Marker and application thereof in construction of preeclampsia prediction model

Abstract

The invention relates to the field of biological detection, in particular to a marker and application thereof in construction of a preeclampsia prediction model. The invention provides a marker, which comprises a red blood cell index and an inflammation marker; the red blood cell indexes comprise red blood cell count, hemoglobin, hematocrit, average red blood cell volume, average red blood cell hemoglobin and red blood cell distribution width variable coefficient; the inflammatory markers comprise leukocytes, neutrophils, lymphocytes, monocytes, platelets, eosinophils, platelet hematocrit, a systemic inflammatory response index and a systemic immune inflammatory index. The invention develops an efficient and easily available preeclampsia prediction tool, realizes early recognition and management of preeclampsia, and finally reduces maternal and infant complications.

Description

Technical Field

[0001] The present invention relates to the field of biological detection, and in particular to markers and their application in constructing a preeclampsia prediction model. Background Art

[0002] Preeclampsia (PE) presents after 20 weeks of gestation and is primarily characterized by hypertension and proteinuria. Its global incidence ranges from 2% to 4%, making it a significant cause of severe maternal morbidity and mortality. Currently, the only treatment for preeclampsia is pregnancy termination, a leading cause of preterm birth, neonatal morbidity, and perinatal mortality. Furthermore, women with a history of preeclampsia are at significantly increased risk of type 2 diabetes (T2D) and cardiovascular disease (CVD) later in life. Early and accurate prediction of preeclampsia facilitates timely diagnosis and intervention. Low-dose aspirin screening for high-risk pregnancies before 16 weeks of gestation reduces the incidence of preterm preeclampsia. Similarly, third-trimester screening reduces the risk of late-onset preeclampsia through enhanced surveillance and timely clinical decision-making. Existing first-trimester screening algorithms that combine maternal risk factors, mean arterial pressure, uterine artery resistance, and maternal serum placental growth factor (PlGF) levels are effective in predicting preterm preeclampsia. However, these methods are costly and not widely available. Therefore, there is an urgent need for cost-effective and easily measurable predictors.

[0003] Dysregulation of the maternal systemic immune response is widely considered to be the primary etiology of preeclampsia. Systemic immune-related diseases, such as systemic lupus erythematosus (SLE), type 1 and type 2 diabetes mellitus (T1D and T2D), and rheumatoid arthritis (RA), are associated with an increased risk of preeclampsia. During the progression of preeclampsia, various immune cells play a key role in regulating maternal immune tolerance, trophoblast invasion, and uterine spiral artery remodeling. Systemic immune-inflammatory biomarkers, including the neutrophil-to-lymphocyte ratio (NLR), derived neutrophil-to-lymphocyte ratio (dNLR), monocyte-to-lymphocyte ratio (MLR), neutrophil-to-monocyte-to-lymphocyte ratio (NMLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), systemic inflammatory response index (SIRI), and systemic immune-inflammatory index (SII), reflect the balance of the immune response and the overall inflammatory status. However, clinicians often focus primarily on overtly abnormal parameters, often overlooking emerging immune-derived markers that may play an important pathogenic role in disease progression. Recent studies have revealed strong associations between these accessible immune-inflammatory markers and cancer, cardiovascular disease, diabetes, and adverse pregnancy outcomes, including preterm birth. Similarly, red blood cell (RBC) indices have been associated with preeclampsia. Notably, RBC counts and hemoglobin (HGB) levels are slightly elevated in patients with preeclampsia at the onset of the disease. These findings suggest that peripheral blood cell counts may have potential predictive value for preeclampsia. Combining these markers with machine learning techniques holds promise for applications in disease diagnosis, prediction, monitoring, and prognosis. Despite these advances, research on the role and clinical relevance of RBC indices, NLR, dNLR, SII, and PLR, in the first trimester prior to the onset of preeclampsia remains limited, and existing findings are inconsistent. Summary of the Invention

[0004] In light of this, the present invention provides markers and their use in constructing a preeclampsia prediction model. This invention develops an efficient and accessible preeclampsia prediction tool to enable early identification and management of preeclampsia, ultimately reducing maternal and fetal complications.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0006] The present invention provides markers, including: red blood cell index and inflammatory markers;

[0007] The red blood cell indices include: red blood cell count, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin and coefficient of variation of red blood cell distribution width;

[0008] The inflammatory markers include: leukocytes, neutrophils, lymphocytes, monocytes, platelets, eosinophils, platelet count, systemic inflammatory response index and systemic immune inflammatory index.

[0009] In some embodiments of the present invention, among the above markers, the monocytes include: monocyte subpopulations; the monocyte subpopulations include: CD11c + CD62L-.

[0010] In some embodiments of the present invention, among the above markers, the CD11c + An increased CD62L- percentage increases the probability of preeclampsia.

[0011] In some embodiments of the present invention, among the above-mentioned markers, the red blood cell count and preeclampsia have a threshold effect; when below the threshold (<3.84): for every 1 unit increase in the red blood cell count, the risk of preeclampsia decreases by 85% (OR=0.15, P=0.024); when above the threshold (≥3.84): for every 1 unit increase in the red blood cell count, the risk of preeclampsia increases by 161% (OR=2.61, P<0.001).

[0012] In some embodiments of the present invention, among the above markers, the hemoglobin has a threshold effect on preeclampsia; the turning point is 119.43; after adjusting for confounding factors, increased hemoglobin is significantly positively correlated with the risk of preeclampsia (OR=1.02, P<0.001).

