The value of FDG combined with PiB PET in the diagnosis of patients with cognitive impairment in a memory clinic

Abstract Aims To analyze the value of 18F‐fluorodeoxyglucose (FDG) positron emission tomography (PET) combined with amyloid PET in cognitive impairment diagnosis. Methods A total of 187 patients with dementia or mild cognitive impairment (MCI) who underwent 11C‐Pittsburgh compound B (PiB) and FDG PET scans in a memory clinic were included in the final analysis. Results Amyloid‐positive and amyloid‐negative dementia patient groups showed a significant difference in the proportion of individuals presenting temporoparietal cortex (p < 0.001) and posterior cingulate/precuneus cortex (p < 0.001) hypometabolism. The sensitivity and specificity of this hypometabolic pattern for identifying amyloid pathology were 72.61% and 77.97%, respectively, in patients clinically diagnosed with AD and 60.87% and 76.19%, respectively, in patients with MCI. The initial diagnosis was changed in 32.17% of patients with dementia after considering both PiB and FDG results. There was a significant difference in both the proportion of patients showing the hypometabolic pattern and PiB positivity between dementia conversion patients and patients with a stable diagnosis of MCI (p < 0.05). Conclusion Temporoparietal and posterior cingulate/precuneus cortex hypometabolism on FDG PET suggested amyloid pathology in patients with cognitive impairment and is helpful in diagnostic decision‐making and predicting AD dementia conversion from MCI, particularly when combined with amyloid PET.

have become increasingly important for facilitating etiology-specific treatment in patients with dementia.5][6][7] Compared with indirect measurement of abnormal molecular expression in the cerebral spinal fluid (CSF), PET scans are noninvasive and able to reflect the topographical distribution of misfolded protein deposition and metabolic changes.Although novel tracers for tau and neuroinflammation have been approved and investigated, respectively, for dementia diagnosis, amyloid and fluorine-18-fluorodeoxyglucose ( 18 F-FDG) tracers are still the most commonly used and available in clinical practice.
Amyloid PET ligands, such as 18 F-florbetapir, 18 F-florbetaben, and 18 F-flutemetamol, which detect and quantify amyloid neurotic plaques in vivo, have been approved for the diagnosis of Alzheimer's disease (AD) by the US FDA.Previously, carbon-11labeled Pittsburgh compound B ( 11 C-PiB), the first tracer specific to β-amyloid (Aβ) applied in human studies, 8 was found to bind with high affinity to fibrillar Aβ deposits in pathological studies. 9,10rebral hypometabolism measured with FDG PET, which reflects synaptic failure at the cellular level, is a downstream marker indicating neuronal injury and neurodegeneration 11,12 and has been included in the AD research framework 4 and in diagnostic criteria for other types of dementia, according to the characteristic hypometabolic locations or patterns, such as predominant posterior cingulate and temporoparietal cortex in typical AD, frontal and anterior temporal hypometabolism in frontotemporal lobar degeneration (FTLD), 5,6 and occipital hypometabolism with or without the cingulate island sign in dementia with Lewy bodies (DLB). 7Based on a literature search using the population, intervention, comparison, and outcome (PICO) model, a previous study used the Delphi method to determine that FDG PET was good for discriminating DLB from AD (sensitivity: 70%-92%, specificity: 74%-100%), fair for discriminating FTLD from AD (sensitivity: 80%-99%, specificity: 63%-98%), and lacking for discriminating other types of dementia. 13mpared with FDG, PiB had higher sensitivity (96% vs. 80%) and similar specificity (86% vs. 84%) for detecting AD neuropathologic changes in autopsy-confirmed participants, and the sensitivity and specificity approached 97% and 98%, respectively, when both imaging modalities were congruent. 14Moreover, in a cohort with cognitive impairment, it was reported that in 23% of cases, neurologists changed their initial diagnosis after considering FDG and PiB PET results, with the combination of PiB and FDG contributing to the most changed diagnoses (68%), followed by PiB only (28%) and FDG only (7%). 15erefore, although amyloid PET has a good performance in predicting AD pathology, FDG PET shows additional value in increasing the diagnostic accuracy, differentiating with other forms of dementia, and influencing subsequent diagnosis and treatment decisions.In this study, the positivity of PiB PET and the hypometabolic pattern of FDG PET were retrospectively analyzed in patients with dementia or mild cognitive impairment (MCI) being treated at a memory clinic.
The diagnostic accuracies of FDG PET for both AD and MCI due to AD were identified based on the amyloid positivity of PiB PET.The value of combining PiB and FDG PET scans on the diagnostic process was further summarized.

