Construction method of pulmonary nodule benign and malignant differential diagnosis model based on single-cell immune spectrum

By constructing single-cell immunoassays using mass cytometry, the problems of high false positive rates and difficulty in identifying malignant nodules in lung nodule screening have been solved. This has enabled highly sensitive and specific lung cancer screening and early diagnosis, providing earlier treatment opportunities and more appropriate surgical approaches.

CN119044493AActive Publication Date: 2024-11-29ZHEJIANG UNIV

Patent Information

Application Number
CN202411096852.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-29
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

Existing methods for screening lung nodules have high false positive rates, make it difficult to accurately identify malignant nodules in the early stages, and lack effective methods for preoperative differentiation between MIA and IA, leading to overtreatment and diagnostic difficulties.

Method used

A single-cell immunoassay was constructed using mass cytometry. Peripheral blood mononuclear cells were obtained by Ficoll separation. Cells were labeled with 40 metal-coupled antibodies. A differential diagnostic model for benign and malignant pulmonary nodules was established by combining PARC clustering and random forest algorithms. 34 immune cell subsets were screened to achieve diagnosis with high sensitivity and high specificity.

Benefits of technology

It improves the sensitivity and specificity of lung cancer screening, reduces the risk of missed diagnosis and misdiagnosis, enables early diagnosis and reasonable surgical planning, and is non-invasive and highly accurate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure HDA0004988185130000011
    Figure HDA0004988185130000011
  • Figure HDA0004988185130000012
    Figure HDA0004988185130000012
  • Figure HDA0004988185130000013
    Figure HDA0004988185130000013
Patent Text Reader

Abstract

The invention provides a construction method of a pulmonary nodule benign and malignant differential diagnosis model based on a single cell immune spectrum. The method comprises the following steps: obtaining PBMC (peripheral blood mononuclear cells) by using a peripheral blood sample, and carrying out mass spectrum flow analysis on the PBMC to obtain a mass spectrum flow analysis data set; a PARC clustering algorithm is adopted, cells are divided into different phenotypes according to marker expression, the expression proportion of each cell subset is used as a modeling feature, a pulmonary nodule benign and malignant differential diagnosis model is obtained, and screening and early diagnosis of lung cancer are achieved. The technology has the characteristics of non-invasiveness, high sensitivity and high specificity, the diagnosis accuracy of lung cancer screening is improved, and an earlier treatment opportunity and a more suitable operation mode are provided for patients.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the medical field, and more specifically relates to a method for establishing a diagnosis model for differential diagnosis of benign and malignant pulmonary nodules based on a single cell immune profile of mass spectrometry technology. Background Art

[0002] Lung cancer is one of the malignant tumors with the highest morbidity and mortality rates worldwide. The vast majority of patients (48%) already have distant metastasis at the time of initial diagnosis, and the 5-year relative survival rate of these patients is only about 8%. The 5-year survival rate of patients initially diagnosed with early lung cancer can reach 62%, but the detection rate is only about 20%. In contrast, for stage I cancer, the 5-year survival rate increases significantly, from 68% to 92%. There is no doubt that increasing the detection rate of lung cancer in the curable stage (stage 0, I, and II) is the most effective way to reduce lung cancer mortality. Improving the early diagnosis rate of lung cancer is crucial to reducing lung cancer mortality and improving lung cancer prognosis.