[0013] In some embodiments of the present invention, among the above markers, the hematocrit has a threshold effect on preeclampsia; the turning point is 34.63; after adjusting for confounding factors, increased hematocrit is significantly positively correlated with the risk of preeclampsia (OR=1.01, P=0.003).

[0014] In some embodiments of the present invention, among the above-mentioned markers, there is a threshold effect between the white blood cell count and preeclampsia; when it is below the threshold (<8.44), there is no significant association with the risk of preeclampsia (OR=0.92, P=0.307); when it is above the threshold (≥8.44), for every increase of 1 unit in the white blood cell count, the risk of preeclampsia increases by 14% (OR=1.14, P<0.001).

[0015] In some embodiments of the present invention, among the above markers, after adjusting for confounding factors, the increased neutrophils are significantly positively correlated with the risk of preeclampsia.

[0016] In some embodiments of the present invention, among the above markers, after adjusting for confounding factors, the increase in lymphocytes is significantly positively correlated with the risk of preeclampsia.

[0017] In some embodiments of the present invention, among the above markers, after adjusting for confounding factors, the increase of monocytes is significantly positively correlated with the risk of preeclampsia.

[0018] In some embodiments of the present invention, among the above markers, the platelets have a threshold effect on the risk of preeclampsia; the turning point is 204; after adjusting for confounding factors, increased platelets are significantly correlated with the risk of preeclampsia.

[0019] In some embodiments of the present invention, among the above markers, after adjusting for confounding factors, an increase in the systemic inflammatory response index is significantly positively correlated with the risk of preeclampsia.

[0020] In some embodiments of the present invention, among the above markers, after adjusting for confounding factors, an increase in the systemic immune inflammatory index is significantly positively correlated with the risk of preeclampsia.

[0021] In some embodiments of the present invention, among the above markers, the calculation formula of the systemic inflammatory response index is the neutrophils × the monocytes / the lymphocytes;

[0022] The calculation formula of the systemic immune inflammation index is the platelets×the neutrophils / the lymphocytes.

[0023] The present invention also provides the use of the above markers in preparing products for diagnosing and / or predicting preeclampsia.

[0024] In some embodiments of the present invention, the product in the above application further comprises: one or more detection indicators of age, BMI, parity, number of single pregnancies, mode of conception, glucose metabolism level during pregnancy and hypothyroidism.

[0025] In some embodiments of the present invention, in the above application, the age stratification standard is 35 years old; advanced maternal age (≥35 years old) is a known risk factor for preeclampsia, which may be related to decreased vascular elasticity and aging of placental function.

[0026] In some embodiments of the present invention, the BMI stratification standard is 30 kg / m 2 ; BMI ≥ 30 is associated with insulin resistance and chronic inflammation, increasing the risk of preeclampsia.

[0027] In some embodiments of the present invention, in the above application, the stratification standard for parity is parity = 0; parity refers to the number of previous deliveries (including live births or stillbirths), and primiparas may have a higher risk of preeclampsia due to insufficient placental immune adaptation.

[0028] In some embodiments of the present invention, in the above application, the classification standard for the number of singleton pregnancies is single or multiple; multiple pregnancies have a significantly increased risk of preeclampsia due to the large placenta volume and heavy maternal circulatory load.

[0029] In some embodiments of the present invention, in the above applications, the conception method includes: in vitro fertilization or natural pregnancy; in vitro fertilization includes: embryo transfer technology, hormone intervention, and increased risk of preeclampsia.

[0030] The present invention also provides the use of the marker in constructing a preeclampsia prediction model and / or system.

[0031] The present invention also provides a model and / or system for predicting preeclampsia, comprising:

[0032] Data collection module: obtains the results of complete blood cell count and test indicators of the sample;

[0033] Feature screening module: screening markers for predicting preeclampsia;

[0034] Blood routine scoring module: constructing a blood routine scoring model based on the markers;

[0035] Preeclampsia prediction model module: constructing a preeclampsia prediction model based on the blood routine scoring model and the detection indicators;

[0036] Calculation module: using the preeclampsia prediction model to calculate the probability of the sample suffering from preeclampsia;

[0037] The marker is the marker mentioned above;

[0038] The detection index is the detection index in the above application.

[0039] In the present invention, the prediction model is constructed using the R language caret package. The specific code is as follows, where Train_set is the blood routine results of 15 training set samples, Train_label is the disease outcome of the training set samples, and the final models are GradientBoosting Machines (GBM), RandomForest+Gradient Boosting Machines (RF+GBM), Partial Least Squares Regression with Generalized Linear Models (plsRglm), Ridge Regression (Ridge), Generalized Linear Model Boosting (glmBoost), Extreme Gradient Boosting (XGBoost), Stepwise GeneralizedLinearModel[Forward] (Stepglm[forward]), Stepglm[both], Stepglm[backward], NaiveBayes, Linear Discriminant Analysis (LDA) and SupportVectorMachine (SVM). The code is:

[0040]

[0041]

[0042] The GBM model was selected as the optimal blood routine score based on the ROC results and model convenience.

[0043]

[0044] The above code is then used to calculate the preeclampsia score, where train_set is the GBM score of the training set sample blood routine, maternal age, BMI, and number of parities, and train_label is the disease outcome of the training set sample. The optimal model, RF+GBM, is selected based on the ROC result.

[0045]

[0046]

[0047] In some embodiments of the present invention, in the above-mentioned model and / or system, in the blood routine scoring module and the preeclampsia prediction model module, the construction adopts a machine learning algorithm; the machine learning algorithm includes: gradient boosting machine, random forest + gradient boosting machine, extreme gradient boosting, ridge regression, stepwise generalized linear model, generalized linear model, naive Bayes, linear discriminant analysis and support vector machine One or more.

