Methods and systems for determining risk for autism spectrum disorders

Abstract

In certain embodiments, the present invention is derived from the discovery that analysis of population distribution curves of metabolite levels in blood can be used to help predict the risk of autism spectrum disorders (ASD) in individuals and / or distinguish ASD from non-ASD developmental delays (DD). In certain aspects, information from assessments of the presence, absence, and / or direction (rising or falling) of tail effects in metabolite distribution curves is used to predict the risk of ASD and / or distinguish ASD from DD.

Description

[0001] This application is a divisional application of invention patent application No. 201580028014.7, whose application date is April 10, 2015 and whose invention name is “Method and system for determining the risk of autism spectrum disorders”.

[0002] CROSS-REFERENCE TO RELATED APPLICATIONS

[0003] This application claims priority to U.S. Patent Application No. 14 / 633,558, filed on February 27, 2015; the U.S. Patent Application No. 14 / 493,141, filed on September 22, 2014, claims priority to U.S. Provisional Patent Application No. 61 / 978,773, filed on April 11, 2014, and U.S. Provisional Patent Application No. 62 / 002,169, filed on May 22, 2014; the contents of each of which are incorporated herein by reference in their entirety. Field of the Invention

[0004] The present invention relates generally to prediction of risk for autism spectrum disorders (ASD) and other disorders. background

[0005] Autism spectrum disorders (ASD) are pervasive developmental disorders characterized by deficits in interactive social interactions, language difficulties, and repetitive behaviors and restricted interests, typically manifesting during the first 3 years of life. The etiology of ASD is poorly understood but is thought to be multifactorial, with both genetic and environmental factors contributing to the development of the disorder.

[0006] Data show that although the average age at which parents begin to suspect their child has ASD is 20 months, the median age of diagnosis is 54 months. From a clinical perspective, an important challenge is to determine as early as possible whether a child has ASD and whether a specialist referral to an autism treatment program is necessary.

[0007] Overview

[0008] The diagnosis of ASD is usually made by developmental pediatricians and other specialists after carefully evaluating the child using the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders. A reliable diagnosis usually requires careful evaluation of the subject by multiple specialists including a developmental pediatrician, neurologist, psychiatrist, psychologist, speech and hearing specialist, and occupational therapist. In addition, the median age of ASD diagnosis is 54 months, despite the fact that the average age of parents suspecting ASD is as early as 20 months. The CDC (Centers for Disease Control) has observed that only 18% of children who are ultimately diagnosed with ASD are determined before 36 months of age.

[0009] Unfortunately, young children with undiagnosed ASD miss the opportunity for early therapeutic intervention during an important window of child development. Medical diagnostic tests that reliably determine risk for ASD are needed to specifically identify younger children earlier, when therapeutic intervention may be more effective.

[0010] Embodiments of the present invention arise from the following discovery: analysis of distribution curves of analyte (such as metabolites) measurements within and between populations provides information that can be used to establish or improve a classifier for predicting the risk of a disorder or condition such as ASD. In particular, analysis of population distribution curves of metabolite levels in blood helps predict the risk of autism spectrum disorders (ASD) in a subject. For example, analysis of population distribution curves of metabolite levels in blood can be used to distinguish autism spectrum disorders (ASD) from non-ASD developmental disorders (such as developmental delay (DD) not caused by autism spectrum disorders) in a subject.

[0011] Statistical analyses of biomarkers that distinguish between two groups typically assume that the mean biomarker levels of the two populations are different and that variations around this mean are attributed to experimental and / or population variations that are best characterized by a Gaussian distribution. In contrast to this baseline model, we observe that for some analytes, the distribution in ASD or sometimes in DD is best characterized as itself consisting of multiple sub-distributions, one of which is indistinguishable from the other healthy state (e.g., where the ASD and DD distributions are undifferentiated) and another of which deviates far from the mean in a small number of subjects, e.g., presenting the "tail" of the combined distribution for that population. This insight leads to an analytical framework that is significantly different from the baseline; it has been found that for some analytes, better results can be achieved by, for example, establishing an ordering that does not require an intrinsic Gaussian distribution model, defining thresholds based on the top or bottom of the population distribution.

[0012] Thus, metabolites are described herein as exhibiting a "tail enrichment" or "tail" effect, wherein there is an enrichment of samples of a particular population (e.g., ASD or DD) in the distal portion of the distribution curve of the metabolite levels for that metabolite. The information obtained by assessing the presence, absence, and / or direction (up or down) of the tail effect in the metabolite distribution curve can be used to predict the risk of ASD. It has been found that for a particular metabolite, the metabolite level corresponding to the top or bottom (e.g., decile) of the distribution curve, i.e., within the "tail" of the distribution curve (whether in the "right tail" or the "left tail"), is highly suggestive of the presence or absence of ASD.

[0013] In addition, it was found that risk prediction was improved when multiple metabolites with low degree of overlap and common information were combined. For example, for the assessment of ASD, there are some specific metabolite groups that provide complementary diagnostic / risk assessment information. That is, ASD-positive individuals that cannot be identified by analyzing the level of a first metabolite (e.g., individuals within the identification tail of the first metabolite) are not the same as ASD-positive individuals that can be identified by analyzing a second metabolite (or there may be a low, non-zero degree of overlap). Without wishing to be bound by a particular theory, this finding may reflect the multifaceted nature of ASD itself.

[0014] Thus, in certain embodiments, the risk assessment method comprises identifying whether a subject falls into any of a plurality of identified metabolite tails involving a plurality of metabolites, e.g., wherein the predictors of the different metabolite tails are at least partially unconnected, e.g., they have low common information, and thus risk prediction improves when multiple metabolites with low common information are introduced. The classifier has a predetermined level of predictability, e.g., in the form of AUC - i.e., the area under the ROC curve where the false positive rate (1-specificity) is plotted against the true positive rate (sensitivity) for the classifier - wherein the AUC increases when metabolites exhibiting tail effects with low common information are added to the classifier.

[0015] In some embodiments, the present invention is derived from the following discovery: certain threshold values ​​of metabolite levels in the blood can be used to help predict the risk of autism spectrum disorders (ASD) in a subject. In some aspects, these threshold values ​​of metabolites derived from the evaluation of the presence, absence and / or direction (up or down) of the tail effect in the metabolite distribution curve are used to predict the risk of ASD. In some aspects, these threshold values ​​may be at the upper or lower end of the distribution of metabolite levels in a population. It has been found that for specific metabolites, metabolite levels above an upper threshold and / or below a lower threshold are highly suggestive of the presence or absence of ASD.

[0016] In some embodiments, the levels of these metabolites can be used to distinguish ASD from other forms of developmental delay (eg, developmental delay (DD) not due to an autism spectrum disorder).

[0017] In one aspect, the present invention relates to a method for distinguishing an autism spectrum disorder (ASD) from a non-ASD developmental delay (DD) in a subject, the method comprising: (i) measuring the level of a first metabolite among a plurality of metabolites in a sample obtained from the subject, the population distribution of the first metabolite having been previously characterized in a first population of subjects with ASD and in a second population of subjects with non-ASD developmental delay (DD), wherein the first metabolite is pre-determined to exhibit an ASD tail effect and / or a DD tail effect, each tail effect comprising an associated right tail or right tail that is enriched in members of the respective (ASD or DD) population; left tail, wherein the first metabolite exhibits an ASD tail effect with a right tail, when the level of the first metabolite in the sample is above a predetermined upper (minimum) threshold of the right tail defined as enrichment in the first (ASD) population members, the level of the first metabolite in the sample is within the ASD tail, and when the first metabolite exhibits an ASD tail effect with a left tail, when the level of the first metabolite in the sample is below a predetermined lower (maximum) threshold of the left tail defined as enrichment in the first (ASD) population members, the level of the first metabolite in the sample is within the ASD tail, and when the first metabolite exhibits a DD tail effect with a right tail, The level of the first metabolite in the sample is within the DD tail when the level of the first metabolite in the sample is above a predetermined upper (minimum) threshold value that defines the right tail of enrichment in members of the second (DD) population, and the level of the first metabolite in the sample is within the DD tail when the level of the first metabolite in the sample is below a predetermined lower (maximum) threshold value that defines the left tail of enrichment in members of the second (DD) population; (ii) measuring the level of at least one additional metabolite from the sample among the plurality of metabolites, wherein the at least one additional metabolite has been previously characterized in the first population and the second population. The method comprises the steps of: providing a population distribution of each of the at least one additional metabolites, wherein the population distribution of each of the at least one additional metabolites is predetermined to exhibit at least one of an ASD tail effect and a DD tail effect, and, for each of the at least one additional metabolite, identifying whether the level of the metabolite in the sample is within the corresponding ASD tail and / or DD tail according to step (i); and (iii) determining with a predetermined level of predictability that (a) the subject has ASD instead of DD or (b) the subject has DD instead of ASD based on the sample being in the ASD tail identified and / or the DD tail identified by analyzing the metabolites in steps (i) and (ii).

[0018] In certain embodiments, the first metabolite is pre-determined to exhibit an ASD tail effect with an associated upper (minimum) or lower (maximum) threshold, the threshold being pre-determined such that the odds that a sample of unknown classification (previously uncharacterized sample) meeting the criteria is ASD rather than DD are no less than 1.6:1, p<0.3. In certain embodiments, the odds are no less than 2:1, or no less than 2.5:1, or no less than 2.75:1, or no less than 3:1, or no less than 3.5:1, or no less than 3.75:1, or no less than 4:1. In any of the foregoing, the p-value (statistical significance value) satisfies p<0.3, or p<0.25, or p<0.2, or p<0.15, or p<0.1, or p<0.05.

[0019] In certain embodiments, the first metabolite is predetermined to show a DD tail effect with an associated upper (minimum) or lower (maximum) threshold, the threshold being predetermined such that the odds that an unknown classified sample (previously uncharacterized sample) meeting the criteria is DD rather than ASD is no less than 1.6:1, p<0.3. In certain embodiments, the odds are no less than 2:1, or no less than 2.5:1, or no less than 2.75:1, or no less than 3:1, or no less than 3.25:1, or no less than 3.5:1, or no less than 3.75:1, or no less than 4:1. In any of the foregoing, the p-value (statistical significance value) satisfies p<0.3, or p<0.25, or p<0.2, or p<0.15, or p<0.1, or p<0.05.

[0020] In certain embodiments, the predetermined level of predictability corresponds to a receiver operating characteristic (ROC) curve plotting the false positive rate (1-specificity) against the true positive rate (sensitivity) having an AUC (area under the curve) of at least 0.70.

[0021] In certain embodiments, the predetermined upper (minimum) threshold value for one or more of the metabolites is a percentile from the 85th to the 95th percentile (e.g., about the 90th percentile, or about the 85th, 86th, 87th, 88th, 89th, 91st, 92nd, 93rd, 94th, or 95th percentile, rounded to the nearest percentile), and wherein the predetermined lower (maximum) threshold value for one or more of the metabolites is a percentile from the 10th to the 20th percentile (e.g., about the 15th percentile, or about the 10th, 11th, 12th, 13th, 14th, 16th, 17th, 18th, 19th, or 20th percentile, rounded to the nearest percentile).

[0022] In certain embodiments, the plurality of metabolites includes at least two metabolites selected from the group consisting of 5-hydroxyindoleacetate (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionate, 3-carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF), 3-indoxyl sulfate, 4-ethylphenyl sulfate, 8-hydroxyoctanoate, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetate, isovalerylglycine, lactate, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate sulfate), pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolate, xanthine, hydroxy-chlorothalonil, octenoylcarnitine, and 3-hydroxyhippurate.

[0023] In certain embodiments, the plurality of metabolites includes at least two metabolites selected from the group consisting of phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfate, pantothenic acid (vitamin B5), and hydroxychlorothalonil.

[0024] In certain embodiments, the plurality of metabolites includes at least 3 metabolites selected from the group consisting of phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfate, pantothenic acid (vitamin B5), and hydroxychlorothalonil.

[0025] In certain embodiments, the plurality of metabolites comprises at least one pair of metabolites selected from the pairs listed in Table 6.

[0026] In certain embodiments, the plurality of metabolites comprises at least one triplet of metabolites selected from the triplet listed in Table 7.

[0027] In certain embodiments, the plurality of metabolites comprises at least one pair of metabolites that, when combined together as a set of two metabolites, provide an AUC of at least 0.62 (e.g., at least about 0.63, 0.64, or 0.65), wherein AUC is the area under the ROC curve plotting the false positive rate (1-specificity) against the true positive rate (sensitivity) for a classifier based on the set of two metabolites alone.

[0028] In certain embodiments, the plurality of metabolites comprises at least one triplet of metabolites that, when combined together as a set of three metabolites, provide an AUC of at least 0.66 (e.g., at least about 0.67 or 0.68), wherein AUC is the area under the ROC curve plotting the false positive rate (1-specificity) against the true positive rate (sensitivity) for a classifier based only on the set of three metabolites.

[0029] In another aspect, the present invention relates to a method for determining a subject's risk for an autism spectrum disorder (ASD), the method comprising: (i) analyzing the level of a first metabolite of a plurality of metabolites from a sample obtained from the subject, the population distribution of the first metabolite having been previously characterized in a reference subject population having a known classification, wherein the first metabolite is predetermined to exhibit an ASD tail effect comprising an associated right tail or left tail enriched in ASD members, and wherein the first metabolite exhibits an ASD tail effect with a right tail, when the level of the first metabolite in the sample is above a predetermined upper (minimum) threshold value defining the right tail enriched in ASD population members, the level of the first metabolite in the sample is within the ASD tail, and wherein the first metabolite exhibits an ASD tail effect with a left tail, When the level of the first metabolite in the sample is below a pre-determined lower (maximum) threshold value that defines the left tail enriched in members of the ASD population, the level of the first metabolite in the sample is within the ASD tail; (ii) measuring the level of at least one additional metabolite from the plurality of metabolites in the sample, the population distribution of each of the at least one additional metabolite being previously characterized in a reference population and predetermined to exhibit an ASD tail effect, and for each of the at least one additional metabolite, identifying whether the level of the metabolite in the sample is within the corresponding ASD tail according to step (i); and (iii) determining the risk of the subject having ASD at a predetermined predictive level based on the sample being located in the ASD tail identified for the metabolites analyzed in steps (i) and (ii).

[0030] In certain embodiments, the first metabolite is predetermined to exhibit an ASD tail effect with an associated upper (minimum) or lower (maximum) threshold, the threshold being predetermined such that the odds that an unknown classified sample (previously uncharacterized sample) meeting the criteria is ASD rather than DD are no less than 1.6:1, p<0.3. In certain embodiments, the odds are no less than 2:1, or no less than 2.5:1, or no less than 2.75:1, or no less than 3:1, or no less than 3.25:1, or no less than 3.5:1, or no less than 3.75:1, or no less than 4:1. In any of the foregoing, the p-value (statistical significance value) satisfies p<0.3, or p<0.25, or p<0.2, or p<0.15, or p<0.1, or p<0.05.

