Method for diagnosing pitt-hopkins syndrome

Abstract

The invention relates to a method for diagnosing Pitt-Hopkins Syndrome (PTHS) using a single test, which comprises the steps of: a) providing an isolated biological sample; b) extracting and processing nucleic acids from the isolated biological sample; c) carrying out real-time PCR (qPCR) analysis on the extracted and processed sample; d) determining in the biological sample, using qPCR, the expression level of at least 4 of 13 genes selected according to the calculation of the number of copies; e) analysing the expression levels of the genes using machine learning; and f) determining the presence of PTHS according to the result obtained from at least four genes, wherein the 4-13 genes are selected from the group consisting of GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1PR5, LPAR1, CHRM4 and GRIN2A.

Description

[0001] METHOD FOR DIAGNOSING PITT HOPKINS SYNDROME

[0002] DESCRIPTIVE MEMORANDUM FIELD OF THE INVENTION

[0003] The present invention relates to the disease diagnostics industry. In particular, the present invention relates to a method for diagnosing Pitt-Hopkins Syndrome (PTHS) through a single-assay test kit that simplifies the diagnosis of this condition.

[0004] In particular, it relates to a method for the diagnosis of PTHS where the method is designed to perform the diagnosis from an isolated biological sample of blood or buccal swab, determine the level of expression based on the calculation of the number of copies of each of at least 4 genes in blood or 13 genes in mucosa selected in said biological sample by real-time PCR (qPCR), perform analysis of the expression levels of said genes by machine learning and determine the presence of PTHS, achieving improved diagnostic effectiveness by means of a single assay / test and with a predictive value greater than 80%.

[0005] STATE OF THE ART

[0006] Pitt-Hopkins syndrome (PTHS) is a neurodevelopmental disorder recognized as a syndromic form of autism spectrum disorder (ASD). It is primarily characterized by moderate to severe intellectual disability, significant developmental delays, and multisystem dysfunctions, including epilepsy, motor dysfunctions, and respiratory abnormalities such as episodic hyperventilation followed by apnea. Additionally, patients present with distinctive facial features, such as a prominent forehead, deep-set eyes, a broad nasal bridge, and a protruding upper lip. PTHS is caused by heterozygous translocations, mutations, or deletions in the TCF4 gene, a transcription factor located on chromosome 18q21. The prevalence of this syndrome is estimated to be between 1 in 34,000 and 1 in 41,000 live births.

[0007] The TCF4 gene encodes a transcription factor belonging to the basic helix-loop-helix (bHLH) protein family. This transcription factor is essential for regulating gene expression during embryonic development and in adulthood, with prominent expression in the brain, intestine, muscles, gonads, kidneys, and glands. TCF4 plays a crucial role in neurogenesis, neuronal differentiation, and the formation of neural circuits—processes that are fundamental for the development and normal function of the nervous system (Espinoza et al., 2024).

[0008] TCF4 exerts its function by binding to specific DNA motifs called E-box sites, present in the promoter or enhancer regions of target genes, thereby regulating their transcription. Furthermore, TCF4 forms heterodimers with other transcription factors of the bHLH family, amplifying its ability to regulate a broad spectrum of genes. TCF4 dysfunction, such as that caused by mutations or deletions, leads to the disruption of numerous biological processes, including tumor suppression, myelination, neuronal development, and nerve function.

[0009] In the context of the nervous system, TCF4 is involved in the regulation of genes crucial for synoptic plasticity, neuronal excitability, and the formation of cortical layers. Loss of TCF4 function disrupts the proliferation and differentiation of neuronal progenitor cells (Espinoza et al., 2024), which can result in abnormal cortical development, including cortical thinning and synoptic dysfunction. These alterations are reflected in cognitive and behavioral deficits, common features in autism spectrum disorders and Pitt-Hopkins syndrome.

[0010] The diagnostic process for Pitt-Hopkins syndrome (PTHS) is complex and requires a multidisciplinary evaluation that includes both the identification of specific clinical characteristics and genetic confirmation. Initially, clinicians must look for a combination of severe intellectual disability, significant developmental delays, and distinctive facial features such as a prominent forehead and a broad nasal bridge, along with other symptoms such as respiratory and motor dysfunctions. Currently, the diagnosis of Pitt-Hopkins syndrome (PTHS) is primarily based on a behavioral assessment, ruling out other conditions. This assessment also considers a series of additional complementary tests depending on the symptoms. These additional tests include electroencephalography (EEG), hearing tests, visual tests, and neuroimaging, among others.Once these signs are identified, genetic analysis is recommended, starting with chromosomal microarray testing to detect chromosomal imbalances associated with the syndrome. If this analysis is negative, exorna sequencing or specific panels are performed to detect mutations in the TCF4 gene. In the absence of these technologies, Sanger sequencing and deletion or duplication analysis using MLPA can be used. Confirmation of a pathogenic variant in TCF4 allows for a definitive diagnosis, and in some cases, genetic testing of the parents is recommended to confirm the de novo occurrence of the mutation. However, the aforementioned evaluations, analyses, and tests have the drawback of being largely inconclusive, as they are based on observations that may be better considered at later ages.This late diagnosis contrasts with the fact that the earlier the diagnosis is made, the earlier an intervention can be started and the more successful it will be.

[0011] On the other hand, in some patients diagnosed with clinical features consistent with PTHS, no direct defects in the TCF4 gene are observed using conventional methods such as next-generation sequencing (NGS) or chromosomal microarrays. However, structural alterations such as translocations, duplications, or chromosomal imbalances can significantly impact TCF4 function, affecting the regulation of genes under its control. These functional abnormalities, although clinically relevant, cannot be detected by the aforementioned techniques, thus limiting their diagnostic utility in this context.

[0012] The method proposed in the present invention represents an innovative solution for evaluating the expression of genes regulated by TCF4, thus providing an indirect functional analysis of their activity. TCF4, a key transcription factor in neurocognitive development, regulates the expression of a network of genes involved in essential processes such as neurogenesis, synaptic plasticity, and cortical maturation. Alterations in this gene network can indicate TCF4 dysfunction, even in the absence of detectable mutations at the genomic level.

[0013] By measuring the expression levels of multiple target genes controlled by TCF4, this technology offers a distinct advantage: it allows for the identification of functional alterations caused by chromosomal imbalances or structural disruptions at the TCF4 locus that impact its regulatory capacity. For example, recent studies have shown that duplications in 18q21, the region that includes TCF4, can be associated with neurological phenotypes by altering the gene dosage of the transcription factor.

[0014] In this respect, the approach of the present invention offers a functional framework that transcends the limitations of sequencing and microarrays. By detecting alterations in specific genes regulated by TCF4, the present method not only more accurately diagnoses TCF4-associated dysfunction but also provides a tool for understanding the molecular consequences of these alterations. This results in a significant diagnostic improvement, especially for patients in whom the underlying genetic etiology cannot be determined by standard methods.

[0015] This advance reinforces the usefulness of regulated gene network analysis as a comprehensive approach to diagnosing neurodevelopmental conditions, while highlighting the relevance of assessing the functional impact of genomic alterations beyond the direct identification of mutations.

