Clinical significance of focal ß-amyloid deposition measured by ¹⁸F-flutemetamol PET

We recruited 310 patients with cognitively unimpaired (CU; n = 125), mild cognitive impairment (MCI; n = 125), and AD dementia (ADD, n = 60) who underwent FMM PET between June 2015 and December 2017. The CU individuals had normal age-, sex-, and education-adjusted performance on standardized cognitive tests [10]. The participants with MCI met the criteria proposed by Petersen et al. [11]: (1) subjective memory complaints, (2) relatively normal performance in other cognitive domains, (3) normal activities of daily living (ADL), (4) objective memory impairment below − 1.5 SD on either verbal or visual memory tests, and (5) not demented. The ADD patients met the criteria for dementia by the Diagnostic and Statistical Manual of Mental Disorders 4th Edition, Text Revision (DSM-IV-TR) [12] and were diagnosed with probable ADD according to the NIA-AA core clinical criteria [1]. The patients were recruited from 9 referral hospitals in South Korea (175 from Samsung Medical Center, 135 from Validation Cohort of Korean Brain Aging Study for the Early Diagnosis and Prediction of AD (KBASE-V) [13]). All participants underwent neurologic examination, neuropsychological test, and Apolipoprotein E (APOE) genotyping. We screened blood tests including a complete blood count, blood chemistry, thyroid function, vitamin B12, folate, and syphilis serology and excluded participants with abnormal findings that could affect their cognition. Participants with previous or current neurological or psychiatric diseases such as brain tumors, encephalitis, epilepsy, and depressive disorders that could affect cognitive function were also excluded. On MRI, patients with structural lesions such as hydrocephalus, brain tumors, or traumatic brain injuries were also excluded. The Institutional Review Boards approved this study at all participating centers. We obtained written, informed consent from patients and caregivers.

We performed FMM PET scans using a Discovery 600 PET/CT scanner (GE), Discovery 690 PET/CT scanner (GE), Discovery STE PET/CT scanner (GE), Biography MCT PET/CT scanner (Siemens) or Gemini TF PET/CT scanner (Philips) on a total number of 310 participants as described in a previous study [13]. The participants underwent a 20 min PET scan (4 × 5 min dynamic frames) starting at 90 min after intravenous injection of 185 MBq ± 10% of ¹⁸F-flutemetamol. Low-dose computed tomography was utilized for attenuation correction before scans. We reconstructed the images with the Ordered Subsets Expectation Maximization algorithm using 4 iterations and 16 subsets.

Visual interpretation of the FMM PET images was performed by two blinded neurologists (referred to as “readers”) who successfully completed the manufacturing company’s electronic training program. Visual interpretation of FMM PET images was performed by systematic review of ten brain regions (frontal, parietal, PPC, striatum, and lateral temporal lobes in each hemisphere) [14]. For each region, the readers used dichotomous assessment in classifying images as either normal or abnormal in a rainbow color scale anchored to the pons. We defined each region to be abnormal when there was increased cortical gray matter signal (above 50–60% peak intensity) and/or reduced (or absent) gray-white matter contrast [15]. Inter-reader agreement of interpretation of FMM PET was excellent (kappa score = 0.94).

We classified patients into three groups. No-FMM (no FMM uptake in any region), Focal-FMM (FMM uptake in 1–9 regions), and Diffuse-FMM (FMM uptake in all 10 regions). Examples of FMM PET images are shown in Additional file 1: Figure S1.

