Anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile

Abstract

The present invention discloses a kind of anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell atlas, including: peripheral blood single cell acquisition unit, mass spectrometry flow detection unit, raw data conversion unit, normalization processing unit, peripheral blood immune cell grouping acquisition unit, filtering unit, peripheral blood immune cell secondary grouping acquisition unit, clustering unit, feature vector acquisition unit and treatment effectiveness evaluation unit. The analysis sample used by the system of the present invention comes from the patient's peripheral blood, which is highly operable, non-invasive, stable and repeatable; the technology used is CyTOF, which has moderate cost; and the evaluation of the anti-PD-1 drug effect can be advanced to 5 to 6 cycles after medication, and compared with the imaging detection method, it can distinguish the ineffective patients earlier, help clinicians make efficacy evaluation earlier, and change the treatment plan in time for the patients with ineffective anti-PD-1 treatment.

Description

Technical Field

[0001] The present invention relates to the technical field of drug treatment effectiveness evaluation, and in particular to an anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell atlas. Background Art

[0002] Hepatocellular carcinoma (HCC) is the most common pathological type of primary liver cancer, accounting for 75% to 85% of cases. Currently, for patients with advanced HCC, targeted immunotherapy through programmed cell death protein 1 (PD-1) / programmed cell death ligand 1 (PD-L1) has become a promising treatment method. Studies have shown that the anti-PD-1 blocking immunotherapy currently used in clinical practice, including camrelizumab, nivolumab and pembrolizumab, can increase the objective response rate (ORRs, an indicator for evaluating the therapeutic effect) to 15% to 20% in patients with advanced hepatocellular carcinoma who have previously received sorafenib pretreatment, which is about 3 times the ORR of sorafenib alone. Despite this, to date, the number of patients with advanced hepatocellular carcinoma who have benefited from the use of PD-1 or PD-L1 blockers as immunotherapy is still small. This is most likely due to the inherent high heterogeneity of the systemic immunity, pathogenesis, histopathology and molecular basis of HCC. Previous studies have shown that biomarkers such as PD-L1 expression or tumor mutation burden (TMB) are correlated with the efficacy of PD-1 / PD-L1 blockade therapy in patients with advanced hepatocellular carcinoma, but they are not sufficient to accurately evaluate the efficacy of blockade therapy. Therefore, we urgently need to find predictive biomarkers or construct methods to help evaluate the efficacy of PD-1 / PD-L1 immunotherapy in patients with advanced hepatocellular carcinoma.

[0003] In recent years, studies have used high-dimensional single-cell analysis to further reveal the immune phenotype composition in the tumor microenvironment (TME) of hepatocellular carcinoma. These findings are important for understanding the expression of specific cell subsets, such as PD-1 + The role of T cells in HCC progression and their response to blocking therapy are of great significance. However, it is often difficult to obtain tumor tissue from patients with advanced hepatocellular carcinoma, and liver biopsy still has risks. For example, it often causes serious complications including bleeding in patients with hepatocellular carcinoma with varying degrees of cirrhosis. Even if a biopsy is performed, local samples do not represent the overall tumor microenvironment well due to intratumor heterogeneity. In contrast, blood-based detection methods are more suitable for studying and dynamically monitoring disease progression, treatment and prognosis as a feasible and sensitive alternative. Summary of the invention

[0004] The purpose of the present invention is to provide an anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell maps that is easy to implement and has few side effects. Compared with the imaging detection methods currently used in clinical practice, it can evaluate patients who are ineffective in treatment earlier to address the shortcomings of the existing technology.

[0005] The present invention adopts the following technical solutions:

[0006] An anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profiles, including:

[0007] A peripheral blood single cell acquisition unit is used to separate mononuclear cells from the peripheral blood of the patient to be diagnosed to obtain single cells of the peripheral blood sample; wherein the peripheral blood sample is a peripheral blood sample of the patient to be diagnosed from the baseline to 5 to 6 treatment cycles of anti-PD-1 monotherapy;

[0008] The mass spectrometry flow cytometry detection unit uses pre-designed metal isotope-coupled protein-specific biomarkers to perform mass spectrometry flow cytometry experiments on single cells in peripheral blood samples to obtain the sample's single-cell mass spectrometry raw data Data raw ;

[0009] The raw data conversion unit is used to convert the raw data of single cell mass spectrometry into raw data raw Perform conversion processing to obtain the converted single-cell mass spectrometry flow data Data trans ; Among them, the conversion processing formula is: Data trans =sinh -1 (Data raw / 10);