[0016] In the prior art, several solutions have been found that partially resolve the technical problem posed. Among the known solutions is US Patent No. 10,002,230 B2, which discloses a method for screening, diagnosing, or prognostically assessing a pediatric population with autism. More specifically, the invention provides a method for performing a weighted test of genes and characteristics associated with autism from a biological sample. This test analyzes the level of gene expression associated with an analyte from at least 20 genes using a specialized reference database that includes individuals with typical neurodevelopment and those with autism spectrum disorder.The expression level of each gene is statistically normalized, and a weighted genetic signature matrix is ​​prepared. A weighted expression level for each gene is calculated using bioinformatics analysis tools, and the divergence of the expression level of each weighted gene with respect to the reference database is established. However, this document does not disclose a method that uses fewer than 20 genes, unlike the method of the present invention, which uses between 4 and 13 genes to establish a diagnosis. Furthermore, this document does not specifically address PTHS syndrome but rather focuses generally on autism spectrum disorder (ASD). None of the genes analyzed in this document correspond to any of the genes analyzed in the present invention. The method covers a broad range of genes for the diagnosis of ASD in general, but it lacks the specificity for a specific diagnosis such as Pitt Hopkins syndrome.Finally, the document states that using sets of 20, 25, and 30 genes, a correct classification of at least 70%, 75%, and 80%, respectively, can be obtained. This result is less efficient than that obtained in the present invention, which, with a set of only 4 genes analyzed in blood, achieves an 86% correlation with patients already diagnosed by traditional methods, and with only 13 genes analyzed in a mucosal sample, achieves an 84% correlation with patients already diagnosed by traditional methods. Furthermore, the qPCR-based method of the present invention offers faster results and lower costs than the microarray-based or high-throughput sequencing method of the aforementioned patent. This is because qPCR is generally more economical and accessible than massive sequencing methods.The qPCR in the method of the present invention allows for simpler implementation in routine clinical laboratories, compared to more complex and expensive methods such as massive sequencing used in the noted patent.

[0017] Another document is the publication of patent application WO / 2020 / 018461, which discloses methods and uses related to the diagnosis, treatment, prevention or improvement of the symptoms of neurodegenerative diseases such as Alzheimer's disease (AD), dementia, age-related dementia, Parkinson's disease (PD), autism spectrum disorder (ASD), among others.In the method, a biological sample is obtained from a subject, the number of copies, the expression level, or the activity level of one or more targets (genes) is determined, and the number of copies, the expression level, or the activity level of said one or more targets detected in steps b) is compared with the number of copies, the expression level, or the activity level of said one or more targets in a control, wherein a significant increase and / or decrease in the number of copies, expression level, or activity level of one or more targets in the sample in question with respect to the control copy number, expression level, or activity level of one or more targets indicates that the subject suffers from a neurological disease or is at risk of developing a neurological disease, but this document does not disclose a method specifically for PTHS, nor does it indicate that the diagnosis can be made using a set of 4-13 genes to make the diagnosis.None of the genetic targets used in this document correspond to any of the genes analyzed in the present invention. Furthermore, in said document, the main origin of the biological sample is cerebrospinal fluid (CSF) and / or interstitial fluid (ISF), and although it discloses that it may be blood, there is no indication that the samples may come from buccal or mucosal swabs.

[0018] Also included is patent application US20210284961 A1, which describes methods for identifying changes in gene expression and mutations in neural organoids to identify neural networks that predict the onset of autism and associated comorbidities. The document discloses a plurality of biomarkers as a diagnostic panel to predict the risk of developing autism in humans. It describes the development of a human neural organoid model from an individual's induced pluripotent stem cells (PSCs) derived from skin or blood cells, which allows for the identification of autism markers in early developmental stages, including at birth. Among the genes mentioned in the document are TCF4, GRIN2A, ALDH1 A3, GLRA2, RXRG, and OXTR, but the document does not specifically describe a particular set of these genes for diagnostic purposes.This document does not disclose that its development allows for the diagnosis of PTHS, nor the gene assay / test (kit) of the present invention. Furthermore, the qPCR-based method of the present invention offers faster results and lower cost than a method based on the development of a neural organoid model as described in that document. In addition, the qPCR of the present invention allows for simpler implementation than organoid generation in routine clinical analyses. SOLUTION TO THE TECHNICAL PROBLEM.

[0019] To address the problem raised, an alternative method for diagnosing Pitt Hopkins Syndrome (PTHS) is presented, improving the effectiveness of the diagnosis through a single trial / test and with a correlation with patients already diagnosed by traditional methods of over 80%.

[0020] SUMMARY DESCRIPTION OF THE INVENTION

[0021] The present method for the diagnosis of PTHS comprises the steps of: a) providing a biological sample of blood or buccal mucosa;

[0022] b) perform extraction and processing of nucleic acids from the isolated biological sample;

[0023] c) perform a real-time PCR (qPCR) analysis on the extracted and processed sample;

[0024] d) determine in said biological sample the level of expression through qPCR of at least a number of 4 to 13 genes selected based on the calculation of copy number;

[0025] e) perform analyses of the expression levels of these genes using machine learning,- and

[0026] f) Determine the presence of PTHS according to the result obtained. g) In a preferred configuration for detection in buccal mucosa samples, the 13 selected genes correspond to: GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4 and GRIN2A.

[0027] h) In a preferred configuration for detection in blood samples, the 4 selected genes correspond to: GLRA1, ALDH2, RXRG1 and S1 PR5.

[0028] DESCRIPTION OF THE FIGURES

[0029] Figure 1 shows the workflow with the stages of the method to carry out the diagnosis of PTHS.

[0030] Figure 2 shows a graph with the results of machine learning analysis using different methods for the qPCR data obtained for the 13 genes analyzed in buccal mucosa samples. The X-axis indicates the techniques applied: PLS-DA: partial least squares discriminant analysis, XGB: Extreme Gradient Boosting, SVM: Support Vector Machine, KNN: K-Nearest Neighbors, and LDA: Linear Discriminant Analysis. Pretreatments used: VarSel (VR>11): Variable selection based on the Regression Vector (VR), VarSel (VIP>1): Variable selection using the Variable Importance in Projection (VIP), VarSel (SR>0): Variable selection based on the Selectivity Ratio (SR), and AE: Autoscaling.

[0031] Figure 3 shows a table indicating the results of a sequential removal analysis of genes from least to most important according to VR (PLS-DA) with the 13 genes for the analyses generated from buccal mucosa samples. NER: No error rate; LV: number of latent variables; VN: true negatives; FP: false positives; VP: true positives; FN: false negatives; Sp: specificity (VN / (VN+FP)); Se: sensitivity (VP / (VP+FN)).

[0032] Figure 4 shows a graph with the results of machine learning analysis using different methods for the qPCR data obtained for the 4–13 genes analyzed in blood samples. The X-axis indicates the techniques applied: PLS-DA: partial least squares discriminant analysis, XGB: Extreme Gradient Boosting, SVM: Support Vector Machine, KNN: K-Nearest Neighbors, and LDA: Linear Discriminant Analysis. The pretreatments used were: VarSel (Score > 0): variable selection based on a score greater than zero, and AE: Autoscaling.

