De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 https://doi.org/10.1186/s13195-021-00899-0 RESEARCH Open Access Cerebrospinal fluid levels of the neurotrophic factor neuroleukin are increased in early Alzheimer’s disease, but not in cerebral amyloid angiopathy Anna M. De Kort1 , H. Bea Kuiperij1, Daniel Alcolea2,3, Iris Kersten1, Alexandra A. M. Versleijen4, Steven M. Greenberg5, Erik Stoops6, Floris H. B. M. Schreuder1, Catharina J. M. Klijn1, Alberto Lleó2,3, Jurgen A. H. R. Claassen7 and Marcel M. Verbeek1,4* Abstract Background: Neuroleukin (NLK) is a protein with neurotrophic properties and is present in a proportion of senile plaques and amyloid laden vessels. It has been suggested that NLK is part of a neuroprotective response to amyloid β-induced cell death. The aim of our study was to investigate the value of cerebrospinal fluid (CSF) NLK levels as a biomarker of vascular amyloid deposition in patients with cerebral amyloid angiopathy (CAA) and in patients with amnestic mild cognitive impairment (aMCI) and Alzheimer’s disease (AD). Methods: CSF NLK levels were quantified by ELISA in CAA patients (n = 25) and controls (n = 27) and in two independent samples of aMCI patients, AD patients, and controls: (1) From the Radboud University Medical Center (Nijmegen), we included n = 19 aMCI patients, n = 40 AD patients, and n = 32 controls. (2) From the Hospital of Sant Pau (Barcelona), we included n = 33 aMCI patients, n = 17 AD patients, and n = 50 controls. Results: CSF NLK levels were similar in CAA patients and controls (p = 0.95). However, we found an elevated CSF concentration of NLK in aMCI (p < 0.0001) and AD patients (p < 0.0001) compared to controls in both samples sets. In addition, we found a correlation of CSF NLK with CSF YKL-40 (age-adjusted-spearman-rank-coefficient = 0.82, p < 0.0001) in aMCI/AD patients, a well-known glial marker of neuro-inflammation. Conclusions: We found that CSF NLK levels are elevated in aMCI and AD patients compared to controls, but are not increased in CAA patients. CSF NLK levels may be related to an increased neuroinflammatory state in early stages of AD, given its association with YKL-40. Keywords: Cerebrospinal fluid, Biomarkers, Alzheimer’s disease, Amyloid, Cerebral amyloid angiopathy, Neuroleukin, Neuro-inflammation * Correspondence: marcel.verbeek@radboudumc.nl 1 Department of Neurology, Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Radboud Alzheimer Centre, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands 4 Department of Laboratory Medicine, Radboud University Medical Center, Nijmegen, The Netherlands Full list of author information is available at the end of the article © The Author(s). 2021, corrected publication 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/. 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 in a credit line to the data. De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 2 of 10 Introduction CSF YKL-40 (also known as chitinase-3 like-protein 1 Alzheimer’s disease (AD) is the leading cause of demen- or cartilage glycoprotein-39) is an established neuroin- tia worldwide. It is neuropathologically characterized by flammatory marker [14]. YKL-40 is mainly expressed by the intracellular accumulation of hyperphosphorylated astrocytes, but also by microglia, often in close vicinity tau protein in neurofibrillary tangles and by the extracel- to senile plaques and neurofibrillary tangles [15–17]. In- lular accumulation of amyloid-beta protein (Aβ). In AD, creased CSF YKL-40 levels have been reported in cogni- Aβ is found in parenchymal senile plaques, and in many tively healthy APOE-ε4 carriers, amnestic mild cognitive cases also in the cerebral vasculature, referred to as cere- impairment (aMCI), and AD patients [14] and also in bral amyloid angiopathy (CAA) [1]. The clinical diagno- frontotemporal dementia [18] and amyotrophic lateral sis of AD is based on a gradual decline of cognition, sclerosis [19]. most commonly with an impairment in encoding and re- The aim of our study was to examine CSF NLK levels call of recently learned information as well as evidence in patients with CAA and in two independent samples of cognitive dysfunction in at least one other cognitive of patients with aMCI or AD diagnosed according to domain [2]. Mild cognitive impairment (MCI) refers to clinical criteria. We also studied the association of CSF an early, but abnormal state of cognitive impairment NLK in aMCI and AD patients with a positive biomarker defined by memory complaints, objective memory status A+T+(N+), and we performed an exploratory impairment, essentially preserved general cognitive study on the association of CSF NLK with CSF YKL-40, function, largely intact functional activities, but not de- and the known Alzheimer CSF biomarkers total tau, mented [3]. Cerebrospinal fluid (CSF) and neuroimaging phosphorylated tau, and Aβ42. biomarkers can be used to support the diagnosis [4] and have been proposed to define a prodromal stage of AD Methods based on the presence of MCI and evidence for amyloid Participants deposition (A), tau accumulation (T), and neurodegener- Sporadic CAA patients and controls ation (N) [4]. Beyond their supporting role in the We included 14 patients with sporadic CAA from the diagnosis, biomarkers may contribute to predicting Massachusetts General Hospital (MGH), Boston, USA, prognosis and may facilitate monitoring of disease and 11 patients with sporadic CAA and 27 controls from progression and treatment response. Furthermore, novel the Radboud University Medical Center, Nijmegen, the biomarkers may advance our comprehension of the Netherlands (RUMC). The inclusion criteria for the different aspects of AD pathophysiology next to tau and CAA patients were a diagnosis of probable or definite amyloid pathology. CAA according to the modified Boston criteria [7]. CAA is the most prevalent co-pathology of