Cerebral folate deficiency (CFD) is a condition defined by low concentration (<41 nM) of
5-methyltetrahydrofolate (5MTHF, the biologically active form of folate) in the
cerebrospinal fluid (CSF) 1. CFD is considered to be of "primary" origin if it is
associated with several genetic diseases involving folate metabolism, leading to severe
motor and cognitive manifestations. CFD can also be "secondary", i.e. associated with
various definite diseases such as genetic mitochondrial diseases and also unidentified
diseases. In secondary CFD, the cause of CFD is not understood. In the center, the team
identified 16 CFD patients with common characteristics: deep secondary CFD (5MTHF<10nM
whereas blood folate is normal), high protein CSF >1g/L (N<0.5), and a specific
involvement of brain white matter, hyperintense on T2-weighted MRI. The team called this
syndrome "LHIPFOLD" for Leukoencephalopathy with CSF HIgh Protein and FOLate Deficiency
(confidential, planned submission in end of 2022). Within the French metabolic
biochemistry labs network, the team identified 6 additional LHIPFOLDpat (LHIPFOLD
patients). They have a mean age of 42 to 14 years (min 15; max 77), and a sex ratio of
1.8 (M/F). 9/22 had either a large mitochondrial DNA deletion (Kearns-Sayre) or
bi-allelic POLG mutations (mitochondrial DNA replication defect). The other 13 patients
were all undiagnosed until recently, but the team found in a research context and
validated in vitro bi-allelic variants in 4/7 tested patients within genes from MRPS
family, coding for sub-units of the mitoribosome.
LHIPFOLD patients clearly have a defect of intracerebral 5MTHF transport, but without
mutation of the brain folate transporters. Since folate is exclusively transported from
blood to the brain by the choroid plexus (CP)2, the team hypothesize that CFD in LHIPFOLD
is due to CP dysfunction. CP, localized in the brain ventricles, is a vascularized
epithelial organ that acts as a blood-CSF barrier, expressing numerous transporters for
molecules transfer from the blood to the CSF and vice versa (removing waste products from
brain cells). CP also secretes CSF and expresses CSF-secreted proteins related to various
functions3.
High protein CSF due to a blood brain barrier alteration, and reported abnormal CP
morphology from brain autopsy in two Kearns-Sayre patients are also in favor of CP
dysfunction in LHIPFOLD 4. In addition, in two LHIPFOLDpat, the team found very high
concentrations of CSF sialic acid without the involvement of sialic acid-related genes
whereas CP strongly expresses the sialic acid efflux transporter. The team also recently
measured strongly decreased CP size after automatic segmentation in two other LHIPFOLD
pat. All these data strongly argue for CP dysfunction in LHIPFOLD .
If the hypothesis is correct, it means that beyond CFD, other CSF abnormalities probably
exist as a consequence of generalized CP dysfunction. This may be very interesting in
terms of therapeutic approach, as folate supplementation has a clear but limited
therapeutic effect, especially in some patients. Unfortunately, to assess our hypothesis,
there is no available clinical exploration to evaluate CP functions, whereas CP
dysfunction/alteration is suggested as a potential pathogenesis contributor in several
common neurological diseases (for the following text the team will use dysfunction for
both dysfunction and alteration). As the team think that in LHIPFOLD the CP dysfunction
is particularly severe, as suggested by the deep CFD, the team consider LHIPFOLD syndrome
a very useful clinical model to validate the capacity of some relevant diagnostic tools
to evaluate CP function. Therefore, the objectives of the project are:
Prove CP dysfunction in LHIPFOLD, through the identification of specific and
CP-related MRI and biochemical signatures, using CP-centered imaging and
metabolomics and proteomics approaches.
Identify CP-related biochemical biomarkers in LHIPFOLD directly amenable to
treatment.
Set-up imaging and biochemical diagnostic tests for evaluating CP function in
clinical practice (biomarkers-based CP-dysfunction score).
Methodology To reach the objectives, the team will conduct a monocentric diagnostic phase
1 case-control study, based on descriptive, deterministic statistic and multidimensional
clustering approach, supported by complementary diagnostic tools specifically designed
for CP exploration of LHIPFOLDpat versus healthy volunteers (HV) and neurological control
patients (NCpat). NCpat will be used to assess the specificity of the CP-related
signatures identified in LHIPFOLDpat. The team expect that findings will be similar to HV
in some NCpat, but will reflect slight to moderate CP dysfunction in others. With AP-HP
as the study sponsor, the team will obtain approval of the Research Ethics Committee for
a "Jardé 1" study, in accordance with the legislation and regulatory requirements in
force.
-Patients and controls selection and recruitment Subjects will be prospectively recruited
during a two-year period: 1)LHIPFOLDpat will be selected as previously described; 2)HV
will be recruited by the usual advertisement procedure, and will be matched for age and
sex with LHIPFOLDpat; 3) NCpat will be age and sex matched patients with definite common
brain diseases frequently explored in the Neurology department: multiple sclerosis,
amyotrophic lateral sclerosis, alzheimer's disease, frontotemporal dementia, idiopathic
intracranial hypertension and normal pressure hydrocephalus. The diagnosis of these
diseases will be made according to the newest guidelines by the neurologists specialised
in each disease. Concerning the sample to be recruited, the team took into account two
constraints: the number of LHIPFOLDpat is limited; the study is not a simple
deterministic comparison vs a control group but more a multidimensional analysis leading
to the identification of CP-related signatures through homogeneous patient clusters.