CSF sampling was performed in 77 participants (49 CU, 16 MCI, and 12 ADD) by procedures as previously described [13]. We obtained CSF in 15 mL polypropylene tubes (Falcon, Corning Science, NY, USA) and centrifuged at 2000×g for 10 min at room temperature (RT). Approximately 10 cc of the CSF supernatant was frozen and transferred to the laboratory at Inha University. For measuring CSF biomarkers, the CSF was thawed and gently extracted into pipettes with polypropylene tips. A total of 0.4 mL aliquots of CSF was frozen in polypropylene tubes (Sarstedt AG & Co., Nümbrecht, Germany) and stored at − 80 °C until analysis. We measured the level of CSF Aß 42, total tau (t-tau), and phosphorylated tau (p-tau) using the multiplex xMAP Luminex platform with INNO-BIA AlzBio3 kits. AlzBio3 kits (Fujirebio Europe, Ghent, Belgium) contained capture monoclonal antibodies for Aß 42, t-tau, and p-tau, which linked to two aqueous quality controls (a-QC) with pre-defined concentration ranges for the three biomarkers. The procedure is described elsewhere [16, 17]. To reduce the effects of sources of variability on the results [18], CSF analysis was followed by the manufacturer’s instructions and standard of procedures that were previously described [19, 20].

All participants underwent a standardized neuropsychological battery called the Korean version of the Consortium to Establish a Registry for Alzheimer’s Disease Assessment Packet (CERAD-K) [21] or Seoul Neuropsychological Screening Battery (SNSB) [22], which consisted of tests of language, visuospatial, memory, and frontal/executive functions.

Tests in CERAD-K included the Korean version of the Boston Naming Test (K-BNT) to assess language function; constructional praxis (copying figures) to assess visuospatial function; 10 word list recall (20-min delayed recall) and constructional recall (20-min delayed figure recall) to assess verbal and visual memory, respectively; Controlled Oral Word Association Test (COWAT: animal naming) and Stroop Test (color reading) to assess frontal/executive function; and the Mini-Mental State Examination (MMSE) to assess global cognitive function.

Tests in SNSB included the K-BNT to assess language function; Rey-Osterrieth Complex Figure Test (RCFT: copying) to assess visuospatial function; Seoul Verbal Learning Test (20-min delayed recall) and RCFT (20-min delayed recall) to assess verbal and visual memory, respectively; COWAT (animal naming) and the Stroop Test (color reading) to assess frontal-executive function; and the MMSE to assess global cognitive function. [23]

For descriptive statistics, we used the chi-square test and analysis of variance (ANOVA) followed by Bonferroni’s post hoc analyses.

To compare the cognitive profile of the three groups (No-FMM, Focal-FMM, and Diffuse-FMM group), we used ANOVA followed by Bonferroni’s post hoc analyses. When we compared the cognitive profile of the Regional-FMM group with the No-FMM or the Diffuse-FMM group, we used ANOVA followed by Bonferroni’s correction for 30 multiple tests (5 regions and 6 cognitive tests). To evaluate the association between cognition and number of FMM uptake regions, we used linear regression analyses.

For comparison of the CSF profile of the three groups, we used analysis of covariance (ANCOVA) followed by Bonferroni’s post hoc analyses after controlling for age. To evaluate the association between CSF profile and number of FMM uptake regions, linear regression analyses were used after adjusting for age. All statistical tests were performed using MedCalc (MedCalc Software version 19, Ostend, Belgium).

To determine the spreading order of FMM uptake, we assumed that regions with earlier appearance of pathology would show abnormal FMM uptake in a greater number of participants, as suggested by previous studies [24, 25]. The different frequency of regional involvement was assessed using a bootstrapping method with 1000 re-samples in R v3.1.3 (Institute for Statistics and Mathematics, Vienna, Austria; www.​R-project.​org), which derived the estimates of 95% confidence intervals and standard error. We calculated asymptotic p values and corrected for multiple comparisons with Bonferroni’s method for all combinations of regional pairs.

The demographic and clinical characteristics are presented in Table 1. Of all the participants, 17.1% (53/310) patients were classified into the Focal-FMM group. The proportion in the Focal-FMM group and the extent of Focal-FMM uptake differed according to cognitive level. 13.6% (17/125) of CU, 16.0% (20/125) of MCI, and 26.7% (16/60) of ADD were classified into the Focal-FMM group. Among the Focal-FMM group, the median number of uptake regions was 1.0 (95% CI = 1.0–4.0) in CU, 3.5 (95% CI = 2.0–5.8) in MCI, and 6.0 (95% CI = 3.6–8.0) in ADD. The Focal-FMM group was older than the No-FMM group and had more APOE ε4 carriers compared to the No-FMM group. In addition, there were statistically significant differences in distribution of cognitive level across the groups.

