Method for detecting metastatic ability of tumor cells based on raman spectroscopy

Abstract

The application provides a tumor cell metastasis ability detection method based on Raman spectrum, relates to the technical field of medicine, and combines single-cell Raman technology and a machine learning method, quickly and accurately predicts the metastasis ability of tumor cells according to Raman signals of the tumor cells, can realize detection on cells without limitation, can detect cultured cells or tissue-derived cells, expands the use scene of detection, and the method is simple to operate, does not need to excessively process cells, can directly utilize a single-cell Raman instrument for detection, and only needs 5s for a single cell to complete detection, is more efficient, and the result is more accurate. Meanwhile, the method can realize identification at a single-cell level, is high in sensitivity, can predict the metastasis ability of tumor tissues that have not occurred metastasis in advance, intervenes in treatment of tumor with high metastasis potential in advance, and improves the cure rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical technology, in particular to a tumor cell metastasis ability detection method based on Raman spectroscopy. BACKGROUND

[0002] Tumor cells with high metastasis ability is one of the important signs of tumor malignancy, and is a major problem in clinical tumor research for patient treatment and prognosis. The metastasis ability of tumor cells is currently determined by several methods in the laboratory and clinic, including the following: 1) transwell-based comparison of the metastasis ability of cultured cells, including cell invasion and migration. This method requires culturing cells in a transwell chamber, and then comparing the number of cells that have crossed the bottom of the chamber. The main disadvantage of this method is that it is not suitable for tissue cells or other cells that cannot be cultured in vitro, and cell culture requires a lot of time, usually 2-3 days, so this method is also time-consuming; 2) determination based on the expression of proteins related to tumor metastasis reported in the literature. Past studies on tumor metastasis have found that tumor metastasis is often accompanied by the process of epithelial-mesenchymal transition (EMT). Current literature has reported that the expression of various EMT-related proteins is associated with tumor metastasis, but this method has the disadvantage that current research has shown that EMT is a complex process with different stages, and tumor cells at different stages have significantly different metastasis abilities. There is no significant correlation between the reported EMT-related proteins and the metastasis ability of tumor cells in the complex EMT process, so there is currently no reliable EMT-related protein that can represent the metastasis ability of tumor cells; 3) circulating tumor cells (CTC) in peripheral blood. Distant metastasis of tumor requires tumor cells to enter blood vessels, then circulate through blood to reach the target organ at a distance, then exit the blood vessels and enter the specific target organ. The part of tumor cells that enter the blood vessels is called CTC. The detection of CTC in blood represents that tumor has metastasized. The disadvantage of this method is that it requires a complex enrichment method to enrich and identify CTC from blood, which is costly and complex to operate. Secondly, the content of CTC in blood is usually low, resulting in insufficient sensitivity of CTC as a detection indicator and easy false negative results; 4) imaging-based diagnosis. This method is one of the gold standards for determining whether tumor has metastasized in clinic. When tumor tissue is observed in an organ through imaging, it is determined that metastasis has occurred. However, its disadvantage is that it cannot be used for early prediction and intervention treatment of tumor metastasis, because it can only be observed through imaging when tumor cells have colonized and developed into tumor tissue in the distant organ after metastasis.

[0003] Therefore, there is an urgent need for a new and more simple and accurate detection method to accurately predict the metastatic ability of tumors.

[0004] Therefore, the present application is proposed. SUMMARY

[0005] One of the purposes of the present application is to provide a tumor cell metastasis detection method based on Raman spectroscopy to at least solve one of the technical problems existing in the prior art.

