Method for intraoperative rapid pathological diagnosis and drug sensitivity evaluation

Abstract

The invention belongs to the technical field of medical information, and relates to a method for intraoperative rapid pathological diagnosis and drug sensitivity evaluation. An imaging nanoprobe (A) with a lipid targeting function is used for dyeing a frozen section (B) of a tumor tissue sample cut off in an operation or dyeing the frozen section of the tumor tissue subjected to drug action, and then fluorescence imaging data are obtained under fluorescence microscopic imaging equipment; through correlation analysis of lipid substances of imaging data, an image processing analysis algorithm related to lipid metabolism capability is established, and definition of tumor site boundaries and evaluation of treatment effectiveness of medicine acting on tumors are realized. The method does not depend on high-end and expensive microscope equipment, and can be used for rapid pathological diagnosis in an operation, tumor classification and effect evaluation after drug treatment. The dyeing process is free of washing, the operation is simple, and the method is an intraoperative pathology and drug sensitivity evaluation method aiming at lipid chemical imaging analysis.

Description

Technical Field

[0001] The present invention belongs to the field of medical information technology, and in particular relates to a method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation. Background Art

[0002] Tumors are closely related to abnormal lipid metabolism. During the malignant transformation process, tumor cells have the malignant characteristics of unlimited proliferation, invasion, metastasis and loss of control. Lipid metabolism is highly reprogrammed in cancer. In human cancers, fat synthesis is significantly upregulated to meet the greater needs of membrane biosynthesis. Changes in lipid composition and abundance are hallmarks of cancer invasiveness. The absorption, synthesis, transport and metabolism of lipids have a profound impact on the pathogenesis of tumors. In particular, the close relationship between lipid metabolism and programmed cell death in tumors provides new insights into the diagnosis and treatment of tumors. The present invention is a novel method for rapid pathological differentiation and drug resistance evaluation based on abnormal lipid metabolism. In view of the correlation between tumors and abnormal lipid metabolism, a fluorescent nanoprobe is used to develop and image lipids in tumors, determine the boundaries of tumor tissues, and evaluate the drug sensitivity of lipid changes after drug action.

[0003] Rapid intraoperative pathological diagnosis refers to the surgeon removing part of the patient's lesion tissue during the operation, fixing it and immediately sending it to the pathology department for rapid frozen section staining, observing it under a microscope by a pathologist, and informing the surgeon of the results so that the surgeon can judge the pathological report of the resection range. Existing rapid intraoperative pathological diagnosis is mainly through tissue biopsy or surgical resection. A small piece of lesion tissue is taken and stained with traditional hematoxylin-eosin staining (H&E). The presence of cancer cells and the degree of differentiation of cancer cells are diagnosed under a microscope. This step is the standard for diagnosing the grade of cancer. However, H&E staining involves multiple steps, and the process includes: tissue fixation, dehydration, embedding, slicing, baking, dewaxing, hematoxylin staining, differentiation, anti-blue, eosin staining, dehydration and sealing, a total of 11 steps. The operator needs to have rich experience and patience, precise operation and time control to ensure the quality and effect of staining. The process is cumbersome and time-consuming, and takes more than 40 minutes. Long waiting poses certain risks to patients during surgery. The present invention develops an intraoperative pathological diagnosis method that can quickly and accurately distinguish tumor cells from normal tissues, and can determine the stage of the tumor, which is of great significance for intraoperative decision-making, improving patient prognosis, and avoiding reoperation.

