Application of serum Raman spectroscopy in rapid and early identification of aplastic anemia and myelodysplastic syndrome

Abstract

The present invention provides a rapid, non-invasive, and antibody-free method for identifying AA and MDS based on Raman spectroscopy. This method utilizes the peak intensity of Raman spectra as a biomarker for the preparation of products for predicting or detecting blood disorders. The specificity and significant Raman shifts of the present invention are valuable biomarkers for early disease identification, improving diagnostic efficiency, reducing testing costs, and promoting the rapid and accurate diagnosis of AA and MDS.

Description

Technical Field

[0001] The present application relates to the field of medical diagnosis, and in particular to markers for early differentiation and / or detection of aplastic anemia and myelodysplastic syndrome. Background Art

[0002] Acquired bone marrow failure syndrome (BMF) is a group of bone marrow hematopoietic disorders caused by abnormal immune function or hematopoietic stem cell quality. It can occur in people of all ages and includes aplastic anemia (AA), myelodysplastic syndrome (MDS), paroxysmal nocturnal hemoglobinuria (PNH), idiopathic cytopenia of undetermined significance (ICUS), and immune-related pancreatic cytopenia (IRP).

[0003] Among them, aplastic anemia (AA) is characterized by decreased proliferation of bone marrow hematopoietic cells and pancytopenia in peripheral blood. Clinically, AA is divided into subtypes such as non-severe aplastic anemia (NSAA), severe AA (SAA), and very severe aplastic anemia (VSAA). Traditional detection methods for AA mainly include blood and bone marrow examinations and cytological examinations.

[0004] Myelodysplastic syndrome (MDS) is characterized by ineffective hematopoiesis, refractory cytopenias, and hematopoietic failure, with a high risk of transformation to acute myeloid leukemia (AML). MDS is clinically divided into subtypes: refractory anemia (RA), refractory anemia with ringed sideroblasts (RAS), refractory anemia with excess blasts (RAEB), and refractory anemia in transformation (RAEB-t). RAEB is divided into RAEB1 and RAEB2, and patients are at risk of bone marrow failure and transformation to AML. Traditional testing for MDS primarily includes blood and bone marrow examinations, cytogenetic testing, immunophenotyping, and genetic analysis. There is currently no "gold standard" for the diagnosis of MDS, and it is a "diagnosis of exclusion." Because its symptoms are similar to those of MDS and AA, with anemia, bleeding, and infection as the main clinical manifestations, the differential diagnosis of MDS from AA requires the support of test results from multiple platforms.

[0005] Raman spectroscopy has been used for rapid, non-invasive and label-free identification of hematological diseases. Tsui et al. found that the Raman spectroscopy at 1637, 1585 and 1372 cm -1 The hemoglobin-related Raman band at the sac has a higher amplitude in atypical cells such as echinocytes and acanthocytes, indicating biochemical abnormalities of red blood cells. Some studies have demonstrated the ability of Raman spectroscopy to distinguish exosomes in the process from monoclonal gammopathy (MGUS) to asymptomatic myeloma (aMM) and symptomatic myeloma (sMM), thus providing useful clinical indications for patient care. Other studies have shown that there is a certain degree of distinction or correlation between MM subtype plasma cells and differentiation clusters (CD45+ / CD38+ / CD138-) and (CD45- / CD38+ / CD138+). Based on a full understanding of the existing technology, there is currently no research and application of Raman spectroscopy for serum analysis of different types of AA and MDS, and there is also a lack of in-depth discussion of serological indicators related to lipid metabolism in patients with AA and MDS.

[0006] Refractory anemia-type MDS is easily confused with atypical aplastic anemia, particularly the RAEB subtype. Current testing methods are time-consuming and expensive due to the complexity of the testing. Furthermore, traditional methods for diagnosing BMF are time-consuming and expensive, require invasive bone marrow sampling, and lack a gold standard for diagnosis for some disease types. Therefore, developing a low-cost diagnostic method that can facilitate early identification of BMF is crucial for the diagnosis of hematologic diseases. Summary of the Invention

[0007] In view of this, the present invention aims to propose an application of serum Raman spectroscopy in the rapid and early identification of aplastic anemia and myelodysplastic syndrome, so as to overcome the problems of existing detection methods that are time-consuming, costly and invasive.

[0008] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0009] Based on Raman spectroscopy, the Raman peaks representing nucleic acids in peripheral blood were detected at 726, 781, 786, 1078, 1190, and 1415 cm -1 , representing the protein Raman peak 1221cm -1 , representing the Raman peak of β-carotene at 1162 cm -1 , representing the Raman peak of phospholipids / cholesterol at 1285 cm -1 and the Raman peaks representing lipids at 1437 and 1446 cm -1 The application of the characteristic peak intensity level as a biomarker in the preparation of a product for predicting or detecting acquired bone marrow failure syndrome.

[0010] Furthermore, by detecting the peak intensity of Raman peak biomarkers in the peripheral blood of the subjects and comparing it with the average value of normal controls, the progression of the disease can be diagnosed and the condition can be assessed.

[0011] Furthermore, the acquired bone marrow failure syndrome is aplastic anemia and / or myelodysplastic syndrome.

[0012] The Raman peaks of representative proteins in peripheral blood were detected based on Raman spectroscopy at 869, 1221, and 1260 cm -1 and the Raman peak representing phospholipids / cholesterol at 1285 cm -1 The invention discloses an application of the characteristic peak intensity level as a biomarker in the preparation of a product for typing acquired bone marrow failure syndrome.

[0013] Furthermore, the method further includes detecting the Raman peak 1344 cm representing collagen in peripheral blood based on Raman spectroscopy. -1 and the Raman peak at 726 cm representing nucleic acids -1 The characteristic peak intensity level of the protein was used as a biomarker.

[0014] Furthermore, the product for typing acquired bone marrow failure syndrome is a product for identifying different subtypes of aplastic anemia. Based on Raman spectroscopy, the Raman peaks 1003 and 1206 cm representing proteins in peripheral blood are detected. -1 The invention discloses an application of the characteristic peak intensity level as a biomarker in the preparation of a product for typing acquired bone marrow failure syndrome.