[0031] In certain embodiments, the predetermined level of predictability corresponds to a receiver operating characteristic (ROC) curve plotting the false positive rate (1-specificity) against the true positive rate (sensitivity) having an AUC (area under the curve) of at least 0.70.

[0032] In certain embodiments, the plurality of metabolites includes at least two metabolites selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine, and 3-hydroxyhippuric acid.

[0033] In another aspect, the present invention relates to a method for determining a subject's risk for autism spectrum disorder (ASD), comprising: (i) analyzing the levels of multiple metabolites in a sample obtained from the subject, the multiple metabolites comprising at least two metabolites selected from the following metabolites: 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine and 3-hydroxyhippuric acid; and (ii) determining that the subject has a risk of ASD based on the quantitative levels of the multiple metabolites.

[0034] In certain embodiments, the subject is no older than about 54 months. In certain embodiments, the subject is no older than about 36 months.

[0035] In certain embodiments, the plurality of metabolites includes at least two metabolites selected from the group consisting of phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfate, pantothenic acid (vitamin B5), and hydroxychlorothalonil.

[0036] In certain embodiments, the plurality of metabolites includes at least 3 metabolites selected from the group consisting of phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfate, pantothenic acid (vitamin B5), and hydroxythiophanate-methyl.

[0037] In certain embodiments, the plurality of metabolites comprises at least one pair of metabolites selected from the pairs listed in Table 6.

[0038] In certain embodiments, the plurality of metabolites comprises at least one triplet of metabolites selected from the triplet listed in Table 7.

[0039] In certain embodiments, the plurality of metabolites comprises at least one pair of metabolites that, when combined together as a set of two metabolites, provide an AUC of at least 0.62 (e.g., at least about 0.63, 0.64, or 0.65), wherein AUC is the area under the ROC curve of a classifier based only on the set of two metabolites plotting the false positive rate (1-specificity) relative to the true positive rate (sensitivity).

[0040] In certain embodiments, the plurality of metabolites comprises at least one triplet of metabolites that, when combined together as a set of three metabolites, provide an AUC of at least 0.66 (e.g., at least about 0.67 or 0.68), wherein AUC is the area under the ROC curve of a classifier based only on the set of three metabolites plotting the false positive rate (1-specificity) relative to the true positive rate (sensitivity).

[0041] In certain embodiments, the sample is a plasma sample.

[0042] In certain embodiments, measuring the level of a metabolite comprises performing mass spectrometry. In certain embodiments, performing mass spectrometry comprises performing a mass spectrometry selected from pyrolysis mass spectrometry, Fourier transform infrared spectroscopy, Raman spectroscopy, gas chromatography-mass spectrometry, high pressure liquid chromatography / mass spectrometry (HPLC / MS), liquid chromatography (LC)-electrospray mass spectrometry, cap-LC-tandem electrospray mass spectrometry, and ultra-high performance liquid chromatography / electrospray ionization tandem mass spectrometry.

[0043] In another aspect, the present invention relates to a method for distinguishing a sample of autism spectrum disorder (ASD) from non-ASD developmental delay (DD) in a subject, comprising: (i) analyzing the levels of a plurality of metabolites in a sample obtained from the subject, wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfate, 4-ethylphenylsulfate, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactate, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiophanate-methyl, octenoylcarnitine and 3-hydroxyhippuric acid, wherein the levels and / or population distribution of the plurality of metabolites have been previously characterized on a reference population; and (ii) determining with a predetermined level of predictability that (a) the subject has ASD rather than DD or (b) the subject has DD rather than ASD by comparing the levels of the plurality of metabolites in a sample from the subject to a predetermined threshold value (e.g., a threshold value determined from a reference population of samples with known classification).

[0044] In certain embodiments, the invention provides methods for analyzing metabolites by assigning weights to different metabolites to reflect their respective functions in risk prediction. In some embodiments, the weight assignments can be derived from the biological functions of the metabolites (e.g., the pathways to which they belong), their clinical utility, or their significance from statistical or epidemiological analyses.

[0045] In certain embodiments, the present invention provides methods for measuring metabolites using various techniques, including but not limited to chromatography, mass spectrometry, fluorimetry, electrophoresis, immunoaffinity, and immunochemical assays.

[0046] In certain embodiments, the present invention provides methods for determining a subject's risk for an autism spectrum disorder (ASD), comprising analyzing the levels of multiple metabolites in a sample from the subject; and determining, based on the quantified levels of the multiple metabolites, whether the subject has ASD rather than a non-ASD developmental disorder at a predetermined predictable level.

[0047] In certain embodiments, the plurality of metabolites includes at least one metabolite selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine, 3-hydroxyhippuric acid, and combinations thereof.

[0048] In certain embodiments, the plurality of metabolites include at least two metabolites selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine, 3-hydroxyhippuric acid, and combinations thereof.

[0049] In certain embodiments, the plurality of metabolites include at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9 or at least 10 metabolites selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine, 3-hydroxyhippuric acid, and combinations thereof.

[0050] In certain embodiments, the plurality of metabolites includes additional metabolites. In some embodiments, the plurality of metabolites includes more than 21 metabolites.

[0051] In certain embodiments, the present invention provides methods for distinguishing an autism spectrum disorder (ASD) from a non-ASD developmental disorder in a subject, comprising the steps of analyzing the levels of a plurality of metabolites in a sample from the subject, comparing the levels of the metabolites with their corresponding population distributions in a reference population, and determining with a predetermined level of predictability whether the subject has ASD rather than a non-ASD developmental disorder by comparing the levels of the plurality of metabolites in the sample from the subject with the levels and / or population distributions of the plurality of metabolites in a previously characterized reference population.

[0052] For example, in certain embodiments, the invention provides diagnostic criteria comprising at least one metabolite that can predict a subject's risk of ASD using a ROC curve having an AUC of at least 0.60, at least 0.65, at least 0.70, at least 0.75, at least 0.80, at least 0.85, or at least 0.90. The AUC is the area under the ROC curve that plots the false positive rate (1-specificity) against the true positive rate (sensitivity) for a classifier.

[0053] In certain embodiments, at least one metabolite for analysis is selected from 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine, 3-hydroxyhippuric acid, and combinations thereof.

[0054] In certain embodiments, the at least one metabolite for analysis comprises at least two or more (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21) members selected from the following members: 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furan Propionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiophanate-methyl, octenoylcarnitine and 3-hydroxyhippuric acid, wherein a non-ASD population distribution curve and an ASD population distribution curve are established for each metabolite (e.g., each of the metabolites showing a tail effect).

[0055] In certain embodiments, the metabolite for analysis is selected from the group consisting of γ-CEHC, xanthine, p-cresol sulfate, octenoylcarnitine, phenylacetylglutamine, and combinations thereof.

[0056] In certain embodiments, one metabolite analyzed is γ-CEHC.

[0057] In certain embodiments, one metabolite analyzed is xanthine.

[0058] In certain embodiments, one metabolite analyzed is p-cresol sulfate.

[0059] In certain embodiments, one metabolite analyzed is octenoylcarnitine.

[0060] In certain embodiments, one metabolite analyzed is phenylacetylglutamine.

[0061] In certain embodiments, one metabolite analyzed is isovalerylglycine.

[0062] In certain embodiments, one metabolite analyzed is pipecolic acid.

[0063] In certain embodiments, one metabolite analyzed is indoleacetic acid.

[0064] In certain embodiments, one metabolite analyzed is octenoylcarnitine.

[0065] In certain embodiments, one metabolite analyzed is hydroxythiophanate-methyl.

[0066] In certain embodiments, the plurality of metabolites comprises at least a first metabolite and a second metabolite that are complementary (e.g., the ASD tail samples of the first and second metabolites are substantially non-overlapping), such that the predictive sub-portions provided by the metabolites are not connected and have little common information. In certain embodiments, risk prediction is improved by introducing a plurality of metabolites with little common information.

[0067] In certain embodiments, the plurality of metabolites includes two metabolites, wherein the two metabolites combined together as a set of two metabolites provide an AUC of at least 0.62, 0.63, 0.64, or 0.65.

[0068] In certain embodiments, the plurality of metabolites includes three metabolites, wherein the three metabolites combined together as a group of three metabolites provide an AUC of at least 0.66, 0.67, or 0.68.

[0069] In certain embodiments, the present invention provides a method for distinguishing autism spectrum disorders (ASD) from non-ASD developmental disorders in a subject by analyzing the levels of two sets of predefined metabolites. In certain embodiments, a first set of metabolites represents metabolites closely associated with ASD, while a second set of metabolites represents metabolites associated with a control condition (e.g., DD). By analyzing two sets of metabolites in a sample from a subject, the risk that the subject has ASD rather than a control condition can be determined by a variety of methods described in the present disclosure. For example, this can be achieved by comparing the aggregated ASD tail effect of the first set of metabolites with the aggregated non-ASD tail effect of the second set of metabolites.

[0070] In certain embodiments, the present invention provides a method for distinguishing an autism spectrum disorder (ASD) from a non-ASD developmental delay (DD) in a subject, the method comprising: (i) measuring the levels of a plurality of metabolites in a sample obtained from the subject, wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid (5-HIAA), indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfate, 4-ethylphenylsulfate, hydroxyisovalerylcarnitine (C5), isovalerylglycine, lactic acid, N1- methyl-2-pyridone-5-carboxamide, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof; and (ii) calculating the number of metabolites in the sample whose levels are equal to or lower than (a) a predetermined threshold concentration indicating ASD (ASD left tail effect) as defined in Table 9A or (b) a predetermined threshold concentration indicating DD (DD left tail effect) as defined in Table 9B; and / or (iii) calculating the number of metabolites in the sample whose levels are equal to or higher than (a) a predetermined threshold concentration indicating ASD (ASD left tail effect) as defined in Table 9A or (b) a predetermined threshold concentration indicating DD (DD left tail effect) as defined in Table 9B; and (iv) determining that the subject has ASD or DD based on the numbers obtained in steps (ii) and / or (iii).

[0071] In certain embodiments, the present invention provides a method for determining that a subject has a risk of or is at risk of developing an ASD, the method comprising: (i) measuring the levels of a plurality of metabolites in a sample obtained from the subject, wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid (5-HIAA), indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol (1,5-AG), 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoleacetic acid (IAA), 5-hydroxy-1,2-dihydro-2-furanpropionic acid (1,5-D), 1,2-dihydro ... doxyl sulfate, 4-ethylphenyl sulfate, hydroxyisovalerylcarnitine (C5), isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof; and (ii) detecting two or more of the following: (a) a xanthine level of 182.7 ng / ml or greater; (b) a hydroxy-chlorothalonil level of 20.3 ng / ml or greater; (c) a 5-hydroxyindoleacetic acid level of 28.5 ng / ml or greater; (d) a lactic acid level of 686600 ng / ml or greater. above; (e) pantothenic acid level of 63.3 ng / ml or above; (f) pipecolic acid level of 303.6 ng / ml or above; (g) γ-CEHC level of 32.0 ng / ml or below; (h) indoleacetic acid level of 141.4 ng / ml or below; (i) p-cresol sulfate level of 182.7 ng / ml or below; (j) 1,5-anhydroglucitol (1,5-AG) level of 11910.3 ng / ml or below; (k) 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF) level of 7.98 ng / ml l or less; (l) 3-indoxyl sulfate level of 256.7 ng / ml or less; (m) 4-ethylphenyl sulfate level of 3.0 ng / ml or less; (n) hydroxyisovalerylcarnitine (C5) level of 12.9 ng / ml or less; (o) N1-methyl-2-pyridone-5-carboxamide level of 124.82 ng / ml or less; and (p) phenylacetylglutamine level of 166.4 ng / ml or less; and (iii) based on the metabolite levels detected in step (ii), determining that the subject has ASD or is at risk of developing ASD.

[0072] In some embodiments, the metabolite levels detected in step (ii) are approximate levels.

[0073] In some embodiments, the present invention provides a method for diagnosing a subject as having an autism spectrum disorder (ASD) or being at risk of developing the disorder, the method comprising (i) determining the level of one or more metabolites in a sample obtained from the subject; (ii) comparing the determined level of the one or more metabolites with a predetermined level of the one or more metabolites, the predetermined level indicating that the subject has ASD or is at risk of developing the disorder; and (iii) based on the difference between the determined level of the one or more metabolites and the predetermined level, diagnosing the subject as having ASD or being at risk of developing the disorder. wherein the one or more metabolites are selected from 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), p-phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine and 3-hydroxyhippuric acid.

[0074] In some embodiments, the present invention provides a method for diagnosing a subject as having an autism spectrum disorder (ASD) or being at risk of developing the disorder, the method comprising (i) determining the level of one or more metabolites in a sample obtained from the subject; (ii) comparing the determined level of the one or more metabolites with a predetermined level of the one or more metabolites, the predetermined level indicating that the subject has ASD or is at risk of developing the disorder; and (iii) diagnosing the subject as having ASD or being at risk of developing the disorder based on the difference between the determined level and the predetermined level of the one or more metabolites; wherein the one or more metabolites are selected from 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfate, 4-ethylphenyl sulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-chlorothalonil, octenoylcarnitine and 3-hydroxyhippuric acid, provided that the metabolite is not 5-hydroxyindoleacetic acid (5-HIAA), 5-anhydroglucitol (1,5-AG), 3-(3- hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indolesulfonic acid, 4-ethylphenylsulfonic acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine or 3-hydroxyhippuric acid.

[0075] In certain embodiments, the present invention provides a method for determining a subject's risk of ASD by measuring the level of certain metabolites and genetic information from the subject. In some embodiments, the genetic information includes copy number variation (CNV) and / or fragile X (FXS) testing.

[0076] In other embodiments, limitations described with respect to certain aspects of the invention may apply to other aspects of the invention. For example, in some embodiments, limitations of a claim dependent on one independent claim may apply to another independent claim. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 The distribution of an exemplary metabolite in two populations (eg, ASD and DD) is illustrated, along with the average shift of the metabolite between the two populations.

[0078] Figure 2 The distribution of an exemplary metabolite in two populations (eg, ASD and DD) is illustrated, as well as the tail effect of this metabolite between the two populations (eg, ASD distribution has a denser tail).

[0079] Figure 3 Illustrated is the distribution of the metabolite 5-hydroxyindoleacetic acid in two populations (eg, ASD and DD), which exhibits a statistically significant mean shift (t-test; p<0.01) and a statistically significant right-tail effect ('extreme value' indicates tail effect, Fisher's test; p=0.001) between the two populations.

[0080] Figure 4 Illustrate the distribution of the metabolite γ-CEHC in two groups (eg, ASD and DD), which exhibited a statistically significant left-tail effect between the two groups ("extreme value" indicates tail effect, Fisher's test; p=0.008)

[0081] Figure 5 The distribution of the metabolite phenylacetylglutamine in two populations (e.g., ASD and DD) is illustrated, which exhibits a statistically significant mean shift (t-test; p=0.001) and statistically significant left and right tail effects ("extreme value" indicates tail effect, Fisher's test; p=0.0001) between the two populations. The distribution appears as a shifted Gaussian curve in both populations.

