WO2012031412A1

WO2012031412A1 - Plasma mirna profile for anticipating therapeutic effect of interferon in treating chronic hepatitis b and detecting kits

Info

Publication number: WO2012031412A1
Application number: PCT/CN2010/077506
Authority: WO
Inventors: 袁正宏; 张小楠; 邬敏; 陈良; 张继明; 张欣欣; 张占卿; 吴景迪; 王介非; 陈晓蓉; 黄涛
Original assignee: 上海市公共卫生临床中心
Priority date: 2010-09-08
Filing date: 2010-09-30
Publication date: 2012-03-15
Also published as: CN102399852A

Abstract

A plasma miRNA profile used for anticipating therapeutic effect of interferon in treating chronic Hepatitis B is provided, which contains one or more miRNAs of SEQ ID NO.1-10. The kit, detecting chip and method used for analyzing expression level of corresponding miRNA profile are also provided. The products and method posses high efficiency and low cost, and are suitable for evaluating therapeutic effect of interferon in clinic. The miRNA profile of present invention is an independent factor for anticipating. Combining with HBV drug resistance mutation analysis, the miRNA profile is helpful in developing individual therapeutic project, reducing therapeutic cost and improving proportion of complete response.

Description

Plasma miRNA profile and detection kit for predicting the efficacy of interferon in the treatment of chronic hepatitis B

Technical field

The present invention relates to the fields of molecular biology and medicine. More specifically, it relates to plasma miRNA expression levels predicting the efficacy of interferon in the treatment of chronic hepatitis B. The invention also relates to methods and kits for detecting miRNA levels.

Background technique

Hepatitis B virus (HBV) infection can cause chronic liver disease and significantly increase the risk of developing hepatocellular carcinoma. Although hepatitis B surface antigen recombinant antigen vaccine has been widely used to prevent infection, HBV infection remains an important health problem in countries all over the world. Interferon alpha and nucleoside analogs, such as lamivudine, adefovir, entecavir, telbivudine, etc., are the main therapeutic agents for HBV infection. Interferon therapy usually only achieves a complete response in a limited number of patients, and is prone to recurrence after discontinuation of treatment. In recent years, the use of pegylated interferon (also known as long-acting interferon, mainly PEG-IFNa2a, PEG-IFNa2b) has significantly improved e antigen seroconversion (PEG-IFNa2a 32%, PEG-IFNa2b 36%) and A small percentage of patients had surface antigen seroconversion (2.9%, PEG-IFNa2a). However, long-acting interferons are prone to cause side effects such as flu-like symptoms, depression, and personality changes. Nucleoside analogs, such as lamivudine, adefovir, entecavir, telbivudine, etc., are potent inhibitors of hepatitis B reverse transcriptase polymerase and block viral genome replication. However, such drug treatments do not achieve a high seroconversion rate, and a large proportion of patients relapse after discontinuation of medication. In addition, long-term use is likely to cause the formation of drug resistance mutations.

Although the treatment of chronic hepatitis B, viral seroconversion rate and virological response have been significantly improved in recent years, the treatment effect of chronic hepatitis B is still not ideal. In particular, long-acting interferons are very expensive for low-income families in developing countries. Therefore, there is an urgent need for a method for predicting the efficacy of interferon therapy prior to treatment, thereby providing a solution for personalized treatment.

The currently known factors associated with prognosis are: 1. Pre-treatment ALT levels, 2. HBV DNA before treatment, 3. Sexuality, 4. Liver histology, 5. Dynamics of HBsAg titer during treatment. However, current indicators have limited predictive power and there is an urgent need for a molecular marker that can be detected in the blood with high predictive accuracy. Summary of the invention

The technical problem to be solved by the present invention is to determine a molecular marker which can be detected in blood and which has a high prediction accuracy.

Recent studies show a small non-coding RNA (about 22 nucleotides), also known as small RNA (microRNA, miRNA) ₀ small RNA present in all types of tissue, by a target mRNA3 'untranslated terminal nucleotide pairs Promote its degradation or inhibit translation. miRNAs are produced by transcription of long precursor RNA and are pretreated with Drosha (nuclease) and transported to the cytosol via exportin_5-mediated nuclear export. The miRNA is further cleaved by the DICER enzyme to form a mature miRNA of 17-24 nucleotides and binds to the RISC complex to perform gene silencing. Small RNAs are a class of short nucleotides present in plasma that have been reported to be very sensitive molecular markers of liver damage and pregnancy.

