Breast cancer molecular typing method based on nanoString platform CTSL2 gene expression quantity

By combining the nanoString platform and mRNA probe hybridization reaction with machine learning methods, we have achieved efficient and accurate molecular subtyping of breast cancer, solving the accuracy problem caused by manual interpretation in existing technologies and improving the automation and accuracy of breast cancer subtyping.

CN118910226APending Publication Date: 2024-11-08HENAN CANCER HOSPITAL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411187944.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing molecular subtyping methods for breast cancer rely on immunohistochemistry, which is subject to human subjectivity in interpreting the results and cannot perform absolute quantification of protein expression, resulting in insufficient subtyping accuracy.

Method used

The expression level of CTSL2 gene was detected using the nanoString platform. Combined with mRNA probe hybridization reaction and nSolver 4.0 software analysis, molecular subtyping of breast cancer was performed using machine learning methods, including quantitative analysis and secondary classification of HER2, ESR1, PGR and Ki-67 genes.

Benefits of technology

It reduces manual operation steps and improves the accuracy and automation of typing results, especially by achieving an accuracy of 92.55% through secondary classification of CTSL2 gene expression levels.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118910226A_ABST
    Figure CN118910226A_ABST
Patent Text Reader

Abstract

The invention discloses a breast cancer molecular typing method based on nanoString platform CTSL2 gene expression quantity, and particularly relates to the technical field of medical treatment, the breast cancer molecular typing method comprises the following specific steps: 1, extracting total RNA in a breast cancer paraffin specimen by using a Qiagen FFPE RNA extraction kit; 2, measuring the total RNA concentration; and step 3, carrying out hybridization reaction on the mRNA probe and a target RNA sequence, sealing a film on a hybridization reaction product, putting the hybridization reaction product into a nanoString digital fluorescent barcode single-molecule analyzer, carrying out gene expression detection through a program set in advance, and carrying out quantitative analysis on expression quantities of HER2, ESR1, PGR and Ki-67 genes by using nSolver4.0 software. The method comprises the following steps: performing luminal type, HER2 + type and triple negative type (TN type for short) molecular typing on each specimen through a machine learning method, and further dividing the luminal type into luminal A type (LA type for short) and luminal B type (LB type for short) and dividing the HER2 + type into luminal B-HER2 + type (LB-HER2 type for short) and HER2 + type (HER2 type for short) through a threshold value of a CTSL2 gene.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a molecular typing method for breast cancer based on CTSL2 gene expression on a nanoString platform. Background Art

[0002] Breast cancer is a highly heterogeneous tumor. The traditional pathological morphological classification has gradually shown its imperfections in current clinical practice. With the development and clinical application of molecular biology technology, the concept of molecular classification based on tumor morphology combined with gene expression characteristics has gradually reached a consensus in the medical community. Common molecular subtypes of breast cancer are: Luminal A, Luminal B, HER-2 overexpression and Basal-like. Among them, Luminal A is the most common molecular subtype of breast cancer, with an incidence of 44.5%-69.0%. ER+ and / or PR+, HER-2-, low expression of Ki-67 protein (expression ≤14%), the best prognosis. Endocrine therapy (± chemotherapy) is often used, and endocrine therapy has the best effect. Tamoxifen and drug-induced castration drugs such as Noradrenaline are often selected before menopause, and aromatase inhibitors such as anastrozole and letrozole are often selected after menopause. Luminal B type: Luminal B type is 7.8%, ER+ and / or PR+, and / or HER-2+, Ki-67 protein is highly expressed (expression>14%), endocrine therapy is still effective, and the prognosis is good. Luminal B-HER2+ breast cancer is less responsive to tamoxifen than luminal A type due to positive HER2 expression, but it is still effective to switch to endocrine therapy with other mechanisms of action. Treatment often uses chemotherapy + endocrine therapy + targeted therapy. HER-2 overexpression type: The incidence is 14.7%, ER- and / or PR-, HER-2+, endocrine therapy is ineffective, chemotherapy is effective, and it is an indication for the HER2 targeted therapy drug Herceptin. The efficacy of HER2(+) breast cancer for cyclophosphamide combined with anthracycline (AC) chemotherapy is significantly better than that of luminal type. The clinical remission rate of the former can reach 70%, while that of the latter is 47%. Although this subtype is more sensitive to chemotherapy, the clinical prognosis is poor. Chemotherapy + targeted therapy is often used. One year of Herceptin treatment can reduce the relative risk of recurrence by 52% and increase the 3-year disease-free survival by 12%. Basal-like type: The incidence rate is 17.1%, ER- and / or PR-, HER-2-, endocrine therapy is ineffective, chemotherapy is effective, and the prognosis is the worst. Its metastasis often occurs in the viscera and central nervous system. Among breast cancer patients receiving preoperative neoadjuvant chemotherapy (AC), there is a high overall response rate and pathological remission rate. 85% of patients have clinical remission, of which 27% achieve pathological complete remission, which is significantly higher than luminal breast cancer. Although it is sensitive to preoperative neoadjuvant chemotherapy and has a high pathological remission rate, its prognosis is still the worst in the molecular classification of breast cancer.