[0015] Furthermore, the method further includes detecting the Raman peak 1344 cm representing collagen in peripheral blood based on Raman spectroscopy. -1 The peaks representing lipids are at 1437 and 1443 cm -1 , representing the peaks of nucleic acids at 726 and 786 cm -1 and 1162 cm representing β-carotene -1 The Raman peak is used as a biomarker. Furthermore, the product for typing acquired bone marrow failure syndrome is a product for identifying different subtypes of myelodysplastic syndrome.

[0016] Furthermore, the Raman spectroscopy detection conditions are as follows: 785 nm laser as excitation light, 40x objective lens, output power of 10 mW, shooting process using a ×40 0.6NA Nikon lens, the sample is illuminated by a laser beam with an output power of 10 mW within a spot diameter of approximately 1.596 μm, a single integration time of 250 s, the number of integrations is one, and the measurement range is 600–1800 cm -1 , 5-10 sites were measured in each group with a resolution of 1cm -1 .

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] The significance of this study lies in establishing a clinical model using a noninvasive method, involving peripheral blood collection only, to differentiate between healthy controls, patients with aplastic anemia (AA), and patients with myelodysplastic syndrome (MDS), as well as between healthy controls and different subtypes of patients with aplastic anemia (AA), and between healthy controls and different subtypes of patients with myelodysplastic syndrome (MDS). Furthermore, we screened Raman peaks that contribute significantly to disease classification as biomarkers for aplastic anemia (AA) and myelodysplastic syndrome (MDS), laying the foundation for better utilization of clinical serological test data for the rapid and early differentiation of AA and MDS. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of Raman measurement of serum;

[0020] Figure 2 is the serum Raman spectrum. Figure 2 a is the Raman spectra of the healthy control group, AA and MDS groups. Figure 2 b is the Raman spectra of the healthy control group, NSAA, SAA, and VSAA. Figure 2 c shows the Raman spectra of the healthy control group, MDS-RAEB1 and MDS-RAEB2;

[0021] Figure 3 OPLS-DA identification of serum samples from healthy controls and acquired bone marrow failure syndrome, Figure 3 a is the OPLS-DA score plot, Figure 3 b is OPLS-DA loading plot, Figure 3 c is OPLS-DA V+S plot, Figure 3 d is the OPLS-DA score plot of the healthy control group vs. the AA group. Figure 3 e is the loading plot of the healthy control group vs. the AA group, and Figure 3f is the V+S plot of the healthy control group vs. the AA group. Figure 3 g is the OPLS-DA scoreplot of the healthy control group vs the MDS group, and Figure 3h is the loading plot of the healthy control group vs the MDS group. Figure 3 i is the V+Splot of the healthy control group vs the MDS group, Figure 3 j is the OPLS-DA score plot of AA group vs MDS group, Figure 3 k is the loading plot of AA group vs MDS group, Figure 3 l is the V+S plot of AA group vs MDS group;

[0022] Figure 4 OPLS-DA identification of healthy control group and AA subtype serum samples, among which, Figure 4 a. Healthy control group, NSAA, SAA and VSAA discriminant scores plotted by OPLS-DA using Hotelling's 95% confidence ellipse; Figure 4 b. Healthy control group, NSAA, SAA and VSAA discrimination loading line plot; Figure 4 c. Healthy control group, NSAA, SAA and VSAA identification V+S diagram; Figure 4 d. OPLS-DA identification score plot of healthy control group and NSAA drawn with Hotelling's 95% confidence ellipse; Figure 4 e. Loading line diagram of the healthy control group and NSAA; Figure 4 f. V+S diagram of healthy control group and NSAA; Figure 4 g. Healthy control group and SAA identified by OPLS-DA score map drawn with Hotelling's 95% confidence ellipse; Figure 4 h. Loading line diagram of healthy control group and SAA; Figure 4 i. V+S diagram of healthy control group and SAA; Figure 4j. Score plot of healthy control group and VSAA identified by OPLS-DA using Hotelling's 95% confidence ellipse; Figure 4 k. Loading line diagram of healthy control group and VSAA discrimination; Figure 4 l. V+S diagram of healthy control group and VSAA identification;

[0023] Figure 5 OPLS-DA identification of serum samples from healthy controls and MDS subtypes, Figure 5 a. OPLS-DA discrimination score plot of healthy control group, MDS-RAEB1 and MDS-RAEB2 drawn with Hotelling's 95% confidence ellipse; Figure 5 b. Healthy control group, MDS-RAEB1 and MDS-RAEB2 differential loading line diagram; Figure 5 c. Healthy control group, V+S diagram for differentiation between MDS-RAEB1 and MDS-RAEB2; Figure 5 d. OPLS-DA identification score diagram of healthy control group and MDS-RAEB1 drawn with Hotelling's 95% confidence ellipse; Figure 5e. Loading line diagram of healthy control group and MDS-RAEB1 identification; Figure 5 f. V+S diagram of the healthy control group and MDS-RAEB1 differentiation; Figure 5 g. Healthy control group and MDS-RAEB2 identified by OPLS-DA score map drawn with Hotelling's 95% confidence ellipse; Figure 5 h. Loading line diagram of the healthy control group and MDS-RAEB2; Figure 5 i. Identification of V+S between healthy controls and MDS-RAEB2;

[0024] Figure 6 shows the statistics of potential biomarkers and serum biochemical results in healthy controls vs AA vs MDS; Figure 6a Figure 6b-6g shows the statistical analysis of the Raman characteristic peak positions with VIP>1.0 for the four models: healthy control group vs. AA group vs. MDS group, healthy control group vs. AA group, healthy control group vs. MDS group, and AA group vs. MDS group. Figures 6b-6g show the six peripheral blood biochemical indicators of total protein, glucose, triglycerides, total cholesterol, high-density lipoprotein, and low-density lipoprotein in the healthy control group, AA group, and MDS group.

