Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 https://doi.org/10.1186/s13195-019-0549-1 RESEARCH Open Access Plasma Aβ42/40 ratio alone or combined with FDG-PET can accurately predict amyloid-PET positivity: a cross-sectional analysis from the AB255 Study Virginia Pérez-Grijalba1, Javier Arbizu2, Judith Romero1, Elena Prieto2, Pedro Pesini1* , Leticia Sarasa1, Fernando Guillen2, Inmaculada Monleón1, Itziar San-José1, Pablo Martínez-Lage3, Josep Munuera4, Isabel Hernández5,6, Mar Buendía5, Oscar Sotolongo-Grau5, Montserrat Alegret5,6, Agustín Ruiz5,6, Lluis Tárraga5,6, Mercè Boada5,6, Manuel Sarasa1 and The AB255 Study Group7 Abstract Background: To facilitate population screening and clinical trials of disease-modifying therapies for Alzheimer’s disease, supportive biomarker information is necessary. This study was aimed to investigate the association of plasma amyloid-beta (Aβ) levels with the presence of pathological accumulation of Aβ in the brain measured by amyloid-PET. Both plasma Aβ42/40 ratio alone or combined with an FDG-PET-based biomarker of neurodegeneration were assessed as potential AD biomarkers. Methods: We included 39 cognitively normal subjects and 20 patients with mild cognitive impairment from the AB255 Study who had undergone PiB-PET scans. Total Aβ40 and Aβ42 levels in plasma (TP42/40) were quantified using ABtest kits. Subjects were dichotomized as Aβ-PET positive or negative, and the ability of TP42/40 to detect Aβ-PET positivity was assessed by logistic regression and receiver operating characteristic analyses. Combination of plasma Aβ biomarkers and FDG-PET was further assessed as an improvement for brain amyloidosis detection and diagnosis classification. Results: Eighteen (30.5%) subjects were Aβ-PET positive. TP42/40 ratio alone identified Aβ-PET status with an area under the curve (AUC) of 0.881 (95% confidence interval [CI] = 0.779–0.982). Discriminating performance of TP42/40 to detect Aβ-PET-positive subjects yielded sensitivity and specificity values at Youden’s cutoff of 77.8% and 87.5%, respectively, with a positive predictive value of 0.732 and negative predictive value of 0.900. All these parameters improved after adjusting the model for significant covariates. Applying TP42/40 as the first screening tool in a sequential diagnostic work-up would reduce the number of Aβ-PET scans by 64%. Combination of both FDG-PET scores and plasma Aβ biomarkers was found to be the most accurate Aβ-PET predictor, with an AUC of 0.965 (95% CI = 0.913–0.100). Conclusions: Plasma TP42/40 ratio showed a relevant and significant potential as a screening tool to identify brain Aβ positivity in preclinical and prodromal stages of Alzheimer’s disease. Keywords: Plasma, Amyloid-beta, FDG-PET, Biomarker, Preclinical Alzheimer’s disease, Amyloid-PET, Mild cognitive impairment * Correspondence: ppesini@araclon.com 1 Araclon Biotech S.L., Vía Hispanidad 21, 50009 Zaragoza, Spain Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 2 of 9 Background score), which has proved to predict and detect AD Alzheimer’s disease (AD) is a progressive neurodegener- dementia with good diagnostic performance [30]. ative disorder and the most common cause of dementia. This work was primarily aimed to investigate the po- Accumulating data from clinical research support that tential value of the TP42/40 ratio as a screening tool for the AD pathophysiologic process starts decades before brain Aβ-PET positivity in a population of 59 CN and the onset of clinical symptoms [1–3]. The disease amnesic MCI (a-MCI) patients. This population was par- develops in a continuum from a preclinical stage in tially explored in a previous work in which clinical per- which amyloid pathology has been defined as the earliest formance of TP42/40 was evaluated (considering clinical Alzheimer’s pathological changes [3–5]. Thus, individ- diagnosis as the gold standard), together with a general uals at the preclinical or prodromal AD stages represent assessment of the correlation of TP42/40 with CSF and an important target group in the context of clinical trials Aβ-PET biomarkers. With this study, we pursued and population screening. New diagnostic procedures confirming previous results from our own group in an can identify measurable brain changes by positron emis- independent AIBL population [22], focusing in the sion tomography (PET) and cerebrospinal fluid (CSF) discriminating performance of TP42/40 in detecting Aβ analysis, even at the preclinical stage. These procedures positivity in the brain. Secondarily, we have explored the have been incorporated in the National Institute of potential value added by the combination of Aβ plasma Aging–Alzheimer’s Association A/T/(N) research frame- levels and AD-Conv score to improve prediction of Aβ- work for the biological definition of AD [5]. PET status at early disease stages. Furthermore, in the A large number of clinical studies very consistently framework of the new biological concept of AD based in show that these neuroimaging or CSF biomarkers pro- the A/T/(N) system [5], we also explored if combination vide relevant information for the diagnosis. Yet, blood- of both the TP42/40 plasma ratio, as a β-amyloid marker based biomarkers are the desirable tool for large-scale (A), and AD-Conv score, reflecting neurodegeneration assessments of patients either in clinical research or, (N), might help in clinical assessment at early stages. eventually, in primary clinical settings, due to their cost- effectiveness