[0006] In order to achieve the above-mentioned purposes of the present application, the following technical solutions are adopted:

[0007] Raman spectroscopy is a non-labeled scattering spectrum, which is a phenomenon that the frequency of the scattered light changes after the incident light is scattered due to the fact that the molecular bond is excited to a virtual state but has not yet returned to the original state. Different molecules can produce specific Raman scattering spectra, so they can be used as a standard for distinguishing different substances. Initially, Raman spectroscopy was mainly used to analyze the composition of chemical substances, but with the development of technology, Raman spectroscopy technology has been gradually applied to the identification of biological materials such as microorganisms, blood, cells, etc. The Raman spectrum of each single cell is a Raman spectrum peak set (also known as Ramanome) composed of more than 1500 Raman spectrum peaks corresponding to different types of chemical bonds, and these signals contain multi-dimensional information such as the composition and content of all substances in the cell, including DNA, RNA, protein, lipid, etc., which can reflect the current life characteristics of the cell. Since Raman spectroscopy technology can be used to detect at the single cell level, and the detection time of the cell is only about 1 second, it has been gradually applied to the biological field. In the field of tumor research, there have been reports that different types of tumor cells, tumor cells before and after drug induction, etc. can be distinguished based on tumor Raman spectroscopy, and tumor tissues can be directly detected by Raman spectroscopy to distinguish different types of tumors. These existing studies have shown that whether it is for different types of tumor cells or for the same type of cells that have undergone EMT under the induction of drugs, Raman spectroscopy can effectively distinguish them. In particular, in skin cancer, tumor tissues in the epidermis and tumor tissues that have invaded the dermis can also be distinguished by Raman spectroscopy, but there have been few reports on the correlation between the metastatic ability of tumor cells and their Raman signals. Based on this, in a first aspect, the present application provides a tumor cell metastasis detection method based on Raman spectroscopy, comprising the following steps:

[0008] a) Single-cell Raman spectroscopy detection of tumor cell lines with different metastatic abilities;

[0009] b) Using a machine learning algorithm, a classification model based on single-cell Raman signal data is constructed to distinguish different cell lines with different metastatic abilities;

[0010] c) screening out Raman signal peaks with high classification weight from the classification model by using Gini algorithm;

[0011] d) calculating a Raman metastasis index (RMI) according to the correlation of the Raman signal peaks with the metastasis ability of tumor cells and the Raman signal intensity thereof.

[0012] Compared with the transwell method, the detection method provided by the present application can be applied to tissue cells and other cells that cannot be cultured in vitro, and each cell detection only takes a few seconds, which is more time-saving and labor-saving. Since the detection is carried out at the single cell level, the amount of cells required is less, and the heterogeneity of tumors can also be reflected. Compared with CTC cell detection, the detection method provided by the present application does not require excessive pretreatment steps, and the in situ tumor cells are detected, which is easier to obtain samples. The content of CTC in blood is extremely low, and a relatively complex enrichment process is required. Compared with the gold standard imaging, the detection method provided by the present application can detect the in situ tumor to determine whether it has a high risk of metastasis before a distant metastatic lesion is formed, so as to intervene in advance.

[0013] Further, in step a), the tumor cell lines with different metastasis abilities include ASPC1 cell line with high metastasis ability, PANC1 and BXPC3 cell lines with medium metastasis ability, and MIAPACA2 cell line with low metastasis ability.

[0014] It can be understood that the present application does not limit the specific selection of tumor cell lines with different metastasis abilities, and other known tumor cells with different metastasis abilities can also be selected as standards.

[0015] The above four tumor cells with verified tumor metastasis ability are used as standards, and the Raman metastasis index (RMI) obtained is more accurate.

[0016] Further, in step a), the laser power for Raman spectrum detection is 100 mW, the wavelength is 532 nm, and the acquisition time is 1 second.

[0017] Further, in step b), the machine learning algorithm includes a random forest algorithm.

[0018] It can be understood that the present application does not limit the specific type of machine learning algorithm, and other algorithms except random forest can also be used for modeling, such as PC-LDA (linear principal component analysis) or SVM (support vector machine).

[0019] Further, in step c), the screening condition of using Gini algorithm is that the Gini score is greater than 0.75.