[0033] Figure 5 shows a table indicating the results of a sequential gene removal analysis, ranked from least to most important, using the XGB method with the 13 genes for the analyses generated from blood samples. NER: No error rate; LV: number of latent variables; TN: true negatives; FP: false positives; TP: true positives; FN: false negatives; Sp: specificity (TN / (TN+FP)); Se: sensitivity (TP / (TP+FN)).

[0034] Figure 6 shows the qPCR results for the 13 genes GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1PR5, LPAR1, CHRM4, and GRIN2A, normalized by the 18S rRNA gene, from a control group patient using buccal mucosa samples. The qPCR results shown are the average of three measurements.

[0035] Figure 7 shows the qPCR results for the 13 genes GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1PR5, LPAR1, CHRM4, and GRIN2A, normalized by the 18S gene, from a patient with Pitt-Hopkins syndrome using a buccal mucosa sample. The qPCR results shown are the average of three measurements. Figure 8 shows the result of the machine learning analysis with the prediction associated with class 1, that is, a patient without Pitt-Hopkins syndrome.

[0036] Figure 9 shows the result of the machine learning analysis with the prediction associated with class 2. That is, a patient with Pitt-Hopkins syndrome.

[0037] Figure 10 shows the qPCR results for the four genes normalized by the 18S rRNA gene: GLRA1, ALDH2, RXRG1, and S1PR5, from a control group patient using a blood sample. The qPCR results shown are the average of three measurements.

[0038] Figure 11 shows the qPCR results for the four genes normalized by the 18S rRNA gene: GLRA1, ALDH2, RXRG1, and S1PR5, from a blood sample of a patient with Pitt-Hopkins syndrome. The qPCR results shown are the average of three measurements.

[0039] Figure 12 shows the result of the machine learning analysis with the prediction associated with class 1. That is, a patient without Pitt-Hopkins syndrome.

[0040] Figure 13 shows the result of the machine learning analysis with the prediction associated with class 2. That is, a patient with Pitt-Hopkins syndrome. DETAILED DESCRIPTION OF THE INVENTION

[0041] The present invention relates to an alternative method for diagnosing Pitt-Hopkins syndrome (PTHS) by qPCR analysis of at least 4 of 13 selected genes related to the pathophysiology of PTHS. The diagnosis is performed using a PCR assay kit that incorporates primers to quantify the expression of these at least 4 genes. The assay / test was developed by analyzing blood samples and buccal mucosal swabs from patients diagnosed with Pitt-Hopkins syndrome in Chile.

[0042] The procedure, once the blood or buccal mucosa sample has been collected, comprises the following steps: (1) nucleic acid extraction; (2) DNase treatment; (3) cDNA synthesis; and (4) real-time PCR (qPCR).

[0043] The present development also includes a reaction plate, where the starters or primers for qPCR are located.

[0044] The at least 4 of 13 genes analyzed by qPCR and the primers developed are defined in the following Table A: Table A: List of the 13 selected genes and their primers

[0045]

[0046] The analysis of data generated from qPCR is based on calculating the number of gene copies detected. This copy number is expressed in relation to the 18S reference gene. Once the copy number is obtained, patients are compared with control subjects.

[0047] Statistically significant differences were found in some genes. For example, the expression of the GRIN2A and S1 PR5 genes in buccal mucosa samples was distinctly increased in diagnosed patients compared to controls. However, the remaining 11 genes analyzed in buccal mucosa and the 13 genes analyzed from blood samples also showed potential differences. For this reason, a machine learning analysis was performed using different methods, such as the PLS-DA (partial least squares discriminant analysis) classification technique applied to the analysis of mucosal samples.

[0048] In summary, the qPCR method used, followed by machine learning analysis, begins with a comparison to an internal control without a sample. This control represents a baseline against which the expression of each gene is determined. There is no established minimum value; instead, a baseline is established in each run, and the expression level of each gene is determined against it. Some genes do not amplify, and their expression value is recorded as zero. The implication of this is that zero values ​​also provide information used by the machine learning analysis to make predictions. In other words, the absence of gene expression is also relevant information that is determined in each run compared to the baseline and then imputed as zero.

[0049] Analysis method applied to mucosa

[0050] qPCR data from 13 genes were used: GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4, and GRIN2A. Values ​​were normalized by the 18S gene.

[0051] A data matrix was generated in Excel 2406 software, organizing the patients in the rows and the normalized genes in the columns. The values ​​tabulated in each cell corresponded to those determined by qPCR and subsequently normalized. Values ​​below the limit of detection were replaced by 0. The matrix also included the class (Pitt-Hopkins Syndrome or control) of each patient.

[0052] The PLS-toolbox 9.3.1 software (Eigenvector Research, Inc) was used in Matlab R2024a (MathWorks, Inc). The data matrix in .xlsx format was imported into PLS-toolbox using the software's import tool.

[0053] The data were analyzed without pretreatment using the PLS-DA (partial least squares discriminant analysis) machine learning classification technique. Given the available sample size, the internal validation strategy "cross-validation leaving one out" was selected.

[0054] The number of latent variables that minimized classification error in internal validation was selected. This number was 13.

[0055] The presence of anomalous data was assessed using the graph "Reduced Residual Q vs. T 2 "Reduced Hotelling." The presence of anomalous data was ruled out.

[0056] The confusion table was evaluated during training and internal validation, and the model's no-error rate was calculated. Values ​​of 100% and 84% were obtained, respectively.

[0057] Method of analysis applied to blood

[0058] qPCR data from 4 genes were used: GLRA1, ALDH2, RXRG1, S1 PR5. Values ​​were normalized by the 18S gene.

[0059] A data matrix was generated in Excel 2406 software, organizing the patients in the rows and the normalized genes in the columns. The values ​​tabulated in each cell corresponded to those determined by qPCR and subsequently normalized. Values ​​below the limit of detection were replaced by 0. The matrix also included the class (Pitt-Hopkins Syndrome or control) of each patient.

[0060] The PLS-toolbox 9.3.1 software (Eigenvector Research, Inc) was used in Matlab R2024a (MathWorks, Inc).

[0061] The data matrix in .xlsx format was imported into PLS-toolbox using the software's import tool.

[0062] The data were analyzed without pretreatment using the XGBoost (extreme gradient boosting) machine learning classification technique. Given the available sample size, the "cross-validation leaving one out" internal validation strategy was selected. The following possible values ​​were selected for the parameters max_depth, num_round, and eta: [1, 2345], [5075, 100, 125150], and [0.1, 0.3, 0.5, 0.7, 0.9]. The values ​​automatically optimized by the software were 2, 50, and 0.9, respectively.

[0063] The confusion table was evaluated during training and internal validation, and the model's no-error rate was calculated. Values ​​of 100% and 86% were obtained, respectively.

