Test method for embryonic chromosomal aneuploidy

Abstract

The present invention provides a method for testing chromosomal aneuploidy of embryos cultured in vitro that is less invasive and more reliable than usual. The present invention is a test method for embryonic chromosomal aneuploidy, the method comprising: establishing a marker RNA group comprising non-coding RNAs that are more abundant in embryonic culture broth of aneuploid embryos than embryos with a normal number of chromosomes, creating an extracellular RNA profile comprising the levels of all non-coding RNAs constituting the marker RNA group in the culture broth for embryos cultured in vitro for a teacher sample, associating the extracellular RNA profile and the embryonic chromosomal aneuploidy information for the teacher sample to generate training data, generating a trained model by performing machine learning based on the training data, and assessing the existence of chromosomal aneuploidy in an in vitro cultured embryo to be tested from the extracellular RNA profile of the embryo to be tested using the trained model.

Description

Methods for testing embryos for chromosomal aneuploidy

[0001] The present invention relates to a method for testing chromosomal aneuploidy in in vitro cultured embryos, and a learning device and a trained model used in the method.

[0002] One in ten couples is infertile, and the incidence has been increasing in recent years. There are many different causes of infertility, but regardless of the cause, reproductive medicine such as in vitro fertilization and intracytoplasmic sperm injection is often effective. However, there are also cases where pregnancy does not occur or where couples experience repeated miscarriages despite undergoing infertility treatment. Many of these miscarriages are caused by chromosomal abnormalities. For this reason, if it were possible to determine whether or not there are chromosomal abnormalities at the fertilized egg (embryo) stage, it is expected that many miscarriages could be prevented.

[0003] One of the chromosomal abnormalities that can cause miscarriage is chromosomal excess or deficiency (chromosomal aneuploidy). Therefore, by using preimplantation embryo aneuploidy testing (PGT-A) to check for the presence or absence of chromosomal aneuploidy before implantation, and selecting embryos without any problems for implantation, it may be possible to improve the chances of pregnancy.

[0004] The present inventors have developed a method, PGT-A, in which trophectoderm cells from embryos 5 days after in vitro fertilization are subjected to biopsy, whole genome amplification, and low-coverage whole genome sequencing using next-generation sequencing (Non-Patent Document 1). This method is currently being performed as a clinical test at a registered public health laboratory established by the present inventors. While this method can prevent miscarriage due to chromosomal aneuploidy, even when chromosomal normality is determined using this method, the pregnancy rate is only 60-70%, and the invasiveness of the biopsy to the embryo is suspected to be one of the reasons for this. A less invasive PGT-A method is also known in which the presence or absence of chromosomal aneuploidy in embryos is determined using genomic DNA derived from dead cells contained in the culture medium during in vitro culture as an indicator (Non-Patent Document 2). However, this method has significant inter-institutional variability, with an average accuracy of less than 80%, and further improvement in accuracy is needed.

[0005] On the other hand, in recent years, it has been reported that various non-coding RNA molecules can be used as indicators for evaluating the state of cells and tissues. In particular, Patent Document 1 discloses a method for examining the expression profile of non-coding RNA molecules in cells in an in vitro cell culture system and evaluating the characteristics of the cells, such as their phenotype and quality, based on the expression profile.

[0006] Special Publication No. 2013-535979

[0007] Kurahashi et al. Reproductive Medicine and Biology, 2016, vol.15(1), p.13-19.Rubio, et al., American Journal of Obstetrics & Gynecology, 2020, 751.e1.

[0008] The present invention aims to provide a method for testing for chromosomal aneuploidy in in vitro cultured embryos, which is less invasive and more reliable than conventional methods, as well as a learning device and the like for use in said method.

[0009] The present inventors comprehensively analyzed non-coding RNAs in the culture medium of in vitro-cultured embryos and found that the abundance of nuclear non-coding RNAs (ncRNAs) in the culture medium is specifically increased in aneuploid embryos compared to euploid embryos. In particular, they found that the presence or absence of chromosomal aneuploidy in in vitro-cultured embryos can be determined with high reliability by using a trained model constructed by machine learning using as training data the abundance profiles in the culture medium of 133 long non-coding RNAs (lncRNAs) that showed statistically significant fluctuations in the culture medium of aneuploid embryos. Based on this finding, the present invention was completed.