[0082] Figure 6 The correlation of two exemplary metabolites is illustrated, demonstrating that the two metabolites have different tail effect distributions and are complementary.

[0083] Figure 7 Illustrated are the tail effects of 12 exemplary metabolites in 180 subjects and their predictive abilities for ASD and DD.

[0084] Fig. 8A Graph illustrating ASD and non-ASD tail effects for 180 samples using an exemplary 12-metabolite panel, demonstrating that samples from ASD patients show clustering of the ASD tail effect.

[0085] Figure 8B Illustrated are plots of ASD and non-ASD tail effects for 180 samples using an exemplary 12-metabolite panel, and an exemplary method for binning the data.

[0086] Figure 8CGraph illustrating ASD and non-ASD tail effects for 180 samples using an exemplary 21-metabolite panel, demonstrating that samples from ASD patients show clustering of the ASD tail effect.

[0087] Fig. 9 The exemplary 12-metabolite panel illustrates that predictability for ASD increases as the number of metabolites evaluated increases.

[0088] Fig. 10A The effect of the trichotomy on the predictability of ASD using an exemplary 12-metabolite panel is illustrated.

[0089] Fig. 10B Illustrated is the effect of trichotomy on the predictability of ASD in the analysis of an exemplary 21-metabolite panel.

[0090] Fig.11A The improvement in predictability of ASD using the voting method compared to a non-voting method for analysis of an exemplary 12-metabolite panel is illustrated.

[0091] Fig. 11B The improvement in predictability of ASD using a voting approach compared to a non-voting approach using an exemplary 21-metabolite panel is illustrated.

[0092] Fig.12 The validation process to achieve high predictability for ASD using an exemplary 12-metabolite panel is illustrated.

[0093] Figures 13A-13U The population distribution of 21 exemplary metabolites in ASD and non-ASD populations is illustrated.

[0094] Fig.14A -B illustrates the impact on ASD predictability by including and excluding an exemplary 12 metabolite panel, an exemplary 21 metabolite panel, and a set of 84 candidate metabolites from a total of 600 metabolites, as assessed by tail effect analysis and mean shift analysis. (blacklist=exclude, whitelist=include, mx_12=exemplary 12 metabolite panel, mx_targeted 21=exemplary 21 metabolite panel, mx-all-candidates=84 candidate metabolites, all features=total set of 600 metabolites)

[0095] Fig. 14C-D illustrates the effect on ASD predictability produced by including (whitelist) and excluding (blacklist) an exemplary 12 metabolite panel and an exemplary 21 metabolite panel from a total of 600 metabolites, as assessed by tail effect analysis and mean shift analysis, and by comparing logistic regression with Bayesian analysis in two sample cohorts (i.e., "Christmas" and "Easter"). (blacklist=exclude, whitelist=include, mx_12=exemplary 12 metabolite panel, mx_targeted 21=exemplary 21 metabolite panel, mx-all-candidates=84 candidate metabolites, all features=total set of 600 metabolites)

[0096] Fig.15 The effect on the predictability of ASD by using increasing numbers of metabolites selected from a subset of the exemplary 21 metabolite panel is illustrated.

[0097] Fig.16A The effect of adding genetic information to the tail effects analysis using an exemplary 12-metabolite panel was illustrated, demonstrating increased power to separate ASD from non-ASD.

[0098] Fig. 16B The effect of adding genetic information to the tail effects analysis using an exemplary 21 metabolite panel was illustrated, demonstrating increased power to distinguish ASD from non-ASD.

[0099] Fig.17A -B illustrates the effect on ASD predictability by including and excluding an exemplary 21-metabolite panel from the total number of metabolites by comparing tail effect analysis with mean shift analysis and by comparing logistic regression with Bayesian analysis. (blacklist=exclude, whitelist=include, mx_12=exemplary 12-metabolite panel, mx_targeted 21=exemplary 21-metabolite panel, mx_all_candidates=84 candidate metabolites, all features=total set of 600 metabolites)

[0100] Fig.18A-B The effect on ASD predictability by including and excluding an exemplary 21-metabolite panel from the total number of metabolites is illustrated by comparing tail effect analysis with mean shift analysis and by using logistic regression in two cohorts (i.e., "Christmas" and "Easter"). (blacklist=exclude, whitelist=include, mx_12=exemplary 12 metabolite panel, mx_targeted 21=exemplary 21 metabolite panel, mx_all_candidates=84 candidate metabolites, all features=total set of 600 metabolites)

[0101] Fig.19A -D Illustrate the effect on ASD predictability by including and excluding an exemplary panel of 21 metabolites from the total number of metabolites using the "Christmas" cohort, or the "Easter" cohort, or a combination of both, comparing tail effect analysis to mean shift analysis, and comparing logistic regression to Bayesian analysis. (blacklist=exclude, whitelist=include, mx_12=exemplary panel of 12 metabolites, mx_targeted 21=exemplary panel of 21 metabolites, mxallcandidates=84 candidate metabolites, allfeatures=total set of 600 metabolites)

[0102] Fig. 20 Representative graphs illustrating the specificity and sensitivity of tail effects analysis of an exemplary 21-metabolite panel for predicting ASD.

[0103] Fig.21 An example scoring system is shown in which a risk score for ASD or DD is calculated based on the sum of the log 2 values ​​of the odds ratios of predictive metabolites for ASD and DD.

[0104] definition

[0105] In order to more easily understand the present invention, certain terms are first defined below. Additional definitions of the following terms and other terms are set forth throughout the specification.

[0106] In this application, unless otherwise indicated by the context, (i) the term "a / a" may be understood to mean "at least one"; (ii) the term "or" may be understood to mean "and / or"; (iii) the terms "comprising" and "including" may be understood to include the specified components or steps, whether presented by themselves or with one or more additional components or steps; and (iv) the terms "about" and "approximately" may be understood to allow for standard variations as understood by a person of ordinary skill in the art; and (v) when a range is provided, the endpoints are included.

[0107] Agent: As used herein, the term "agent" may refer to a compound or entity of any chemical class, including, for example, polypeptides, nucleic acids, carbohydrates, lipids, small molecules, metals, or combinations thereof.

[0108] Approximately: As used herein, the terms "approximately", "approximately" or "about" are intended to encompass normal statistical variations as understood by those of ordinary skill in the art based on the relevant context. In certain embodiments, unless otherwise indicated or apparent from the context (except where such a number exceeds 100% of the possible value), the terms "approximately", "approximately" or "about" refer to a range of values ​​falling within 25%, 20%, 19%, 18%, 17%, 16%, 15%, 14%, 13%, 12%, 11%, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1% or less of the reference value in either direction (greater or less).

[0109] Area Under the Curve (AUC): A classifier has an associated ROC curve (Receiver Operating Characteristic Curve) that plots the false positive rate (1-specificity) against the true positive rate (sensitivity). The area under the ROC curve (AUC) is a measure of how well a classifier can distinguish between two diagnostic groups. A perfect classifier would have an AUC of 1.0, compared to a random classifier with an AUC of 0.5.

[0110] Associated with: Two events or entities are "associated with" one another, as that term is used herein, if the presence, level, and / or form of one event or entity is correlated with the presence, level, and / or form of another event or entity. For example, a particular entity is considered to be associated with a particular disease, disorder, or condition if the presence, level, and / or form of the entity is correlated with the incidence and / or susceptibility of the disease, disorder, or condition (e.g., across a relevant population).

[0111] Autism spectrum disorder: As used herein, the term "autism spectrum disorder" is considered by those skilled in the art to refer to a developmental disorder related to autism "spectrum", characterized by one or more of interactive social deficits, language difficulties, repetitive behaviors and restricted interests. Autism spectrum disorder has been described in DSM-V (May 2013) as a disorder comprising a series of symptoms, including, for example, communication disorders, such as inappropriate responses in conversation, misreading non-verbal interactions, difficulty in establishing age-appropriate friendships, over-reliance on routine, high sensitivity to changes in their environment, and / or intense focus on inappropriate items. Autism spectrum disorder has additionally been characterized, for example, by DSM-IV-TR as including autism, Asperger's disease, Rett's disease, childhood disintegrative disorder and otherwise not described pervasive developmental disorders (including atypical autism). In some embodiments, standardized testing instruments such as questionnaires and observation schedules are used to characterize autism spectrum disorders (ASD). For example, in some embodiments, ASD is characterized by (i) a score that meets the cutoff for Communication plus Social Interaction Total on the Autism Diagnostic Observation Schedule (ADOS), and a score that meets the cutoff for Social Interaction, Communication, Behavior Patterns, and Developmental Abnormality at ≤36 months on the Autism Diagnostic Interview-Revised (ADI-R); and / or (ii) a score that meets the ASD cutoff for Communication and Social Interaction Total on the ADOS, and a score that meets the cutoff for Social Interaction, Communication, Behavior Patterns, and Developmental Abnormality at ≤36 months on the ADI-R, and (ii)(a) a score that meets the cutoff for Social Interaction and Communication on the ADI-R, or (ii)(b) a score that meets the cutoff for Social Interaction or Communication on the ADI-R and a score that is within 2 points of the cutoff for Social Interaction or Communication (which does not meet the cutoff), or (ii)(c) a score that is within 1 point of the cutoff for Social Interaction and Communication on the ADI-R.

[0112] Classification: As used herein, "classification" is the process of learning to separate data points into different categories by finding common features between collected data points within known categories, and then assigning the data points to one of these different categories using mathematical or other methods. In statistics, classification is the problem of identifying the subpopulation to which a new observation belongs based on a training data set containing observations whose subpopulations are known (where the identity of the subpopulation is unknown). Therefore, the requirement is to classify new individual items into groups based on quantitative information on one or more measurements, traits, or characteristics, etc., and based on a training set in which previously determined groupings have been established. Classification has many applications. In some cases, it is used as a data mining process, while in other cases, more detailed statistical modeling is performed.

[0113] Classifier: As used herein, a "classifier" is a method, algorithm, computer program or system for performing data classification. Examples of widely used classifiers include, but are not limited to, neural networks (multilayer perceptrons), logistic regression, support vector machines, k-nearest neighbors, Gaussian mixture models, Gaussian naive Bayes, decision trees, partial least squares determinant analysis (PSL-DA), Fisher linear discriminant, logistic regression, naive Bayes classifier, perceptron, support vector machine, quadratic classifier, Kernet estimation, Boosting, neural networks, Bayesian networks, hidden Markov models, and learning vector quantization.

[0114] Determine: Many of the methods described herein include a step of "determining". Upon reading this specification, one of ordinary skill in the art will understand that such "determining" can be accomplished using or by using any of a variety of techniques available to those skilled in the art, including, for example, specific techniques explicitly mentioned. In some embodiments, determining involves manipulation of a physical sample. In some embodiments, determining involves consideration and / or manipulation of data or information, such as using a computer or other processing unit suitable for performing relevant analysis. In some embodiments, determining involves receiving relevant information and / or material from a source. In some embodiments, determining includes comparing one or more characteristics of a sample or entity to a comparable reference.

[0115] Determine risk: As used herein, determining risk includes calculating or quantifying the probability that a given subject has or does not have a particular condition or disorder. In some embodiments, a positive or negative diagnosis of a disorder or disease (e.g., autism spectrum disorder (ASD) or developmental delay (DD)) can be made based in whole or in part on a determined risk or risk score (e.g., odds ratio or range).

[0116] Developmental delay: As used herein, the phrase developmental delay (DD) refers to one or more processes of child development, including, for example, a persistent or greater or less delay in physical development, cognitive development, communication development, social or emotional development, or adaptive development, which is not caused by autism spectrum disorders. Even though individuals with ASD may be considered developmentally delayed, the classification of ASD used herein will be considered superior to the classification of DD, so the classification of ASD and DD is mutually exclusive. In other words, unless otherwise specified, the classification of DD is considered to refer to developmental delays of non-ASD. In some embodiments, DD is characterized by non-autism (AU) and non-ASD, but (i) a score of 69 or less on the Mullen scale, a score of 69 or less on the Vineland scale, and a score of 14 or less on the SCQ, or (ii) a score of 69 or less on Mullen or Vineland, and a score of 77 or less on another assessment within half of the standard deviation of the cutoff value.

[0117] Diagnostic information: As used herein, diagnostic information or information used for diagnosis is any information useful in determining whether a patient has a disease or condition and / or classifying a disease or condition into a phenotypic category or into any category that is important for the prognosis of the disease or condition, or the likely response to a treatment (general treatment or any specific treatment) of the disease or condition. Similarly, diagnosis refers to providing any type of diagnostic information, including but not limited to whether a subject is likely to have a disease or condition (such as an autism spectrum disorder), the state, stage or characteristics of the disease or condition manifested in the subject, information related to the nature or classification of the disorder, information related to prognosis and / or information useful in selecting an appropriate treatment. The choice of treatment may include the selection of a specific therapeutic agent or other treatment modality, such as behavioral therapy, dietary changes, etc., a choice as to whether to forgo or proceed with treatment, a choice related to a dosing regimen (e.g., the frequency or level of one or more doses of a specific therapeutic agent or combination of therapeutic agents), etc.

[0118] Marker: As used herein, a marker refers to an agent whose presence or level is associated with or has a correlation with a particular disease or condition. Alternatively or in addition, in some embodiments, the presence or level of a particular marker is associated with, for example, the activity (or activity level) of a particular signaling pathway, which may be a characteristic of a particular disorder. The marker may or may not play an etiological role in the disease or disorder. The statistical significance of the presence or absence of a marker may vary depending on the specific marker. In some embodiments, the detection of a marker is highly specific because it reflects a high probability that the disorder belongs to a particular subclass. According to the present invention, useful markers do not need to distinguish between disorders of a particular subclass with 100% accuracy.

[0119] Metabolite: As used herein, the term metabolite refers to a substance produced by a chemical or physical process in the body. The term "metabolite" includes any chemical or biochemical product of a metabolic process, such as any compound produced by the processing, cleavage or consumption of a biological molecule. Examples of such molecules include, but are not limited to: acids and related compounds; mono-, di- and tri-carboxylic acids (saturated, unsaturated and aliphatic and cyclic, aromatic, alkaryl); aldehyde acids, keto acids; lactone forms; gibberellins; abscisic acid; alcohols, polyols, derivatives and related compounds; ethanol, benzyl alcohol, methanol; propylene glycol, glycerol, phytol; inositol, furfuryl alcohol, menthol; aldehydes, ketones, quinones, derivatives and related compounds; acetaldehyde, butyraldehyde, benzaldehyde, acrolein, furfural, glyoxal; acetone, butanone; anthraquinone; carbohydrates; monosaccharides, disaccharides, trisaccharides; alkaloids, amines and other bases; pyridines (including nicotinic acid, nicotinamide); pyrimidines (including cytosine, thymine); purines (including guanine , adenine, xanthine / hypoxanthine, kinetin); pyrroles; quinolines (including isoquinolines); morphinans, tropanes, cinchonans, nucleotides, oligonucleotides, derivatives and related compounds; guanosine, cytosine, adenosine, thymidine, inosine; amino acids, oligopeptides, derivatives and related compounds; esters; phenols and related compounds; heterocyclic compounds and derivatives; pyrroles, tetrapyrroles (corrinoids and porphines / porphyrins, w / w / o metal ions); flavonoids; indoles; lipids (including fatty acids and triglycerides), derivatives and related compounds; carotenoids, phytoene and sterols, isoprenoids, including terpenes; and modified forms of the above molecules. In some embodiments, metabolites are the products of metabolism of endogenous substances. In some embodiments, metabolites are the products of metabolism of exogenous substances. In some embodiments, metabolites are the products of metabolism of endogenous substances and exogenous substances. As used herein, the term "metabolome" refers to the chemical distribution or fingerprint of metabolites in a body fluid, cell, tissue, organ, or organism.