[0064] Definitions

[0065] In the present invention, a biological sample is understood to be any material obtained from a living organism, in this particular case, through buccal swab collection techniques or venous blood sampling. These samples contain cells or biological fluids that allow for the extraction of ribonucleic acid (RNA) for subsequent analysis using quantitative polymerase chain reaction (qPCR). These samples are essential for the detection and quantification of specific genetic material, and are key in molecular diagnostic processes, biomedical research, and the development of new therapies or diagnostic methods.

[0066] In the present invention, a blood sample is defined as a specific volume of venous blood obtained from a living organism by venipuncture. This sample contains blood cells and plasma, which allow for the extraction of ribonucleic acid (RNA) for subsequent analysis using quantitative polymerase chain reaction (qPCR).

[0067] In the present invention, a buccal mucosa sample is defined as epithelial cells collected by buccal swabbing using a sterile swab rubbed in the oral cavity of a living organism. This sample allows for the extraction of ribonucleic acid (RNA) for analysis using the quantitative polymerase chain reaction (qPCR) technique.

[0068] In the present invention, the concept of an isolated sample refers to a biological fraction that has been separated from its original source, either by mechanical, chemical, or biotechnological means, in order to obtain a specific component, such as ribonucleic acid (RNA). In this context, an isolated sample may refer to cells, fluids, or fragments of genetic material extracted from a venous blood or buccal mucosa sample, which are subsequently processed and purified for analysis using quantitative polymerase chain reaction (qPCR). In the present invention, an assay is understood to be an experimental procedure designed to measure or detect a specific substance, molecule, or biological reaction in a biological sample.In this context, the test may include methods such as quantitative polymerase chain reaction (qPCR), intended for the quantification and analysis of nucleic acids, such as RNA, extracted from buccal mucosa or venous blood samples.

[0069] A test is defined as a standardized procedure or series of procedures applied to a biological sample to detect, identify, or quantify a substance or biological activity. In the context of this patent, the test includes the use of techniques such as qPCR to analyze RNA samples extracted from buccal mucosa or venous blood, enabling the precise detection of genetic alterations associated with specific conditions such as Pitt-Hopkins Syndrome or Autism Spectrum Disorder.

[0070] A kit is defined as a set of reagents, tools, and devices packaged and prepared together for use in molecular diagnostic or analytical procedures, along with instructions for their use. In the context of this patent, the kit may include specific reagents for RNA extraction from biological samples, such as buccal swabs or venous blood, along with the components necessary to perform quantitative polymerase chain reaction (qPCR). The kit is designed to facilitate the detection of genetic alterations for diagnostic or research purposes. In the present invention, Pitt-Hopkins syndrome (PTHS) is defined as a rare genetic neurodevelopmental disorder characterized by moderate to severe intellectual disability, delayed motor and language development, distinctive facial features, and respiratory problems such as episodes of hyperventilation and apnea.PTHS is primarily caused by mutations in the TCF4 gene that affect normal brain development. This syndrome is frequently associated with autism spectrum disorders and exhibits significant phenotypic variability. The diagnosis of PTHS can be confirmed through genetic testing, including sequencing techniques.

[0071] In the present invention, ASD, also known as Autism Spectrum Disorder (ASD), refers to a group of neuropsychiatric conditions characterized by persistent difficulties in communication and social interaction, along with restricted and repetitive patterns of behavior, interests, or activities. ASD is a spectrum that encompasses a wide range of manifestations, from mild to severe, and may be associated with intellectual disabilities or higher cognitive abilities. Genetic factors play a significant role in the etiology of ASD, and its diagnosis can be supported by genetic testing.In the present invention, machine learning is understood to be a discipline within the field of Artificial Intelligence, which, through algorithms, provides computers with the ability to identify patterns in massive amounts of data and make predictions (predictive analytics), using iterations and selection of results. This learning allows computers to perform specific tasks autonomously, in other words, without needing to be programmed.

[0072] Different options described for different technical characteristics may be combined with each other, or with other options known to a person normally versed in the subject, without this limiting the scope of the present application.

[0073] In the context of this request, and without limiting its scope, "at least one" shall be understood to mean one or more of the elements referenced. Therefore, the number of elements referenced does not limit the scope of this request. Furthermore, if more than one element is provided, those elements may or may not be identical, without limiting the scope of this request.

[0074] The grammatical articles "a," "an," "the," and "the," as used herein, are intended to include "at least one," "at least one," "one or more," or "one or more," unless the context indicates or requires otherwise. Therefore, the articles are used herein to refer to one or more of the grammatical objects of the article. By way of example, "a component" means one or more components, and thus more than one component may be contemplated and used in an implementation of the invention. Furthermore, the use of a singular noun includes the plural, and the use of a plural noun includes the singular, unless the context of use requires otherwise.

[0075] The use of terms such as: "includes", "which includes", "including", "has", "which has", "having", "contains", "which contains", "containing", "comprising" or "comprising", even incorporating some grammatical equivalents of these, should generally be understood as open and non-limiting, e.g., not excluding additional unmentioned elements or steps, unless explicitly stated or understood otherwise in the described context.

[0076] In the context of this application, without limiting its scope, "plurality" shall mean two or more of the elements referred to. Accordingly, the number of elements of the plurality referred to does not limit the scope of this application as long as it is greater than or equal to two. Furthermore, such elements of the plurality may or may not be identical to one another without limiting the scope of this application. When the term "approximately" or "around" is used before a quantitative value, these teachings also include the specific quantitative value, unless specifically stated otherwise. As used herein, the term "approximately" or "around" refers to a variation of ±10% of the stated nominal value, unless a range is explicitly stated herein.Unless otherwise stated, if the term "approximately" or "around" is mentioned before the first endpoint of a numerical interval, or a set of numbers, regardless of how they are represented (for example, ratios of the type A:B or A / B, where A and B are whole numbers or decimals, among other numerical representations), this term refers to all the numbers mentioned, and in the case of numerical intervals, to both the first and second endpoints of the interval. For example, an interval mentioned as "approximately X to Y" should be read as "approximately X to approximately Y."

[0077] To designate intervals and / or ranges, various expressions can be used, such as "X - Y", "from X to Y", "from X to Y", "from X - Y", "between X and Y", and others used for this purpose.

[0078] Although this application mentions separate modes of implementation, it should be understood that any mode of implementation, and its characteristic features, may be freely combined with any other mode of implementation and its characteristic features, even in the absence of an explicit statement to that effect. It should be understood that the order of the steps or the order in which certain actions are performed is irrelevant as long as these teachings remain operative.

[0079] Furthermore, two or more steps or actions may be performed simultaneously. The use of any and all examples, or exemplary language in this document, such as "as" or "including," is intended solely to better illustrate the disclosure herein and does not limit the scope of the invention unless expressly stated. Nothing in the description should be construed as indicating that any element not claimed is essential to the practice of the invention herein.