[0020] The higher the Gini score, the higher the corresponding Raman characteristic peak plays a role in distinguishing tumor metastasis ability, that is, the stronger the correlation between the Raman characteristic peak and tumor metastasis. Therefore, the higher the Gini score, the more accurate the calculation of RMI by the Raman peak. However, the higher the Gini score, the fewer the number of Raman characteristic peaks. In order to consider the accuracy of RMI calculation by Raman peaks and prevent overfitting, the diversity of Raman characteristic peaks should be considered, thereby reducing the application range. Therefore, after multiple attempts, 0.75 is the most appropriate screening condition.

[0021] Further, in step c), the Raman signal peaks with high classification weight include 1465 cm -1 , 1129 cm -1 , 1435 cm -1 , 854 cm -1 , 1433 cm -1 , 1699 cm -1 , 1078 cm -1 , 1302 cm -1 , 1449 cm -1 , 1461 cm -1 , 1559 cm -1 , 1352 cm -1 , 1065 cm -1 , 1668 cm -1 , and 1799 cm -1 .

[0022] In addition to the above-mentioned Raman signal peaks, methods for calculating RMI by selecting Raman signal peaks other than the 15 Raman signal peaks, or containing part of the 15 Raman peaks, and adding other new Raman peaks can also be used.

[0023] Further, the tumor cells include pancreatic cancer cells, cervical cancer cells, and choroidal melanoma cells.

[0024] Further, the tumor cells include natural cells, genetically edited cells, and drug-treated cells.

[0025] The inventors verified the effect of RMI through in vivo and in vitro experiments, and proved that RMI can effectively characterize the metastasis ability of in vivo and in vitro cells, and different types of tumor cell lines, including pancreatic cancer cell lines (ASPC1, P1E6, P1G8, P2D6, P2G8), cervical cancer cell lines (Siha) and choroidal melanoma cell lines (MUM2B). In addition, the method also found that, whether the tumor metastasis ability is different through gene editing methods, such as ASPC1-KO and-NC, Siha-OE and-NC, or the metastasis ability is different through drug treatment, such as MUM2B-ART and-NC (ART represents artemether treatment), and natural cells without any treatment (P1G8, P1E6, P2G8, P2D6), the metastasis ability can be accurately indicated, which shows that the method is not affected by the type of tumor and has better practicability.

[0026] Further, the Raman transfer index RMI is the abbreviation of Ramanome-based metastasis index, that is, the coefficient calculated based on the single cell Raman signal for characterizing the metastasis ability of tumor cells, and the calculation method is as follows: wherein w is the real Raman signal of each Raman peak in the cell, I is the correlation between each Raman peak and the metastasis ability of tumor, and the I value is 1 for positive correlation and-1 for negative correlation.

[0027] .

[0028] Further, only simple and rapid Raman signal detection of cultured cells is needed through a single cell Raman instrument, and then the RMI value can be calculated based on the signal intensity of 15 Raman characteristic peaks and the correlation between the signal intensity and the metastasis of tumor, and the size of the RMI value represents the size of the metastasis ability of tumor cells. The greater the RMI value, the stronger the metastasis ability of tumor cells, so that the metastasis ability of tumor can be quickly and accurately predicted.

[0029] Compared with the prior art, the present application has the following beneficial effects:

[0030] The present application combines single-cell Raman technology and machine learning method, and quickly and accurately predicts the metastasis ability of tumor according to the Raman signal of tumor cells, which has the following advantages: 1) Compared with transwell, the present method has no restriction on cell source, and can detect cultured cells or tissue-derived cells, thus expanding the use scenario of detection, and the present method is simple to operate and does not require excessive processing of cells, and can be directly detected by a single-cell Raman instrument, and only 5s is needed for a single cell to complete the detection, which is more efficient; 2) Compared with EMT-related proteins, the present method analyzes the overall Raman signal of cells, rather than simply several proteins, so that the metastasis ability of cells can be more accurately reflected, and the result is also more accurate; 3) Compared with CTC, the present method can be identified at the single-cell level, so the sensitivity is higher than that of CTC, and the present method can directly detect tumor cells in situ tissue, and the metastasis ability of tumor is quantified through the calculated RMI value, so that the metastasis of tumor can be predicted earlier; 4) Compared with imaging methods, the present method can predict the metastasis ability of tumor tissue before metastasis, and intervene in the treatment of tumor with high metastasis potential in advance, so as to improve the cure rate. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the specific embodiments of the present application or the prior art, the drawings needed in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0032] Figure 1Accuracy verification of RMI of different tumor cell types provided by the embodiments of the present application. Among them, A) Comparison of RMI of ASPC1-hnRNPAB-NC and ASPC1-hnRNPAB-KO cells with cell invasion and migration assays based on transwell experiments. The "Raman" histogram represents the RMI calculated according to the Raman spectrum, and the y-axis on the right represents the RMI value. B) Comparison of RMI between four single clone cells isolated from PANC-1 cells. C) Comparison of RMI between MUM2B cells with and without ART treatment and between SiHa cells with and without overexpression of S100A7 gene. D) Comparison of RMI in ASPC1-NC-RFP and ASPC1-KO-RFP cells with cell invasion and migration assays. ns represents p ≥ 0.05, * represents p < 0.05, ** represents p < 0.01, and *** represents p < 0.001. The column and error bar represent SD ± mean. N represents the number of cells, and n represents the number of biological replicates.

[0033] Figure 2 Experimental results of the metastatic ability of the naturally screened cell strains P1G8, P2G8, P1E6 and P2D6 provided by the embodiments of the present application;

[0034] Figure 3 Accuracy verification of RMI in pancreatic tumor tissues provided by the embodiments of the present application. Among them, A) Schematic diagram of the process of establishing a PDAC mouse orthotopic model, in which ASPC1-NC-RFP and ASPC1-KO-RFP cells are inoculated into the pancreas of mice, respectively. B) Dissecting pancreatic tumor tissues from mice injected with ASPC-NC-RFP cells and ASPC1-KO-RFP cells. The light white tissue on the right represents the pancreas containing tumor cells, and the dark red tissue on the left represents the spleen. C) Comparison of RMI between mouse pancreatic tumor cells derived from injection of ASPC1-NC-RFP cells and ASPC1-KO-RFP cells. D) H&E staining of liver tissues obtained from mice injected with ASPC1-NC-RFP cells and ASPC1-KO-RFP cells, which is used as a standard for whether pancreatic cancer liver metastasis occurs. The proportion of mice with liver metastasis is calculated by dividing the number of mice with liver metastasis by the total number of mice in the corresponding group. DETAILED DESCRIPTION

[0035] Unless otherwise defined, scientific and technical terms used in connection with the present application shall have the meanings that are commonly understood by those of ordinary skill in the art. If not apparent from the context, the meaning of a term will be explained. The meaning and scope of the term should be clear; however, in the event of any latent ambiguity, the above interpretation should be preferred to any dictionary or extrinsic definition.

[0036] In this application, the use of "or" means "and / or" unless specifically stated otherwise. Moreover, the use of the term "including" as well as other forms, such as "include", is intended to be non-limiting.

[0037] Generally, the nomenclature used in connection with, and the techniques of, cell and tissue culture, molecular biology, immunology, microbiology, genetics and protein and nucleic acid chemistry and hybridization described herein are those well-known and commonly used in the art. The methods and techniques of the present application are generally performed according to conventional methods well-known in the art and as described in various general and more specific references that are cited throughout the present specification unless otherwise indicated. Enzymatic reactions and purification techniques are performed according to the manufacturer's instructions, as commonly accomplished in the art or as described herein. The nomenclature used in connection with, and the techniques of, analytical chemistry, synthetic organic chemistry, and medicinal and pharmaceutical chemistry described herein are those well-known and commonly used in the art.

[0038] The technical solutions of the present application will be described clearly and completely in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application. Unless otherwise specified, the materials in the embodiments are prepared according to the existing methods or directly purchased from the market.

