Application of marker in preparation of related products for predicting intrahepatic cholestasis in gestation period

Abstract

The invention provides an application of a marker in preparation of a related product for predicting intrahepatic cholestasis in a gestation period. Wherein the marker is a gene marker; and the gene marker is one or more of SLC7A2, PXK, RNF141 or FCHO2. By adopting the marker for detection, high-specificity and high-sensitivity risk prediction of intrahepatic cholestasis in the gestation period can be realized.

Description

Technical Field

[0001] The present invention relates to the field of pregnancy complications, and more particularly, to the use of gene markers in the preparation of related products for predicting intrahepatic cholestasis of pregnancy. Background Art

[0002] Intrahepatic cholestasis of pregnancy (ICP) is a unique complication in the second and third trimesters of pregnancy (clinically, pregnancy is divided into three trimesters. The first trimester is before 13 weeks of gestation; the second trimester is from 14 to 27 weeks of gestation; the third trimester is 28 weeks of gestation and later). Clinically, it is characterized by skin pruritus and elevated bile acids, mainly endangering the fetus, increasing the preterm birth rate and stillbirth rate. Early prediction and early intervention can improve pregnancy outcomes. Currently, there is no clinical product for predicting intrahepatic cholestasis of pregnancy.

[0003] Currently, it is mainly to combine serum alpha-fetoprotein and age as markers to predict the risk of intrahepatic cholestasis of pregnancy, but the accuracy is very low, and the prediction sensitivity and specificity are less than 0.65.

[0004] So far, the main research in this field has focused on assisting the diagnosis of intrahepatic cholestasis of pregnancy, lacking risk prediction. Risk prediction can support the research on the mechanism of intrahepatic cholestasis of pregnancy and targeted intervention treatment methods. Therefore, there is an urgent need to develop a program that can accurately predict the risk of intrahepatic cholestasis of pregnancy. Summary of the Invention

[0005] The main object of the present invention is to provide the use of a marker in the preparation of related products for predicting intrahepatic cholestasis of pregnancy, so as to improve the accuracy of predicting the risk of intrahepatic cholestasis of pregnancy.

[0006] To achieve the above object, according to one aspect of the present invention, there is provided the use of a marker in the preparation of related products for predicting intrahepatic cholestasis of pregnancy, wherein the marker includes a clinical index marker and / or a gene marker. The clinical index marker includes a conception method index and any one or more of the following clinical detection indexes: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase. The conception method index is whether it is assisted reproduction; the gene marker includes any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0007] Further, the application includes using a detection reagent to detect the expression level of clinical detection indicators and / or gene markers. Preferably, when the marker is a gene marker, the detection reagent is a reagent for detecting the mRNA expression level or protein expression level of the gene marker; more preferably, the detection reagent for detecting the mRNA expression level of the gene marker is a probe and / or primer, and further preferably, it is a reagent related to preparing a high-throughput sequencing library from the mRNA of the gene marker.

[0008] Further, in combination with the conception mode index, the clinical detection indicators and / or gene markers in the biological sample derived from the pregnant woman are detected, wherein the biological sample is selected from one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; preferably, the biological sample is collected during the 13th - 25th gestational week of the pregnant woman; preferably, the related product includes a kit or an electronic prediction device.

[0009] To achieve the above object, according to the second aspect of the present invention, a kit for predicting intrahepatic cholestasis of pregnancy is provided. The kit includes a detection reagent for the marker of intrahepatic cholestasis of pregnancy, and the marker is a clinical index marker and / or a gene marker. Among them, the clinical index marker includes the conception mode index and any one or more of the following clinical detection indicators: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase. The conception mode index is whether it is assisted reproduction; the gene marker includes any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0010] Further, the detection reagent is a reagent for detecting the expression level of clinical detection indicators and / or gene markers; preferably, when the marker is a gene marker, the detection reagent is a reagent for detecting the mRNA or protein expression level of the gene marker; preferably, the detection reagent for detecting the mRNA expression level of the gene marker is a probe and / or primer, and further preferably, it is a reagent related to preparing a high-throughput sequencing library from the mRNA of the gene marker.

[0011] According to the third aspect of the present invention, an electronic device for predicting intrahepatic cholestasis of pregnancy is provided. The electronic device is built-in with a risk prediction model for intrahepatic cholestasis of pregnancy in pregnant women. The risk prediction model is generated by training a computer using the expression levels of markers in biological samples from pregnant women with positive intrahepatic cholestasis of pregnancy. The markers include clinical index markers and / or gene markers. Among them, the clinical index markers include the conception mode index and any one or more of the following clinical detection indexes: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase. The conception mode index is whether assisted reproduction is used; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0012] To achieve the above object, according to the fourth aspect of the present invention, a method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy is provided. The construction method includes: dividing the pregnant women group with positive intrahepatic cholestasis of pregnancy and the healthy pregnant women group into a training set and a validation set; obtaining the index values of markers in the biological samples of the pregnant women group with positive intrahepatic cholestasis of pregnancy and the healthy pregnant women group in the training set, and performing machine learning training using a variety of risk prediction models to select the optimal parameters of each risk prediction model; under the optimal parameters of each risk prediction model, using the index values of the markers in the validation set to verify the prediction effect of the risk prediction model; selecting the risk prediction model with the best prediction effect as the risk prediction model for predicting intrahepatic cholestasis of pregnancy; among them, the markers include clinical index markers and / or gene markers. The clinical index markers include the conception mode index and any one or more of the following clinical detection indexes: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase. The conception mode index is whether assisted reproduction is used; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0013] Further, the variety of risk prediction models include one or more of the following: generalized linear model, gradient boosting machine, random forest, and support vector machine; preferably, the biological sample is one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; preferably, the biological sample is collected when the pregnant woman is at 13 to 25 weeks of gestation.