[0010] The method for testing for chromosomal aneuploidy in an embryo according to the present invention is as follows: [1] A method for testing for chromosomal aneuploidy in an embryo, comprising: a marker setting step of setting a group of marker RNAs consisting of aneuploidy marker RNAs, where the aneuploidy marker RNAs are non-coding RNAs whose abundance in the culture medium of in vitro culture is higher in chromosomally aneuploid embryos than in embryos with a normal number of chromosomes; a profile creation step of creating an extracellular RNA profile for an in vitro-cultured embryo used as a teacher sample, the extracellular RNA profile consisting of the abundance in the culture medium of all aneuploidy marker RNAs constituting the group of marker RNAs; a training data generation step of generating training data by associating the extracellular RNA profile obtained in the profile creation step with chromosomal aneuploidy information on the teacher sample embryo; a learning step of generating a trained model by performing machine learning on the training data; and a testing step of determining the presence or absence of chromosomal aneuploidy in an in vitro-cultured embryo to be tested from the extracellular RNA profile of the embryo to be tested using the trained model. [2] The method for testing for chromosomal aneuploidy in an embryo according to [1], wherein the embryo is a human embryo. [3] The aneuploidy marker RNA constituting the group of marker RNAs isLINC00174、LINC00235、LINC00381、LINC00492、LINC00547、LINC00685、LINC00837、LINC00861、LINC00869、LINC00945、LINC01005、LINC01049、LINC01120、LINC01122、LINC01126、LINC01202、LINC01277、LINC01326、LINC01338、LINC01358、LINC01493、LINC01507、LINC01526、LINC01592、LINC01641、LINC01680、LINC01684、LINC01742、LINC01764、LINC01807、LINC01825、LINC01888、LINC01939、LINC01947、LINC01963、LINC02043、LINC02053、LINC02058、LINC02110、LINC02234、LINC02330、LINC02334、LINC02357、LINC02392、LINC02430、LINC02470、LINC02485、LINC02505、LINC02507、LINC02549、LINC02562、LINC02627、LINC02635、LINC02675、LINC02696、LINC02781、LINC02822、LINC02835、ASMTL-AS1, BCL2L1-AS1, C4A-AS1, C4B-AS1, CBR3-AS1, CHL1-AS1, CNTN4-AS1, CSNK1G2-AS1, ECE1-AS1, FEZF1-AS1, FOXO6-AS1, GORAB- AS1, GYG2-AS1, HMGN3-AS1, IFNG-AS1, INTS6L-AS1, ITPK1-AS1, JMJD1C-AS1, LMNTD2-AS1, MAGI2-AS3, MAPK10-AS1, MRAP-AS1, MYB-AS1 , NECTIN4-AS1, OGFR-AS1, OVCH1-AS1, PCDH9-AS1, PCF11-AS1, PCSK6-AS1, PITPNA-AS1, PPP1R26-AS1, PRKCA-AS1, PRKG1-AS1, PRKG2-A S1, PRR7-AS1, RGPD4-AS1, SIM1-AS1, SLFNL1-AS1, SMIM10L2B-AS1, SRI-AS1, TAPT1-AS1, TBC1D22A-AS1, TNK2-AS1, TP73-AS2, TPM1-AS, PPM1K-DT, RIC3-DT, RPP38-DT, TMED2-DT, VPS13B-DT, ATP2B2-IT1, BACH1-IT2, CACNA1C-IT1, CPS1-IT1, KCNH1-IT1, OPCML-IT1, TTLL7-IT1, The method for testing for chromosomal aneuploidy in an embryo according to [2], wherein the aneuploidy genes are two or more selected from the group consisting of AATBC, BCYRN1, CECR3, CERNA1, DLEU2, DLEU2L, FAM215B, KCNQ1OT1, MEG3, MEG8, PRNCR1, RMST, TEX41, TYMSOS, UCA1, MIR302CHG, MIR4527HG, and MIR4713HG. [4] The method for testing for chromosomal aneuploidy in an embryo according to any one of [1] to [3], wherein the culture medium for in vitro culture is a culture medium for a blastocyst stage embryo. [5] The method for testing for chromosomal aneuploidy in an embryo according to any one of [1] to [4], wherein the machine learning is performed using a naive Bayes algorithm. [6] a marker setting unit that sets a marker RNA group consisting of aneuploidy marker RNAs, the non-coding RNAs being present in a culture solution of in vitro culture in embryos with a normal number of chromosomes being higher than in embryos with a normal number of chromosomes as aneuploidy marker RNAs;a learning unit that reads, as training data, an extracellular RNA profile consisting of the abundances of all aneuploidy marker RNAs that make up the marker RNA group in the culture medium of an in vitro-cultured embryo used as a teacher sample, and chromosomal aneuploidy information of the embryo, and generates a trained model by performing machine learning based on the training data. [7] A testing device that includes: a marker setting unit that sets, as aneuploidy marker RNA, a non-coding RNA that is more abundant in the in vitro culture culture medium of an embryo with a chromosomal aneuploidy than in an embryo with a normal number of chromosomes, and sets a marker RNA group consisting of the aneuploidy marker RNAs; an extracellular RNA profile input unit that inputs an extracellular RNA profile consisting of the abundances of all aneuploidy marker RNAs that make up the marker RNA group in the culture medium of an in vitro-cultured embryo to be tested; and a testing unit that determines chromosomal aneuploidy of the embryo to be tested, based on the extracellular RNA profile inputted to the extracellular RNA profile input unit, using a trained model generated by performing machine learning based on the extracellular RNA profile of the in vitro-cultured embryo and the chromosomal aneuploidy information of the embryo. [8] A trained model used to determine chromosomal aneuploidy in in vitro cultured embryos, the trained model being generated by machine learning based on an extracellular RNA profile of an in vitro cultured embryo and chromosomal aneuploidy information of the embryo, the extracellular RNA profile consisting of the abundance in the embryo culture medium of all aneuploidy marker RNAs constituting a predetermined group of marker RNAs, the aneuploidy marker RNAs being non-coding RNAs whose abundance in the in vitro culture medium is greater in embryos with chromosomal aneuploidy than in embryos with a normal number of chromosomes, and the trained model for causing a computer to function to determine chromosomal aneuploidy of the embryo based on the extracellular RNA profile of the embryo input as a test subject.

[0011] The present invention enables the non-invasive and highly reliable determination of the presence or absence of chromosomal aneuploidy in in vitro cultured embryos, and is therefore useful for quality testing of in vitro cultured embryos before implantation, and is particularly suitable for testing chromosomal aneuploidy in preimplantation embryos in infertility treatment.