| Participants
Detailed written informed consent was obtained from all subjects and their relatives.A total of 201 consecutive patients with cognitive impairment who visited the memory clinic and received PiB and FDG PET scans at Tianjin Medical University General Hospital between July 2014 and August 2021 were retrospectively reviewed.
All patients also underwent a comprehensive evaluation, such as a medical history collection based on both subject and caregiver interviews, physical and neurological examinations, laboratory tests (e.g., thyroid function, vitamin B12, folate, and syphilis serology), brain magnetic resonance imaging (MRI), and neuropsychological assessments.After careful review of the medical records of all participants, 187 patients were included in the final analysis and 12 patients without detailed clinical data were excluded.
All participants were diagnosed by dementia specialists at the memory clinic according to specific diagnostic criteria, such as the revised National Institute on Aging and the Alzheimer's Association (NIA-AA) criteria for probable AD dementia, 16 Petersen's criteria for MCI, 17 the revised Frontotemporal Dementia Consensus criteria for behavioral variant frontotemporal dementia (bvFTD), 5 the classification recommendations for primary progressive aphasia (PPA), 6 the Third or Fourth report of DLB Consortium for probable DLB, 7,18 the clinical criteria for corticobasal syndrome (CBS), 19 the Vascular Behavioral and Cognitive Disorders criteria for vascular dementia (VaD), 20 and the fifth edition of the diagnostic and statistical manual of mental disorders (DSM-5) criteria for depression. 21r further analysis, persons with dementia and persons with MCI were divided into two groups, with persons with depression included in the dementia group because they presented pseudodementia.

| PET imaging
PiB and FDG PET scans were conducted at the PET/CT center of Tianjin Medical University General Hospital on a Discovery PET/ CT 710 scanner (GE Healthcare) in the three-dimensional scanning mode.PiB was injected into an antecubital vein as a bolus injection, with a mean dose of 370-555 MBq.Images were acquired during a 90-min dynamic PET scan (34 frames: 4 × 15 s, 8 × 30 s, 9 × 60 s, 2 × 180 s, 8 × 300 s, 3 × 600 s).At a minimum of 1 h after PiB injection, subjects were intravenously injected with 259 MBq of FDG and then received a 10-min static PET scan at 40 mins after FDG injection.Each frame produced 47 slices with 3.75 mm thickness, which covered the whole brain.Both PiB and FDG images were reconstructed to a 256 × 256 matrix (pixel size of 1.37 mm 2 ).

| PET interpretation
Positron emission tomography scans were visually read by two experienced nuclear medicine physicians (Li Cai and Ying Wang) according to a procedure described in our previous publications. 22,23e mean values for all the specific regions were calculated from the integral PiB image.The positivity or negativity of PiB PET was determined by the ratio of the mean value of the target region to that of the cerebellum with a cutoff value of 1.5 (the upper 95% confidence interval from a cluster analysis of healthy individuals).
Fluorodeoxyglucose frames for each subject were summed and normalized to the mean activity in the pons, then were presented in the NIH color scale and could be windowed and viewed in three planes at the rater's discretion.In addition to examination of specific brain regions, such as the medial frontal lobes, lateral frontal lobes, anterior temporal cortex, temporoparietal cortex, posterior cingulate/precuneus cortices, and occipital lobe, the results of FDG PET images were further classified into two patterns according to their hypometabolic topography: the "AD-typical pattern" was defined as hypometabolism that was mostly observed in the temporoparietal cortex and posterior cingulate cortex and the "non-AD pattern" was characterized by either hypometabolism mainly in other brain areas or nonsignificant hypometabolism relative to normal controls.

| Follow-up
Diagnostic changes after considering PET results were recorded for all patients with dementia.For patients with MCI, long-term followup was performed, mainly including medical histories and neuropsychological testing, to determine the conversion to dementia, which was diagnosed according to the criteria for major neurocognitive disorder of DSM-5 21 and diagnostic criteria for specific diseases or syndromes mentioned above.