[0014] To achieve the above object, according to the fifth aspect of the present invention, a screening method for markers for predicting intrahepatic cholestasis of pregnancy is provided. The screening method includes: collecting clinical index information of each sample in a group of pregnant women with positive intrahepatic cholestasis of pregnancy and a group of healthy pregnant women. The clinical index information includes the following multiple types: age, BMI, pregnancy and delivery history, whether the conception method is assisted reproduction, pregnancy complications, liver function indexes, and blood routine indexes; taking the clinical index information as the first candidate feature and filling in the missing clinical index information; filtering the clinical index information with a correlation higher than the correlation threshold in the clinical preparation information to obtain the second candidate feature; using the Lasso regression model to screen the second candidate feature and performing cross-validation evaluation, and finally retaining the second candidate feature with a frequency higher than the frequency threshold as the clinical index marker; and / or detecting the expression profile differences in biological samples from a group of pregnant women with positive intrahepatic cholestasis of pregnancy and a group of healthy pregnant women, thereby preliminarily screening out candidate gene markers; using at least two machine learning models to screen the candidate gene markers, and retaining the gene markers with a frequency higher than the frequency threshold that commonly appear in all machine learning models as the final gene markers; the clinical index marker and / or the gene marker are the markers for predicting intrahepatic cholestasis of pregnancy. Among them, the clinical index marker includes the conception method index and any one or more of the following clinical detection indexes: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase. The conception method index is whether it is assisted reproduction; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0015] According to the sixth aspect of the present invention, a computer-readable storage medium is provided. The storage medium includes a stored program. When the program runs, it controls the device where the storage medium is located to execute the above method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy or the above screening method for markers for predicting intrahepatic cholestasis of pregnancy.

[0016] According to the seventh aspect of the present invention, a processor is provided. The processor is used to run a program. When the program runs, it executes the above method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy or the above screening method for markers for predicting intrahepatic cholestasis of pregnancy.

[0017] Applying the technical solution of the present invention, by discovering 5 clinical index markers and 17 gene markers specifically associated with intrahepatic cholestasis of pregnancy, and using the markers of the present application as the detection object, the risk prediction of intrahepatic cholestasis of pregnancy with high specificity and high sensitivity is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings forming a part of this application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0019] Figure 1A It shows a schematic diagram of the screening process of clinical markers in Example 1 of the present invention;

[0020] Figure 1B It shows a schematic diagram of the screening process of gene markers in Example 1 of the present invention;

[0021] Figure 2 It shows a schematic diagram of the construction method and verification process of the gene marker model in Example 1 of the present invention;

[0022] Figure 3A It shows the sorting result of the screening frequencies of each feature in the clinical indicators in Example 1 of the present invention;

[0023] Figure 3B It shows the risk prediction effect of 5 clinical index markers on ICP in Example 1 of the present invention;

[0024] Figure 4A It shows the sorting result of the screening frequencies of each feature in the candidate gene markers in Example 1 of the present invention;

[0025] Figure 4B It shows the risk prediction effect of 17 gene markers on ICP in Example 1 of the present invention;

[0026] Figure 5 It shows the risk prediction effect of the combined use of 5 clinical index markers and 17 gene markers on ICP in Example 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the embodiments.

[0028] It should be noted that in this application, "diagnosis" refers to determining a disease during the onset period by means of certain methods; while "prediction" refers to determining a disease before the typical phenotype appears. In this application, the clinically diagnosed indicator for intrahepatic cholestasis of pregnancy is that the total bile acid TBA > 10. However, in the clinical sample population of this application, the TBA of all individuals is < 10, and intrahepatic cholestasis cannot be diagnosed. But through the prediction scheme of the markers in this application, a part of the patients who will be clinically diagnosed with intrahepatic cholestasis later can be distinguished in advance. That is to say, "prediction" in this application refers to detection at an earlier stage of clinical onset (in this application, it refers to before the 25th week of pregnancy, preferably between the 13th and 25th gestational weeks, with an average of the 17th gestational week).

[0029] As mentioned in the background art section, there is currently a need for clinical early risk prediction of intrahepatic cholestasis of pregnancy in pregnant women. This application compares the differential characteristics of the plasma-free RNA expression profiles of pregnant women who are healthy in the early and mid-pregnancy and diagnosed with intrahepatic cholestasis in the late pregnancy by using the antenatal examination indicators of the mother in the early and mid-pregnancy and the expression profile of plasma-free RNA generated from the mother's peripheral blood plasma, screens out the risk prediction markers for intrahepatic cholestasis of pregnancy, and constructs a risk prediction model for intrahepatic cholestasis of pregnancy by using machine learning algorithms (generalized linear model, gradient boosting machine, random forest, and support vector machine), improving the sensitivity and specificity of the prediction of intrahepatic cholestasis in the mid-pregnancy and achieving early intervention.