[0048] In some embodiments of the present invention, in the above-mentioned model and / or system, the blood routine scoring module adopts a gradient boosting machine; the preeclampsia prediction model module adopts a random forest + gradient boosting machine.

[0049] In some embodiments of the present invention, in the above-mentioned model and / or system, the formula of the preeclampsia prediction model is: the optimal cutoff point is Youden index = sensitivity + specificity - 1;

[0050] The optimal cutoff point of the preeclampsia prediction model is 0.027; when the score is ≥ 0.027, it is judged as a high-risk preeclampsia; when the score is < 0.027, it is judged as a low-risk preeclampsia.

[0051] In some embodiments of the present invention, in the above-mentioned model and / or system, the formula of the routine blood test scoring model is: the optimal cutoff point is Youden Index = Sensitivity + Specificity - 1; the optimal cutoff point of the routine blood test scoring model is 0.037.

[0052] The present invention also provides a method for constructing the above-mentioned model and / or system, comprising the following steps:

[0053] S1: Obtain the results of the complete blood cell count and test indicators of the sample;

[0054] S2: constructing the blood routine scoring model based on the markers;

[0055] S3: constructing a preeclampsia prediction model based on the blood routine scoring model and the detection indicators;

[0056] S4: Calculate the probability of the sample suffering from preeclampsia using the preeclampsia prediction model.

[0057] In some embodiments of the present invention, the sample collection time in the above construction method is between 8 and 14 months of pregnancy. +6 week.

[0058] The beneficial effects of the present invention include:

[0059] (1) Based on a cohort of 17,955 pregnant women from two centers, we analyzed routine blood data in early pregnancy to explore the changing patterns of novel inflammatory indicators before the onset of preeclampsia.

[0060] (2) Identify immune cells that may be involved in the progression of preeclampsia through MR analysis.

[0061] (3) Leveraging the advantages of machine learning, we evaluated the predictive performance of combining inflammatory markers with maternal risk factors. By integrating these innovative approaches, we developed an efficient and accessible tool for predicting preeclampsia, enabling early identification and management of preeclampsia, ultimately reducing maternal and fetal complications. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for describing the embodiments or the prior art.

[0063] Figure 1 Shown is a flow chart of the present invention;

[0064] Figure 2 The proportion of patients with PE is ranked by quartiles; where: WBC, white blood cell count; NEUT, neutrophil count; LYMPH, lymphocyte count; MONO, monocyte count; PLT, platelet count; SIRI, systemic inflammatory response index; SII, systemic immune inflammatory index; RBC, red blood cell count; HGB, hemoglobin; HCT, hematocrit;

[0065] Figure 3 Figure 2 shows restricted cubic spline analysis of inflammatory markers and red blood cell indices; A is WBC, white blood cell count; B is NEUT, neutrophil count; C is PLT, platelet count; D is RBC, red blood cell; E is HGB, hemoglobin; F is HCT, hematocrit;

[0066] Figure 4 The relationship between inflammatory markers and red blood cell indices and the risk of PE was shown in the subgroups of singleton (singleton and non-singleton) and parity (primipara and multipara) pregnancy; where: WBC, white blood cell count; NEUT, neutrophil count; LYMPH, lymphocyte count; PLT, platelet count; RBC, red blood cell count;

[0067] Figure 5 The forest plot shows the causal relationship between immune cells and PE;

[0068] Figure 6 CD11c + CD62L - Mediation analysis of the genetic association between the percentage of monocytes and preeclampsia; A represents the CD11c + CD62L -Forest plot of the causal relationship between monocyte percentage, phosphate to 5-hydroxyproline and PE; B is the expression of phosphate-5-hydroxyproline in CD11c + CD62L - The mediating role of monocytes in the causal effect of PE;

[0069] Figure 7 Figure 3 shows the PE prediction model constructed by machine learning algorithms; A is the AUC value of the blood routine scoring model constructed by 12 machine learning algorithms; B is the AUC value of the PE prediction model constructed by 12 machine learning algorithms; C is the ROC curve of the best blood routine scoring model (GBM) and the PE prediction model (RF+GBM) in the training group; D is the ROC curve of the best blood routine scoring model (GBM) and the PE prediction model (RF+GBM) in the experimental group;

[0070] Figure 8 CD11c on PE + CD62L - Mendelian randomization (MR) results of monocyte percentage; where: A is CD11c in preeclampsia + CD62L - MR scatter plot of monocyte ratio; B is CD11c on PE + CD62L - MR forest plot of the percentage of monocytes in the mononuclear cell population; C is CD11c on PE + CD62L - MR funnel plot of monocyte ratio; D is CD11c on PE + CD62L - The percentage of monocytes in the mononuclear cell population was analyzed by leave-one-out MR. DETAILED DESCRIPTION

[0071] The present invention discloses markers and their application in constructing a preeclampsia prediction model.

[0072] It should be understood that the expression "one or more of" includes individually each of the items recited after the expression and various combinations of two or more of the recited items, unless otherwise apparent from the context and usage. The expression "and / or" in conjunction with three or more recited items should be understood to have the same meaning, unless otherwise apparent from the context.

[0073] The terms "comprising", "having" or "containing", including their grammatical synonyms, should generally be understood as open and non-restrictive, e.g., not excluding other unrecited elements or steps, unless otherwise specifically stated or understood from the context.

[0074] It should be understood that the order of steps or the order in which certain actions are performed are not important as long as the present invention remains operable. Additionally, two or more steps or actions may be performed simultaneously.