and easiness of procedures [6]. Methods Despite previous studies yielding contradictory results Participants [7–9], recent works have shown an association of low The AB255 Study is a multicenter longitudinal study plasma Aβ42/40 ratio with AD [10–13]. Plasma Aβ42/40 with evaluations of the cognitive status of individuals at ratio has also demonstrated value in detecting brain Aβ 0, 12, and 24 months, including 83 cognitively normal pathological changes [14–16], even using different meth- (CN) and 145 age-paired subjects with probable a-MCI odological approaches as mass spectrometry (MS) and [31], all over 64 years of age. The study was designed to new-generation immunological methods [17–19]. Our evaluate the potential of blood-based Aβ biomarkers to group has been largely working on the development of detect AD. A complete description of the AB255 Study reliable and informative plasma biomarkers based on Aβ protocols and population details have been previously measurements [20]. We have previously reported a consist- referred [21]. Clinical diagnosis of each participant was ent association between a low TP42/40 ratio and both clin- performed using an extensive neurological examination ical MCI diagnosis [21] and Aβ accumulation in the brain [32] and a battery of neuropsychological tests, including [22, 23]. Reduced TP42/40 levels were also found to predict evaluation of global cognition using the Mini-Mental State higher rates of Aβ accumulation in the brain [22, 24]. Fur- Examination [33, 34] and verbal learning and memory by thermore, it has been recently reported that lower TP42/40 The Word List Learning test from the Wechsler Memory ratio levels were also associated with increased cortical up- Scale-Third Edition (WMS-III) [35], delayed recall and a take of the [18F]Flortaucipir tau-PET marker in AD-related recognition task without list of interference [36], and the regions [25]. Free and Cued Selective Reminding Test (FCSRT) [37]. A PET imaging, using the most widely available radio- subpopulation of 39 cognitively normal subjects and 20 tracer ([18F]fluorodeoxyglucose PET, FDG-PET) as a patients with a-MCI who had undergone PiB-PET scans measure of cerebral glucose metabolism, is a marker of and fulfilled criteria for inclusion were considered for the neurodegeneration that has been established as a sensi- present work. Participants were all recruited and assessed tive tool for detecting neuronal dysfunction [26]. FDG- from 2010 to 2013 at The Memory Clinic from Fundació PET diagnostic performance could be upgraded by the ACE (Barcelona, Spain) [38]. use of quantitative indices that have been developed to account for inter-observer variability and to support Plasma analyses challenging evaluation [27–29]. Arbizu et al. have EDTA blood was obtained after overnight fasting, and previously described FDG-based automated quantitative samples were immediately cooled to 2–8 °C until proc- scores, such as the AD conversion score (AD-Conv essed within 30 h of collection by centrifugation at Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 3 of 9 2500×g for 15 min at 4 °C. Plasma was appropriately ali- were used as the predictive variables. Demographic char- quoted in polypropylene tubes and stored at − 80 °C acteristics and biomarker levels were first compared be- until analysis, avoiding freeze-thaw cycles. ABtest40 and tween groups using chi-squared and Mann-Whitney ABtest42 ELISA kits (Araclon Biotech Ltd., Spain) were tests as appropriate. Statistically significant differences in used for the quantification of total Aβ40 and Aβ42 in TP42/40 levels between Aβ-PET(+) and Aβ-PET(−) indi- plasma after proprietary treatment of the sample. The viduals were evaluated using generalized linear regres- specific analytical procedures and performance charac- sion models (GLM) adjusted for significant demographic teristics of these tests are described elsewhere [20]. covariates (age, APOE genotype, and clinical group). Samples were randomized and encoded by an external Correlation of quantitative measures of both plasma CRO to guarantee the validity of results. TP42/40 ratio and AD-Conv score with Aβ-PET SUVR was assessed by Spearman rank correlation analyses. PiB-PET analysis The association of TP42/40 and AD-Conv score with Cortical Aβ burden was assessed by PET using 11C- PET-based abnormal amyloid status was further investi- Pittsburg compound B (PiB-PET). Detailed procedures gated using logistic regression followed by receiver oper- of neuroimaging acquisition and analysis are described ating characteristic (ROC) curve analyses. Predicted by Espinosa et al. [31]. Imaging data were analyzed using values of binary logistic regression models were used to the Fundació ACE Pipeline for Neuroimaging Analysis, combine variables in ROC analysis. All TP42/40 models available at http://detritus.fundacioace.com/. Participants were adjusted for age, APOE genotype (APOEε4 allele were classified as β-amyloid positive (PET-Aβ(+)) or β- carriers versus non-carriers), and clinical diagnosis. AD- amyloid negative (PET-Aβ(−)) with relation to a cutoff Conv score was determined by a multivariate model of 1.4 SUVR in PiB-PET scans [39]. already including age and APOE ε4 genotype, so it was not readjusted for these covariates in the analysis. FDG-PET analysis Plasma and FDG-PET biomarkers were evaluated alone PET acquisition was performed after 4 h of fasting and or in combination, so that improvement in classification once it was confirmed that blood glucose