[0003] However, due to the lack of clinical symptoms and sensitive technology, it is difficult to detect lung cancer in these early stages. Randomized controlled trials, including the National Lung Cancer Screening Trial (NLST) and the Netherlands-Belgium Lung Cancer Screening Trial (NELSON) trials, have shown that screening with low-dose CT (LDCT) can significantly reduce the mortality of lung cancer, but there is still uncertainty in the improvement of clinical effectiveness and cost-effectiveness of LDCT, especially the diagnosis of subsolid nodules is the most difficult and challenging, and there is also a problem with LDCT screening, which is the detection of a large number of false-positive nodules: most of the nodules detected by LDCT in high-risk populations are benign nodules. A large-scale NLST study from the United States showed that although 24.2% of the people who underwent LDCT examination had lung nodules, after follow-up, up to 96.4% of the positive nodules were diagnosed as benign (false positives); other studies have also shown that the proportion of benign lung nodules after surgical resection is as high as 20%, and 38% after biopsy. Several studies have confirmed that the average nodule detection rate of LDCT screening is about 20%, while >90% of nodules are benign. Too high a false positive rate may lead to overdiagnosis, overtreatment, waste of medical resources, and increase anxiety in the examinee.

[0004] The Mayo model based on imaging features (the earliest was the Mayo Clinical Lung Cancer Prediction Model developed by the Mayo Clinic in 1997 for patients with nodules ranging from 4-30 mm based on their age, smoking history, nodule diameter, spicule sign, location, etc.), the Veterans Association (VA) model and the BROCK model are used to judge the nature of nodules. The area under the curve (AUC) is 0.65-0.83, and the model accuracy is not high. Moreover, the Mayo Clinical Lung Cancer Prediction Model is based on Western patients and may not be suitable for Chinese patients. There are still certain limitations in its practical application.

[0005] Therefore, the key to the diagnosis and treatment of lung nodules is to effectively further differentiate and triage lung nodules, quickly determine whether they are benign or malignant, and remove malignant nodules as soon as possible, while avoiding unnecessary overtreatment and reducing the proportion of benign nodules that require surgical removal. In order to meet the huge unmet clinical needs, highly sensitive and specific methods are urgently needed to accurately identify malignant lung nodules.

[0006] On the other hand, the World Health Organization (WHO) Classification of Lung Tumors (5th edition, 2021) classifies atypical adenomatous hyperplasia (AAH) and adenocarcinoma in situ (AIS) as glandular precursor lesions, which are not included in the category of lung adenocarcinoma, and classifies microinvasive adenocarcinoma (MIA) and invasive adenocarcinoma (IA) as two types of lung cancer with different degrees of invasiveness. The Association for the Study of Lung Cancer / American Thoracic Society / European Respiratory Society multidisciplinary classification of lung adenocarcinoma shows that MIA often does not spread to regional lymph nodes or metastasize, and its postoperative disease-free survival rate is almost 100%. Therefore, MIA can choose sublobar resection and does not require systematic lymph node dissection, while the surgical method for IA patients needs to be combined with clinical practice. Therefore, a method that can effectively distinguish between MIA and IA before surgery is needed to formulate a surgical plan in advance.

[0007] The human immune system is closely related to the occurrence and development of tumors. During the occurrence and development of tumors, the tumor immune microenvironment (TME) is also constantly changing. Tumor immune changes are not only limited to the tumor, but are often accompanied by systemic immune disorders. These findings highlight the complex interaction between tumors and peripheral immunity. Studies have shown that there are many types of infiltrating immune cells in lung tumors, which play an extremely important role in the occurrence and development of tumors. Macrophages and T cell populations have potential interactions in the tumor immune microenvironment, and lung tumors are also rich in other myeloid components, including neutrophils, non-classical monocytes and intermediate monocytes; at the same time, studies have shown that the presence of B cells is associated with protective immunity in lung cancer patients; some other clinical studies have shown that high density of tumor-infiltrating T cells is associated with an increase in the median survival of cancer patients. Therefore, comprehensive and detailed monitoring of the human peripheral immune status may provide great help in distinguishing benign and malignant lung nodules.