[0004] Drug sensitivity evaluation can provide help for clinicians and is of great significance for the implementation of individualized treatment. By evaluating the drug resistance of tumors, doctors can more accurately understand the tumor's response to specific treatment plans, adjust treatment strategies, formulate personalized treatment plans, improve treatment effects, avoid ineffective treatments, reduce side effects and economic burdens, and prolong patient survival. Existing drug sensitivity evaluations mainly judge drug sensitivity by observing changes in clinical symptoms of patients before and after chemotherapy. For example, changes in tumor marker levels in the patient's blood can be detected. However, the fluctuation of tumor marker levels is affected by many factors. Changes in tumor size can also be detected by imaging methods such as B-ultrasound and CT; sampling for histopathological examinations and observing morphological changes and apoptosis of tumor cells to evaluate drug sensitivity. These morphological evaluations are relatively subjective and have a large number of human factors. The present invention develops a rapid drug sensitivity evaluation method associated with the tumor lipid metabolism mechanism to provide an effective means for drug sensitivity evaluation.

[0005] Three patents closely related to the present invention were found:

[0006] (1) "A method for preparing and using a wash-free fluorescent imaging nanoprobe targeting lipid droplets", patent number 202210141414.0, application date: February 16, 2022. It relates to a method for preparing and using a nanoprobe that specifically targets intracellular lipid droplets. At the same time, the probe has two-photon fluorescence characteristics, and can achieve high-quality imaging of lipid droplets in living cells without washing under 780nm picosecond laser excitation; and has the advantages of high quantum yield, good stability, excellent biocompatibility, and simple operation. Its advantage is that it can be washed without washing and has a higher imaging contrast compared to the background signal.

[0007] The present invention is significantly different from the ZL202210141414.0 patent, and their usage scenarios and innovations are different. The probe involved in the ZL202210141414.0 patent is used for cell imaging. The probe involved in the present invention is used for the use of frozen sections of tissue pathology. The ZL202210141414.0 patent is for imaging special organelles in cells - lipid droplets. This patent involves a method for intraoperative pathological diagnosis that quickly and accurately distinguishes tumor cells from normal tissues. It is also a rapid method for evaluating drug sensitivity associated with the lipid metabolism mechanism of tumors. It involves a new clinically usable diagnostic and evaluation basis, which is based on the analysis results related to abnormal lipid metabolism.

[0008] (2) “A system and method for rapid intraoperative pathological diagnosis” Patent No.: 201510967267.2 Application date: December 22, 2015. This patent obtains the functional map and optical section map of the diseased sample tissue through three major systems. The data analysis compares and analyzes the functional map and optical section map with the existing disease database data to generate a diagnosis result report.

[0009] The present invention is significantly different from Patent 201510967267.2. Patent 201510967267.2 uses the method of obtaining the diagnosis result by comparing the similarity of morphological information with the known database; the present invention obtains the diagnosis result by distinguishing the chemical imaging characteristic information of lipids.

[0010] (3) "System and method for rapid intraoperative remote pathology diagnosis" Patent No.: 201810693334.X Application date: June 29, 2018. This patent claims that remote pathology experts do not need to travel to grassroots hospitals; as long as there is an Internet connection, remote experts can access digital pathology images at any time and make pathology diagnoses in a timely manner, greatly improving the efficiency of intraoperative pathology diagnosis.

[0011] The present invention is significantly different from 201810693334.X. The 201810693334.X invention requires pathology professionals with rich knowledge reserves to perform pathological diagnosis. The present invention does not require professionals and can provide a diagnosis conclusion through digital discrimination of chemical imaging information. Summary of the invention

[0012] The present invention relates to a method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation, so as to solve the problems that the methods provided by the prior art are cumbersome in film preparation and require diagnosis by pathologists with rich experience.

[0013] To achieve the above purpose, the present invention adopts the following technical solutions:

[0014] (1) Weigh 40 mg of N,N-diethyl-p-phenylenediamine and 10 mL of anhydrous ethanol to form a mixed solution, put it into a 20 mL reactor and fully react it in an oven at 200°C under sealed conditions for 12 hours to obtain a brown solution. The obtained crude product is purified by silica gel chromatography, and dichloromethane and methanol are used as eluents in the process to obtain a purified red fluorescent component, namely, the imaging nanoprobe (A). The purified probe is rotary evaporated and stored in a dimethyl sulfoxide solution to become the imaging nanoprobe (A) stock solution for standby use.