levels were performance using both amyloid and neurodegeneration below 110 mg/dl. The dose of 18FDG was established ac- information derived from them was assessed. Youden’s cording to weight (150 μCi/kg), being the standard adult index maximizing cutoffs were considered for evaluation dose 10 mCi (370 MBq) in a volume of 1–10 ml saline. of biomarker performance in detecting brain Aβ positiv- Dosimetry was established in accordance with ICRP 53. ity, either individually or combined. This same proced- For adults, critical organ dose (bladder) was 0.16 mGy/ ure was applied to the total AB255 population to assess MBq and effective dose was 0.019 mSv/MBq. Acquisition TP42/40 and AD-Conv score clinical discrimination started 40 min after 18FDG administration and lasted for ability (CN versus a-MCI). For visualization purposes, 20 min. FDG-PET scans were analyzed following the heat maps representing predicted probability of brain Aβ procedure previously described by Arbizu et al. [30]. positivity depending on the chosen TP42/40 cutoff were Using this automated voxel-based analytical method, depicted with regard to age and APOE genotype. quantitative indices were developed to compute the hypo- metabolic pattern of each subject. The AD-Conv score Results integrates the information provided by PET into a multi- Demographic characteristics of the population included variate model including age, gender, Mini-Mental State in this study are summarized in Table 1. Eighteen Examination (MMSE) score, and APOE ε4 genotype. (30.5%) individuals were classified as Aβ-PET(+), and most of them were clinically diagnosed as a-MCI (14 a- APOE genotyping MCI Aβ-PET(+) versus 6 a-MCI Aβ-PET(−)). Likewise, Genomic DNA was isolated from EDTA blood. Target 35 out of the 39 CN individuals were classified as Aβ- DNA was amplified by PCR and digested with the re- PET(−). Comparing Aβ-PET(+) and Aβ-PET(−) groups, striction enzyme HhaI. APOE genotyping was carried subjects with abnormal Aβ-PET status were on average out by subsequent restriction analysis of the pattern of older and were more frequently APOE 4 carriers. Age, fragments obtained after electrophoresis in a polyacryl- APOE 4, and clinical diagnosis were significantly amide gel [40]. different between groups, whereas no statistically signifi- cant differences were found with regard to gender or Statistical analysis education level (Table 1). Both AD-Conv score and Statistical analysis was performed using SPSS version 22 TP42/40 were significantly different between Aβ-PET for Windows (SPSS Inc., Chicago, IL). A probability level groups (Mann-Whitney p < 0.001) and correlated with of p < 0.05 was considered statistically significant. Total Aβ-PET SUVR measures (TP42/40 rs = − 0.464; AD- in plasma Aβ42/40 ratio (TP42/40) and AD-Conv score Conv score rs = 0.581; both p < 0.001). Figure 1 shows Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 4 of 9 Table 1 Demographic and biomarker features of this population Aβ-PET (−) Aβ-PET (+) p value N (%) 41 (69.5%) 18 (30.5%) Age, mean (SD) 71.6 (4.11) 75.2 (5.65) p 0.014 Gender Female, n (%) 18 (43.9%) 9 (50.0%) p 0.665 APOE4 carriers, n (%) 6 (14.6%) 12 (66.7%) p < 0.001 Education level, mean years (SD) 12.33 (3.96) 10.56 (4.44) p 0.113 CN/a-MCI, n (% a-MCI) 35/6 (14.6%) 4/14 (77.8%) p < 0.001 TP42/40, mean (SD) 0.1329 (0.0208) 0.0997 (0.0197) p < 0.001 AD-Conv score, mean (SD) 0.133 (0.162) 0.384 (0.259) p < 0.001 p value was obtained from Mann-Whitney or chi-squared tests as appropriate. SD standard deviation, CN cognitively normal, a-MCI amnestic mild cognitive impairment, Aβ amyloid-beta, TP42/40 total plasma Aβ42/40 ratio, AD-Conv score AD conversion score based on FDG-PET these associations including information of the clinical improved yielding an AUC of 0.963 (95% CI 0.917– status of each participant. 1.000). Sensitivity, specificity, and NPV were all over Plasma TP42/40 ratio was in average 25% lower in 0.920 (Fig. 3b, Table 2). We used the logistic regression PET-Aβ(+) subjects compared to PET-Aβ(−) individuals model to predict probabilities of Aβ-PET positivity with (Table 1 and Fig. 2). After adjusting for significant regard to age in both APOE ε4 non-carriers (Fig. 4a) demographic covariates (age, APOE 4, and clinical and carriers (Fig. 4b). This figure represents a statistical diagnosis), association of TP42/40 with Aβ-PET status inference since the actual study population does not remained statistically significant (GLM β = 87.67, p = cover all the age ranges represented. As expected, the 0.005). ROC analysis revealed an AUC of 0.881 (95% predicted probability increased with age and was always confidence interval [CI] 0.779–0.982) for the prediction higher in APOE ε4 carriers compared to non-carriers. of Aβ-PET positivity using TP42/40 alone (Fig. 3a). The Low TP42/40 ratio considerably augmented likelihood of Youden’s cutoff of plasma TP42/40 ratio was 0.1049 and abnormal Aβ-PET status for every genotype and age yielded a sensitivity of 77.8% and a specificity of 87.5% condition. Indeed, the high NPV and low LR− character- in detecting PET-Aβ(+) subjects (Table 2). When study- ized TP42/40 as an ideal marker for pre-screening. ing independently CN and a-MCI subjects, we found Considering a hypothetical clinical trial targeting Aβ- excellent performance of TP42/40 in the CN group PET-positive subjects, a recruitment procedure based in (AUC unadjusted TP42/40 0.957, 95% CI 0.890–1.000), Aβ-PET alone would require 3.3 PET scans per success- although