[0075] The use of any and all examples or exemplary language, such as "such as" or "including," herein is intended merely to better illustrate the invention and does not limit the scope of the invention unless otherwise claimed. No language in this specification should be construed as indicating any non-claimed element as essential to the practice of the invention.

[0076] In addition, the numerical ranges and parameters used to define the present invention are approximate values. The relevant numerical values ​​in the specific examples have been presented as accurately as possible. However, any numerical value inherently inevitably contains standard deviations due to individual testing methods. Therefore, unless otherwise expressly stated, all ranges, amounts, values, and percentages used in this disclosure should be understood to be modified by the word "about." As used herein, "about" generally means that the actual value is within plus or minus 10%, 5%, 1%, or 0.5% of a specified value or range.

[0077] The present invention relates to the following English and Chinese parameters: red blood cell count RBC, hemoglobin HGB, hematocrit HCT, mean corpuscular volume MCV, mean corpuscular hemoglobin MCH, red blood cell distribution width variation coefficient RDW-CV; white blood cell WBC, neutrophil NEUT, lymphocyte LYMPH, monocyte MONO, eosinophil EO, platelet PLT, platelet hematocrit PCT, systemic inflammatory response index SIRI, systemic immune inflammatory index SII.

[0078] In Examples 1 to 6 and Effect Examples 1 to 7 of the present invention, all the raw materials and reagents used can be purchased from the market.

[0079] The present invention will be further described below in conjunction with the embodiments:

[0080] Example 1 Data Collection and Processing

[0081] The main data were collected from pregnant women who gave birth in the First Affiliated Hospital of Soochow University and Suzhou Municipal Hospital from 2015 to 2024. All data were obtained from the hospital information system. +6 Blood routine data (at 24 weeks) were collected and used as features for model construction. In the non-preeclampsia population, individuals with aneuploidy or major fetal malformations, termination of pregnancy before 24 weeks due to miscarriage, fetal death, or other reasons, as well as cases of preterm birth, gestational hypertension, and intrauterine growth restriction were excluded. However, patients with gestational diabetes, hypothyroidism, and intrahepatic cholestasis of pregnancy (ICP) were included in the study.

[0082] By definition, the diagnosis of preeclampsia requires the presence of new-onset hypertension (systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg at 20 weeks of gestation or later), or chronic hypertension with at least one of the following: proteinuria (24-hour urine protein ≥300 mg, urine protein-to-creatinine ratio ≥30 mg / mmol, or urine dipstick result ≥2+), renal dysfunction (serum creatinine >97 μmol / L), hepatic dysfunction (transaminases ≥65 IU / L), or hematologic dysfunction (platelet count <100,000 / μL).

[0083] Example 2 Complete blood cell count analysis

[0084] Between 8 and 14 gestational years +6 During the week, fasting venous blood was collected into EDTA tubes. On the same day, a complete blood cell count was performed using the Mindray BC-6800Plus fully automatic hematology analyzer, which uses semiconductor laser fluorescence flow cytometry and fluid dynamic focusing technology to obtain a complete blood cell count, including white blood cell classification, in a dedicated channel. The calculation method for inflammation-related markers is as follows: white blood cell count (WBC), neutrophil count (NEUT), lymphocyte count (LYMPH), monocyte count (MONO) and platelet count (PLT) (unit: 10 9 The NEUT / LYMPH (NLR), NEUT / (WBC-NEUT) (dNLR), MONO / LYMPH (MLR), (MONO + NEUT) / LYMPH (NMLR), PLT / LYMPH (PLR), and LYMPH / MONO (LMR) ratios were calculated. SIRI was calculated as NEUT × MONO / LYMPH, and SII as PLT × NEUT / LYMPH.

[0085] Example 3

[0086] First, we used multivariate logistic regression to assess the associations of inflammatory markers, red blood cell indices, and preeclampsia using trend analysis. Inflammatory markers and red blood cell indices were categorized into quartiles, with the lowest quartile (Q1) serving as the reference. The median of each group was then included as a continuous variable in the logistic regression model. Results are presented as odds ratios (ORs) with 95% confidence intervals (95% CIs). Three models were constructed, adjusting for various confounding variables. The crude model remained unadjusted. Model 1 adjusted for confounders such as age, body mass index (BMI), singleton pregnancy, in vitro fertilization (IVF) pregnancy, and parity. Model 2 further adjusted for gestational diabetes and hypothyroidism. To further explore potential nonlinear relationships between inflammatory markers and preeclampsia risk, we performed restricted cubic spline (RCS) regression analysis. Nonlinearity was tested using the likelihood ratio test. A two-stage linear regression model was used to further analyze the threshold effect of inflammatory markers on preeclampsia risk. Furthermore, we performed subgroup analyses using logistic regression to explore the associations of inflammatory markers, red blood cell indices, and preeclampsia in different subgroups. Stratification factors included age (<35 years and ≥35 years), BMI (<30 kg / m 2 and ≥30kg / m 2 ), parity (primipara <0 and multipara >0), gestational diabetes (no and yes), hypothyroidism (no and yes), singleton pregnancy (no and yes), and in vitro fertilization pregnancy (no and yes). Interaction analyses were also performed to explore differences between subgroups.