[0008] In recent years, mass spectrometry technology combines traditional flow cytometry analysis methods with mass spectrometry detection methods, uses metal isotope labels to replace fluorescent labels, and uses mass spectrometry to quantify labels, which can achieve simultaneous detection of more than 40 target proteins at the single-cell level, without mutual interference between channels, and without complicated and cumbersome compensation calculations. It significantly enhances the ability to evaluate complex cell systems and processes and is a multi-parameter, high-throughput single-cell protein detection technology platform. By detecting multiple marker combinations, CyTOF technology can distinguish multiple cell subpopulations, construct cell maps of health or disease states, conduct a comprehensive analysis of intracellular signal transduction networks, and analyze the basic immune status of the human body in a high-dimensional manner to obtain a full range of immune cell composition, phenotype and functional information. Summary of the invention

[0009] In view of this, the purpose of the present invention is to provide a method for constructing a model for differential diagnosis of benign and malignant pulmonary nodules based on single-cell immune profiles, and a method for establishing a model for differential diagnosis of benign and malignant pulmonary nodules based on single-cell immune profiles of mass spectrometry technology. Peripheral blood mononuclear cells (PBMC) are obtained by treating peripheral blood samples with Ficoll separation (Ficoll density gradient centrifugation), and PBMCs are subjected to mass spectrometry flow analysis to obtain a mass spectrometry flow analysis data set; the PARC clustering algorithm is used to divide cells into different phenotypes according to marker expression, and the expression ratio of each cell subgroup is used as a modeling feature to obtain a model for differential diagnosis of benign and malignant pulmonary nodules.

[0010] In order to achieve the above object, the present invention adopts the following technical solution:

[0011] 1. PBMCs were obtained by treating peripheral blood samples with Ficoll separation, and suspended in 5 ml of pre-cooled FACS buffer (1×PBS+0.5% BSA), then centrifuged at 400×g for 5 min at 4°C, the supernatant was discarded, and the cell pellet was resuspended in buffer. Cell count and quality assessment of PBMCs were performed before mass cytometry (CyTOF, Mass Cytometry) detection to ensure that the count was greater than 3×106 and the viability was greater than 85%.

[0012] 2. 40 metal-conjugated antibodies were selected as markers for labeling cells; PBMCs were washed with PBS buffer and then stained with 0.5mM cisplatin, and the cells were blocked from binding to antibodies with Fc receptors for 30min, and unbound antibodies were removed by centrifugation. The PBMCs were then fixed in 200μL of insertion solution overnight. The cells were washed and resuspended in distilled water, added to 20% EQ beads, and further analyzed on a mass spectrometry flow cytometer. The markers of the 40 metal-conjugated antibodies include: CD45, CD3, CD56, TCRγ / δ, CD196 (CCR6), CD14, IgD, CD123 (IL-3Rα), CD85j (ILT2), CD19, CD25 (IL-2Rα), CD274 (PD-L1), CD278 (ICOS), CD39, CD27, CD24, CD45RA, CD86, CD28, CD19 7(CCR7), CD11c, CD33, CD152(CTLA-4), CD161, CD185(CXCR5), CD66b, CD183(CXCR3), CD94, CD57, CD45RO, CD127(IL-7Rα), CD279(PD-1), CD38, CD194(CCR4), CD20, CD16, HLA-DR, CD4, CD8a, CD11b.

[0013] 3. The PARC clustering algorithm was used to divide cells into different phenotypes according to marker expression. The expression ratio of each cell subgroup was used as the modeling feature, and the random forest algorithm and 10-fold cross validation were used to screen the features.

[0014] 4. Through the random forest method, a model for differential diagnosis of benign and malignant lung nodules was obtained.