[0015] (2) extracting diseased sample tissue from the patient's body, fixing the diseased sample tissue using a freezing microtome technique, and preparing frozen section samples with a thickness of 5-20 microns.

[0016] (3) The frozen sections of the samples in step (2) were immersed in a 0.01 M phosphate buffer solution (C) with a pH of 7.2-7.4 containing 50-100 μg / mL imaging nanoprobe (A) and incubated at 37° C. for 5 min. After washing three times with the phosphate buffer solution (C), further fluorescence imaging was performed.

[0017] (4) Performing microscopic fluorescence imaging on the stained sample tissue slices obtained in step (3) to obtain a fluorescence image. (5) Performing data analysis on the fluorescence image obtained in step (4), thereby obtaining a rapid intraoperative pathological diagnosis and drug sensitivity evaluation. For the pathological tissue slices obtained during the operation, fluorescence image data is collected after staining, and the metabolic correlation characteristics of lipid substances are obtained through image analysis; and the tumor site or normal tissue site is determined based on the characteristic discrimination basis. For the tumor biopsy tissue samples of patients after medication or the tissue sample slices obtained during surgery, fluorescence image data is collected after staining, and the metabolic correlation characteristics of lipid substances are obtained through image analysis; and the drug resistance or drug inresistance of the anti-tumor drug used is determined based on the characteristic discrimination basis.

[0018] The present invention provides a fluorescent probe that can be used for rapid pathological diagnosis during surgery and a method for evaluating drug sensitivity.

[0019] The significant advantages and features of the present invention are:

[0020] (1) The raw materials of the probe synthesized by the present invention are low-cost and easy to obtain, the synthesis method of the probe is simple, the purification operation is also very convenient, and the overall cost is low. In addition, the probe does not need to be subsequently modified or rely on expensive and complex equipment, so it is very easy to achieve mass production.

[0021] (2) The staining method of the present invention is rapid for staining clinical pathological frozen sections, does not require washing, does not require other operations, and has simple operating steps.

[0022] (3) The stained pathological sections of the present invention only require the most common fluorescence microscope for imaging, which is low-cost and easy to obtain, and can meet the needs of rapid pathological diagnosis and drug sensitivity evaluation in clinical practice.

[0023] (4) The present invention is an analytical method established for the correlation between tumors and abnormal lipid metabolism. Different from the existing clinical pathological diagnosis technology based on nuclear morphology analysis, it adopts a new mechanism for evaluating tumor tissue boundaries and drug sensitivity based on chemical information association. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a schematic diagram of the synthesis of the fluorescent probe prepared by the present invention;

[0025] Figure 2A transmission electron microscope image of the probe prepared by the present invention;

[0026] Figure 3 The fluorescence spectra of the probe prepared by the present invention at different excitation wavelengths;

[0027] Figure 4 This is a fluorescence photo / spectral difference comparison diagram of the probe prepared by the present invention in n-octanol and water;

[0028] Figure 5 This is a fluorescence imaging diagram of the carbon dots prepared in the present invention staining sections of breast cancer at different stages, with a scale of 50 μm.

[0029] Figure 6 This is the result of the first classification method, which is the combination of pixel distribution map and machine learning algorithm to classify fluorescence imaging.

[0030] Figure 7 This is the result of the second classification method, namely, the convolutional neural network classification of fluorescence imaging. DETAILED DESCRIPTION

[0031] The present invention is further described in conjunction with the embodiments. The described embodiments are only a part of the embodiments of the present invention, and cannot include all the embodiments. The purpose of combining the embodiments is to explain the present invention in detail and systematically, rather than to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0032] Example 1. Preparation and characterization of probes