[0120] Metabolite distribution curve: As used herein, a metabolite distribution curve is a probability distribution curve determined by a function that plots metabolite levels against population density (e.g., ASD or DD). In some embodiments, the distribution curve is a standard curve fit to the data. In some embodiments, the distribution curve is a least squares polynomial curve fit. In some embodiments, the distribution curve is asymmetric or non-Gaussian. In some embodiments, the distribution curve is simply a plot (e.g., a "plot") of cases versus metabolite values ​​with associated diagnostic categories, without a curve fit.

[0121] Mutual Information: As used herein, mutual information refers to a measure of the common dependence of two variables (i.e., the degree to which knowing one variable reduces uncertainty about the other variable). High mutual information indicates a large reduction in uncertainty; low mutual information indicates a small reduction; zero mutual information between two random variables means that the variables are independent.

[0122] Non-autism spectrum disorder (non-ASD): As used herein, non-autism spectrum disorder (non-ASD) refers to the classification of children or adults who do not have autism spectrum disorder. In some embodiments, "non-ASD" is generally a developing subject. In some embodiments, a non-ASD population consists of or includes subjects with developmental delay (DD). In some embodiments, "non-ASD" consists of or includes subjects with DD and normal development.

[0123] Patient: As used herein, the term "patient" or "subject" refers to any organism to which a test or composition is or can be administered, e.g., for experimental, diagnostic, preventive, and / or therapeutic purposes. In some embodiments, the patient suffers from or is susceptible to one or more disorders or conditions. In some embodiments, the patient displays one or more symptoms of a disorder or disorder. In some embodiments, the patient is suspected of suffering from one or more disorders or conditions.

[0124] Predictability: As used herein, predictability refers to the degree to which a correct prediction or forecast of a subject's disease state can be made, qualitatively or quantitatively. Complete predictability implies strict certainty, but lack of predictability does not necessarily mean lack of certainty. Limitations in predictability may be caused by factors such as lack of information or excessive complexity.

[0125] Prognostic and predictive information: As used herein, the terms prognostic information and predictive information are used interchangeably and refer to any information that can be used to indicate any aspect of the course of a disease or condition, whether with or without treatment. Such information can include, but is not limited to, the likelihood that a patient's disease will be cured, the likelihood that a patient's disease will respond to a particular treatment (where response can be defined in any of a variety of ways). Prognostic and predictive information is included in the broad category of diagnostic information.

[0126] With reference to: the term "with reference to" is generally used in this article to describe a standard or control reagent, individual, colony, sample, sequence or numerical value for which a target reagent, individual, colony, sample, sequence or numerical value is compared. In some embodiments, the test or determination of a target reagent, individual, colony, sample, sequence or numerical value is substantially tested and / or determined simultaneously with the test or determination of the target reagent, individual, colony, sample, sequence or numerical value. In some embodiments, the reference reagent, individual, colony, sample, sequence or numerical value is a historical reference, which is optionally embodied in a tangible medium. Typically, as will be appreciated by those skilled in the art, the reference reagent, individual, colony, sample, sequence or numerical value is determined or characterized under conditions comparable to those conditions for determining or characterizing the target reagent, individual, colony, sample, sequence or numerical value.

[0127] Regression analysis: As used herein, "regression analysis" includes any technique for modeling and analyzing several variables, wherein the focus is on the relationship between the dependent variable and one or more independent variables. More specifically, regression analysis helps to understand how the typical value of the dependent variable changes when any one independent variable changes and other independent variables remain fixed. Most commonly, regression analysis estimates the conditional expected value of the dependent variable when the independent variable is kept fixed-that is, the average value of the dependent variable when the independent variable is kept fixed. Less commonly, the focus is on the quantile or other location parameters of the conditional distribution of the dependent variable when the independent variable is given. In all cases, the estimated target is a function of the independent variable, called the regression function. In regression analysis, it is also of interest to characterize the dependent variable changes around the regression function, which can be described by a probability distribution. Regression analysis is widely used in prediction and forecasting, wherein the use of regression analysis has substantial overlap with the field of machine learning. Regression analysis is also used to understand which independent variables are related to the dependent variable, and to explore the form of these relationships. In limited cases, regression analysis can be used to infer the causal relationship between the independent variable and the dependent variable. A large number of techniques for performing regression analysis have been developed. Familiar methods such as linear regression and ordinary least squares regression are parametric in that the regression function is defined in terms of a finite number of unknown parameters estimated from the data. Nonparametric regression refers to techniques that allow the regression function to exist in terms of a specified set of functions, which may be infinite-dimensional.

[0128] Risk: As will be understood from the context, the "risk" of a disease, disorder or condition is the degree to which a particular individual will be diagnosed as suffering from or will develop the possibility of a disease, disorder or condition. In some embodiments, the risk is expressed as a percentage. In some embodiments, the risk is 0%, 1%, 2%, 3%, 4%, 5%, 6%, 7%, 8%, 9% or 10% until 100%. In some embodiments, the risk is expressed as a risk relative to the risk associated with a reference sample or a reference sample group. In some embodiments, a reference sample or a reference sample group has a known risk of a disease, disorder or condition. In some embodiments, a reference sample or a reference sample group is from an individual comparable to a particular individual. In some embodiments, the relative risk is 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more. In some embodiments, the relative risk can be expressed as a relative risk (RR) or an odds ratio (Odds Ratio, OR).

[0129] Sample: As used herein, the term "sample" generally refers to a biological sample obtained from or derived from a target source as described herein. In some embodiments, the target source includes an organism, such as an animal or a human. In some embodiments, a biological sample is or comprises a biological tissue or fluid. In some embodiments, a biological sample can be or comprises bone marrow; blood; plasma; serum; blood cells; ascites; tissue or fine needle biopsy samples; body fluids containing cells; free floating nucleic acids; sputum; saliva; urine; cerebrospinal fluid, peritoneal fluid, pleural fluid, feces, lymph fluid, gynecological fluid, skin swab, vaginal swab, oral swab, nasal swab; washing or lavage fluids such as catheter lavage fluid or bronchoalveolar lavage fluid; aspirates; scrapings; bone marrow samples; tissue biopsy samples; surgical samples; feces, other body fluids, secretions and / or excretions; and / or cells therefrom, etc. In some embodiments, a biological sample is or comprises a cell obtained from an individual. In some embodiments, the obtained cell is or comprises a cell from an individual from whom the sample was obtained. In some embodiments, the sample is a "primary sample" obtained directly from a target source by any appropriate means. For example, in some embodiments, a primary biological sample is obtained by a method selected from the following: biopsy (e.g., fine needle aspiration or tissue biopsy), surgery, body fluid collection (e.g., blood, lymph, feces, etc.), etc. In some embodiments, as will be clear from the context, the term "sample" refers to a preparation obtained by processing a primary sample (e.g., by removing one or more components therein and / or by adding one or more reagents thereto). For example, filtering is performed using a semipermeable membrane. Such a "processed sample" may include, for example, nucleic acids or proteins extracted from a sample, or nucleic acids or proteins obtained by subjecting a primary sample to techniques such as amplification or reverse transcription of mRNA, separation and / or purification of certain components.

[0130] Subject: "Subject" refers to a mammal (e.g., a human, including prenatal human forms in some embodiments). In some embodiments, the subject suffers from a disease, disorder, or condition of interest. In some embodiments, the subject is susceptible to the disease, disorder, or condition. In some embodiments, the subject displays one or more symptoms or characteristics of the disease, disorder, or condition. In some embodiments, the subject does not display any symptoms or characteristics of the disease, disorder, or condition. In some embodiments, the subject is a person having one or more characteristics that characterize a susceptibility or risk for a disease, disorder, or condition. The subject may be a patient, which refers to a person who goes to a medical provider for diagnosis or treatment of a disease. In some embodiments, the subject is an individual to whom treatment is administered.

[0131] Substantially: As used herein, the term "substantially" refers to the qualitative condition of exhibiting an overall or nearly overall degree or level of a target feature or property. One of ordinary skill in the biological field will appreciate that biological and chemical phenomena rarely, if ever, reach completeness and / or proceed to completeness or achieve or avoid absolute results. Therefore, the term "substantially" is used herein to cover the potential lack of completeness inherent in many biological and chemical phenomena.

[0132] Suffering from: An individual who is "suffering from" a disease, disorder, or condition has been diagnosed as having it and / or exhibits or has exhibited one or more symptoms or characteristics of the disease, disorder, or condition.

[0133] Susceptible to: An individual who is "susceptible" to a disease, disorder, or condition is at risk for developing the disease, disorder, or condition. In some embodiments, such an individual is known to have one or more susceptibility factors that are statistically correlated with an increased risk of developing the relevant disease, disorder, and / or condition. In some embodiments, an individual who is susceptible to a disease, disorder, or condition does not display any symptoms of the disease, disorder, or condition. In some embodiments, an individual who is susceptible to a disease, disorder, or condition has not been or has not yet been diagnosed as having the disease, disorder, and / or condition. In some embodiments, an individual who is susceptible to a disease, disorder, or condition is an individual who has been exposed to conditions associated with the development of the disease, disorder, or condition. In some embodiments, the risk of developing a disease, disorder, and / or condition is based on the risk of a population (e.g., family members of an individual with an allergy, etc.)

[0134] Tail enrichment and tail effect: As used herein, the term "tail enrichment" or "tail effect" refers to a classification-enhancing property exhibited by a metabolite (or other analyte) that has a relatively high concentration in a sample from a particular population at the distal portion of a distribution curve of metabolite levels. "Upper tail" or "right tail" refers to a distal portion of a distribution curve that is greater than the mean. "Lower tail" or "left tail" refers to a distal portion of a distribution curve that is below the mean. In some embodiments, the tail is determined by a predetermined threshold based on a ranking. For example, if the measured value of a sample for a certain metabolite is higher than the value corresponding to the 85th to 95th (e.g., 90th) percentile of the metabolite in the population, or lower than the value corresponding to the 10th to 20th (e.g., 15th) percentile of the metabolite in the population, the sample is designated as being within the tail.

[0135] Therapeutic agent: As used herein, the phrase "therapeutic agent" refers to any agent that has a therapeutic effect and / or induces a desired biological and / or pharmacological effect when administered to a subject. In some embodiments, an agent is considered a therapeutic agent if its administration to a population of interest is statistically correlated with a desired or beneficial therapeutic outcome in the population, regardless of whether a particular subject to whom the agent is administered experiences the desired or beneficial therapeutic outcome.

[0136] Training set: As used herein, a "training set" is a set of data used in various fields of information science to discover potential predictive relationships. Training sets are used in artificial intelligence, machine learning, genetic programming, intelligent systems, and statistics. In all of these fields, the training set serves the same purpose and is often used in conjunction with a test set.

[0137] Test set: As used herein, a "test set" is a set of data used in various fields of information science to evaluate the strength and utility of predictive relationships. Test sets can be used in artificial intelligence, machine learning, genetic programming, intelligent systems, and statistics. In all of these fields, test sets serve the same purpose.

[0138] Treatment: As used herein, the term "treatment" (also referred to as "medical treatment" or "medical treatment") refers to any administration of a substance or treatment (e.g., behavioral treatment) that partially or completely alleviates, ameliorates, alleviates, inhibits, delays the onset of one or more symptoms, features, and / or causes of a particular disease, disorder, and / or condition, reduces its severity, and / or reduces its frequency, incidence, or severity. Such treatment may be treatment of subjects who do not show signs of the relevant disease, disorder, and / or condition and / or subjects who only show early signs of the disease, disorder, and / or condition. Alternatively or in addition, such treatment may be treatment of subjects who show one or more determined signs of the relevant disease, disorder, and / or condition. In some embodiments, treatment may be treatment of subjects who have been diagnosed with the relevant disease, disorder, and / or condition. In some embodiments, treatment may be treatment of subjects who are known to have one or more susceptibility factors that are statistically associated with an increased risk of developing the relevant disease, disorder, and / or condition. DETAILED DESCRIPTION OF THE INVENTION

[0139] The present invention provides methods and systems for determining the risk of autism spectrum disorder (ASD) in a subject based on specific analysis of metabolite levels in a sample (e.g., a blood sample or a plasma sample). Various aspects of the present invention are described in detail in the following sections. The use of sections and headings is not intended to limit the present invention. Each section may be applied to any aspect of the present invention. In this application, unless otherwise indicated, the use of "or" means "and / or".

[0140] Autism Spectrum Disorders

[0141] The criteria for the clinical diagnosis of autism spectrum disorders (ASD) have been set forth in the Diagnostics and Statistical Manual of Mental Disorders, 5th edition (DSM-V, published in May 2013).

[0142] Additionally ASD has been characterized, for example, by DSM-IV-TR as including autism, Asperger's disorder, Rett's disorder, childhood disintegrative disorder, and pervasive developmental disorder not otherwise specified (including atypical autism).

[0143] In some embodiments, ASD is characterized by (i) a score that meets the cutoff for Communication plus Social Interaction Total in ADOS, and a score that meets the cutoff for Social Interaction, Communication, Behavior Patterns, and Developmental Abnormality at ≤36 months in the Autism Diagnostic Interview-Revised (ADI-R); and / or (ii) a score that meets the ASD cutoff for Communication and Social Interaction Total in ADOS, and a score that meets the cutoff for Social Interaction, Communication, Behavior Patterns, and Developmental Abnormality at ≤36 months in the ADI-R, and (ii)(a) a score that meets the cutoff for Social Interaction and Communication in the ADI-R, or (ii)(b) a score that meets the cutoff for Social Interaction or Communication in the ADI-R and a score that is within 2 points of the cutoff for Social Interaction or Communication (which does not meet the cutoff), or (ii)(c) a score that is within 1 point of the cutoff for Social Interaction and Communication in the ADI-R.