AD, with MMSE scores were available for a subset (n = 15) of almost 80% of the patients having some degree of CAA these patients (Table 1). Controls underwent a lumbar [5], likely because Aβ deposition is a shared feature of puncture as part of the diagnostic workup of neurologic senile plaques and CAA [6]. Vascular Aβ depositions symptoms or to exclude central nervous system involve- can also occur in the absence of AD and is recognized ment of a systemic disease at the RUMC. They neither had as a designated type of cerebral small vessel disease, re- the suspected neurological disease nor a neurodegenerative ferred to as sporadic CAA. Sporadic CAA is associated with cerebral (micro) hemorrhages and can also result in Table 1 Characteristics of the sporadic CAA patients and a decline in cognitive function [5]. Sporadic CAA is controls diagnosed during life using the MRI-based modified CAA (n = 25) Controls (n = 27) p-value Boston criteria [7], which rely on the hemorrhagic Age (years) 66 ± 10 63 ± 10 p = 0.32a manifestations of CAA. However, these are indirect signs Sex, M/F (n) 13/12 12/15 p = 0.59b of relatively late-stage manifestations of CAA and they do not provide definite proof of the disease. MMSE score 28 [26–29]d N.A. - Neuroleukin (NLK) may serve as biomarker for CAA or TP (mg/ml) 0.91 [0.85–0.96] 0.83 [0.78–1.03] p = 0.12c AD. In an immunohistochemical study in AD patients, Aβ40 (pg/ml) 3745 [2200–4694] 6312 [4888–7381] p < 0.0001c NLK expression was found in senile plaques and in Aβ- NLK (ng/ml) 3.39 [2.94–4.15] 3.31 [2.41–5.40] p = 0.95c laden vessels [8]. In addition, messenger RNA levels of Values are medians and [IQR] except for age (mean ± SD) and sex (n). TP, NLK, NLK were increased in cultured human brain pericytes and Aβ40 were measured in CSF after exposure to Aβ40 [8], which may represent an at- Abbreviations: Aβ40 amyloid beta-40, CAA cerebral amyloid angiopathy, F female, IQR interquartile range, M male, MMSE Mini-Mental State Examination, tempt to survive Aβ-induced cell death. Furthermore, it N.A. not available, NLK neuroleukin, SD standard deviation, TP total protein was shown that NLK can serve as a neurotrophic growth a Student’s t-test b Chi-square test factor [9] and may support survival of sensory and spinal c Mann-Whitney test neurons and stimulate axonal growth [10–13]. d n = 15 patients De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 3 of 10 disease, known cognitive impairment, sepsis, a recent stroke 32 controls from RUMC, fulfilling the criteria as men- (< 6 months), or a malignancy in the central nervous tioned above (“ Sporadic CAA patients and controls”). system. Barcelona aMCI/patients and controls For validation aMCI and AD patients and controls of our findings, we included 33 patients with aMCI, 17 We included two independent samples of patients with patients with probable AD, and 50 cognitively normal amnestic mild cognitive impairment (aMCI), patients controls from the SPIN cohort (Memory Unit at with AD, and controls: we performed an initial analysis Hospital Sant Pau, Barcelona). Inclusion criteria were a including participants from the RUMC (the “Nijmegen diagnosis of aMCI according to the Petersen criteria [3] aMCI/AD patients and controls”) and a confirmatory or probable AD according to the NINCDS-ADRDA cri- analysis in participants from the Memory Unit at teria [20]. Controls were spouses or children of patients, Hospital Sant Pau, Barcelona (the “Barcelona aMCI/AD and inclusion criteria were the absence of memory com- patients and controls”). plaints or significant impairment in other domains or daily living activities, a clinical dementia rating global Nijmegen aMCI/AD patients and controls We in- score of 0, an MMSE score of > 27, and a Free and Cued cluded 19 patients with aMCI, 40 patients with probable Selective Reminding Test (FCSRT) total immediate score AD, and 32 controls (different than the above- within normal range for age and education [23]. Mini mentioned n = 27), all from the RUMC. The inclusion Mental State Examination (MMSE) scores [22] were avail- criteria were either a clinical diagnosis of aMCI accord- able for all aMCI, AD patients, and controls (Table 2). ing to the Petersen criteria [3] at the moment of the We used previously published local cutoff values to lumbar puncture followed by a diagnosis of probable AD stratify the patients and controls for their “ATN status” according to NINCDS-ADRDA criteria [20] in a later (Aβ42 (A+), < 916 pg/ml; phosphorylated tau (T+), > 63 stage, or a diagnosis of probable AD according to the pg/ml; total tau (N+), > 456 pg/ml) [24]. same criteria. Diagnoses were established during multi- disciplinary memory clinic meetings. We used prede- CSF analysis fined local cutoff values of CSF analysis to stratify the All participants in this study underwent a lumbar patients and controls for their “ATN status” (Aβ42 (A+), puncture according to state-of-the-art local protocols. < 500 pg/ml; phosphorylated tau (T+), > 85 pg/ml; total At all locations, the CSF was collected in polypropylene tau (N+), > 350 pg/ml) [21]. Mini Mental State Examin- tubes, centrifuged, aliquoted, and stored in polypropyl- ation (MMSE) scores [22] were available for all aMCI and ene tubes at − 80 °C. For all CSF analyses, the technician AD patients (Table 2). These patients were compared with who performed the analysis was blinded for the clinical Table 2 Characteristics of Nijmegen aMCI/AD patients and controls aMCI (n = 19) AD (n = 40) Controls (n = 32) p-value Age (years) 72 [70–78] 74 [65–77] 61 [56–67] p < 0.0001a Sex, M/F (n) 7/12 11/29 15/17 p = 0.24b MMSE score 26 [23–27] 20 [17–24] N.A. p < 0.0001a APOE genotype ε4+: n = 10 ε4+: n = 0 N.A. p = 0.58b ε4−: n = 3 ε4−: n = 8 N.A.: n = 6 N.A.: n = 2 TP (mg/ml) 0.87 [0.79–0.93] 0.85 [0.76–0.96] 0.91 [0.79–.02] p = 0.51a T-tau (pg/ml) 504 [419–691] 627 [321–956] 267 [179–386] p < 