[0030] Based on the research results, the applicant proposed the technical solution of this application. In a typical implementation manner, there is provided an application of a detection reagent for markers in the preparation of related products for predicting intrahepatic cholestasis of pregnancy, where the markers include clinical index markers and / or gene markers. Among them, the clinical index markers include the conception method index and any one or more of the following clinical detection indicators: total bile acid, gamma-glutamyl transpeptidase, folic acid, and aspartate aminotransferase, and the conception method, and the conception method index is whether it is assisted reproduction; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0031] The present application first discovers that the above-mentioned markers in pregnant women's biological samples have a significant correlation with intrahepatic cholestasis of pregnancy in pregnant women, and thus can be used as risk prediction markers for predicting intrahepatic cholestasis of pregnancy in pregnant women. Using the markers of the present application as marker factors for risk prediction realizes the risk prediction of intrahepatic cholestasis of pregnancy in the second trimester of pregnancy, and both the prediction sensitivity and specificity are higher than those of the prior art.

[0032] In the application of the present application, the detection reagent is a detection reagent for detecting the expression level of the marker. When the marker is a gene marker, the detection reagent is a detection reagent for the mRNA or protein expression level of the gene marker; more preferably, the detection reagent for detecting the mRNA expression level of the gene marker is a probe and / or primer, and further preferably, it is a reagent related to preparing a high-throughput sequencing library from the mRNA of the gene marker.

[0033] Specifically, the detection reagents for clinical index markers can be detected using commonly used reagents or reagent kits in the art. For example, folic acid can be detected by chemiluminescent microparticle immunoassay. The specifically applicable reagent kit is a folic acid assay kit (for example, product number: National Medical Device Approval Import 20182402392), and the detection instrument is Abbott I2000; total bile acid can be detected by the enzyme cycling method, and the applicable reagent kit is a total bile acid assay kit (product number: Shanghai Medical Device Approval Standard 20172400590), and the detection instrument is Roche Cobas c702; γ-glutamyl transpeptidase can be detected by the enzyme colorimetric method, and the applicable reagent kit is a γ-glutamyl transpeptidase detection kit (product number: Jiangsu Medical Device Approval Standard 20202401213), and the detection instrument is Roche Cobas c702; aspartate aminotransferase can be detected by the enzyme colorimetric method, and the applicable reagent kit is an aspartate aminotransferase detection kit (product number: National Medical Device Approval Import 20162404192), and the detection instrument is Roche Cobas c702. It should be noted that the detection steps of all the above clinical index markers are carried out according to the instructions of the reagent kit. In addition, it should be noted that whether the conception method in the above clinical index is assisted reproduction can be obtained from the basic information of the pregnant woman's antenatal examination.

[0034] Since RNA sequencing usually includes a reverse transcription step for generating cDNA molecules for sequencing, when using RNA sequencing, the detection reagent of the present invention can also include a reagent for converting RNA in a biological sample into a cDNA fragment library.

[0035] Extracellular free RNA in biological samples can be extracted using commonly used methods or reagent kits in the art or a combination of both. For example, the standard RNA extraction procedure of TRIzol LS can be used to extract extracellular free RNA from plasma biological samples.

[0036] In a specific embodiment, quantitative analysis of extracellular free RNA preferably includes using whole transcriptome sequencing to sequence extracellular free RNA in a biological sample (preferably a plasma sample) of a pregnant woman using next-generation sequencing. This method can sequence plasma free mRNA and free lncRNA simultaneously. Analysis can also be performed using the RT-PCR method. Other methods known in the art, such as qPCR, can also be used to quantitatively analyze the expression profile of extracellular free RNA.

[0037] Preferably, quantitative analysis of extracellular free RNA further includes a step of quality control of the original extracellular free RNA sequencing data, preferably including trimming adapters, removing low-quality reads, removing reads with a length of <17 bp, removing rRNA sequences and value RNA and Y RNA sequences, and aligning the remaining reads to the human transcriptome (in the order of miRNA, tRNA, and piRNA, mRNA and lncRNA, and finally other RNAs). Then the remaining reads are aligned to the human genome. The expression level of long RNAs (including mRNA and lncRNA) is corrected to TPM, and the formula is as follows:

[0038] TPM = (Ni / Li) * 1000000 / (sum(N1 / L1 + N2 / L2 + N3 / L3 + … + Nn / Ln)),

[0039] where Ni is the number of reads aligned to the i-th gene; Li is the length of the i-th gene;

[0040] sum(N1 / L1 + N2 / L2 +... + Nn / Ln) is the sum of the values after normalization of all (n) genes by length.

[0041] In the application of the present application, a biomarker in a biological sample derived from a pregnant woman is detected using a detection reagent, wherein the biological sample derived from a pregnant woman is selected from one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; preferably, the biological sample is collected at the 13th - 25th gestational week of the pregnant woman. Plasma, serum, or whole blood derived from a pregnant woman is preferably used. The biological sample is most preferably plasma. For example, peripheral blood can be obtained from a pregnant woman and plasma separation can be performed to obtain the plasma biological sample to be used. In addition to plasma, serum, or whole blood, other body fluid samples, such as urine, amniotic fluid, etc., can also be used. The acquisition of the biological sample can be implemented using conventional methods in the art.

[0042] In the present invention, the biological sample can be collected during the 13th to 25th gestational week of the pregnant woman. By using the above-mentioned markers of the present application as the risk prediction markers for intrahepatic cholestasis of pregnancy. With the markers of the present application, the risk prediction of intrahepatic cholestasis of pregnancy can be achieved in the second trimester. The present invention can predict intrahepatic cholestasis of pregnancy before the onset. Therefore, the method of the present invention has a wider range of applicable populations and greater clinical application value.