The present invention has been extensively studied to determine a plasma miRNA profile predicting the efficacy of interferon in the treatment of chronic hepatitis B, which includes one or more of the miRNAs of SEQ ID NO.

Further, according to the miRNA profile, a kit for predicting the therapeutic effect of interferon for treating chronic hepatitis B is provided, the kit comprising one or more miRNA-specific probe nucleotides for SEQ ID NO.

At the same time, a chip for predicting the efficacy of interferon in the treatment of chronic hepatitis B is provided, which contains one or more miRNA-specific probe nucleotides in SEQ ID NO 1-10. Most preferably, the chip contains a miRNA-specific probe nucleotide for SEQ ID NO 1-10.

The present invention also provides a method for predicting the therapeutic effect of interferon in the treatment of chronic hepatitis B, which is to detect the expression level of one or more miRNAs in SEQ ID NO 1-10 before use of interferon plasma and serum samples, and express the expression in the sample. The difference between the level and the standard level was analyzed by statistical methods to predict the efficacy of interferon.

Expression levels of at least one of the miRNAs are above or below a known standard level.

Specifically, the method comprises the steps of: (1) extraction of plasma, serum sample small RNA, (2) nucleic acid sample hybridization with a chip having one or more miRNA-specific probes as in SEQ ID NO. (3) Compare the hybridization signal with the hybridization sample.

The statistical method described is the analysis of the significance of the difference between the expression level and the standard level in the sample.

The statistical methods include: miRNA expression profiling scores each miRNA by maximum correlation degree minimum redundancy analysis; using neighborhood classification algorithm as a prediction engine for category judgment; eigenvalue increment analysis method and leave one method cross validation to optimize selected miRNAs quantity. In the present invention, the expression profile of a part of miRNA before treatment of hepatitis B patients can predict the therapeutic effect of interferon alpha, and accordingly, a method for detecting miRNA, a method for establishing a predictive model, and a detection kit are disclosed.

The present invention provides a unique miRNA expression profile that is capable of distinguishing whether a patient with HBV has a response to long-acting interferon or a common interferon treatment (initial virological response, 12 weeks greater than 21 og HBV DNA reduction). This expression profile consists of hsa-miR-1268 (SEQ ID NO. 1), has-miR-150* (SEQ ID NO. 2), has-miR-22 (SEQ ID NO. 3), has-miR-197 (SEQ ID NO. 4), hsa-miR- 26a (SEQ ID NO. 5), has- miR-188- 5p (SEQ ID NO. 6), has- miR-1471 (SEQ ID NO. 7), has- miR - 484 (SEQ ID NO. 8), has- miR-1181 (SEQ ID NO. 9), Has-miR-194 (SEQ ID NO. 10) composition. The present invention provides a method for detecting a serum miRNA expression profile. Specifically, the method includes the steps of: (1) extracting total plasma RNA, and (2) detecting the expression profile of a specific miRNA in the total RNA by using a microarray. It can be operated in accordance with conventional methods in the art.

The statistical method includes one or more of the following: (1) miRNA expression profiles are scored for each miRNA by maximum relevance minimum redundancy analysis (mRMR), (2) using a neighbor classification algorithm (NN) As a predictive engine for class judgment, (3) Eigenvalue Incremental Analysis (IFS) and leave-one-out cross-validation optimize the number of selected miRNAs.

Using the miRNA expression profile of the present invention, 66 samples of HBV patients receiving long-acting interferon therapy were analyzed with a sensitivity of 63.33% and a specificity of 69.44%. This miRNA profile consisted of 10 specific nucleotide sequences, and its expression was significantly different in the non-response and fast response groups.