[0144] Developmental delay

[0145] Developmental delay is one or more processes of child development, including, for example, a large or small delay in physical development, cognitive development, communication development, social or emotional development, or adaptive development, which is not caused by ASD. In some embodiments, DD is characterized by non-autism (AU) and non-ASD, which (i) scores 69 or less on the Mullen scale, 69 or less on the Vineland scale, 14 or less on the SCQ, or (ii) scores 69 or less on Mullen or Vineland, and within half a standard deviation of the cutoff value on another assessment (score 77 or less). Even if the individual with ASD can be considered to be developmentally delayed, the classification of ASD used herein will be considered to be prior to the classification of DD, so that the classification of ASD and DD is mutually exclusive.

[0146] Risk Assessment for ASD

[0147] Children who present with symptoms of impaired language, behavior, or social development are often seen by clinicians (most often in primary care settings), but they cannot determine whether the child has ASD or some other condition, disorder, or classification (e.g., DD). Diagnosing children, especially at an age before significant language development, is difficult, and many primary care physicians do not have the skills or resources to perform differential diagnoses on their patients. For example, ASD may not be easily distinguished from other developmental disorders, conditions, or classifications (e.g., DD).

[0148] It is useful to assess the risk of ASD in subjects (including the probability of non-ASD and DD) and to distinguish ASD from DD. Risk assessment of ASD provides opportunities for early intervention and treatment. For example, non-specialists can use ASD risk assessment to initiate referral to specialists. Specialists can use ASD risk assessment to prioritize further evaluation of patients. The assessment of ASD risk can also be used to establish a temporary diagnosis before the final diagnosis, during which time convenient services can be provided to high-risk children and his or her family.

[0149] Described herein are methods for determining the risk of ASD in a subject. In some embodiments, determining the risk of ASD includes determining whether the chance of a subject having ASD is greater than about 50%. In some embodiments, determining the risk of ASD includes determining that the chance of a subject having ASD is greater than about 60%, 65%, 70%, 74%, 80%, 85%, 90%, 95% or 98%. In some embodiments, determining the risk of ASD includes determining that the subject has ASD. In some embodiments, determining the risk of ASD includes determining that the subject does not have ASD (i.e., non-ASD).

[0150] In some embodiments, the present invention provides methods for distinguishing ASD from non-ASD classifications (e.g., DD) of a subject. In some embodiments, distinguishing ASD from non-ASD classifications / conditions comprises determining that the chance that a subject has ASD rather than a non-ASD classification is greater than about 60%, 65%, 70%, 74%, 80%, 85%, 90%, 95%, or 98% (i.e., the chance of having ASD and not having a non-ASD classification). In some embodiments, the non-ASD classification is DD. In some embodiments, the non-ASD classification is "normal".

[0151] In some embodiments, the invention provides methods for determining that a subject does not have ASD or DD.

[0152] Analytical methods

[0153] Methods for assessing ASD risk or distinguishing ASD from other non-ASD developmental disorders are described herein. In some embodiments, the risk assessment is based (at least in part) on the measurement and characterization of metabolites in a sample, such as a blood sample, from a subject. In some embodiments, a plasma sample is derived from a blood sample, and the plasma sample is analyzed.

[0154] Metabolites can be detected in a variety of ways, including chromatography and / or mass spectrometry-based assays, fluorescence assays, electrophoresis, immunoaffinity, hybridization, immunochemistry, ultraviolet spectroscopy (UV), fluorescence analysis, radiochemical analysis, near infrared spectroscopy (near IR), nuclear magnetic resonance spectroscopy (NMR), light scattering analysis (LS), and turbidimetry.

[0155] In some embodiments, metabolites are analyzed by liquid or gas chromatography or ion mobility (electrophoresis) alone or in combination with mass spectrometry or by mass spectrometry alone. Such methods have been used to identify and quantify biomolecules, such as cellular metabolites. (See, e.g., Li et al., 2000; Rowley et al., 2000; and Kuster and Mann, 1998). Mass spectrometry can be based on, for example, quadrupole ion trap or time-of-flight mass spectrometry, with single, double or triple mass-to-charge ratio scanning and / or filtering (MS, MS / MS or MS 3 ), and after a suitable ionization method such as electrospray ionization, atmospheric pressure chemical ionization, atmospheric pressure photoionization, matrix-assisted laser desorption ionization (MALDI) or surface enhanced laser desorption ionization (SELDI). (See, e.g., International Patent Application Publication Nos. WO 2004056456 and WO 2004088309). In some embodiments, the first separation of metabolites from a biological sample can be achieved using gas or liquid chromatography or ion mobility / electrophoresis. In some embodiments, the ionization of a mass spectrometry method can be achieved by electrospray ionization, atmospheric pressure chemical ionization or atmospheric pressure photoionization. In some embodiments, the mass spectrometer comprises a quadrupole, an ion trap or a time-of-flight or Fourier transform instrument.

[0156] In some embodiments, metabolites are analyzed on a mass scale by non-targeted ultra-high performance liquid or gas chromatography / electrospray or atmospheric pressure chemical ionization tandem mass spectrometry platforms optimized for identification and relative quantification of small molecule complements of biological systems (see, e.g., Evans et al., Anal. Chem., 2009, 81, 6656-6667).

[0157] In some embodiments, the first separation of metabolites from the biological sample can be achieved by using gas or liquid chromatography or ion mobility / electrophoresis. In some embodiments, the ionization of the mass spectrometry method can be achieved by electrospray ionization, atmospheric pressure chemical ionization, or atmospheric pressure photoionization. In some embodiments, the mass spectrometry instrument includes a quadrupole, ion trap, or time-of-flight or Fourier transform instrument.

[0158] In some embodiments, the blood sample containing the target metabolite is centrifuged to separate plasma from other blood components. In certain embodiments, an internal standard is unnecessary. In some embodiments, a specified amount of internal standard is added to plasma (a portion thereof), and then methanol is added to precipitate plasma components such as proteins. The precipitate is separated from the supernatant by centrifugation, and the supernatant is harvested. If the concentration of the target metabolite is increased for more accurate detection, the supernatant is evaporated, and the residue is dissolved in an appropriate amount of solvent. If the concentration of the target metabolite is undesirably high, the supernatant is diluted in a suitable solvent. An appropriate amount of sample containing metabolites is loaded onto a liquid chromatography column balanced with a suitable mixture of mobile phase A and mobile phase B. In the case of reversed phase liquid chromatography, mobile phase A is typically water with or without a small amount of additives such as formic acid, and mobile phase B is typically methanol or acetonitrile. Appropriate gradients of mobile phase A and mobile phase B are pumped through the column to achieve separation of the target metabolite by retention time or the time of elution from the column. When metabolites are eluted from the column, they are ionized and brought into the gas phase, and ions are detected and quantified by mass spectrometry. The specificity of detection is achieved by double filtering of specific precursor ions and specific product ions produced by the precursor ions. Absolute quantification can be obtained by standardizing the ion counts from the target metabolite to the ion counts produced by the known amount of the internal standard of the given metabolite, and by comparing the standardized ion counts with the calibration curve established using the known amount of pure metabolites and the internal standard. The internal standard is usually a stable isotope labeled form of the pure metabolite or a pure form of the structural analog of the metabolite. Alternatively, the relative quantification of a given metabolite expressed in arbitrary units can be calculated by standardizing for a selected internal reference value (e.g., the median value of the metabolite levels of all samples run in a given group).

[0159] In some embodiments, one or more metabolites are measured by immunoassay.

[0160] Many specific immunoassay formats and their variants can be used for the measurement of metabolites. (See, e.g., E. Maggio, Enzyme-Immunoassay, (1980) (CRC Press, Inc., Boca Raton, Fla.); see also U.S. Pat. No. 4,727,022 "Methods for Modulating Ligand-Receptor Interactions and their Application"; U.S. Pat. No. 4,659,678 "Immunoassay of Antigens"; U.S. Pat. No. 4,376,110 "Immunometric Assays Using Monoclonal Antibodies"; U.S. Pat. No. 4,275,149 "Macromolecular Environment Control in Specific Receptor Assays"; U.S. Pat. No. 4,233,402 "Reagents and Method Employing Channeling" and U.S. Pat. No. 4,230,767 "Heterogenous Specific Binding Assay Employing a Coenzyme as Label"). The antibodies can be conjugated to a solid support suitable for diagnostic assays (e.g., beads such as protein A or protein G agarose, microspheres, plates, slides, or wells formed from materials such as latex or polystyrene) according to known techniques, such as passive binding. Likewise, the antibodies described herein can be conjugated to a detectable label or group, such as a radiolabel (e.g., 35 S. 125 I. 131 I), enzyme labels (e.g., horseradish peroxidase, alkaline phosphatase), and fluorescent labels (e.g., fluorescein, Alexa, green fluorescent protein).

[0161] Determining ASD Risk

[0162] In some embodiments, the methods of the present invention allow one of skill in the art to identify, diagnose, or otherwise assess a subject based, at least in part, on measuring metabolite levels in a sample obtained from a subject who may not currently exhibit signs or symptoms of ASD and / or other developmental disorders, but who nonetheless may be at risk for having or developing ASD and / or other developmental disorders.

[0163] In certain embodiments, the level of a metabolite or other analyte (e.g., proteome or genomic information) can be measured in a test sample, and the level can be compared to a normal control level or to a level in a subject with a developmental disorder, illness, or classification (e.g., non-ASD developmental delay, DD) that is not ASD. In some embodiments, the term "normal control level" refers to the level of one or more metabolites or other analytes or indices typically found in subjects who do not suffer from ASD or are unlikely to suffer from ASD or other developmental disorders. In some embodiments, the normal control level is a range or index. In some embodiments, the normal control level is determined from a database of previously tested subjects. The difference in the level of one or more metabolites or other analytes compared to the normal control level can indicate that the subject has ASD or is at risk of developing ASD.

[0164] Conversely, a lack of difference in the level of one or more metabolites compared to a normal control level of one or more metabolites or other analytes can indicate that the subject does not have an ASD or is at low risk for developing an ASD.

[0165] In some embodiments, the reference value is a value obtained from a control subject or population whose diagnosis is known (i.e., diagnosed as having or identified as having ASD, or diagnosed as not having or identified as not having ASD). In some embodiments, the reference value is an index value or a baseline value, such as a "normal control level" as described herein. In some embodiments, the reference sample or index value or baseline value is taken from or derived from one or more subjects who have been exposed to ASD treatment, or may be taken from or derived from one or more subjects at low risk of developing ASD, or may be taken from or derived from subjects who have shown improvement in ASD risk factors due to exposure to treatment. In some embodiments, the reference sample or index value or baseline value is taken from or derived from one or more subjects who have not been exposed to ASD treatment. In some embodiments, samples are collected from subjects who have received initial treatment for ASD and / or subsequent treatment for ASD to monitor treatment progress. In some embodiments, the reference value has been derived from a risk prediction algorithm or calculated index from a population study of ASD. In some embodiments, the reference value is from a subject or population having a disease or condition other than ASD, such as another developmental disorder, eg, non-ASD developmental delay (DD).

[0166] In some embodiments, the difference in metabolite levels measured by the methods of the present invention includes an increase or decrease in metabolite levels compared to normal control levels, reference values, index values, or baseline values. In some embodiments, an increase or decrease in metabolite levels relative to a reference value from a normal control population, a general population, or a population with another disease indicates the presence of ASD, the progression of ASD, the aggravation of ASD, or the improvement of ASD or ASD symptoms. In some embodiments, an increase or decrease in metabolite levels relative to a reference value from a normal control population, a general population, or a population with another disease indicates an increase or decrease in the risk of developing ASD or complications associated therewith. An increase or decrease may indicate the success of one or more treatment regimens for ASD, or may indicate an improvement or regression of ASD risk factors. An increase or decrease may be, for example, at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, or at least 50% of a reference value.

[0167] In some embodiments, the difference in metabolite levels as described herein is a statistically significant difference. "Statistically significant" refers to a difference that is greater than the expected difference that occurs only randomly. Statistical significance can be determined by any method known in the art. For example, statistical significance can be determined by p-value. The p-value is a probability measure of the difference that occurs by chance between groups during the experiment. For example, a p-value of 0.01 means that the probability of the result occurring by chance is 1 / 100. The lower the p-value, the less likely the difference in the measurement between the groups is to be accidental. If the p-value is equal to or less than 0.05, the difference is considered to be statistically significant. In some embodiments, the statistically significant p-value is equal to or less than 0.04, 0.03, 0.02, 0.01, 0.005 or 0.001. In some embodiments, the statistically significant p-value is equal to or less than 0.30, 0.25, 0.20, 0.15 or 0.10 (for example, in the case of identifying whether a single specific metabolite has a predictive value of additiveness when used in a classifier including other metabolites). In some embodiments, the p-value is determined by a t-test. In some embodiments, the p-value is obtained by Fisher's test.In some embodiments, statistical significance is achieved by analyzing a combination of several metabolites in the panel and combining with a mathematical algorithm to achieve a statistically significant risk prediction.

[0168] Classification tests, assays or methods have an associated ROC curve (receiver operating characteristic curve) that plots the false positive rate (1-specificity) against the true positive rate (sensitivity). The area under the ROC curve (AUC) is a measure of the degree to which a classifier can distinguish between two diagnostic groups. The maximum AUC is 1.0 (a perfect test) and the minimum area is 0.5 (e.g., an area in which normal is not distinguished from disease). It should be understood that when the AUC approaches 1, the accuracy of the test increases.

[0169] In some embodiments, a high degree of risk prediction accuracy is a test or assay in which the AUC is at least 0.60. In some embodiments, a high degree of risk prediction accuracy is a test or assay in which the AUC is at least 0.65, at least 0.70, at least 0.75, at least 0.80, at least 0.85, at least 0.90, or at least 0.95.

[0170] Predicting ASD risk by assessing tail effects

[0171] In some embodiments, the average difference in metabolite levels is assessed between populations or between populations (e.g., between an ASD population and a DD population) or compared to a normal control population. In some embodiments, the enrichment of metabolites from samples of a given population (i.e., ASD) in the tail of the distribution curve is assessed. That is, it is determined whether a larger proportion of samples from a specified population (e.g., ASD) are present in the tail of the distribution curve compared to a second population (e.g., DD) (i.e., "tail effect"). In some embodiments, the mean difference and tail effect are identified and utilized. In some embodiments, the tail is determined by a predetermined threshold. For example, if the measured value of a sample for a certain metabolite is higher than the value corresponding to the 90th percentile (right tail or upper tail) of the metabolite in the population, or lower than the value corresponding to the 15th percentile (left tail or lower tail), it is designated as being located in the tail. In some embodiments, the threshold for the right (upper) tail of a given metabolite is a value corresponding to the 80th, 81st, 82nd, 83rd, 84th, 85th, 86th, 87th, 88th, 89th, 90th, 91st, 92nd, 93rd, 94th, 95th, 96th, 97th, 98th, or 99th percentile (e.g., if a sample's measurement for a given metabolite is above the value associated with that percentile, it is designated in the right tail). In some embodiments, the threshold for the left (lower) tail of a given metabolite is a value corresponding to the 25th, 24th, 23rd, 22nd, 21st, 20th, 19th, 18th, 17th, 16th, 15th, 14th, 13th, 12th, 11th, 10th, 9th, 8th, 7th, 6th, 5th, 4th, 3rd, 2nd, or 1st percentile (e.g., if a sample's measurement for a given metabolite is below the value associated with that percentile, it is assigned to the left tail). The percentile values ​​displayed include decimal values.