[0043] By using the clinical index markers and / or gene markers of the present invention as the markers for predicting intrahepatic cholestasis of pregnancy in pregnant women, according to the preparation principle of the existing kits, a prediction kit for the gene markers of the present invention can be prepared. Detection probes, chips, etc. for predicting the risk of premature rupture of fetal membranes and preterm birth in pregnant women can also be prepared for these gene markers. Therefore, in some preferred application scenarios of the present application, the related products can be kits or electronic prediction devices.

[0044] In the second typical embodiment of the present application, a kit for predicting intrahepatic cholestasis of pregnancy is provided. The kit includes a detection reagent for the markers of intrahepatic cholestasis of pregnancy, the marker clinical index markers and / or gene markers. Among them, the clinical index markers include the conception mode index and any one or more of the following clinical detection indexes: total bile acid, γ-glutamyl transpeptidase, folic acid, and aspartate aminotransferase and the conception mode. The conception mode index is whether it is assisted reproduction; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0045] The above-mentioned markers have a significant correlation with intrahepatic cholestasis of pregnancy in pregnant women, and thus can be used as risk prediction markers for predicting intrahepatic cholestasis of pregnancy in pregnant women. Therefore, the kit including the markers of the present application, by using these markers as the marker factors for risk prediction, can improve the specificity and sensitivity of the risk prediction of intrahepatic cholestasis of pregnancy in pregnant women.

[0046] In some preferred embodiments, the detection reagent is a detection reagent for detecting the expression level of the marker. When the marker is a gene marker, the detection reagent is a detection reagent for detecting the mRNA or protein expression level of the above-mentioned gene marker; preferably, the detection reagent for detecting the mRNA expression level of the detection gene marker is a probe and / or primer, and further preferably, it is a related reagent for preparing a high-throughput sequencing library from the mRNA of the gene marker.

[0047] Specifically, detection reagents for clinical indicator markers can be detected using commonly used reagents in the art. For example, folic acid can be detected by chemiluminescent microparticle immunoassay, and the specific kit applicable is the folic acid assay kit (for example, product number: National Medical Device Approval Import 20182402392), and the detection instrument is Abbott I2000; total bile acid can be detected by the enzyme cycling method, and the applicable kit is the total bile acid assay kit (product number: Shanghai Medical Device Approval Standard 20172400590), and the detection instrument is Roche Cobas c702; γ-glutamyl transpeptidase can be detected by the enzyme colorimetric method, and the applicable kit is the γ-glutamyl transpeptidase detection kit (product number: Jiangsu Medical Device Approval Standard 20202401213), and the detection instrument is Roche Cobas c702; aspartate aminotransferase can be detected by the enzyme colorimetric method, and the applicable kit is the aspartate aminotransferase detection kit (product number: National Medical Device Approval Import 20162404192), and the detection instrument is Roche Cobas c702. It should be noted that the detection steps of all the above clinical indicator markers are carried out according to the instructions of the kit. Among them, whether the conception method is assisted reproduction can be obtained from the basic information of the pregnant woman's antenatal examination.

[0048] Since RNA sequencing usually includes a reverse transcription step to generate cDNA molecules for sequencing, when using RNA sequencing, the detection reagent of the present invention can also include a reagent for converting RNA in a biological sample into a cDNA fragment library.

[0049] In the third typical embodiment of the present application, an electronic device for predicting intrahepatic cholestasis of pregnancy is also provided. The electronic device is built-in with a risk prediction model for intrahepatic cholestasis of pregnancy. The risk prediction model is generated by training a computer using the expression levels of markers in biological samples from pregnant women with positive intrahepatic cholestasis of pregnancy. The markers are clinical indicator markers and / or gene markers. Among them, the clinical indicator markers include the conception method index and any one or more of the following clinical detection indicators: total bile acid, γ-glutamyl transpeptidase, folic acid, and aspartate aminotransferase and the conception method. The conception method index is whether it is assisted reproduction; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0050] In the fourth typical embodiment of the present application, a method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy is provided. The construction method includes:

[0051] Pregnant women with intrahepatic cholestasis of pregnancy (ICP) positive and healthy pregnant women are divided into a training set and a validation set;

[0052] Obtain the index values of markers in biological samples of pregnant women with ICP positive and healthy pregnant women in the training set, and perform machine learning training using a variety of risk prediction models to select the optimal parameters of each risk prediction model;

[0053] Under the optimal parameters of each risk prediction model, use the index values of markers in the validation set to verify the prediction effect of the risk prediction model;

[0054] Select the risk prediction model with the best prediction effect as the risk prediction model for predicting intrahepatic cholestasis of pregnancy;

[0055] Among them, the markers include clinical index markers and / or gene markers. The clinical index markers include the conception method index and any one or more of the following clinical detection indexes: total bile acid, γ-glutamyl transpeptidase, folic acid, aspartate aminotransferase, and the conception method. The conception method index is whether it is assisted reproduction; the gene markers include any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0056] In a preferred embodiment, the variety of risk prediction models include one or more of the following: generalized linear model, gradient boosting machine, random forest, and support vector machine.