[0010] Normalization processing unit, used to normalize the converted single cell mass spectrometry data trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization ;

[0011] Peripheral blood immune cell clustering acquisition unit uses a pre-trained 10*10 self-organizing network SOM-net 1 to normalize the single-cell mass spectrometry data after normalization normalization Classify and obtain immune cell clustering data of peripheral blood samples PBMC-1 ;

[0012] Filtering unit, used to filter immune cell clustering data from peripheral blood samples PBMC-1 CD15 + CD16 +Granulocytes and abnormal cell populations, including lymphocytes, dendritic cells and monocytes Data PBMC-2 ;

[0013] The peripheral blood immune cell secondary clustering acquisition unit uses the pre-trained 10*10 self-organizing network SOM-net2 to PBMC-2 Classify and obtain the secondary grouping data of immune cells in peripheral blood samples som-cluster ;

[0014] Clustering unit, secondary clustering data of immune cells in peripheral blood samples som-cluster Merge and cluster to obtain the peripheral blood sample immune cell subset information, i.e., the peripheral blood immune cell atlas Data immune-cell ;

[0015] Feature vector acquisition unit, based on the peripheral blood immune cell atlas Data immune-cell , , Calculate the Spearman correlation coefficient between the proportion of T02, T05, T10, T13, T15 and T27 immune cell subsets and the treatment cycle in samples of different treatment cycles of the patients to be diagnosed to form the characteristic vector R;

[0016] The treatment effectiveness evaluation unit combines the feature vector R and the pre-obtained feature vector R N Hierarchical clustering was performed, and those who responded to the treatment were clustered into the same branch, and those who did not respond to the treatment were clustered into another branch. The effectiveness of anti-PD-1 monotherapy was evaluated based on the clustering results.

[0017] Furthermore, the peripheral blood single cell acquisition unit separates mononuclear cells from the peripheral blood of the patient to be diagnosed by using a Ficoll separation solution gradient centrifugation method or an ACK lysis solution centrifugation method to obtain a peripheral blood sample single cell.

[0018] Furthermore, the peripheral blood single cell acquisition unit collects peripheral blood samples from the patient to be diagnosed once during each treatment cycle from baseline to 5 to 6 treatment cycles of anti-PD-1 monotherapy.

[0019] Furthermore, the metal isotope-coupled protein-specific biomarkers pre-designed in the mass spectrometry flow detection unit are shown in the following table:

[0020]

[0021] Furthermore, the normalization processing unit uses the z-score method to normalize the converted single-cell mass spectrometry data Data trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization .

[0022] Furthermore, the peripheral blood immune cell clustering acquisition unit SOM-net 1 acquisition steps are as follows:

[0023] a) 60 peripheral blood samples from 10 patients with advanced hepatocellular carcinoma were selected, and the expression of each metal isotope-coupled protein-specific biomarker in each cell was used as the original feature data Data raw-train ;

[0024] b) Convert the original feature data of each sample to obtain Data trans-train ,Data trans-train =sinh -1 (Data raw-train / 10);

[0025] c) Convert the processed data of each sample trans-train Each biomarker channel data is normalized separately to obtain normalized high-dimensional feature data Data normalization-train ;

[0026] d) On average, 4000 cells are randomly selected from each sample, for a total of 240,000 cells, to obtain the selected normalized high-dimensional feature data Data normalization-train-selected , and perform 10*10 self-organizing network training on it to obtain 10*10 self-organizing network SOM-net 1.

[0027] Furthermore, step c) converts the processed data Data of each sample trans-train Each biomarker channel data in is normalized using the z-score method to obtain normalized high-dimensional feature data Data normalization-train .

[0028] Furthermore, the peripheral blood immune cell secondary clustering acquisition unit SOM-net 2 acquisition steps are as follows:

[0029] a) The normalized high-dimensional feature data Data obtained from 60 peripheral blood samples of 10 patients with advanced hepatocellular carcinoma normalization-train Substitute into SOM-net 1 to obtain immune cell clustering;

[0030] b) Based on the immune cell clustering information, filter out CD15 + CD16 + Granulocytes and abnormal cell populations, including lymphocytes, dendritic cells and monocytes Data PBMC-2-train ;

[0031] c) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data DataPBMC-2-train-selected , and perform 10*10 self-organizing network training on it to obtain 10*10 self-organizing network SOM-net 2.