CBS (n = 1)
Age, years 67.72.62%, specificity: 55.93%), the "AD-typical pattern" on FDG PET had a higher AUC of 0.753 (p < 0.05), with sensitivity and specificity of 72.61% and 77.97% for identifying AD pathology in dementia patients.In terms of MCI patients, the "AD-typical pattern" on FDG PET showed an AUC of 0.685 (p < 0.05), with a sensitivity and specificity of 60.87% and 76.19%, respectively.

| Diagnostic changes after PET assessment in patients with dementia
The initial diagnosis was changed after considering PET results in   In this study, we retrospectively analyzed the value of FDG PET combined with PiB PET in the diagnosis of patients with cognitive impairment at memory clinics.Amyloid deposition measured with PiB PET was observed in not only patients clinically diagnosed with AD (72.34%) and MCI (52.27%) but also patients with other types of dementia, such as DLB (66.67%),VaD (66.67%), and FTLD (28.00%).

| Prediction of conversion to dementia from MCI for PiB and FDG PET
A typical pattern featuring hypometabolism in the temporoparietal cortex and posterior cingulate cortex showed moderate-to-high accuracy in predicting amyloid deposition.Both PiB and FDG had effects on diagnostic decisions in patients with dementia.Moreover, although to a lesser degree than PiB, the typical hypometabolic pattern on FDG also effectively predicted MCI conversion to AD dementia.
Similar to previous reports, 24  27.66% of patients with clinically diagnosed AD showed an absence of PiB binding in the present study, most of whom had their diagnosis changed to other types of dementia at follow-up.On the contrary, false-negative findings are possible and negative PiB results cannot rule out AD diagnosis since PiB PET may not be able to detect more soluble species of Aβ42 or atypical amyloid deposits. 25Therefore, a few patients who were PiB TA B L E 2 Diagnostic changes in dementia patients stratified by FDG and PiB PET results.negative in our memory clinic still retained an AD diagnosis according to their clinical presentations.7][28] Currently, for clinically diagnosed AD patients with negative amyloid biomarkers, it is still difficult to determine a definite diagnosis unless neuropathological evidence is obtained.

AD
A fraction of patients with clinically diagnosed FTLD also presented with PiB-positive results, with 35.71% of bvFTD and 20.00% of PPA (including nfvPPA and svPPA) patients in this study.
Postmortem studies showed that ~ 15%-20% of clinically diagnosed bvFTD cases could be caused by AD pathology, 29,30 also known as "the fvAD." 31 Alternatively, comorbid FTLD and AD pathology may be present, with FTLD pathology as the dominant force driving the clinical presentation and amyloid pathology as a by-product of aging. 32It has been suggested that svPPA and nfvPPA are generally caused by FTLD (mainly tau and TDP-43 proteinopathies), 33 while lvPPA is mainly caused by AD pathology. 34,35However, a considerable proportion of nfvPPA (20%) and svPPA (16%) cases presented Aβ pathology according to a meta-analysis using individual participant data from 36 centers. 36Although CBS caused by AD pathology has been recently recognized and considered an atypical variant of AD, 37 the CBS patient in this study had negative results on PiB.
Accordingly, amyloid PET is useful in identifying AD pathology as the underlying cause or comorbidity in patients with complex presentations, such as PPA variants or CBS, 37,38 particularly in the condition of DMT targeting Aβ.
We found a large percentage (66.67%) of DLB patients presenting a positive result on PiB PET.This result was supported by previous findings that elevated Aβ on PET scans is a common observation in patients with DLB (up to 60% of patients). 24,39,40It has been demonstrated that diffuse plaques primarily comprising Aβ42 are typically abundant in patients with Lewy body disease at autopsy and contribute to elevated PiB binding. 41In addition, 66.67% of patients with VaD presented as PiB positive in the present study.These findings supported that concurrent cerebrovascular, Lewy body and AD pathology are very common in elderly patients with dementia.
Compared with amyloid-negative patients, a higher proportion of amyloid-positive patients showed significant hypometabolism in the temporoparietal and posterior cingulate/precuneus cortex on FDG PET images, and this effect was even present in patients with MCI.6][47] Amyloid positivity was 26.5% in the nonamnestic MCI (naMCI) group and 64.7% in the amnestic MCI (aMCI) group in a previous study. 47We did not differentiate these two subtypes in our participants, although most patients were classified as aMCI based on medical records.Moreover, consistent a meta-analysis showing that amyloid PET had a sensitivity of 93% and a specificity of 56% for predicting MCI conversion to AD, 48 PiB results showed a high value in predicting AD dementia conversion from MCI at long-term follow-up with a median time of 32 months.Moreover, the "AD-typical pattern" of hypometabolism on FDG PET also predicted AD conversion in persons with MCI.However, the sensitivity and specificity of FDG PET varied from 25% to 100% and from 15% to 100% for predicting conversion from MCI to AD in previous studies. 49Another study found that FDG showed higher specificity (100% vs. 62%) but lower sensitivity (79% vs. 100%) than PiB PET in predicting AD conversion. 50The observed variability was likely due to variability in demographic variables, follow-up durations, and more importantly, the heterogeneity in the analysis methodology to define PiB positivity and FDG regions and patterns.In addition, as a downstream marker for neuronal injury, changes in FDG likely appear later than amyloid deposition on PET and therefore show a relatively high specificity and low sensitivity in predicting AD conversion.