[0080] DETAILED MODALITIES:

[0081] The present invention relates to a method for diagnosing Pitt-Hopkins Syndrome (PTHS) through a single-test assay (kit) that simplifies the diagnosis of this condition. In particular, it relates to a method for diagnosing Pitt-Hopkins Syndrome (PTHS) using a single assay, comprising the steps of: a) providing an isolated biological sample;

[0082] b) perform extraction and processing of nucleic acids from the isolated biological sample;

[0083] c) perform a real-time PCR (qPCR) analysis on the extracted and processed sample;

[0084] d) determine in said biological sample the level of expression through qPCR of at least 4 of 13 genes selected based on the calculation of copy number;

[0085] e) perform analyses of the expression levels of these genes using machine learning,- and

[0086] f) determine the presence of PTHS according to the result obtained from at least four genes;

[0087] where the 4-13 genes are selected from the group consisting of GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4 and GRIN2A.

[0088] In a preferred modality, in stage a) the isolated biological sample is blood.

[0089] In another preferred modality in stage d) the expression level is determined based on the calculation of the number of copies of at least 4 of 13 selected genes, where the 13 genes are selected from the group consisting of GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4 and GRIN2A.

[0090] In another preferred modality, in stage a) the isolated biological sample is from buccal mucosa.

[0091] In an additional preferred modality in stage a) the isolated biological sample is blood and / or buccal mucosa.

[0092] In another preferred modality, in stage d) the analysis of gene expression levels is performed using PLS-DA (partial least squares discriminant analysis).

[0093] In another preferred modality, in stage d) the analysis of gene expression levels is performed using XGBoost (extreme gradient boosting). In an additional preferred modality in stage e) the analysis of expression levels was performed by machine learning comprising automatic learning through the identification of patterns in massive data, using iterations and selection of results, to obtain the expression levels of at least 4 genes out of a total of 13 selected genes.

[0094] APPLICATION EXAMPLES EXAMPLES OF PREFERRED EMBODIMENTS

[0095] The following are examples of applications of the present invention. These examples are provided for illustrative purposes only to facilitate a better understanding of the invention, but in no case should they be considered as limiting the scope of the protection sought. Furthermore, specifications of different technical features described in the examples may be combined with each other, or with other technical features previously described, without limiting the scope of the protection sought.

[0096] EXAMPLE 1. Implementation of the diagnostic method and validation of its efficiency. Briefly, the procedure involved the development of primers for qPCR amplification of the genes used, sample collection, RNA extraction, qPCR analysis, and data processing using machine learning methods. Figure 1 describes the workflow and analytical strategy.

[0097] 1.1 Design and validation of specific starters

[0098] Primers were designed for the 13 genes used (GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1PR5, LPAR1, CHRM4, and GRIN2A) and the 18S control gene using the NIH Primer Design Tool. Each sequence was evaluated to ensure specificity and efficiency in qPCR amplification. Details of the primer sequences, sequence identifiers, and amplicon size are shown in Table B. Table B: List of the 13 genes used and details of the designed primers

[0099]

[0100] 1.2 Sample collection and processing

[0101] 1.2.1 Obtaining the sample. Buccal mucosa sample: Epithelial cells were collected by buccal swab using a sterile swab rubbed for 30 seconds in the oral cavity, ensuring sufficient contact with the mucous membranes.

[0102] Blood sample: Venous blood was obtained by puncture, collecting a volume of 3 mL in tubes with EDTA anticoagulant.

[0103] 1.2.2 Initial storage:

[0104] The samples were immediately transferred to dry ice and stored at -80°C until processing.

[0105] In this way, blood or buccal mucosa samples were collected from 6 and 9 patients diagnosed in Chile with Pitt Hopkins syndrome, respectively.

[0106] Blood samples were obtained from 6 patients, while buccal swab samples were obtained from 9 patients. These data were compared with samples obtained from 8 and 10 control subjects, who collectively donated blood and buccal mucosa, respectively.

[0107] 1.3 Preparation of nucleic acids

[0108] Once the blood or buccal mucosa sample was collected, it was subjected to the following procedures:

[0109] 1.3.1 Extraction of nucleic acids and obtaining RNA.

[0110] Total RNA was extracted using the Trizol-based method as detailed below:

[0111] 1. The samples were processed according to their origin. For mucosal samples, the swab was directly inserted into an Eppendorf tube to which 350 pL of TRIzol was added. The swab was repeatedly rotated within the solution to ensure complete dissolution of the sample.

[0112] 2. For whole blood samples, 500 pL of blood were transferred to an Eppendorf tube. Then, 350 pL of TRIzol were added directly onto the sample, using a plastic grinder to thoroughly lyse the contents.

[0113] 3. Once the samples were lysed in TRIzol, 200 pL of chloroform were added. The tube was vigorously shaken for 15 seconds using a vertex mixer.

[0114] 4. The tubes were left to stand for 3 minutes at room temperature to allow phase separation. 5. Subsequently, the samples were centrifuged at 12,000 rpm for 15 minutes at 4 °C, resulting in the formation of three phases: an upper aqueous phase, an interface, and a lower organic phase.

[0115] 6. The aqueous phase (containing the nucleic acids) was carefully removed and transferred to a new Eppendorf tube.

[0116] 7. To the aqueous phase obtained, an equal volume of isopropanol was added, mixing gently until a homogeneous solution was achieved.

[0117] 8. The mixture was left to stand for 10 minutes at room temperature to promote the precipitation of nucleic acids.

[0118] 9. The samples were centrifuged at 12,000 rpm for 10 minutes at 4 °C, which allowed the formation of a visible pellet at the bottom of the tube.

[0119] 10. The supernatant was carefully removed without disturbing the pellet.

[0120] 11. 750 pL of 70% ethanol was added to wash the pellet and mixed gently.

[0121] 12. The washing was followed by centrifugation at 12,000 rpm for 10 minutes at 4 °C. This washing procedure was repeated once more.

[0122] 13. After the last wash, the supernatant was removed with extreme care, leaving the pellet as dry as possible.

[0123] 14. The tubes were left to dry at room temperature for 15 minutes to completely remove any ethanol residue.

[0124] 15. Finally, the dried pellets were resuspended in 30 pL of nuclease-free water preheated to 70 °C. The samples were stored at -80 °C until their subsequent treatment with DNase.

[0125] This procedure ensured the obtaining of high-quality RNA suitable for the subsequent stages of analysis.

[0126] 1.3.2 DNase Treatment

[0127] The BioLabs DNase I enzyme (M0303S) was used for the treatment of total RNA samples according to the following protocol.

[0128] 1. For each 30 pL sample of total RNA, a reaction mixture was prepared with the following components:

[0129] 10X buffer: 3 pLDNase I: 1 pL

[0130] Nuclease-free water: 6 pL

[0131] Total volume per tube: 10 pL

[0132] 2. The prepared mixture was added to each total RNA sample, ensuring a homogeneous mixture.

[0133] 3. The samples were incubated at 37 °C for 1 hour to allow enzymatic digestion of the contaminating DNA.

[0134] 4. At the end of the digestion, 0.5 pL of 0.5 M EDTA was added to each tube to stop the enzyme activity.

[0135] 5. The tubes were heated to 75 °C for 10 minutes, ensuring complete inactivation of DNase.

[0136] 6. Finally, the total RNA was quantified using standard spectrophotometry or fluorometry methods, verifying the purity and concentration of the treated RNA.

[0137] This procedure ensured the efficient removal of contaminating DNA, leaving the RNA free of impurities for subsequent analysis.