0.0001a P-tau (pg/ml) 91.0 [71.0–114.0] 97.5 [79–129] 47.5 [32.3–56.8] p < 0.0001a Aβ42 (pg/ml) 459 [404-516] 468 [389–564] 864 [664–1142] p < 0.0001a NLK (ng/ml) 4.66 ± 1.40 4.84 ± 1.58 3.07 ± 1.2 p < 0.0001d GCA score n = 7; 2 [2–2] n = 28; 1.5 [1–2] n = 4; 0.5 [0–1] p = 0.006c Mean MTA n = 5; 1.5 [1–2.25] n = 28; 1 [1–2] n = 2; 0 [0–0] p = 0.125c Medians and [IQR] are reported for all characteristics and markers, except for NLK (mean ± SD) and sex (n). T-tau, P-tau, Aβ42, and NLK were measured in CSF Abbreviations: AD Alzheimer’s disease, Aβ42 amyloid beta-42, APOE Apolipoprotein E, aMCI amnestic mild cognitive impairment, F female, GCA global cortical atrophy, IQR interquartile range, M male, MMSE Mini-Mental State Examination, MTA medial temporal lobe atrophy, N.A. not available, NLK neuroleukin, P-tau phosphorylated tau, SD standard deviation, TP total protein, T-tau total tau a Kruskal-Wallis test b Chi-square test c Fisher’s exact test d ANOVA De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 4 of 10 diagnoses, and patient and controls samples were ran- reached by consultation of a skilled vascular geriatrician domly analyzed to overcome bias. (JC). NLK was quantified in CSF using the human glucose- 6-phosphate isomerase ELISA kit (cat no. Ab171575; Data analysis Abcam, Cambridge, UK) according to the manufac- We used the software programs IBM SPSS statistics for turers’ instructions [25]. A standard curve ranging from Windows, version 25.0 (Armonk, NY: IBM Corp) and 156 to 10,000 pg/ml was constructed using serial two- GraphPad Prism 5.03 (La Jolla, CA). If parameters had a fold dilutions. CSF samples were diluted six times prior Gaussian distribution, parameters were depicted as mean to analysis. ± standard deviation and group differences were ana- Total protein levels in CSF were determined using lyzed with a Student’s t-test or an ANOVA. Otherwise, PierceTM BCA protein assay kit (Thermo Fisher Scien- parameters were stated as medians with interquartile tific, Waltham, MA, USA). In all NLK and total protein ranges and differences were analyzed with a Mann- CSF analyses, five quality controls were included on each Whitney U test or a Kruskal-Wallis test. The Shapiro- plate in each test run to correct for any inconsistencies Wilk test was used to analyze the normality of the data. between plates. These controls consisted of pooled CSF Sex frequency was analyzed by a chi-square test. GCA samples that were stored in aliquots at −80 °C. For each and MTA were compared using Fisher’s exact test. analysis, a fresh aliquot was used. When comparing group differences of NLK levels, we Aβ42, tau phosphorylated at threonine 181 (p-tau), adjusted for age and sex by performing multiple regres- and total tau (t-tau) levels in CSF of Nijmegen aMCI/ sion analysis with patient group, age and sex as inde- AD patients and controls were analyzed using Innotest pendent variables. In a second model, we adjusted for ELISAs (Fujirebio, Gent, Belgium). Test characteristics age, sex, and CSF total protein levels. To explore poten- of these assays have been described previously [26]. In tial specific relationships between NLK and Alzheimer Barcelona aMCI/AD patients and their controls, these pathology, we performed a subgroup analysis of aMCI/ biomarkers were analyzed using automated versions of AD patients who were “AD biomarker positive” (i.e., A+ the same assays using a Lumipulse apparatus (Fujirebio, T+(N+)) compared with “AD biomarker-negative” (i.e., Gent, Belgium). A−T−(N−)) controls. We pooled aMCI and AD bio- Aβ40 levels in CSF of CAA patients and their controls marker positive individuals to increase the power of this were analyzed using an ELISA (Euroimmun, Lübeck, analysis. To determine the diagnostic accuracy of NLK Germany). APOE genotyping was performed as described for the distinction between aMCI patients and controls, earlier [27]. and between AD patients and controls, we determined CSF YKL-40 measurements were available for a subset the area under the curve (AUC) using a receiver operat- (n = 29 controls, n = 13 aMCI, n = 5 AD patients) of the ing characteristic curve (ROC) with 95% confidence Barcelona patients and controls. These were analyzed interval (CI). Spearman rank correlation (rSP) was used as described earlier (MicroVueTM, Quidel, San Diego, to evaluate correlation between age, total protein, and CA) [28]. MMSE score with NLK. We used partial correlation with age as a covariate to adjust for age when investigating MRI acquisition and analysis the association between t-tau, p-tau, Aβ42, YKL-40, MRIs were obtained in a clinical setting, and field GCA, MTA, and NLK. To further explore the associ- strength was either 1.5 or 3.0 T. MRI analysis included ation of NLK with YKL-40, we also performed this axial and coronal T1-weighted and T2-weighted se- analysis in the controls and the MCI/AD patients quences. MRI images were reviewed blinded to all clin- separately. ical and CSF data and scored for medial temporal lobe atrophy (MTA) [29] and global cortical atrophy (GCA) Ethical statement [30]. The MTA scale assesses the width of the choroidal Lumbar punctures and vena punctures were performed fissure and the temporal horn, and the height of the hip- after informed consent from the patients themselves or pocampal formation on a 5-grade rating scale [29]. Both from the patients’ legal representatives. The CSF of AD hemispheres were scored, and the mean score of the left patients, controls, and the majority of the CAA patients and right hemisphere was used. The GCA score is the from the RUMC (n = 7) originated from a clinical diag- mean score for cortical atrophy throughout the entire nostic work-up. Four CAA patients underwent a lumbar cerebrum and was rated by a 4-point scale [30]. Each puncture in the context of a cohort study on biomarkers MRI was rated independently by two trained raters (AK for CAA (CAVIA; no. 733050202). This study was and LR), and there was a good agreement between the approved by the Medical Ethics Committee Arnhem- two raters: a weighted Cohen’s κ of 0.69 for GCA and Nijmegen (file nos. 2016-3011 and 2014-1401), Ethical 0.73 for MTA. In case of a disagreement, consensus was Committee of IIB-Sant Pau (16/2013), and Partners De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 5 of 10 Human Research Committee of Boston (study ID PHRC CSF NLK levels in Barcelona aMCI/AD patients and #2006P000664). controls Sex distribution was similar, whereas aMCI (mean age Results 70 ± 3 year, 30% male), and AD patients (mean age 67 ± NLK assay for quantification in CSF 7 years, 47% male) were older (p < 0.0001 and p = 0.002) The lower limit of detection (LLOD) of the NLK ELISA than the control subjects (mean age 63 ± 4 years, 50% was 61 pg/ml. All CSF measures were above the LLOD. male; Table 3). When adjusted for age and sex, there The mean intra-assay coefficient of variation (CV) for CSF was a significant difference between aMCI patients and samples (duplicate measures) was 3.6 ± 3.6% and the controls (p = 0.018; Table 3 and Fig. 1D), and between mean inter-assay CV was 4.9 ± 2.9%. The variation be- AD patients and controls (p = 0.008). Additional adjust- tween the ELISA plates, based on quality control measure- ment for CSF total protein yielded similar results (p = ments, ranged from a factor 0.89 to 1.08, for which the 0.013 and p = 0.005, respectively). CSF measures were corrected accordingly, by recalculating We found a higher mean CSF NLK (4.44 ± 1.28 ng/ the results using the results of one plate as a reference. ml) in n = 40 aMCI/AD patients with positive AD biomarkers (A+T+(N+)) than in n = 41 controls with CSF NLK levels in sporadic CAA patients and controls normal AD biomarker status (A−T−(N−), 3.15 ± 0.91 Age and sex distribution of the CAA patients (age 66 ± ng/ml, p = 0.001; Fig. 1E), when adjusted for age and 10 years, 52% male) and controls (age 63 ± 10 years, 44% sex. This difference remained after additional adjustment male) was similar (Table 1). Patients were assigned to for total protein (p = 0.001). the following categories of certainty: definite CAA (aut- Again, NLK correlated with age (rSP = 0.43, p < 0.0001) opsy confirmed, n = 2), probable CAA with supporting and with total protein (rSP = 0.32, p < 0.0001). pathology (n = 6), and probable CAA (n = 17). CSF NLK levels, adjusted for age and sex, were similar (p = 0.40, ROC analysis see Table 1 and Fig. 1A) in CAA patients and controls. We found an AUC of 0.81 (95% CI 0.68–0.93; Fig. 2A) Additional adjustment for CSF total protein did not to discriminate aMCI patients from controls in the affect the results (p = 0.34). Nijmegen sample, and an AUC of 0.74 (95% CI 0.63– We observed an increase in CSF NLK with increasing age 0.85) in the Barcelona sample. (rSP = 0.37, p = 0.014), but we found no correlation with We found an AUC of 0.82 (95% CI 0.73–0.92; Fig. 2B) total protein (rSP = 0.008, p = 0.95). Furthermore, CSF NLK to discriminate AD patients from controls in the correlated with MMSE score (rSP = −0.76, p = 0.001). Nijmegen sample, and an AUC of 0.83 (95% CI 0.73– 0.93) in the Barcelona sample. CSF NLK levels in Nijmegen aMCI/AD patients and controls Sex distribution was similar, whereas aMCI (median age Correlation of CSF NLK with other CSF biomarkers and 72 [70–78] years, 37% male) and AD patients (median MRI markers age 74 [65–77] years, 28% male) were older (both p < CSF NLK correlated with CSF YKL-40 (Spearman rank 0.0001) than the control subjects (median age 61 [56– coefficient, adjusted for age (rSPc) = 0.34, p < 0.019) in 67] years, 47% male; Table 2). CSF NLK levels were, the combined Barcelona aMCI/AD patients and controls adjusted for age and sex, increased in aMCI patients (Fig. 3). This correlation was prominent in the aMCI/ compared to controls (p = 0.009; Table 2 and Fig. 1B) AD patients (rSPc = 0.82, p < 0.0001), but not in the con- and in AD patients compared to controls (p < 0.0001). trols (rSPc = −0.14, p = 0.47). Additional adjustment for CSF total protein did not CSF NLK levels correlated with CSF t-tau levels affect the results (p = 0.005 and p < 0.0001, (Nijmegen aMCI/AD patients and controls: rSPc = 0.65, respectively). p < 0.0001; Barcelona aMCI/AD patients and controls Furthermore, in n = 28 aMCI/AD patients with rSPc = 0.60, p < 0.0001) and CSF p-tau levels (Nijmegen positive AD biomarkers (A+T+(N+)) the mean CSF aMCI/AD patients and controls: rSPc = 0.65, p < 0.0001; NLK concentration (5.21 ± 1.08 ng/ml) was higher than Barcelona aMCI/AD patients and controls: rSPc = 0.58, p in n = 20 controls with normal AD biomarker status (A < 0.0001). CSF NLK correlated with Aβ42 levels in the −T−(N−); 2.72 ± 1.0 ng/ml; when adjusted for age and Nijmegen aMCI/AD patients and controls (rSPc = −0.29, sex (p < 0.0001; Fig. 1C), and after additional adjustment p = 0.005), but not in the Barcelona aMCI/AD patients for total protein (p < 0.0001). and controls (rSPc = −0.014, p = 0.89). CSF NLK significantly correlated with age (rSP = 0.39, There was neither a correlation between CSF NLK and p < 0.0001) and with CSF total protein levels (rSP = 0.22, GCA score (rSPc = 0.23, p = 0.17), nor with mean MTA p = 0.039). score (rSPc = −0.036, p = 0.84). De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 6 of 10 Fig. 1 CSF NLK levels in the sporadic CAA patients and controls (A) and in the aMCI/AD patients and controls (B–E), with and without stratification for AT(N) status. Box and scatterplots in all panels (depicting median, interquartile range and range). A NLK levels in CAA patients and controls. The NLK levels were similar (p = 0.95). B CSF NLK levels in aMCI patients, AD patients, and controls from Nijmegen. These levels were increased in aMCI (p = 0.001) and AD patients (p < 0.0001) compared to controls. C NLK levels