Depuis l’introduction du terme de schizophrénie à début tardif (SDT) puis de celui, plus récent, de very-late-onset schizophrenia-like psychosis (VLOSP), les processus neuropathologiques sous-tendant ces formes tardives de schizophrénie n’ont pas été déterminés. Cet article se propose de faire le point sur les connaissances actuelles concernant d’éventuels processus neurodégénératifs dans les formes tardives. Pour cela, une revue de la littérature sur la base de donnée PubMed entre le 1^er janvier 1995 et le 30 avril 2008 a été effectuée. Les résultats reprennent divers champs d’investigation : clinique, anatomopathologie, imagerie. Au travers de ces divers champs d’investigation, deux hypothèses principales ont été testées : (1) : la SDT est-elle prodromique de maladie d’Alzheimer ? (2) : la SDT est-elle secondaire à une pathologie cérébrovasculaire ? L’interprétation des résultats permet quelques pistes. La SDT comme prodrome de la maladie d’Alzheimer semble peu probable. Cependant une évolution démentielle à dix ans est possible chez les sujets ayant un début très tardif des troubles. Au plan anatomopathologique, les SDT pourraient être atteints d’une tauopathie limbique. Les sujets les plus vulnérables décompenseraient spontanément avec l’âge. D’autres sujets décompenseraient plus tardivement sous l’impact d’une pathologie cérébrovasculaire. Cette tauopathie limbique pourrait être en lien avec des circuits cérébraux spécifiques à la SDT. Ces circuits seraient différents de ceux impliqués dans la schizophrénie débutant chez l’adulte jeune. L’atteinte de ces circuits spécifiques pourrait expliquer les différences cliniques et électroencéphalographiques existant entre formes tardives de schizophrénie et schizophrénie débutant chez l’adulte jeune.

Since 1943 when, for the first time, M. Bleuler used the term late-onset schizophrenia (LOS) to refer to diagnostic groups gathering some clinical features of schizophrenia, but with onset after the age of 40, opinions on this entity are divided. The main question is whether LOS and schizophrenia, with onset in early adulthood (early-onset schizophrenia EOS), have the same etiopathogeny. This discussion became more complex with the introduction, in 2000, of a new entity, the “very-late-onset schizophrenia-like psychosis” (VLOSP), which took the place of “late paraphrenia” and grouped together schizophrenia, delusional disorders and paranoid psychosis with age of onset after 60 years. Neuropathological processes underlying these entities have not beed determined. In particular, neurodegenerative processes could be explored.

A literature review between 1 January 1995 and 30 April 2008, based on a research on PubMed with the terms “late-onset schizophrenia”, “paraphrenia”, “late paraphrenia”, “VLOSP”, and “late psychosis”, takes stock of the various studies and hypotheses which have investigated the link between LOS/VLOSP and neurodegenerative processes.

Clinical approach: there is no greater family history of dementia disorders in LOS/VLOSP than in the general population. Neuropsychological pattern between LOS and Alzheimer’s disease (AD) seems different, with more impairment in delayed recall in AD, and in short-term memory in LOS. Some longitudinal studies, however, have argued that a part of patients with LOS would develop dementia at 10 years. These patients would have a later onset of disorders (> 60 years). Anatomopathological approach: anatomopathological studies show that LOS is not consecutive to AD, and might be related to a restricted limbic tauopathy. Neuroimaging approach: Magnetic Resonance Imaging (MRI) morphological neuroimaging studies show little differences between LOS and EOS. The thalamus volume was significantly smaller, and cortical atrophy was more important in LOS. MRI neuroimaging studies regarding white matter hyperintensities (WMH), which are considered as a macroscopic manifestation of cerebrovascular disease (CD), show inconsistent results. Positive results are observed with LOS subjects with later onset of disease. In addition, WMH were not located on cerebral tracts, which are implicated in EOS, but were periventricular.

It is notable that the studies conducted do not differentiate between LOS and VLOSP, since these two groups of patients are generally mixed. Furthermore, an attempt to analyse the data has to take into account only the studies which have included subjects with strict criteria of schizophrenia, independently of the disorders’ age of onset. In addition, we found no study comparing LOS and Lewy Bodies Dementia, which is however an important differential diagnosis. The two main hypotheses tested appear to be: (1): is LOS a prodrome of AD? (2): is LOS consecutive to CD? This review permits a partial answer to these questions. First, LOS as a prodrome of AD seems very unlikely, considering anatomopathological studies. However, it is possible that subjects with later age of onset of the disease develop dementia at 10 years. Second, CD seems not to be altered by the cerebral tracts which are implicated in EOS. A hypothesis could be that the tracts implicated in LOS and EOS differ. This might explain the differences found in clinical (less formal thought disorders and negative symptoms in LOS compared with EOS) and electroencephalographical studies (no reduction of P300’s amplitude in LOS, which has however been proposed as trait marker of schizophrenia in EOS) between LOS and EOS. These tracts could be related to a limbic tauopathy found in anatomopathological studies. More vulnerable subjects would develop LOS with ageing. Other subjects would develop LOS secondary to CD. This could explain the later age of onset of the disease in these subjects.