[0012] Fig. 1 is a schematic block diagram showing the configuration of a learning device according to the present embodiment; Fig. 2 is a flow diagram showing an example of processing by the learning device according to the present embodiment; Fig. 3 is a schematic block diagram showing the configuration of an inspection device according to the present embodiment; Fig. 4 is a flow diagram showing an example of processing by execution acknowledgement according to the present embodiment.

[0013] A method for testing for chromosomal aneuploidy in an embryo according to an embodiment of the present disclosure is a method for testing whether the number of chromosomes in an in vitro-cultured embryo is normal (diploid in humans) or aneuploid, using non-coding RNA in the culture medium as a marker. Because the culture medium of an in vitro-cultured embryo (hereinafter sometimes simply referred to as "embryo culture medium") is used as the test sample, the method is non-invasive and can avoid damage to the embryo compared to conventional methods that use cells constituting the embryo (inner cell mass or trophoblast cells) as the test sample. Therefore, by performing preimplantation embryo chromosomal aneuploidy testing using the test method according to the embodiment, increased embryo implantation and pregnancy rates can be expected. In other words, the method for testing for chromosomal aneuploidy in an embryo according to an embodiment of the present disclosure can provide extremely useful information for preimplantation diagnosis of in vitro-cultured embryos.

[0014] In the method for testing for chromosomal aneuploidy in an embryo according to an embodiment of the present disclosure, the embryo to be tested (the embryo for which the presence or absence of chromosomal aneuploidy is to be determined) is not particularly limited as long as it is an embryo that has been cultured in vitro, and it is possible to use an embryo that has been cultured from a fertilized egg obtained by artificial insemination under a microscope using a conventional method, etc. The embryo to be subjected to this testing method is not particularly limited as long as it is a pre-implantation embryo, but is preferably an embryo at the blastocyst stage, and more preferably an embryo 5 to 6 days after artificial insemination.

[0015] In the method for testing embryos for chromosomal aneuploidy according to an embodiment of the present disclosure, the culture medium and culture method used to culture the embryos to be tested are not particularly limited as long as they are culture media and culture methods commonly used in embryo culture. This may be a one-step culture method in which the culture medium is not changed from artificial insemination to implantation, or a continuous culture method in which the embryos are cultured in an early embryo medium from artificial insemination to the 4- to 8-cell stage, and then cultured in a blastocyst medium thereafter. These culture media preferably contain glucose, pyruvic acid, lactic acid, amino acids, and inorganic salts necessary for embryo development, and are adjusted to be isotonic and neutral in osmotic pressure. If necessary, they may also preferably contain albumin, vitamins, various growth factors, cytokines, antibiotics, and the like. Commercially available embryo culture media, either as is or with appropriate modifications, may be used to culture the embryos to be tested, depending on the animal species. For example, in the case of human embryos, examples of single-stage culture media include "Continuous Single Culture Complete" (manufactured by Fujifilm Wako Pure Chemical Industries, Ltd.) and "SSM" (manufactured by Irvine Scientific Co., Ltd.), examples of early embryo culture media used in continuous culture methods include "ECM" (manufactured by Irvine Scientific Co., Ltd.), "Cleavage" (manufactured by Cook Medical Co., Ltd.), and "ONESTEP Medium" (manufactured by NAKA Medical Co., Ltd.), and examples of blastocyst culture media include "MultiBlast" (manufactured by Irvine Scientific Co., Ltd.) and "Blastocyst" (manufactured by Cook Medical Co., Ltd.).

[0016] In the testing method according to an embodiment of the present disclosure, non-coding RNA is used as a marker instead of genomic DNA in embryo culture medium. Genomic DNA in embryo culture medium is primarily derived from dead cells, and therefore does not necessarily correlate with chromosomal aneuploidy. This results in low accuracy in determining aneuploidy, and false positives are likely to occur, particularly in the case of mosaic embryos. In contrast, the testing method according to an embodiment of the present disclosure uses non-coding RNA, which is correlated with chromosomal aneuploidy, as a marker, enabling highly reliable determination of the presence or absence of chromosomal aneuploidy.

[0017] In a testing method according to an embodiment of the present disclosure, a non-coding RNA is used as an aneuploidy marker, the amount of which in an embryo culture medium correlates with the presence or absence of chromosomal aneuploidy. Specifically, a non-coding RNA that is more abundant in aneuploid embryos than in embryos with a normal number of chromosomes is used as the aneuploidy marker RNA. Although the origin of this non-coding RNA is unclear, because aneuploid embryos are more susceptible to apoptosis than euploid embryos, it is presumed that this non-coding RNA is specifically leaked into the culture medium by aneuploid embryos that have undergone apoptosis during culture.

[0018] The aneuploidy marker RNA used in the embodiments of the present disclosure can be obtained by comprehensively and quantitatively analyzing RNA in the embryo culture medium of a eudiploid embryo and that of an aneuploid embryo, and selecting non-coding RNAs whose abundance in the embryo culture medium of an aneuploid embryo is statistically significantly higher than that of a eudiploid embryo. Comprehensive and quantitative analysis of RNA in embryo culture medium can be performed, for example, by RNA-Seq (RNA analysis using a next-generation sequencing (NGS) system).

[0019] In the testing method according to an embodiment of the present disclosure, a marker RNA group consisting of two or more types of aneuploidy marker RNA is set as a marker setting step. By using a plurality of types of aneuploidy marker RNA as the marker RNA group, rather than a single type, the accuracy of determining the presence or absence of chromosomal aneuploidy can be improved.