ACK N OWLED G M ENTS
This work was financially supported by the Tianjin Health Science and Technology Project (ZC20230) and Tianjin Key Medical

3 . 1 |
Continuous variables were expressed as the mean ± standard deviation.Categorical data were expressed as numbers and percentages.The differences in proportions of hypometabolic regions and hypometabolic patterns between amyloid-positive persons and amyloid-negative persons were examined with chi-squared tests for the dementia group and MCI group, respectively.Receiver operating characteristic (ROC) curves were used to determine the performance of FDG PET in predicting AD pathology in patients with dementia or MCI according to the PiB PET results.Chi-squared tests were used to examine the differences in proportions of hypometabolic regions and hypometabolic patterns between persons with MCI remaining stable and those converting to dementia.p values <0.05 were considered statistically significant.3 | RE SULTS Demographic and clinical characteristics and results of PiB PET Participants were initially diagnosed with AD (n = 94, including 92 participants with typical AD and two with posterior variant of AD), MCI (n = 44), FTLD (n = 25, including 14 participants with bvFTD, six with a semantic variant of PPA [svPPA], four with a nonfluent variant of PPA [nfvPPA], and one with CBS), DLB (n = 3), VaD (n = 3, including one patient with VaD and two with mixed dementia), depression (n = 6), and dementia of unclear etiology (not otherwise specified [Dem NOS], n = 12) before PET scan measurements.There were 107patients presenting a positive result on PiB PET (Table1).The proportions of amyloid positivity were 72.34% (68/94), 52.27% (23/44), TA B L E 1 Demographic and clinical characteristics according to the initial clinical diagnosis prior to PET assessment.

Figure 1
Figure 1 shows representative FDG and PiB images of four participants with dementia and two participants with MCI.Compared with amyloid-negative patients, a higher proportion of amyloidpositive patients had hypometabolism in the temporoparietal cortex (p < 0.001) and posterior cingulate/precuneus cortex (p < 0.001) in the dementia group; no statistically significant difference in the hypometabolic region was observed in the MCI group (TableS1).There was a significant difference in the proportion of hypometabolic patterns between amyloid-positive and amyloid-negative patients with dementia (p < 0.05) and MCI (p < 0.05).Specifically, the "AD-typical pattern" was shown in 61 (72.62%) dementia patients and 14 (60.87%)MCI patients with a positive result on PiB PET and in 13 (22.03%)dementia patients and five (23.81%)MCI patients with a negative result on PiB PET (Figure 2A,B).Receiver operating characteristic curve analysis was performed for hypometabolic regions and patterns on FDG with significant differences in group comparison, taking PiB positivity as the gold standard for AD diagnosis (Figure 2C,D).Compared with regional hypometabolism in the temporoparietal cortex (area under the ROC curve [AUC]: 0.563, sensitivity: 38.10%, specificity: 74.58%) and posterior cingulate/precuneus cortex (AUC: 0.643, sensitivity:

