds to hydroxyapatite in vascular calcifications, and is able to detect microcalcifications below the resolution of CT⁴². Such microcalcifications have been hypothesized to be a significant contributor to plaque instability⁴³ given cellular data suggesting that the earliest stages of atherosclerotic vascular microcalcification is a direct response to an intense inflammatory stimulus⁴⁴. Furthermore, studies suggest that while macroscopic calcium deposits in plaques confer additional mechanical stability, microcalcifications can significantly increase mechanical stress in the fibrous cap⁴⁵. In fact, ¹⁸F-NaF uptake has been reported to highlight calcifications in a distribution that does not always overlap with macrocalcifications as seen on CT⁴⁶, without significant regional overlap between macrocalcifications and ¹⁸F-NaF uptake⁸. These pathophysiologic insights suggest that ¹⁸F-NaF may be similarly effective to ¹⁸F-FDG in identifying in culprit plaques, though investigations in larger studies are required to confirm this hypothesis. More studies comparing ¹⁸F-FDG and ¹⁸F-NaF uptake distributions may shed light on each tracer’s clinical utility, reflecting the different pathophysiological processes they proxy.

Given that our quantitative analysis revealed substantial heterogeneity, our study suggests that definitive conclusions regarding the differences of ¹⁸F-FDG uptake in symptomatic versus asymptomatic plaques will require that future studies consider more standardized methodology in terms of image acquisition techniques, image analysis and outcome reporting. At present, though the evidence is insufficient to generalize the significant results of this meta-analysis to all stroke patients or for the use of PET imaging as a routine diagnostic tool in such patients, directed future research efforts are needed to allow for better validation and interpretation of test results. Despite the current limitations in the literature, our systematic review and meta-analysis may suggest a promising role for PET imaging in non-invasive interrogation of atherosclerosis, as part of a multimodal diagnostic risk stratification strategy in patients suffering cerebral ischemia likely due to carotid atherosclerosis. In addition to valuable diagnostic information provided by MRI or CT^13,47, PET may help identify active atherosclerotic lesions, especially in the case of several suspect plaques or non-stenotic atherosclerosis, for medical or targeted surgical therapy.