[0020] When the embryo to be tested is a human embryo, examples of aneuploidy marker RNAs include those listed in Tables 1 to 7 below. All of these are non-coding RNAs registered in the HGNC (HUGO Gene Nomenclature Committee) database (address: https: / / www.genenames.org / data / genegroup / #! / group / 788). In the tables, the "HGNC ID" column indicates the ID number registered in the HGNC database for each RNA.

[0021]

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028] As the marker RNA group used for human embryos, any two or more of the 133 types of non-coding RNAs can be used in combination. In order to determine the presence or absence of chromosomal aneuploidy with higher accuracy, it is preferable to use all of the 133 types as the marker RNA group in the testing method according to the embodiment of the present disclosure.

[0029] In a testing method according to an embodiment of the present disclosure, the abundance of each of all aneuploidy marker RNAs constituting a marker RNA group in the embryo culture medium is measured, and the resulting dataset of measurements, i.e., the dataset of the abundance of all aneuploidy marker RNAs constituting the marker RNA group in the embryo culture medium, is used as the extracellular RNA profile of the embryo. In a testing method according to an embodiment of the present disclosure, the abundance of each individual aneuploidy marker RNA in the embryo culture medium is not used independently as an aneuploidy marker, but rather the extracellular RNA profile, which is information on the abundance of multiple aneuploidy marker RNAs in the embryo culture medium, is used as an aneuploidy marker. This allows for the presence or absence of chromosomal aneuploidy to be determined with greater accuracy and sensitivity.

[0030] In a testing method according to an embodiment of the present disclosure, the presence or absence of chromosomal aneuploidy based on an embryo's extracellular RNA profile is determined using a trained model constructed using a machine learning algorithm for determining embryonic aneuploidy. Although functional evaluation of non-coding RNA is difficult, constructing a determination algorithm using machine learning and using a trained model equipped with this algorithm makes it possible to determine with high specificity the presence or absence of chromosomal aneuploidy based on the extracellular RNA profile, which is information on a population of non-coding RNA.

[0031] In a testing method according to an embodiment of the present disclosure, in order to construct a trained model for determining aneuploidy, an in vitro-cultured embryo whose presence or absence of chromosomal aneuploidy has been confirmed by other means is prepared as a teacher sample. Then, in a profile creation step, an extracellular RNA profile is created for the in vitro-cultured embryo for the teacher sample, consisting of the abundance in the culture medium of all aneuploidy marker RNAs that make up the marker RNA group set in the marker setting step. The teacher sample embryo is of the same animal species as the embryo to be tested.

[0032] The abundance of each aneuploidy marker RNA constituting the marker RNA group in the embryo culture medium can be measured by various methods capable of quantitatively detecting RNA. RNA extracted from the embryo culture medium may be directly subjected to quantitative measurement, but since this allows for more stable measurement, it is preferable to perform reverse transcription using the RNA extracted from the embryo culture medium as a template to produce cDNA, and then subject the resulting cDNA to quantitative measurement. RNA extraction from the embryo culture medium and cDNA synthesis by reverse transcription can be performed by conventional methods.

[0033] As a quantitative measurement method, various methods such as qPCR (quantitative polymerase chain reaction), microarray method, and RNA-Seq can be used. When the number of aneuploidy marker RNAs constituting the marker RNA group is large, measurement by microarray method or RNA-Seq is preferable because the target extracellular RNA profile can be created relatively easily, and measurement by RNA-Seq is more preferable because it has relatively excellent quantitative properties. On the other hand, when the number of aneuploidy marker RNAs is small, measurement by qPCR is also preferable because it has excellent quantitative properties. These methods can be performed by conventional methods.

[0034] For example, when generating an extracellular RNA profile using RNA-Seq, a reverse transcription reaction is first performed using RNA extracted from embryo culture medium as a template, and the resulting cDNA is fragmented to an appropriate read length. Next, adapter sequences containing molecular barcodes are ligated to both ends of the fragmented cDNA, followed by PCR to obtain the resulting amplification product. The resulting amplification product is mapped based on genome sequence data, and the number of reads of the target aneuploidy marker RNA is quantified and normalized to calculate the abundance value in the embryo culture medium. Fragmentation, adapter sequence ligation, and amplification product analysis can be performed by referring to the recommended protocol for the NGS system used. Examples of NGS systems include the "Genome Analyzer" (manufactured by Illumina), "NextSeq 550Dx" (manufactured by Illumina), and the "Genome Sequencer FLX system" (manufactured by Roche).

[0035] Next, in the training data generation step, the extracellular RNA profile of the teacher sample embryo is associated with the chromosomal aneuploidy information of the embryo to generate training data. The chromosomal aneuploidy information of the teacher sample embryo can be obtained by collecting a portion of the trophectoderm cells from the embryo and directly examining the chromosomes. Specifically, the created extracellular RNA profile and chromosomal aneuploidy information for the teacher sample embryo are input, and the extracellular RNA profile and chromosomal aneuploidy information are generated as training data.

[0036] The extracellular RNA profile is affected by the culture conditions. In other words, the training data used to generate a trained model for determining the presence or absence of chromosomal aneuploidy in in vitro-cultured embryos (hereinafter referred to as the "trained model for aneuploidy determination") is affected by the culture conditions of the embryos. For this reason, for the extracellular RNA profile of an embryo used as a training sample, it is preferable that the embryo culture conditions, particularly the culture medium used, the manner of medium replacement, and the culture time from artificial insemination in the culture medium for creating the extracellular RNA profile, are the same as those of the embryos to be tested.