[0057] Training the computer can be implemented by machine learning methods. The machine learning methods are selected from regression methods, classification methods, or a combination thereof. "Machine learning" generally refers to an algorithm that gives a computer the ability to learn without explicit programming, including algorithms that learn from data and make predictions on data. The machine learning methods used in the present invention may include random forest, least absolute shrinkage and selection operator logistic regression, regularized logistic regression, XGBoost, decision tree learning, artificial neural network, deep neural network, support vector machine, rule-based machine learning, generalized linear model, gradient boosting machine, etc. Preferred machine learning methods include one or more of the following: generalized linear model, gradient boosting machine, random forest, support vector machine.

[0058] In the model construction method of the present invention, the training set and the validation set can be split according to a certain ratio as needed. Preferably, all pregnant women with intrahepatic cholestasis of pregnancy are randomly split into a training set and a validation set according to a ratio of 7:3 in terms of the number of people, and all healthy pregnant women are randomly split into a training set and a validation set according to a ratio of 7:3 in terms of the number of people. The screening of the best markers is completed in the training set, and the validation set is used to test the prediction effects of the best markers and the model.

[0059] In a preferred embodiment, candidate gene markers are initially screened by comparing the differences in gene expression profiles between the group of pregnant women with intrahepatic cholestasis of pregnancy and the group of healthy pregnant women. The gene markers may include mRNA genes and lncRNA genes. This step can be implemented using, for example, the DESeq2 package (R software package). Finally, they are sorted in ascending order of p-values, and the top 1000 genes are selected as candidate molecular markers and further screened in the subsequent prediction model. The candidate molecular markers are first screened using the lasso model and random forest, and the high-frequency features that appear in both are used as the final gene markers.

[0060] In a preferred embodiment, in the training set, based on the finally screened best gene markers and / or best clinical index markers, four machine learning methods: generalized linear model (GLM), gradient boosting machine (GBM), random forest (RF), and support vector machine (SVM) are used to predict the risk of intrahepatic cholestasis of pregnancy. Preferably, for each algorithm, 7-fold cross-validation is used to select the optimal parameters for constructing the prediction model. The formed model can be verified in the validation set.

[0061] Preferably, through the effect verification of the validation set, the model with the best effect can be selected and the feature importance can be calculated.

[0062] Preferably, the gene markers and clinical index markers can be used alone as markers to evaluate the risk prediction effect, or can be used together as markers to verify the risk prediction effect, so as to construct a risk prediction model.

[0063] In a preferred embodiment, the prediction model constructed by the method of the present invention can achieve risk prediction of intrahepatic cholestasis of pregnancy before the onset of the disease. Therefore, it is only necessary to combine clinical antenatal examination indicators and / or collect peripheral blood of pregnant women to perform non-invasive risk prediction of intrahepatic cholestasis of pregnancy. The AUC (area under the receiver operating characteristic curve) of predicting ICP using only clinical indicator markers was 0.81 and 0.751 in the training set and the validation set respectively, with a sensitivity of 64% and a specificity of 79% in the validation set. The AUC of predicting ICP using only gene markers was 0.94 and 0.74 in the training set and the validation set respectively, with a sensitivity of 55% and a specificity of 81% in the validation set. When the two types of markers are used in combination, the AUC of predicting ICP is 0.97 and 0.84 in the training set and the validation set respectively, with a sensitivity of 73% and a specificity of 84% in the validation set, which are significantly higher than the prior art level.

[0064] It should be noted that in the risk prediction model, the risk score can be automatically calculated by the model to evaluate and predict the level of intrahepatic cholestasis risk during pregnancy. The optimal threshold is determined according to the ROC curve and the Youden index.

[0065] In a preferred embodiment, the biological sample is one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; preferably, the biological sample is collected when the pregnant woman is at 13 to 25 weeks of gestation. The most preferred biological sample is plasma. For example, peripheral blood can be obtained from the pregnant woman and plasma separation can be performed to obtain the plasma biological sample to be used. In addition to plasma, serum or whole blood, other body fluid samples such as urine, amniotic fluid, etc. can also be used. The acquisition of the biological sample can be implemented by conventional methods in the art.

[0066] In the present invention, the biological sample can be collected when the pregnant woman is at 13 to 25 weeks of gestation. By using the above-mentioned markers of the present application as markers for predicting the risk of intrahepatic cholestasis of pregnancy. Using the markers of the present application, the risk prediction of intrahepatic cholestasis of pregnancy can be achieved in the second trimester of pregnancy. The present invention can predict intrahepatic cholestasis of pregnancy before the onset of the disease. Therefore, the method of the present invention has a wider range of applicable populations and greater clinical application value.

[0067] In the fifth typical embodiment of the present application, a method for screening markers for predicting intrahepatic cholestasis of pregnancy is provided, and the screening method includes:

[0068] Collect the clinical index information of each sample in the pregnant woman group positive for intrahepatic cholestasis of pregnancy and the healthy pregnant woman group. The clinical index information includes the following multiple types: age, BMI, pregnancy and delivery history, pregnancy complications, liver function indexes, and blood routine indexes;

[0069] Take the clinical index information as the first candidate feature and fill in the missing clinical index information;

[0070] Filter the clinical index information in the clinical preparation information with a correlation higher than the correlation threshold to obtain the second candidate feature;

[0071] Use the Lasso regression model to screen the second candidate feature and conduct cross-validation evaluation. Finally, retain the second candidate feature with a frequency higher than the frequency threshold as the clinical index marker; and / or