[0032] Furthermore, the clustering unit uses a hierarchical clustering method to perform secondary clustering of immune cell data from peripheral blood samples. som-cluster Merge clustering was performed to obtain a peripheral blood immune cell atlas Data containing 27 T cell subsets, 3 B cell subsets, 2 NK cell subsets, 1 dendritic cell subset and 5 monocyte subsets, a total of 38 immune cell subsets immune-cell .

[0033] Furthermore, the feature vector R obtained in advance by the treatment effectiveness evaluation unit N is the feature vector R of 10 patients with advanced hepatocellular carcinoma for whom the efficacy of anti-PD-1 monotherapy is known N (N=1, 2, 3, ..., 10).

[0034] Beneficial effects of the present invention:

[0035] The analysis samples used in the system of the present invention come from the patient's peripheral blood, which is highly operable, non-invasive, stable and repeatable; the technology used is CyTOF, which is moderately costly; and the evaluation of the effect of anti-PD-1 drugs can be advanced to 5 to 6 cycles after medication. Compared with imaging detection methods, it can identify patients with ineffective treatment earlier, help clinicians make efficacy evaluation earlier, and timely change the treatment plan for patients with ineffective anti-PD-1 treatment, thereby improving patient survival rate. The present invention will provide more abundant efficacy judgment indicators for the precise anti-PD-1 treatment of advanced hepatocellular carcinoma. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a schematic diagram of the sample collection and mass spectrometry flow detection experiment of the present invention: 1 peripheral blood sample, 2 Ficoll separation solution, 3 mixed solution after ACK lysis solution and peripheral blood sample are fully mixed, 4 peripheral blood cells, 5 metal isotope coupled protein specific biomarkers (Table 1), 6 CyTOF instrument, 7 CyTOF offline raw data.

[0037] Figure 2 It is a heat map of immune cell subset classification and protein-specific biomarker expression of the present invention.

[0038] Figure 3 It is the Spearman correlation coefficient of the present invention comparing the proportion of specific cell subpopulations in DCB and NDB patient groups with changes in treatment cycles.

[0039] Figure 4This is a clinical information chart of 10 patients receiving PD-1 immunotherapy at different time points in the present invention: SD-disease stability, PD-disease progression, PR-partial remission, last dose-the last time anti-PD-1 monotherapy was used, Blood-collection-the time point for blood sample collection, + is the time when PD is first determined, and when PD is determined, the patient stops taking the medicine; wherein, SD, PD, and PR are objective responses determined according to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) standard.

[0040] Figure 5 It is a layer clustering heat map of the dynamic changes of peripheral blood immune components of patients of the present invention.

[0041] Figure 6 These are the tumor imaging images of two patients who received ineffective PD-1 immunotherapy in the present invention. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. The embodiments are only used to explain the present invention and do not limit the protection scope of the present invention.

[0043] An anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profiles, including:

[0044] A peripheral blood single cell acquisition unit is used to separate mononuclear cells from the peripheral blood of the patient to be diagnosed by using a Ficoll separation solution gradient centrifugation method or an ACK lysis solution centrifugation method to obtain a peripheral blood sample single cell; wherein the peripheral blood sample is a peripheral blood sample of the patient to be diagnosed from the baseline to 5 to 6 treatment cycles of anti-PD-1 monotherapy, and preferably, a peripheral blood sample is collected once in each treatment cycle;

[0045] The mass spectrometry flow cytometry detection unit uses pre-designed metal isotope-coupled protein-specific biomarkers to perform mass spectrometry flow cytometry experiments on single cells in peripheral blood samples to obtain the sample's single-cell mass spectrometry raw data Data raw ; Pre-designed metal isotope coupled protein-specific biomarkers, as shown in the table below:

[0046]

[0047] The raw data conversion unit is used to convert the raw data of single cell mass spectrometry into raw data raw Perform conversion processing to obtain the converted single-cell mass spectrometry flow data Data trans ; The conversion processing formula is:

[0048] Data trans =sinh -1 (Data raw / 10);

[0049] Normalization processing unit, using z-score method to transform the single cell mass spectrometry data Data trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization ;

[0050] Peripheral blood immune cell clustering acquisition unit uses a pre-trained 10*10 self-organizing network SOM-net 1 to normalize the single-cell mass spectrometry data after normalization normalization Classify and obtain immune cell clustering data of peripheral blood samples PBMC-1 ; Among them, the steps to obtain SOM-net 1 are as follows:

[0051] a) 60 peripheral blood samples from 10 patients with advanced hepatocellular carcinoma were selected, and the expression of each metal isotope-coupled protein-specific biomarker in each cell was used as the original feature data Data raw-train ;

