The present invention relates to the field of human medicine and specifically to the diagnosis of cognitive disorder including Alzheimer disease. This invention relates to a biomarker for cognitive disorders, in particular Alzheimer's disease, consisting of circulating non-coding RNAs (ncRNA) such as microRNA (miRNA) and long non-coding RNAs (lncRNA), or their combination in a peripheral body fluid and representing a non-invasive method for diagnosing the cognitive disorder in particular Alzheimer's disease or for monitoring its development or progression using this biomarker. It further relates to associated kits, methods, protocols and transmittable forms of information for diagnosis purposes and/or for other human medicine applications including the patient stratification in clinical trials and/or monitoring the efficacy of therapeutic strategies using this biomarker and peripheral body fluids from patients. The present invention will thus have a strong impact on the quality of life of patients and a significant impact on the healthcare system and economy.
Alzheimer's disease (AD) is a chronic neurodegenerative disease characterized by progressive loss of cognitive function and pathologically by extracellular deposition of amyloid-beta peptide (Aβ) and intracellular deposition of hyper-phosphorylated tau protein in neurofibrillary tangles in the brain, associated with progressive neuronal degeneration (Marcus et al. J Neurogenet. 2011; 25(4):127-33; Gotz et al, Br J Pharmacol. 2012; 165(5):1246-59).
AD is the most common form of dementia. Approximately 46.8 million people worldwide currently live with AD or other type of dementia. With an ageing population, this number is estimated to increase to 131.5 million by 2050 (World Alzheimer Report, 2015: http://www.alz.co.uk/research/WorldAlzheimerReport2015.pdf). As such, AD is becoming an increasingly important burden on the affected individuals and their families as well as economic and social costs on medical and healthcare resources in both developed and emerging countries.
Currently, there is no cure for AD. Symptomatic treatments exist for this disease all trying to counterbalance neurotransmitter's disturbance. Although a number of new potential disease-modifying therapeutic candidates are currently being studied in clinical trials, none has been approved yet. From 2002 to 2012, there was a failure of 99.6% of AD clinical trials that were 289 Phases 2 and 3 trials on symptomatic agents (36.6%), disease-modifying small molecules (35.1%) and disease-modifying immunotherapies (18%) (Cummings et al., Expert Rev Clin Pharmacol. 2014;7(2):161-5). Among the strategies proposed by worldwide experts in the AD field and by pharmaceutical industries to improve the success rate for AD drug development, are: a) intervening earlier in the disease process before neurodegeneration begins, b) identifying and developing accurate biomarkers for early diagnosis, non-invasive and suitable for stratification of subject populations and for longitudinal monitoring of drug efficacy in clinical trials.
The AD pathological features are defined in post-mortem histopathological analysis. The presence of amyloid plaques remains an absolutely required feature for the definitive diagnosis of AD, as accepted by the American Academy of Neurology, American Psychiatric Association (DSM-IV) and both the CERAD and NIA-Reagan Institute neuropathological criteria.
Current diagnosis of the disease remains uncertain (Dubois et al Lancet Neurol. 2014;13(6):614-29) and it is based on combination of battery of clinical and neuropsychological tests and neuroimaging, such as structural imaging using Magnetic Resonance Imaging (MRI) and/or glucose metabolism using Positron emission tomography (PET) of fluorodeoxyglucose F18 (18F-FDG), with an accuracy of less than 70%-85% depending on the method and the severity stage of the disease (Frisoni et al., Nat Rev Neurol. 2010; 6(2): 67-77; Tahmasian et al., J Nucl Med. 2016 ;57(3):410-5).
Sometimes, detection of AD-associated biomarkers Aβ42 and tau or phosphorylated-tau in cerebrospinal fluid (CSF) is additionally performed (Sunderland et al., JAMA. 2003;289(16):2094-2103), but CSF tests request a lumbar puncture, which is an invasive procedure and often requires hospitalization of subjects. In addition, the accuracy of these CSF tests remains insufficient as up to 30% of subjects can be misdiagnosed and the test cannot be longitudinally practiced to be followed up for confirmation of diagnostic purposes.
Longitudinal confirmatory diagnosis test is essential for applications to early detection of the preclinical stages of AD and mild cognitive impairment (MCI; defined as an early stage during which the first subtle symptoms manifest) and thus detecting early and calculating the risk of conversion of MCI into AD. Due to its limits (invasive), a CSF test is not rarely used as a biomarker for early diagnosis in preclinical AD stages before symptoms appear, nor it is suitable as a companion biomarker repeatedly practiced at several time points in same subjects recruited in longitudinal clinical trials.
Recently, specific neuroimaging methods relevant for AD pathology are being developed, notably with ligands for in vivo amyloid-beta (Aβ) ligands for Positron Emission Tomography (PET) neuroimaging including F-18 florbetapir and F-18 flutemetamol which have been approved by FDA and/or EMA (Choi et al., 2009 and Wong et al., 2010). However, their practice is costly and very limited (available only in some hospitals of large cities equipped with PET technology, still mainly for research purposes as not reimbursed by health insurances) and their diagnosis accuracy is still under studies for further understanding. Thus, the use of these Aβ PET neuroimaging methods remains very limited and the vast majority of patients do not profit from such tests even in rich countries.
There is a need for better definition of AD patient population to be included for the drug-development trials: PET imaging using Aβ ligands showed that up to 30% of subjects diagnosed with AD show a negative Aβ scan and that up to 35% of subjects with normal cognition status show a positive Aβ scan (Gaël Chételat et al, NeuroImage Clinical 2013; 2:356-65). Thus, PET Aβ scan is being used to guide for recruitment of subjects in the desired cohorts in some clinical trials by the large pharmaceutical companies. Anti-Aβ antibody e.g. Solanezumab tested in patients with mixed mild to moderate AD failed to show clear efficacy, however with some encouraging data in only mild AD patients subgroup. In mild AD patients selected based on CSF biomarker profile, the results showed a promising efficacy (Carlson et al., Alzheimers Dement (Amst). 2016; 2:75-85; Siemers et al., Alzheimers Dement. 2016;12(2):110-20). However, the use of CSF biomarker is invasive, dramatically limiting its use as a biomarker for drug development. For example, it is not suitable for repeated use to monitor the stability and/or the progression of the disease, nor it is suitable to monitor in clinical longitudinal trials the efficacy of disease-modifying therapeutic drug candidates or preventive strategies.