[0015] Optionally, step 3 also includes screening out 34 immune cell subsets and markers as modeling features. The 19 features of the differential diagnosis model for benign and malignant pulmonary nodules include: CD33-CD14-CD3+CD4+CD28+, CD33-CD14-CD3+CD4+CD274+, CD33-CD14-CD3+CD4+CD197+CD45RA-, CD33-CD14-CD3+CD4+HLA_DR+CD38+, CD33-CD14-CD3+CD4+CXCR5-CD 183-CCR6-, CD33-CD14-CD3+CD4+CD25+CD127-CD161-CD45RA+, CD33-CD14-CD3+CD8+CD197+-C D45RA+, CD33-CD14-CD3-CD19+CD24+CD38+, CD33-CD14-CD3-CD20-CD38+CD27+, CD33-CD14-CD3 -CD56+CD16+CD94+, CD33-CD14-CD3-CD56+CD16+CD161+, CD3-CD19-CD56-CD14-CD123+CD11c+, CD33-CD14-CD3-CD56+CD16-, CD86, CD11c, CD183, CD94, CD4, CD11b; The 15 features of the lung cancer infiltration degree judgment model include: CD3 3-CD14-CD3+CD8+CD85j+, CD33-CD14-CD3+CD8+CD161-, CD33-CD14-CD3+CD4+, CD33-CD14-CD3 +CD4+CD197-CD45RA+, CD33-CD14-CD3+CD4+HLA_DR+CD38+, CD33-CD14-CD3+CD4+HLA_DR+CD38- , CD33-CD14-CD3+CD8+, CD33-CD14-CD3+CXCR5+, CD33-CD14-CD3+CD8+CD197+CD45RA-, CD33-CD14-CD3-CD19+CD24+CD38+, CD33-CD14-CD3-CD56+CD16+ , CD33-CD14-CD3-CD56+CD16+CD57+, CD33-CD14-CD3-CD56+CD16+HLA_DR+, CD3-CD19-CD56-CD14-HLA_DR-, CD56.

[0016] Optionally, step 3 also includes using a training set and a validation set, which is to divide the samples into a training set and a validation set according to the chronological order of group entry.

[0017] Optionally, step 2 also includes using a combination of 40 antibody markers.

[0018] Optionally, when selecting samples for inclusion in step 1, random inclusion is adopted, covering samples of different nodule sizes (such as ≤10 mm, 11-20 mm, 21-30 mm, etc.), different types of nodules (such as solid nodules, partially solid nodules, pure ground glass density nodules), and different types of adenocarcinoma infiltration degrees (such as atypical adenomatous hyperplasia AAH, carcinoma in situ AIS, minimally invasive adenocarcinoma MIA, invasive adenocarcinoma IA, etc.).

[0019] Through the above technical solutions, it can be known that the present invention discloses a method for establishing a single-cell immune profile differential diagnosis model for benign and malignant pulmonary nodules, which has good detection efficiency in the differential diagnosis of lung cancer (CA confirmed by pathology) and non-cancerous group samples (including Non-CA confirmed by imaging and Non-CA confirmed by pathology), and the AUC of the training set and the validation set can reach 0.95 and 0.96, which are better than the existing clinical models: Mayo model (training set and validation set AUC are 0.75 and 0.70, respectively), Veterans Association (VA) model (training set and validation set AUC are 0.73 and 0.65, respectively) and BROCK model (training set and validation set AUC are 0.84 and 0.85, respectively); and it also performs well in the most difficult to identify pathological non-cancerous group and lung cancer group, with the AUC of the training set and the validation set being 0.92 and 0.90, respectively, which are significantly better than the existing clinical models: Mayo model (training set and validation set AUC are 0.69 and 0.61, respectively), Veterans Association (VA) model (training set and validation set AUC are 0.73 and 0.65, respectively) and BROCK model (training set and validation set AUC are 0.84 and 0.85, respectively); and it also performs well in the most difficult to identify pathological non-cancerous group and lung cancer group, with the AUC of the training set and the validation set being 0.92 and 0.90, respectively, which are significantly better than the existing clinical models: Mayo model (training set and validation set AUC are 0.69 and 0.61, respectively), Veterans Association (VA) model (training set and validation set AUC are 0.63 and 0.65, respectively). Association (VA) model (the AUC of the training set and validation set were 0.68 and 0.61, respectively) and BROCK model (the AUC of the training set and validation set were 0.72 and 0.65, respectively).