[0037] Based on the generated training data, machine learning, particularly deep learning, is performed to generate a trained model for aneuploidy discrimination (training step). The amount of training required to generate a trained model with sufficient accuracy for aneuploidy discrimination is not particularly limited, but it is preferable to perform machine learning on training data with a number of training sample embryos equal to or greater than 50, preferably equal to or greater than 80.

[0038] The machine learning is not particularly limited, and can be performed using various machine learning algorithms. Examples of machine learning algorithms that can be used in the testing method according to the embodiment of the present disclosure include Naive Bayes, Extra Trees Classifier, Light Gradient Boosting Machine (LightGBM), Random Forest Classifier, Logistic Regression, Adaptive Boosting Classifier, Linear Discriminant Analysis, Ridge Regression, K-Neighbors Classifier, and Gradient Boosting Classifier. Among these, it is preferable to use the Naive Bayes algorithm for machine learning in the testing method according to the embodiment of the present disclosure, since it is possible to construct a trained model for aneuploidy detection that can achieve extremely high detection accuracy.

[0039] Using the trained model for determining aneuploidy constructed in this way, the presence or absence of chromosomal aneuploidy in the in vitro cultured embryo to be tested is determined from the extracellular RNA profile of the embryo (testing step). This trained model for determining aneuploidy enables the presence or absence of chromosomal aneuploidy in the embryo to be determined with high accuracy.

[0040] A method for testing for chromosomal aneuploidy in an embryo according to an embodiment of the present disclosure can be implemented, for example, using a testing system S including a learning device 1 and a testing device 2. The learning device 1 performs machine learning processing based on extracellular RNA profile information of a plurality of ex vivo cultured embryos and chromosomal aneuploidy information of each ex vivo cultured embryo. The learning device 1 generates a trained model as a result of the machine learning processing. The testing device 2 uses the trained model generated by the learning device 1 to determine the presence or absence of chromosomal aneuploidy in the embryo based on the input extracellular RNA profile information of the ex vivo cultured embryo to be tested.

[0041] 1 is a schematic block diagram showing the configuration of a learning device 1 according to this embodiment. The learning device 1 includes an input / output unit I1, a memory unit M1, and a processing unit P1.

[0042] The input / output unit I1 includes a data acquisition unit 111 and an input / output unit 112. The data acquisition unit 111 acquires data via communication or an external storage device. The input / output unit 112 accepts information input by a user operation such as keyboard input or marking using a pointing device.

[0043] The storage unit M1 includes a data storage unit 121, a learning data storage unit 122, and a learning result storage unit 123.

[0044] The data storage unit 121 stores the data acquired by the data acquisition unit 111. The data includes extracellular RNA profile information for each of a plurality of ex vivo cultured embryos. The data storage unit 121 may further include, as additional information, information on each ex vivo cultured embryo and information on aneuploidy marker RNA used in the extracellular RNA profile. Examples of the information on the ex vivo cultured embryo include information on the culture medium used in the in vitro culture, the culture conditions, and the time when the embryo culture medium was collected.

[0045] The training data storage unit 122 stores training data used for machine learning. The training data is data in which, for each in vitro cultured embryo, chromosomal aneuploidy information of the embryo is added to the extracellular RNA profile information and its additional information. The training result storage unit 123 stores a trained model generated as a result of machine learning processing based on the training data.

[0046] The processing unit P1 is configured to include a training data generation unit 131 and a training unit 132. The training data generation unit 131 links extracellular RNA profile information of each in vitro cultured embryo with chromosomal aneuploidy information of the in vitro cultured embryo to generate training data. The training unit 132 performs machine learning processing based on the training data generated by the training data generation unit 131. The training unit 132 stores the generated trained model in the training result storage unit 123. The machine learning processing can be performed by a conventional method.

[0047] The processing of the learning device 1 will be described below, focusing on the processing performed by each unit in Fig. 1. Fig. 2 is a flow diagram showing an example of processing performed by the learning device 1 according to this embodiment.

[0048] (Step S101) The data acquisition unit 111 acquires extracellular RNA profile information and additional information for each of the multiple in vitro cultured embryos. Then, the processing of step S102 is performed. (Step S102) The training data generation unit 131 displays the extracellular RNA profile information and additional information acquired in step S101, and generates training data for each in vitro cultured embryo by linking the chromosomal aneuploidy information of the embryo to the extracellular RNA profile information and the chromosomal aneuploidy information. The training data generation unit 131 stores the generated training data in the training data storage unit 122. Then, the processing of step S103 is performed. (Step S103) The learning unit 132 performs machine learning processing based on the training data generated in step S102. As a result of the machine learning processing, a trained model is generated.

[0049] 3 is a schematic block diagram showing the configuration of the inspection device 2 according to this embodiment. The inspection device 2 includes an input / output unit I2, a memory unit M2, and a processing unit P2.

[0050] The input / output unit I2 is configured to include an extracellular RNA profile information acquisition unit 211, an input unit 212, and a display unit 213. The extracellular RNA profile information acquisition unit 211 acquires extracellular RNA profile information of the ex vivo cultured embryo as the test subject, for example, from an RNA-Seq device, communication, or an external storage device. The input unit 212 accepts information entered by the user via a keyboard or pointing device. The display unit 213 is a display that displays test results, etc.