[0039] Example 1 Preparation of cell suspension

[0040] 1. Preparation of single cell suspension of in vitro cultured cells:

[0041] Commercial reagents commonly used for in vitro cultured cells: DMEM medium, 1640 medium, fetal bovine serum FBS, penicillin-streptomycin-amphotericin B antibiotic solution, PBS buffer, trypsin solution;

[0042] Commercial reagents that need to be prepared additionally in this embodiment: 4% tissue fixative;

[0043] Specific experimental steps:

[0044] a) The cells are cultured in vitro in a suitable culture medium, and the cells are generally collected after 24-48 hours of culture. For adherent cells, after washing the residual culture medium solution with PBS buffer, an appropriate amount of trypsin is added for digestion (reaction in a 37°C incubator for 1 min), so that the adherent cells are released from the bottom of the culture dish, and then the cells are collected in a 1.5 mL EP tube. The trypsin and the participating culture medium solution are washed away with PBS buffer again, and the supernatant solution is discarded after centrifugation; for suspension cells, the trypsin digestion process can be omitted, and the cells are collected in a 1.5 mL EP tube, and the participating culture medium solution is washed away with PBS, and the supernatant solution is discarded after centrifugation;

[0045] b) 1 mL of 4% tissue fixative is added to the EP tube containing the cells, and after reaction at room temperature for 15 min, the supernatant is discarded after centrifugation at 1000 rpm for 5 min, and then 1 mL of PBS buffer is added to resuspend the cells, and the supernatant is discarded after centrifugation at 1000 rpm for 5 min. After ensuring that the residual fixative is washed away, the cells are resuspended in 200 μL of PBS solution, and thus the single-cell suspension is prepared.

[0046] 2. Preparation of single-cell suspension of tumor tissue:

[0047] Required commercial reagents: PBS buffer, collagenase IV solution (2 mg / mL), 100 μm cell filter, 4% tissue fixative;

[0048] Specific experimental steps: The tumor tissue is cut into pieces with scissors and placed in a 1.5 mL EP tube. The residual blood is washed away with 1 mL of PBS, and then the supernatant is discarded after centrifugation at 1000 rpm for 10 min. Then 1 mL of collagenase IV solution is added to resuspend the cells, and the reaction is carried out at 37°C for 30 min. After the reaction is completed, the suspension is filtered with a 100 μm cell filter to remove the undigested tissue pieces into single cells. Then 4% tissue fixative is added and reacted at room temperature for 15 min. The supernatant is discarded after centrifugation at room temperature at 1000 rpm for 5 min. The cells are resuspended with 1 mL of PBS buffer, and the supernatant is discarded again after centrifugation at room temperature at 1000 rpm for 5 min. After ensuring that the residual fixative is washed away, the cells are finally resuspended in 200 μL of PBS solution to obtain the prepared single-cell suspension of the tissue.

[0049] Example 2 Single-cell Raman identification

[0050] Commercial instrument: Horiba LabRam HR Raman spectrometer, 532 nm Nd:YAG laser, 60x / 100NA water immersion objective lens.

[0051] Detection method: 100 μL single cell cell suspension was dropped on CaF2

[0052] material slide, then placed under a microscope equipped with a 60x water lens, slowly adjust the distance between the water lens and the slide, when the cells on the slide can be clearly seen through the microscope, select different positions of the cells to collect Raman signals through the software: each cell is selected with a wavelength of 532 nm laser 5 points for data collection, respectively, the upper, lower, left, right, and middle five positions of the cell, the power is 100 mW, the collection time is 1 second, at least 25 cells are collected for each sample.

[0053] Example 3: Construction of a prediction model based on single cell Raman technology and tumor metastasis ability

[0054] According to the existing literature reports, four cell lines with significant differences in metastatic ability have been selected, namely high metastatic ability ASPC1 cell line, medium metastatic ability PANC1 and BXPC3 cell lines, and low metastatic ability MIAPACA2 cell line. The cell culture and single cell suspension preparation method is the same as that of Example 1, and the cell Raman signal collection method is the same as that of Example 2. After multiple single cell Raman detection, the Random Forest algorithm is used to construct a classification model based on single cell Raman signals for distinguishing different cell lines with high, medium and low metastatic ability, wherein 70% of the data is used to construct the model, and 30% is used to verify the classification accuracy of the model.