[0052] b) Convert the original feature data of each sample to obtain Data trans-train ,Data trans-train =sinh -1 (Data raw-train / 10);

[0053] c) Convert the processed data of each sample trans-train Each biomarker channel data in is normalized using the z-score method to obtain normalized high-dimensional feature data Data normalization-train ;

[0054] d) On average, 4000 cells are randomly selected from each sample, for a total of 240,000 cells, to obtain the selected normalized high-dimensional feature data Data normalization-train-selected , perform 10*10 self-organizing network training on it, and obtain 10*10 self-organizing network SOM-net 1;

[0055] Filtering unit, used to filter immune cell clustering data from peripheral blood samples PBMC-1 CD15 + CD16 +Granulocytes (this step is not required if the Ficoll separation medium gradient centrifugation method is used) and abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment), and data containing lymphocytes, dendritic cells and monocytes are obtained. PBMC-2 ;

[0056] The peripheral blood immune cell secondary clustering acquisition unit uses the pre-trained 10*10 self-organizing network SOM-net2 to PBMC-2 Classify and obtain the secondary grouping data of immune cells in peripheral blood samples som-cluster ; Among them, the steps to obtain SOM-net 2 are as follows:

[0057] a) The normalized high-dimensional feature data Data obtained from 60 peripheral blood samples of 10 patients with advanced hepatocellular carcinoma normalization-train Substitute into SOM-net 1 to obtain immune cell clustering;

[0058] b) Based on the immune cell clustering information, filter out CD15 + CD16 + Granulocytes (this step is not required if the Ficoll separation medium gradient centrifugation method is used) and abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment), and data containing lymphocytes, dendritic cells and monocytes are obtained. PBMC-2-train ;

[0059] c) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data Data PBMC-2-train-selected , perform 10*10 self-organizing network training on it, and obtain 10*10 self-organizing network SOM-net 2;

[0060] Clustering unit, using hierarchical clustering method to secondary cluster data of immune cells in peripheral blood samples som-cluster Merge clustering was performed to obtain a peripheral blood immune cell atlas Data containing 27 T cell subsets, 3 B cell subsets, 2 NK cell subsets, 1 dendritic cell subset and 5 monocyte subsets, a total of 38 immune cell subsets immune-cell ;

[0061] Feature vector acquisition unit, based on the peripheral blood immune cell atlas Data immune-cell , calculate the Spearman correlation coefficient between the proportion of T02, T05, T10, T13, T15 and T27 immune cell subsets in samples of different treatment cycles of the patients to be diagnosed and the treatment cycle to form the characteristic vector R;

[0062] The treatment effectiveness evaluation unit combines the feature vector R and the pre-obtained feature vector R N Hierarchical clustering was performed, and those who were effective were clustered into the same branch, and those who were ineffective were clustered into another branch. The effectiveness of anti-PD-1 monotherapy was evaluated based on the clustering results. Among them, the feature vector R obtained in advance by the treatment effectiveness evaluation unit N is the feature vector R of 10 patients with advanced hepatocellular carcinoma for whom the efficacy of anti-PD-1 monotherapy is known N (N=1, 2, 3, ..., 10).

[0063] The effectiveness evaluation of anti-PD-1 monotherapy using the above system includes the following steps:

[0064] 1) Separating mononuclear cells from the peripheral blood of the patient to be diagnosed by Ficoll separation solution gradient centrifugation method or ACK lysis solution centrifugation method to obtain single cells of peripheral blood samples; wherein the peripheral blood samples are peripheral blood samples of the patient to be diagnosed from baseline to 5 to 6 treatment cycles of anti-PD-1 monotherapy, preferably, a peripheral blood sample is collected once in each treatment cycle;

[0065] 2) Use the pre-designed metal isotope-coupled protein-specific biomarkers (Table 2) to perform mass spectrometry flow cytometry on single cells in peripheral blood samples to obtain the single-cell mass spectrometry raw data of the samples. raw ;

[0066] 3) Single cell mass spectrometry raw data of samples raw Perform conversion processing to obtain the converted single-cell mass spectrometry flow data Data trans ,Data trans =sinh -1 (Data raw / 10);

[0067] 4) Use the z-score method to analyze the converted single-cell mass spectrometry data trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization ;

[0068] 5) Use the pre-trained 10*10 self-organizing network SOM-net 1 to normalize the single-cell mass spectrometry data Data normalization Classify and obtain immune cell clustering data of peripheral blood samples PBMC-1 ; Among them, the steps to obtain SOM-net 1 are as follows:

[0069] a) 60 peripheral blood samples from 10 patients with advanced hepatocellular carcinoma were selected, and the expression of each metal isotope-coupled protein-specific biomarker in each cell was used as the original feature data Data raw-train ;

[0070] b) Convert the original feature data of each sample to obtain Data trans-train ,Data trans-train =sinh -1 (Data raw-train / 10);

[0071] c) Convert the processed data of each sample trans-train Each biomarker channel data in is normalized using the z-score method to obtain normalized high-dimensional feature data Data normalization-train ;

[0072] d) On average, 4000 cells are randomly selected from each sample, for a total of 240,000 cells, to obtain the selected normalized high-dimensional feature data Data normalization-train-selected , perform 10*10 self-organizing network training on it, and obtain 10*10 self-organizing network SOM-net 1;

[0073] 6) Filter the immune cell clustering data of peripheral blood samples PBMC-1 CD15 + CD16 + Granulocytes (if the Ficoll separation medium gradient centrifugation method is used, no need to filter out) and abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment) are obtained to obtain data including lymphocytes, dendritic cells and monocytes. PBMC-2 ;

[0074] 7) Use the pre-trained 10*10 self-organizing network SOM-net 2 to train Data PBMC-2 Classify and obtain the secondary grouping data of immune cells in peripheral blood samples som-cluster ; Among them, the steps to obtain SOM-net 2 are as follows:

[0075] a) The normalized high-dimensional feature data Data obtained from 60 peripheral blood samples of 10 patients with advanced hepatocellular carcinoma normalizalion-train Substitute into SOM-net 1 to obtain immune cell clustering;

[0076] b) Based on the immune cell clustering information, filter out CD15 + CD16 +Granulocytes (if the Ficoll separation medium gradient centrifugation method is used, no need to filter out) and abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment) are obtained to obtain data including lymphocytes, dendritic cells and monocytes. PBMC-2-train ;

[0077] c) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data Data PBMC-2-train-selected , perform 10*10 self-organizing network training on it, and obtain 10*10 self-organizing network SOM-net 2;

[0078] 8) Use hierarchical clustering method to secondary cluster data of immune cells in peripheral blood samples som-cluster Merge clustering was performed to obtain a peripheral blood immune cell atlas Data containing 27 T cell subsets, 3 B cell subsets, 2 NK cell subsets, 1 dendritic cell subset and 5 monocyte subsets, a total of 38 immune cell subsets immune-cell ;

[0079] 9) According to the peripheral blood immune cell atlas Data immune-cell , calculate the Spearman correlation coefficient between the proportion of T02, T05, T10, T13, T15 and T27 immune cell subsets in samples of different treatment cycles of the patients to be diagnosed and the treatment cycle to form the characteristic vector R;

[0080] 10) Combine the eigenvector R and the pre-obtained eigenvector R N Hierarchical clustering was performed, and those who were effective were clustered into the same branch, and those who were ineffective were clustered into another branch. The effectiveness of anti-PD-1 monotherapy was evaluated based on the clustering results. Among them, the feature vector R obtained in advance by the treatment effectiveness evaluation unit N is the feature vector R of 10 patients with advanced hepatocellular carcinoma for whom the efficacy of anti-PD-1 monotherapy is known N (N=1, 2, 3, ..., 10).

[0081] Example 1

[0082] 1) Peripheral blood samples were obtained from 10 patients with advanced hepatocellular carcinoma who received PD-1 immunotherapy (anti-PD-1) monotherapy from baseline (i.e., before the first immunotherapy) to 5 to 6 cycles of anti-PD-1 monotherapy (Table 1). Among them, 5 patients (N1 to N5) had not received other treatments before immunotherapy, and their peripheral blood samples were processed by Ficoll separation medium gradient centrifugation. The other 5 patients (C1 to C5) had undergone chemotherapy or systemic systemic treatment before immunotherapy, and their peripheral blood samples were processed by ACK lysis solution centrifugation to obtain cell precipitates, i.e., mononuclear cells.

[0083] Table 1

[0084]

[0085] 2) Use the pre-designed metal isotope-coupled protein-specific biomarkers (Table 2) to perform mass spectrometry flow cytometry (CyTOF) detection experiments on peripheral blood cell sediments to obtain the single-cell mass spectrometry raw data of the samples. raw :

[0086]

[0087] Where m is the number of metal isotope-coupled protein-specific biomarkers (Table 2), n is the number of cells in a single sample, and α ij is the expression level of the metal isotope coupled protein-specific biomarker in the jth cell in a single sample.