[0172] In some embodiments, a distribution curve is generated from a graph of metabolite levels of one or more populations. In some embodiments, a distribution curve is generated from a single reference population (e.g., a general population). In some embodiments, a distribution curve is generated from two populations, such as an ASD population and a non-ASD population (such as DD). In some embodiments, a distribution curve is generated from three or more populations (e.g., an ASD population, a non-ASD population but with another developmental disorder / disorder / classification such as DD, and a healthy control population such as a control population without developmental disorders). The metabolite distribution curve from each population can be used to perform more than one risk assessment (e.g., diagnosing ASD, diagnosing DD, distinguishing ASD from DD). The assessment method using the tail effect described herein can be applied to more than two populations.

[0173] In some embodiments, multiple metabolites and their distribution are used for risk assessment. In some embodiments, the levels of two or more metabolites are used to predict ASD risk. In some embodiments, at least two metabolites are selected from the metabolites listed in Table 1. In some embodiments, at least three metabolites are selected from the metabolites listed in Table 1. In some embodiments, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or 21 metabolites selected from the metabolites listed in Table 1 are used to predict ASD risk.

[0174] Further discussion of Table 1 (Tables 1A to 1C) is provided in the Examples section below.

[0175] Table 1A. Exemplary 21-metabolite panel with tail effects predictive of ASD vs. DD

[0176]

[0177]

[0178] Table 1B. Exemplary metabolites with tail enrichment predictive of ASD

[0179]

[0180] Table 1C. Exemplary metabolites with tail enrichment predictive of DD

[0181]

[0182]

[0183] In some embodiments, the at least two metabolites analyzed are selected from phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfuric acid, pantothenic acid (vitamin B5), hydroxy-thiophanate-methyl, and combinations thereof.

[0184] In some embodiments, the at least three metabolites analyzed are selected from phenylacetylglutamine, xanthine, octenylcarnitine, p-cresol sulfate, isovalerylglycine, γ-CEHC, indoleacetic acid, pipecolic acid, 1,5-anhydroglucitol (1,5-AG), lactic acid, 3-(3-hydroxyphenyl)propionic acid, 3-indoxylsulfuric acid, pantothenic acid (vitamin B5), hydroxy-thiophanate-methyl, and combinations thereof.

[0185] In some embodiments, information about the lack of tail effects for a particular set of metabolites is used for risk assessment. In some embodiments, the lack of tail effects is determined to provide a null result (i.e., no information, as opposed to negative information). In some embodiments, the lack of tail effects is determined to indicate one classification over another (e.g., more indicative of DD relative to ASD).

[0186] In some embodiments, the distribution curve is asymmetric or non-Gaussian.

[0187] In some embodiments, the distribution curve does not follow a parametric distribution pattern.

[0188] In some embodiments, information from the mean difference (eg, mean deviation) is combined with tail effect information for risk assessment. In some embodiments, information from the mean difference is used for risk assessment without using tail effect information.

[0189] In some embodiments, analysis of metabolites is combined with other types of information, such as genetic information, demographic information, and / or behavioral assessments, to determine a subject's risk for developing ASD or other disorders.

[0190] In some embodiments, an ASD risk assessment is performed based at least in part on the measured amounts of certain metabolites in a biological sample (e.g., blood, plasma, urine, saliva, feces) obtained from a subject, wherein certain metabolites are found herein to exhibit a "tail effect". The inventors have found that there is not necessarily a statistically significant mean shift between two populations associated with a tail effect. Therefore, the tail effect is a specific phenomenon different from the mean shift.

[0191] In certain embodiments, a particular metabolite exhibits a right tail effect indicative of an ASD population rather than a non-ASD population (e.g., a DD population) when the metabolite is characterized as:

[0192] ·● establishing a non-ASD population distribution curve for the metabolite in a non-ASD population (e.g., a DD population), wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population;

[0193] ·● establishing an ASD population distribution curve for the metabolite in the ASD population, wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population; and

[0194] The non-ASD population distribution curve and the ASD population distribution curve are characterized by one or both of (A) and (B):

[0195] o (A) the ratio of (i) the area under the distribution curve for the ASD population for x>level n of the metabolite compared to (ii) the area under the distribution curve for the non-ASD population for x>level n of the metabolite is greater than 150% (e.g., >200%, >300%, >500%, >1000%, etc.), thereby providing predictive utility for distinguishing the ASD classification from the non-ASD classification for samples having >level n of the metabolite, and

[0196] ○(B) if n' is the minimum threshold metabolite level corresponding to the top decile (or any cutoff from about 5% to about 20%) of the combined non-ASD and ASD populations used to generate the distribution curve, then for an unknown sample (e.g., a random sample selected from a population having equal numbers of ASD and non-ASD members) with a metabolite level of at least n', the odds that the sample is ASD rather than non-ASD are no less than 1.6:1 (e.g., no less than 2:1, no less than 3:1, no less than 4:1, no less than 5:1, no less than 6:1, no less than 7:1, no less than 8:1, no less than 9:1, or no less than 10:1) (e.g., where p<0.3, p<0.2, p<0.1, p<0.05, p<0.03, or p<0.01, e.g., a statistically significant classification), thereby providing predictive utility for distinguishing ASD classification from non-ASD classification for samples having levels > n' of that metabolite.

[0197] In certain embodiments, a particular metabolite exhibits a left-tail effect indicative of an ASD but not a non-ASD population (e.g., a DD population) when the metabolite is characterized as:

[0198] ● establishing a non-ASD population distribution curve for the metabolite in a non-ASD population (e.g., a DD population), wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population;

[0199] ● establishing an ASD population distribution curve for the metabolite in the ASD population, wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population; and

[0200] The non-ASD population distribution curve and the ASD population distribution curve are characterized by one or both of (A) and (B):

[0201] o (A) the ratio of (i) the area under the distribution curve for the ASD population for x < level m of the metabolite compared to (ii) the area under the distribution curve for the non-ASD population for x < level m of the metabolite is greater than 150% (e.g., >200%, >300%, >500%, >1000%, etc.), thereby providing predictive utility for distinguishing the ASD classification from the non-ASD classification for samples having < level m of the metabolite, and

[0202] ○(B) If m' is the maximum threshold metabolite level corresponding to the bottom decile (or any cut-off from about 5% to about 20%) of the combined non-ASD and ASD populations used to generate the distribution curve, then for an unknown sample having a metabolite level below m' (e.g., a random sample selected from a population having an equal number of ASD and non-ASD members), the odds that the sample is ASD rather than non-ASD are not less than 1.6:1 (e.g., not less than 2:1, not less than 3:1, not less than 4:1, not less than 5:1, not less than 6:1, not less than 7:1, not less than 8:1, not less than 9:1 or not less than 10:1) (e.g., where p < 0.3, p < 0.2, p < 0.1, p < 0.05, p < 0.03 or p < 0.01, e.g., a statistically significant classification), thereby providing predictive utility for differentiating ASD classification from non-ASD classification of samples having a metabolite level < m'.

[0203] In certain embodiments, when a metabolite is characterized as follows, the particular metabolite exhibits a right-tail effect indicative of a non-ASD population (e.g., a DD population) rather than an ASD population:

[0204] ·● Establish a non-ASD population distribution curve for the metabolite in the non-ASD population (e.g., the DD population), where the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population;

[0205] ·● Establish an ASD population distribution curve for the metabolite in the ASD population, where the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population; and

[0206] ·● The non-ASD population distribution curve and the ASD population distribution curve are characterized by one or both of (A) and (B):

[0207] ○(A)(i) The ratio of the area under the non-ASD population distribution curve for x > the metabolite level n to (ii) the area under the ASD population distribution curve for x > the metabolite level n is greater than 150% (e.g., > 200%, > 300%, > 500%, > 1000%, etc.), thereby providing predictive utility for differentiating non-ASD classification from ASD classification of samples having > the metabolite level n, and

[0208] ○(B) if n' is the minimum threshold metabolite level corresponding to the top decile (or any cutoff from about 5% to about 20%) of the combined non-ASD and ASD populations used to generate the distribution curve, then for an unknown sample (e.g., a random sample selected from a population having equal numbers of ASD and non-ASD members) with a metabolite level of at least n', the odds that the sample is non-ASD rather than ASD are no less than 1.6:1 (e.g., no less than 2:1, no less than 3:1, no less than 4:1, no less than 5:1, no less than 6:1, no less than 7:1, no less than 8:1, no less than 9:1, or no less than 10:1) (e.g., where p<0.3, p<0.2, p<0.1, p<0.05, p<0.03, or p<0.01, e.g., a statistically significant classification), thereby providing predictive utility for distinguishing a non-ASD classification from an ASD classification for samples having a level > n' of the metabolite.

[0209] In certain embodiments, a particular metabolite exhibits a left-tail effect indicative of a non-ASD population (e.g., a DD population) rather than ASD when the metabolite is characterized as:

[0210] ● establishing a non-ASD population distribution curve for the metabolite in a non-ASD population (e.g., a DD population), wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population;

[0211] ● establishing an ASD population distribution curve for the metabolite in the ASD population, wherein the x-axis represents the level of the first metabolite and the y-axis represents the corresponding population; and

[0212] The non-ASD population distribution curve and the ASD population distribution curve are characterized by one or both of (A) and (B):

[0213] o (A) the ratio of (i) the area under the distribution curve for the non-ASD population for x < level m of the metabolite compared to (ii) the area under the distribution curve for the ASD population for x < level m of the metabolite is greater than 150% (e.g., >200%, >300%, >500%, >1000%, etc.), thereby providing predictive utility for distinguishing the non-ASD classification from the ASD classification for samples having < level m of the metabolite, and

[0214] ○(B) If m' is the maximum threshold metabolite level corresponding to the bottom decile (or any cut-off value from about 5% to about 20%) of the combined non-ASD and ASD populations used to generate the distribution curve, then for an unknown sample having a metabolite level below m' (e.g., a random sample selected from a population having an equal number of ASD and non-ASD members), the odds that the sample is non-ASD rather than ASD are not less than 1.6:1 (e.g., not less than 2:1, not less than 3:1, not less than 4:1, not less than 5:1, not less than 6:1, not less than 7:1, not less than 8:1, not less than 9:1 or not less than 10:1) (e.g., where p < 0.3, p < 0.2, p < 0.1, p < 0.05, p < 0.03 or p < 0.01, e.g., a statistically significant classification), thereby providing predictive utility for differentiating non-ASD classification from ASD classification of samples having a metabolite level < m'.

[0215] In certain embodiments, risk assessment is performed using multiple metabolites that exhibit tail effects. It has been observed that for the assessment of ASD, there are certain metabolite groups (e.g., two or more metabolites) that can provide complementary diagnostic / risk assessment information. For example, ASD-positive individuals identified by analyzing the level of a first metabolite (e.g., individuals within the identified tail of the first metabolite) are different from (or there may be a low, non-zero overlap with) ASD-positive individuals identified by analyzing a second metabolite. The tail of the first metabolite predicts certain ASD individuals, while the tail of the second metabolite predicts other ASD individuals. Without wishing to be bound by a particular theory, this finding may reflect the multifaceted nature of ASD itself.

[0216] Accordingly, in certain embodiments, the risk assessment method includes determining whether a subject falls into any of multiple identified metabolite tails involving multiple metabolites, e.g., when the predictors of different metabolite tails are at least partially unconnected, e.g., they have low mutual information, and thus, when multiple metabolites with low mutual information are introduced, risk prediction is improved.

[0217] In some embodiments, the present invention provides one or more binding members for use in a method of determining / diagnosing a subject having ASD or at risk of ASD, each binding member being capable of specifically binding to a metabolite selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionate, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoleoxysulfate, 4-ethylphenylsulfate, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine and 3-hydroxyhippuric acid.

[0218] In some embodiments, the present invention provides for the use of one or more binding members for diagnosing / determining a subject having ASD or at risk of having ASD, wherein the binding member is capable of specifically binding to a metabolite selected from the group consisting of 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfate, 4-ethylphenylsulfate, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine and 3-hydroxyhippuric acid.

[0219] The binding member may be selected from nucleic acid molecules, proteins, peptides, antibodies or fragments thereof, all of which are capable of binding to a specific metabolite. In certain embodiments, the binding member may be labeled to aid in the detection and determination of the level of the bound metabolite.

[0220] The present invention also provides a plurality of binding members fixed to a solid support, wherein the plurality of binding members are selected from 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (C5), indoleacetic acid, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylate, octenoylcarnitine and 3-hydroxyhippuric acid, wherein the plurality of binding members account for at least 20%, 30%, 40%, 50%, 60%, 70%, 80% or 90% of the binding member population on the solid support.

[0221] In some embodiments, the present invention provides kits for implementing the methods described herein, particularly kits for determining / diagnosing that a subject has ASD or is at risk of having ASD. The kit allows the user to determine the presence, level (increase or decrease) of one or more metabolites selected from the following metabolites in a sample under test: 5-hydroxyindoleacetic acid (5-HIAA), 1,5-anhydroglucitol (1,5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfate, 4-ethylphenylsulfate, 8-hydroxyoctanoic acid, γ-CEHC, hydroxyisovalerylcarnitine (5), indoleacetic acid, isovalerylglycine Acid, lactic acid, N1-methyl-2-pyridone-5-carboxamide, p-cresol sulfate, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, xanthine, hydroxy-thiocarboxylic acid, octenylcarnitine and 3-hydroxyhippuric acid; the kit comprises (a) a solid support having a plurality of binding members, wherein each of the binding members is independently specific for one of the one or more metabolites immobilized thereon; (b) a developer comprising a label; and optionally (c) one or more components selected from a washing solution, a diluent and a buffer. Example

[0222] Subjects

[0223] Blood samples were collected from subjects aged 18 to 60 months who were referred to 19 developmental assessment centers for evaluation of possible developmental disorders other than isolated activity problems. Informed consent was obtained from all subjects. Subjects with a prior diagnosis of ASD at a clinic specializing in pediatric developmental assessment or who were unable or unwilling to complete the study procedures were excluded from the study.

[0224] The subjects were those enrolled in the SynapDx Autism Spectrum Disorders Gene Expression Analysis (STORY) study. The STORY study was conducted in accordance with current ICH guidelines on Good Clinical Practice (GCP) and applicable regulatory requirements. GCP is an international ethical and scientific quality standard for designing, conducting, recording and reporting studies involving human subjects. Compliance with this standard provides public assurance that the rights, safety and health of research subjects are protected and that clinical research data are credible, in accordance with the principles derived from the Declaration of Helsinki.

[0225] Figures 1 to 12 The results shown in are based on 180 blood samples from males in the STORY study. The sample set includes 122 ASD samples and 58 DD (non-ASD) samples. ASD diagnosis follows the DSM-V diagnostic criteria. The other results are based on a wider collection of 299 blood samples from male subjects in the STORY study. The wider sample group includes 198 ASD samples and 101 DD samples.