[0055] Example 4: Construction of Raman index RMI (Ramanome-based metastasis index) for indicating tumor cell metastasis ability

[0056] After successfully constructing the classification model that can be used to distinguish different metastatic abilities of tumors in Example 3, the Gini algorithm is used to screen out the Raman signal peaks with high classification weight from the model. Each Raman signal peak has a corresponding Gini score, and the higher the Gini score, the greater the weight of the Raman peak in the classification model. The method provided in this embodiment is tested, and finally 15 Raman peaks with Gini scores greater than 0.75 (1465, 1129, 1435, 854, 1433, 1699, 1078, 1302, 1449, 1461, 1559, 1352, 1065, 1668, 1799 cm -1), according to the correlation between the 15 Raman peaks and the tumor metastasis ability and the Raman signal intensity itself, a cell metastasis index RMI based on Raman signal is constructed. The calculation method is as follows: wherein w is the true Raman signal of each Raman peak in the cell, I is the correlation between each Raman peak and the tumor metastasis ability, and I value is 1 for positive correlation and -1 for negative correlation.

[0057]

[0058] Example 5 Verification of the effect of RMI indicating tumor metastasis ability

[0059] Verification at the level of in vitro cultured cells: this example uses 12 cell lines that have been screened, genetically edited or treated with drugs, and whose metastasis ability has been verified by transwell experiments or reported in the literature, and whose metastasis ability has significant differences, respectively Figure 1 The cells described in the above. Six genetically edited cells, including ASPC1-hnRNPAB-NC and -KO, whose metastasis results are shown in Figure 1 a; Siha-S100A7-NC and -OE, whose metastasis results are shown in the literature: Tian Tian, Xukun Li, Zhen Hua, Jianlin Ma, Xiaowei Wu, Zhihua Liu, Hongyan Chen, Zhumei Cui, Oncotarget 2017, 8, 24964-24977; and ASPC1-NC-RFP and -KO-RFP, whose metastasis results are shown in Figure 1 d. Four naturally screened cells, P1G8, P2G8, P1E6 and P2D6, whose metastasis ability experimental screenshots are shown in Figure 2 Two drug-treated cells, MUM2B-NC and MUM2B-ART, whose effects of ART on tumor cell metastasis ability are shown in the literature: S. A. K. Rasheed, T. Efferth, I. A. Asangani, H. Allgayer, International Journal of Cancer 2010, 127, 1475-1485. Single-cell Raman data acquisition was performed on each of them, and then their corresponding RMI values were calculated, proving that the cell lines with high metastasis ability have significantly higher RMI values than the cell lines with low metastasis ability, as shown in Figure 1 .

[0060] Figure 1 ASPC1-hnRNPAB-KO in Fig. A, Figure 1 Siha-S100A7-OE in Fig. C, and Figure 1The ASPC1-KO-RFP in the middle D is a cell in which the corresponding gene is knocked out (KO) or the corresponding gene is overexpressed (OE) by gene editing technology. Figure 1 The P1G8, P2G8, P1E6, and P2D6 in the middle B are tumor cells naturally grown without any gene editing and drug treatment. Figure 1 The MUM2B-ART in the middle C is an experimental group in which the MUM2B cells are treated with artesunate (ART). Figure 1 The results show that, regardless of the cell type, the tumor metastasis ability predicted by the RMI calculated based on the single-cell Raman technology is consistent with the transwell results or the tumor metastasis ability reported in the literature.