[0088] Among them, the CyTOF detection experimental process is as follows Figure 1 As shown, the specific steps are as follows:

[0089] a) Count the cells in the cell pellet obtained in step 1) and take 1-3×10 6 Cell number;

[0090] b) Prepare a 194Pt (1mM) live / death stain solution with a final concentration of 0.25μM using phosphate buffered saline (PBS) (pH7.4), take 50μL to 1.5mL of the 194Pt live / death stain solution to resuspend the cells, preferably 100μL, and stain on ice for 5 minutes to distinguish live / death cells in subsequent data analysis;

[0091] c) Add 100 μL to 1.5 mL FACS Buffer to each sample, preferably 500 μL, resuspend the cells, centrifuge at 400 g / 5 min at 2-8°C, and discard the supernatant;

[0092] d) Add 20 μL to 1.5 mL of blocking solution to each sample, preferably 50 μL, resuspend the cells, and block on ice for 20 min;

[0093] e) adding a mixture of metal isotope-coupled protein-specific biomarkers (Table 2) (0.5 to 4 μL for each metal antibody) and staining on ice for 20 to 60 minutes to perform extracellular staining on the sample;

[0094] f) Add 100 μL to 1.5 mL FACS Buffer to each sample, preferably 1 mL, resuspend the cells, centrifuge at 400 g / 5 min at 2-8°C, discard the supernatant, and repeat 2-3 times;

[0095] g) Prepare a final concentration of 100-500 nM Ir dye solution with Fix and Perm Buffer, take 200 μL-1.5 mL of each sample to resuspend the cells, incubate at room temperature for 1 hour to stain and fix the DNA;

[0096] h) Add 100 μL to 1.5 mL FACS Buffer to each sample, preferably 1 mL, resuspend the cells, centrifuge at 800 g / 5 min at 2-8°C, and discard the supernatant;

[0097] i) Add 0.5-1 mL ddH2O to each sample to resuspend the cells and transfer to a 5 mL filter tube (12*75 mm) using a pipette, and filter 1-2 times;

[0098] j) The filtered cell suspension was detected by mass spectrometry flow cytometry.

[0099] Table 2

[0100]

[0101] 3) Data raw To perform the conversion:

[0102] Data trans =sinh -1 (Data raw / 10)

[0103] Data trans Represents single-cell mass cytometry data after transformation.

[0104] 4) Data trans Each biomarker channel data in is normalized using the z-score method to obtain Data normalization :

[0105]

[0106]

[0107] Among them, Datanormalization represents the single-cell mass spectrometry data after z-score normalization, m is the number of metal isotope-coupled protein-specific biomarkers (Table 2), n is the number of cells in a single sample, and α ij is the specific biomarker expression of the metal isotope coupling protein of the i-th cell in a single sample, is the mean expression value of the i-th metal isotope coupling protein-specific biomarker in a single sample, σ j is the standard deviation of the expression value of the i-th metal isotope coupled protein-specific biomarker in a single sample, β ij is the normalized value of the expression level of the metal isotope coupled protein-specific biomarker in the jth cell in a single sample.

[0108] 5) Use the pre-trained 10*10 self-organizing network SOM-net 1 to train Data normalization Classify and obtain immune cell clustering data of peripheral blood samples PBMC-1 .

[0109] The steps to obtain SOM-net 1 are as follows:

[0110] a) 60 peripheral blood samples from 10 patients with advanced hepatocellular carcinoma were selected (Table 1), and the expression of each metal isotope-coupled protein-specific biomarker in each cell was used as the original feature data Data raw-train ;

[0111] b) Convert the original feature data of each sample to obtain Data trans-train ,Data trans-train =sinh -1 (Data raw-train / 10);

[0112] c) Convert the processed data of each sample trans-train Each biomarker channel data in is normalized using the z-score method to obtain normalized high-dimensional feature data Data normalization-train ;

[0113] d) On average, 4000 cells are randomly selected from each sample, for a total of 240,000 cells, to obtain the selected normalized high-dimensional feature data Data normalization-train-selected , and perform 10*10 self-organizing network training on it to obtain 10*10 self-organizing network SOM-net 1.

[0114] 6) Filter out CD15 + CD16 +Granulocytes (this step is not required if the Ficoll separation medium gradient centrifugation method is used), as well as abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment), to obtain data including lymphocytes, dendritic cells and monocytes PBMC-2 .

[0115] 7) Use the pre-trained 10*10 self-organizing network SOM-net 2 to train Data PBMC-2 Classify and obtain the secondary grouping data of immune cells in peripheral blood samples som-cluster .