The present invention further validated this predictive model in 28 patients with HBV who received general interferon therapy and achieved an accuracy of 60.7%. The present invention provides a miRNA expression profile capable of predicting the efficacy of interferon in the treatment of chronic hepatitis B with higher accuracy. This expression profile is an independent predictor compared to other known efficacy-related factors. This expression profile combined with HBV-resistant mutation analysis will facilitate the development of individualized treatment regimens, ultimately reducing treatment costs and increasing the proportion of complete responses. The corresponding detection method is convenient, fast, and low in cost. The invention also provides an efficient and inexpensive corresponding detection kit and chip. DRAWINGS

Figure 1. Optimization of the miRNA profile flow chart that predicts the efficacy of interferon.

Figure 2. Ten miRNA expression profiles are shown in the heat map of the fast response and non-response groups. Green: Less than mean - standard deviation, black: between mean - standard deviation and mean + standard deviation, red: greater than average + standard deviation. The ten miRNAs are: hsa- miR- 1268 (SEQ ID NO. 1), has- miR- 150* (SEQ ID NO. 2), has- miR- 22 (SEQ ID NO. 3), has- miR- 197 (SEQ ID NO. 4), hsa-miR-26a (SEQ ID NO. 5), has- miR-188-5p (SEQ ID NO. 6), has- miR-1471 (SEQ ID NO. 7), Has- miR-484 (SEQ ID NO. 8), has-miR-1181 (SEQ ID NO. 9), has-miR-194 (SEQ ID NO. 10).

Figure 3. IFS (incremental eigenvalue) plot of interferon efficacy prediction. X-axis, number of eigenvalues, Y-axis, go to an interactive test accuracy. The arrival accuracy is highest when 10 eigenvalues are selected.

Figure 4. Correlation analysis of liver-plasma miRNA expression profiles. detailed description

Example 1: Detection of miRNA expression profiles

I. Experimental materials:

The present invention obtains plasma and serum samples from 94 chronic hepatitis B patients receiving long-acting interferon and common interferon. The study was reviewed by an ethics committee and all patients signed informed consent. The training set included 66 patients with HBV who received EG-IFN a2a (180μ / week) or PEG-IFN a2b (100μ / week). The test set included 28 patients receiving normal interferon.

HBV patients (IFN-a2b or IFN-alb, 3-5 MU/qod). The basic clinical data of the enrolled patients are shown in Table 1, in which 81.9% are e antigen positive, and the average viral load is 7. 05 log ₁₀ copies/ml. Enrollment exclusion criteria: HBV patients enrolled were not treated with nucleoside analogues or interferon 6 months prior to initiation of treatment, viral load was above 5 X 10 ⁴ copies/ml, and liver alanine aminotransferase was abnormal 1. 0 ULN) . Exclusion criteria include HIV or HCV co-infection. All patients were tested for HBV surface antigen, e antigen (Abbott AXSYM HBsAg (normal: 0 - 2S/N) and HBe 2. 0 ME IA Kit (normal: 0

- 1. OS/CO), viral load was detected by quantitative PCR (PPI). The rapid response was defined as a decrease in HBV DNA greater than 21 og 12 weeks after the start of treatment and a decrease in less than 21 og was defined as no response. 2. Experimental methods

RNA extraction and miRNA chips

Total RNA was extracted from 400 μl plasma samples using the mirVanaTM PARISTM (Ambion, Austin, TX) kit. RNA concentrations were quantified by NanoDrop 1000 spectrometer (NanoDrop Technologies, Waltham, MA). The Qiagen RNAeasy FFPE kit was used to extract RNA from formalin fixed paraffin-embedded specimens (FFPE). Briefly, the procedure is as follows: Scrape the FFPE specimen from the slide with a scalpel, dewax it with xylene, remove xylene by centrifugation, and add 100% ethanol to remove residual xylene. After full-speed centrifugation and air drying, the protease K passed through the FFPE specimen was digested at 55 ° C for 15 minutes and at 80 ° C for 15 minutes. After removing the genomic DNA by a gDNA spin column, the sample was passed through an RNeasy column, washed twice, and eluted with nuclease-free water. miRNA chip detection

The present invention compared 94 plasma samples and 13 FFPE specimens using a human miRNA chip (Agilent Technologies, Santa Clara, CA). This chip contains the Sanger database V12. 0 851 human miRNAs. The sample total RNA was labeled with Cy3 and hybridized with the chip. After washing, the chip was scanned by G2505C (Agilent Technologies, Santa Clara, CA) scanner. The labeling and hybridization procedures were performed according to the standard operation of the agilent miRNA chip system. The information was analyzed by the scanning software Rev. 5. 0 (Agilent Technologies, Santa Clara, CA) and converted into intensity information. Through background removal, the chip signals were exported to GeneSpring GX10 software (Agilent Technologies, Standardization was carried out by Santa Clara, CA).