[0003] NanoString nCounter technology is a new technology for mRNA differential expression analysis, which has the following advantages. First, the technology is based on the principle of nucleic acid hybridization, without PCR amplification, and can achieve absolute quantification of a small amount of RNA samples. Second, the technology can detect 12 samples at the same time, and the results are reproducible. Third, this method has few manual operation steps, a high degree of automation, and saves time.

[0004] Breast cancer with the same pathological morphology is highly heterogeneous at the molecular level due to molecular genetic changes, which leads to great differences in tumor prognosis and response to treatment. The molecular classification of breast cancer based on gene expression profiles and gene chips can more accurately reflect the biological behavior of tumors, judge prognosis, and facilitate the selection and research of more targeted personalized treatment methods. An individualized and systematic treatment plan should be formulated based on the molecular classification and other relevant factors of each patient.

[0005] The existing technology has the following deficiencies: In the molecular classification of breast cancer, the existing molecular classification method of breast cancer is based on immunohistochemistry to determine the expression of ER, PR, HER2 and Ki-67 proteins. The method is mostly based on manual interpretation of the results. The subjectivity of manual interpretation can easily affect the interpretation results, especially the interpretation of borderline cases. Secondly, the immunohistochemistry method cannot absolutely quantify the expression of proteins. Summary of the invention

[0006] To this end, the present invention provides a molecular typing method for breast cancer based on the expression of CTSL2 gene on the nanoString platform. The method comprises the following steps: hybridization reaction is carried out between mRNA probe and target RNA sequence, and the hybridization reaction product is sealed and placed in a nanoString digital fluorescent barcode single molecule analyzer to carry out gene expression detection through a pre-set program. Then, nSolver4.0 software is used to quantitatively analyze the expression of HER2, ESR1, PGR and Ki-67 genes. The molecular typing of each sample is carried out through a machine learning method, namely, TN type, ER+ type and HER2+ type. Then, the ER-positive and HER2-positive types are further typed based on the expression of CTSL2 gene.

[0007] In order to achieve the above object, the embodiment of the present invention provides the following technical solution: a method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform, comprising the following specific steps:

[0008] Step 1: Extract total RNA from breast cancer paraffin samples using the Qiagen FFPE RNA extraction kit;

[0009] Step 2: Determine the total RNA concentration and evaluate the quality of RNA extraction by the ratio of OD260 / 280 and OD260 / 230;

[0010] Step 3: Use specific probes (including reporter probes and capture probes) of the 21 gene panel to carry out hybridization reaction with the target RNA sequence, that is, Genelist is the source reference (Table 1). The hybridization reaction system is as follows:

[0011] S1, hybridization solution 70 μl + reporter probe one tube (total volume 42 μl) = Master mixture;

[0012] S2. Add 8 μl of Master Mix + 5 μl of RNA (150 ng, the insufficient amount is supplemented with RNase-free H2O) to each PCR tube.

[0013] S3, continue to add 2 μl of capture probe to each tube, the total volume is 15 μl;

[0014] S4. Mix gently and place in 65℃ PCR instrument for 16-24h.

[0015] Step 4: Add the hybridization reaction product in step 3 to the dedicated card holder of the nanoString digital fluorescent barcode single molecule analyzer, seal the film and place it in the nanoString digital fluorescent barcode single molecule analyzer to perform gene expression detection according to the pre-set program;

[0016] Step 5: nSolver4.0 software was used to quantitatively analyze the expression of HER2, ESR1, PGR, Ki-67 and CTSL2 genes, and molecular typing was performed on each specimen through machine learning and set thresholds. The specific typing method is as follows:

[0017] S1. The expression levels of HER2, ESR1, PGR and Ki-67 genes were quantitatively analyzed using nSolver4.0 software;

[0018] S2. Use machine learning methods, such as random forest or decision tree, to train a typing model for the expression of HER2, ESR1, PGR and Ki-67 genes in samples with clear clinical typing results, and predict the TN, ER+ and HER2+ subtypes of these samples based on the model;

[0019] S3. For patients predicted to be ER+ and HER2+, further secondary classification is performed; ER+ patients can be further divided into luminal A and luminal B, and HER2+ patients can be further divided into HER2+ and luminalB-HER2+.