[0020] At the same time, the model established in this scheme can effectively distinguish MIA and IA before surgery, and the AUC of the training set and the validation set are 0.97 and 0.93 respectively. The model of the present invention improves the sensitivity and specificity of lung cancer screening, reduces the risk of missed diagnosis and misdiagnosis, and can be used for auxiliary diagnosis and screening of lung cancer

[0021] The present invention provides a lung cancer screening technology based on cellular immune profile analysis of mass spectrometry technology, which detects the expression profile of tumor-related immune cells in the patient's peripheral blood, classifies the peripheral blood sample data, combines artificial intelligence algorithms, builds a learning model, and inputs the peripheral blood sample data into the model to obtain the patient's risk score for lung cancer, thereby achieving screening and early diagnosis of lung cancer. The technology is non-invasive, highly sensitive, and highly specific, and can quickly and accurately detect the expression profile of lung cancer-related immune cells in peripheral blood, improving the diagnostic accuracy of lung cancer screening and providing patients with earlier treatment opportunities and the best treatment methods.

[0022] The advantages of the present invention are: (1) improving the sensitivity and specificity of lung cancer screening and reducing the risks of missed diagnosis and misdiagnosis; (2) avoiding invasive diagnostic operations and reducing clinical risks; (3) achieving early diagnosis of lung cancer and providing patients with earlier treatment opportunities; (4) further classifying the degree of lung cancer infiltration and providing patients with more suitable surgical methods; (5) using CyTOF technology, which has the characteristics of high throughput, high resolution and high sensitivity, and can quickly and accurately detect the expression spectrum of lung cancer-related immune cells; (6) combining artificial intelligence algorithms to improve the accuracy and reliability of lung cancer screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0024] Figure 1 This is a flow chart of lung cancer screening based on CyTOF cellular immune profiles of the present invention.

[0025] Figure 2 It is a box plot of risk scores of lung cancer and non-cancer group samples in the training set and validation set of the lung cancer diagnosis model of the present invention.

[0026] Figure 3 It is the AUC graph of the training set and the validation set of the lung cancer diagnosis model of the present invention.

[0027] Figure 4 It is a risk score box plot of the training set and validation set of the lung cancer diagnosis model of the present invention in the most difficult to distinguish pathological non-cancer group and lung cancer group samples.

[0028] Figure 5 It is the AUC graph of the training set and validation set of the lung cancer diagnosis model of the present invention in the most difficult to distinguish pathological non-cancer group and lung cancer group samples.

[0029] Figure 6 It is a box plot of risk scores of lung cancer and non-cancer group samples in the training set and validation set of the surgical method decision model of the present invention.

[0030] Figure 7 It is the AUC graph of the training set and validation set of the surgical method decision model of the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is described clearly and completely below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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] Example 1 A method for establishing a diagnosis model for differential diagnosis of benign and malignant pulmonary nodules based on single cell immune profiles of mass spectrometry technology

[0033] The specific steps are as follows: 1. Collect peripheral blood samples (5 ml / case) and send them to the laboratory for further processing within 12 hours at room temperature or within 48 hours at 4°C. The subjects should meet the following requirements: over 18 years old; clear diagnosis of the disease (patients with lung nodules planned for surgical resection and further confirmed by histopathology, or no changes in lung nodules after 3 years of follow-up, or lung nodule diameter ≤4mm); sign the informed consent form. The exclusion criteria for collection include: history of cancer-related treatment; acute infection; blood transfusion therapy within 6 months; use of drugs that affect peripheral blood components in the past 2 weeks; local recurrence of tumors; organ decompensated dysfunction; immunodeficiency syndrome; blood precancerous diseases; receiving immunosuppressive therapy; and coagulation dysfunction. A total of 1,032 peripheral blood samples were collected for this project. Figure 1 As shown, the collected peripheral blood samples were transferred to the laboratory and processed as follows.