Quantitative RT-PCR-.

Small RNA was obtained from 100_200μ1 plasma samples using the mirVANA miRNA extraction kit, as described in the product manual. The eluted RNA was concentrated by the Labconco freeze-drying system, and the RNA was reverse transcribed and amplified using a megaplex RT primer library and an amplification primer library (ABI). The obtained cDNA was quantified using a Taqman miRNA detection kit. III. Experimental results

miRNA expression profiling was performed on 94 patients with HBV who received interferon therapy.

The strategy of the present invention to distinguish between different initial virological responses is shown in FIG. A total of 66 well-documented cases were enrolled as training

II ^

The training group was validated in 28 test cases by training the miRNA expression profile. The basic clinical levels of the two groups were similar (Table 1), and the initial response rates were 45.5% and 35.7%, respectively (p value 0. 4948, Fi sher' s exact test)

Table 1. Clinical basic data of the patient.

Note: ^a is a percentage in parentheses. ^b Fisher's exact test. ^ε unpaired t-test. ^d only counts genotypes B and C.

Clinical variable entire queue test group P value ^b

(n=94) (η=28) (training group versus test group ratio a

Comparative)

Age-year

Median 29 29 31

Range 18-54 18-47 18-54 0.1694 ^c gender

Number of males (%) 60 (64) 40 (61) 20 (71) 0.3179 female 34 26 8

ALT, xULN 3.42±2.06 3.40±2.07 3.48±2.08 0.7552 ^c

HBV-DNA 7.05±1.13 7.22±1.14 6.65±1.01 0.0664 ^c

(loglO copies/ml)

HBeAg positive one (%) 77 (81.9) 58 (87.9) 19 (67.9) 0.8197

HBV genotype

B 25 (26.6) 19 (28.8) 6 (21.4)

C 68 (72.3) 46 (69.7) 22 (78.6) 0.6108 ^d

D 1 (1.1) 1 (1-5) 0

IVR (initial response rate) 42.6 45.5 35.7 0.4948 Example 2: Statistical analysis and establishment of predictive models

I. Test method:

Maximum Correlation Minimum Redundancy Analysis (mRMR)

The analysis of the minimum degree of redundancy minimum redundancy was first developed by Peng et al. To evaluate and analyze eigenvalues, we used the (mRMR) method to evaluate the relevance of each MiRNA to patient outcomes and the redundancy of other miRNAs. A miRNA with a lower correlation and lower redundancy can be considered a good miRNA eigenvalue. To measure correlation and redundancy, interactive information (MI) is used. Finally, each miRNA gets a Maxrel score based on its association.

Nearest neighborhood classification algorithm (NN)

A neighbor classification algorithm (NN) is used in the present invention as a prediction algorithm. The NN algorithm is a simple but very effective machine learning method that is widely used for various classification problems. It makes a decision by calculating the distance between the vector to be detected and each vector in the training set. The interrogation vector will be grouped into one class that is closest to the vector distance.

The present invention uses the leave-one-cross test to evaluate the performance of the predictive model. In this method, each data is sequentially removed as test data of other data, and the prediction engine is trained by multiple detections. The final total prediction accuracy is calculated as follows:

_ TP+TN

― TP+TN+FP+FN

TP is true positive, TN is true negative, FP false positive, FN false negative

Eigenvalue Incremental Analysis (IFS)

The score of each eigenvalue for the predicted relevance is obtained after the mRMR analysis. Next, in order to select the optimal number of eigenvalues, we used the IFS method. We have established different NNA prediction models by using different numbers of eigenvalues. The leave-one-out cross test further tests and evaluates different predictive models. By calculating the overall prediction accuracy of each model, we obtained the IFS curve and obtained the optimal number of eigenvalues when the IFS curve reached its maximum.