in A+T+(N+) aMCI/AD patients and A−T−(N −) controls from Nijmegen. NLK levels were significantly higher in A+T+(N+) aMCI/AD patients (p < 0.0001). D CSF NLK levels in aMCI patients, AD patients, and controls from Barcelona. Levels were significantly higher in aMCI patients (p < 0.0001) and AD patients (p < 0.0001) compared to controls. E NLK levels in A+T+(N+) aMCI/AD patients and A−T−(N−) controls from Barcelona. NLK levels were significantly higher in A+T+(N+) aMCI/AD patients (p < 0.0001). Abbreviations: AD, Alzheimer’s disease; CAA, cerebral amyloid angiopathy; CSF, cerebrospinal fluid; aMCI, amnestic mild cognitive impairment; ns, non-significant. **p < 0.01, ***p < 0.001 Discussion positive A+T+(N+)” aMCI and AD patients compared The main findings of our study are as follows: (1) to A−T−(N−) controls. (3) CSF NLK levels were asso- CSF NLK levels were increased in aMCI patients and ciated with CSF YKL-40 levels in aMCI/AD patients, in AD patients, as established in two independent which suggests an association of NLK with neuroin- samples from two different centers. (2) The increase flammation. (4) CSF NLK levels were not increased in in CSF NLK was particularly strong in “AD biomarker sporadic CAA patients versus controls. De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 7 of 10 Table 3 Characteristics of the Barcelona patients and controls aMCI AD Controls p-value n = 33 n = 17 n = 50 Age (years) 70 ± 3 67 ± 7 63 ± 4 p < 0.0001a Sex, M/F (n) 10/23 8/9 25/25 p = 0.20b MMSE score 26 [25–28] 20 [19–25] 29 [29–30] p < 0.000d APOE genotype ε4+: n = 24 ε4+: n = 6 ε4+: n = 10 p < 0.0001c ε4−: n = 8 ε4−: n = 10 ε4−: n = 38 N.A.: n = 1 N.A.: n = 1 N.A.: n = 2 TP (mg/ml) 0.78 [0.72–0.88] 0.79 [0.71–0.88] 0.78 [0.69–1.04] p = 0.99d T-tau (pg/ml) 647 [411–823] n = 15; 742 [598–1163] 265.0 [227–305] p < 0.0001d P-tau (pg/ml) 101.3 [78.0–143.8] 114.1 [79.9–147.6] 40.0 [29.3–47.8] p < 0.0001d Aβ42 (pg/ml) 511 [421–623] 420 [372–575] 1240 [1036–1600] p < 0.0001d NLK (ng/ml) 3.88 [3.00–5.24] 4.17 [3.61–4.60] 2.82 [2.30–3.60] p < 0.0001d YKL-40 (ng/ml) n = 13; 324 ± 63 n = 5; 323 ± 40 n = 29; 226 ± 51 p < 0.0001a Characteristics of the Barcelona AD-control group. Medians and [IQR] are reported, except for age and YKL-40 (mean ± SD), and sex (n). TP, T-tau, P-tau, Aβ42, NLK and YKL-40 were measured in CSF Abbreviations: AD Alzheimer’s disease, Aβ42 amyloid beta-42, APOE Apolipoprotein E, aMCI amnestic mild cognitive impairment, F female, IQR interquartile range, M male, MMSE Mini-Mental State Examination, N.A. not available, NLK neuroleukin, P-tau phosphorylated tau, SD standard deviation, TP total protein, T-tau total tau, YKL-40 protein associated with neuro-inflammation and Alzheimer’s disease, named after the first three N-terminal amino acids: tyrosine (Y), lysine (K), and leucine (L) and its molecule mass (40 kDa), also known as chitinase-3 like-protein 1 a ANOVA b Chi-square test c Fisher’s exact test d Kruskal-Wallis test NLK is a 56-kDa multifunctional protein named after myeloid leukemia cells [33]. In an immunohistochemical two of its functions: it can act as a neurotrophic factor study in AD patients, NLK expression was found in neu- and as a lymphokine [12]. In the extracellular space rons, microglia, and oligodendroglia, and in the vessel NLK supports neuronal survival and stimulates axonal wall, specifically in cerebrovascular cells such as smooth growth [10–13], but NLK may also play a role in the in- muscle cells [8]. NLK has also been investigated in the duction of B-cell differentiation into antibody-secreting context of other neurodegenerative diseases: in a trans- cells [9]. NLK is also known as glucose-6-phosphate genic mouse model of Huntington’s disease, the NLK isomerase or phosphohexose isomerase, an intracellular gene was upregulated [34]. Furthermore, in patients with enzyme involved in glycolysis and gluconeogenesis [31], Parkinson’s disease and multiple system atrophy, CSF as autocrine motility factor, a tumor maturation factor NLK levels were normal [25]. in cancer cells, playing a role in metastasis [32] and as a Our finding that CSF NLK is increased in aMCI pa- maturation factor, mediating differentiation of human tients suggests that altered levels are already present in Fig. 2 A ROC analysis showed moderately high accuracy levels for discrimination of aMCI from control in the Nijmegen aMCI patients and controls (gray line) and the Barcelona aMCI patients and controls. B ROC analysis showed consistently high accuracy levels for discrimination of AD from control in the Nijmegen AD patients and controls (gray line) and the Barcelona AD patients and controls. Abbreviations: AD, Alzheimer’s disease; AUC area under the curve. The Barcelona cohort serves as a validation cohort De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 8 of 10 as a missense mutation in the gene encoding for TREM2, a receptor expressed on microglia) and in- creased AD risk [36]. Furthermore, Aβ may initiate microglial activation, resulting in microglial migration to plaques, phagocytosis, and enzymatic degradation of Aβ [37]. Prolonged microglial activation may be detrimental, since it eventually leads to a self-perpetuating cycle in- volving the production of neurotoxic pro-inflammatory cytokines, damaging neurons, inducing more recruit- ment of microglia, further increasing their activation and release of pro-inflammatory cytokines [35]. Fig. 3 Scatter plot of the correlation between CSF NLK levels and Inflammation does, however, not seem to be a major YKL-40 levels in a subset (n = 47) of the Barcelona aMCI/AD patients component in CAA-related microhemorrhage or and controls. NLK and YKL-40-levels significantly correlated in the microinfarction [38], with the exception of a rare combined aMCI/AD and controls (rSP = 0.51, p < 0.0001, gray line) and in the aMCI/AD patients separately (rSP = 0.79, p < 0.0001, black