We performed a longitudinal investigation in 316 SMC participants of the French monocentric “INveStIGation of AlzHeimer’s PredicTors in Subjective Memory Complainers” (INSIGHT-preAD) cohort (Pitié-Salpêtrière University Hospital, Paris) [19]. All participants underwent two Aβ-positron emission tomography (Aβ-PET) scans, at baseline and at 2-year follow-up. A subset of 40 individuals received a baseline lumbar puncture. The concentrations of plasma NFL and t-Tau were measured at three time points using an ultrasensitive technology (N = 79 at baseline, 1-year, and 3-year follow-up).
Study participants
We designed a large-scale mono-centric research program using a cohort of SMC recruited from the INSIGHT-preAD study, a French academic university-based cohort [19] which is part of the Alzheimer Precision Medicine Initiative (APMI) and its Cohort Program (APMI-CP) [20,21,22,23]. Participants were enrolled between May 25, 2013, and January 20, 2015, at the Institute of Memory and Alzheimer’s disease (Institut de la Mémoire et de la Maladie d’Alzheimer, IM2A) at the Pitié-Salpêtrière University Hospital in Paris, France [19]. The study was conducted in accordance with the tenets of the Declaration of Helsinki of 1975 and approved by the local Institutional Review Board at the participating center. All participants gave written informed consent for use of their clinical data for research purposes.
PET data acquisition and processing
All Florbetapir-PET scans were acquired in a single session on a Philips Gemini GXL CT-PET scanner 50 (± 5) min after injection of approximately 370 MBq (333–407 MBq) of Florbetapir. Images acquisition, reconstruction including correction algorithms and reallination, averaging, and quality check were performed by the CATI team (Centre d’Acquisition et Traitement des Images) (http://cati-neuroimaging.com) [19, 24, 25] and were calculated for each of 12 cortical regions of interest (right and left posterior cingulate, right and left anterior cingulate, right and left superior frontal, right and left inferior parietal, right and left precuneus, right and left middle temporal cortices), as well as the global average standard uptake value ratio (SUVR).
CSF sampling and biomarkers assessment
CSF sampling was performed by lumbar puncture in a subsample of 40 individuals. All CSF samples were collected in polypropylene tubes, centrifuged at 1000 g for 10 min at + 4 °C. The collected supernatant was stored at − 80 °C for pending biochemical analysis. The immunoassays for CSF core biomarkers are reported in previous studies [26, 27].
Blood sampling and collection tube storage
Ten (10) mL of venous blood were collected in one BD Vacutainer® spray-coated K2 tube, which was employed for all subsequent immunological analyses. Blood samples were taken in the morning, after a 12-h fast, handled in a standardized way, and centrifuged for 15 min at 2000 G-force at + 4 °C. Per sample, plasma fraction was collected, homogenized, aliquoted into multiple 0.5 mL cryovial-sterilized tubes, and finally stored at − 80°Cwithin 2 h from collection.
Immunoassays for plasma biomarkers
All analyses of plasma t-Tau and NFL concentrations were performed at the Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Sweden [28,29,30]. In particular, a volume of 0.5 mL of plasma for each individual was required for performing the analyses.
Plasma t-Tau concentrations were measured using the Human Total Tau 2.0 kit on the ultrasensitive single molecule arrays (Simoa) platform (Quanterix, Lexington, MA), according to the manufacturer instructions.
For plasma t-Tau, both repeatability and intermediate precision were 12.2% for an internal QC plasma sample with a concentration of 1.9 pg/mL [2, 30]. The t-Tau assay was originally developed in a collaboration between the Clinical Neurochemistry Laboratory and Quanterix [31]. However, the commercially available assay was built on this work but with another set of antibodies.
Plasma NFL concentrations were measured using the ultrasensitive Simoa technology, according to the manufacturer instructions. Repeatability was 9.6% and 10.6% and intermediate precision was 14.6% and 11.6% for two internal QC plasma samples with concentrations of 12.9 pg/mL and 107 pg/mL [28,29,30]. The NFL assay was originally developed by the Clinical Neurochemistry Laboratory and then commercialized by Quanterix [32].
All samples were analyzed on one occasion using one batch of reagents by board-certified laboratory technicians who were blinded to clinical data.
Statistical analysis
Statistical analysis was performed using IBM-SPSS® Statistics, Version 20, and Addinsoft, XLSTAT Statistical and Data Analysis Solution (2019), Long Island, NY, USA, statistical packages for Mac OS X.
Kolmogorov-Smirnov test was applied to check for normality. Gaussian distributed values were expressed as mean and standard deviation (SD) or standard error (SE); otherwise, median and interquartile range (IR) were used for quantitative variables, while categorical data were expressed as frequency.
To evaluate the impact of age, sex, and APOE ε4 carrier status on NFL and t-Tau evolution in a 3-year follow-up, respectively, we conducted two independent linear mixed-effects models (LMM) on the whole sample [33]. Age, sex, and APOE ε4 carrier status were included as fixed effects in the model and each individual as random effect. We also included interaction between age and sex, age and APOE ε4 carrier status, and sex and APOE ε4 carrier status. Type III likelihood ratio tests were used to test each fixed effect and interaction. The statistical models used were verified for normal distribution of residuals, random effects, and homoscedasticity of residuals. Subsequently, the analysis was repeated after stratifying the sample into amyloid-PET-negative and positive individuals.
On the whole sample, cross-sectional data were also elaborated to identify the variables maximally contributing to group separation of individuals, based on the following outcome variables Y: baseline plasma NFL and baseline plasma t-Tau. To this end, two partial least square (PLS) models were generated, and the variables with Variable Importance in Projection (VIPs, expressing a measure of a variable’s relevance in the model) greater than 1.50 were considered significant for separation of the sample [34, 35]. In the first model, Y was defined by placing NFL = 1 whether NFL concentrations were above or below the third quartile, and 0 if below. Input variables were sex, age, APOE ε4 carrier status, arterial hypertension, (HTA), atrial fibrillation, heart disease, dyslipidemia, diabetes, obstructive sleep apnea syndrome (OSAS), head trauma, mood disorders, vitamin B12 deficiency, body mass index (BMI), global SUVR, MMSE and FCSRT at baseline, and baseline plasma t-Tau. In the second model, input variables were the same, as previously described, but the outcome variable Y was t-Tau, categorized as 1 if t-Tau concentrations were upper than the third quartile, and 0 otherwise.
The associations of the two plasma biomarkers at baseline with baseline global and regional brain Aβ deposition, Δ Aβ deposition, baseline MMSE score, FCSRT, the corresponding CSF biomarkers, and between each other were evaluated by Spearman correlation testing. If the relationship was significant, a subsequent stepwise forward regression was performed. Spearman test was used to explore the association of Δ NFL and Δ t-Tau with Δ MMSE and Δ FCSRT (cognitive measures). Δ value has been defined as the difference between the baseline and the 3-year follow-up value, excluding for Aβ deposition where we considered 2-year follow-up value, because we had the baseline data available and after 2 years.
Variables with skewed distribution were log-transformed for use in all parametric analyses. P values < 0.05 were considered significant in all statistical elaboration.