Consequently, the strategy was to identify a global sample size that would lead to a
homogeneous classification of patients based on 5 or 6 biomarkers. According to Donicar's
formula5 2^6 patients properly selected by a stratified sampling technique should allow
this. These 64 patients will include 15 LHIPFOLDpat and an equal number of 25 HV and 25
NCpat. Furthermore, keeping the deterministic approach, and based on a simulation study
integrating mean and standard deviation of CSF 5MTHF in LHIPFOLDpat and NCpat (3.85 +/-
2.67 and 49.53 +/- 13.35 respectively) with respectively 15 and 25 subjects, this sample
size provides a power greater than 95% to demonstrate a difference with a Wilcoxon Rank
Sum test between these two groups (of note, values in HV are currently unknown).
The team will acquire brain MRI data (with blood gadolinium injection), blood and CSF
samples from all subjects. LHIPFOLDpat and NCpat will undergo these procedures as part of
standard care in the Neurology Department of Pitié-Salpêtrière Hospital. HV will be
hospitalized one day in our Clinical Investigation Center. All subjects will sign
informed consent and agree to undergo MRI examination and blood and CSF sampling for
research purpose for HV, and additional MRI time acquisition and sample volumes for
patients.
Brain MRI study Brain MRI will be performed at 3 TESLA (Siemens PRISMA) at the
Centre for Neuroimaging Research (CENIR), the imaging platform of the Paris Brain
Institute (ICM). The team will assess both CP morphology and function: 1/for CP
morphological analysis, the team will measure CP volumes and extract textural
parameters. CP segmentation will be performed using automated software developed at
the site. Texture analysis will be performed using the MP2RAGE MRI sequence. 2/CP
function assessment will include quantitative measurements of the capillary
permeability and macrovascular perfusion. To explore CP permeability, dynamic T1
perfusion sequence with gadolinium enhancement will be used; t-TRANS and other
regional values will be extracted. Post processing will be performed using Syngovia
(Siemens) software. To explore CP macrovascular perfusion, the team will use a
shorter version of the dynamic Arterial Spin Labeling perfusion (without gadolinium)
published previously 6. With this sequence the team will be able to appreciate the
arterial transit time and apparent blood flow. Data analysis will be performed at
CENIR.
Discovery of CP dysfunction biomarker candidates in CSF/blood compartments Two mixed
approaches will be used to select the top 10 biomarkers candidates (a priori 5
metabolites and 5 proteins) that will be investigated by a targeted and fully
quantitative analysis: i)untargeted metabolomics and proteomics (the findings will
be reviewed to assess a possible link with CP function); and ii)hypothesis-driven
approach: in addition to 5MTHF, based on data from the literature and publicly
available biological database, the team also identified metabolites (including
ascorbic acid and sialic acid) that seem to be specifically transported by the CP,
and proteins specifically expressed by the CP (including TRPM3 and Klotho). As blood
concentrations can influence CSF concentrations for some molecules, the team plan to
analyze blood and CSF sampled at the same time, with calculation of a CSF/Blood
ratio of concentrations.
For untargeted metabolomics, metabolic extracts of CSF and plasma samples will be
obtained following methanol-assisted protein precipitation and analysed by liquid
chromatography coupled to high resolution mass spectrometry (LC-HRMS) as the team
routinely perform 7. A combination of two complementary LC-HRMS platforms will be used to
profile both hydrophobic and polar metabolites8. Data pre-treatment and statistical
analyses will be achieved by using Workflow4Metabolomics (W4M), which is an open-source
collaborative online platform for computational metabolomics. Annotation of data sets and
identification of biologically relevant signals will be performed by using 'in-house'
spectral libraries, and MS/MS (tandem mass spectrometry) experiments. Relative
quantification of ~250 metabolites will be achieved thanks to normalization using quality
control samples. For targeted metabolomics, CSF 5MTHF is routinely measured in Necker
laboratory specialized in Neurometabolism, using very sensitive and specific analytical
methods (liquid chromatography hyphenated to tandem MS). The team are in the process to
include in the same run ascorbic acid and sialic acid measures. Depending on the chemical
nature of the metabolite biomarker candidates identified in the untargeted study, one or
few new methods for robust quantification using stable isotopes will be developed.
A preliminary proteomic study on the CSF of a patient with high CSF protein and a control
will allow us to establish a sample preparation protocol, select a mass spectrometry
analysis method and data analyses workflow. The objective is twofold: to optimize the
extraction of proteins and to allow the comparison of very different CSFs in order to
develop an analytical strategy that will allow us to identify biomarkers. The study will
be carried out by using a label-free approach to minimize variability, the number of
steps and especially not to restrict the dynamic range of the sample. All bioinformatics
analyses will be performed with myProMS9, our proteomic data management and analysis
software developed at the Institut Curie in collaboration with the INSERM U900
bioinformatics platform directed by Dr E. Barillot. For the targeted proteomic approach,
the team will use the expertise on targeted MS with stable isotope-labeled internal
standard peptides, the gold standard approach for protein quantification in biological
fluid, resulting in high robustness, high sensitivity and multiplex analysis10.
-Cross data analysis The team will centrally manage, analyze and integrate clinical and
biological datasets and knowledge generated by the MRI and biological samples analysis to
identify a global CP-related signature in LHIPFOLD patients. With these data, the team
will establish a CP-dysfunction score for use in routine clinical practice, intended to
be used to explore CP in neurological patients.