[0138] 1.3.3. cDNA preparation.

[0139] In this stage, the Promega M-MLV Reverse Transcriptase enzyme (M1701) was used to perform cDNA synthesis.

[0140] 1. For each sample, the necessary volume of total RNA was calculated to reach a final amount of 2 pg of RNA.

[0141] 2. The volume of each sample was completed with molecular biology quality water until a total of 9 pL was obtained in each tube.

[0142] 3. 1 pL of oligo dT (200 ng / pL) was added to each tube, ensuring a homogeneous mixture.

[0143] 4. The tubes were heated to 70 °C for 5 minutes to denature RNA secondary structures.

[0144] 5. Subsequently, the samples were rapidly cooled in ice to condense and preserve the components, followed by a brief spin in a microcentrifuge to concentrate the contents at the bottom of the tube. 6. A master reaction mixture was prepared for each sample, with the following components:

[0145] MLV Tampon 5X: 5 pl

[0146] dNTPs (10 mM): 2 pL

[0147] Nuclease-free water: 2 pL

[0148] Total mix volume: 9 pL

[0149] 7. 9 pL of the prepared mixture was added to each tube, obtaining a total volume of 19 pL per sample.

[0150] 8. The samples were incubated at 37 °C for 2 minutes to equilibrate the conditions before adding the enzyme.

[0151] 9. One pL of M-MLV Reverse Transcriptase was added to each tube, resulting in a final volume of 20 pL per sample. For a negative control (no reverse transcription), a tube was prepared in which the enzyme was replaced with 1 pL of water.

[0152] 10. The tubes were incubated at 42 °C for 1 hour to allow cDNA synthesis.

[0153] 11. The reaction was stopped by heating the samples to 70 °C for 5 minutes to inactivate the enzyme.

[0154] 12. 20 pL of cDNA were obtained per sample, which were stored or used directly for further analysis, such as real-time PCR.

[0155] This protocol guarantees the efficient and reproducible synthesis of cDNA from total RNA samples, adapted for use in gene expression experiments.

[0156] 1.3.4 Performing real-time PCR (qPCR).

[0157] In this stage, the KAPA SYBR FAST qPCR Master Mix (KK4602) DNA polymerase enzyme was used to perform the qPCR reactions as detailed below:

[0158] 1. For each gene to be analyzed, a master mixture was prepared with the following components, adjusted to the final volume of each well (9 pL):

[0159] KAPA SYBR FAST Master Mix: 5 pL

[0160] Forward primer (10 pM): 0.125 pL (125 nM final concentration) Reverse primer (10 pM): 0.125 pL (125 nM final concentration) Nuclease-free water: 3.75 pL

[0161] 2. A volume of 9 pL of the prepared mix was distributed into the corresponding wells of a qPCR plate.

[0162] 3. A 1:5 dilution of the previously synthesized cDNA was prepared using nuclease-free water.

[0163] 4. 1 pL of the cDNA dilution was added to each well containing the mix, for a final total volume of 10 pL per reaction.

[0164] 5. The plate was carefully sealed with an adhesive film to prevent evaporation and contamination.

[0165] 6. A rapid centrifugation (spin) of the plate was performed to ensure homogeneity of the samples and avoid bubbles.

[0166] 7. The plate was placed in the real-time thermocycler, configuring the specific thermal program for the qPCR reaction, which typically included:

[0167] Initial denaturation: 95 °C for 3 minutes

[0168] Repeated cycle (40 cycles):

[0169] ■ Denaturation: 95 °C for 10 seconds

[0170] ■ Alignment / Extension: Specific temperature for starters (normally 60 °C) for 30 seconds

[0171] Dissociation curve: Temperature increases to verify the specificity of the amplified products (starting at 60°C for 1 minute and ending at 95°C for 15 seconds, increasing 0.1°C per second).

[0172] 8. Data acquisition was carried out during the extension phase to monitor the fluorescence emitted by the SYBR Green.

[0173] This protocol allowed for the relative quantification of gene expression through the analysis of amplification curves, ensuring specificity through the dissociation curve. The data obtained were used to interpret expression levels in different genes analyzed.

[0174] The analysis of the data generated from qPCR is based on calculating the number of gene copies detected. This copy number is expressed relative to the 18S reference gene. Once the copy number is obtained, patients are compared with control subjects. 1.4 Data Analysis

[0175] The data were analyzed according to the nature of the sample, as detailed below:

[0176] 1.4.1 Analysis method applied to oral mucosa

[0177] qPCR data from 13 genes were used: GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4, and GRIN2A. Values ​​were normalized by the 18S gene.

[0178] For each sample, qPCR analyses were performed at least three times independently.

[0179] A data matrix was generated in Excel 2406 software, organizing the patients in the rows and the normalized genes in the columns. The values ​​tabulated in each cell corresponded to those determined by qPCR and subsequently normalized. Values ​​below the limit of detection were replaced by 0. The matrix also included the class (Pitt-Hopkins Syndrome or control) of each patient.

[0180] The PLS-toolbox 9.3.1 software (Eigenvector Research, Inc) was used in Matlab R2024a (MathWorks, Inc).

[0181] The data matrix in .xlsx format was imported into PLS-toolbox using the software's import tool.

[0182] The data were analyzed without pretreatment using the PLS-DA (partial least squares discriminant analysis) machine learning classification technique. Given the available sample size, the internal validation strategy "cross-validation leaving one out" was selected.

[0183] The number of latent variables that minimized classification error in internal validation was selected. This number was 13.

[0184] The presence of anomalous data was assessed using the graph "Reduced Residual Q vs. T 2 "Reduced Hotelling." The presence of anomalous data was ruled out.

[0185] The confounding table was evaluated during training and internal validation, and the model's no-error rate was calculated. Values ​​of 100% and 84% were obtained, respectively. 1.4.2 Analysis method applied to blood

[0186] qPCR data from 4 genes were used: ALDH2, S1 PR5, RXRG1, GLRA1. Values ​​were normalized by the 18S gene.

[0187] For each sample, qPCR analyses were performed at least three times independently.

[0188] A data matrix was generated in Excel 2406 software, organizing the patients in the rows and the normalized genes in the columns. The values ​​tabulated in each cell corresponded to those determined by qPCR and subsequently normalized. Values ​​below the limit of detection were replaced by 0. The matrix also included the class (Pitt-Hopkins Syndrome or control) of each patient.

[0189] The PLS-toolbox 9.3.1 software (Eigenvector Research, Inc) was used in Matlab R2024a (MathWorks, Inc).

[0190] The data matrix in .xlsx format was imported into PLS-toolbox using the software's import tool.

[0191] The data were analyzed without pretreatment using the XGBoost (extreme gradient boosting) machine learning classification technique. Given the available sample size, the "cross-validation leaving one out" internal validation strategy was selected.

[0192] The following possible values ​​were selected for the parameters max_depth, num_round, and eta: [1 2345], [5075 100 125150], and [0.1, 0.3, 0.5, 0.7, 0.9]. The values ​​automatically optimized by the software were 2, 50, and 0.9, respectively.