[0226] For all tests, about 3 mL of blood samples were collected in EDTA tubes and plasma was prepared by centrifuge tubes. The plasma was then frozen and transported to the laboratory for analysis. Methanol extraction of the samples was performed in the laboratory, and the extracts were analyzed by optimized ultra-high performance liquid or gas chromatography / tandem mass spectrometry (UHPLC / MS / MS or GC / MS / MS) methods (see, e.g., Anal. Chem., 2009, 81, 6656-6667).

[0227] Data analysis

[0228] Metabolites were quantified in blood samples of male and female subjects.

[0229] The samples were analyzed for metabolite levels and quantified as concentrations in arbitrary units normalized to the median concentration measured for all samples on a given day. For example, a unit greater than 1 refers to the amount of the metabolite greater than the median of the samples for that day, and a unit less than 1 refers to the amount less than the median. A cross-validation was then performed, in which the samples were randomly divided into non-overlapping training / test sets, for which the unbiased performance of the machine learning classifier was evaluated. 21 metabolites have been identified that are highly informative in predicting ASD, both individually and collectively, particularly in male subjects.

[0230] Example 1: Identifying metabolite level information

[0231] This example shows that valuable information for ASD risk assessment can be discerned from the identification and analysis of tail effects in sample distributions that would otherwise be missed by conventional analyses (eg, analyses based on mean shifts).

[0232] Once metabolite levels are determined, there are multiple ways to implement this information for risk assessment, including mean shift and tail effects. Surprisingly, mean shift was found to provide some, but not optimal, predictive information. Figure 1 An exemplary mean shift is shown in . In this figure, the ASD distribution is shifted to the right of the non-ASD distribution (DD).

[0233] In addition to conventional mean shift analysis, the inventors also discerned additional information from the samples. Metabolite distribution curves were plotted for ASD and non-ASD (here DD) samples, and it was found that for the subgroups of metabolites measured, samples from the ASD or DD population were enriched in the right (upper) or left (lower) tail (i.e., tail effect). Representative tail effects are shown in Figure 2 Of note, both distributions share nearly identical means (i.e., minimal or no mean shift or average shift). Therefore, the predicted value of the metabolites cannot be discerned from conventional analysis of the mean shift.

[0234] Metabolites may show right (upper) tail effects or left (lower) tail effects, or both. The ASD and non-ASD (here DD) distribution curves of the representative metabolite 5-HIAA are shown in Figure 3 A clear right tail effect was observed, e.g., the ASD distribution had a larger AUC in the right tail. Thus, it was demonstrated that samples with high levels of this metabolite were highly enriched in ASD population members. With this metabolite, both the mean shift (indicated by the t-test value) and the right tail (indicated by the "extreme" Fisher test value) were statistically significant.

[0235] The ASD and non-ASD (here DD) distribution curves of another illustrative metabolite, γ-CEHC, are shown in Figure 4 A clear left tail effect was observed, e.g., the ASD distribution had a larger AUC in the left tail. Thus, it was demonstrated that samples with low levels of this metabolite were highly enriched in ASD population members. For this metabolite, the mean shift (indicated by the t-test value) was not statistically significant, while the left tail was statistically significant.

[0236] These data illustrate that the identification and analysis of tail effects provides additional information for risk assessment that cannot be obtained through conventional mean shift analysis.

[0237] Example 2: Selected metabolites showing tail effects strongly predict ASD

[0238] This example illustrates the evaluation of tail effects for predicting ASD. The inventors identified statistically significant tail effects for many metabolites in samples obtained from male subjects. The tail effects provide information about which group (i.e., ASD group or DD group) the subject belongs to, both individually and collectively. Table 1 shows an exemplary panel of 21 metabolites that exhibit ASD vs. DD tail effects with high predictive power.

[0239] Table 1B shows metabolites of 21 metabolite panels with tail effects predicting ASD. The statistical significance (p-value) of each tail effect and its position on the distribution curve (i.e., left tail effect or right tail effect) are indicated. An odds ratio greater than 1 indicates predictive power for ASD. For example, 5HIAA has a right tail with an odds ratio of 4.91, indicating that in the STORY research data set (wherein the ratio of ASD to DD samples is 2:1), there are about 10 ASD samples in the right tail for each DD sample. Confidence intervals were estimated by the bootstrap method. Resampling was performed by replacement, and 1000 separate bootstraps were generated from the STORY data. For each bootstrap, the position of the tail and the corresponding odds ratio were determined. 90% confidence intervals were calculated from the distribution of the observed odds ratios.

[0240] Based on these criteria, 19 metabolites out of the 21 metabolite panel were found to be predictive for ASD.

[0241] Table 1C shows metabolites with tail effects that predict DD. The statistical significance (p-value) of each tail effect and its position on the distribution curve (i.e., left tail effect or right tail effect) are indicated. An odds ratio less than 1 indicates the predictive power of DD. Based on these criteria, 8 metabolites in the 21 metabolite panels were found to predict DD. For ASD, considering that the ratio of DD to ASD samples in the STORY study was 1:2, the odds ratio and 90% confidence interval were similarly determined.

[0242] It is noteworthy that some metabolites showed a single-tail effect (either left or right) that was predictive for ASD or DD, while other metabolites showed both left and right tail effects, which together provided predictive power for ASD and DD. For example, phenylacetylglutamine and p-cresol sulfate showed both left and right tail effects.

[0243] The tail effects of the 21 metabolites listed in Table 1 are Figures 13A to 13U For each figure, the distribution of one metabolite in both the ASD and DD populations is shown. The legend at the top of each figure shows the statistical significance (p-value generated by Fisher's test) for the left and right tails of that metabolite.

[0244] Some metabolites, such as phenylacetylglutamine, showed mean changes and tail effects. Figure 5 As shown, phenylacetylglutamine showed a statistically significant mean shift (t-test; p=0.001) and statistically significant left and right tail effects between the two groups ("extreme value" indicates tail effect, Fisher test; p=0.0001). The distribution showed a shifted Gaussian curve between the ASD and DD groups.

[0245] Table 2 shows the thresholds used to determine the tail effects of the 21 metabolite panel based on the potential population distribution of each metabolite in the ASD and non-ASD populations. Exemplarily, the upper threshold corresponds to the 90th percentile distribution, while the lower threshold corresponds to the 15th percentile distribution. The absolute measurement of the threshold (e.g., ng / ml, nM, etc.) can be calculated by using the values ​​in Table 2 with the average concentration of the metabolite in the population.

[0246] Table 2. Threshold levels for the left tail (equal to or below the 15th percentile) and right tail (equal to or above the 90th percentile) of the metabolite distribution curve

[0247]

[0248]

[0249] Example 3: Prediction of ASD using multiple metabolites

[0250] The information provided by multiple metabolites (e.g., those listed in Table 1) can be used individually or as a group to assist in disease risk prediction. Particularly informative metabolite groups include members that are uncorrelated with each other and have low collinearity (i.e., low mutuality). For example, Figure 6 5HIAA levels compared to γ-CEHC levels are shown, indicating a lack of correlation between the informative levels of the two metabolites. For example, ASD individuals identified in the tail of 5HIAA ( Figure 3 ) are usually not identical to the ASD individuals identified in the tail of γ-CEHC. Therefore, metabolites 5HIAA and γ-CEHC are considered to provide complementary information. The tail enriched for metabolites with low commonality provides complementary classification information.

[0251] Figure 7 is a graph indicating, for each of the 180 samples, whether the sample is within the tail of each metabolite of the 12-metabolite panel. In this exemplary panel, the tails of two metabolites (xanthine and p-cresol sulfate) are predictive for non-ASD (e.g., DD), while the tails of the other 10 metabolites predict ASD.

[0252] When multiple metabolites are evaluated, the number of combinations of tail effect summed counts increases, and the potential tail effect summed counts increase. When unknown samples are measured, the distribution of tail effect summed counts from ASD and non-ASD populations can be plotted and the resulting distribution used to determine the appropriate separation between ASD and non-ASD. Fig. 8A As shown, ASD and non-ASD (here DD) samples can be further analyzed by using a voting (e.g., binning) scheme to further exploit the complementary information provided by the metabolites for which tail effects are observed. Data for a total of 12 metabolites are shown. In one specific scheme, for a given sample, the number of metabolites for which the sample falls within the predictive tail of ASD is summed, and the number of metabolites for which the sample falls within the predictive tail of non-ASD (here DD) is also summed. These two values ​​are plotted as x and y coordinates ( Fig. 8A ). It is noteworthy that as the number of ASD-enriched metabolites increases (higher on the y-axis) and as the number of non-ASD-enriched metabolites decreases (lower on the x-axis), there appears to be less mixing of non-ASD points in the ASD points, for example, suggesting a lower likelihood of a false-positive diagnosis of ASD. On the other hand, as the number of ASD-enriched metabolites decreases (lower on the y-axis) and as the number of non-ASD-enriched metabolites increases (higher on the x-axis), there appears to be less mixing of ASD points in the non-ASD points, for example, suggesting a lower likelihood of a false-positive diagnosis of DD.

[0253] The samples were divided into four different bins, e.g. Figure 8B The top and bottom right boxes in particular show clear separation, facilitating ASD or DD risk assessment.

[0254] Of the four bins shown in FIG8 , the bin that is most strongly predictive of ASD includes samples with 2 or more ASD-enriched features and 0 or 1 non-ASD-enriched features. Bins with 1 ASD-enriched feature and 0 or 1 non-ASD-enriched features also predict ASD, although not as strongly as the above bins. Bins with 1 or more non-ASD-enriched features and 0 ASD-enriched features strongly predict non-ASD. In some cases, sample bins with no ASD-enriched features and no non-ASD-enriched features can also provide predictive information.

[0255] In an exemplary voting scheme, a vote is performed for a given sample, for example, metabolites enriched in ASD receive a point score and metabolites enriched in non-ASD lose a point score. Samples with positive results (e.g., equal to or greater than 1) can be considered ASD (or have a significant risk of ASD), and samples with negative results (equal to or less than -1) can be considered non-ASD (or have a significant likelihood of non-ASD). Samples with zero results can be considered likely to be non-ASD or ASD, depending on the distribution of ASD compared to non-ASD in the sample, or can be returned as uncertain or "unclassified result" samples. Similarly, Figure 8C Voting results for the panel of 21 metabolites described in Table 1 are shown.

[0256] In another exemplary scoring system ( Fig.21 For each metabolite, the log 2 values ​​of the odds ratios (log2 OR) of ASD and DD signatures were summed to calculate the risk score for ASD or DD.

[0257] The tail effect information can be used to distinguish subjects with ASD or non-ASD conditions. Likewise, the tail effect information can be used to predict the risk of a subject developing another disease or condition (e.g., DD).

[0258] For example, Fig. 8A and 8C The tail effect distribution of the non-ASD population (e.g., DD) shown can be used to establish a reference value for the sum of the average tail effects of a given number of metabolites in the population. This average can be used as a reference to compare the sum of the average tail effects of samples from unknown subjects and can be used to assess the risk of a subject suffering from ASD without having to obtain a population distribution curve for the metabolite in the ASD and non-ASD populations.

[0259] Tail effect information may also be combined with conventional mean shift information and / or other classification information for improved classification results, for example, as described in the exemplary voting scheme above or similar schemes.

[0260] This paper demonstrates that the predictability of ASD risk can be improved by analyzing a combination of certain metabolites. For example, Figure 9-1 1 and Figures 13-14A-D illustrate how using a voting scheme can increase the AUC of the classifier and improve the predictive power. When using subgroups of the 12 metabolite panel, the ASD predictive power increases (y-axis) as the number of metabolites in the subgroup increases (from 1 to 12) ( Fig. 9). Using different classifiers (i.e., logistic regression, naive Bayes, or support vector machine (SVM)) and different feature selection also affect the AUC ( Fig. 9 ). Fig. 10A showed a trichotomous prediction of ASD risk using a 12-metabolite panel for the same population using different signatures and classifiers, whereas Fig. 10B Results using a panel of 21 metabolites are shown. Fig.11A and 11B The results show that the use of a 12-metabolite panel ( Fig.11A ) and 21 metabolite panels ( Fig. 11B ) voting scheme. These analyses suggest that a high degree of confidence in ASD risk assessment can be achieved by selecting target metabolites and using appropriate statistical tools. Fig.12 As shown, an AUC of at least 0.74 was obtained using 12 metabolites according to the above method.

[0261] Example 4: Selection of high impact metabolites from metabolomics data

[0262] Samples from ASD and DD subjects were screened for detection of approximately 600 known metabolites (shown in Table 3). From the initial set of 600 metabolites, 84 candidate metabolites were identified that exhibited a tail effect. Subgroups of the 84 metabolites detected in the samples were elucidated and their names are listed in Table 4. Metabolite groups (e.g., 12 and 21 metabolite panels) were selected from the set of 84 candidate metabolites based on high individual metabolite AUCs. Certain candidate metabolites were excluded from the panel based on factors such as association with medication or age.

[0263] Table 3: Four hundred sixty five (465) elucidated metabolites from the initial set of 600 metabolites assayed

[0264]

[0265]

[0266]

[0267]

[0268]

[0269]

[0270] Table 4: Identified candidate metabolites showing tail effects

[0271]

[0272]

[0273]

[0274] Two groups of metabolites were tested (by Figure 7 The 12 metabolite panels consisting of the metabolites of Table 1 and the 21 metabolite panels consisting of the metabolites of Table 1 were used for ASD risk prediction. The results showed that the 12 and 21 metabolite panels strongly contributed to the prediction of ASD. An overview of the effects of including and excluding the metabolites of the 12 or 21 metabolite panels on ASD prediction is shown in Fig.14A -D. The white list represents the AUC value of the classifier using data from only 12 or 21 metabolite panels, while the black list represents the AUC value of the classifier using 600 metabolites from a group of 84 candidate metabolites or a complete group (all candidates = 84 candidate metabolites; all features = 600 metabolites) without including 12 or 21 metabolite panels. Average shift (upper figure) and tail analysis (lower figure) were performed. These data show that the predictive information of ASD is attributed to the metabolites in the 12 or 21 metabolite panels, whether evaluated by average shift or tail analysis. Therefore, it is observed that the ability of metabolites (metabolites in 12 and 21 metabolite panels) showing strong tail effects to predict ASD over DD is much greater than other metabolites from 600 metabolite panels (which do not show strong tail effects).

[0275] Fig. 14C -D expands the Fig.14A -B and include additional analyses using naive Bayes analysis in addition to logistic regression. Fig. 14B Also shown are the results broken down into different sample groups (i.e., "Christmas" and "Easter"). The leftmost panel shows the AUC results where the classifier was trained on 192 samples and cross-validated only in the Christmas group; the middle left panel shows the AUC results where the classifier was trained on 299 samples and cross-validated on the Christmas and Easter groups; the middle right panel shows the AUC results where the classifier was trained only on samples from the Easter group and cross-validated only in the Easter group; the rightmost panel shows the AUC results where the classifier was trained on samples from the Christmas and Easter groups and cross-validated on the Easter group. The highest AUC was obtained using metabolites in the 12 or 21 metabolite panels (e.g., metabolites that exhibit a tail effect).