Overall, the existing tests either lack an easy accessibility and simplicity for use for diagnosis of the large AD population and/or lack accuracy (sensitivity and specificity). This represents a major impediment and bottleneck to develop reliable and rapid diagnosis test for AD. Another impediment is the identification of a biomarker that does not require invasive sample collecting, such as a spinal tap. The lack of such an accessible, sensitive and specific biomarker that could be validated by cellular, animal model, pre-clinical models, and human testing impedes the development of therapies and drugs for AD or for the studies on pathological processes triggering AD or involved in the progression of AD.
Today, clearly there is an unmet need for an accurate and non-invasive peripheral biomarker test for diagnosis of AD including preclinical and early AD and for applications in drug development (patient stratification and monitoring drug efficacy in clinical trials).
ncRNAs include small microRNAs (miRNAs) and long non-coding RNAs (lncRNAs).
Expression levels of miRNAs may have potential as diagnostic biomarkers as they are known to circulate and tissue specific profiles can be identified in a number of body fluids such as plasma, CSF and urine. Recent developments in deep sequencing technology present a viable approach to develop biomarker pipelines in order to profile peripheral miRNA signatures specific to neurodegenerative diseases.
We previously showed that the lncRNAs expressed or highly enriched in tissues such as cardiac tissue or brain tissue, can be released into the peripheral circulation and be easily quantifiable by classical RT-PCR in different peripheral samples (PCT/EP2018/065492). In addition of the RT-PCR, we performed also total RNA sequencing on peripheral samples and quantified significant proportion of lncRNAs in the peripheral samples such as serum, plasma and Paxgene-RNA-tube collected whole blood (PCT/EP2018/065492).
According to the invention, new miRNA signatures and lncRNA signatures specific of neurodegenerative diseases, especially of AD, have been identified, using a method set up combining most recent technology and samples collected in non-invasively manner, including Paxgene whole blood tube, serum and plasma.
Alzheimer (AD) and age-matched healthy controls. Y axis: mean+/−standard error of the mean of concentration of the lncRNAs in CPM (count per million). X axis: AD: Alzheimer patient group, HV: Healthy control group.
The invention relates to a method for the diagnosis of a cognitive disorder including but not limited to Alzheimer disease in a subject at risk of having or developing mild cognitive impairment or a cognitive disorder, comprising:
In some embodiments, the ncRNA signature comprises or consists of ncRNAs selected from the group consisting of:
The invention also relates to a method for monitoring the development or progression of a cognitive disorder in a subject suffering from a loss or impairment of cognitive function or dementia, said method comprising the following steps:
In some embodiments, the ncRNA signature comprises or consists of ncRNAs selected from the group consisting of:
LINC01206:20, lnc-05orf67-3:1, HAND2-AS1:58, lnc-PRDM9-20:1, lnc-CLK1-1:7, lnc-DNALI1-5:4, RORB-AS1:6, lnc-TPPP-1:3, lnc-BMS1-2:1, lnc-ADRB1-4:1, lnc-XXYLT1-5:1, MIR99AHG:104, LINC01748:17, and lnc-AKR1E2-15:1;
A biological sample may be a small part of a subject obtained from a body fluid, e.g. blood, plasma, serum, urine or cerebrospinal fluid. Suitably the biological sample are plasma or serum or Paxgene-RNA-tube.
By cognitive disorder is meant e.g. progressive impairment or progressive loss of a cognitive function, for example AD.
The “reference” may be suitable control sample such as for example a sample from a normal, healthy subject having no cognitive impairment symptoms and no abnormal neuroimaging findings and being age-matched to the patient to be diagnosed with the method of the present invention. The reference may be a sample from the same subject prior to demonstration of disorder or disease symptoms or prior to diagnosis of the impairment or loss of cognitive function or dementia. The reference may be a standardized sample, e.g. a sample comprising material or data from several samples of healthy subjects who have no cognitive impairment symptoms and no abnormal neuroimaging findings.
For Differential Diagnosis of a Specific Dementia Type or Its Mild Cognitive Impairment (MCI) Versus Other Dementia Types and Their MCI:
Dementia types and subtypes are numerous and include e.g.
Dementia with Lewy body (DLB) and MCI of DLB type
For differential diagnosis of one dementia subtype, the reference sample may be sample(s) from patient(s) suffering dementia of other type(s). For example, for differential diagnosis of AD or MCI of AD type, the sample(s) shall be from patient(s) with MCI/dementia of non-AD type (i.e. sample(s) from subject(s) with FTD and/or DLB and/or PDD and/or vascular dementia and/or PSP and/or CBD and/or other dementia), or from subject(s) who may suffer another disease but have no cognitive impairment or dementia.
The invention also relates to a method of diagnosing a cognitive disorder including but not limited to Alzheimer disease and monitoring its development or progression in a subject suffering from progressive impairment or loss of cognitive function or dementia, comprising:
The invention also relates to a method of diagnosing a cognitive disorder including but not limited to early MCI and very mild Alzheimer disease in a subject with no apparent and/or no measurable symptoms when using the available methods such as MMSE, MoCA and other neurocognitive tests, but with abnormal changes of biomarkers of the present invention preceding the symptoms, comprising:
The invention further relates to a method of predictive diagnosing a cognitive disorder including but not limited to early MCI and very mild Alzheimer disease in a subject with no apparent and/or no measurable abnormality by neuroimaging methods as MRI and/or CT scans, comprising:
The invention also relates to a method for treating a subject suffering from a progressive impairment or loss of cognitive function or dementia, said method comprising:
Examples of therapeutic treatment of cognitive disorders may comprise the administration of:
Furthermore, the invention comprises a kit for diagnosing and/or monitoring a cognitive disorder including but not limited to AD, comprising at least one reagent for the determination of a ncRNA expression profile comprising or consisting of at least one ncRNA (miRNA and/or lncRNA) selected from 406 miRNAs listed in Table 1 and 1008 lncRNAs listed in Table 5 in the biological sample.