it declined in the a-MCI group (AUC un- ful recruitment. However, following a sequential strategy adjusted TP42/40 0.814, 95% CI 0.617–1.000) (Table 2). including a TP42/40 pre-screening stage would require When adjusting TP42/40 with relevant covariates (age, 1.2 Aβ-PET scans per successful recruitment, implying a APOE genotype, and clinical group), discrimination was 64% saving in PET scans. Fig. 1 Correlation of TP42/40 and AD-Conv score with Aβ-PET. Scatterplots of plasma TP42/40 ratio (a) and AD-Conv score (b) levels with regard to PiB-PET. Blue: cognitively normal (CN) subjects. Red: amnestic mild cognitive impairment (a-MCI) individuals. Both TP42/40 and AD-Conv score biomarkers showed a significant correlation with Aβ-PET, although in inverse directions. rs Spearman rank correlation coefficient Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 5 of 9 (Table 2). Thus, we hypothesized that a combination of both plasma TP42/40 and neurodegeneration (FDG- PET) biomarkers could outperform the single-predictor models in detecting Aβ-PET(+) subjects. The combined model of plasma Aβ and neurodegener- ation biomarkers (AdjTP42/40+AD-Conv score) im- proved performance of both individual biomarkers and covariates alone (Fig. 3c) to discriminate Aβ-PET status. AUC for this model was found to be 0.965 (95% CI 0.921–1.000), with very high sensitivity and specificity (> 92%) and NPV of 1.00 (Table 2). On the other hand, TP42/40 performed poorly at dis- criminating between CN and a-MCI patients in the whole AB255 study population (AUC = 0.639, 95% CI 0.564–0.714). Nevertheless, combining TP42/40 with AD-Conv score improved subjects’ clinical classification up to an AUC of 0.915 (95% CI 0.875–0.955). The effect is mainly driven by AD-Con score which by itself alone presented an AUC of 0.893 (95% CI 0.849–0.937). Discussion In this study, we aimed at assessing the potential value Fig. 2 Distribution of TP42/40 levels between Aβ-PET groups. Dots of TP42/40 ratio in the prediction of brain Aβ pathology represent the individual TP42/40 ratio obtained for each patient with measured by Aβ-PET in a preclinical and prodromal AD regard to their Aβ-PET status. Continuous lines represent mean TP42/40 values and standard mean error. The dashed line represents cohort of 59 patients from the AB255 Study. In addition, the Youden’s cutoff (0.1094) PET-Aβ positivity discrimination using the present study evaluated the potential benefit of the TP42/40 alone. p value from the generalized linear model (GLM) was combination of amyloid and neurodegeneration bio- also included markers, reflected by plasma TP42/40 and the FDG- PET-derived AD-Conv score [30]. Both individually and AD-Conv score was also evaluated as a predictor of combined, these biomarkers significantly detected Aβ Aβ-PET positivity in this cohort, showing good diagnos- pathology in the brain. Combination of adjusted TP42/ tic performance (AUC 0.898, 95% CI 0.817–0.979) 40 ratio and AD-Conv score was found to be the most (Fig. 3b), and sensitivity and specificity values at max- accurate Aβ-PET predictor, with sensitivity of 100% and imum Youden’s cutoff of 100% and 67.5%, respectively specificity of 92.3%. Fig. 3 Receiver operating characteristic (ROC) curves discriminating Aβ-PET(+) from Aβ-PET(−) subjects. ROC curve and the corresponding area under the curve (AUC) with 95% confidence interval (CI) are depicted for the following models: a TP42/40 alone; b TP42/40 adjusted for age, APOE genotype, and clinical group (AdjTP42/40) and AD-Conv score; c models including only significant demographic covariates (age, APOE genotype, and clinical group) and the combined model of amyloidosis and neurodegeneration (AdjTP42/40+AD-Conv score) Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 6 of 9 Table 2 Diagnostic performance of individual and combined models in detecting Aβ-PET positivity at Youden’s cutoff Sensitivity Specificity PPV NPV LR+ LR− Whole population (N = 59) TP42/40 0.778 0.875 0.732 0.900 6.22 0.25 Age+APOE4+Clinical Group 0.944 0.805 0.680 0.945 4.84 0.07 TP42/40+Age+APOE4+Clinical Group 0.944 0.925 0.847 0.974 12.59 0.06 AD-Conv score 1.000 0.675 0.706 1.000 3.08 0.00 AdjustedTP42/40+AD-Conv score 1.000 0.923 0.851 1.000 13.00 0.00 Only CN (N = 39) TP42/40 1.000 0.886 0.500 1.000 8.75 0.00 Age+APOE4 0.750 0.943 0.600 0.971 12.12 0.26 TP42/40+Age+APOE4 1.000 1.000 1.000 1.000 – 0.00 AD-Conv score 1.000 0.765 0.327 1.000 4.25 0.00 AdjustedTP42/40+AD-Conv score 1.000 1.000 1.000 1.000 – 0.00 Only a-MCI (N = 20) TP42/40 0.714 1.000 1.000 0.653 – 0.29 Age+APOE4 0.357 0.833 0.799 0.411 2.14 0.77 TP42/40+Age+APOE4 0.643 1.000 1.000 0.601 – 0.36 AD-Conv score 0.846 0.500 0.759 0.636 1.70 0.31 AdjustedTP42/40+AD-Conv score 0.615 1.000 1.000 0.583 – 0.38 Cases in which LR+ was not quantifiable were presented as –. Prevalence considered for each group calculation was corresponding to the present cohort: 30.5% in the whole population, 10.25% in the CN group, and 65% in the a-MCI group. PET PPV-positive predictive value, NPV negative predictive value, LR+ positive likelihood ratio, LR− negative likelihood ratio Aβ-PET(+) patients showed significantly lower levels findings also replicate our previous results showing the of TP42/40, with an average reduction of 25% compared potential of TP42/40 as a surrogate biomarker of Aβ- to the Aβ-PET(−) group. This significant inverse associ- PET status in an independent preclinical population ation of plasma Aβ42/40 ratio with brain Aβ deposition [22]. TP42/40 