clinical subtype: CAA-related inflammation [6]. On dots and black line), but not in the controls (rSP = 0.033, p = 0.87, the other hand, previous immunohistochemical stud- white dots, dotted line). Abbreviations: AD, Alzheimer’s dementia; rSP, ies in AD patients and in a transgenic rat model of Spearman rank coefficient; aMCI, amnestic mild cognitive CAA have found that CAA type 1, i.e., amyloid de- impairment; NLK, neuroleukin; YKL-40, chitinase-3 like-protein 1 position in the capillary vessels, was associated with perivascular inflammation [39, 40]. prodromal AD, and that NLK might serve as an early We previously demonstrated by immunohistochemis- biomarker for AD. Levels were also altered in AD pa- try that NLK may be associated with CAA pathology [8], tients. We found similar effects sizes and AUC’s of CSF but apparently this association is not reflected by an in- NLK in aMCI patients and AD patients in two inde- crease of NLK in the CSF. We could speculate that CAA pendent samples of patients and controls, indicating that pathology, in contrast to AD pathology, does not lead to our results are consistent. However, there is overlap of a neuroprotective response involving NLK to an extent NLK levels between aMCI patients and controls. More in which that is reflected by an increase of NLK in the studies will be needed to confirm this association of CSF. The correlation between NLK and MMSE levels NLK with different stages of AD as well as its potential might suggest that NLK levels in CAA patients are influ- biomarker value for early diagnosis. enced by cognition (and possibly additional AD path- We observed a marked increase of NLK in aMCI/AD ology). However, since we only have the availability of with a positive AD biomarker status, and a clear correl- MMSE scores in a subset of CAA patients, we should in- ation with CSF p-tau and t-tau, whereas we found no or terpret these data with caution. only a weak correlation of NLK with CSF Aβ42. To- gether, these findings suggest that an increased cerebral NLK production is tightly associated with tau-mediated Strengths and limitations neurodegeneration. A limitation of this study is that we did not have In line with this, we found that CSF NLK strongly cor- YKL-40 levels available for the Nijmegen CAA, AD related with CSF YKL-40 (also known as chitinase-3 patients, and controls. However, the correlation in like-protein 1 or cartilage glycoprotein-39), an estab- the Barcelona samples is quite strong. In addition, lished neuroinflammatory marker [14], expressed by the lack of detailed information on the cognitive astrocytes and microglia. Furthermore, our finding that status of CAA patients obstructed us from systemat- this association is very strong in the aMCI/AD group ically analyzing this as a possible confounding factor suggests that a neuroinflammatory process involving in- for the analysis of NLK levels. Furthermore, controls creased production of NLK and YKL-40 may be particu- were significantly younger than the aMCI patients larly associated with the early phases of the disease and AD patients. However, when we adjusted for process. age, difference between patients and controls Chronic microglial inflammation has been observed in remained significant. Our group sizes were relative the brains of AD patients at various disease stages, and small, but a strength of our study is the inclusion of pro-inflammatory cytokines such as IL-1β, IL-2, IL-6, aMCI/AD and controls subjects with thorough clin- and tumor necrosis factor-α have been shown to be ical evaluation, for many of whom imaging and AD upregulated in AD [35]. In addition, genome-wide asso- biomarker data was also available. Another important ciation studies have identified associations between in- strength is that we were able to validate our findings flammation- and immune-related variants in genes (such in two independent aMCI/AD samples. De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 9 of 10 Conclusions Spain. 3Center of Biomedical Investigation Network for Neurodegenerative We conclude that CSF NLK is increased in aMCI and Diseases (CIBERNED), Madrid, Spain. 4Department of Laboratory Medicine, Radboud University Medical Center, Nijmegen, The Netherlands. AD patients, but not in CAA patients, which indicates 5 Department of Neurology, Massachusetts General Hospital, Harvard Medical that it is a marker associated with early stage AD. NLK School, Boston, MA, USA. 6ADx NeuroSciences, Ghent, Belgium. 7Department is a neurotrophic factor, possibly involved in a neuropro- of Geriatrics, Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Radboud Alzheimer Centre, Nijmegen, The tective response to neuronal damage caused by AD Netherlands. pathology. The strong correlation of NLK with YKL-40 suggests an inflammatory role for NLK as well. Future Received: 26 May 2021 Accepted: 8 September 2021 studies will have to confirm whether an increase in CSF NLK is associated with early stages of AD and whether References NLK has an additive value as biomarker for the differen- 1. Mattson MP. Pathways towards and away from Alzheimer’s disease. Nature. tial diagnosis of dementia syndromes. 2004;430(7000):631–9. https://doi.org/10.1038/nature02621. 2. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr, Kawas Acknowledgements CH, et al. The diagnosis of dementia due to Alzheimer’s disease: We thank Levi Romanesco for his help with rating the MRIs. recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's Authors’ contributions disease. Alzheimers Dement. 2011;7(3):263–9. https://doi.org/10.1016/j.ja HBK and MMV designed the study, and IK and AAMV performed the lz.2011.03.005. experiments. ADK, HBK, and MMV analyzed and/or interpreted the data. ADK, 3. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med. DA, SMG, FHBMS, CJMK, AL, and JC contributed to the acquisition of patient 2004;256(3):183–94. https://doi.org/10.1111/j.1365-2796.2004.01388.x. data and samples. ADK drafted the manuscript. All authors critically revised 4. Jack CR, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. the manuscript. The authors read and approved the final manuscript. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement. 2018;14(4):535–62. https://doi.org/10.1016/j.ja Funding lz.2018.02.018. This study was supported by the BIONIC project (no. 733050822), which has 5. Boyle PA, Yu L, Nag S, Leurgans S, Wilson RS, Bennett DA, et al. Cerebral been made possible by ZonMW. This project is part of “Memorabel,” the amyloid angiopathy and cognitive outcomes in community-based older research and innovation program for dementia, as part of the Dutch national persons. Neurology. 2015;85(22):1930–6. https://doi.org/10.1212/WNL. “Deltaplan for Dementia”: zonmw.nl/dementiaresearch, and the CAFÉ project 0000000000002175. (the National Institutes of Health, USA, grant number 5R01NS104147-02). The 6. Greenberg SM, Bacskai BJ, Hernandez-Guillamon M, Pruzin J, Sperling R, van BIONIC project is a consortium of Radboudumc, LUMC, ADX Neurosciences, Veluw SJ. Cerebral amyloid angiopathy and Alzheimer disease - one and Rhode Island University. CJM Klijn and FHBM Schreuder are supported peptide, two pathways. Nat Rev Neurol. 2020;16(1):30–42. https://doi.org/1 by a clinical established investigator grant of the Dutch Heart Foundation 0.1038/s41582-019-0281-2. (grant 2012 T077) and CJM Klijn by an ASPASIA grant from The Netherlands 7. Linn J, Halpin A, Demaerel P, Ruhland J, Giese AD, Dichgans M, et al. Organization for Health Research and Development, ZonMW (grant Prevalence of superficial siderosis in patients with cerebral amyloid 015008048). FHBM Schreuder is supported by a senior clinical scientist grant angiopathy. Neurology. 2010;74(17):1346–50. https://doi.org/10.1212/WNL. of the Dutch Heart Foundation (grant 2019 T060). 0b013e3181dad605. The SPIN cohort has received funding from CIBERNED; Instituto de Salud 8. Rensink AAM, Otte-Höller I, Donkelaar HJ, De Waal RMW, Kremer B, Verbeek Carlos III; jointly funded by Fondo Europeo de Desarrollo Regional (FEDER), MM. Differential gene expression in human brain pericytes induced by Unión Europea, “Una manera de hacer Europa”; Generalitat de Catalunya; amyloid-β protein. Neuropathol Appl Neurobiol. 2004;30(3):279–91. https:// Fundació “La Marató TV3” Fundació Bancària Obra Social La Caixa; Fundación doi.org/10.1111/j.1365-2990.2004.00536.x. BBVA; Fundación Española para el Fomento de la Investigación de la 9. Gurney M, Apatoff B, Spear G, Baumel M, Antel J, Bania M, et al. Esclerosis Lateral Amiotrófica (FUNDELA); Global Brain Health Institute; Neuroleukin: a lymphokine product of lectin-stimulated T cells. Science. Fundació Catalana Síndrome de Down; and Fundació Víctor Grífols i Lucas. 1986;234(4776):574–81. https://doi.org/10.1126/science.3020690. 10. Gurney ME, Apatoff BR, Heinrich SP. Suppression of terminal axonal Availability of data and materials sprouting at the neuromuscular junction by monoclonal antibodies against The datasets used and/or analyzed during the current study are available a muscle-derived antigen of 56,000 daltons. J Cell Biol. 1986;102(6):2264–72. from the corresponding author upon reasonable request. https://doi.org/10.1083/jcb.102.6.2264. 11. Deng L, Shi B, Zhuang Y, Chu J, Shi X, Zhang S, et al. Performance and Declarations mechanism of neuroleukin in the growth and survival of sertoli cell-induced neurons in a coculture system. Cell Transplanti J. 2014;23(3):381–94. https:// Ethics approval and consent to participate doi.org/10.3727/096368913X663578. Lumbar punctures and vena punctures were performed after informed 12. Gurney MEM. Molecular cloning and expression of neuroleukin, a consent from the patients themselves or from the patients’ legal neurotrophic factor for spinal and sensory neurons. Science. 1986;234(4776): representatives. This study was approved by the Medical Ethics Committee 566–74. https://doi.org/10.1126/science.3764429. Arnhem-Nijmegen (file nos. 2016-3011 and 2014-1401), Ethical Committee of 13. Tanie Y, Tanabe N, Kuboyama T, Tohda C. Extracellular neuroleukin IIB-Sant Pau (16/2013) and Partners Human Research Committee of Boston enhances neuroleukin secretion from astrocytes and promotes axonal (study ID PHRC #2006P000664). growth in vitro and in vivo. Front Pharmacol. 2018;9:1228. https://doi.org/1 0.3389/fphar.2018.01228. Competing interests 14. Baldacci F, Lista S, Cavedo E, Bonuccelli U, Hampel H. Diagnostic function of E. Stoops is a full-time paid employee and shareholder of ADx NeuroSci- the neuroinflammatory biomarker YKL-40 in Alzheimer’s disease and other ences. All other authors report no declaration of interest. neurodegenerative diseases. Expert Rev Proteomics. 2017;14(4):285–99. https://doi.org/10.1080/14789450.2017.1304217. Author details 15. Craig-Schapiro R, Perrin RJ, Roe CM, Xiong C, Carter D, Cairns NJ, et al. YKL- 1 Department of Neurology, Radboud University Medical Center, Donders 40: a novel prognostic fluid biomarker for preclinical Alzheimer’s disease. Institute for Brain, Cognition and Behaviour, Radboud Alzheimer Centre, P.O. Biol Psychiatry. 2010;68(10):903–12. https://doi.org/10.1016/j.biopsych.2010. Box 9101, 6500 HB Nijmegen, The Netherlands. 2Sant Pau Memory Unit, 08.025. Department of Neurology, Hospital de la Santa Creu i Sant Pau, Biomedical 16. Querol-Vilaseca M, Colom-Cadena M, Pegueroles J, San Martín-Paniello C, Research Institute Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Clarimon J, Belbin O, et al. YKL-40 (Chitinase 3-like I) is expressed in a subset De Kort et al. Alzheimer's Research & Therapy (2021) 13:160 Page 10 of 10 of astrocytes in Alzheimer’s disease and other tauopathies. J of Huntington’s disease. J Neurochem. 2000;75(2):830–9. https://doi.org/10.1 Neuroinflammation. 