[0025] Figure 7 This is a validation analysis of the healthy control group vs different AA subtypes, in which Figure 7 a is the statistical analysis of the Raman characteristic peak positions with VIP>1.0 of the four models: healthy control group vs AA subtype, healthy control group vs NSAA, healthy control group vs SAA, and healthy control group vs VSAA; Figure 7 b- Figure 7 g is the six peripheral blood biochemical indicators of TP, glucose, TG, TC, HDL and LDL in the healthy control group, NSAA, SAA and VSAA groups;

[0026] Figure 8 shows the validation analysis of the healthy control group vs different MDS subtypes, where Figure 8a Statistical analysis of Raman characteristic peak positions with VIP>1.0 for the three models: healthy control group vs MDS subtype, healthy control group vs MDS-RAEB1, and healthy control group vs MDS-RAEB2; Figure 8b- Figure 8g Six peripheral blood biochemical indicators of TP, glucose, TG, TC, HDL and LDL in the healthy control group, MDS-RAEB1 and MDS-RAEB2 groups;

[0027] Figure 9 shows the establishment and validation of the identification model for serum samples in the control and BMF groups; Figure 9a For cluster analysis, Figure 9b For Permutation analysis, Figure 9c-9d A dot plot showing both the training set and prediction set samples, where Figure 9c External validation of the OPLS model for healthy controls vs AA. Figure 9d This was an external validation of the OPLS model for healthy controls vs MDS;

[0028] Figure 10 External validation of the OPLS model for healthy controls vs different AA subtypes or different MDS subtypes, Figure 10 a. External validation of the OPLS model in healthy controls vs. NSAA. Figure 10 b. External validation of the OPLS model for healthy control group vs SAA; Figure 10c. External validation of the OPLS model for healthy control group vs VSAA; Figure 10 d. External validation of the OPLS model in healthy controls vs. MDS-RAEB1; Figure 10 e. External validation of the OPLS model in healthy controls vs MDS-RAEB2. DETAILED DESCRIPTION

[0029] Unless otherwise defined, the technical terms used in the following examples have the same meanings as commonly understood by those skilled in the art to which this invention belongs. The experimental reagents used in the following examples, unless otherwise specified, are conventional biochemical reagents; the experimental methods described, unless otherwise specified, are conventional methods.

[0030] The present invention will be described in detail below with reference to the embodiments and accompanying drawings.

[0031] A total of 83 patients who visited the Hematology Hospital of the Chinese Academy of Medical Sciences (Institute of Hematology, Chinese Academy of Medical Sciences) in 2021 were recruited, including 38 males and 45 females, ranging in age from 1 to 72 years. They were divided into three groups: 35 patients in the AA group, including 17 patients with NSAA, 11 patients with SAA, and 7 patients with VSAA; 25 patients with MDS, including 14 patients with MDS-RAEB1 and 11 patients with MDS-RAEB2; and 23 patients in the healthy control group.

[0032] All patients in the experimental group underwent blood and bone marrow examinations, cytogenetic analysis, immunophenotyping, and genetic analysis, and were confirmed by experienced hematologists. This study was approved by the Ethics Committee of the Blood Disease Hospital, Chinese Academy of Medical Sciences (Institute of Hematology, Chinese Academy of Medical Sciences). Serum samples used in the study were obtained from residual clinical samples collected from patients with BMF and controls. All BMF patients and controls underwent routine serum biochemistry testing. Serum biochemistry data were obtained from the Clinical Testing Center of the Blood Disease Hospital, Chinese Academy of Medical Sciences (Institute of Hematology, Chinese Academy of Medical Sciences).

[0033] 1. Experimental methods

[0034] 1.1 Collection of relevant serological indicators

[0035] The subjects fasted for 10 hours and collected serum. The peripheral blood total protein (TP), glucose, triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL) and low-density lipoprotein (LDL) were measured using an automatic biochemical analyzer.

[0036] Clinical data of AA, MDS, and control groups (normally distributed data are expressed as x ± s [mean ± standard deviation (SD)], and mean values between groups are compared using one-way analysis of variance; non-normally distributed data are expressed as M (Q1-Q3) [median (25th-75th percentile)], and the Kruskal-Wallis test is used for comparison between groups. P values are used to compare clinical data results between AA, MDS, and control groups. P < 0.05 is considered statistically significant):

[0037]

[0038] 1.2 Peripheral serum Raman spectroscopy analysis

[0039] 5 μL of serum was dropped onto a quartz glass slide and measured using a confocal Raman spectrometer, an XploRA Raman microscope. A 785 nm laser with a 10 mW output power was used as the excitation light. A 40x objective lens was selected, and the specimen was fixed on an XYZ three-dimensional stage. A ×40 0.6 NA Nikon lens was used for imaging. A laser beam with a 10 mW output power was applied to a spot diameter of approximately 1.596 μm on the sample. The single integration time was 250 s, the number of integrations was one, and the measurement range was 600–1800 cm. -1 , 5-10 sites were measured in each group with a resolution of 1cm -1 At the same time, the Raman spectrum of the quartz glass was measured as the background. Labspec6 software was used for data processing such as smoothing and baseline correction. All spectra were analyzed at their respective 1450 cm -1 The Raman peaks were intensity normalized to the internal standard.

[0040] 1.3 Establishment of OPLS-DA diagnostic model based on Raman spectroscopy data analysis

[0041] Raman spectra were obtained by Raman spectroscopy, and the data sets were analyzed by multivariate statistical analysis using SIMCA 14.1.