[0193] The confusion table was evaluated during training and internal validation, and the model's no-error rate was calculated. Values ​​of 100% and 86% were obtained, respectively.

[0194] 1.5 Results

[0195] 1.5.1 Analysis of buccal mucosa samples

[0196] Table C presents the analytical methods, parameters evaluated, and results obtained in the analysis of buccal mucosa samples for the 13 genes evaluated by qPCR. The techniques applied were PLS-DA: partial least squares discriminant analysis, XGB: Extreme Gradient Boosting, SVM: Support Vector Machine, KNN: K-Nearest Neighbors, and LDA: Linear Discriminant Analysis.

[0197] The pretreatments used were VarSel (VR>|11): Variable selection based on the Regression Vector (VR), VarSel (VIP>1): Variable selection using the Variable Importance in Projection (VIP), VarSel (SR>0): Variable selection based on the Selectivity Ratio (SR) and AE: Autoscaling.

[0198] The best-performing machine learning analysis was obtained by applying the PLS-DA (Partial Least Squares Discriminant Analysis) classification technique without data pretreatment. Given the available sample size, the internal validation strategy of "cross-validation leaving one out" was used.

[0199] Table C. Details of the results of machine learning analysis by different methods for the qPCR data obtained for the 13 genes analyzed in buccal mucosa samples, with NER being the no error rate. LV: number of latent variables; VN: true negatives; FP: false positives; VP: true positives; FN: false negatives; Sp: specificity (VN / (VN+FP)); Se: sensitivity (VP / (VP+FN)).

[0200] >

[0201] > >

[0202]

[0203] Figure 2 shows the results obtained by applying different machine learning methods to the qPCR data of the 13 genes in buccal mucosa samples. Based on the data summarized in Table C and the graphical results in Figure 2, it is concluded that PLS-DA without data pretreatment was the most efficient method for identifying patients with Pitt-Hopkins syndrome (PTHS), achieving 84% accuracy, measured by the NER (No Error Rate), which corresponds to the proportion of correct predictions made by the model out of the total number of predictions.

[0204] Subsequently, a sequential gene removal analysis was performed to assess whether it was possible to omit any of the 13 selected genes. This analysis consisted of iteratively repeating the model while excluding each gene, evaluating the impact on the model's accuracy. Figure 3 details the results of this analysis, ordering the genes according to their relative importance based on the VR (Variable Importance in PLS-DA). The results show that removing any gene decreases the model's accuracy, leading to the conclusion that all genes are necessary to maintain optimal performance.

[0205] According to the results presented in Figure 3, removing any of the 13 genes from the analysis impacts the accuracy of the result, decreasing the 84% accuracy found when incorporating all 13 genes.

[0206] 1.5.2 Blood sample analysis

[0207] Table D presents the analytical methods, parameters evaluated, and results obtained in the analysis of blood samples for the 4 to 13 selected genes. In this case, the best performance was obtained by applying the XGBoost (Extreme Gradient Boosting) classification technique without data pretreatment. As in the analysis of mucosal samples, the internal validation strategy of "cross-validation leaving one out" was used.

[0208] Table D. Details of the machine learning analysis results using different methods for the qPCR data obtained for the 4–13 genes analyzed in blood samples. The techniques applied were PLS-DA (partial least squares discriminant analysis), XGB (extreme gradient boosting), SVM (support vector machine), KNN (k-nearest neighbors), and LDA (linear discriminant analysis). The pretreatments used were VarSel (score > 0): variable selection based on a score greater than zero, and AE (autoscaling). LV: number of latent variables; TN: true negatives; FP: false positives; VP: true positives; FN: false negatives; Sp: specificity (TN / (TN+FP)); Se: sensitivity (VP / (VP+FN)).

[0209]

[0210] >

[0211] >

[0212] >

[0213]

[0214] Figure 4 shows the results of machine learning analyses using different methods for the qPCR data obtained for the 4 to 13 genes analyzed in blood samples. Details of these analyses are listed in Table D.

[0215] Based on Table D and the graph in Figure 4, it can be concluded that the XGB method performed best in identifying patients with PTHS. This method achieved 86% accuracy, as evidenced by the NER (no-error rate) metric, which refers to the proportion of correct predictions made by the model out of the total number of predictions.

[0216] Figure 5 describes the analysis based on the best model obtained with the 13 genes, performing a sequential removal of genes from least to most important according to the XGB method for analyses generated from blood samples. According to the analysis presented in Figure 5, it is concluded that there are four genes that, when retained in the analysis, allow for a classification with 86% accuracy.

[0217] EXAMPLE 2. Application of the diagnostic method developed to predict Pitt-Hopkins Syndrome from 13 genes determined in mucosa obtained by buccal swab by qPCR.

[0218] This example describes the application of the diagnostic method implemented to predict Pitt-Hopkins Syndrome (PTHS) using data obtained from oral mucosa of a control subject and a patient previously diagnosed with PTHS, using qPCR and analyzed using machine learning.2.1 Provision of isolated biological sample.

[0219] Buccal mucosa samples were collected using a sterile swab rubbed against the individual's oral cavity for 30 seconds, ensuring sufficient contact with the mucosa. The samples were immediately stored at -80 °C until processing.

[0220] 2.2 Performing extraction and processing of nucleic acids from the isolated biological sample.

[0221] 2.2.1 RNA Extraction and Preparation.

[0222] Total RNA was extracted as detailed in Example 1, using the TRIzol-based protocol, followed by DNase treatment to remove DNA contaminants. The RNA obtained was quantified and verified for purity and quality.

[0223] 2.2.2 Synthesis of complementary DNA (cDNA).

[0224] The treated RNA was reverse transcribed to cDNA using the M-MLV Reverse Transcriptase enzyme as detailed in Example 1. The cDNA products were stored at -80 °C for further analysis by qPCR.

[0225] 2.3 Performing real-time PCR (qPCR) analysis on the extracted and processed sample.

[0226] qPCR analysis was performed, using the specific primers, as defined in Example 1, for 13 selected genes (GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4, GRIN2A) normalized against the 18S reference gene. The KAPA SYBR FAST DNA polymerase enzyme and a standard qPCR protocol were used, as detailed in Example 1.

[0227] 2.4 Determination by qPCR of the expression level in said biological sample of the 13 selected genes based on the calculation of the number of copies obtained.

[0228] The expression level of each gene was determined based on the number of gene copies detected by qPCR. This number of copies is expressed in relation to the 18S reference gene.

[0229] For the individuals analyzed in the present example, the expression level of the 13 genes analyzed by qPCR and normalized by the 18S gene are shown in Figure 6 (control subject) and Figure 7 (patient with Pitt-Hopkins Syndrome), respectively.2.5 Performing analysis of the expression levels of these genes using machine learning.

[0230] 2.5.1 Preparation of data for the patient with an unknown diagnosis

[0231] The expression data for the 13 selected genes—GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1PR5, LPAR1, CHRM4, and GRIN2A—were pooled. The data were normalized against the 18S reference gene and organized into an Excel file according to the required nomenclature. Values ​​below the limit of detection were replaced with 0. This file was saved for later analysis.