[0087] Example 4 Genetic association between immune cells and preeclampsia

[0088] The genetic association between immune cells and preeclampsia and the mediating role of plasma metabolites were investigated by Mendelian randomization (MR). This study utilized genetic data related to 731 immune cells from a 2020 study, divided into absolute cell counts (ACC), median fluorescence intensity (MFI), morphological parameters (MP), and relative cell counts (RCC). In addition, 1,400 metabolites (1,091 plasma metabolites and 309 metabolite ratios) from a 2023 study and genetic data on preeclampsia from the IEU Open GWAS project (2,355 cases and 264,887 controls) were included. Instrumental variables (IVs) were selected according to three core assumptions (association, independence, and exclusion assumptions), and those that met P < 1 × 10^-5 and linkage disequilibrium (LD) thresholds (R 2<0.001, distance = 10,000 kb) single nucleotide polymorphisms (SNPs), with weak instrumental variables (F-statistic <10), were excluded. Inverse variance weighting (IVW) was used as the primary method for two-sample MR analysis, with Egger, weighted median, simple mode, and weighted mode methods used as supplementary methods. Mediation effects were calculated using the coefficient product method to estimate the proportion of the effect of immune cells on preeclampsia mediated by plasma metabolites. Sensitivity analyses were performed to assess the robustness of the results. All statistical analyses and visualizations were performed in R software using the TwoSampleMR, MRPRESSO, and ggplot2 packages.

[0089] Example 5: Machine Learning Algorithm to Construct a Preeclampsia Prediction Model

[0090] In this study, 12 machine learning methods were used, including gradient boosting machine (GBM), random forest + gradient boosting machine (RF+GBM), partial least squares regression and generalized linear model (plsRglm), ridge regression (Ridge), generalized linear model boosting (glmBoost), extreme gradient boosting (XGBoost), stepwise generalized linear model

[0091] Forward

[0092] We investigated the diagnostic utility and predictive value of inflammatory markers and red blood cell indices in preeclampsia using forward (Stepglm), bidirectional (Stepglm), backward (Stepglm), naive Bayes, linear discriminant analysis (LDA), and support vector machine (SVM) methods. Participants were randomly divided into a training set and a test set in a 5:5 ratio. The training set was used to screen variables and construct models, while the test set was used to evaluate the performance of the final model. First, a blood test score model was constructed using 12 machine learning algorithms based on 15 statistically significant blood test parameters. Subsequently, a comprehensive preeclampsia prediction model was constructed by combining the blood test score predicted by the optimal algorithm with maternal factors (including age, BMI, parity, singleton pregnancy, in vitro fertilization pregnancy, gestational diabetes, and hypothyroidism). The diagnostic performance of each model was evaluated by calculating the area under the curve (AUC). The optimal model was selected based on the AUC value, and the receiver operating characteristic (ROC) curve was plotted.

[0093] Example 6 Statistical Analysis

[0094] Non-normally distributed continuous variables were evaluated using the Mann-Whitney U test and are presented as medians (interquartile range Q1, Q3 Q1, Q3). Categorical variables were compared using the chi-square test and reported as counts (percentages). All statistical analyses were performed using R software (version 4.4.0), and statistical significance was defined as P < 0.05.

[0095] Baseline characteristics of pregnant women in effect case 1

[0096] A total of 17,955 pregnant women were included in this study, of whom 17,381 were non-preeclamptic and 574 developed preeclampsia (Table 1). Inflammatory markers were measured at a mean gestational age of 12.43 weeks. Compared with non-preeclamptic women, women with preeclampsia had a higher age, BMI, proportion of singleton and in vitro fertilization pregnancies, and lower parity. In terms of inflammatory markers and red blood cell indices, women with preeclampsia showed higher WBC, NEUT, LYMPH, MONO, PLT, SIRI, SII, RBC, HGB, and hematocrit (HCT) during early pregnancy before the onset of the disease. However, no significant differences were found in NLR, dNLR, MLR, NMLR, PLR, and LMR. In addition, WBC, NEUT, LYMPH, MONO, PLT, SIRI, and SII were divided into quartiles. As the quartile level increased, the prevalence of preeclampsia increased ( Figure 2 ).

[0097] Table 1 Baseline characteristics of pregnant women

[0098]

[0099]

[0100] In Table 1, Z, Mann-Whitney test, χ 2 , χ 2 Test; M(Q1, Q3), median (first quartile, third quartile); n(%), number (percentage).

[0101] Effect Example 2: Association between inflammatory markers, red blood cell indices and the risk of preeclampsia

[0102] Three models were constructed by adjusting different confounding variables to evaluate the association of WBC, NEUT, LYMPH, MONO, PLT, SIRI, SII, RBC, HGB and HCT with the risk of preeclampsia. After adjusting for all confounding variables, WBC (OR = 1.09, 95% CI: 1.04-1.13, P < 0.001), NEUT (OR = 1.09, 95% CI: 1.04-1.14, P < 0.001), LYMPH (OR = 1.27, 95% CI: 1.05-1.53, P = 0.013), MONO (OR = 2.57, 95% CI: 1.31-5.03, P = 0.006), PLT (OR = 1.01, 95% CI: 1.01-1.01, P < 0.001), SIRI, SII, RBC, HGB and HCT were significantly associated with the risk of preeclampsia. RI (OR=1.11, 95%CI: 1.01-1.21, P=0.032), SII (OR=1.01, 95%CI: 1.01-1.01, P=0.002), RBC (OR=2.12, 95%CI: 1.64-2.75, P<0.001), HGB (OR=1.02, 95%CI: 1.01-1.03, P<0.001) and HCT (OR=1.07, 95%CI: 1.03-1.11, P<0.001) were significantly positively correlated with the prevalence of preeclampsia (Table 2).

[0103] Subsequently, WBC, NEUT, LYMPH, MONO, PLT, SIRI, and SII were converted into categorical variables based on quartiles. In the fully adjusted model, the risk of preeclampsia increased significantly with increasing quartile levels of WBC, NEUT, LYMPH, MONO, and PLT compared with Q1 (P < 0.05). However, this trend was not observed for SIRI and SII (Table 2).