In some embodiments, the ncRNA expression profile comprises or consists of ncRNAs selected from the group consisting of:
The kit of the invention allows performing the measurement of the miRNA signature and/or the lncRNA signature of the invention, wherein the kit comprises at least one reagent for measuring at least one ncRNA (miRNA and/or lncRNA) as indicated above.
By “reagent” is meant a reagent which specifically allows the determination of the miRNA/gene expression profile, i.e. a reagent specifically intended for the specific determination of the expression level of the miRNA/gene present in the miRNA/gene expression profile. Examples include e.g. amplification primer pairs (forward and reward) and/or probes specific for the miRNA/gene present in the miRNA/gene expression profile. This definition excludes generic reagents useful for the determination of the expression level of any other miRNA/gene.
By “reagent” is also meant a reagent which specifically allows the determination of the lncRNA expression profile, i.e. a reagent specifically intended for the specific determination of the expression level of the lncRNA present in the lncRNA expression profile. Examples include e.g. amplification primer pairs (forward and reward) and/or probes specific for the lncRNA present in the lncRNA expression profile. This definition excludes generic reagents useful for the determination of the expression level of any other lncRNA.
In some embodiments, the kit for diagnosing and/or monitoring a cognitive disorder comprises one or more oligonucleotide probes specific for ncRNAs of interest and a reagent for purifying the probe- target nucleic acid complexes. The oligonucleotide probes comprise a sequence complementary to a region of the ncRNAs of interest. The oligonucleotide probes may be DNA or RNA. The oligonucleotide probes are preferably DNA. In a preferred embodiment, the oligonucleotide probes are biotinylated and the reagent for purifying the probe-target complexes is a streptavidin-coated substrate, e.g, a streptavidin-coated magnetic particle, e.g., T1 streptavidin coated magnetic bead. In a preferred embodiment, the ncRNAs of interest is lncRNAs. The length of oligonucleotide probes specific for lncRNAs may be from 30 to 80 nucleotides, e.g., from 40 to 70, from 40 to 60, or about 50 nucleotides.
A further embodiment of the invention relates to a targeted sequencing panel for next generation sequencing, comprising nucleic acids specific for at least one ncRNA (miRNA and/or lncRNA) selected from 406 miRNAs listed in Table 1 and 1008 lncRNAs listed in Table 5 in the biological sample.
In some embodiments, the at least one ncRNA (miRNA and/or lncRNA) comprises or consists of ncRNAs selected from the group consisting of:
The invention also relates to the applications of the method for development of new therapeutic strategies (drug candidates) tested in clinical trials for the treatment of a cognitive disorder including but not limited to Alzheimer disease, wherein the method can be used (a) before starting the treatment for the selection of patients who would then be recruited in clinical trial; thus the test will enhance the likelihood for treating patient population that benefits from the tested therapeutic strategy(ies) while avoiding recruitment of patient subpopulation patients who do not benefit from this tested new drug candidate(s) and/or (b) for monitoring the response(s) including efficacy of the tested therapeutic strategy(ies) once treatment with the new tested drug candidate starts.
In a further aspect, the invention also relates to methods of detection. For example, the invention contemplates a method for the detection of at least one ncRNA (miRNA and/or lncRNA). In one aspect, the method comprises:
In some embodiments, the ncRNA signature comprises or consists of ncRNAs selected from the group consisting of:
The expression level of miRNAs and/or lncRNAs may be determined by any technology known by a man skilled in the art. In particular, the expression level of miRNAs and/or lncRNAs is determined by measuring the amount of nucleic acid transcripts of each miRNA or lncRNAs. The amount of nucleic acid transcripts can be measured by any technology known by a man skilled in the art. The measure may be carried out directly on an extracted RNA sample or on retrotranscribed complementary DNA (cDNA) prepared from extracted RNA by technologies well-known in the art. From the RNA or cDNA sample, the amount of nucleic acid transcripts may be measured using any technology known by a man skilled in the art, including nucleic acid microarrays, quantitative PCR, sequencing (e.g., next generation sequencing), FIMAP quantification, and hybridization with a labeled probe.
In some embodiments, the expression level of miRNAs and/or lncRNAs is determined using sequencing, e.g., next generation sequencing. Sequencing may be carried out after converting extracted RNA to cDNA using reverse transcriptase or RNA molecules may be directly sequenced. In a particular embodiment, which should not be considered as limiting the scope of the invention, the measurement of the expression level using next generation sequencing may be performed as follows. Briefly, RNA is extracted from a sample (e.g., blood sample). After removing rRNA, RNA samples are then reverse transcribed into cDNA. To ensure strand specificity, single stranded cDNA is first synthetized using Super-Script II reverse transcriptase and random primers in the presence of Actinomycin D, and then converted to double stranded cDNA with the second strand marking mix that incorporates dUTP in place of dTTP. Resulting blunt ended cDNA are purified using AMPure XP magnetic beads. After a 3′end adenylation step, adaptor is attached to cDNA. So obtained cDNA (sequencing library) may be amplified by PCR. The sequencing libraries can be sequenced by any next generation sequencing technology known by a man skilled in the art.
In some embodiments, the measurement of the expression level of ncRNAs, e.g., by sequencing (e.g., next generation sequencing), is facilitated by capturing and enriching nucleic acids (RNA or cDNA) corresponding to ncRNAs of interest prior to the measurement. As used herein, enrichment refers to increasing the percentage of the nucleic acids of interest in the sample relative to the initial sample by selectively purifying the nucleic acids of interest. The enrichment of nucleic acids corresponding to ncRNAs of interest can be carried out on extracted RNA sample or cDNA sample prepared from extracted RNA. In some embodiments, nucleic acids corresponding to ncRNAs of interest are captured and enriched by hybridizing RNA or cDNA sample to oligonucleotide probes specific for ncRNAs of interest (e.g. oligonucleotide probes comprising a sequence complementary to a region of ncRNAs of interest) under conditions allowing for hybridization of the probes and target nucleic acids to form probe-target nucleic acid complexes. Probes may be DNA or RNA, preferably DNA. The probe-target nucleic acid complexes can be purified by any technology known by a man skilled in the art. In a preferred embodiment, probes are biotinylated. The biotinylated probe-target nucleic acid complexes can be purified by using a streptavidin-coated substrate, e.g, a streptavidin-coated magnetic particle, e.g., T1 streptavidin coated magnetic bead. In a preferred embodiment, the ncRNAs measured by this method are lncRNAs. The length of probes specific for lncRNAs may be from 30 to 80 nucleotides, e.g., from 40 to 70, from 40 to 60, or about 50 nucleotides.