ratio adjusted for significant covariates is in accordance with previous studies [14–19]. These detected Aβ-PET positivity in the present cohort with Fig. 4 Heat maps showing predicted probability of being Aβ-PET(+) based on TP42/40 and age stratified for APOE genotype. Predicted probabilities in APOE Ɛ4 non-carriers (a) and APOE Ɛ4 carriers (b). Probabilities are displayed as percentages. Note that TP42/40 decreases upward in axis for optimal visualization purposes. It should be noted that probabilities were predicted from a logistic regression model modeling expected probabilities in the age range most likely present during pre-screening or clinical assessment. These predicted probabilities could be more accurately determined in a larger population including the complete age range Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 7 of 9 94.4% sensitivity and 92.5% specificity. This means that significant correlation between PiB-PET and FDG-PET in a sequential screening scenario aiming to detect sub- [55], corresponding well to our findings. jects with abnormal Aβ-PET status, a first step assessing Only few studies have explored combination of FDG- TP42/40 would allow a 64% reduction in the number of PET and Aβ biomarkers, mostly in CSF and strictly Aβ-PET scans. Considering the restricted availability and focused on conversion to AD dementia [43, 46–48]. Inter- costs of Aβ-PET neuroimaging [41], this sequential estingly, combination of TP42/40 and AD-Conv score im- strategy would imply a significant reduction of the pa- proves subjects’ clinical classification into CN or a-MCI. tient burden and procedural costs, both for clinical trials As could be expected, the effect is mainly driven by the and patient management at primary care settings. AD-Conv score which itself integrates into a multivariate Clinical performance of plasma Aβ42/Aβ40 in the model FDG-PET neurodegeneration (downstream to β- present study is comparable to that obtained using MS- amyloid changes as per the amyloid cascade hypothesis) based methods [17, 18] or new-generation immuno- with APOE genotype, MMSE score, age, and gender. This logical assays [19] and are congruent with other studies approach is in agreement with the recent research frame- demonstrating a consistent inverse association of the work based on the A/T/(N) system [5], and our findings plasma Aβ42/Aβ40 with the disease [10–13, 42, 43]. support the suitability of a biological definition of AD, Nevertheless, there still exists considerable overlap of measured by biomarkers that are reflecting different pro- plasma Aβ ratio between diagnostic groups or amyloid cesses underlying AD clinical manifestations. status groups, which limits the use of Aβ plasma ratios Larger and varied populations need to be explored in for clinical diagnosis at the individual level. On the other order to confirm the potential utility of FDG-PET and hand, clinical performance of plasma biomarkers is also TP42/40 combination, which may vary depending on hampered by the fact that Aβ-PET classification, used as population characteristics and disease stage. In fact, the the gold standard, neither is utterly precise. In line with overall improvement of the adjusted TP42/40 model by this, plasma TP42/40 false-positive results could corres- AD-Conv score, although relevant, was relatively modest pond to a-MCI subjects who are already accumulating in our cohort. Another weakness of our study is the ab- Aβ in the brain, but have not reached the threshold for sence of a PiB-PET follow-up, which would be of great abnormal status yet. If this were the case, plasma Aβ utility in ascertaining the potential prognostic value of biomarkers would turn out to be abnormal before Aβ- plasma biomarkers to predict the rate of brain Aβ de- PET does. Alternatively, misclassification could be due position and could have provided relevant evidence for to heterogeneity of the underlying disease in a-MCI elucidating the possible source of bias in the classifica- patients. When using only CN to explore TP42/40 po- tion performance. tential in detecting Aβ-PET+ subjects, results are better compared to the a-MCI group or the mixed population. Conclusions Adjusted TP42/40 can accurately detect the 4 AB-PET+ This study demonstrates a consistent inverse association individuals in the cohort of 39 CN, which confirms the of plasma TP42/40 and FDG-PET biomarkers with Aβ- preclinical potential of this biomarker when detecting PET status. Findings from this work replicate our previous Aβ abnormality in the brain, detected also in previous results in different cohorts, hence confirming TP42/40 as work from our group [22]. However, due to the small a useful biomarker to rule out brain β-amyloidosis in pre- size of the CN and a-MCI groups in the present popula- clinical and early AD stages. Plasma TP42/40 tests may be tion, together with the different prevalence of Aβ-PET used as an initial screening tool for patient management; positivity, these findings should be taken cautiously and only those resulting TP42/40 positives would have to further replicated. undergo diagnostic confirmation by CSF analysis or PET AD-Conv score has proven a significant correlation neuroimaging. Integration of neurodegeneration and with Aβ-PET and a good performance in detecting brain blood-based Aβ biomarkers could contribute to early AD Aβ pathology within this cohort, being the AUC values diagnosis in compliance with the A/T/(N) research frame- obtained within the range