[0034] 2. Sample pretreatment, including: Isolation of peripheral blood mononuclear cells (PBMCs) from blood using Ficoll-Paque density gradient centrifugation. Suspend them in 5 ml pre-cooled FACS buffer (1×PBS+0.5% BSA), then centrifuge at 400×g for 5 min at 4°C, discard the supernatant, and resuspend the cell pellet in buffer. Perform cell counting and quality assessment of PBMCs before CyTOF detection to ensure that the count is greater than 3×106 and the viability is greater than 85%.

[0035] 3. CyTOF staining and data analysis, including: 40 metal-coupled antibodies were selected as markers for labeling cells; PBMCs were washed with PBS buffer and then stained with 0.5mM cisplatin, and the cells were blocked from binding to antibodies for 30min with Fc receptors, and unbound antibodies were removed by centrifugation. The PBMCs were then fixed in 200μL of insertion solution overnight. The cells were washed and resuspended in distilled water, added to 20% EQ beads, and further analyzed on a mass spectrometry flow cytometer. The FCS file was normalized using the bead normalization method. Each sample data was debarcoded using a dual-state filtering scheme with a unique mass marker barcode. FlowJo software was used to exclude debris, dead cells, and double cells, leaving only live single immune cells.

[0036] 4. Feature screening: The samples were divided into training sets and validation sets according to the chronological order of enrollment. The training set included 178 lung cancer samples and 218 non-cancer controls. First, we evaluated the negative and positive expression of markers on each cell in the training set. Then, based on the expression profiles of these markers, the random forest algorithm and 10-fold cross-validation were used to select characteristic cell subpopulations. The features with an importance level exceeding 0.01 in each successful random forest model construction were recorded. If a feature occurred more than 350 times during 1,000 cross-validations, it would be counted. Finally, a total of 19 characteristic cell subpopulations were screened for model construction.

[0037] 5. Model establishment: Using 178 lung cancer samples in the training set and 218 non-cancer control group samples, the features screened by the above method were used to build a lung cancer diagnosis model through random forest method. The risk score of each participant was calculated by the model. The score is the average probability of the positive probability of the sample judged by each decision tree in the random forest model, ranging from 0 to 1.

[0038] Example 2: A validation method for a single cell immune profile based on mass spectrometry technology for differential diagnosis of benign and malignant pulmonary nodules

[0039] 1. Using the protocol provided in Example 1, 251 untrained non-cancerous samples and 283 untrained lung cancer samples in the validation set were used for CyTOF staining and data analysis.

[0040] 2. Input the new peripheral blood sample into the lung cancer diagnostic model jointly constructed using 19 cell subsets, predict and evaluate the blood sample, and determine whether the sample is a lung cancer patient based on the prediction results of the lung cancer diagnostic model.

[0041] Example 3 Establishment and verification of a model for judging the degree of lung cancer invasion based on single cell immune profiles of mass spectrometry technology

[0042] 1. Using the scheme provided in Example 1, 113 lung nodule samples pathologically diagnosed as MIA and 105 lung nodule samples pathologically diagnosed as IA in the training set were used for CyTOF staining and data analysis.

[0043] 2. The scheme provided in Example 1 was adopted to select characteristic cell subpopulations using random forest algorithm and 10-fold cross validation. A total of 15 cell subpopulations were screened as modeling features to construct a new model for judging the degree of infiltration of malignant pulmonary nodules.

[0044] 3. The scheme provided in Example 2 was adopted to perform model validation using 111 lung nodule samples with pathological diagnosis of MIA and 106 lung nodule samples with pathological diagnosis of IA in the validation set.

[0045] Obviously, the above embodiments of the present invention are merely examples for more clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made on the basis of the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.