[0020] Further, in step 5, the specific steps of the secondary classification are as follows: According to the CTSL2 gene expression level of the existing embodiment and the corresponding known clinical classification, the ROC curve of the classification is constructed ( Figure 1 and Figure 2 ), the known clinical classifications include luminal A and luminal B in the ER+ group and HER2 and luminal B-HER2 in the HER2+ group. According to the ROC curve, the cut-off value corresponding to the maximum value of the Youden index of the curve is selected as the classification value for determining luminal A and luminal B as well as HER2 and luminal B-HER2. When CTSL2 is greater than the cut-off value, the sample is luminal B or luminal B-HER2, otherwise it is luminal A or HER2.

[0021] Furthermore, the maximum value of the Youden index represents the maximum sum of the sensitivity and specificity of using the CTSL2 gene expression to distinguish luminal A type from luminal B type and HER2+ type from luminal B-HER2 type ( Figure 3 ).

[0022] Furthermore, in step 4, the time required for the entire operation is 30 minutes.

[0023] Furthermore, the manual operation in step three only takes 30 minutes, and 12 samples can be tested simultaneously at one time.

[0024] The present invention has the following advantages:

[0025] 1. The present invention uses a probe for mRNA quantification to perform hybridization reaction with the target RNA sequence, and then seals the hybridization reaction product and places it in the nanoString digital fluorescent barcode single molecule analyzer to perform gene expression detection through a pre-set program, effectively reducing the manual operation steps. Then, nSolver4.0 software is used to perform absolute quantitative analysis of HER2, ESR1, PGR, Ki-67 and CTSL2 genes, and molecular typing of each specimen is performed through machine learning and set thresholds. Compared with the prior art, the present invention can automatically predict the typing results through the model to reduce the impact of manual subjective interpretation on the accuracy of the typing results.

[0026] 2. The present invention adopts a two-step typing and identification approach, first predicting TN type, ER+ type and HER2+ type, and then further typing ER+ type and HER2+ type based on the expression level of the CTSL2 gene. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 : is the ROC curve for distinguishing LA type and LB type in the ER+ group of the embodiment.

[0028] Figure 2 : is the ROC curve for distinguishing HER2+ type and LB-HER2 type in the HER2 positive group in the embodiment.

[0029] Figure 3 It is a comparison of the mRNA typing and immunohistochemical typing results.

[0030] Figure 4 is a clinical information table of the embodiment. DETAILED DESCRIPTION

[0031] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The present invention provides a method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform, comprising the following specific steps:

[0033] Step 1: Extract total RNA from breast cancer paraffin specimens using Qiagen FFPE RNA extraction kit;

[0034] Step 2: Determine the total RNA concentration and evaluate the quality of total RNA by the values ​​of OD260 / 280 and OD260 / 230;

[0035] Step 3: Use gene-specific probes to carry out hybridization reaction with target RNA sequences, that is, Genelist is the source reference, but the reporter probe and capture probe for detecting gene expression are designed by yourself, see Table 1. The hybridization reaction system is as follows:

[0036] a. Hybridization solution 70 μl + reporter probe one tube (total volume 42 μl) = Master mixture;

[0037] b. Add 8 μl Master Mixture + 5 μl RNA (150 ng, the insufficient amount is supplemented with RNase-free H2O) to each PCR tube;

[0038] c. Add 2 μl of capture probe to each tube, with a total volume of 15 μl.

[0039] d. Mix gently and incubate in a 65℃ PCR machine for 16-24 hours.

[0040] This step only takes 30 minutes to perform manually, and 12 samples can be tested simultaneously.

[0041] Step 4: Add the hybridization reaction product in step 3 to the dedicated card holder of the nanoString digital fluorescent barcode single molecule analyzer, seal the membrane and place it in the instrument to perform gene expression detection according to the pre-set program. This step of the experimental operation only takes 30 minutes.

[0042] Step 5: Use nSolver4.0 software to perform absolute quantitative analysis on the expression of HER2, ESR1, PGR, Ki-67 and CTSL2 genes, and perform molecular typing of each specimen through machine learning and set thresholds. The specific threshold determination and typing methods are as follows:

[0043] (1) nSolver4.0 software was used to obtain the absolute expression levels of HER2, ESR1, PGR and Ki-67 genes.

[0044] (2) Using machine learning methods, such as random forest or decision tree, the expression levels of HER2, ESR1, PGR, and Ki-67 genes in samples with clear clinical classification results are used to train classification models, and the TN type, ER+ type, and HER2+ type of these samples are predicted based on the model.

[0045] Example: According to the existing sample implementation examples, the typing results of the samples (TN type, ER+ type, HER2+ type) and the corresponding genes (HER2, PGR, ESR1, Ki-67) are obtained. As input, a machine learning method (random forest, decision tree, etc.) is used to build a classification model. For new samples to be tested, the expression level of the corresponding gene is input, and the existing classification model is input. The model will automatically predict the typing results.