previously defined [30]. These work, although it requires further validation. results reinforce the suitability of quantitative indices Abbreviations derived from automated FDG-PET measures replacing AD: Alzheimer’s disease; PET: Positron emission tomography; standard qualitative brain FDG-PET imaging [27–29]. CSF: Cerebrospinal fluid; CN: Cognitively normal; MCI: Mild cognitive Previous studies evaluating the association of FDG- impairment; Aβ-PET: Amyloid-beta PET; MS: Mass spectrometry; Aβ: Amyloid- beta; TP42/40: Total in plasma Aβ42/40; AIBL: Australian Imaging Biomarkers PET and Aβ-PET measures have shown discrepant and Lifestyle Study of Aging; FDG-PET: [18F]fluorodeoxyglucose PET; AD- results. Several works did not find a correlation in AD Conv score: AD conversion score (FDG-PET); a-MCI: Amnestic MCI; [44–46], MCI [47], and CN individuals [48, 49], although CRO: Contract research organization; ROC: Receiver operating characteristic; CI: Confidence interval; GLM: Generalized linear regression models; it was found in other cohorts [50–54]. Another study PPV: Positive predictive value; NPV: Negative predictive value; LR: Likelihood including both CN and MCI individuals showed a ratio Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 8 of 9 Acknowledgements Instituto de Salud Carlos III, Madrid, Spain. 7https://www.araclon.com/ The authors would like to thank the participants and collaborators of this alzheimer/abtest/current-situation/. study and the entire AB255 Study Group. Other AB255 Study Group collaborators: Miguel Goñi7; Francesc Pujadas8; Received: 11 July 2019 Accepted: 22 October 2019 Alberto Villarejo9; Ana Frank10; Jordi Peña-Casanova11; Manuel Fernández12; Gerard Piñol13; Rafael Blesa14; Pedro Gil15; Luis F. Pascual16; Miquel Aguilar17; Giovanni B Frisoni18; Jorge Matias-Guiu15; Niels Andreasen9; Carmen Antúnez20. References 7. Hospital Divino Vallés, Burgos, Spain 1. Buchhave P, Minthon L, Zetterberg H. Cerebrospinal fluid levels of β- 8. Hospital Universitari Vall d’Hebrón, Barcelona, Spain amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before 9. Hospital Doce de Octubre, Madrid, Spain the onset of Alzheimer dementia. Arch Gen Psychiatry. 2012;69(1):9. 10. Hospital La Paz, Madrid, Spain 2. Jack CR, Holtzman DM. Biomarker modeling of Alzheimer’s disease. Neuron. 11. Hospital del Mar, Barcelona, Spain 2013;80(6):1347–58. 12. CAE Oroitu Algorta, Vizcaya, Spain 3. Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, Andrieu S, et al. 13. Hospital Santa María de Lleida, Lleida, Spain Preclinical Alzheimer’s disease: definition, natural history, and diagnostic 14. Hospital de la Santa Creu i Sant Pau, Barcelona, Spain criteria. Alzheimers Dement. 2016;12(3):292–323. 15. Hospital Clínico San Carlos, Madrid, Spain 4. Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, et al. 16. Hospital Lozano Blesa, Zaragoza, Spain Toward defining the preclinical stages of Alzheimer’s disease: 17. Hospital Universitari Mútua recommendations from the National Institute on Aging-Alzheimer’s Terrassa, Terrassa, Spain Association workgroups on diagnostic guidelines for Alzheimer’s disease. 18. IRCCS Centro San Giovanni di Dio FBF, Brescia, Italy Alzheimers Dement. 2011;7(3):280–92. 19.Karolinska Institutet, Stockholm, Sweden 5. Jack CR, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. 20. Hospital Virgen de la Arrixaca, Fundación Alzheimur, Murcia, Spain NIA-AA research framework: toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535–62. 6. Bateman RJ, Blennow K, Doody R, Hendrix S, Lovestone S, Salloway S. Authors’ contributions Plasma biomarkers of AD emerging as essential tools for drug development: VPG and PP carried out the data analysis and interpretation and prepared an EU/US CTAD task force report. J Prev Alzheimer’s Dis JPAD. 2019;6(3): and wrote the manuscript. JR performed Aβ quantification. LS contributed to 169–73. the APOE analysis. IM and ISJ contributed to the study design, sample 7. Hansson O, Zetterberg H, Vanmechelen E, Vanderstichele H, Andreasson U, collection, and processing. JA contributed to the study design, recruitment, Londos E, et al. Evaluation of plasma Aβ40 and Aβ42 as predictors of and FDG-PET acquisition and data analysis. IH, MA, and MB participated in conversion to Alzheimer’s disease in patients with mild cognitive the data collection and analysis. PML, JM, LT, AR, MB, and MS conceived and impairment. Neurobiol Aging. 2010;31(3):357–67. designed the study. All authors reviewed the manuscript. All authors read 8. Lopez OL, Kuller LH, Mehta PD, Becker JT, Gach HM, Sweet RA, et al. Plasma and approved the final manuscript. amyloid levels and the risk of AD in normal subjects in the Cardiovascular Health Study. Neurology. 2008;70(19):1664–71. Funding 9. Lövheim H, Elgh F, Johansson A, Zetterberg H, Blennow K, Hallmans G, et al. This work was funded by Araclon Biotech, Fundació ACE Memory Clinic, as well Plasma concentrations of free amyloid β cannot predict the development as the Spanish Ministry of Health through Instituto de Salud Carlos III (Madrid) of Alzheimer’s disease. Alzheimers Dement. 2017;13(7):778–82. (FISS PI10/00954) and by Agència d’Avaluació de Tecnologia i Recerca Mèdiques, 10. Graff-Radford NR, Crook JE, Lucas J, Boeve BF, Knopman DS, Ivnik RJ, Departament de Salut de la Generalitat de Catalunya (RECERCALIA grant 390/06/ et al. Association of low plasma Aβ42/Aβ40 ratios with increased 2009). Araclon Biotech was the main funder of the study, participating in imminent risk for mild cognitive impairment and Alzheimer disease. conceptualization, design, data collection, and manuscript preparation. Arch Neurol. 