Compared to the No-FMM group, the Focal-FMM group showed significantly lower performance in all cognitive domains except for visuospatial function. Compared to the Diffuse-FMM group, the Focal-FMM group showed better performance in verbal memory and visual memory functions. Global cognitive function (measured by MMSE) of the Focal-FMM group was better than the Diffuse-uptake group but worse than the No-FMM group (Table 2).

Among the Focal-FMM group, as the number of FMM uptake regions increased, z-scores decreased in all cognitive domains such as K-BNT (β = − 0.264, p = 0.004), visuospatial function (β = − 0.290, p = 0.010), verbal memory (β = − 0.105, p = − 0.105), visual memory (β = − 0.138, p = 0.021), COWAT (β = − 0.162, p = 0.004), Stroop Test (β = − 0.239, p = 0.004), and MMSE (β = − 0.306, p = 0.016) (Fig. 1.).

We further divided Focal-FMM group into patients with less FMM uptake (1–5 regions involved) group and patients with more FMM uptake (6–9 regions involved) group. Focal-FMM group with 1–5 regions involved did not show significant difference compared to the No-FMM group, while it showed better performance in all cognitive domains except for language when compared to Diffuse-FMM group. Focal-FMM group with 6–9 regions involved showed worse performance in all cognitive domains compared to No-FMM group, while it did not show significant difference when compared to Diffuse-FMM group (Additional file 1: Table S1).

Then, we compared each regional Focal-FMM group with the Focal- or Diffuse-FMM group to evaluate the regional effects of focal FMM uptake on cognitive function. Compared to the Diffuse-FMM group, verbal memory scores were higher in the Focal-FMM group with frontal, lateral temporal, parietal, or PPC regional involvement whereas no difference was found in the Focal-FMM group with striatal involvement. Compared to the Diffuse-FMM group, visual memory scores were higher in the Focal-FMM group with frontal, lateral temporal, or parietal regional involvement whereas no difference was found in the Focal-FMM group with PPC or striatal involvement. Compared to the No-FMM group, verbal and visual memory scores were lower in the Focal-FMM group with PPC or striatal involvement (Table 2).

Comparisons of cognitive scores between the No-, Focal-, and Diffuse-FMM groups in each cognitive level (CU, MCI and ADD) are shown in Additional file 1: Table S2). Among CU individuals, the cognitive score did not differ among the No-, Focal-, and Diffuse-FMM groups. However, among MCI patients, the Focal-FMM group showed better performance in verbal and visual memory function, as well as in MMSE score, compared to the Diffuse-FMM group. Among ADD patients, Focal-FMM patients performed worse in language and frontal function compared to those of the No-FMM group and performed better in verbal memory than those of the Diffuse-FMM group. We provided a breakdown of cases by clinical designation and number of regions in Additional file 1: Table S3).

Levels of CSF AD biomarkers (Aß 42, p-tau, t-tau, p-tau/ Aß 42, and t-tau/ Aß 42) in No-, Focal-, and Diffuse-FMM groups are shown in Fig. 2. The Focal-FMM group showed increased levels of CSF Aß 42 and decreased levels of CSF t-tau and t-tau/Aß 42, compared to the Diffuse-FMM group. However, there were no differences between Focal-FMM-uptake and No-FMM-uptake groups except for p-tau level.

Among the Focal-FMM group, Aß deposition was most frequently observed in the frontal (62.3%, 95% CI = 48.8–75.8) followed by PPC (60.4%, 95% CI = 46.8–74.0), parietal (60.4%, 95% CI = 46.8–74.0), striatum (43.4%, 95% CI = 29.6–57.2), and lateral temporal (39.6%, 95% CI = 26.0–53.2) regions (Fig. 3).