2017;14(1):118. https://doi.org/10.1186/s12974-017- 046/j.1471-4159.2000.0750830.x. 0893-7. 35. Kinney JW, Bemiller SM, Murtishaw AS, Leisgang AM, Salazar AM, Lamb BT. 17. Bonneh-Barkay D, Bissel SJ, Kofler J, Starkey A, Wang G, Wiley CA. Astrocyte Inflammation as a central mechanism in Alzheimer’s disease. Alzheimers and macrophage regulation of YKL-40 expression and cellular response in Dement. 2018;4(1):575–90. https://doi.org/10.1016/j.trci.2018.06.014. neuroinflammation. Brain Pathol. 2012;22(4):530–46. https://doi.org/10.1111/ 36. Karch CM, Goate AM. Alzheimer’s disease risk genes and mechanisms of j.1750-3639.2011.00550.x. disease pathogenesis. Biol Psychiatry. 2015;77(1):43–51. https://doi.org/10.1 18. Alcolea D, Vilaplana E, Suárez-Calvet M, Illán-Gala I, Blesa R, Clarimón J, et al. 016/j.biopsych.2014.05.006. CSF sAPPβ, YKL-40, and neurofilament light in frontotemporal lobar 37. Heneka MT, Carson MJ, Khoury JE, Landreth GE, Brosseron F, Feinstein DL, degeneration. Neurology. 2017;89(2):178–88. https://doi.org/10.1212/WNL. et al. Neuroinflammation in Alzheimer’s disease. Lancet Neurol. 2015;14(4): 0000000000004088. 388–405. https://doi.org/10.1016/S1474-4422(15)70016-5. 19. Illán-Gala I, Alcolea D, Montal V, Dols-Icardo O, Muñoz L, de Luna N, et al. 38. van Veluw SJ, Scherlek AA, Freeze WM, Ter Telgte A, van der Kouwe AJ, CSF sAPPβ, YKL-40, and NfL along the ALS-FTD spectrum. Neurology. 2018; Bacskai BJ, et al. Different microvascular alterations underlie microbleeds 91(17):e1619–e28. https://doi.org/10.1212/WNL.0000000000006383. and microinfarcts. Ann Neurol. 2019;86(2):279–92. https://doi.org/10.1002/a 20. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. na.25512. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work 39. Zhu X, Hatfield J, Sullivan JK, Xu F, Van Nostrand WE. Robust Group under the auspices of Department of Health and Human Services neuroinflammation and perivascular pathology in rTg-DI rats, a novel model Task Force on Alzheimer's Disease. Neurology. 1984;34(7):939–44. https://doi. of microvascular cerebral amyloid angiopathy. J Neuroinflammation. 2020; org/10.1212/WNL.34.7.939. 17(1):78. https://doi.org/10.1186/s12974-020-01755-y. 21. Vos SJB, Visser PJ, Verhey F, Aalten P, Knol D, Ramakers I, et al. Variability of 40. Richard E, Carrano A, Hoozemans JJ, van Horssen J, van Haastert ES, CSF Alzheimer’s disease biomarkers: implications for clinical practice. PLoS Eurelings LS, et al. Characteristics of dyshoric capillary cerebral amyloid One. 2014;9(6):e100784–e. angiopathy. J Neuropathol Exp Neurol. 2010;69(11):1158–67. https://doi. 22. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practical method org/10.1097/NEN.0b013e3181fab558. for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–98. https://doi.org/10.1016/0022-3956(75)90026-6. 23. Alcolea D, Clarimón J, Carmona-Iragui M, Illán-Gala I, Morenas-Rodríguez E, Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in Barroeta I, et al. The Sant Pau Initiative on Neurodegeneration (SPIN) cohort: published maps and institutional affiliations. a data set for biomarker discovery and validation in neurodegenerative disorders. Alzheimers Dement. 2019;5(1):597–609. https://doi.org/10.1016/j. trci.2019.09.005. 24. Alcolea D, Pegueroles J, Muñoz L, Camacho V, López-Mora D, Fernández- León A, et al. Agreement of amyloid PET and CSF biomarkers for Alzheimer’s disease on Lumipulse. Ann Clin Transl Neurol. 2019;6(9):1815–24. https://doi.org/10.1002/acn3.50873. 25. Santaella A, Kuiperij HB, van Rumund A, Esselink RAJ, van Gool AJ, Bloem BR, et al. Cerebrospinal fluid monocyte chemoattractant protein 1 correlates with progression of Parkinson’s disease. NPJ Parkinson’s Dis. 2020;6(1):21. 26. Verbeek MM, Kremer BP, Rikkert MO, Van Domburg PH, Skehan ME, Greenberg SM. Cerebrospinal fluid amyloid beta(40) is decreased in cerebral amyloid angiopathy. Ann Neurol. 2009;66(2):245–9. https://doi.org/10.1002/a na.21694. 27. Wilhelmus MM, Otte-Höller I, Davis J, Van Nostrand WE, de Waal RM, Verbeek MM. Apolipoprotein E genotype regulates amyloid-beta cytotoxicity. J Neurosci. 2005;25(14):3621–7. https://doi.org/10.1523/ JNEUROSCI.4213-04.2005. 28. Alcolea D, Carmona-Iragui M, Suárez-Calvet M, Sánchez-Saudinós MB, Sala I, Antón-Aguirre S, et al. Relationship between β-Secretase, inflammation and core cerebrospinal fluid biomarkers for Alzheimer’s disease. J Alzheimers Dis. 2014;42(1):157–67. https://doi.org/10.3233/JAD-140240. 29. Scheltens P, Leys D, Barkhof F, Huglo D, Weinstein HC, Vermersch P, et al. Atrophy of medial temporal lobes on MRI in “probable” Alzheimer’s disease and normal ageing: diagnostic value and neuropsychological correlates. J Neurol Neurosurg Psychiatry. 1992;55(10):967–72. https://doi.org/10.1136/ jnnp.55.10.967. 30. Pasquier F, Leys D, Weerts JG, Mounier-Vehier F, Barkhof F, Scheltens P. Inter- and intraobserver reproducibility of cerebral atrophy assessment on MRI scans with hemispheric infarcts. Eur Neurol. 1996;36(5):268–72. https:// doi.org/10.1159/000117270. 31. Cordeiro AT, Godoi PHC, Silva CHTP, Garratt RC, Oliva G, Thiemann OH. Crystal structure of human phosphoglucose isomerase and analysis of the initial catalytic steps. Biochim Biophys Acta (BBA) - Proteins Proteomics. 2003;1645(2):117–22. https://doi.org/10.1016/S1570-9639(02)00464-8. 32. Watanabe H, Takehana K, Date M, Shinozaki T, Raz A. Tumor cell autocrine motility factor is the neuroleukin/phosphohexose isomerase polypeptide. Cancer Res. 1996;56(13):2960–3. 33. Xu W, Seiter K, Feldman E, Ahmed T, Chiao J. The differentiation and maturation mediator for human myeloid leukemia cells shares homology with neuroleukin or phosphoglucose isomerase. Blood. 1996;87(11):4502–6. https://doi.org/10.1182/blood.V87.11.4502.bloodjournal87114502. 34. Iannicola C, Moreno S, Oliverio S, Nardacci R, Ciofi-Luzzatto A, Piacentini M. Early alterations in gene expression and cell morphology in a mouse model