In some embodiments, the expression level of miRNAs and/or lncRNAs may be determined using quantitative PCR. Quantitative, or real-time, PCR is a well-known and easily available technology for those skilled in the art and does not need a precise description. In a particular embodiment, which should not be considered as limiting the scope of the invention, the determination of the expression profile using quantitative PCR may be performed as follows. Briefly, the real-time PCR reactions are carried out using the TaqMan Universal PCR Master Mix (Applied Biosystems). 6 μl cDNA is added to a 9 μl PCR mixture containing 7.5 μl TaqMan Universal PCR Master Mix, 0.75 μl of a 20X mixture of probe and primers and 0.75μ1 water. The reaction consists of one initiating step of 2 min at 50 deg. C., followed by 10 min at 95 deg. C., and 40 cycles of amplification including 15 sec at 95 deg. C. and 1 min at 60 deg. C. The reaction and data acquisition can be performed using the ABI 7900HT Fast Real-Time PCR System (Applied Biosystems). The number of template transcript molecules in a sample is determined by recording the amplification cycle in the exponential phase (cycle threshold or CQ or CT), at which time the fluorescence signal can be detected above background fluorescence. Thus, the starting number of template transcript molecules is inversely related to CT.
In some embodiments, the expression level of miRNAs and/or lncRNAs may be determined by the use of a nucleic acid microarray. A nucleic acid microarray consists of different nucleic acid probes that are attached to a substrate, which can be a microchip, a glass slide or a microsphere-sized bead. A microchip may be constituted of polymers, plastics, resins, polysaccharides, silica or silica-based materials, carbon, metals, inorganic glasses, or nitrocellulose. Probes can be nucleic acids such as cDNAs (“cDNA microarray”) or oligonucleotides (“oligonucleotide microarray”). To determine the expression profile of a target nucleic acid sample, said sample is labelled, contacted with the microarray in hybridization conditions, leading to the formation of complexes between target nucleic acids that are complementary to probe sequences attached to the microarray surface. The presence of labelled hybridized complexes is then detected. Many variants of the microarray hybridization technology are available to the man skilled in the art.
In some embodiments, the expression level of miRNAs and/or lncRNAs may be determined by FIMAP quantification. Briefly, lncRNAs and miRNAs are amplified by PCR using primers with the same sequences as for qPCR that are chemically modified. Forward primers are phosphorylated in 5′ and Reverse primers biotinylated in 5′. PCR products are digested with exonuclease to eliminate the phosphorylated strand and keep only the biotinylated ones. Biotinylated PCR products are incubated with coded silica microdisks coated with oligos complementary to the RNAs of interest (one code per target oligo), and hybridized products are revealed by addition of a fluorescent streptavidin conjugate. The expression level of miRNAs and/or lncRNAs is determined by measuring fluorescence signal.
In a still further aspect, the method of detection contemplates the use of a reagent to measure or detect the one or more miRNAs and/or one or more lncRNA. By “reagent” is meant a reagent that specifically allows the determination of the miRNA/gene or the lncRNA expression profile, i.e. a reagent specifically intended for the specific determination of the expression level of the miRNA/gene present in the miRNA/gene expression profile or the lncRNA present in the lncRNA expression profile. Examples include e.g. amplification primer pairs (forward and reward) and/or probes specific for the miRNA/gene present in the miRNA/gene expression profile and/or or the lncRNA present in the lncRNA expression profile. This definition excludes generic reagents useful for the determination of the expression level of any other miRNA/gene and generic reagents useful for the determination of the expression level of any other lncRNA.
In yet a further aspect, the method of detection comprises detecting one or more miRNAs and/or lncRNA from a biological sample, wherein the biological sample is selected from blood, plasma, serum, urine, cerebrospinal fluid, tear, saliva. In some embodiments, the biological sample is blood.
By “treating” or “treatment” of a subject being at risk to develop or having a cognitive disorder, particularly a progressive impairment or loss of cognitive function or dementia, e.g. AD, is meant administering or administration of a regimen to the subject in need thereof such that at least one symptom of the disorder or disease is cured, alleviated, remedied or improved. According to the present invention, depending on the diagnosing results, the subject is submitted to an adapted therapeutic treatment and is further monitored. The monitoring of the subject based on the miRNA signature and/or lncRNA signature of the invention allows to further adapt or modify the therapeutic treatment, e.g. increasing the drug amount, administering a combination of drugs to obtain a synergy effect, or replacing the drug by an effective amount of another drug.
The invention also relates to the modification of a therapeutic strategy in subjects suffering from cognitive disorder who have been diagnosed and/or monitored using a method for (in vitro) diagnosis or monitoring of a progressive impairment or loss of cognitive function or dementia according to the invention.
According to the invention, the miRNAs and/or the lncRNAs of the invention may be used in combination with one or more biomarkers currently used for diagnosing a cognitive disorder including but not limited to AD. Examples of such biomarkers include without any limitation the biomarkers as disclosed in U.S. Pat. Nos. 9,377,472, 7,897,786 and in U.S. Pat. No. 6,703,212, the contents thereof being included herein by reference. The method of the invention combining the miRNAs and/or the lncRNAs of the invention with one or more known biomarkers may allow enhanced accuracy of diagnosis; or differential diagnosis; and a more efficient and successful drug development, e.g.
enhanced accuracy for patient stratification before recruitment in clinical trials and monitoring of the efficacy of approved therapies or novel drugs under development.
According to the invention, the miRNAs and/or the lncRNAs of the invention may be used in combination with one or more tests currently used for diagnosing a cognitive disorder including but not limited to AD and/or for differential diagnostic purposes. Examples of such tests include without any limitation the neuropsychological tests, such as MMSE, MoCA, and the neuroimaging methods, in particular volumetric MRI. The method of the invention combining the miRNAs and/or the lncRNAs of the invention with one or more known biomarkers may allow enhanced accuracy of diagnosis; or differential diagnosis; and a more efficient and successful drug development, e.g. enhanced accuracy for patient stratification before recruitment in clinical trials and monitoring of the efficacy of approved therapies or novel drugs under development.