[0072] Detect the expression profile differences in biological samples from pregnant women with intrahepatic cholestasis of pregnancy (ICP) positive and healthy pregnant women to initially screen out candidate gene markers;

[0073] Use at least two machine learning models to screen the candidate gene markers, and retain the gene markers with a frequency higher than the frequency threshold that commonly appear in all machine learning models as the final gene markers;

[0074] The clinical index marker and / or gene marker is the marker for predicting intrahepatic cholestasis of pregnancy. Among them, the clinical index marker includes the conception method index and any one or more of the following clinical detection indexes: total bile acid, γ-glutamyl transpeptidase, folic acid, aspartate aminotransferase, and the conception method. The conception method index is whether it is assisted reproduction; the gene marker includes any one or more of the following: ABCA2, ALDH4A1, ANKRA2, BLVRB, ENTR1, EZR, FCHO2, HIST1H2AE, IQCE, LARS, MMP15, PXK, RNF141, SLC7A2, TGFBR2, TMEM135, and TMEM144.

[0075] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the present invention.

[0076] From the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary hardware devices such as detection instruments. Based on this understanding, the data processing part of the technical solution of the present application can be embodied in the form of a software product, and this computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of various embodiments or some parts of the embodiments of the present application.

[0077] The present application can be used in many general or special computing system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on.

[0078] Obviously, those skilled in the art should understand that some modules or steps of the above-mentioned present application can be implemented on a general computing device. They can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices. Optionally, they can be implemented with program codes executable by the computing device, so that they can be stored in a storage device and executed by the computing device, or they can be separately made into individual integrated circuit modules, or multiple modules or steps of them can be made into a single integrated circuit module to implement. In this way, the present application is not limited to any specific combination of hardware and software.

[0079] In a preferred embodiment, a computer-readable storage medium is provided. The storage medium includes a stored program, wherein when the program runs, it controls the device where the storage medium is located to execute the above-mentioned method for constructing a risk prediction model for intrahepatic cholestasis of pregnancy or the above-mentioned method for screening markers for intrahepatic cholestasis of pregnancy.

[0080] In a preferred embodiment, a processor is provided. The processor is used to run a program, wherein when the program runs, it executes the above-mentioned method for constructing a risk prediction model for intrahepatic cholestasis of pregnancy or the above-mentioned method for screening markers for intrahepatic cholestasis of pregnancy.

[0081] The beneficial effects of the present application will be further described below in conjunction with specific embodiments.

[0082] Example 1

[0083] (1) Sample enrollment

[0084] Single - fetus pregnant women samples were collected from the hospital. After excluding pregnant women lacking clinical information, a total of 203 pregnant women were enrolled. Among them, 149 pregnant women were enrolled in 2017 and 2018 and served as the training set, including 43 pregnant women who subsequently developed intrahepatic cholestasis of pregnancy (ICP) and 106 healthy pregnant women. The validation set included 54 pregnant women enrolled in 2016, among whom 11 subsequently developed ICP and 43 were healthy.

[0085] (2) Clinical information collection and processing

[0086] Clinical information of samples was collected from all pregnant women, including basic information (age, BMI, etc.), pregnancy and delivery history, pregnancy complications, liver function indicators and blood routine. The detailed clinical information is shown in Table 1.

[0087] Table 1:

[0088]

[0089]

[0090] (3) Maternal plasma acquisition

[0091] Peripheral blood of 203 single - fetus pregnant women was obtained from the hospital. The average gestational week for blood collection was 17 weeks. All blood samples were immediately stored at 4°C and plasma separation was performed within 8 hours. Plasma separation was carried out by a two - step centrifugation method. Centrifuge at 1,600g for 10 minutes at 4°C and then at 12,000g for 10 minutes. After plasma separation, it was immediately stored at - 80°C waiting for the next step of processing.

[0092] (4) Extraction and sequencing of cfRNA

[0093] Trizol LS was added to the plasma and immediately shaken and mixed evenly. Subsequent cfRNA extraction steps can use methods familiar in the art. Sequencing of cfRNA utilized whole - transcriptome sequencing, and next - generation sequencing was used to sequence plasma samples of ICP and healthy pregnant women.

[0094] (5) Quantitative analysis of cfRNA expression profiles

[0095] The original cfRNA sequencing data is quality controlled, including trimming adapters, removing low-quality reads, removing reads with a length of <17 bp, removing rRNA sequences, value RNA, and Y RNA sequences. The remaining reads are first aligned to the human transcriptome (in the order of miRNA, tRNA, and piRNA, mRNA and lncRNA, and finally other RNAs), and then the remaining reads are aligned to the human genome. The expression levels of long RNAs (including mRNA and lncRNA) are corrected to TPM, and the formula is as follows.

[0096] TPM = (Ni / Li) * 1000000 / (sum(N1 / L1 + N2 / L2 + N3 / L3 + … + Nn / Ln))

[0097] Ni is the number of reads aligned to the i-th gene; Li is the length of the i-th gene; sum(N1 / L1 + N2 / L2 +... + Nn / Ln) is the sum of the values after normalizing all (n) genes by length.

[0098] (6) Molecular biomarker screening

[0099] All molecular biomarker screening is completed in the training set, and the validation set is used to test the prediction effects of molecular biomarkers and models.