[0116] To obtain SOM-net 2, follow these steps:

[0117] a) The normalized high-dimensional feature data Data obtained from 60 peripheral blood samples of 10 patients with advanced hepatocellular carcinoma (Table 1) normalization-train Substitute into SOM-net 1 to obtain immune cell clustering;

[0118] b) Based on the immune cell clustering information, filter out CD15 + CD16 + Granulocytes (this step is not required if the Ficoll separation medium gradient centrifugation method is used), as well as abnormal cell populations (cell populations with abnormal protein-specific biomarker combinations, which may be abnormally stained cells that were not filtered out in the previous pretreatment), to obtain data including lymphocytes, dendritic cells and monocytes PBMC-2-train ;

[0119] c) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data Data PBMC-2-train-selected , and perform 10*10 self-organizing network training on it to obtain 10*10 self-organizing network SOM-net 2.

[0120] 8) Then use the hierarchical clustering method to som-cluster Merge clustering was performed to obtain a peripheral blood immune cell atlas Data containing 27 T cell subsets, 3 B cell subsets, 2 NK cell subsets, 1 dendritic cell subset and 5 monocyte subsets, a total of 38 immune cell subsets immune-cell (like Figure 2 shown):

[0121]

[0122] β ij is the normalized value of the expression of the metal isotope coupling protein-specific biomarker in the jth cell in a single sample, β cj(j=1, 2, 3, ..., n) is the cell subpopulation to which the j-th cell in a single sample belongs.

[0123] 9) For a single sample, according to its respective Data immune-cell Calculate the proportion of each immune cell subset q k (k=1, 2, 3,..., 38), q k =Total number of cells belonging to immune cell subset k in a single sample / total number of cells in a single sample; for a single patient, calculate the Spearman correlation coefficient r between the proportion of each immune cell subset in samples of different treatment cycles and the treatment cycle k .

[0124] 10) According to the clinical judgment criteria, 10 patients with advanced hepatocellular carcinoma who received anti-PD-1 monotherapy (Table 1) were divided into patients who benefited from immunotherapy (DCB) and patients who did not benefit from immunotherapy (NDB). k Unpaired sample T test was performed with DCB and NDB as groups; r k The coefficient in the DCB group is significantly higher than that in the NDB group, such as Figure 3 Therefore, the r values ​​of the designated immune cell subsets T02, T05, T10, T13, T15 and T27 are selected to form the feature vector R (r2, r5, r 10 , r 13 , r 15 , r 27 ).

[0125] 11) The feature vector R of 10 patients with advanced hepatocellular carcinoma who received anti-PD-1 monotherapy N (N=1, 2, 3, ..., 10) hierarchical clustering was performed and the clustering results were observed. Patients C1 and N5 were effective patients and were clustered into the same branch. The rest were ineffective patients and were clustered into another branch (e.g. Figure 5 shown).

[0126] 12) Taking N2 and C3 patients as an example, it can be seen that compared with imaging detection of changes in tumors in patients with hepatocellular carcinoma, the dynamic peripheral blood detection method proposed by the system of the present invention can determine earlier whether patients who receive anti-PD-1 monotherapy are ineffective (such as Figure 4 and Figure 6 shown).

Claims

1. A system for evaluating the effectiveness of anti-PD-1 monotherapy based on peripheral blood immune cell profiles, characterized in that: include: A peripheral blood single cell acquisition unit is used to separate mononuclear cells from the peripheral blood of the patient to be diagnosed to obtain single cells of the peripheral blood sample; wherein the peripheral blood sample is a peripheral blood sample of the patient to be diagnosed from the baseline to 5 to 6 cycles of anti-PD-1 monotherapy; The mass spectrometry flow cytometry detection unit uses pre-designed metal isotope-coupled protein-specific biomarkers to perform mass spectrometry flow cytometry experiments on single cells in peripheral blood samples to obtain the sample's single-cell mass spectrometry raw data Data raw ; The raw data conversion unit is used to convert the raw data of single cell mass spectrometry into raw data raw Perform conversion processing to obtain the converted single-cell mass spectrometry flow data Data trans ; Among them, the conversion processing formula is: Data trans =sinh -1 (Data raw / 10); Normalization processing unit, used to convert the single cell mass spectrometry data trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization ; Peripheral blood immune cell clustering acquisition unit uses a pre-trained 10*10 self-organizing network SOM-net 1 to normalize the single-cell mass spectrometry data after normalization. normalization Classify and obtain immune cell clustering data of peripheral blood samples PBMC-1 ; Filtering unit, used to filter immune cell clustering data from peripheral blood samples PBMC-1 CD15 + CD16 + Granulocytes and abnormal cell populations, including lymphocytes, dendritic cells and monocytes Data PBMC-2 ; The peripheral blood immune cell secondary clustering acquisition unit uses the pre-trained 10*10 self-organizing network SOM-net 2 to PBMC-2 Classify and obtain the secondary grouping data of immune cells in peripheral blood samples som-cluster ; Clustering unit, secondary clustering data of immune cells in peripheral blood samples som-cluster Merge and cluster to obtain the peripheral blood sample immune cell subset information, i.e., the peripheral blood immune cell atlas Data immune-cell ; Feature vector acquisition unit, based on the peripheral blood immune cell atlas Data immune-cell , calculate the Spearman correlation coefficient between the proportion of T02, T05, T10, T13, T15 and T27 immune cell subsets in samples of different treatment cycles of the patients to be diagnosed and the treatment cycle to form the characteristic vector R; The treatment effectiveness evaluation unit combines the feature vector R and the pre-obtained feature vector R N Hierarchical clustering was performed, and those who responded to the treatment were clustered into the same branch, and those who did not respond to the treatment were clustered into another branch. The effectiveness of anti-PD-1 monotherapy was evaluated based on the clustering results.

2. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The peripheral blood single cell acquisition unit uses Ficoll separation solution gradient centrifugation method or ACK lysis solution centrifugation method to separate mononuclear cells from the peripheral blood of the patient to be diagnosed, and obtains single cells of the peripheral blood sample.

3. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The peripheral blood single cell acquisition unit collects peripheral blood samples from the undiagnosed patients once every treatment cycle from baseline to 5 to 6 cycles of anti-PD-1 monotherapy.

4. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The metal isotope-coupled protein-specific biomarkers pre-designed for the mass spectrometry flow detection unit are shown in the following table:

5. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The normalization processing unit uses the z-score method to normalize the converted single-cell mass spectrometry data. trans Each biomarker channel data is normalized to obtain the normalized single-cell mass spectrometry data Data normalization .

6. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The steps for obtaining the peripheral blood immune cell clustering acquisition unit SOM-net 1 are as follows: a) 60 peripheral blood samples from 10 patients with advanced hepatocellular carcinoma were selected, and the expression of each metal isotope-coupled protein-specific biomarker in each cell was used as the original feature data Data raw-train ; b) Convert the original feature data of each sample to obtain Data trans-train ,Data trans-train =sinh -1 (Data raw-train / 10); c) Convert the processed data of each sample trans-train Each biomarker channel data is normalized separately to obtain normalized high-dimensional feature data Data normalization-train ; d) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data Data normalization-train-selected , and perform 10*10 self-organizing network training on it to obtain 10*10 self-organizing network SOM-net 1.

7. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: Step c) Convert the processed data of each sample to Data trans-train Each biomarker channel data in is normalized using the z-score method to obtain normalized high-dimensional feature data Data normalization-train .

8. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The steps for obtaining the peripheral blood immune cell secondary clustering acquisition unit SOM-net 2 are as follows: a) The normalized high-dimensional feature data Data obtained from 60 peripheral blood samples of 10 patients with advanced hepatocellular carcinoma normalization-train Substitute into SOM-net 1 to obtain immune cell clustering; b) Based on the immune cell clustering information, filter out CD15 + CD16 + Granulocytes and abnormal cell populations, including lymphocytes, dendritic cells and monocytes Data PBMC-2-train ; c) Randomly select 4000 cells from each sample on average, a total of 240,000 cells, and obtain the selected normalized high-dimensional feature data Data PBMC-2-train-selected , and perform 10*10 self-organizing network training on it to obtain the 10*10 self-organizing network SOM-net2.

9. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The clustering unit uses hierarchical clustering method to secondary cluster data of immune cells in peripheral blood samples som-cluster Merge clustering was performed to obtain a peripheral blood immune cell atlas Data containing 27 T cell subsets, 3 B cell subsets, 2 NK cell subsets, 1 dendritic cell subset and 5 monocyte subsets, a total of 38 immune cell subsets immune-cell .

10. The anti-PD-1 monotherapy effectiveness evaluation system based on peripheral blood immune cell profile according to claim 1, characterized in that: The feature vector R obtained in advance by the treatment effectiveness evaluation unit N is the feature vector R of 10 patients with advanced hepatocellular carcinoma for whom the efficacy of anti-PD-1 monotherapy is known N (N=1, 2, 3, ..., 10).

Images