2007;64:9. 11. Lambert J-C, Schraen-Maschke S, Richard F, Fievet N, Rouaud O, Berr C, et al. Availability of data and materials Association of plasma amyloid with risk of dementia: the prospective Three- The datasets used and/or analyzed during the current study are available City Study. Neurology. 2009;73(11):847–53. from the corresponding author on reasonable request. 12. Chouraki V, Beiser A, Younkin L, Preis SR, Weinstein G, Hansson O, et al. Plasma amyloid-β and risk of Alzheimer’s disease in the Framingham Heart Study. Alzheimers Dement. 2015;11(3):249–257.e1. Ethics approval and consent to participate 13. van Oijen M, Hofman A, Soares HD, Koudstaal PJ, Breteler MM. Plasma Aβ1– The study was approved by the CEIC ethic committee 2009/5455, and all 40 and Aβ1–42 and the risk of dementia: a prospective case-cohort study. participants provided written informed consent prior to inclusion. Patient Lancet Neurol. 2006;5(8):655–60. recruitment and collection protocols were in accordance with ethical 14. Lui JK, Laws SM, Li Q-X, Villemagne VL, Ames D, Brown B, et al. Plasma standards according to WMA Declaration of Helsinki-Ethical Principles for amyloid-beta as a biomarker in Alzheimer’s disease: the AIBL study of aging. Medical Research Involving Human Subjects. J Alzheimers Dis. 2010;20(4):1233–42. 15. Devanand DP, Schupf N, Stern Y, Parsey R, Pelton GH, Mehta P, et al. Plasma Consent for publication A and PET PiB binding are inversely related in mild cognitive impairment. Consent for publication was provided by every participant of the study. Neurology. 2011;77(2):125–31. 16. Rembach A, Faux NG, Watt AD, Pertile KK, Rumble RL, Trounson BO, et al. Competing interests Changes in plasma amyloid beta in a longitudinal study of aging and VPG, PP, JR, ISJ, LS, IM, and MS were employees of Araclon Biotech at the Alzheimer’s disease. Alzheimers Dement. 2014;10(1):53–61. moment the study and data analysis were carried out. All other co-authors 17. Nakamura A, Kaneko N, Villemagne VL, Kato T, Doecke J, Doré V, et al. High declare that they have no competing interests. performance plasma amyloid-β biomarkers for Alzheimer’s disease. Nature. 2018;554(7691):249–54. Author details 18. Ovod V, Ramsey KN, Mawuenyega KG, Bollinger JG, Hicks T, Schneider T, 1 Araclon Biotech S.L., Vía Hispanidad 21, 50009 Zaragoza, Spain. 2Servicio de et al. Amyloid β concentrations and stable isotope labeling kinetics of Medicina Nuclear, Clínica Universidad de Navarra, Pamplona, Spain. 3Center human plasma specific to central nervous system amyloidosis. Alzheimers for Research and Advanced Therapies and Memory Clinic, Fundación Dement. 2017;13(8):841–9. CITA-Alzheimer, San Sebastián, Spain. 4Institut de recerca Sant Joan de Déu, 19. Janelidze S, Stomrud E, Palmqvist S, Zetterberg H, van Westen D, Jeromin A, Hospital Infantil Sant Joan de Déu, Barcelona, Spain. 5Research Center and et al. Plasma β-amyloid in Alzheimer’s disease and vascular disease. Sci Rep. Memory Clinic, Fundació ACE, Institut Català de Neurociències Aplicades, 2016;6(1):26801. Universitat Internacional de Catalunya-Barcelona, Barcelona, Spain. 20. Pérez-Grijalba V, Fandos N, Canudas J, Insua D, Casabona D, Lacosta AM, 6 Networking Research Center on Neurodegenerative Diseases (CIBERNED), et al. Validation of immunoassay-based tools for the comprehensive Pérez-Grijalba et al. Alzheimer's Research & Therapy (2019) 11:96 Page 9 of 9 quantification of Aβ40 and Aβ42 peptides in plasma. J Alzheimers Dis. 2016; 43. Swaminathan S, Risacher SL, Yoder KK, West JD, Shen L, Kim S, et al. 54(2):751–62. Association of plasma and cortical amyloid beta is modulated by APOE ε4 21. Pérez-Grijalba V, Romero J, Pesini P, Sarasa L, Monleón I, San-José I, et al. status. Alzheimers Dement. 2014;10(1):e9–18. Plasma Aβ42/40 ratio detects early stages of Alzheimer’s disease and 44. Furst AJ, Rabinovici GD, Rostomian AH, Steed T, Alkalay A, Racine C, et al. correlates with CSF and neuroimaging biomarkers in the AB255 study. Cognition, glucose metabolism and amyloid burden in Alzheimer’s disease. J Prev Alzheimers Dis. 2019;6(1):34–41. Neurobiol Aging. 2012;33(2):215–25. 22. Fandos N, Pérez-Grijalba V, Pesini P, Olmos S, Bossa M, Villemagne VL, et al. 45. La Joie R, Perrotin A, Barré L, Hommet C, Mézenge F, Ibazizene M, et al. Plasma amyloid β 42/40 ratios as biomarkers for amyloid β cerebral Region-specific hierarchy between atrophy, hypometabolism, and β-amyloid deposition in cognitively normal individuals. Alzheimers Dement Diagn (Aβ) load in Alzheimer’s disease dementia. J Neurosci. 2012;32(46):16265–73. Assess Dis Monit. 2017;8:179–87. 46. Tauber C, Beaufils E, Hommet C, Ribeiro MJ, Vercouillie J, Vierron E, et al. 23. de Rojas I, Romero J, Rodríguez-Gomez O, Pesini P, Sanabria A, Pérez- Brain [18F]FDDNP binding and glucose metabolism in advanced elderly Cordon A, et al. Correlations between plasma and PET beta-amyloid levels healthy subjects and Alzheimer’s disease patients. J Alzheimers Dis. 2013; in individuals with subjective cognitive decline: the Fundació ACE Healthy 36(2):311–20. Brain Initiative (FACEHBI). Alzheimers Res Ther. 2018;10(1):119. 47. Hatashita S, Yamasaki H, Suzuki Y, Tanaka K, Wakebe D, Hayakawa H. 24. Doecke J, Pérez-Grijalba V, Fandos N, Fowler C, Villemagne VL, Masters CL, [18F]Flutemetamol amyloid-beta PET imaging compared with [11C]PIB et al. Total Aβ42/Aβ40 ratio in plasma predict amyloid-PET status, across the spectrum of Alzheimer’s disease. Eur J Nucl Med Mol Imaging. independent of clinical AD diagnosis. Neurology. 