[0051] The memory unit M2 is configured to include a terminal data memory unit 221, a setting memory unit 222, and a learning model memory unit 223. The terminal data memory unit 221 stores the extracellular RNA profile information of the test subject embryo acquired by the extracellular RNA profile information acquisition unit 211, information received by the input unit 212, and the test results of the testing unit 232. The setting memory unit 222 stores setting information for the test. The setting information includes, for example, information on the aneuploidy marker RNA used for the determination. The learning model memory unit 223 stores the trained model generated by the learning device 1.

[0052] The processing unit P2 is configured to include a setting unit 231, an inspection unit 232, and a display control unit 233. The setting unit 231 allows setting information such as information on the aneuploidy marker RNA to be input by a user operation, and stores the setting information in the setting storage unit 222.

[0053] The testing unit 232 determines the presence or absence of chromosomal aneuploidy for the input extracellular RNA profile information (testing process) using the trained model stored in the training model storage unit 223. The display control unit 233 causes the display unit 213 to display test results, etc., including the determination result of the presence or absence of chromosomal aneuploidy.

[0054] <Processing of Inspection Apparatus> The processing of the inspection apparatus 2 will be described below, focusing on the processing performed by each unit in Fig. 3. Fig. 4 is a flow chart showing an example of processing performed by the inspection apparatus 2 according to this embodiment.

[0055] (Step S201) The extracellular RNA profile information acquisition unit 211 acquires extracellular RNA profile information of the in vitro cultured embryo to be tested. Then, the processing of step S201 is performed. (Step S202) The testing unit 232 determines the presence or absence of chromosomal aneuploidy for the extracellular RNA profile information acquired in step S201 using a trained model generated by machine learning processing (step S103 in Figure 2). Then, the processing of step S203 is performed. (Step S203) The display control unit 233 causes the display unit 213 to display the determination result (test result) of the presence or absence of chromosomal aneuploidy in step S202.

[0056] Note that a portion of the learning device 1 or the inspection device 2 in the above-described embodiments may be implemented by a computer. In this case, a program for implementing this control function may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed. Note that the term "computer system" as used herein refers to a computer system built into the learning device 1 or the inspection device 2, including hardware such as an OS and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include media that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or media that store programs for a certain period of time, such as volatile memory within a computer system that serves as a server or client in such cases. The program may be designed to implement a portion of the above-described functions, or may be capable of implementing the above-described functions in combination with a program already stored in the computer system. Furthermore, part or all of the learning device 1 and inspection device 2 in the above-described embodiments may be realized as an integrated circuit such as an LSI (Large Scale Integration). Each functional block of the learning device 1 and inspection device 2 may be individually implemented as a processor, or part or all of them may be integrated into a processor. Furthermore, the integrated circuit implementation method is not limited to LSI, and may be implemented using a dedicated circuit or a general-purpose processor. Furthermore, if an integrated circuit implementation technology that can replace LSI emerges due to advances in semiconductor technology, an integrated circuit based on that technology may be used.

[0057] One embodiment of the present invention has been described in detail above with reference to the drawings, but the specific configuration is not limited to that described above, and various design changes and the like are possible within the scope that does not deviate from the gist of the present invention.

[0058] The present invention will now be described in more detail with reference to examples, but the present invention is not limited to the following examples.

[0059] [Example 1] In vitro-cultured blastocyst-stage embryos were examined for non-coding RNAs whose abundance in the embryo culture medium correlates with the presence or absence of chromosomal aneuploidy, and whether chromosomal aneuploidy in in vitro-cultured embryos can be determined using the embryo culture medium as a test sample was investigated.

[0060] (1) Identification of non-coding RNAs correlated with chromosomal aneuploidy 58 samples of blastocyst-stage embryos cultured in 30 μL of culture medium on days 5-6 after artificial insemination were subjected to chromosomal aneuploidy analysis and comprehensive analysis of non-coding RNAs in the embryo cells themselves and in the embryo culture medium. For chromosomal aneuploidy analysis, karyotype analysis was performed on a portion of the trophectoderm cells. For comprehensive analysis of non-coding RNAs in the embryo cells themselves, RNA-Seq was performed on RNA extracted from all cells constituting the embryo, excluding the trophectoderm cells subjected to chromosomal aneuploidy analysis. For comprehensive analysis of non-coding RNAs in the embryo culture medium, RNA-Seq was performed on RNA extracted from the entire volume of the embryo culture medium.

[0061] Karyotype analysis using an NGS system. Of the 53 whole-genome samples (ASC01-34, ESC01-19), 5-10 cells were biopsied from the trophectoderm (TE) region and subjected to whole-genome amplification using the PicoPLEX method. The DNA obtained by whole-genome amplification was prepared using the Nextera XT DNA Library Preparation Kit (Illumina) and then sequenced using the MiSeq Sequence System with 37-bp single-end reads, resulting in a data volume of at least 1 million reads per sample. All reagents used in the Nextera XT and MiSeq sequencing systems for whole-genome amplification and sample preparation were those included in the VeriSeq PGS Kit (Illumina). Sequencing data was generated from FASTQ files to BAM files using "MiSeq Reporter" software, and the BAM files were imported into "BlueFuse" software for analysis to determine chromosome copy number for karyotype analysis. "MiSeq Reporter" and "BlueFuse" software were obtained from Illumina.

[0062] As a result, of the 58 samples, 19 were euploid embryos, 5 were mosaic embryos, and 34 were aneuploid embryos. One of the aneuploid embryos was a triploid embryo.