[0042] Supervised orthogonal partial least-squares discrimination analysis (OPLS-DA) was performed on the serum Raman spectroscopy data of BMF patients and controls using SIMCA14.1 software. The goodness-of-fit parameter R was used to 2 and Q 2The performance of the OPLS model was evaluated separately. Under the null hypothesis, the model was resampled 200 times by random changes in the y matrix for model validation. In order to find statistically significant Raman peaks in the classification model as potential biomarkers, we used cluster analysis and V+S analysis. Based on comprehensive consideration of parameters such as correlation coefficient, loading, and distance from the center in the V+S plot, peaks with Variable Importance (VIP)>1.0 were selected as potential biomarkers. The obtained potential biomarkers were subjected to significance tests, and potential biomarkers with P<0.05 were considered statistically significant. The V+S plot can provide a list of Raman peaks in descending order of VIP values. Peaks with VIP>1.0 and biological significance were screened as potential markers. Origin software was used for relevant data processing. IBM SPSS Statistics 20 was used for statistical analysis, and Graphpad Prism 5 was used to draw statistical correlation graphs. From serum sample collection, Raman spectroscopy detection, multivariate analysis to the establishment of the identification model, a complete set of research ideas for the identification of AA and MDS was formed ( Figure 1 )

[0043] Statistical analysis

[0044] SPSS 26.0 statistical software package was used to process the peripheral serum Raman spectroscopy data of different groups and the data of peripheral blood total protein (TP), glucose, triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL) and low-density lipoprotein (LDL). The frequency data were compared between groups using the chi-square test (Fisher's exact test); the data that met the normal distribution were analyzed using the Data are expressed as [mean ± standard deviation (SD)]. Means between groups were compared using one-way analysis of variance. Pairwise comparisons between groups with equal variance were performed using the LSD method, and between groups with unequal variance were performed using the Tamhane's T2 test. Data that did not conform to a normal distribution were expressed using the M(Q1-Q3) [median (25th–75th percentile)] test. Intergroup comparisons were performed using the Kruskal-Wallis test. P < 0.05 was considered statistically significant.

[0045] 1.5 Validation of the sensitivity and specificity of the BMF diagnostic model

[0046] Based on the established differentiation model for AA and MDS based on differences in serum substance levels, a validation model was developed to verify the effectiveness of the differentiation method. The validation model consisted of a training set and a prediction set. The training set included 12 BMF patient spectra and 12 control spectra, with the spectral data groups explicitly labeled during validation model development. The prediction set included another 8 BMF patient spectra and 8 control spectra, with the spectral data groups not labeled during validation model development. Using this method, seven validation models were constructed for AA, MDS, NSAA, SAA, VSAA, MDS-RAEB1, and MDS-RAEB2, respectively, to determine the sensitivity and specificity of the validation models. SIMCA-P software assigned classification scores to each spectrum in the training and prediction sets based on the Raman spectral data. The healthy subjects and BMF groups in the training set were assigned classification scores based on their spectral data groups. The healthy subjects and BMF groups in the prediction set were assigned scores based on the similarity of their spectral data with the spectral data in the training set. In the training set and prediction set, the healthy group was assigned a positive score and the BMF group was assigned a negative score, which was considered to be correctly classified. Otherwise, it was considered to be incorrectly classified. In order to more intuitively represent the sensitivity and specificity of the diagnostic model, a dot plot showing the samples of the training set and the prediction set was drawn according to the classification scores of the training set and the prediction set ( Figure 9c -d, Figure 10 By setting the predicted value to 0 as the cutoff value, the sensitivity and specificity of the BMF diagnostic model were obtained (Table 3).

[0047] 2. Experimental results:

[0048] 2.1 Raman spectroscopy analysis results of serum from patients with acquired bone marrow failure syndrome and healthy controls

[0049] To study the Raman spectral characteristics of serum from BMF patients and healthy controls, the present invention obtained 229, 235, and 155 serum Raman characteristic spectra from the healthy control group, AA group, and MDS group, respectively;

[0050] The AA group included 120 spectra from NSAA patients, 73 spectra from SAA patients, and 42 spectra from VSAA patients;

[0051] The MDS group included 88 spectra from MDS-RAEB1 patients and 67 spectra from MDS-RAEB2 patients.

[0052] Peak assignments of the relevant serum Raman spectra presented in this study are reported in Table 1.

[0053] Table 1 Peak assignments of serum Raman spectra:

[0054]

[0055] Figure 2 a-2b show the wavelengths of 600-1800 cm -1 Serum Raman spectra of healthy controls, AA and MDS, healthy controls and AA subtypes, and healthy controls and MDS subtypes were within the range. Figure 2 a shows the Raman spectra of the healthy control group, AA and MDS samples. The pink, yellow and blue vertical lines in the figure represent the Raman spectra of proteins (643, 759, 1003, 1260, 1603, 1654 cm -1 ), nucleic acid (826, 1579cm -1 ) and lipids (1446cm -1 ) related peak positions. Figure 2 b shows the Raman spectra of the healthy control group, NSAA, SAA, and VSAA. Figure 2 c shows the Raman spectra of the healthy control group, MDS-RAEB1 and MDS-RAEB2, and the spectral graphs show similar morphologies.

[0056] It can be seen that it is difficult to identify the differences in serum substances between the acquired bone marrow failure syndrome patient group and the control group based on the above Raman spectral patterns and peak positions. It is necessary to combine the classification model established by the OPLS-DA method to further screen out peak positions that can effectively distinguish the healthy control group and the BMF group as potential biomarkers.

[0057] 2.2 Preliminary screening of potential biomarkers using the OPLS-DA model

[0058] SIMCA-P software was used to apply OPLS-DA to the sample data and a detailed analysis and comparison was carried out; the effectiveness of the supervised OPLS-DA model established based on Raman spectroscopy data can be verified by cluster analysis ( Figure 9a ) and permutation analysis ( Figure 9b ) for evaluation.

[0059] 2.2.1 Healthy control group vs AA group vs MDS group

[0060] To establish a method to distinguish the two major categories of blood system diseases in the healthy control group, AA group, and MDS group, the OPLS-DA model was used to preliminarily screen potential biomarkers.

[0061] Figure 3(a) OPLS-DA score plot. The three sample groups are clearly separated: the healthy control group is located on the positive x-axis, the MDS group is located on the negative x-axis, and the AA group is located between the healthy control and MDS groups. This indicates that the healthy control, AA, and MDS groups are clearly differentiated. This differentiation indicates that OPLS-DA can effectively discriminate serum spectral data from healthy controls, AA, and MDS patients, laying the foundation for analyzing the material characteristics of these three groups.