[0232] 2.5.2 Predictive model load

[0233] The predictive model, for these purposes named “SPH_Mucosa_PLSDA13.mat”, was loaded into the MATLAB working environment and opened using the PLS-toolbox software. This file contains the PLS-DA-based model previously trained and validated with known samples in Example 1.

[0234] 2.5.3 Importing patient data

[0235] The data was imported into the PLS toolbox from the prepared Excel file. During the import, the cells were configured to distinguish between labels and normalized expression data.

[0236] 2.5.4 Prediction configuration and analysis

[0237] The predictive model was used to classify the patient based on the provided data. The analysis yielded a result indicating whether the patient belongs to the control group (Class 1) or presents PTHS (Class 2).

[0238] • In the control subject of the present example, the model predicted Class 1, confirming that the patient did not have PTHS (see Figure 8).

[0239] • In the second case, using data from a patient with confirmed PTHS, the model predicted Class 2, confirming the presence of the syndrome (see Figure 9).

[0240] EXAMPLE 3. Application of the Diagnostic Method developed to predict Pitt-Hopkins Syndrome from 4 genes determined in blood by qPCR. This example describes the application of the diagnostic method implemented to predict Pitt-Hopkins Syndrome (PTHS) using data obtained from venous blood of a control subject and a patient previously diagnosed with PTHS by qPCR and analyzed by machine learning.

[0241] 3.1 Provision of isolated biological sample.

[0242] Three milliliters of venous blood were collected from an individual in tubes containing EDTA anticoagulant. The samples were immediately stored at -80 °C until processing.

[0243] 3.2 Performing extraction and processing of nucleic acids from the isolated biological sample.

[0244] 3.2.1 RNA Extraction and Preparation.

[0245] Total RNA was extracted as detailed in Example 1, using the TRIzol-based protocol, followed by DNase treatment to remove DNA contaminants. The RNA obtained was quantified and verified for purity and quality.

[0246] 3.2.2 Synthesis of Complementary DNA (cDNA)

[0247] The treated RNA was reverse transcribed to cDNA using the M-MLV Reverse Transcriptase enzyme as detailed in Example 1. The cDNA products were stored at -80 °C for further analysis by qPCR.

[0248] 3.3 Performing real-time PCR (qPCR) analysis on the extracted and processed sample.

[0249] qPCR analysis was performed, using the specific primers, as defined in Example 1, for 4 selected genes (GLRA1, ALDH2, RXRG1, S1 PR5) normalized against the 18S reference gene. The KAPA SYBR FAST DNA polymerase enzyme and a standard qPCR protocol were used, as detailed in Example 1.

[0250] 3.4 Determination by qPCR of the expression level in said biological sample of the 4 selected genes based on the calculation of the number of copies obtained.

[0251] The expression level of each gene was determined based on the number of gene copies detected by qPCR. This number of copies is expressed in relation to the 18S reference gene.

[0252] For the individuals analyzed in the present example, the expression level of the 4 genes analyzed by qPCR and normalized by the 18S gene, are shown in Figure 10 (control subject) and Figure 11 (patient with Pitt-Hopkins Syndrome), respectively.

[0253] 3.5 Performing an analysis of the expression levels of these genes using machine learning.

[0254] 3.5.1 Preparation of data for the patient with an unknown diagnosis

[0255] Expression data were collected for four selected genes: GLRA1, ALDH2, RXRG1, and S1 PR5. The data were normalized against the 18S reference gene and organized into an Excel file according to the required nomenclature. Values ​​below the limit of detection were replaced with 0. This file was saved for later analysis.

[0256] 3.5.2 Predictive model load

[0257] The predictive model, for these purposes named “SPH_Sangre_XGB4.mat”, was loaded into the MATLAB working environment and opened using the PLS-toolbox software. This file contains the XGBoost-based model previously trained and validated with known samples in Example 1.

[0258] 3.5.3 Importing patient data

[0259] The data was imported into the PLS toolbox from the prepared Excel file. During the import, the cells were configured to distinguish between labels and normalized expression data.

[0260] 3.5.4 Prediction Configuration and Analysis

[0261] The predictive model was used to classify the patient based on the provided data. The analysis yielded a result indicating whether the patient belongs to the control group (Class 1) or presents PTHS (Class 2).

[0262] • In the control subject of the present example, the model predicted Class 1, confirming that the patient did not have PTHS (see Figure 12).

[0263] • In the second case, using data from a patient with confirmed PTHS, the model predicted Class 2, confirming the presence of the syndrome (see Figure 13).References

[0264] Espinoza, F., Carrazana, R., Retamal-Fredes, E., Ávila, D., Papes, F., Muotri, AR, Ávila, A., 2024. Tcf4 dysfunction alters dorsal and ventral cortical neurogenesis in Pitt-Hopkins syndrome mouse model showing sexual dimorphism. Biochim Biophys Acta Mol Basis Dis 1870, 167178. https: / / doi.org / 10.1016 / j-bbadis.2024.167178

Claims

CLAIMS 1- A method for the diagnosis of Pitt Hopkins syndrome (PTHS) by means of a single assay, CHARACTERIZED in that it comprises the steps of: a) provide an isolated biological sample; b) perform extraction and processing of nucleic acids from the isolated biological sample; c) perform a real-time PCR (qPCR) analysis on the extracted and processed sample; d) determine in said biological sample the level of expression through qPCR of at least 4 of 13 genes selected based on the calculation of copy number; e) perform analyses of the expression levels of these genes using machine learning,- and f) determine the presence of PTHS according to the result obtained from at least four genes; where the 4-13 genes are selected from the group consisting of GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4 and GRIN2A. 2- The method for the diagnosis of PTHS according to claim 1, CHARACTERIZED in that in step a) the isolated biological sample is blood. 3- The method for the diagnosis of PTHS according to claim 1, CHARACTERIZED in that in step d) the level of expression is determined based on the calculation of the number of copies of at least 4 of 13 selected genes, where the 13 genes are selected from the group consisting of GLRA1, GLRA2, GLRA3, TCF4, ALDH1, ALDH2, RXRG1, APOD, OXTR, S1 PR5, LPAR1, CHRM4 and GRIN2A. 4- The method for the diagnosis of PTHS according to claim 3, CHARACTERIZED in that in step a) the isolated biological sample is from buccal mucosa. 5- The method for the diagnosis of PTHS according to claim 3, CHARACTERIZED in that in step a) the isolated biological sample is from blood and / or buccal mucosa. 6- The method for the diagnosis of PTHS according to claim 1, CHARACTERIZED in that in step d) the analysis of the expression levels of the genes is performed by PLS-DA (partial least squares discriminant analysis). 7- The method for the diagnosis of PTHS according to claim 1, CHARACTERIZED in that in step d) the analysis of the expression levels of the genes is performed by XGBoost (extreme gradient boosting). 8- The method for the diagnosis of PTHS according to claim 1, CHARACTERIZED in that in step e) analysis of expression levels was performed by machine learning comprising automatic learning through pattern identification in massive data, using iterations and selection of results, to obtain the expression levels of at least 4 genes from a total of 13 selected genes.

Images