To identify miRNA signatures involved specifically in the pathogenesis of AD and mild cognitive impairment (MCI) of AD type, a total of 2083 miRNAs in samples from different groups, including a group of patients with AD or MCI of AD type and group of cognitively intact healthy controls with no brain imaging abnormalities, have been screened. Subsequently miRNA profiling in all samples, 406 miRNAs have been detected above the threshold. By comparison of miRNA-expression levels measured in the samples of the different groups, the miRNA differentially expressed in the samples of the control group as compared to the expression level in the samples of patient group diagnosed as having AD or cognitive impairment of AD type (at the Neurology department of clinical site, based on neurocognitive and neuropsychological tests and neuroimaging tests and on cerebrospinal fluid biomarkers: beta-amyloid peptide 1-42 (Aβ42) and tau (total and/or phosphorylated) were identified as miRNA biomarker candidates.
To identify lncRNA signatures involved specifically in the pathogenesis of AD and mild cognitive impairment (MCI) of AD type, a total of 127802 lncRNAs in samples from different groups, including a group of patients with AD or MCI of AD type and group of cognitively intact healthy controls with no brain imaging abnormalities, have been screened. Subsequently lncRNA profiling in all samples, 19867 lncRNAs have been detected above the threshold. By comparison of lncRNA-expression levels measured in the samples of the different groups, the lncRNA differentially expressed in the samples of the control group as compared to the expression level in the samples of patient group diagnosed as having AD or cognitive impairment of AD type (at the Neurology department of clinical sites, based on neurocognitive and neuropsychological tests and neuroimaging tests and on cerebrospinal fluid biomarkers: beta-amyloid peptide 1-42 (Aβ42) and tau (total and/or phosphorylated) were identified as lncRNA biomarker candidates.
Expression levels were analyzed using a two-tailed Welch t test and/or Wilcoxon Mann-Whitney test between two groups. Significant differential expression was identified as p<0.05. Fold change and AUC (Area Under the Curve) are calculated for each miRNA and lncRNA and for each tested condition. miRNA candidates were also selected when differential expression was determined based on fold-changes ≥1.49-fold or ≤0.75-fold changes and/or AUC>0.6 or <0.4, are indicated in Tables 2, 3 and 4.
lncRNA candidates were also selected when differential expression was determined based on fold-changes ≥1.6 (or ≤0.6) and/or an AUC≥0.80 (or <0.20), are indicated in Tables 6 and 7. Random Forest algorithm (Breimann 2001, Breiman and Cutler 2001) was used to build the model and also used to select the top miRNAs and/or the top lncRNAs. A predictive model based on the combination of the top 2-20 miRNAs and/or lncRNAs enables to predict the disease with an accuracy of ≥80%.
Out of the 2083 miRNA measured, 406 miRNA showed expression levels above threshold. Sequences of the 406 miRNAs are shown in Table 1.
Out of the 406 detectable miRNAs analyzed, 28 miRNAs showed a statistically significant difference (Welch T-test p value <0.05), and among these candidates, 7 candidates showed a p value <0.01. The 28 miRNAs are shown in Table 2.
Among the 406 detected miRNA, a total of 74 miRNA candidates also showed an AUC value ∝0.6 or ≤0.4; of which a total of 11 candidates showed a good AUC value of ≥0.65 or ≤0.35; and 1 candidates showed a very good AUC value of ≥0.7 or ≤0.30. The 74 candidate miRNAs are shown in Table 3.
Among the 406 detected miRNAs, a total of 15 candidates showed both an AUC value of ≥0.6 or ≤0.4 and a fold change value of ≥1.49 or ≤0.75. The list of the 15 miRNAs are shown in Table 4.
Out of the 127802 lncRNAs sequenced, 19867 lncRNAs were selected based on their threshold expression level for statistical analysis. The comparison of the AD patient with healthy control populations showed that 1008 lncRNAs are differentially expressed with a statistical significance (p value <0.05, Wilcoxon test) (Table 5). The sequences of these 1008 lncRNAs are shown in the sequence listing included in this application.
Out of the 1008 lncRNAs differentially expressed with a statistical significance (p value <0.05), 33 lncRNAs showed a fold change of >2 or <0.5 and are shown in the Table 6.
Out of the 1008 lncRNAs differentially expressed with a statistical significance (p value <0.05), 60 lncRNAs showed an AUC of ≥0.85 and are shown in Table 7.
The results from the predictive modelling based on the random forest algorithm to discriminate between AD patient and healthy control populations when using the selected miRNAs with a p value <0.05 comprising signatures of 2 or more miRNAs showed a good accuracy. For example, the signature comprising three miRNAs, miR1220.3p, miR378d and miR99a.5p, enabled a good AUC of 0.839 and an accuracy of 81.8%.
The predictive modelling based on the random forest algorithm to discriminate between AD patient and healthy control populations when using the total of the 19867 lncRNAs analyzed, enabled to show that the AUC in function of the number of lncRNA reached a plateau with the following 13 lncRNAs. These 13 lncRNAs were used to construct the model. The results show that this lncRNA signature enabled a discrimination between the 2 populations (mild AD patient and healthy control populations) with an AUC value=0.993, an accuracy=95.8%, sensitivity=100% and specificity=91.7%.
Out of the 1008 lncRNAs differentially expressed with a statistical significance (p value <0.05), the 90 lncRNAs with a fold change of >2 (or <0.5) or an AUC of ≥0.85 (or ≤0.15) were used for the predictive modelling based on the random forest algorithm. The results show that with the following 7 top ranked candidates enabled a full discrimination between AD and HV groups, with an AUC value of 100% as well as 100% accuracy, 100% sensitivity and 100% specificity.
When applying another filter considering a list of the lncRNAs which have a fold change of ≥1.6 and an AUC of ≥0.85, the use of random forest algorithm enabled to select the following 12-top ranked lncRNAs to construct the model and this enabled to discriminate between the AD and HV populations with still an excellent AUC=0.979, excellent accuracy of 95.8% sensitivity of 91.7% and specificity of 100%.