[0062] Figure 3 b is the OPLS-DA loading plot, which is used to preliminarily screen the Raman peaks that contribute to the identification model of the healthy control group vs the AA group vs the MDS group. The red, blue, green and purple peak numbers in the figure are related to nucleic acids, proteins, β-carotene and lipids, respectively. There is a correlation between the loading plot and the score plot in representing the relationship between the sample substance content, that is, the substance represented by the peak located on the positive semi-axis of the vertical coordinate of the loading plot has a relatively higher content in the group on the positive semi-axis of the horizontal coordinate of the score plot, and the negative semi-axis of the vertical coordinate of the loading plot and the negative semi-axis of the horizontal coordinate of the score plot also have a similar corresponding relationship. Figure 3 b can be seen that nucleic acid (726, 786, 1190cm -1 ), protein (1003cm -1 ), β-carotene (1162cm -1 ) played an important role in the identification of the three groups of samples, reflecting that the contents of nucleic acid, protein and β-carotene in the healthy control group were higher than those in the acquired bone marrow failure syndrome patient group. Figure 3 Figure c represents the OPLS-DA V+S plot, which integrates indicators such as VIP and correlation coefficient, which characterize the contribution of peak positions to the classification model. This plot was used to further screen potential biomarkers. In subsequent analyses, significance tests were performed on characteristic peak positions to identify peak positions that could effectively distinguish between the healthy control group and the bone marrow failure syndrome patient group as potential biomarkers.

[0063] Based on the healthy control group vs AA vs MDS model, the samples of the healthy control group, AA, and MDS groups were combined in pairs for OPLS-DA analysis ( Figure 3 dl). Figure 3d, 3g, and 3j are the OPLS-DA score plots of the three models: healthy control group vs AA group, healthy control group vs MDS group, and AA group vs MDS group. The two groups of samples in the three figures are located on the positive and negative half axes of the X-axis, respectively. The samples are clearly clustered in the scatter plots, reflecting that OPLS-DA can better extract the difference information in the spectrum. The established identification method can identify the differences in the metabolic components of serum samples. The three models can well identify any two groups of samples in the model ( Figure 3 d, 3g, 3j). Figure 3 e, 3h, and 3k are the loading plots of the three models: healthy control group vs AA, healthy control group vs MDS, and AA vs MDS. The figures show that the peak intensities of proteins and nucleic acids in the healthy control group are generally higher than those in the BMF group, while the peak intensities of lipids are lower than those in the BMF group ( Figure 3 e, 3h, 3k).

[0064] 2.2.2 Healthy Control Group vs Different AA Subtypes

[0065] Based on the biomarkers that identified differences between AA and healthy individuals in Section 2.2.1, the AA subtype OPLS-DA model was used to analyze differences in sample substance content and screen potential biomarkers for AA subtypes.

[0066] Depend on Figure 4 The OPLS-DA score plot shown in Figure a shows a clear separation between the four sample groups. The healthy control group and the VSAA group are located on the positive x-axis, the NSAA group and the SAA group are located on the negative x-axis, and the healthy control group and the VSAA group are located on the positive and negative y-axis, respectively, reflecting the differentiation between the healthy control group and the AA subtype groups. This differentiation demonstrates that OPLS-DA can effectively discriminate serum spectral data from the healthy control group, NSAA, SAA, and VSAA patients, providing the basis for analyzing the material characteristics of these four groups.

[0067] Figure 4 (b) OPLS-DA loading plot, used to preliminarily screen Raman peaks contributing to the identification model of AA subtypes in healthy controls. Peak numbers in red, blue, yellow, green, and purple are associated with nucleic acids, proteins, collagen, β-carotene, and lipids, respectively. Figure 4 b can see protein (869, 1221, 1260cm -1 ) and collagen (1344cm -1 ) played an important role in the identification of the five groups of samples, reflecting that the protein and collagen contents of the healthy control group were higher than those of the AA subtype group.

[0068] Figure 4 c is the OPLS-DA V+S plot, which was used to further screen potential biomarkers. In subsequent analysis, the characteristic peak positions were tested for significance, and the peak positions that could effectively distinguish the healthy control group and AA subtypes were determined as potential biomarkers.

[0069] Figure 4 dl shows that based on the healthy control group vs AA subtype model, the healthy control group was combined with the three groups of AA subtype samples for OPLS-DA analysis. Figure 4 d, 4g, and 4j are the OPLS-DA score plots of the three models: healthy control group vs NSAA, healthy control group vs SAA, and healthy control group vs VSAA, respectively. The two groups of samples in the three figures are located on the positive and negative half axes of the X-axis, respectively. The samples are clearly clustered in the scatter plots, reflecting that the three models have good discrimination ability for the two groups of samples in the model. Figure 4 e, 4h, and 4k are the loading plots of the healthy control group vs NSAA, healthy control group vs SAA, and healthy control group vs VSAA models, respectively. The figures show that the peak intensities representing collagen, nucleic acid, protein, and phospholipid / cholesterol in the healthy control group are generally higher than those in the AA subtype group ( Figure 4 e, 4h, 4k) Specifically, the peak intensities of collagen (1344 cm-1), nucleic acid (786, 1078 cm-1) and protein (1260 cm-1) in the healthy control group were higher than those in the NSAA group ( Figure 4 e). The peak intensities of collagen (1344 cm-1), nucleic acid (786 cm-1) and protein (1221 and 1260 cm-1) in the healthy control group were higher than those in the SAA group ( Figure 4 h). The peak intensities of collagen (1344 cm-1), nucleic acid (726, 1190 cm-1) and protein (1221 cm-1) in the healthy control group were higher than those in the VSAA group ( Figure 4 k). Figure 4 f, 4i, and 4l are the V+S plots of the healthy control group vs NSAA, healthy control group vs SAA, and healthy control group vs VSAA models, respectively. The V+S plot provides the main basis for determining the potential biomarkers in the healthy control group and AA subtype models ( Figure 4 c, 4f, 4i, 4l). Biomarker screening and validation were performed using the same methods as above in the healthy control group versus AA subtype, healthy control group versus NSAA, healthy control group versus SAA, and healthy control group versus VSAA models. A list of peaks with VIP > 1.0 was derived from the V+S plot, and biologically significant peaks were identified as potential biomarker peak ranges.

