This work shows that ventricular and lumbar levels of Aβ42 are closely related in communicating hydrocephalus; they are low in iNPH, and comparable to those of other surgical hydrocephalus.
The main limit of this study is that, for ethic reasons, we cannot compare the obtained data to the concentrations existing in "normal" subjects. CSF is a transport medium for numerous molecules (nutriments, growing factors, terminal products of brain cell metabolism). The ventricular and lumbar protein concentrations are often different, depending upon their origin. Indeed, brain proteins are considered to generally exhibit a higher level in intraventricular CSF, whereas blood proteins are more concentrated in lumbar CSF [13, 14]. However, there are many exceptions; for example, concentrations of transthyretin, a protein synthesized by choroid plexus, are nearly the same in ventricular and lumbar CSF .
There are no available reference values well established for Aβ42 ventricular levels in healthy individuals. Therefore, we measured Aβ42 levels in lumbar and ventricular CSF of 5 patients with chronic communicating hydrocephalus and we found that ventricular and lumbar Aβ42 values are not significantly different. Despite the small sample size, this fact suggests the ventricular CSF Aβ42 concentrations measured in this study seem to be relevant for a comparison with those existing in lumbar CSF of control subjects.
In iNPH, several studies [16–19] demonstrated that the lumbar Aβ42 levels are lowered. Our results are in good agreement with these previous data: more than 75% of the studied patients, whatever the aetiology of the hydrocephalus, exhibit reduced Aβ42 levels in their ventricular CSF. The fact that the ventricular Aβ42 levels were similar iNPH and oCH suggests some similarities in the pathophysiological process despite the existing differences in CSF dynamics and aetiologies. The same findings were reported for other proteins (neurofilament light protein, tau, sulfatide, vasoactive intestinal peptide and neuropeptide Y) in case of iNPH and aqueductal stenosis . Several hypotheses could explain this decrease of Aβ42 levels.
Several hypothesis might be put forward to explain the decrease in Aβ42 levels. First, oligomer formation, favoured by the altered CSF turnover in NPH, could partially mask the antigenic sites of the peptide, especially the C-terminal part, which is hidden inside the hydrophobic core of the aggregate. This therefore could lead to an underestimation of the Aβ42 levels when using an ELISA technique . Second, Aβ42 glycation could contribute to an underestimation of the protein concentration by ELISA. Indeed, the CSF glucose concentration is relatively high (about 0.5g/l), and CSF stasis promotes Amadori products formation , which could also mask some antigenic sites of the peptide. Third, a reduced production of Aβ42 could be suspected, as described in Creutzfeldt-Jacob's disease. In this affection, there are no Aβ42 deposits and CSF Aβ42 levels are significantly reduced , this probably due to the inhibition of β-secretase cleavage by PrP(C) .
However, an intra-cerebral sequestration of amyloid protein most likely occurs. Strozyk et al, investigating the relationship between amyloid neuropathology and post-mortem CSF Aβ42 levels in an autopsy sample of 155 patients with dementia found that a high number of plaques is associated with reduced levels of ventricular Aβ42 .
The « B compound of Pittsburgh » (PIB), a thioflavine derivative, which is a ligand of Aβ42, is able, after i.v. injection, to cross the blood brain barrier and bind to Aβ peptides. Molecular neuroimaging allows to detect amyloid pathology in vivo using PET scanning with PIB . Several authors [27, 28] using the PIB PET scanning, demonstrated, in patients with Alzheimer disease, that Aβ42 lumbar levels are lowered and inversely correlated to the Aβ load of brain. With the same technique, Leinonen et al  demonstrated the occurrence of amyloid deposits in 5/10 patients with iNPH, and cerebral biopsies performed during shunt surgery confirmed the existence of amyloid deposits in these patients. In a large follow-up study of NPH, idiopathic or secondary, Leinonen et al  also showed, on small cortical brain biopsies obtained during surgical treatment, that 186/433 (43%) patients displayed Aβ42 deposits. These lesions of AD are more frequent among older subjects with iNPH , which could be explained partially by a reduced transport via LRP-I [2, 32] and lower activities of some enzymes involved in the catabolism of the Aβ42 peptide ; this suggests an increasing role of the CSF turnover in Aβ42 clearance during aging.
It is interesting to note, based on data derived from the ADNI database, an inverse correlation between ventricular size and CSF Aβ42 levels is observed in controls, MCI and AD, which also supports the hypothesis that the CSF turnover could play an important role in the regulation of the CSF Aβ42 peptide concentration .