[0104] Table 2 Relationship between inflammatory indicators, red blood cell indicators and the risk of PE

[0105]

[0106]

[0107]

[0108] In Table 2, the original model was the unadjusted model; Model 1 was adjusted for age, BMI, singleton pregnancy, IVF pregnancy, and parity; and Model 2 was adjusted for age, BMI, singleton pregnancy, IVF pregnancy, parity, diabetes, and hypothyroidism.

[0109] OR: odds ratio; CI: confidence interval.

[0110] Effect Example 3 Nonlinear relationship and threshold effect between inflammatory markers, red blood cell indices and preeclampsia risk

[0111] To further ensure the robustness of the results, potential nonlinear relationships between inflammatory markers and the risk of preeclampsia were examined. In the RCS regression model, after adjusting for all confounding factors, no significant nonlinear relationships were found between LYMPH, MONO, SIRI, and SII and preeclampsia (nonlinear P>0.05). However, significant nonlinear associations were found between WBC, NEUT, PLT, RBC, and HGB and preeclampsia (nonlinear P<0.05) (e.g. Figure 3 A nonlinear association trend was also observed between HCT and preeclampsia (nonlinear P = 0.07) (as shown in Figure 2). Figure 3 F).

[0112] In Table 3, further analysis revealed a threshold effect between WBC and the risk of preeclampsia (likelihood ratio test P = 0.034), with an inflection point of 8.44. When WBC was less than 8.44, no significant association with the risk of preeclampsia was observed (OR = 0.92, 95% CI: 0.79-1.08, P = 0.307). However, when WBC exceeded 8.44, each unit increase was associated with a 0.14-fold increase in the risk of preeclampsia (OR = 1.14, 95% CI: 1.07-1.22, P < 0.001). A threshold effect was also observed for PLT (likelihood ratio test P = 0.004), with an inflection point of 204. For RBC, a significant threshold effect was detected (likelihood ratio test P < 0.001), with an inflection point of 3.84. When RBC count was less than 3.84, each unit increase was associated with a 0.85-fold lower risk of preeclampsia (OR = 0.15, 95% CI: 0.03-0.78, P = 0.024). However, when RBC count exceeded 3.84, each unit increase was associated with a 1.61-fold increased risk of preeclampsia (OR = 2.61, 95% CI: 1.95-3.50, P < 0.001). In addition, threshold effects were also detected for HGB (likelihood ratio test P < 0.001) and HCT (likelihood ratio test P = 0.021), with inflection points of 119.43 and 34.63, respectively.

[0113] Table 3 Analysis of the threshold effects of inflammatory markers and red blood cell indices on PE using a two-stage regression model

[0114]

[0115]

[0116] In Table 3, OR: odds ratio; CI: confidence interval.

[0117] Subgroup analysis of effect case 4

[0118] Subgroup analyses were performed to examine whether the relationships between inflammatory markers, red blood cell indices, and the risk of preeclampsia differed by age, BMI, parity, gestational diabetes, hypothyroidism, singleton pregnancy, and in vitro fertilization pregnancy. Significant interactions were found between WBC, NEUT, LYMPH, and PLT and singleton pregnancy (P for interaction < 0.05), and between RBC and parity (P for interaction < 0.05) (e.g. Figure 4 shown).

[0119] Table 4 Relationship between inflammatory indicators, red blood cell indicators and the risk of PE in different subgroups

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128] In Table 4, n(%): number (percentage); OR: odds ratio; CI: confidence interval.

[0129] Effect Example 5: Genetic Association between Immune Cells and Preeclampsia

[0130] To evaluate the genetic association of immune cells with preeclampsia, a two-sample MR approach was used to assess the causal relationship between 731 immune cell types and preeclampsia. The analysis identified 15 immune cell types as risk factors for preeclampsia, including CD20-% lymphocytes, CD11c+CD62L- monocytes % monocytes, lymphocytes % leukocytes, HLA DR+ NK % CD3- lymphocytes, PDL-1 on CD14-CD16+ monocytes, and CD16 on CD14+CD16+ monocytes (e.g., Figure 5 shown).

[0131] Because the OR value of CD11c+CD62L- monocytes% monocytes was high, genetic association analysis was further performed for this immune cell type. No heterogeneity or horizontal pleiotropy was detected in the sensitivity analysis, which was confirmed by the Cochran Q test, MR-Egger intercept test, and MR pleiotropy residuals and outliers (MR-PRESSO) (P>0.05, Table 5). The funnel plot further verified the stability of the results (such as Figure 8 In addition, the leave-one-out analysis showed that no single IV significantly affected the causal inference (e.g. Figure 8 This finding supports the validity of the causal relationship between CD11c+CD62L- monocytes and the risk of preeclampsia.

[0132] To investigate how CD11c+CD62L- monocytes% monocytes affect the progression of preeclampsia, related metabolites were included in the analysis. The results showed that elevated levels of CD11c+CD62L- monocytes% monocytes may increase the ratio of phosphate to 5-oxoproline, thereby increasing the risk of preeclampsia (e.g. Figure 6 The estimated mediation effect was 0.009 (95% CI: -0.000174 to 0.0182), and the mediation effect ratio was 6.6% (95% CI: -0.128% to 13.3%) (Table 6, as shown in Figure 6). Figure 6 (as shown in B).