[0063] <Comprehensive analysis of non-coding RNAs in embryo cells themselves or embryo culture medium> Comprehensive analysis of non-coding RNAs in embryo cells themselves and embryo culture medium was performed for 19 euploid embryos and 34 aneuploid embryos. However, triploid embryos were excluded from the comprehensive analysis of non-coding RNAs in the embryo culture medium.

[0064] For 53 whole embryo cell samples (ASC01-34, ESC01-19), total RNA was extracted and reverse-transcribed using the GenNext® RamDA-seq® Single Cell Kit (Code No. RMD-101T or RMD-101, TOYOBO). cDNA obtained by reverse transcription was prepared using the Nextera XT DNA Library Preparation Kit (Code No. FC-131-1024 or FC-131-1096, Illumina). For 52 embryo culture fluid samples (ASM01-33, ESM01-19), RNA was extracted from the embryo culture fluid using the miRNeasy Serum / Plasma Advanced Kit (Cat. No. / ID: 217204). The extracted RNA was used as a template for reverse transcription using the GenNext® RamDA-seq® Single Cell Kit, and the resulting cDNA was prepared using the Nextera XT DNA Library Preparation Kit. Each sample was assigned an index, and after sample preparation, the samples were pooled to equalize their concentrations and sequenced using the NoVaseq sequencing system (Illumina) with 2 x 75 bp paired-end reads. The sequence data was analyzed using the ramdaq pipeline (https: / / github.com / rikenbit / ramdaq), a dedicated analysis pipeline for RamDA-seq®. RNA expression levels obtained by sequencing were adjusted between samples using the TCC software package (https: / / bioconductor.org / packages / release / bioc / html / TCC.html), allowing for comparison and analysis of expression levels.

[0065] Among the non-coding RNAs detected in the 52 embryo culture medium samples, 133 were identified as aneuploidy marker RNAs, which were statistically significantly more abundant in chromosomally aneuploid embryos than in euploid embryos. These non-coding RNAs are shown in Table 8.

[0066]

[0067] (2) Extracellular RNA Profile of Each Aneuploidy Marker RNA For each of the 105 samples subjected to RNA-Seq, the dataset of the amounts of the 133 non-coding RNAs listed in Table 8 was used to create the extracellular RNA profile of each sample. These 133 non-coding RNAs were compiled by statistically extracting differentially expressed genes (DEGs) in non-coding RNAs between embryos with a normal chromosome number and embryos with aneuploidy by simultaneously analyzing all RNA sequencing data from whole embryo cells and embryo culture medium samples using the TCC software package.

[0068] (3) Generation of Trained Models for Aneuploidy Determination by Machine Learning From the 105 analyzed samples, 15 embryo culture medium samples were excluded, leaving a total of 90 samples. Information on the presence or absence of aneuploidy in each embryo was linked to the extracellular RNA profiles of these 90 samples, and this information was prepared as training data. A machine learning determination algorithm was constructed using this training data. The machine learning algorithms used were the naive Bayes algorithm, the extra-tree algorithm, the LightGBM algorithm, the random forest algorithm, the logistic regression algorithm, the adaptive boosting algorithm, the linear discriminant analysis algorithm, the ridge regression algorithm, the K-nearest neighbor algorithm, and the gradient boosting algorithm. The trained models constructed by each learning algorithm were each subjected to 10-fold cross-validation and adjusted to be optimal.

[0069] The results of 10-fold cross-validation of trained models constructed using each learning algorithm are shown in Table 9. In the table, "AUC" means the area under the receiver operating characteristic curve (ROC curve) plotted with the true positive rate (TPR) on the vertical axis and the false positive rate (FPR) on the horizontal axis, and "MCC" means the Matthews correlation coefficient. The trained model constructed using the naive Bayes algorithm had the highest accuracy.

[0070]

[0071] Using the trained model for aneuploidy determination constructed using the naive Bayes method, which showed the highest accuracy, 15 embryo culture fluid samples that were not used to generate the trained model were tested, and chromosomal aneuploidy was determined using these extracellular RNA profiles as test data. The results are shown in Table 10. As a result, as with cross-validation, the chromosomal aneuploidy determination for the test data showed extremely high accuracy, with an accuracy of 93.3%, a sensitivity of 100%, and a positive predictive value of 90.9%.

[0072]

[0073] 1...Learning device, I1...Input / output unit, M1...Memory unit, P1...Processing unit, 111...Data acquisition unit, 112...Input / output unit, 121...Data memory unit, 122...Learning data memory unit, 123...Learning result memory unit, 131...Learning data generation unit, 132...Learning unit, 2...Inspection device, I2...Input / output unit, M2...Memory unit, P2...Processing unit P2, 211...Extracellular RNA profile information acquisition unit, 212...Input unit, 213...Display unit, 221...Terminal data memory unit, 222...Setting memory unit, 223...Learning model memory unit, 231...Setting unit, 232...Inspection unit, 233...Display control unit.

Claims

1. A method for testing for chromosomal aneuploidy in embryos, comprising: a marker setting step of setting a group of marker RNAs consisting of aneuploidy marker RNAs, where the non-coding RNAs present in greater amounts in the culture medium of in vitro culture in chromosomally aneuploid embryos than in embryos with a normal number of chromosomes are designated as aneuploidy marker RNAs; a profile creation step of creating an extracellular RNA profile consisting of the amounts present in the culture medium of all aneuploidy marker RNAs constituting the group of marker RNAs for an in vitro-cultured embryo used as a teacher sample; a training data generation step of generating training data by associating the extracellular RNA profile obtained in the profile creation step with chromosomal aneuploidy information for the teacher sample embryo; a learning step of generating a trained model by performing machine learning based on the training data; and a testing step of determining the presence or absence of chromosomal aneuploidy in an in vitro-cultured embryo to be tested from the extracellular RNA profile of the embryo to be tested using the trained model.