[0100] (6.1) Clinical information as features

[0101] The clinical information used as candidate features is shown in Table 1 ("Yes" in the last column). The screening process is as Figure 1A shown. Since some samples lack some clinical information, the bagImpute function (caret package in R) is used for filling. Then, highly correlated variables are excluded (here it refers to high correlation between features. Removing highly correlated variables is to provide more information with fewer features. The threshold for high correlation here is a Pearson correlation coefficient > 0.6). Then, the lasso algorithm is used to screen features. Cross-validation is used for evaluation, and the most frequently occurring of all candidate features are used as the final clinical information features.

[0102] (6.2) Plasma-free RNA as features

[0103] First, candidate molecular biomarkers are preliminarily screened by comparing the expression profile differences between the ICP group and the healthy group. As Figure 1BAs shown, differential analysis was performed on each mRNA gene in two groups, and this step was implemented using the DESeq2 package (R software package). Finally, they were sorted from smallest to largest according to the p-value, and the top 1000 genes were selected as candidate molecular markers and further screened in the subsequent prediction model. The candidate molecular markers were first screened using the lasso model and random forest, and the high-frequency features (here, the frequency threshold was set as the number of occurrences greater than or equal to 7 times) that appeared in both were used as the final gene markers, as shown in Table 2.

[0104] Table 2:

[0105]

[0106]

[0107] (7) Construction and validation of the marker model

[0108] As Figure 2 shown, in the training set, based on the finally selected features (mRNA molecules, clinical information), 4 machine learning algorithms (generalized linear model - GLM, generalized linear model; gradient boosting machine - GBM, Gradient Boosting Machine; random forest - RF, Random Forest; and support vector machine - SVM, support vector machine) were used to evaluate the risk of preterm birth. Each algorithm used 7-fold cross-validation to select the optimal parameters for constructing the prediction model. The final model was verified in the validation set, and the best model was selected as the final model and the feature importance was calculated. At the same time, the mRNA molecular markers and clinical indicators were jointly applied to test and verify the effect and build the model together.

[0109] (8) Evaluation of the prediction effect of molecular markers on the risk of ICP

[0110] (8.1) Prediction effect of clinical information on the risk of ICP

[0111] After feature selection of clinical indicators, 5 clinical indicator markers were generated, as Figure 3A shown. The AUC for predicting ICP in the training set and validation set were 0.81 and 0.751 respectively, the sensitivity in the validation set was 64%, and the specificity was 79%, as shown in Table 4 and Figure 3B shown.

[0112] (8.2) Prediction effect of molecular markers on the risk of ICP

[0113] After feature selection of mRNA, 17 clinical indicators were generated, asFigure 4A As shown, when training and predicting using a single mRNA as a feature, it was found that PXK, RNF141, and FCHO2 had the highest AUC in the seven-fold cross-validation of the training set, as shown in Table 3. At the same time, ABCA2 and SLC7A2 are functionally related to the liver. Therefore, these 5 genes were gradually added for training and prediction later, and the results are shown in Table 4 and Figure 4B . The prediction effect of combining PXK, RNF141, FCHO2, ABCA2, and SLC7A2 on ICP is better than that of combining 17 genes. The AUCs of the 5-gene combination in the training set and the validation set are 0.86 and 0.80 respectively, the sensitivity in the validation set is 64%, and the specificity is 84%, as shown in Table 4 and Figure 4B as shown.

[0114] (8.3) Prediction effect of combined markers on ICP risk

[0115] After combining clinical indicators and mRNA features, the prediction effect on ICP is as Figure 5 shown in and Table 4. The prediction effect of combining the 5 genes with clinical information is similar to that of combining 17 genes with clinical information. The predicted AUCs of the 5 genes combined with clinical information in the training set and the validation set are 0.89 and 0.87 respectively, the sensitivity in the validation set is 73%, and the specificity is 86%. The predicted AUCs of the 17 genes combined with clinical information in the training set and the validation set are 0.97 and 0.84 respectively, the sensitivity in the validation set is 73%, and the specificity is 84%, as Figure 5 shown in and Table 4.

[0116] Table 3:

[0117] Feature 7-fold cross-validation AUC PXK 0.745479025 RNF141 0.727919501 FCHO2 0.716914683 IQCE 0.700325964 SLC7A2 0.696605726 BLVRB 0.695521542 ENTR1 0.692410714 MMP15 0.689965986 TMEM135 0.682794785 TGFBR2 0.67154195 TMEM144 0.663378685 ANKRA2 0.649964569 EZR 0.635962302 HIST1H2AE 0.632312925 LARS 0.630626417 ALDH4A1 0.626388889 ABCA2 0.623511905

[0118] Table 4

[0119]

[0120] In the above table, RNA group X, that is, RNA group X. For example, RNAgroup17 is RNA group 17, referring to the combination of 17 mRNAs, and so on for others. RNAclin17 is the combination of 17 mRNAs and all clinical indicator markers.

[0121] From the above results, it can be seen that the above embodiments of the present invention achieve the following technical effects: Since plasma-free RNA is derived from various tissues throughout the body, it provides a means to monitor the health status of tissues, including reproductive tissues such as the placenta. Based on plasma-free RNA and clinical prenatal examination indicators during pregnancy, combined with machine learning algorithms, molecular markers for predicting the risk of intrahepatic cholestasis of pregnancy can be screened, and the prediction of intrahepatic cholestasis of pregnancy in the second trimester can be achieved by constructing a model.