[0061] In addition, Figure 1 The MUM2B and Siha in the middle C are choroidal melanoma cells and cervical cancer cells, respectively. Artesunate (ART in the figure) has been reported in the literature to reduce the metastasis ability of tumor cells, and the overexpression of the S100A7 gene (S100A7-OE in the figure) has been reported to promote the metastasis ability of Siha, and the results of the middle C show that the RMI calculated based on the single-cell Raman technology is consistent with the literature in predicting the tumor metastasis ability of MUM2B and Siha under the conditions of ART treatment and S100A7 overexpression, thus indicating that RMI can be applied to tumor cell types other than pancreatic cancer cells. Figure 1 The results of the middle C show that the RMI calculated based on the single-cell Raman technology is consistent with the literature in predicting the tumor metastasis ability of MUM2B and Siha under the conditions of ART treatment and S100A7 overexpression, thus indicating that RMI can be applied to tumor cell types other than pancreatic cancer cells.

[0062] Verification at the tumor tissue cell level: After successful verification in vitro cultured cells, the mouse pancreatic cancer orthotopic tumor liver metastasis model was constructed in this embodiment. The ASPC1-NC-RFP cells and the ASPC1-KO-RFP cells were injected into the pancreases of 7 and 8 mice, respectively. After the mice continued to grow for 6 weeks, the mice were sacrificed, and the pancreatic cancer tissues grown in the pancreases of the mice were taken. The single-cell suspension was prepared using the method in Example 2, and then the single-cell Raman detection was performed to calculate the RMI values of the two groups. Meanwhile, the liver tissues of each mouse were collected, and HE staining was performed to determine whether liver metastasis occurred, as a positive control. The results prove that the RMI values are consistent with the proportion of liver metastasis in the two groups, as shown in Figure 3

[0063] The above data prove that the metastasis ability index RMI based on Raman in this method can quickly and accurately predict the metastasis ability of in vitro cultured tumor cells and in vivo grown tumor tissue cells.

[0064] ​It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting the metastatic potential of tumor cells based on Raman spectroscopy, characterized in that, Includes the following steps: a) Single-cell Raman spectroscopy detection of tumor cell lines with different metastatic abilities; b) Using machine learning algorithms, construct a classification model based on single-cell Raman signal data that can be used to distinguish different cell lines with different metastatic abilities; c) Use the Gini algorithm to select Raman signal peaks that contribute significantly to the discrimination from the classification model; d) The Raman metastasis index (RMI) is calculated based on the correlation between the Raman signal peak and the metastatic ability of tumor cells and the Raman signal intensity. The Raman metastasis index (RMI) is calculated using the following formula (1), where w is the actual Raman signal of each Raman peak in the cell, and I is the correlation between each Raman peak and the tumor metastasis ability. If there is a positive correlation, the value of I is 1, and if there is a negative correlation, the value of I is -1. ； The higher the RMI value, the stronger the metastatic ability of tumor cells.

2. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, In step a), the tumor cell lines with different metastatic abilities include the ASPC1 cell line with high metastatic ability, the PANC1 and BXPC3 cell lines with medium metastatic ability, and the MIAPACA2 cell line with low metastatic ability.

3. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, In step a), the laser power for Raman spectroscopy detection is 100mW, the wavelength is 532 nm, and the acquisition time is 1 second.

4. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, In step b), the machine learning algorithm mentioned includes the random forest algorithm.

5. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, In step c), the screening criterion using the Gini algorithm is a Gini score greater than 0.

75.

6. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 5, characterized in that, In step c), the Raman signal peaks that contribute significantly to the discrimination include those at 1465 cm⁻¹. -1 1129 cm -1 1435 cm -1 854 cm -1 1433 cm -1 1699 cm -1 1078 cm -1 1302 cm -1 1449 cm -1 1461 cm -1 1559 cm -1 1352 cm -1 1065 cm -1 1668 cm -1 and 1799 cm -1 .

7. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, The tumor cells include pancreatic cancer cells, cervical cancer cells, and choroidal melanoma cells.

8. The method for detecting tumor cell metastasis based on Raman spectroscopy according to claim 1, characterized in that, The tumor cells include natural cells, gene-edited cells, and drug-treated cells.

Images