[0133] Table 5

[0134]

[0135] Table 6 The mediating role of the ratio of phosphate to 5-hydroxyproline in the causal effect of CD11c+CD62L- monocytes as a percentage of monocytes on PE

[0136]

[0137] Effect Example 6 Predicting the risk of preeclampsia before symptoms appear by using inflammation score and red blood cell index changes

[0138] The 17,955 samples were randomly divided into a training set and a test set in a 5:5 ratio. No statistically significant differences were found between the two groups, except that gestational diabetes was more common in the test set.

[0139] Twelve machine learning methods were used to construct scoring models for blood routine indicators. Among them, the model constructed using GBM performed best in predicting preeclampsia, with an AUC value of 0.72 in the training set and an AUC value of 0.65 in the test set (e.g. Figure 7This performance was significantly better than that of a single indicator, indicating that the scoring model can effectively identify the risk of preeclampsia during early pregnancy.

[0140] Subsequently, maternal risk factors such as age, BMI, and parity were incorporated into the model to construct a comprehensive preeclampsia prediction model. In the training and test sets, the combined prediction model showed high predictive performance before 14 weeks of gestation, with AUC values ​​of 0.82 and 0.73, respectively (e.g., Figure 7 BD shown in Figure 2).

[0141] Effect Example 7

[0142] In this study, we comprehensively investigated the relationship between routine blood inflammatory markers, red blood cell (RBC) parameters, and the risk of preeclampsia (PE) using large-scale cohort data from two centers. Using machine learning methods, we developed a scoring system based on routine blood-related indicators and combined it with other PE-related risk factors to establish a joint prediction model applicable before 14 weeks of gestation. The model showed strong predictive performance with an area under the receiver operating characteristic (ROC) curve of 0.82. This study is the first to explore the relationship between routine blood markers and PE risk in early pregnancy (before the disease develops), thereby significantly expanding the range of early risk factors for PE. By incorporating these indicators, the accuracy of early PE prediction was significantly improved.

[0143] The exact etiology of PE remains unclear, but endothelial dysfunction is a core mechanism of its progression. This dysfunction results from exposure of maternal endothelial cells to factors released by the placenta, leading to elevated levels of systemic vasoconstrictors, oxidative stress, anti-angiogenic factors, and inflammatory mediators. Activation of maternal innate and adaptive immune cells and their mediators is closely associated with the development of PE. Recent studies have suggested that elevated white blood cell (WBC) and neutrophil (NEUT) counts, as well as a high neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), may serve as potential predictors of liver injury and coagulation abnormalities in patients with PE. Furthermore, several small studies have shown that elevated WBC counts during the first and second trimesters of pregnancy are associated with a higher prevalence of late-onset PE. Despite the availability of these indicators, large-scale studies to validate these findings are lacking, and the predictive value of these biomarkers remains to be confirmed.

[0144] This study systematically evaluated the relationship between inflammatory markers and the development of PE using a large-scale analysis of routine blood tests performed before the onset of PE. Unlike previous studies that typically examined these markers as separate indicators, this study found that white blood cell count (WBC), neutrophil urea nitrogen (NEUT), lymphocytes (LYMPH), monocytes (MONO), platelets (PLT), systemic inflammatory response index (SIRI), and systemic immune inflammatory index (SII) were all strongly associated with PE. This association remained significant even after adjustment for confounding factors. However, ratios such as NLR, dNLR, MLR, NMLR, PLR, and LMR were not significantly associated with PE. Further analysis revealed a threshold effect between WBC and PE (likelihood ratio test, P = 0.034), with a turning point of 8.44. Below this threshold, WBC was not significantly associated with the risk of PE (OR = 0.92, 95% CI: 0.79-1.08, P = 0.307). However, above this threshold, each unit increase in WBC count was associated with a 0.14-fold increased risk of PE (OR = 1.14, 95% CI: 1.07-1.22, P < 0.001). A similar threshold effect was detected between PLT and PE (likelihood ratio test P = 0.004), with a turning point of 204. No significant correlations were found with the ratios of NLR, dNLR, MLR, NMLR, PLR, and LMR. Furthermore, MR and sensitivity analyses further confirmed the potential causal relationship between inflammatory cells and PE, with the percentage of CD11c+CD62L- monocytes playing a particularly prominent role in the pathogenesis of PE. A recent study of peripheral blood mononuclear cells from patients with PE similarly demonstrated that monocytes undergo remodeling and activation in PE. Among these monocyte subsets, immune pathways such as complement response, myeloid leukocyte activation, increased interferon receptor activity, and macrophage activation were particularly upregulated, consistent with the well-established hallmarks of systemic inflammation in PE. Furthermore, mediation analysis showed that an increase in the percentage of CD11c+CD62L- monocytes may lead to an increase in the phosphate to 5-oxoproline ratio, further increasing the risk of PE. These findings not only significantly enhance the scientific reliability of the conclusions, but also highlight the potential application value of cell indicators in the early prediction of PE.

[0145] Previous studies have shown that the pathophysiology of PE alters the biophysical properties of erythrocytes, including increased aggregation and decreased deformability. These changes may impede normal perfusion of the intervillous space of the placenta, thereby affecting fetal oxygen supply. Damaged erythrocytes may exacerbate the progression of endothelial dysfunction and maternal circulatory disturbances. These studies have also shown that erythrocyte count and hemoglobin (HGB) levels are slightly increased in patients with PE, but little is known about changes in erythrocyte-related parameters before the onset of PE. In this study, we observed that before the onset of the disease (at 14 weeks of gestation), RBC count (OR = 2.12, 95% CI: 1.64-2.75, P < 0.001), HGB (OR = 1.02, 95% CI: 1.01-1.03, P < 0.001), and hematocrit (HCT) (OR = 1.01, 95% CI: 1.01-1.01, P = 0.003) were significantly positively associated with the incidence of PE after adjusting for all confounding variables. Furthermore, significant nonlinear relationships were observed between RBC, HGB, and HCT and the risk of PE (nonlinearity P < 0.05). A significant threshold effect was detected for RBC (likelihood ratio test P < 0.001), with a turning point of 3.84. When RBC was below 3.84, each unit increase was associated with a 0.85-fold decrease in the risk of PE (OR = 0.15, 95% CI: 0.03-0.78, P = 0.024). However, when RBC exceeded 3.84, each unit increase was associated with a 1.61-fold increase in the risk of PE (OR = 2.61, 95% CI: 1.95-3.50, P < 0.001). Furthermore, threshold effects were observed for HGB (likelihood ratio test P < 0.001) and HCT (likelihood ratio test P = 0.021), with turning points of 119.43 and 34.63, respectively.