Further to select the best lncRNA candidates, modeling using Random Forest algorithm was applied to a specific set of lncRNA candidates having both a statistically significant differential expression in AD patients versus healthy control populations and a good correlation (Pearson R coefficient) with scores of neurocognitive tests including MMSE and/or MOCA out of 7 neuropsychological tests performed (Table 8): The results show that the signature of the following 3 top ranked lncRNAs enabled an excellent discrimination between AD patients and healthy control populations with AUC=0.953, accuracy, sensitivity and specificity of 91.7%, 91.7% and 91.7%.
Modeling using Random Forest algorithm was also applied to a specific set of lncRNA candidates from Table 9 having both a statistically significant differential expression in AD patients versus healthy control populations and a good correlation (Pearson) with neuroimaging scores (volume of brain structures of relevance for cognition and memory such as the mediotemporal area, left and right hippocampus, left and right amygdala, entorhinal cortex out of more than 120 structures measured): The results show that the signature of the following 7 top ranked lncRNAs enabled an excellent discrimination between AD patients and healthy control population with AUC=0.993, accuracy=95.8%, sensitivity=91.7% and specificity=100%.
Modeling using Random Forest algorithm was also applied to a specific set of lncRNA candidates from Table 10 having both a statistically significant differential expression in AD patients versus healthy control populations and a good correlation (Pearson) with level of the CSF biomarkers of high relevance to AD pathology: Aβ42, tau or phosphorylated-tau: The results show that the signature of the following 18 top ranked lncRNAs enabled an excellent discrimination between AD patients and healthy control population with AUC=0.972 accuracy=0,917 sensitivity =0.83 and specificity=1.
To identify miRNA in plasma or serum or whole blood samples of human subjects, miRNA profiling of 2083 miRNAs was performed using HTG EdgeSeq miRNA whole transcriptome V2 targeted sequencing assay (HTG WTA V2, HTG Molecular, Tuscon, United States). This technology is based on nuclease protection targeted RNA sequencing assay that uses an extraction free lysis process followed by a nuclease protection assay (NPA) to prepare a stoichiometric library of nuclease protection probes (NPP) for measurement.
To identify lncRNAs in serum, plasma or whole blood samples of human subjects, circulating total RNA was first extracted and sequencing libraries were prepared by removal of ribosomic RNA (RiboZero TruSeq library preparation kit, Illumina Inc. San Diego, USA) and sequenced on Illumina NextSeq500 with 2×75 bp read length.
ncRNA (miRNA and lncRNA) were also measured by qPCR using specific primers. For this, circulating total RNA was first extracted, reverse transcription and real time PCR using specific primers were performed. Using FiMAP, a proprietary platform based on hybridization of PCR products on coated microdiscs with complementary oligonucleotide coupled to detection probes. Total RNA was extracted followed by reverse transcription and PCR step allowing multiplexing of several targets using specific primers to the ncRNAs. Quantification was therefore performed on the FiMAP platform using coded microdiscs specific of each ncRNA.
Serum or plasma samples were prepared from blood samples collected in lithium-heparin tubes and whole blood samples were collected in Paxgene RNA tubes. Samples were from healthy volunteers (HV), donors at the “Etablissement Français du Sang” (EFS) of Mulhouse, France, and from cognitively intact healthy control subjects and patients with mild cognitive impairment or with different stages of Alzheimer's disease (AD) recruited according to the protocols of the Amoneta Diagnostics sponsored prospective studies registered to the Agence Nationale de Sécurité du Médicament et des Produits de Santé (ANSM) including ADKIT study under the ID RCB: 2015-A00118-41 on Jan. 22, 2015, the chronobiological study under the ID RCB: 2016-A200227-44 on Feb 4., 2016.
Method for miRNAs
HTG whole transcriptome miRNA (WTA) kit was used (HTG WTA V2, HTG Molecular, Tuscon, United States). Samples were prepared accordingly to the following protocol: 15 μl of plasma lysis buffer and 15 μl of plasma sample and 3μl of Proteinase K are mixed and incubated at 50° C. for 60 min with orbital shaking. 25 μl of the mix is transferred to the HTG sample plate and loaded into the HTG processor to perform the nuclease protection assay and prepare the stoichiometric NPP.
For plasma samples, barcoding is performed using Hemo KlenTaq (MO332S, NEB, Evry, France) enzyme. For each sample, we mix 2.4 μl of Hemo KlenTaq, 0.6 μl of dNTPs (10 nM) (NEB, N04475), 6 μl of OneTaq PCR GC Buffer 5X (B90235, NEB, Evry, France), 3 μl of Forward and Reverse Primers (HTG WTA, HTG Molecular, Tuscon, United States), 3μl of sample preparation and 12 μl of H2O. PCR step was performed on ABI 2720 Thermocycler using the following cycling profile: 95° C. for 4 min followed by 16 cycles of 95° C. for 15 sec, 55° C. for 45 sec and 68° C. for 45 sec. Protocol is ended with 68° C. for 10 min.
In order to remove excess of primer from the library, Agentcour AMPure XP beads (A63880, Beckmancoulter, Villepinte, France) were used. For each sample, 37.5 μl of AMPure XP beads are used in combination with 15 μl of the PCR product. After mixing 10 times with pipette, the solution is incubated for 5 min at room temperature and then placed on the magnetic stand. After 2 minutes to separate beads, the cleared solution is carefully removed without disturbing the beads. Beads are washed 2 times with 200 μl of Ethanol 80%. Elution of PCR product link to the beads is performed with 25 μl of H2O. The purified solution of PCR product is place in a new tube while the plate is on the magnetic stand to separate PCR products and beads.
To determine concentration of library for each sample, Kapa Biosystems qPCR Kit (KK4824, Cliniscience, Nanterre, France) is used. For each reaction, the mix is composed of 12 μl of Mastermix, 0.4 μl of ROX dye, 3.6 μl of H2O and 4 μl of template (standards or library diluted at 1/10000). The samples were run on a ABI PRISM 7900HT (High ROX) will the following cycles: 95° C. for 5 min, 35 cycles of 95° C. for 30 sec and 65° C. for 45° C. with data collection followed by a dissociation curve. Standards are corresponding to an amplicon of 452 bp whereas the amplicon of NPP with barcode correspond to 115 bp. The ratio is applied to determine the concentration of each library.