[0122] The present invention only needs to collect the peripheral blood of pregnant women, and can use a non-invasive method to predict the risk of intrahepatic cholestasis earlier with relatively high accuracy. The prediction sensitivity in the second trimester of pregnancy is above 0.7 and the specificity reaches above 0.8, both higher than the prior art level. The method of the present invention can be applied to predict in the second trimester of pregnancy, which helps to achieve early intervention.

[0123] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. Use of a biomarker in the preparation of a related product for predicting intrahepatic cholestasis of pregnancy, characterized in that, The marker is a gene marker; The gene marker is one or more of SLC7A2, PXK, RNF141, or FCHO2.

2. The application according to claim 1, characterized in that, The application includes using a detection reagent to detect the expression levels of the clinical detection index and the gene marker. Preferably, when the marker is the gene marker, the detection reagent is a detection reagent for detecting the mRNA expression level or protein expression level of the gene marker; More preferably, the detection reagent for detecting the mRNA expression level of the gene marker is a probe and / or primer, and further preferably, it is a reagent related to preparing the mRNA of the gene marker into a high-throughput sequencing library.

3. The application according to claim 1 or 2, characterized in that, Combined with the conception mode index, the marker in the biological sample from the pregnant woman is detected, wherein the biological sample is selected from one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; Preferably, the biological sample is collected during the 13th to 25th gestational week of the pregnant woman; Preferably, the related product includes a kit or an electronic prediction device.

4. A kit for predicting intrahepatic cholestasis of pregnancy, characterized in that, The kit includes a detection reagent for the marker of intrahepatic cholestasis of pregnancy, and the marker is a gene marker; The gene marker is one or more of SLC7A2, PXK, RNF141, or FCHO2.

5. The kit according to claim 4, wherein The detection reagent is a detection reagent for detecting the expression levels of the clinical detection index and the gene marker; Preferably, the detection reagent is a detection reagent for detecting the mRNA expression level or protein expression level of the gene marker; Preferably, the detection reagent for detecting the mRNA expression level of the gene marker is a probe and / or primer, and further preferably, it is a reagent related to preparing the mRNA of the gene marker into a high-throughput sequencing library.

6. An electronic device for predicting intrahepatic cholestasis of pregnancy, characterized in that, The electronic device is built-in with a risk prediction model for intrahepatic cholestasis of pregnancy in pregnant women. The risk prediction model is generated by training a computer using the expression levels of the markers in the biological samples from pregnant women with intrahepatic cholestasis of pregnancy, and the marker is a gene marker; The gene marker is one or more of SLC7A2, PXK, RNF141, or FCHO2.

7. A method for constructing a risk prediction model for intrahepatic cholestasis of pregnancy, characterized in that, The construction method includes: Dividing the pregnant women population with intrahepatic cholestasis of pregnancy and the healthy pregnant women population into a training set and a validation set; Obtaining the index values of the markers in the biological samples from the pregnant women population with intrahepatic cholestasis of pregnancy and the healthy pregnant women population in the training set, and performing machine learning training using multiple risk prediction models to select the optimal parameters of each risk prediction model; Under the optimal parameters of each risk prediction model, using the index values of the markers in the validation set to verify the prediction effect of the risk prediction model; Selecting the risk prediction model with the best prediction effect as the risk prediction model for predicting intrahepatic cholestasis of pregnancy; The marker is a gene marker; The gene marker is one or more of SLC7A2, PXK, RNF141, or FCHO2.

8. The construction method according to claim 15, characterized in that, The multiple risk prediction models include one or more of the following: generalized linear model, gradient boosting machine, random forest, and support vector machine. Preferably, the biological sample is one or more of the following: plasma, serum, whole blood, urine, amniotic fluid; preferably, the biological sample is collected during the 13th to 25th gestational week of the pregnant woman.

9. A screening method for a marker for predicting intrahepatic cholestasis of pregnancy, characterized in that, The screening method includes: Collecting the clinical index information of each sample in the pregnant woman group positive for intrahepatic cholestasis of pregnancy and the healthy pregnant woman group, and the clinical index information includes multiple items as follows: age, BMI, pregnancy history, whether the conception method is assisted reproduction, pregnancy complications, liver function indexes, and blood routine indexes; Taking the clinical index information as the first candidate feature and filling in the missing clinical index information; Filtering the clinical index information with a correlation higher than the correlation threshold in the clinical preparation information to obtain the second candidate feature; Detecting the expression profile differences in the biological samples from the pregnant woman group positive for intrahepatic cholestasis of pregnancy and the healthy pregnant woman group, so as to preliminarily screen out candidate gene markers; Using at least two machine learning models to screen the candidate gene markers, and retaining the gene markers with a frequency higher than the frequency threshold that commonly appear in all the machine learning models as the final gene markers; The marker is a gene marker; The gene marker is one or more of SLC7A2, PXK, RNF141, or FCHO2.

10. A computer-readable storage medium, characterized in that, The storage medium includes a stored program, wherein when the program runs, it controls the device where the storage medium is located to execute the method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy according to any one of claims 7 to 8 or the method for screening a marker for predicting intrahepatic cholestasis of pregnancy according to claim 9.

11. A processor, characterized in that, The processor is used to run a program, wherein when the program runs, it executes the method for constructing a risk prediction model for predicting intrahepatic cholestasis of pregnancy according to any one of claims 7 to 8 or the method for screening a marker for predicting intrahepatic cholestasis of pregnancy according to claim 9.

Images