Other analysis

Correlation analysis uses the lowess and lm functions of the R software. We used univariate and multivariate logistic regression analysis to evaluate the known efficacy-related factors and the odds ratios of the miRNA prediction models. The prediction of miRNA target genes uses miRanda, RNAhybrid and TargetScan online prediction services. Result

Looking for interferon therapeutic effects related miRNA profiles

The present invention first pre-treats the miRNA expression profile of 66 HBV patients before treatment, and removes miRNAs which are not detected in more than 50% of the samples, so that genes with relatively high expression levels can be enriched for further analysis. The invention uses mRMR The analytic method evaluated the scores most relevant to the efficacy (Maxrel score, Table 2), and the grading miRNAs further compared the distribution of the three miRNA expression levels in the fast response and non-responder patients by Fisher test. (1) less than the mean - standard deviation, (2) between the mean - standard deviation and the mean + standard deviation, (3) greater than the mean + standard deviation. Figure 2 shows the expression patterns of these miRNAs.

Next, the present invention uses the Nearest Neighbor Classification Algorithm (NN) for prediction, and the leave-one-out cross-validation further verifies the prediction performance. To determine the optimal number of MiRNAs included in the prediction engine, IFS curve analysis evaluated the prediction accuracy of the number of different features (Figure 3). The highest prediction accuracy was highest in the calculation of 10 miRNAs (Table 2). miR-1268, miR_150*, miR-22 and miR-197 are the most significant efficacy-related genes (p<0.01, Table 2). The overall prediction accuracy of the model consisting of ten miRNAs was 66.7%.

Table 2. Maxrel scores and fisher test results for selected miRNAs. miRNA Fisher Test Maxrel

Sequence target gene

p value * score

1 hsa-miR-1268 0.00015 0.199 ND.

2 hsa-miR-150* 0.00293 0.143 ND.

PIK3CG, ABHD12., ACCN4, FZD8, CACNA2D1,

3 hsa-miR-22 0.00490 0.132 PIP5K1B, ROCK2, PRKCB, TLN2, STMN1,

PIK3CG, ABHD12, ACCN4, FZD8, CACNA2D,

4 hsa-miR-197 0.00964 0.111 ROCK2, PRKCB, TLN2, PRKG2

ABHD12, ROCK2, PRKCB, STMN1, DAPK1

5 hsa-miR-26a 0.02464 0.102

6 hsa-miR-188-5p 0.01509 0.102 ND.

7 hsa-miR-1471 0.06961 0.064 ND.

PIK3CG, ABHD12, ACCN4, FZD8, CACNA2D1,

8 hsa-miR-484 0.06961 0.064 PIP5K1B, ROCK2, PRKCB, TLN2, DAPK1

9 hsa-miR-1181 0.06282 0.064 ND.

10 hsa-miR-194 0.08222 0.063 CACNA2D1, PIP5K1B, TLN2, STMN1

* The p-value of the Fisher test was compared between the three miRNA expression levels in RR/NR patients (1. less than mean-standard deviation, 2. between mean-standard deviation and mean + standard deviation, 3. greater than mean + standard deviation) The distribution of the condition is calculated. Example 3 Prediction of 28 patients receiving general interferon treatment samples

The present invention further tests whether the predicted models of the ten miRNAs can effectively distinguish the initial virological response in a test set consisting of 28 patients. As shown in Table 1, the basic clinical data of the test set and the training set are similar, except that the test set patients received general interferon therapy. Since different interferon preparations cause differences in the concentration of effective drugs in the body, this may affect the predicted performance. Despite this, we still get an overall forecast accuracy of 60.7% in the training set.

Comparison of miRNA prediction models with known efficacy factors

The present invention further performs univariate and multivariate logistic regression analysis to assess whether the miRNA model is an independent predictor associated with efficacy.