[0276] AUC predictions are shown by including panels of 12 or 21 metabolites (white list) and excluding them (black list), depending on the number of features added to the statistical analysis. Fig.17A -B, 18A-B and 19A-D further expanded Fig.14A -D. The top panels show the results of the mean shift analysis, while those at the bottom show the tail effect analysis. In each individual panel, the bars represent different metabolite panels, as indicated in the symbols and legends below.

[0277] An exemplary graph depicting the cumulative AUC for ASD risk prediction when evaluating subgroups of a total of 21 metabolites is shown in Fig.15 In this figure, the x-axis shows the number of metabolites in a subgroup selected from a group of 21 metabolites. The y-axis shows the predictive power of ASD. For each number on the x-axis, many random metabolite combinations are analyzed and their AUC values ​​(points) are plotted. The curve shows the increase in AUC caused by an increase in the number of metabolites used (selected from a group of 21 metabolites). On the other hand, the figure demonstrates that even subgroups with a small number of metabolites (e.g., 3 or 5) show high AUCs. Therefore, some metabolites seem to have particularly important predictive tails.

[0278] An exemplary table describing representative subgroups of 3, 4, 5, 6, and 7 metabolites from the 21 metabolites in Table 1 that produce high AUC values ​​is shown in Table 5. For each subgroup size (3, 4, 5, 6, or 7), 50 random selections of metabolite groups were analyzed. For example, for 3 metabolite subgroups from a small group of 21 metabolites, 50 random combinations of the 3 metabolite subgroups were evaluated (combinations in a total of 1330 possible permutations). Combinations from 50 random groups with the highest AUC are shown. Therefore, certain metabolite combinations containing less than 21 metabolites produce high AUC values. Metabolites such as γ-CEHC, p-cresol sulfate, xanthine, phenylacetylglutamine, isovalerylglycine, octenoylcarnitine, and hydroxy-thiophanate-methyl appear in multiple subgroups that produce high AUC values, indicating that these metabolites may be closely related to the patient's ASD status. Therefore, these metabolites (alone or in combination with each other) or other metabolites together appear to be particularly useful for predicting the patient's ASD risk.

[0279] Table 5: Exemplary subgroups of metabolites and prediction of ASD

[0280]

[0281]

[0282] The predictability of the two-metabolite subsets of the 21 metabolites from Table 1 for ASD was evaluated in pairwise combinations. Representative pairwise combinations with robust AUCs are shown in Table 6. Similarly, the predictability of the three metabolite subsets of the 21 metabolites from Table 1 for ASD was evaluated in ternary combinations. Representative ternary combinations with robust AUCs are shown in Table 7.

[0283] Table 6. Exemplary metabolite pairs providing robust AUC

[0284]

[0285]

[0286] Table 7. Exemplary metabolite triplets providing robust AUC

[0287]

[0288]

[0289] Example 5: Validation of 12 metabolite panel classifiers

[0290] based on Figure 7 Data from 180 samples tested (about two-thirds of which were ASD) were used to generate a classifier based on the 12 highly informative metabolites shown. The ability of the classifier to distinguish ASD from non-ASD areas (here, DD) was tested in a second cohort of 130 samples. This approach provided an unbiased estimate of the true predictive performance, corresponding to an AUC of 0.74. A schematic diagram of the process is shown in Fig.12 shown.

[0291] Example 6: Adding genetic information to metabolites can improve ASD risk prediction

[0292] We found that adding genetic information to metabolite information improved ASD risk prediction for some groups. For example, combining copy number variation (CNV) data with metabolite information significantly reduced the confidence interval of ASD risk prediction, such as Fig.16A and 16B As shown. Fig.16A As shown, adding genetic information further enhanced the separation between the ASD and non-ASD groups. In addition to CNVs, other genetic information including but not limited to fragile X (FXS) status may also further aid in diagnostic testing, which can predict ASD risk with improved accuracy and reduced type I and / or type II errors. Fig. 16B As shown, including such additional information (eg, "PathoCV") increased the separation between the ASD and DD groups, thus helping to distinguish between the two disorders.

[0293] Example 7: Significant biological pathways emerging from metabolite analysis

[0294] Further analysis of the metabolite information revealed metabolite clusters presented in Table 1, which play significant roles in different biological pathways. For example, 7 of the 21 metabolites were associated with intestinal microbial activity (33%), as shown in Table 8. All 7 were amino acid metabolites. 6 of the 7 were metabolites of aromatic amino acids and had a benzene ring.

[0295] Table 8: Seven metabolites involved in intestinal microbial activity

[0296]

[0297] Analysis of metabolites strongly associated with ASD, as shown in Table 1, revealed links to certain biological pathways. For example, specific metabolites that provide predictive information for ASD suggest impaired phase II biotransformation, impaired ability to metabolize benzene rings, renal reabsorption disorders, disorders of carnitine metabolism, and imbalances in the transport of large neutral amino acids to the brain. Biological pathway information can be further used to improve ASD risk assessment and / or explore the etiology and pathophysiology of ASD. Such information can also be used to develop pharmacotherapeutic agents for the treatment of ASD.

[0298] Example 8: Elucidation of metabolite concentrations in blood

[0299] Absolute metabolite concentrations in plasma samples were determined by mass spectrometry for 19 of the 21 metabolites described in Example 2. Absolute metabolite concentrations in plasma (ng / ml) were calculated using calibration curves generated from standard samples containing known amounts of the metabolites.

[0300] Table 9A shows 17 metabolites that predict ASD. The direction of the tail effect (left or right), the threshold for determining the presence of the tail effect (i.e., the left tail effect is the 15th percentile, the right tail effect is the 90th percentile) and the odds ratio (log 2) are provided. A positive odds ratio indicates that the tail effect of the metabolite predicts ASD.

[0301] Table 9B shows 7 metabolites that predict DD. The direction of the tail effect (left or right), the threshold for determining the presence of the tail effect (i.e., the left tail effect is the 15th percentile, the right tail effect is the 90th percentile) and the odds ratio (log2) are provided. A negative odds ratio indicates that the tail effect of the metabolite predicts DD.

[0302] Table 9A. Exemplary metabolites with tail enrichment predictive of ASD

[0303]

[0304] Table 9B. Exemplary metabolites with tail enrichment predictive of DD

[0305]

[0306] The present invention relates to the following embodiments:

[0307] 1. A method of distinguishing an autism spectrum disorder (ASD) from a non-ASD developmental delay (DD) in a subject, the method comprising:

[0308] (i) measuring the levels of a plurality of metabolites in a sample obtained from a subject,

[0309] wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid (5-HIAA), indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol (5-AG), 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, hydroxyisovalerylcarnitine (C5), isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof;

[0310] (ii) calculating the number of metabolites in the sample whose levels are at or below a predetermined threshold concentration indicative of:

[0311] (a) Indicated ASD as defined in Table 9A (ASD left tail effect), or

[0312] (b) Indicated DD (DD left tail effect) as defined in Table 9B; and / or

[0313] (iii) calculating the number of metabolites in the sample whose levels are equal to or above a predetermined threshold concentration indicating:

[0314] (a) Indicated ASD as defined in Table 9A (ASD right tail effect), or

[0315] (b) indicated DD as defined in Table 9B (DD right tail effect); and

[0316] (iv) determining that the subject has ASD or DD based on the number obtained in steps (ii) and / or (iii).

[0317] 2. The method of embodiment 1, wherein the plurality of metabolites comprises xanthine and γ-CEHC.

[0318] 3. The method of embodiment 1, wherein the sample is a plasma sample.

[0319] 4. The method of embodiment 1, wherein the metabolite level is measured by mass spectrometry.

[0320] 5. The method of embodiment 1, wherein the subject is no older than about 54 months.

[0321] 6. The method of embodiment 1, wherein the subject is no older than about 36 months.

[0322] 7. A method for determining that a subject has ASD or is at risk for ASD, the method comprising:

[0323] (i) measuring the levels of a plurality of metabolites in a sample obtained from the subject,

[0324] wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid (5-HIAA), indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol (5-AG), 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF), 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, hydroxyisovalerylcarnitine (C5), isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, pantothenic acid (vitamin B5), phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof;

[0325] (ii) detecting two or more of the following:

[0326] (a) Xanthine level is 182.7 ng / ml or above;

[0327] (b) Hydroxychlorothalonil level of 20.3 ng / ml or above;

[0328] (c) 5-hydroxyindoleacetic acid level of 28.5 ng / ml or above;

[0329] (d) lactate level of 686600.0 ng / ml or above;

[0330] (e) pantothenic acid level of 63.3 ng / ml or above;

[0331] (f) pipecolic acid level of 303.6 ng / ml or above;

[0332] (g) γ-CEHC level was 32.0 ng / ml or less;

[0333] (h) indoleacetic acid level of 141.4 ng / ml or less;

[0334] (i) p-cresol sulfate level is 182.7 ng / ml or below;

[0335] (j) 1,5-anhydroglucitol (1,5-AG) level of 11910.3 ng / ml or less;

[0336] (k) 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid (CMPF) level of 7.98 ng / ml or less;

[0337] (l) 3-indoxylsulfate level of 256.7 ng / ml or less;

[0338] (m) 4-ethylphenyl sulfate level of 3.0 ng / ml or less;

[0339] (n) Hydroxyisovalerylcarnitine (C5) level of 12.9 ng / ml or less;

[0340] (o) N-methyl-2-pyridone-5-carboxamide level of 124.82 ng / ml or less; and

[0341] (p) phenylacetylglutamine level of 166.4 ng / ml or less;

[0342] (iii) determining that the subject suffers from or is at risk of ASD based on the metabolite level detected in step (ii).

[0343] 8. The method of embodiment 7, wherein the plurality of metabolites comprises xanthine and γ-CEHC.

[0344] 9. A method according to embodiment 7, wherein the sample is a plasma sample.

[0345] 10. The method of embodiment 7, wherein the metabolite level is measured by mass spectrometry.

[0346] 11. The method of embodiment 7, wherein the subject is no older than about 54 months.

[0347] 12. The method of embodiment 7, wherein the subject is no older than about 36 months.

Claims

1. Use of a reagent for determining the levels of a plurality of metabolites in the preparation of a preparation for distinguishing between autism spectrum disorders (ASD) and non-ASD developmental delays (DD) in a subject, wherein the distinction comprises: (i) measuring the levels of a plurality of metabolites in a sample obtained from a subject using the reagent, wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid, indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol, 3-(3-hydroxyphenyl)propionic acid, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid, 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, hydroxyisovalerylcarnitine, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, pantothenic acid, phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof, wherein the at least two metabolites comprise 3-indoxylsulfuric acid; wherein the agent is one or more binding members independently selected from nucleic acid molecules, proteins or fragments thereof, each binding member being capable of specifically binding to one of the at least two metabolites; and (ii) calculating the number of metabolites in the sample whose levels are at or below a predetermined threshold concentration indicative of: (a) ASD left tail effect indicating ASD as defined in the following table: or (b) The DD left tail effect of the indicated DD as defined in the following table: and / or (iii) calculating the number of metabolites in the sample whose levels are equal to or above a predetermined threshold concentration indicating: (a) ASD right tail effect indicating ASD as defined in the following table: or (b) The DD right tail effect of the indicated DD as defined in the following table: and (iv) determining that the subject has ASD or DD based on the number obtained in steps (ii) and / or (iii).

2. The method of claim 1, wherein the at least two metabolites include 3-indoxylsulfuric acid and 3-(3-hydroxyphenyl)propionic acid.

3. The use according to claim 1, wherein the sample is a plasma sample.

4. The use of claim 1, wherein the metabolite level is measured by mass spectrometry.

5. The method of claim 1, wherein the subject is no older than 54 months.

6. The method of claim 1, wherein the subject is no older than 36 months.

7. Use of reagents for measuring the levels of a plurality of metabolites in the preparation of a preparation for determining that a subject has ASD or is at risk of ASD, wherein the determination comprises: (i) measuring the levels of a plurality of metabolites in a sample obtained from the subject using the preparation, wherein the plurality of metabolites comprises at least two metabolites selected from the group consisting of xanthine, γ-CEHC, hydroxy-thiocarbamide, 5-hydroxyindoleacetic acid, indoleacetic acid, p-cresol sulfate, 1,5-anhydroglucitol, 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid, 3-indoxylsulfuric acid, 4-ethylphenylsulfuric acid, hydroxyisovalerylcarnitine, isovalerylglycine, lactic acid, N1-methyl-2-pyridone-5-carboxamide, pantothenic acid, phenylacetylglutamine, pipecolic acid, 3-hydroxyhippuric acid, and combinations thereof, wherein the at least two metabolites comprise 3-indoxylsulfuric acid; wherein the agent is one or more binding members independently selected from nucleic acid molecules, proteins or fragments thereof, each binding member being capable of specifically binding to one of the at least two metabolites; and (ii) detecting two or more of the following: (a) Xanthine level is 182.7 ng / ml or above; (b) Hydroxychlorothalonil level of 20.3 ng / ml or above; (c) 5-hydroxyindoleacetic acid level of 28.5 ng / ml or above; (d) lactate level of 686600.0 ng / ml or above; (e) pantothenic acid level of 63.3 ng / ml or above; (f) pipecolic acid level of 303.6 ng / ml or above; (g) γ-CEHC level was 32.0 ng / ml or less; (h) indoleacetic acid level of 141.4 ng / ml or less; (i) p-cresol sulfate level is 1220.0 ng / ml or less; (j) 1,5-anhydroglucitol level of 11910.3 ng / ml or less; (k) 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid level of 7.98 ng / ml or less; (l) 3-indoxylsulfate level of 256.7 ng / ml or less; (m) 4-ethylphenyl sulfate level of 3.0 ng / ml or less; (n) Hydroxyisovalerylcarnitine level of 12.9 ng / ml or less; (o) N-methyl-2-pyridone-5-carboxamide level of 124.82 ng / ml or less; and (p) phenylacetylglutamine level of 166.4 ng / ml or less; (iii) determining that the subject suffers from or is at risk of ASD based on the metabolite level detected in step (ii).

8. The use of claim 7, wherein the at least two metabolites include 3-indoxylsulfuric acid and 3-(3-hydroxyphenyl)propionic acid.

9. The use according to claim 7, wherein the sample is a plasma sample.

10. The use of claim 7, wherein the metabolite level is measured by mass spectrometry.

11. The method of claim 7, wherein the subject is no older than 54 months.

12. The method of claim 7, wherein the subject is no older than 36 months.

13. The use of claim 1 or 7, wherein the one or more binding members are independently peptides.

14. The use of claim 1 or 7, wherein the one or more binding members are independently selected from antibodies or fragments thereof.

Images