[0033] Weigh 40 mg of N,N-diethyl-p-phenylenediamine, add it to 10 mL of anhydrous ethanol, and sonicate for 10 minutes to mix it evenly. Then transfer the liquid into a 20 mL stainless steel reactor lined with polytetrafluoroethylene, tighten the lid, place the reactor in an oven that has been heated to 200°C, and take out the reactor after keeping it warm for 12 hours. Allow the reactor to cool naturally to room temperature to obtain a brown crude product, which is then purified using a silica gel column. During the process, dichloromethane and methanol are used as eluents, and the solvent polarity is continuously adjusted to obtain red light emitting carbon dots. The target product is rotary evaporated to obtain a purified imaging nanoprobe (A). The preparation process of the probe and the results of probe purification are shown in Figure 2. Figure 1 shown. Figure 2 The transmission electron microscope photo of the probe is shown, and its particle size is about 2-6 nanometers. The imaging nanoprobe (A) has good dispersion and is in a uniform spherical shape. Figure 3 Figure 2 is the fluorescence spectrum of the probe at different excitation wavelengths, which proves that the fluorescence emission peak of the imaging nanoprobe (A) is located at 630 nanometers. Figure 4The fluorescence spectra of the probe in n-octanol and water are compared. The probe was dissolved in n-octanol to simulate the intracellular lipid environment to detect its lipid response properties. Under ultraviolet light, there is a significant fluorescence change between the two solvents, indicating that the imaging nanoprobe (A) exhibits strong fluorescence in the lipid environment and confirms its lipid targeting ability.

[0034] Example 2: Fluorescence imaging of breast cancer sections stained with probes

[0035] Intraoperative breast cancer tissue was made into frozen sections with a thickness of about 15 μm. The frozen sections were immersed in a 0.01 M phosphate buffer solution (C) with a pH of 7.2-7.4 containing 100 μg / mL imaging nanoprobe (A) and incubated at 37°C for 10 min. After washing three times with (C), the frozen sections were further subjected to fluorescence imaging (Olympus, Japan). A 10× objective was used, and the z-scan mode was set to 0.5 μm with an interval of 67 times. The emission channel was 560-750 nm. Figure 5 Comparison of the same position of H&E staining and fluorescent staining imaging of breast cancer tissue sections. In normal breast tissue sections, a small amount of fluorescence emission was detected in the red channel, indicating low fluorescence activity. The fluorescence intensity of pathological stage I breast cancer tissue sections after staining was slight, while the fluorescence signal of stage II breast cancer tissue sections gradually increased. The strongest fluorescence was observed in stage III breast cancer tissue sections. NT is normal breast tissue. This shows that the imaging nanoprobe (A) can effectively distinguish normal breast tissue (NT) and intraoperative pathological frozen tissue sections of breast cancer of different cancer stages (stages I, II, and III).

[0036] Example 3: Imaging data processing and judgment basis

[0037] Two classification methods are used to classify imaging data. According to the results of the two machine learning methods, accurate positioning diagnosis can be performed on unknown clinical pathological sections.

[0038] The first method is to classify fluorescence images by combining pixel distribution maps with machine learning algorithms, such as Figure 6 As shown in Figure 2. The distribution of pixel values ​​in the same interval shows the differences between different stages. Figure 6 As shown in A, 6B, 6C, and 6D. Figure 6 E is the result of t-distributed stochastic neighbor embedding (t-SNE) dimensionality reduction, and the four classifications have obvious distinction. Figure 6 F is the confusion matrix of the test set of the random forest classifier, which has good classification results and can be used to predict the classification results of unknown images. Figure 6G is the group clustering heat map, which shows that the three machine learning algorithms can well distinguish the pixel distribution maps of the four categories. The pixel distribution maps show great differences in different stages.