Univariate regression analysis showed that the odds ratio of the miRNA model reached 3.27 (p=0.007), which was a significant independent correlation factor. This analysis also found that ALT was a significant factor (odds ratio 1.47, p=0. 002), whereas female (p=0.127), HBVDNA «5 X 107 compared to ^5 X 107, p=0 078) No significant difference was obtained between the genotypes (both B and C = p = 290). Further multivariate regression showed that both miRNA and ALT were independently associated with efficacy, with an odds ratio of 2.98 (p=0. 022, Table 3). Table 3. Univariate and multivariate logistic regressions of factors associated with virological response. analysis. Clinical variable odds ratio 950/0 confidence interval

Low value

Univariate analysis

miRNA prediction model 3.27 1.39 7.69 0.007 Age-year 0.99 0.94 1.05 0.84 Female 1.94 0.83 4.57 0.127 Serum ALT, XULN 1.47 1.15 1.88 0.002

HBV-DNA 2.35 0.91 6.09 0.078

<5Xl(f compared with 5Xl(f

HBV genotype (B vs C) 1.64 0.65 4.14 0.290 Multivariate analysis

Serum ALT, XULN 1.40 1.09 1.81 0.009 miRNA prediction model 2.84 1.11 7.25 0.029

This analysis was analyzed throughout the entire queue (n=94). Liver and plasma miRNA expression profiles are highly correlated

The miRNA present in plasma is considered to be a product secreted by various organs. The liver is an important organ responsible for metabolism in the human body. It has been reported that miRNA can be released into peripheral blood in acute drug-induced liver injury. We examined the miRNA profiles of formalin-fixed formaldehyde-encapsulated liver biopsy specimens (FFPE) from 13 patients enrolled in this study. We first confirmed that 133 miRNAs were detected in all samples in FFPE specimens, and miRNAs in 210 were expressed in at least 11 of 13 samples. We further analyzed whether plasma miRNAs reflect the expression of liver MiRNA to some extent. Indeed, we found a high correlation between miRNA expression profiles in liver and plasma in each individual (r=0. 22-0. 64 p= 0. 013-4. 7 X 10-15, pearson correlation, graph 4A), in the overall analysis of the correlation coefficient 0 · 42, ρ <10- ²⁵⁰ of FIG 4Β. When only ten selected MiRNAs were considered, there was still a significant correlation (r = 0.25, p = 0.003). The above data suggest that the liver is the main provider of miRNA in plasma, and the plasma miRNA profile may partially indicate changes in expression in the liver.

Claims

Claim

A miRNA profile predicting the efficacy of interferon in the treatment of chronic hepatitis B, characterized in that the miRNA profile comprises one or more miRNAs of SEQ ID NO.

2. A kit for predicting the efficacy of interferon in the treatment of chronic hepatitis B, characterized in that the kit contains one or more miRNA-specific probe nucleotides for SEQ ID NO.

3. A chip for predicting the efficacy of interferon in the treatment of chronic hepatitis B, characterized in that the chip comprises one or more miRNA-specific probe nucleotides for SEQ ID NO.

4. The chip of claim 3, wherein the chip comprises a miRNA-specific probe nucleotide for SEQ ID NO. 1-10.

5. A method for predicting the efficacy of interferon in the treatment of chronic hepatitis B, characterized in that the expression level of one or more miRNAs in SEQ ID NO. 1-10 in a plasma or serum sample is detected prior to the use of interferon, The difference in the expression level of the above miRNA in the sample and its standard level was analyzed by statistical methods to predict the efficacy of interferon.

6. The method of claim 5, wherein the expression level of at least one of the miRNAs is above or below a known standard level.

7. The method of claim 5, wherein the method comprises the steps of:

(1) Extraction of small RNA from plasma and serum samples,

(2) the nucleic acid sample is hybridized to a chip having one or more miRNA-specific probes as in SEQ ID NO. 1-10,

(3) Compare the hybridization signal with the hybridization sample.

8. The method according to claim 5, wherein the statistical method is used to analyze the significance of the difference between the expression level and the standard level in the sample.

9. The method according to claim 5, wherein the statistical method comprises: miRNA expression spectrum is scored by maximum correlation degree minimum redundancy analysis for each miRNA; using a neighbor classification algorithm as a prediction engine for class determination ; eigenvalue incremental analysis and leave-one-out cross-validation optimize the number of selected miRNAs.