[0070] 2.2.3 Healthy controls vs different MDS subtypes

[0071] Based on the potential biomarkers of MDS screened out in Section 2.2.1, the MDS subtype model was used to analyze the differences in sample substance content and screen potential biomarkers for different MDS subtypes. Figure 5 (a) OPLS-DA score plot. The three sample groups are clearly separated. The healthy control group is located on the positive x-axis, while the MDS-RAEB1 and MDS-RAEB2 groups are located on the negative x-axis, reflecting the distinction between the healthy control group and the MDS group. MDS-RAEB1 and MDS-RAEB2 are located on the positive and negative y-axis, respectively, reflecting the distinction between the two MDS groups. This differentiation demonstrates that OPLS-DA can effectively discriminate serum spectral data from healthy controls, MDS-RAEB1, and MDS-RAEB2 patients, providing the basis for analyzing the material characteristics of these three groups. Figure 5 b is the OPLS-DA loading plot, which is used to preliminarily screen the Raman peaks that contribute to the identification model of healthy control group vs MDS subtype. The red, blue, yellow, green, and purple peak numbers in the figure are related to nucleic acids, proteins, collagen, β-carotene, and lipids, respectively. Protein (1003 cm -1 ), nucleic acid (726, 786, 1078cm -1 ), lipids (1437, 1443cm -1 ), β-carotene (1162cm -1 ) and collagen (1344cm -1 ) played an important role in the identification of the five groups of samples, reflecting that the protein, nucleic acid, β-carotene and collagen contents of the healthy control group were higher than those of the MDS subtype group, while the lipid content was lower than that of the MDS subtype group ( Figure 5 b). Figure 5 c is the OPLS-DA V+S plot, which was used to further screen potential biomarkers. In subsequent analysis, the characteristic peak positions were tested for significance, and the peak positions that could effectively distinguish the healthy control group and MDS subtypes were determined as potential biomarkers.

[0072] Figure 5 di showed that based on the healthy control group vs MDS subtype model, the healthy control group was combined with the two groups of MDS subtype samples for OPLS-DA analysis. Figure 55d and 5g are the OPLS-DA score plots of the two models: healthy control group vs MDS-RAEB1 and healthy control group vs MDS-RAEB2. The two groups of samples in the two figures are located on the positive and negative half axes of the X-axis, respectively. The samples are clearly clustered in the scatter plots, reflecting that the two models have good discrimination ability for the two groups of samples in the model. Figure 5 e and 5h are the loading plots of the healthy control group vs MDS-RAEB1 and healthy control group vs MDS-RAEB2 models, respectively. The figures show that the peak intensities of collagen, nucleic acid, β-carotene, and protein in the healthy control group are generally higher than those in the MDS subtype group ( Figure 5 e, 5h), specifically: collagen (1344cm -1 ), nucleic acid (726, 786cm -1 ), β-carotene (1162cm -1 ) and protein (1003, 1206cm -1 ) had a higher peak intensity than that of the MDS-RAEB1 group ( Figure 5 e). Nucleic acid of healthy control group (1190cm -1 ) and protein (1003cm -1 ) had a higher peak intensity than that of the MDS-RAEB2 group, while lipids (1437 cm -1 ) content was lower than that in the MDS-RAEB2 group ( Figure 5 h). Figure 5 f, 5i are the V+S plots of the healthy control group vs MDS-RAEB1 and healthy control group vs MDS-RAEB2 models, respectively. The V+S plot provides the main basis for determining the potential biomarkers in the healthy control group and MDS subtype models ( Figure 5 c, 5f, 5i). Biomarker screening and validation were performed using the same methods in the healthy control group versus MDS subtype, healthy control group versus MDS-RAEB1, and healthy control group versus MDS-RAEB2 models. A list of peaks with VIP > 1.0 was derived using V+Splot. Based on previous literature, biologically significant peaks were selected as potential biomarker peak ranges.

[0073] 2.3 Validation of biomarker validity and specificity

[0074] 2.3.1 Healthy control group vs AA group vs MDS group

[0075] The serum lipid levels of BMF patients reflect the hematopoietic activity in the bone marrow. AA and anemia secondary to MDS will increase the risk of hypocholesterolemia in patients. Lower HDL and higher TG levels may lead to cardiovascular and cerebrovascular complications, which are not conducive to the prognosis of BMF patients. Due to the excessive proliferation of immature cells in MDS patients, a large amount of TC in the plasma is reduced due to the synthesis of the plasma membrane, while HDL and LDL are reduced due to the reduced source of TC that needs to be transported by them. On the other hand, BMF patients have reduced serum protein content and lack apolipoproteins, which leads to the occurrence of hypocholesterolemia and hypertriglyceridemia. Under the influence of the interaction of these lipid substances, the serum TC, HDL and LDL levels of MDS patients are lower than those of the control group, while the TG level is higher [7-8]. Based on the above research background, the present invention combines the potential biomarkers obtained by the initial screening of OPLS-DA model in 2.2 with serological indicators related to lipid metabolism for correlation and statistical analysis to further identify and verify the effectiveness and specificity of biomarkers based on Raman spectroscopy.

[0076] Figure 6a The statistical analysis of the Raman characteristic peak positions with VIP>1.0 of the four models, namely the healthy control group vs the AA group vs the MDS group, the healthy control group vs the AA group, the healthy control group vs the MDS group, and the AA group vs the MDS group, is presented.

[0077] Through statistical analysis, we can see that Figure 6a The peak intensity of the representative proteins in the healthy control group was higher than that in the BMF group, among which 1221 cm -1 There were significant statistical differences compared with the AA and MDS groups. The peak intensities of nucleic acids in the healthy control group were higher than those in the BMF group, including 726, 781, 786, 1078, 1190, and 1415 cm -1 There were significant statistical differences compared with the AA and MDS groups ( Figure 6a The peak position of the healthy control group represents β-carotene (1162 cm -1 ) was higher than that of the BMF group, and had a statistically significant difference compared with the AA and MDS groups ( Figure 6a ).

[0078] Figure 6b- g shows six types of peripheral blood biochemical indicators of the healthy control group, AA and MDS groups: total protein (TP), glucose, triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL) and low-density lipoprotein (LDL).