[0039] The second classification method is to use convolutional neural network to classify fluorescence imaging. Figure 7 A is a scatter plot of the accuracy changes of 4 sets of parameter models (a, b, c, d correspond to normal tissue, stage I, stage II, and stage III malignant breast cancer tissue) randomized 10 times. The black corresponds to the training set and the gray corresponds to the test set. Figure 7 B is a scatter plot of the loss value changes of 4 sets of parameters randomized 10 times, black is the training set, and gray is the test set. The imaging data are brought into the convolutional neural network respectively, and the ratio of the training set to the test set is 8:2. Batch size = 10, number of iterations = 30. Randomly repeat 10 times, using different training sets and test sets each time. Figure 7 C and 7D are the accuracy and loss curves of 30 iterations of the four classification models. In 30 iterations, the loss of the training set approaches 0 from 1.4, and the loss of the test set approaches 0.2 from 1.38 ( Figure 7 D). The accuracy of the training set was optimized from 0.26 to 1, and the accuracy of the test set was optimized from 0.21 to 0.95 ( Figure 7 C). Figure 7 E and 7F are the confusion matrices of the classification model ( Figure 7 E is the training set, Figure 7 F is the test set). For the training set, 100% accuracy is shown for the four classifications; for the test set, the accuracy for each stage is 100%, 94.44%, 96%, and 90.91%, respectively. Such high classification accuracy demonstrates the effectiveness of the convolutional neural network model in distinguishing the four stages of breast cancer, indicating that the classification performance of the model is very good.

Claims

1. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation, characterized in that: An imaging nanoprobe with lipid targeting function (A) is used to stain the frozen sections of tumor tissue samples removed during surgery (B), or to stain the frozen sections of tumor tissue after the action of drugs, and then fluorescence imaging data is obtained under a fluorescence microscopy device; by analyzing the correlation of lipid substances in the imaging data, the boundaries of the tumor site can be defined and the therapeutic efficacy of the drug on the tumor can be evaluated.

2. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, characterized in that: The preparation of the imaging nanoprobe (A) is specifically as follows: placing an appropriate amount of N,N-diethyl-p-phenylenediamine in anhydrous ethanol, and placing the mixed solution in a high-pressure reactor for sufficient reaction; the probe purification is specifically as follows: using a silica gel chromatography column for purification, further purifying by adjusting the polarity of the eluent, and finally drying to obtain the imaging nanoprobe (A).

3. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1 or 2, characterized in that: The storage conditions of the imaging nanoprobe (A) are as follows: the purified probe is rotary evaporated and then stored in a dimethyl sulfoxide solution to form a storage solution of the imaging nanoprobe (A) for future use.

4. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, characterized in that: The processing steps of the tumor tissue sample are: using a freezing microtome to quickly freeze and embed the tissue sample during surgery, and then cutting the embedded tissue into slices with a thickness of 5-20 microns (B).

5. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, characterized in that: The staining step is specifically as follows: the probe (A) stock solution obtained in claim 3 is prepared into a 50-100 μg / mL imaging probe working solution (A-1) using a 0.01M phosphate buffer solution (C) with a pH of 7.2-7.4; the frozen section is immersed in the imaging probe working solution (A-1) and incubated at 37°C for 5 minutes; and then washed three times with a phosphate buffer solution (C) to obtain a stained tissue section (B-1) for the next step of fluorescence imaging.

6. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, characterized in that: The correlation analysis of lipid substances is used to define the tumor boundary. The specific steps are as follows: the pathological tissue sections obtained during the operation are stained and fluorescent image data is collected. Through image analysis, the metabolic correlation characteristics of lipid substances are obtained; Based on the characteristic discrimination basis, it is determined to be the tumor site or the normal tissue site.

7. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, characterized in that: The evaluation of the therapeutic efficacy of the drug on the tumor comprises the following steps: slicing the tumor biopsy tissue sample or the surgical tissue sample of the patient after drug administration, collecting fluorescence image data after staining, and obtaining the metabolic correlation characteristics of lipid substances through image analysis; Based on characteristic discrimination, the resistance or inresistance to the anti-tumor drugs used can be determined.

8. A method for rapid intraoperative pathological diagnosis and drug sensitivity evaluation as claimed in claim 1, 6 or 7, characterized in that: The basis for feature discrimination refers to an image processing and analysis algorithm related to lipid metabolism capacity.

Images