[0046] Combination Figure 2-Figure 3 It can be seen that the differential diagnosis model for benign and malignant pulmonary nodules disclosed in the present invention has good detection efficiency in distinguishing lung cancer (CA confirmed by pathology) and non-cancerous group samples (including Non-CA confirmed by imaging and Non-CA confirmed by pathology), and the AUC of the training set and the validation set can reach 0.95 and 0.96, which is better than the existing clinical models: Mayo model (AUC of the training set and the validation set are 0.75 and 0.70, respectively), Veterans Association (VA) model (AUC of the training set and the validation set are 0.73 and 0.65, respectively) and BROCK model (AUC of the training set and the validation set are 0.84 and 0.85, respectively);

[0047] Combination Figure 4-Figure 5 It can be seen that the diagnostic model for differential diagnosis of benign and malignant pulmonary nodules disclosed in the present invention has excellent performance in the most difficult to distinguish pathological non-cancerous group and lung cancer group (AUC of the training set and validation set are 0.92 and 0.90 respectively), which is better than the existing clinical models: Mayo model, Veterans Association (VA) model and BROCK model.

[0048] Combination Figure 6-Figure 7 The model established in this scheme can effectively distinguish MIA and IA before surgery, and the AUC of the training set and validation set are 0.97 and 0.93, respectively.

[0049] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the steps disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0050] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a diagnostic model for benign and malignant pulmonary nodules based on single-cell immune profiles, characterized in that: The following steps are involved: Peripheral blood samples were processed by Ficoll separation method to obtain peripheral blood mononuclear cells, which were then subjected to mass spectrometry flow cytometry analysis to obtain a mass spectrometry flow cytometry analysis data set. The PARC clustering algorithm was used to divide cells into different phenotypes according to marker expression, and the expression ratio of each cell subgroup was used as a modeling feature to obtain a differential diagnosis model for benign and malignant pulmonary nodules.

2. The construction method according to claim 1, characterized in that: This is achieved through the following steps: (1) Peripheral blood samples were treated with Ficoll separation to obtain peripheral blood mononuclear cells, which were suspended in 5 ml of pre-cooled flow cytometry sorting buffer, and then centrifuged at 400 × g for 5 min at 4 °C, the supernatant was discarded, and the cell pellet was resuspended in buffer. The cell count and quality assessment of peripheral blood mononuclear cells were performed before CyTOF mass spectrometry flow cytometry detection to ensure that the count was greater than 3 × 106 and the viability was greater than 85%; (2) 40 metal-conjugated antibodies were selected as markers for labeling cells; PBMCs were washed with PBS buffer and then stained with 0.5 mM cisplatin. Cells were blocked for 30 min with Fc receptors to prevent antibody binding. Unbound antibodies were removed by centrifugation. PBMCs were then fixed in 200 μL of insertion solution overnight. Cells were washed and resuspended in distilled water, added to 20% EQ magnetic beads, and further analyzed on a mass cytometer. (3) The PARC clustering algorithm was used to divide cells into different phenotypes according to marker expression, and the expression ratio of each cell subgroup was used as the modeling feature. The random forest algorithm and 10-fold cross validation were used to screen the features. (4) The random forest method was used to establish a diagnostic model for differential diagnosis of benign and malignant pulmonary nodules.

3. The construction method according to claim 2, characterized in that: When selecting samples for inclusion in step (1), random inclusion is adopted to cover samples of different nodule sizes, different nodule types, and different types of adenocarcinoma infiltration degrees.

4. The construction method according to claim 3, characterized in that: The different nodule sizes are ≤10mm, 11-20mm, and 21-30mm. The different types of nodules are solid nodules, partially solid nodules, and pure ground glass density nodules. The samples with different types of adenocarcinoma infiltration degrees are atypical adenomatous hyperplasia AAH, carcinoma in situ AIS, minimally invasive adenocarcinoma MIA, and invasive adenocarcinoma IA.

5. The construction method according to claim 2, characterized in that: The flow cytometry sorting buffer in step (1) is 1×PBS+0.5% BSA.