[0146] Currently, there is no effective treatment for PE other than delivery. However, early identification of high-risk individuals and initiation of aspirin therapy before 16 weeks of gestation can significantly reduce its risk. Existing assessment models primarily rely on clinical risk factors. Although simple and feasible, this approach has low sensitivity, with detection rates of only 41% and 34% for preterm and term PE, respectively, and a false-positive rate of 10%. Another assessment method combines maternal factors, mean arterial pressure measurement, uterine artery pulsatility index, and serum placental growth factor (PlGF). When implemented before 14 weeks of gestation, this model can effectively identify 90% of cases of premature PE. However, its predictive ability for overall PE (including preterm and term types) is limited, with a receiver operating characteristic (ROC) value of only 0.78, and its performance is even worse in primiparous women. Therefore, the development of new biomarkers for PE is crucial. New research biomarkers such as cfDNA and cfRNA have shown potential for early prediction of PE. For example, a prediction model based on cfDNA methylation analysis at 14 weeks of gestation can predict premature PE with an accuracy of 0.792, but the testing cost of this technology is relatively high. In this study, we found that a routine blood test at 14 weeks of gestation can predict PE, with receiver operating characteristic (ROC) values ​​of 0.72 in the training set and 0.65 in the test set. Furthermore, this test is already included in routine prenatal screening, is low-cost, and has high clinical feasibility.

[0147] In conclusion, our large cohort study revealed significant associations between systemic immune-inflammatory biomarkers and red blood cell indices measured before 14 weeks of gestation and the risk of PE. After adjustment for confounders, elevated WBC, NEUT, LYMPH, MONO, PLT, SIRI, SII, RBC, HGB, and HCT were identified as key risk factors for PE. Furthermore, clear nonlinear relationships were observed between these biomarkers, red blood cell indices, and the risk of PE. Importantly, systemic immune-inflammatory biomarkers and red blood cell indices measured before 14 weeks of gestation are proposed as cost-effective and reliable predictors for the early detection of PE.

[0148] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A marker, characterized in that include: red blood cell indices and inflammatory markers; The red blood cell indices include: red blood cell count, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin and coefficient of variation of red blood cell distribution width; The inflammatory markers include: leukocytes, neutrophils, lymphocytes, monocytes, platelets, eosinophils, platelet count, systemic inflammatory response index and systemic immune inflammatory index.

2. The marker according to claim 1, wherein The monocytes include: monocyte subpopulations; the monocyte subpopulations include: CD11c + CD62L-.

3. Use of the marker according to claim 1 or 2 in the preparation of a product for diagnosing and / or predicting preeclampsia.

4. The use according to claim 3, characterized in that The product also includes one or more detection indicators of age, BMI, parity, number of single pregnancies, mode of conception, glucose metabolism level during pregnancy and hypothyroidism.

5. Use of the marker according to claim 1 or 2 in constructing a preeclampsia prediction model and / or system.

6. A model and / or system for predicting preeclampsia, characterized in that: include: Data collection module: obtains the results of complete blood cell count and test indicators of the sample; Feature screening module: screening markers for predicting preeclampsia; Blood routine scoring module: constructing a blood routine scoring model based on the markers; Preeclampsia prediction model module: constructing a preeclampsia prediction model based on the blood routine scoring model and the detection indicators; Calculation module: using the preeclampsia prediction model to calculate the probability of the sample suffering from preeclampsia; The marker is the marker according to claim 1 or 2; The detection indicator is the detection indicator in the application as claimed in claim 3.

7. The model and / or system according to claim 6, characterized in that In the routine blood test scoring module and the preeclampsia prediction model module, the construction adopts a machine learning algorithm; the machine learning algorithm includes: gradient boosting machine, random forest + gradient boosting machine, extreme gradient boosting, ridge regression, stepwise generalized linear model, generalized linear model, naive Bayes, linear discriminant analysis and support vector machine One or more.

8. The model and / or system according to claim 6 or 7, characterized in that The formula of the preeclampsia prediction model is: the optimal cutoff point is Youden index = sensitivity + specificity - 1; The optimal cutoff point of the preeclampsia prediction model is 0.027; when the score is ≥ 0.027, it is judged as a high-risk preeclampsia; when the score is < 0.027, it is judged as a low-risk preeclampsia.

9. The method for constructing a model and / or system according to any one of claims 6 to 8, characterized in that: The steps include: S1: Obtain the results of the complete blood cell count and test indicators of the sample; S2: constructing the blood routine scoring model based on the markers; S3: constructing a preeclampsia prediction model based on the blood routine scoring model and the detection indicators; S4: Calculate the probability of the sample suffering from preeclampsia using the preeclampsia prediction model.

10. The construction method according to claim 9, wherein: The samples were collected between 8 and 14 gestational years. +6 week.

Images