2. A method for testing chromosomal aneuploidy in an embryo according to claim 1, wherein the embryo is a human embryo.

3. The aneuploidy marker RNAs constituting the marker RNA group are: LINC00174, LINC00235, LINC00381, LINC00492, LINC00547, LINC00685, LINC00837, LINC00861, LINC00869, LINC00945, LINC01005, LINC01049, LINC01120, LINC01122, LINC0 1126, LINC01202, LINC01277, LINC01326, LINC01338, LINC01358, LINC01493, LINC015 07, LINC01526, LINC01592, LINC01641, LINC01680, LINC01684, LINC01742, LINC01764, LINC01807, LINC01825, LINC01888, LINC01939, LINC01947, LINC01963, LINC02043, LI NC02053, LINC02058, LINC02110, LINC02234, LINC02330, LINC02334, LINC02357, LINC0 2392, LINC02430, LINC02470, LINC02485, LINC02505, LINC02507, LINC02549, LINC025 62, LINC02627, LINC02635, LINC02675, LINC02696, LINC02781, LINC02822, LINC02835,ASMTL-AS1, BCL2L1-AS1, C4A-AS1, C4B-AS1, CBR3-AS1, CHL1-AS1, CNTN4-AS1, CSNK1G2-AS1, ECE1-AS1, FEZF1-AS1, FOXO6-AS1, GORAB- AS1, GYG2-AS1, HMGN3-AS1, IFNG-AS1, INTS6L-AS1, ITPK1-AS1, JMJD1C-AS1, LMNTD2-AS1, MAGI2-AS3, MAPK10-AS1, MRAP-AS1, MYB-AS1 , NECTIN4-AS1, OGFR-AS1, OVCH1-AS1, PCDH9-AS1, PCF11-AS1, PCSK6-AS1, PITPNA-AS1, PPP1R26-AS1, PRKCA-AS1, PRKG1-AS1, PRKG2-A S1, PRR7-AS1, RGPD4-AS1, SIM1-AS1, SLFNL1-AS1, SMIM10L2B-AS1, SRI-AS1, TAPT1-AS1, TBC1D22A-AS1, TNK2-AS1, TP73-AS2, TPM1-AS, PPM1K-DT, RIC3-DT, RPP38-DT, TMED2-DT, VPS13B-DT, ATP2B2-IT1, BACH1-IT2, CACNA1C-IT1, CPS1-IT1, KCNH1-IT1, OPCML-IT1, TTLL7-IT1, The method for testing chromosomal aneuploidy in an embryo according to claim 2, wherein the aneuploidy is two or more selected from the group consisting of AATBC, BCYRN1, CECR3, CERNA1, DLEU2, DLEU2L, FAM215B, KCNQ1OT1, MEG3, MEG8, PRNCR1, RMST, TEX41, TYMSOS, UCA1, MIR302CHG, MIR4527HG, and MIR4713HG.

4. A method for testing chromosomal aneuploidy in an embryo according to claim 1, wherein the culture medium for in vitro culture is a culture medium for a blastocyst stage embryo.

5. A method for testing chromosomal aneuploidy in an embryo as described in claim 1, wherein the machine learning is performed using a naive Bayes algorithm.

6. A learning device comprising: a marker setting unit that sets a marker RNA group consisting of aneuploidy marker RNAs, using non-coding RNAs that are present in greater amounts in in vitro culture culture medium in chromosomally aneuploid embryos than in embryos with a normal number of chromosomes as aneuploidy marker RNAs; and a learning unit that reads, as learning data, an extracellular RNA profile consisting of the abundance of all aneuploidy marker RNAs that make up the marker RNA group in the culture medium of an in vitro cultured embryo used as a teacher sample, and chromosomal aneuploidy information of the embryo, and generates a trained model by performing machine learning based on the learning data.

7. A testing device comprising: a marker setting unit that sets a group of marker RNAs consisting of aneuploidy marker RNAs, where the non-coding RNAs present in greater amounts in the culture medium of in vitro culture in embryos with chromosomal aneuploidy than in embryos with a normal number of chromosomes are designated as aneuploidy marker RNAs; an extracellular RNA profile input unit that inputs an extracellular RNA profile consisting of the abundances of all aneuploidy marker RNAs that make up the group of marker RNAs in the culture medium of an in vitro-cultured embryo to be tested; and a testing unit that determines the chromosomal aneuploidy of the embryo to be tested based on the extracellular RNA profile input to the extracellular RNA profile input unit, using a trained model generated by performing machine learning based on the extracellular RNA profile of the in vitro-cultured embryo and chromosomal aneuploidy information of the embryo.

8. A trained model used to determine chromosomal aneuploidy in in vitro cultured embryos, the trained model being generated by machine learning based on an extracellular RNA profile of an in vitro cultured embryo and chromosomal aneuploidy information of the embryo, the extracellular RNA profile consisting of the abundance in the embryo culture medium of all aneuploidy marker RNAs that make up a predetermined group of marker RNAs, the aneuploidy marker RNAs being non-coding RNAs whose abundance in the in vitro culture medium is greater in embryos with chromosomal aneuploidy than in embryos with a normal number of chromosomes, and causing a computer to function to determine the chromosomal aneuploidy of the embryo based on the extracellular RNA profile of the embryo input as the test subject.

Images