6. The construction method according to claim 2, characterized in that: The 40 metal-coupled antibody markers in step (2) include: CD45, CD3, CD56, TCRγ / δ, CD196, CD14, IgD, CD123, CD85j, CD19, CD25, CD274, CD278, CD39, CD27, CD24, CD45RA, CD86, CD28, CD197, CD11c, CD33, CD152, CD161, CD185, CD66b, CD183, CD94, CD57, CD45RO, CD127, CD279 (PD-1), CD38, CD194, CD20, CD16, HLA-DR, CD4, CD8a, and CD11b.

7. The construction method according to claim 2, characterized in that: The step (2) uses a combination of 40 metal-coupled antibody markers.

8. The construction method according to claim 2, characterized in that: The step (3) screened out 34 immune cell subsets and markers as modeling features. The 19 features of the benign and malignant differential diagnosis model of pulmonary nodules included: CD33-CD14-CD3+CD4+CD28+, CD33-CD14-CD3+CD4+CD274+, CD33-CD14-CD3+CD4+CD197+CD45RA-, CD33-CD14-CD3+CD4+HLA_DR+CD38+, CD33-CD14-CD3+CD4+CXCR5-CD183-CCR6-, CD33-CD14-CD3+CD4+CD25+CD127-CD161-CD45RA+, CD33-CD14-CD3+CD8+CD197+-CD45RA+, CD33-CD14-CD3-CD19+CD24+CD38+, CD33-CD14-CD3-CD20-CD38+CD27+, CD33-CD14-CD3-CD56+CD16+CD94 +, CD33-CD14-CD3-CD56+CD16+CD161+, CD3-CD19-CD56-CD14-CD123+CD11c+, CD33-CD14-CD3-CD56+CD16-, CD86, CD11c, CD183, CD94, CD4, CD11b; The 15 features of the lung cancer infiltration degree judgment model include: CD33-CD14-CD3+CD8+CD85j+, CD33-CD14-CD3+CD8+CD161-, CD33-CD14-CD3+CD4+, CD33-CD14-CD3+CD4+CD197-CD45RA+, CD33-CD14-CD3+CD4+HLA_DR+CD38+, CD33-CD14-CD3+CD4+HLA_DR+CD38-, CD33-CD14-CD3+CD8+, CD33-CD14-CD3+CXCR5+, CD33-CD14-CD3+CD8+CD197+CD45RA-, CD33-CD14-CD3-CD19+CD24+CD38+, CD33-CD14-CD3-CD56 +CD16+, CD33-CD14-CD3-CD56+CD16+CD57+, CD33-CD14-CD3-CD56+CD16+HLA_DR+, CD3-CD19-CD56-CD14-HLA_DR-, CD56.

9. The construction method according to claim 2, characterized in that: The step (3) uses a training set and a validation set, and divides the samples into the training set and the validation set according to the time sequence of group entry.

Citation Information

Patent Citations

  • Auxiliary diagnosis method for benign and malignant properties and infiltration degree of pulmonary nodules based on artificial intelligence

    CN107977966A

  • Method for establishing pancreatic cancer screening model based on mass spectrum flow technology

    CN114936449A

  • Application of kit based on mass spectrum flow detection technology in preparation of benign and malignant pulmonary nodule screening product

    CN116026748A

  • Non-small cell lung cancer early typing system based on peripheral blood immune cell map

    CN116337727A

  • Diagnostic method for improving wettability of lung adenocarcinoma of 3 cm or below in operation through multi-modal fusion

    CN118135564A

Cited By

  • Peripheral blood leucocyte DNA methylation gene network model for evaluating benign and malignant and wettability risks of pulmonary nodules and construction method of peripheral blood leucocyte DNA methylation gene network model

    CN122256506A

  • Peripheral blood leukocyte dna methylation gene network model for lung nodule benign and malignant, invasive risk assessment and construction method thereof

    CN122256506B

  • Method for constructing diagnostic model for distinguishing between benign and malignant pulmonary nodules on basis of single-cell immune profile

    WO2026036646A1