- June 24, 2026
- Sylvain Lehmann et al. | Université de Montpellier
Topic/Product:
CNS Disease Panel 120
Disease Area:
Mild Cognitive Impairment (MCI)
Sample Type:
CSF, Plasma
Abstract
Accurate prediction of which patients with mild cognitive impairment (MCI) will progress to dementia remains a major challenge. Current biomarkers detect amyloid pathology with high accuracy but offer limited prognostic value for disease progression. We conducted a prospective analysis in the multicentre BALTAZAR cohort, all diagnosed with MCI at baseline and followed for 3 years. Paired cerebrospinal fluid (CSF) and plasma samples were analysed with the NULISA ultrasensitive multiplex platform quantifying more than 120 central nervous system biomarkers. Prognostic performance was assessed using area under the curve (AUC) and hazard ratios (HRs), both for individual markers and for elastic-net-derived biomarker combinations validated by bootstrap and survival analyses. During the 3-year follow-up, 36% of participants converted to dementia. Plasma p-tau biomarkers showed strong accuracy for detecting amyloid positivity (AUC > 0.90) but limited prognostic value for conversion (AUC < 0.75). In CSF, markers of neurodegeneration (tau, NfL) and synaptic dysfunction (NPTX2 encoding the Neuronal Pentraxin 2) predicted conversion with higher accuracy, exceeding p-tau217 performance. The best-performing CSF combination (IL-16, tau, NPTX2) achieved an AUC of 0.86 (95%CI 0.80-0.91) and an HR of 39.8 (95%CI 9.6-165.2). Plasma combinations (p-tau181 or p-tau217 with YWHAG encoding for 14-3-3 protein gamma, a member of the 14-3-3 protein family) provided only modest improvement, likely reflecting the absence of robust synaptic markers in blood. Prognostic assessment of MCI progression to dementia is best achieved through CSF biomarker combinations reflecting neurodegeneration and synaptic dysfunction, complemented by inflammatory markers. These findings emphasize the clinical and pathophysiological relevance of downstream processes beyond amyloid and tau, and support the implementation of multimarker panels for prognosis and therapeutic monitoring.
Authors & Affiliations
Constance Delaby¹˒², Susanna Schraen-Maschke³, Claire Paquet⁴, Frédéric Blanc⁵, Jean-Sébastien Vidal⁶, Christophe Hirtz Said Assou⁷, Bernadette Allinquant⁸, Stéphanie Bombois⁹, Audrey Gabelle¹⁰, Olivier Hanon⁶˒¹¹, and Sylvain Lehmann¹˒¹¹
¹ LBPC-PPC, Université de Montpellier, INM INSERM, IRMB CHU de Montpellier, Montpellier, France
² Sant Pau Memory Unit, Hospital de la Santa Creu i Sant Pau-Biomedical Research Institute Sant Pau-Universitat Autònoma de Barcelona, Barcelona, Spain
³ Univ. Lille, Inserm, CHU Lille, UMR-S-U1172, LiCEND, Lille Neuroscience & Cognition, LabEx DISTALZ, F-59000 Lille, France
⁴ Université Paris Cité, GHU APHP Nord Lariboisière Fernand Widal, Centre de Neurologie Cognitive, F-75010 Paris, France
⁵ Université de Strasbourg, CHRU de Strasbourg, Memory Resource and Research Centre of Strasbourg/Colmar, French National Centre for Scientific Research (CNRS), ICube Laboratory and Fédération de Médecine Translationnelle de Strasbourg (FMTS), Team Imagerie Multimodale Intégrative en Santé (IMIS)/Neurocrypto, F-67000 Strasbourg, France
⁶ Université Paris Cité, UMR-S 1144, APHP, Hospital Broca, Memory Resource and Research Centre of de Paris-Broca-Ile de France, F-75013 Paris, France
⁷ Institute for Regenerative Medicine and Biotherapy (IRMB), University of Montpellier, INSERM, CHU Montpellier, 34295 Montpellier, France
⁸ UMR-S1266, Université Paris Cité, Institute of Psychiatry and Neuroscience, Inserm, Paris, France
⁹ Assistance Publique-Hôpitaux de Paris (AP-HP), Département de Neurologie, Centre des Maladies Cognitives et Comportementales, GH Pitié Salpêtrière, Paris, France
¹⁰ Université de Montpellier, Memory Research and Resources Center, Department of Neurology, Inserm INM NeuroPEPs team, F-34000 Montpellier, France
¹¹ These authors jointly supervised this work: Olivier Hanon, Sylvain Lehmann