[0046] (3) For patients predicted to be ER+ and HER2+, secondary classification was further performed. ER+ patients were further divided into luminal A and luminal B, and HER2+ patients were further divided into HER2-positive and luminal B-HER2-positive. In the secondary classification, we found that the CTSL2 gene can be better used for classification (compared to the clinically commonly used Ki67), with an accuracy of 92.55% ( Figure 3), specifically, the following steps: construct a ROC curve for classification based on the CTSL2 gene expression in the existing embodiment and the corresponding known clinical classification (such as luminal A type and luminal B type in the ER+ group), and select the cut-off value (median in the embodiment: 48.09) corresponding to the maximum value of the Youden index of the curve (representing the maximum sum of the sensitivity and specificity of using the CTSL2 gene expression to distinguish luminal A type and luminal B type) as the classification value for determining luminal A type and luminal B type according to the ROC curve. When CTSL2 is greater than the cut-off value, the sample is luminal B type, otherwise it is luminal A type ( Figure 1 ). Similarly, for the sample predicted to be HER2 positive, when the expression of CTSL2 gene is 47.335, the Youden index of the ROC curve of HER2+ classification is the largest, that is, when CTSL2 is greater than the cut-off value, it is predicted to be LB-HER2+ type, otherwise it is HER2+ type ( Figure 2 ). Therefore, the CTSL2 expression thresholds of 48.09 and 47.335 can be used to perform secondary typing for ER+ and HER2+ types, respectively.

[0047] Table 1 21 gene panel reporter probes and capture probes and their targeted RNA sequences

[0048]

[0049]

[0050]

[0051]

Claims

1. A method for molecular typing of breast cancer based on the expression of CTSL2 gene on nanoString platform, characterized in that: The specific steps are as follows: Step 1: Extract total RNA from breast cancer paraffin specimens using Qiagen FFPE RNA extraction kit; Step 2: Determine the concentration of total RNA and evaluate the quality of total RNA by the ratio of OD260 / 280 and OD260 / 230; Step 3: Use the probes dedicated to the nanoString platform to perform hybridization reaction with the target RNA sequence (i.e., Genelist is the source reference). The hybridization reaction system is as follows: S1, 70 μl of hybridization solution + one tube of reporter probe (total volume 42 μl) = Master mixture; S2, add 8 μl of Master mixture + 5 μl of RNA (150 ng, the insufficient part is supplemented with RNase-free H2O) to each PCR tube; S3, continue to add 2 μl of capture probe to each tube, with a total volume of 15 μl; S4, gently mix and incubate in a 65°C PCR instrument for 16-24 hours; Step 4: Add the hybridization reaction product in step 3 to the nanoString special clamp, seal the membrane and place it in the nanoString digital fluorescent barcode single molecule analyzer to perform gene expression detection according to the pre-set program; Step 5: nSolver4.0 software was used to quantitatively analyze the expression of HER2, ESR1, PGR, Ki-67 and CTSL2 genes, and molecular typing of each specimen was performed using machine learning and set thresholds; Among them, the molecular typing method according to the expression level of the gene in step 5 is as follows: S1. The expression levels of HER2, ESR1, PGR and Ki-67 genes were quantitatively analyzed using nSolver4.0 software; S2. Use machine learning methods, such as random forest or decision tree, to train the typing model for the expression of HER2, ESR1, PGR and Ki-67 genes in samples with clear clinical typing results, and predict the TN, ER+ and HER2+ subtypes of the specimens based on the model; S3. For patients predicted to be ER+ and HER2+ positive, further secondary classification is performed; ER+ positive patients can be further divided into luminal A type and luminal B type, and HER2+ positive patients can be further divided into HER2 type and luminalB-HER2 type.

2. The method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform according to claim 1, characterized in that: In step five, the specific steps of the secondary classification are as follows: a ROC curve of classification is constructed according to the CTSL2 gene expression level of the existing embodiment and the corresponding known clinical classification. The known clinical classifications include luminal A type and luminal B type in the ER+ group, and HER2 type and luminal B-HER2 type in the HER2+ group. According to the ROC curve, the cut-off value corresponding to the maximum value of the Youden index of the curve is selected as the classification value for determining luminal A type and luminal B type, HER2 type and luminal B-HER2 type, respectively. When CTSL2 is greater than the cut-off value, the sample is luminal B type or luminal B-HER2 type, otherwise it is luminal A type or HER2 type.

3. The method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform according to claim 2, characterized in that: The maximum value of the Youden index represents the maximum sum of the sensitivity and specificity of using the CTSL2 gene expression to distinguish between luminal A type and luminal B type, and between HER2 type and luminal B-HER2 type.

4. The method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform according to claim 1, characterized in that: In step 4, the entire operation takes 30 minutes.

5. The method for molecular typing of breast cancer based on the expression of CTSL2 gene on the nanoString platform according to claim 1, characterized in that: The manual operation in step 3 only takes 30 minutes, and 12 samples can be tested simultaneously at one time.