Each sample are pooled in order to generate a pooled library at 4 nM. From this pooled library, 5 μl where mixed with 0.2N NaOH freshly prepared and incubated for 2 min. The solution was vortexed briefly and centrifuged at 280g for 1 minute and mixed with 990 μl of pre-chiller HT1 buffer (Illumina NextSeq Reagent v2 kit, Illumina, Paris, France). 15% PhiX (PhiX control v3, Illumina, Paris, France) at 20 pM was prepared. 260 μl of prepared denatured library at 20 pM was mix with 39 μl of 20 pM PhiX and 1001 μl of HT1 Buffer was loaded into Illumina NextSeq500 Mid Output V2 150 cycles kit and sequenced.
Data reconstruction and analysis was performed using directly FASTQ files from the Illumina Sequencer and processed by the HTG Parser software.
The data are normalized using the method recommended by the supplier. This normalization is based on total count of sequences after filtering out all values below 70 (threshold recommended by the supplier). After normalization, all miRNAs with 75th percentile of this threshold are considered as noise and have been deleted.
In order to determine miRNAs that discriminate Alzheimer disease patients from healthy control subjects, the following statistical methods are applied to the normalized data: two-tailed Welch t test and computation fold change were performed on normalized data. Individual ROC Curve was performed for each miRNA detected in the samples. The random Forest algorithm with the best miRNA in term of AUROC was performed to establish a predictive model for the diagnosis of Alzheimer disease. The final list of miRNAs of interest was chosen with p value <0.05 or fold change ≥1.49 or ≤0.75 in Log2 and/or individual AUC determined by the selection of all the miRNAs that are at least in one of the three lists of selected biomarkers (tables 1, 2 and 3).
Method for lncRNA
Ribonucleic acid (RNA) extraction is performed starting from 1.5 ml of serum, using Norgen Serum/plasma extraction and RNA Clean-Up and Concentration Micro-Elute Kits according to the manufacturer's instructions.
Sequencing libraries are prepared from the total amount of extracted RNA, using the Illumina TruSeq stranded total RNA library preparation kit combined with the human/mouse/rat RiboZero rRNA removal kit (Illumina Inc. San Diego, USA, C). All steps are performed with the low-throughput protocol and according to the manufacturer's instructions, with no fragmentation step. Briefly, cytoplasmic ribosomal RNA (rRNA) are hybridized to biotinylated target-specific oligos and removed using streptavidin coated magnetic beads. rRNA depleted RNA samples are then reverse transcribed into complementary deoxyribonucleic acid (cDNA). To ensure strand specificity, single stranded cDNA is first synthetized using Super-Script II reverse transcriptase (Invitrogen) and random primers in the presence of Actinomycin D, and then converted to double stranded cDNA with the second strand marking mix that incorporates dUTP in place of dTTP. Resulting blunt ended cDNA are purified using AMPure XP magnetic beads. After a 3′end adenylation step, Illumina's adapters ligation is performed. So, obtained singled indexed libraries are washed twice using AMPure XP beads to remove excess adapters and enriched by PCR (15 cycles). PCR products are purified with a final AMPure XP beads wash and sequencing ready libraries are eluted in 30 μl of resuspension buffer. For quality control, 1 μl of each library is run on the Agilent Technologies 2100 Bioanalyzer using a DNA 1000 chip according to the manufacturer's recommendations. Absence of adapter dimers is checked and the average library size is determined by a region table. Libraries are quantified on Qubit 2.0 using Qubit dsDNA High Sensitivity assay kit (Invitrogen). Library size previously determined on the Bioanalyzer is used to calculate molar concentrations from mass concentrations. All libraries are sequenced with the Illumina NextSeq500 (2×75bp).
RNA-seq data analysis is performed using Partek Flow (Partek Inc., St Louis, Mo., USA build 6). The pre-alignment QA/QC module of Partek Flow is used to visualize the read quality of the FASTQ files. All reads are examined. The raw FASTQ files are trimmed at the 3′ end in function of their quality score (Phred score). The parameters used are an end minimum quality level of 30 and a minimum trimmed read length of 50. Unaligned reads are mapped using the Homo sapiens hg19 genome. This mapping is done using the software STAR version 2.5.3 (Dobin, A., Davis, C. A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, M., and Gingeras, T. R. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinforma. Oxf. Engl. 29, 15-21). The default parameters are used. The post-alignment QC module of Partek Flow was used to visualize the average base quality score per position as well as the mapping quality per alignment. The mapped reads were quantified using the GTF file with the patented lncRNA annotation for quantification using the Partek Expectation/Maximization (E/M) algorithm (Xing, Y., Yu, T., Wu, Y. N., Roy, M., Kim, J., and Lee, C. (2006). An expectation-maximization algorithm for probabilistic reconstructions of full-length isoforms from splice graphs. Nucleic Acids Res. 34, 3150-3160). The default parameters are used. The transcripts where their medians are under the median read density in intergenic regions are discarded for the next steps of the analysis. The transcript counts were normalized by CPM (counts per million). Only transcripts showing high expression (CPM≥10) in at least half the samples of one group are considered.
To determine differentially expressed lncRNA, a statistical analysis is performed using Wilcoxon Mann-Whitney parametric test. A lncRNA with a p value ≤0.05 is considered as differentially expressed. In order to build classification models for the 2 classifications, the Classification for MicroArrays (CMA) package of R (Slawski M, Daumer M, Boulesteix A L. CMA: a comprehensive Bioconductor package for supervised classification with high dimensional data. BMC Bioinformatics 2008, 9:439) with a leave-one-out cross-validation has been used. The algorithms used for this predictive modelling are (a) random forest, (b) linear discriminant analysis and (c) naïve Bayes (Breiman L. Random forests. Machine Learning, 45(1): 5-32, 2001).
The rank of the RNA candidate in a model was calculated by the mean rank of 10 candidates selections (per CV-fold). Per model (and RNA selection method) the AUC was plotted as a function of the number of RNAs in the model. The optimal number of RNAs per model was determined graphically. ROC curves and confusion matrices were generated to assess the predictive performance of our models. The values of AUC, accuracy, sensitivity, specificity, positive and negative predictive values are reported.