32 .
17% of patients with dementia.Clinically diagnosed patients with AD and other types of dementia were stratified into four groups based on both PiB and FDG PET results: PiB+ and ADtypical pattern, PiB+ and non-AD pattern, PiB-and AD-typical pattern, and PiB-and non-AD pattern.Table2shows the diagnostic changes before and after PET assessments in all groups.For clinically diagnosed AD patients, all 49 patients with PiB+ and AD-typical patterns maintained their original AD diagnosis; the diagnoses of two of 19 patients with PiB+ and non-AD patterns were changed to DLB or svPPA (each n = 1); the diagnoses of 10 of 11 patients with PiB-and AD-typical patterns were changed to bvFTD (n = 3), svPPA (n = 1), VaD (n = 2), encephalopathia alcoholica (n = 1), or Dem NOS (n = 3), respectively; the diagnoses of 14 of 15 patients with PiB-and non-AD patterns were changed to bvFTD (n = 5), svPPA (n = 1), depression (n = 3), and Dem NOS (n = 5), respectively.Considering patients with an initial diagnosis of non-AD dementia, the diagnoses of all 12 patients with PiB+ and AD-typical patterns were revised to AD, with one bvFTD and one nfvPPA patient finally being diagnosed as frontal variant of AD (fvAD) and logopenic F I G U R E 1 FDG-PET and PiB-PET images of six representative participants.(A) An AD patient with an "AD-typical pattern" on the FDG-PET image featuring predominant hypometabolism in the temporoparietal cortex and posterior cingulate cortex and with a positive PiB-PET result.(B) A patient with a semantic variant of PPA due to FTLD who was initially diagnosed with AD before PET examination showed hypometabolism in the anterior temporal lobe and prefrontal lobes ("non-AD pattern") on the FDG image and had negative PiB-PET results.(C) A patient with logopenic variant of AD who was initially diagnosed with a nonfluent variant of PPA before PET examination showed an "AD-typical pattern" but left predominant hypometabolism on the FDG image and had positive PiB results.(D) A patient with posterior variant of AD who was initially diagnosed with DLB before PET examination showed hypometabolism in the occipital lobes and posterior parietal lobes on the FDG image and had positive PiB results.(E) An MCI patient converted to AD dementia 14 months after PET examination and showed an "AD-typical pattern" on the FDG image with positive PiB results.(F) An MCI patient remained stable 17 months after PET examination and showed a "non-AD pattern" with nonspecific hypometabolism in a few temporal and frontal areas on the FDG image and had negative PiB results.AD, Alzheimer's disease; DLB, dementia with Lewy bodies; FDG, fluorodeoxyglucose; MCI, mild cognitive impairment; PET, positron emission tomography; PiB, Pittsburgh compound B; PPA, primary progressive.variant of PPA (lvPPA), respectively; in four patients with PiB+ and non-AD patterns, the diagnosis of one bvFTD patient was changed to fvAD, one DLB patient to posterior variant of AD, one Dem NOS patient to AD, and one patient with depression to Dem NOS; in two patients with PiB-and AD-typical patterns, one patient maintained their original bvFTD diagnosis and one Dem NOS patient's diagnosis was changed VaD; in 31 patients with PiB-and non-AD patterns, the diagnosis of one patient with VaD was changed to DLB, the diagnoses of two patients with mixed VaD were excluded to comorbid with AD pathology, one Dem NOS patient's diagnoses was changed to bvFTD, and the other patients maintained their original diagnosis.

1 | 7 of 9 LIU
Abbreviations: AD, Alzheimer's disease; bvFTD, behavioral variant frontotemporal dementia; Dem NOS, dementia not otherwise specified; DLB, dementia with Lewy bodies; FDG, fluorodeoxyglucose; NA, not applicable; nfvPPA, non-fluent variant of primary progressive aphasia; PET, positron emission tomography; PiB, Pittsburgh compound B; svPPA, semantic variant of primary progressive aphasia; VaD, vascular dementia.a One bvFTD patient was revised to a diagnosis of frontal variant of AD and the nfvPPA patient was revised to a diagnosis of logopenic variant of AD. b The bvFTD patient was revised to a diagnosis of frontal variant of AD and the DLB patient was revised to diagnosis of posterior variant of AD. c Two patients with mixed VaD were excluded to comorbid with AD pathology.
There were several limitations in the present study.First, participants in this study, who received PET scans sometimes for specific reasons, such as diagnostic uncertainty, could not represent the entire population of patients, although PET scans are a current dayto-day practice in memory clinics.Second, since the diagnosis of all participants was based on the clinical diagnostic criteria only and lacked pathological confirmation, it is still impossible to validate the diagnostic accuracy of FDG PET in patients with dementia.Third, since this was a retrospective study, we did not follow up all dementia patients in the long term.These patients' diagnosis could have changed further during long-term follow-up.Finally, although the PET images were analyzed and read by two experienced radiologists for a consensus, the interpretations depended heavily on individual experience and training; additionally, the radiologists were not completely blinded to clinical information.Amyloid deposition measured with PiB PET could be observed in clinically diagnosed patients with many types of dementia, such as DLB, VaD, and FTLD, in addition to AD and MCI.A typical pattern featuring temporoparietal and posterior cingulate/precuneus cortex hypometabolism visually identified on FDG PET could predict amyloid deposition and was also helpful in the diagnosis of patients with dementia and predicting AD dementia conversion for persons with MCI in addition to PiB PET.AUTH O R CO NTR I B UTI O N SFang Liu and Yudi Shi collected the data.Fang Liu, Yudi Shi, and Huifeng Chen analyzed the data.Qiuyan Wu prepared Figures1-3.Ying Wang and Li Cai performed the PET procedure and evaluated the images.Fang Liu wrote the manuscript.Nan Zhang conceived and designed the study and reviewed and revised the manuscript.