[0079] The peak position of the healthy control group represents phospholipid / cholesterol (1285cm -1 ) was higher than that of the BMF group and had statistically significant differences compared with the AA and MDS groups ( Figure 6a ), which is consistent with the serological results (Figure 6f). The peak intensity of lipids in the healthy control group was lower than that in the BMF group, with 1437 and 1446 cm -1 There were significant statistical differences compared with the AA and MDS groups ( Figure 6a ), which was consistent with the serological results (Figure 6d).

[0080] The above results showed that the healthy control group and the BMF patient group were well discriminated by the OPLS-DA model, the cluster analysis was effective, and the serum-based Raman spectroscopy analysis was highly consistent with the serological test results.

[0081] 2.3.2 Healthy Control Group vs Different AA Subtypes

[0082] Figure 7 a is the statistical analysis of the Raman characteristic peak positions with VIP>1.0 of the four models: healthy control group vs AA subtype, healthy control group vs NSAA, healthy control group vs SAA, and healthy control group vs VSAA.

[0083] Figure 7 bg shows the six peripheral blood biochemical indicators of TP, glucose, TG, TC, HDL and LDL in the healthy control group, NSAA, SAA and VSAA groups. Statistical analysis showed that the peak positions of the representative proteins in the healthy control group were 869, 1221 and 1260 cm -1 The intensity is higher than that of the AA subtype group, especially the peak at 1344 cm representing collagen. -1 The intensity was higher than that of AA subtype and had statistically significant differences compared with NSAA, SAA and VSAA ( Figure 7 a) The peak representing phospholipids / cholesterol in the healthy control group is located at 1285 cm -1 The intensity is higher than that of AA subtype, which is consistent with the serological results ( Figure 7 f). The peak intensity representing nucleic acid is higher than that of AA subtype, among which 726cm -1 There were significant statistical differences compared with NSAA, SAA and VSAA ( Figure 7 a).

[0084] It is known from existing technologies that lipid metabolism is associated with the hematopoietic process, and blood lipids can be used as a useful parameter to predict the effect of AA patients on immunosuppressive drugs, corticosteroids, and anabolic steroids [8]. Based on the consistent trends of Raman spectra of TC, HDL, and LDL and serological indicators, it was concluded that the OPLS-DA model can well identify different subtypes of the healthy control group and AA patient group, and the cluster analysis effect is good. The serum-based Raman spectroscopy analysis is highly consistent with the serological test results.

[0085] 2.3.3 Healthy controls vs different MDS subtypes

[0086] Figure 8a Statistical analysis of the Raman characteristic peak positions with VIP>1.0 of the three models: healthy control group vs MDS subtype, healthy control group vs MDS-RAEB1, and healthy control group vs MDS-RAEB2.

[0087] Figure 8 shows the six peripheral blood biochemical indicators of the healthy control group, MDS-RAEB1 and MDS-RAEB2 groups, namely TP, glucose, TG, TC, HDL and LDL. Statistical analysis showed that the peaks of the representative proteins of the healthy control group were at 1003 and 1206 cm -1 The intensity is higher than that of the MDS subtype group, especially the peak at 1344 cm representing collagen. -1 The intensity was higher than that of MDS subtypes and had statistically significant differences compared with MDS-RAEB1 and MDS-RAEB2 ( Figure 8a ).

[0088] The peaks representing lipids in the healthy control group are located at 1437 and 1443 cm -1 The intensity was lower than that of MDS subtypes and was statistically significantly different from MDS-RAEB1 and MDS-RAEB2 ( Figure 8a ), which is consistent with the serological results (Figure 8d). The peak intensities of nucleic acid and β-carotene in the healthy control group were higher than those in the MDS subtype, among which nucleic acid (726, 786 cm -1 ) and β-carotene (1162cm -1 ) showed statistically significant differences compared with MDS-RAEB1 and MDS-RAEB2 ( Figure 8a ).

[0089] Based on the existing technology, it is known that MDS patients have myelofibrosis, which leads to the deposition of fibrin and collagen in the bone marrow, thereby reducing the serum collagen content [9]. It can be seen that the conclusions of the present invention based on Raman spectroscopy analysis are well consistent with the serological test results disclosed in the prior art.

[0090] 2.4 Evaluation of the effectiveness of the supervised OPLS-DA model based on Raman spectroscopy data through cluster analysis and permutation analysis

[0091] In order to establish a method for distinguishing the healthy control group from the two major categories of blood system diseases, AA and MDS, 36 characteristic spectra (12 for each category) were randomly selected from the serum Raman spectra of the healthy control group, AA, and MDS to form three sets of data.

[0092] In order to further establish a method for distinguishing different subtypes of AA from the healthy control group, 24 characteristic spectra (6 for each type) were randomly selected from the serum Raman spectra of the healthy control group, NSAA, SAA and VSAA to form four groups of data.

[0093] To further establish a method for distinguishing the control group from MDS subtypes, 36 characteristic spectra (12 per type) were randomly selected from the Raman spectra of the control, MDS-RAEB1, and MDS-RAEB2 serum samples to form three datasets. OPLS-DA was applied to each sample using SIMCA-P software and detailed comparative analysis was performed.

[0094] The cluster analysis method is presented in Figure 9a , Figure 9a The cluster tree first forks into two clusters from left to right. The first cluster is the MDS group, and the second cluster includes the healthy control group and the AA group. The second cluster further forks into two clusters from left to right, the healthy control group and the AA group. The cluster analysis under the OPLS-DA model distinguished the three types of serum sample Raman spectra of the healthy control group (11 correct and 1 incorrect), AA (12 correct and 0 incorrect), and MDS (12 correct and 0 incorrect) with an accuracy rate of 97.2%. Figure 9a The cluster analysis results are summarized in Table 3, which shows that the established OPLS-DA model can better classify control, AA and MDS. Figure 9b In the figure, the permutation analysis shows that the intercept of Q2 on the Y-axis is negative, indicating that the OPLS-DA model is valid and not overfitting.