Additional method for quantification of miRNA and lncRNA
Total RNA was extracted using Paxgene Blood RNA kit (Qiagen, France) and Serum/Plasma mini kit (Norgen Biotek, Canada) respectively, according to manufacturer's instructions.
RNA quality was examined on an Agilent 2100 Bioanalyzer with the RNA 6000 Nano Kits (Agilent, France). RNA quantity was measured on Qubit 3.0 fluorometer using RNA High sensitivity kit (Fisher Scientific, France).
cDNA Synthesis
Total RNA was used for miRNA and lncRNA cDNA synthesis as follows. For miRNA, 100 ng of RNA was reverse transcribed in a final volume of 10 μl including 1 μl of 10× poly(A) polymerase buffer, 0.1 mM of ATP, 1 μM of universal miRNA RT-primer, 0.1 mM of each deoxynucleotide (dATP, dCTP, dGTP and dTTP), 100 units of MuLV reverse transcriptase (New England Biolabs, USA) and 1 unit of poly(A) polymerase (New England Biolabs, USA) was incubated at 42° C. for 1 hour followed by enzyme inactivation at 95° C. for 5 minutes. The sequence of the RT-primer was 5′-CAGGTCCAGTTTT TTTTTTTTTTTVN, where V is A, C and G and N is A, C, G and T. The primer was purchased from IDT (Integrated DNA Technologies, Belgium). For lncRNAs, Reverse transcription was performed using high capacity cDNA reverse transcription kit (Fisher scientific, France) according to the manufacturer instructions.
qPCR Quantification
For lncRNA, preamplification reactions were prepared using Applied Biosystems preamplification master mix with 0.1X (100 nM) of each of the primers pairs corresponding to the lncRNAs. 16 preamplification cycles were performed as preconized by the furnisher (50° C. 2 minutes, 96° C. 10 minutes, 40 cycles at 95° C. for 15 seconds and 60° C. for 1 minute). For miRNA amplification, primers were designed using miRprimer tool (PMID: 24472427). Quantitative PCR of biological samples was done in 10 μl total volume with 1μl of cDNA or preamplification product diluted 1/20 in TE buffer (for miRNA and lncRNA respectively), 5 μl of 2× Soadvanced SYBR green PCR master mix (Biorad, USA) and 250 nM of each primer. Cycling conditions were 95° C. for 5-10 min followed by 40 cycles of 95° C. for 10-30 sec and 60° C. 30-60 sec. A melting curve analysis (60° C. to 99° C.) was performed after the thermal profile to ensure specificity in the amplification. Relative expression level was determined against a standard curve realized on a 5 log scale using CFX maestro Software (Biorad,USA).
lncRNA and miRNAs of interest were also quantified using the FIMAP/QMAP platform developed by Quantamatrix (South Korea). Briefly, lncRNAs and miRNAs were amplified by PCR using primers with the same sequences as for qPCR that were chemically modified. Forward primers were phosphorylated in 5′ and Reverse primers biotinylated in 5′. Multiplexed PCR reactions were performed on Biorad T100 thermocycler in 20 μl reactions containing 2 μl of miRNA or lncRNA cDNA, 250 nM of each primers and 10 μl of 2X One Taq Hot Start 2X master Mix (New England Biolabs, USA) with the following conditions: 30 seconds at 94° C., 25 cycles of 30s at 94° C., 1 min at 60° C. and 1 min at 68° C., followed by a final extension at 68° C. for 5 minutes. PCR products were then digested with 25 U of lambda exonuclease for 30 minutes at 37° C. to eliminate the phosphorylated strand and keep only the biotinylated ones. So digested products were quantified on QMAP using coded silica microdisks coated with oligos complementary to the RNAs of interest (one code per target oligo). Briefly, biotinylated PCR products were incubated with the coated microdisks in a 96 well plate, and hybridized products revealed by addition of a fluorescent streptavidin conjugate, SAPE (Prozyme, Denmark) after washing steps. Plates were imaged on QMAP that takes 2 images per spot: one dark image to quantify the fluorescence and 1 white light image to read the microdisks codes. Each target relative expression is then calculated by QMAP software by assigning fluorescence signal to each target according to the associated microdisk code.
Profiling of 2083 miRNAs based on the miRbase v20 was performed on plasma samples from patients suffering of mild impairment or moderate Alzheimer disease and cognitively intact healthy subjects. After normalization of the results generated with HTG EdgeSeq technology, averages of 406 miRNAs par sample were positively identified. From these 406 miRNAs identified, statistical analysis was performed in order to identify miRNAs with predictive value of Alzheimer disease.
Random Forest algorithm was used for the classification model. From this, several sets of 2 to 20 miRNAs with high predictive value were identified and provide an AUC ranged from 0.839 to 0.793 with an accuracy over 0.77, specificity over 0.77 and sensitivity superior to 0.74.
A list of miRNAs useful for diagnosis of Alzheimer disease was generated. This list was determined using multiple statistical analyses: individual AUC, fold change and t-test for p value (Tables 2, 3 and 4). 74 miRNAs were selected from plasma lithium heparin samples, from the data set with a p value under 0.05. They were further analyzed based on fold change ≥1.49 or ≤0.75 and individual AUC >0,6 or <0,4. See Tables 2, 3 and 4.
Profiling 127,802 transcripts based on LNCipedia v5.2 was performed on serum samples from patients suffering of Alzheimer disease and cognitively intact healthy subjects using total-RNAseq technology. Above 127,802 transcripts, 19867 lncRNAs were identified with expression level of over 10 CPM in at least half the samples of one group.
Out of 19867 lncRNAs, 1008 lncRNAs useful for diagnosis of Alzheimer disease was selected based on the statistical significance (p value <0.05, Wilcoxon test). Random Forest algorithm was used for the classification model. From this, several sets of 2-20 lncRNAs with high predictive value were identified. The sets of selected 2-20 lncRNAs provide an AUC ranged from 0.7 and up to 1 with an accuracy, sensitivity and specificity ranged from 0.7 to 1.
This application claims the benefit and priority of U.S. provisional application 62/554,427, filed on Sep. 5, 2017.
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/EP2018/073905 | 9/5/2018 | WO | 00 |
Number | Date | Country | |
---|---|---|---|
62554427 | Sep 2017 | US |