2019; in press. 2014;41(2):290–300. 25. Risacher SL, Fandos N, Romero J, Sherriff I, Pesini P, Saykin AJ, et al. Plasma 48. Wirth M, Oh H, Mormino EC, Markley C, Landau SM, Jagust WJ. The effect of Aβ levels are associated with cerebral amyloid and tau deposition. amyloid β on cognitive decline is modulated by neural integrity in Alzheimers Dementia. 2019;11:510–9. cognitively normal elderly. Alzheimers Dement J Alzheimers Assoc. 2013; 26. Jagust W, Reed B, Mungas D, Ellis W, Decarli C. What does 9(6):687–698.e1. fluorodeoxyglucose PET imaging add to a clinical diagnosis of dementia? 49. Mosconi L, Andrews RD, Matthews DC. Comparing brain amyloid Neurology. 2007;69(9):871–7. deposition, glucose metabolism, and atrophy in mild cognitive impairment 27. Shaffer JL, Petrella JR, Sheldon FC, Choudhury KR, Calhoun VD, Coleman RE, with and without a family history of dementia. J Alzheimers Dis. 2013;35(3): et al. Predicting cognitive decline in subjects at risk for Alzheimer disease 509–24. by using combined cerebrospinal fluid, MR imaging, and PET biomarkers. 50. Lowe VJ, Weigand SD, Senjem ML, Vemuri P, Jordan L, Kantarci K, et al. Radiology. 2013;266(2):583–91. Association of hypometabolism and amyloid levels in aging, normal 28. Landau SM, Harvey D, Madison CM, Koeppe RA, Reiman EM, Foster NL, et al. subjects. Neurology. 2014;82(22):1959–67. Associations between cognitive, functional, and FDG-PET measures of 51. Kadir A, Almkvist O, Forsberg A, Wall A, Engler H, Långström B, et al. decline in AD and MCI. Neurobiol Aging. 2011;32(7):1207–18. Dynamic changes in PET amyloid and FDG imaging at different stages of 29. Chen K, Ayutyanont N, Langbaum JBS, Fleisher AS, Reschke C, Lee W, et al. Alzheimer’s disease. Neurobiol Aging. 2012;33(1):198.e1–14. Characterizing Alzheimer’s disease using a hypometabolic convergence 52. Frings L, Spehl TS, Weber WA, Hüll M, Meyer PT. Amyloid-β load predicts index. NeuroImage. 2011;56(1):52–60. medial temporal lobe dysfunction in Alzheimer dementia. J Nucl Med Off 30. Arbizu J, Prieto E, Martínez-Lage P, Martí-Climent JM, García-Granero M, Publ Soc Nucl Med. 2013;54(11):1909–14. Lamet I, et al. Automated analysis of FDG PET as a tool for single-subject 53. Edison P, Archer HA, Hinz R, Hammers A, Pavese N, Tai YF, et al. Amyloid, probabilistic prediction and detection of Alzheimer’s disease dementia. Eur J hypometabolism, and cognition in Alzheimer disease: an [11C]PIB and Nucl Med Mol Imaging. 2013;40(9):1394–405. [18F]FDG PET study. Neurology. 2007;68(7):501–8. 31. Espinosa A, Alegret M, Pesini P, Valero S, Lafuente A, Buendía M, et al. 54. Zhou Y, Yu F, Duong TQ. Alzheimer’s Disease Neuroimaging Initiative. White Cognitive composites domain scores related to neuroimaging biomarkers matter lesion load is associated with resting state functional MRI activity within probable-amnestic mild cognitive impairment-storage subtype. J and amyloid PET but not FDG in mild cognitive impairment and early Alzheimers Dis. 2017;57(2):447–59. Alzheimer’s disease patients. J Magn Reson Imaging. 2015;41(1):102–9. 32. Alegret M, Espinosa A, Valero S, Vinyes-Junqué G, Ruiz A, Hernández I, et al. 55. Drzezga A, Becker JA, Van Dijk KRA, Sreenivasan A, Talukdar T, Sullivan C, Cut-off scores of a brief neuropsychological battery (NBACE) for Spanish et al. Neuronal dysfunction and disconnection of cortical hubs in non- individual adults older than 44 years old. PLoS One. 2013;8(10):e76436. demented subjects with elevated amyloid burden. Brain J Neurol. 2011; 33. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical 134(Pt 6):1635–46. method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–98. Publisher’s Note 34. Blesa R, Pujol M, Aguilar M, Santacruz P, Bertran-Serra I, Hernández G, et al. Springer Nature remains neutral with regard to jurisdictional claims in Clinical validity of the “mini-mental state” for Spanish speaking published maps and institutional affiliations. communities. Neuropsychologia. 2001;39(11):1150–7. 35. Wechsler D. WAIS–III. Wechsler adult intelligence scale–third edition. Technical manual. San Antonio: The Psychological Corporation; 1997. 36. Alegret M, Espinosa A, Vinyes-Junqué G, Valero S, Hernández I, Tárraga L, et al. Normative data of a brief neuropsychological battery for Spanish individuals older than 49. J Clin Exp Neuropsychol. 2012;34(2):209–19. 37. Buschke H. Cued recall in amnesia. J Clin Neuropsychol. 1984;6(4):433–40. 38. Boada M, Tárraga L, Hernández I, Valero S, Alegret M, Ruiz A, et al. Design of a comprehensive Alzheimer’s disease clinic and research center in Spain to meet critical patient and family needs. Alzheimers Dement J Alzheimers Assoc. 2014;10(3):409–15. 39. Nordberg A, Carter SF, Rinne J, Drzezga A, Brooks DJ, Vandenberghe R, et al. A European multicentre PET study of fibrillar amyloid in Alzheimer’s disease. Eur J Nucl Med Mol Imaging. 2013;40(1):104–14. 40. Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31(3):545–8. 41. Insel PS, Mattsson N, Mackin RS, Schöll M, Nosheny RL, Tosun D, et al. Accelerating rates of cognitive decline and imaging markers associated with β-amyloid pathology. Neurology. 2016;86(20):1887–96. 42. Schupf N, Tang MX, Fukuyama H, Manly J, Andrews H, Mehta P, et al. Peripheral A subspecies as risk biomarkers of Alzheimer’s disease. Proc Natl Acad Sci. 2008;105(37):14052–7.