[0095] Table 3 Sensitivity and specificity of the BMF diagnostic model

[0096]

[0097]

[0098] The results showed that the sensitivity range of the BMF diagnostic model was 63%-81%, and the specificity range was 92%-100%, reflecting that the Raman spectral characteristic peaks can better distinguish the healthy group and the BMF group, as well as different subtypes of AA and MDS.

[0099] 3. Final Conclusion:

[0100] This study is the first to use Raman spectroscopy (RS) to analyze serum from patients with different types of BMF. Preliminary Raman spectroscopy analysis revealed biomolecular differences specific to BMF, which are attributed to metabolic changes in the patients. The Raman peaks are located at 726, 781, 786, 1078, 1162, 1190, 1221, 1285, 1415, 1437, and 1446 cm -1 , 726 and 1344cm -1 , and 726, 786, 1003, 1344, 1437, and 1443 cm -1 They play an important role in differentiating healthy controls from BMF, healthy controls from AA subtypes, and healthy controls from MDS subtypes. In particular, combined with statistical analysis of serum indicators related to glucose and lipid metabolism, the 1437, 1443, and 1446 cm-1 values, which are closely related to TG, HDL, and glucose, were found to be significantly higher in the 1437, 1443, and 1446 cm-1 values, respectively. -1 The isopeak position can be used as a biomarker for BMF, AA subtypes, and MDS subtypes. The results of this exploratory study show the great promise of developing RS serum analysis into a clinical tool for noninvasive detection and screening of biomarkers for BMF, AA subtypes, and MDS subtypes, and demonstrate the potential correlation of the vast amount of serological test information from BMF patients with disease classification and prognosis.

[0101] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0102] References

[0103] [1]Becktell K,Berlyne D,Pagliuca S,et al.Aplastic Anemia&MDSInternational Foundation(AA&MDSIF):Bone Marrow Failure Disease ScientificSymposium 2018[J].Leukemia research,2019,80:19-25.

[0104] [2]Wang L,Liu H.Pathogenesis of aplastic anemia[J].Hematology,2019,24(1):559-566.

[0105] [3]Locatelli F,Strahm B.How I treat myelodysplastic syndromes ofchildhood[J].Blood, The Journal of the American Society of Hematology,2018,131(13):1406-1414.

[0106] [4]Huang J,Huang L,Liu S,et al.Clinical Outcome of Acquired Post-Immunosuppressive-Therapy Aplastic Anemia in Pediatric Patients:A 13-YearExperience in Two Southern China Tertiary Care Centers[J].InternationalJournal of General Medicine, 2021,14:3133.

[0107] [5]Huang G,Zhang Y,Wei X,et al.CD8+GITR+T cells may negativelyregulate T cell overactivation in aplastic anemia[J].ImmunologicalInvestigations,2021,50(4):406-415.

[0108] [6]Dutta A,De R,Dolai T K,et al.Changes in Different Cytokines(IL-2,TNF-α,and IFN-γ)Profile in Acquired Aplastic Anemia Patients:A Study FromEastern India[J].Journal of pediatric hematology / oncology,2020,42(3):185-192.

[0109] [7]Tanaka T N,Bejar R.MDS overlap disorders and diagnostic boundaries[J].Blood,The Journal of the American Society of Hematology,2019,133(10):1086-1095.

[0110] [8]Yokoyama M,Suto Y,Sato H,et al.Low serum lipids suggest severebone marrow failure in children with aplastic anemia[J].PediatricsInternational,2000,42(6):613-619.

[0111] [9]Beham-Schmid C,Schmitt-Graeff A.Myelodysplastic Syndromes(MDS)[M] / / Bone Marrow Biopsy Pathology.Springer,Berlin,Heidelberg,2020:343-381.

Claims

1. Based on Raman spectroscopy, the Raman peaks representing nucleic acids in peripheral blood are detected at 726, 781, 786, 1078, 1190, and 1415 cm -1 , representing the protein Raman peak 1221cm -1 , representing the Raman peak of β-carotene at 1162 cm -1 , representing the Raman peak of phospholipids / cholesterol at 1285 cm -1 and the Raman peaks representing lipids at 1437 and 1446 cm -1 The application of the characteristic peak intensity level as a biomarker in the preparation of a product for predicting or detecting acquired bone marrow failure syndrome.

2. The use according to claim 1, characterized in that The acquired bone marrow failure syndrome is aplastic anemia and / or myelodysplastic syndrome.

3. Based on Raman spectroscopy, the Raman peaks of representative proteins in peripheral blood are detected at 869, 1221, and 1260 cm -1 , representing the Raman peak of phospholipids / cholesterol at 1285 cm -1 , representing the Raman peak of collagen at 1344 cm -1 and the Raman peak at 726 cm representing nucleic acids -1 The use of the characteristic peak intensity level as a biomarker in the preparation of a product for typing acquired bone marrow failure syndrome is characterized in that, The product for classifying acquired bone marrow failure syndrome is a product for identifying different subtypes of aplastic anemia.

4. Detection of Raman peaks 1003 and 1206 cm representing proteins in peripheral blood based on Raman spectroscopy -1 , representing the Raman peak of collagen at 1344 cm -1 The peaks representing lipids are at 1437 and 1443 cm -1 , representing the peaks of nucleic acids at 726 and 786 cm -1 and 1162 cm representing β-carotene -1 The use of the characteristic peak intensity level of the Raman peak as a biomarker in the preparation of a product for typing acquired bone marrow failure syndrome is characterized in that: The product for classifying acquired bone marrow failure syndrome is a product for identifying different subtypes of myelodysplastic syndrome.

5. The use according to any one of claims 1 to 4, characterized in that The Raman spectroscopy detection conditions are as follows: 785 nm laser as excitation light, 40x objective lens, output power of 10 mW, and the image was taken using a ×40 0.6NA Nikon lens. The sample was illuminated by a laser beam with an output power of 10 mW within a spot diameter of approximately 1.596 μm. The single integration time was 250 s, the number of integrations was one, and the measurement range was 600–1800 cm. -1 , 5-10 sites were measured in each group with a resolution of 1cm -1 .

Images